Untitled - CSIR – National Institute for Interdisciplinary Science ...
Toward an interdisciplinary perspective
-
Upload
independent -
Category
Documents
-
view
4 -
download
0
Transcript of Toward an interdisciplinary perspective
ADVANCES IN LIFE COURSE
RESEARCH
Series Editor: Timothy Owens
Recent Volumes:
Volume 1: Work, Retirement and Social Policy, 1986
Edited by Zena Smith Blau
Volume 2: Family Relations in Life Course Perspective, 1986
Edited by David I. Kertzer
Volume 3: Personal History Through the Life Course, 1993
Edited by R. S. Olusegun
Volume 4: Delinquency and Disrepute in the Life Course,
1995
Edited by Zena Smith Blau and John Hagen
Volume 5: Self and Identity Through the Life Course in
Cross-Cultural Perspective, 2000
Edited by Timothy J. Owens
Volume 6: Children at the Millennium: Where Have
We Come From, Where Are We Going?, 2001
Edited by Sandra L. Hofferth and
Timothy J. Owens
Volume 7: New Frontiers in Socialization, 2002
Edited by Richard A. Settersen, Jr. and
Timothy J. Owens
Volume 8: Changing Life Patterns in Western Industrial
Societies, 2004
Edited by Janet Zollinger Giele and Elke Holst
Volume 9: The Structure of the Life Course: Standardized?
Individualized? Differentiated?, 2005
Edited by Ross Macmillanii
ADVANCES IN LIFE COURSE RESEARCH VOLUME 10
TOWARDS ANINTERDISCIPLINARYPERSPECTIVE ON THE
LIFE COURSE
EDITED BY
RENE LEVYUniversite de Lausanne
PAOLO GHISLETTAUniversite de Geneve
JEAN-MARIE LE GOFFUniversite de Lausanne
DARIO SPINIUniversite de Lausanne
ERIC WIDMERUniversite de Lausanne
Amsterdam – Boston – Heidelberg – London – New York – Oxford
Paris – San Diego – San Francisco – Singapore – Sydney – Tokyo
iii
r 2005 Elsevier Ltd. All rights reserved.
This work is protected under copyright by Elsevier Ltd, and the following terms and conditions apply to its use:
PhotocopyingSingle photocopies of single chapters may be made for personal use as allowed by national copyright laws. Permissionof the Publisher and payment of a fee is required for all other photocopying, including multiple or systematic copying,copying for advertising or promotional purposes, resale, and all forms of document delivery. Special rates areavailable for educational institutions that wish to make photocopies for non-profit educational classroom use.
Permissions may be sought directly from Elsevier’s Rights Department in Oxford, UK: phone (+44) 1865 843830, fax(+44) 1865 853333, e-mail: [email protected]. Requests may also be completed on-line via the Elsevierhomepage (http://www.elsevier.com/locate/permissions).
In the USA, users may clear permissions and make payments through the Copyright Clearance Center, Inc., 222Rosewood Drive, Danvers, MA 01923, USA; phone: (+1) (978) 7508400, fax: (+1) (978) 7504744, and in the UKthrough the Copyright Licensing Agency Rapid Clearance Service (CLARCS), 90 Tottenham Court Road, LondonW1P 0LP, UK; phone: (+44) 20 7631 5555; fax: (+44) 20 7631 5500. Other countries may have a local reprographicrights agency for payments.
Derivative WorksTables of contents may be reproduced for internal circulation, but permission of the Publisher is required for externalresale or distribution of such material. Permission of the Publisher is required for all other derivative works, includingcompilations and translations.
Electronic Storage or UsagePermission of the Publisher is required to store or use electronically any material contained in this work, including anychapter or part of a chapter.
Except as outlined above, no part of this work may be reproduced, stored in a retrieval system or transmitted in anyform or by any means, electronic, mechanical, photocopying, recording or otherwise, without prior written permissionof the Publisher.
Address permissions requests to: Elsevier’s Rights Department, at the fax and e-mail addresses noted above.
NoticeNo responsibility is assumed by the Publisher for any injury and/or damage to persons or property as a matter ofproducts liability, negligence or otherwise, or from any use or operation of any methods, products, instructions orideas contained in the material herein. Because of rapid advances in the medical sciences, in particular, independentverification of diagnoses and drug dosages should be made.
First edition 2005
British Library Cataloguing in Publication DataA catalogue record is available from the British Library.
ISBN: 0-7623-1251-3ISSN: 1040-2608 (Series)
∞ The paper used in this publication meets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper).Printed in The Netherlands.
ELSEVIER B.V.
Radarweg 29
P.O. Box 211
1000 AE Amsterdam,
The Netherlands
ELSEVIER Ltd
84 Theobalds Road
London
WC1X 8RR
UK
ELSEVIER Ltd
The Boulevard, Langford
Lane, Kidlington
Oxford OX5 1GB
UK
ELSEVIER Inc.
525 B Street, Suite 1900
San Diego
CA 92101-4495
USA
Working together to growlibraries in developing countries
www.elsevier.com | www.bookaid.org | www.sabre.org
iv
CONTENTS
PREFACE ix
LIST OF CONTRIBUTORS xi
INTRODUCTION
WHY LOOK AT LIFE COURSES IN ANINTERDISCIPLINARY PERSPECTIVE?
Rene Levy and the Pavie Team 3
PART I: AGENCY AND STRUCTURE
STRUCTURE, AGENCY, AND THE SPACE BETWEEN:ON THE CHALLENGES AND CONTRADICTIONS OFA BLENDED VIEW OF THE LIFE COURSE
Richard A. Settersten, Jr. and Lynn Gannon 35
AGENCY, EVENTS, AND STRUCTURE AT THE ENDOF THE LIFE COURSE
Victor W. Marshall 57
LOOKING AT AMBIVALENCES: THECONTRIBUTION OF A ‘‘NEW-OLD’’ VIEW OFINTERGENERATIONAL RELATIONS TO THE STUDYOF THE LIFE COURSE
Kurt Luscher 93
v
PART II: TRANSITIONS
AGENCY AND STRUCTURE IN EDUCATIONALATTAINMENT AND THE TRANSITION TOADULTHOOD
Jeylan T. Mortimer, Jeremy Staff and Jennifer C. Lee 131
NON-NORMATIVE LIFE COURSE TRANSITIONS:REFLECTIONS ON THE SIGNIFICANCE OFDEMOGRAPHIC EVENTS ON LIVES
Frank F. Furstenberg 155
THE SECRET OF TRANSITIONS: THE INTERPLAY OFCOMPLEXITY AND REDUCTION IN LIFE COURSEANALYSIS
Katherine Bird and Helga Kruger 173
PART III: BIOGRAPHICAL RECONSTRUCTION
LIFE COURSE TRANSITIONS AND SOCIALIDENTITY CHANGE
Nicholas Emler 197
THE IMPACT OF PERSONALITY AND LIVINGCONTEXT ON REMEMBERING BIOGRAPHICALTRANSITIONS
Pasqualina Perrig-Chiello and Walter J. Perrig 217
STUDYING LIVES IN TIME: A NARRATIVEAPPROACH
Dan P. McAdams 237
CONTENTSvi
PART IV: METHODOLOGICAL INNOVATIONS
LIFE COURSE ANALYSIS: TWO (COMPLEMENTARY)CULTURES? SOME REFLECTIONS WITH EXAMPLESFROM THE ANALYSIS OF THE TRANSITION TOADULTHOOD
Francesco C. Billari 261
LIFE COURSE DATA IN DEMOGRAPHY ANDSOCIAL SCIENCES: STATISTICAL AND DATA-MINING APPROACHES
Gilbert Ritschard and Michel Oris 283
FIVE STEPS IN LATENT CURVE MODELING WITHLONGITUDINAL LIFE-SPAN DATA
John J. McArdle 315
AFTERTHOUGHTS
INCITATIONS FOR INTERDISCIPLINARITY IN LIFECOURSE RESEARCH
Rene Levy, Paolo Ghisletta, Jean-Marie Le Goff, DarioSpini and Eric Widmer
361
Contents vii
PREFACE
This volume has grown out of an international colloquium on interdisciplinarity
in life course research organized by the Center for life course and life style studies
(PAVIE) at the Universities of Lausanne and Geneva (Switzerland) in October
2003. The contributors were asked to react to a working document prepared by
the organizers but had entire freedom as to how they wanted to do it.
The authors represent four disciplines directly involved in life course research:
developmental and personality psychology, social psychology, sociology, and
social demography. The focus on these four social-science disciplines in a ven-
ture into interdisciplinarity should certainly not be read as a programmatic
limitation of our wished-for disciplinary scope of collaboration. On the con-
trary, we fully acknowledge that other disciplines have made important con-
tributions to life course research and that they are by no means located at its
margins – let us only think of social history or political science. The potential
role of biology and social medicine may be more controversial, but there again,
our selection is pragmatic and provisional, not exclusive and even less ideolog-
ical. The aim of this volume is to contribute to the enlargement and strength-
ening of collaboration between all disciplines able to develop an interest in life
course or life span research.
The colloquium as well as some of the subsequent work necessary for this
volume to materialize has been financed by the Swiss National Science Foun-
dation, and a series of services and facilities have been contributed by the Uni-
versity and the Municipality of Lausanne, an indispensable help for which we
are very grateful. Beyond the contributors who are, of course, the main actors of
this volume, we owe special thanks to Tatiana Lazzaro who did a great job all
along this enterprise, beginning with the organization of the colloquium and
ending with a large amount of administrative and editing services she delivered
with exceptional enthusiasm, competence and efficacy. We also thank Tim
Owens, the editor of this series, who was highly interested in our endeavor from
the outset and extremely helpful for its taking the shape our readers can ap-
preciate now.
Rene Levy, Paolo Ghisletta, Jean-Marie Le Goff, Dario Spini and
Eric Widmer (Editors)
Lausanne/Geneva
ix
LIST OF CONTRIBUTORS
Francesco C. Billari Istituto di Metodi Quantitativi, Universita
Bocconi and Innocenzo Gasparini Institute for
Economic Research, Milano, Italy
Katherine Bird Universitat Bremen, EMPAS, Bremen,
Germany
Nicholas Emler Department of Psychology, University of
Surrey, Guildford, UK
Frank F. Furstenberg University of Pennsylvania, Department of
Sociology, Philadelphia, PA, USA
Lynn Gannon Department of Sociology, Case Western
Reserve University, Cleveland, OH, USA
Paolo Ghisletta Centre Interfacultaire de Gerontologie (CIG) et
Faculte de Psychologie et des Sciences de
I’education, Universite de Geneve, Centre
lemanique d’etude des parcours et modes de vie,
Universites de Lausanne et de Geneve, Geneve,
Switzerland
Helga Kruger Universitat Bremen, EMPAS, Bremen,
Germany
Jennifer C. Lee Life Course Center, Department of Sociology,
University of Minnesota, Minneapolis, MN,
USA
Jean-Marie Le Goff Centre lemanique d’etude des parcours et
modes de vie, Universite de Lausanne,
Lausanne, Switzerland
Rene Levy Centre lemanique d’etude des parcours et
modes de vie, Universite de Lausanne,
Lausanne, Switzerland
xi
Kurt Luscher Fachgruppe Soziologie, Universitat Konstanz,
Germany
Victor W. Marshall UNC Institute on Aging, University of North
Carolina at Chapel Hill, NC, USA
Dan P. McAdams Northwestern University, Evanston, IL,
USA
John J. McArdle Department of Psychology, University of
Virginia, Charlottesville, VA, USA
Jeylan T. Mortimer Life Course Center, Department of Sociology,
University of Minnesota, Minneapolis, MN,
USA
Michel Oris Department of Economic History, University
of Geneva, Geneva, Switzerland
Walter J. Perrig Institut fur Psychologie, Universitat Bern, Bern,
Switzerland
Pasqualina Perrig-
Chiello
Institut fur Psychologie, Universitat Bern, Bern,
Switzerland
Gilbert Ritschard Department of Econometrics, University of
Geneva, Geneva, Switzerland
Richard A. Settersten,
Jr.
Department of Sociology, Case Western
Reserve University, Cleveland, OH, USA
Dario Spini Centre lemanique d’etude des parcours et
modes de vie, Universite de Lausanne,
Lausanne, Switzerland
Jeremy Staff Department of Sociology and Crime, Law and
Justice, The Pennsylvania State University,
University Park, PA, USA
Eric Widmer Centre lemanique d’etude des parcours et
modes de vie, Universite de Lausanne,
Lausanne, Switzerland
LIST OF CONTRIBUTORSxii
WHY LOOK AT LIFE COURSES IN
AN INTERDISCIPLINARY
PERSPECTIVE?
Rene Levy and the Pavie Team1
CHALLENGES OF LIFE COURSE RESEARCH
After decades of a rather marginal existence and little coherence in its de-
velopment, life course research is definitely coming of age. There recent
signs of consolidation and first attempts at reaping the scattered harvest of
research in various disciplines, especially in the form of a first handbook of
life course research (Mortimer & Shanahan, 2003), of first attempts at
interdisciplinary dialogue around specific approaches such as Baltes’ life-
span psychology2 (Staudinger & Lindenberger, 2003) and of a specialized
annual review (in which this volume is published). Nevertheless, life course
scholars still seem to be a small handful ‘‘digging’’ on the fringe of their
disciplinary mainstreams, as yet with little influence on more established
fields of research. Why are we, i.e., life course researchers, so keen on life
courses? What is there so special about life course research? We feel in fact
that there are a number of specific challenges life course researchers have to
confront and answer.
A first, global and not very differentiated reason for finding it particularly
interesting could be that it encompasses all we find important about human
life and that everything humanly relevant is in the life course.
Towards an Interdisciplinary Perspective on the Life Course
Advances in Life Course Research, Volume 10, 3–32
Copyright r 2005 by Elsevier Ltd.
All rights of reproduction in any form reserved
ISSN: 1040-2608/doi:10.1016/S1040-2608(05)10014-8
3
A second answer, of the same global kind but a bit more specific, could be
that only a life course approach takes fully into account the fact that our
lives are ongoing processes and not just single states or events that can be
adequately captured and understood using snapshots.
A third reason of attractiveness, even more fascinating but also demand-
ing, is the cross-cutting and integrative nature of the life course perspective
with respect to most of our more conventional, institutionalized disciplinary
specialties. To illustrate this statement, take the case of sociology.3 There are
special sociologies of childhood, of youth or adolescence, (maybe soon also
of post-adolescence,) of retirement, and of aging. Each ‘‘age group’’ is
treated in a more or less static perspective and therefore reified as a some-
how homogenous social category and not as a phase in the process of life
course unfolding. There are special sociologies of the family, of work, of
labor markets, of leisure, of stratification and mobility, of voluntary asso-
ciations, of social movements and so on. Each social field is mainly treated
as an isolated entity with its inner logic and specific pathways for the in-
dividuals who participate in these fields. Life-phase specific and sector spe-
cific sociologies have developed their specialized perspectives, slicing up the
lives and contexts of the ‘‘whole individuals’’ we pretend to be into various
aspects, types of social relations, phases, or fields of action. Likewise, in
psychology, ontogeny has been regarded by many as occurring almost
differently in distinct fragments of the life span. Much as in sociology, the
American Psychological Association, among its 53 divisions, has Division
7 for Developmental Psychology and Division 20 for Adult Development
and Aging. Similarly, Division 12 is for the Society of Clinical Psychology,
while Division 53 for the Society of Clinical Child and Adolescent Psy-
chology. Again, the fragmentation of the life course has been, for some
aspects at least, institutionalized. Against this reductionistic, fragmenting
tendency, Settersten (2003b, p. 196) has rightly made a strong point of ‘‘The
importance of understanding people in whole (over time) and as wholes
(studying larger profiles of traits and characteristics rather than single var-
iables)’’. Without an encompassing perspective, life course research cannot
meet the challenge of its very raison d’etre.
On a more fundamental level, there is a fourth challenge which is already
included in these remarks but not explicitly spelled out: life course analysis is
one of the rare ways social sciences have developed so far to conceptualize
time, not just as a physical happening whose whereabouts escape our the-
oretical understanding, but as something that is culturally, socially and also
individually molded, reworked and ‘‘constructed’’, something that has both
a subjective and an objective existence, and whose very objectivity is the
RENE LEVY AND THE PAVIE TEAM4
result of social objectiveness rather than of mere ‘‘being there’’ like the
eternal rhythm of atomic oscillation. Settersten (1999) has entitled his recent
stock-taking book ‘‘Lives in Time and Place’’, referring to two basic di-
mensions of the real world that both are not just physical ‘‘givens’’, but
socially constructed: social space and social time. In classical developmental
psychology, most research was concerned with charting along the age axis
the newly acquired skills and mastered developmental tasks by children.
Refined descriptions of what the child could achieve at what age flourished,
and in various domains age-related norms were established. Wohlwill (1970)
was one of the first to question the limitations of the mere descriptive role of
chronological age. (We shall take up the question of what status to give to
age in our theorizing about life courses in the final chapter.)
In short, in both a humanistic and in a scientific view, life course research
is fascinating because it forces us to break down traditional limitations of
understanding.
THE NECESSITY OF INTERDISCIPLINARITY
If adopting a life course perspective in a single study does not exclude fo-
cusing on a specific aspect of peoples’ lives, the perspective as such is nec-
essarily integrative: it has to bridge a number of institutionalized chasms in
and between social science disciplines. These chasms are sometimes openly
discussed, even questioned, and sometimes rather taken for granted. At any
rate they tend to limit our awareness, inducing what we may call an interna-
listic perspective, artificially bounded by the boundaries of the specialty in
which our scholarly activity is organized. We have already mentioned the two
dimensions life course researchers have to bridge in their work, the age-group
axis and the life-spheres axis. At least three others are of equal importance:
� What sociologists, and others as well, are used to call the macro–micro
link, i.e., more generally, the fact that one basic dimension in the com-
plexity of reality is the nesting of encompassing systems of various levels;4
this dimension (we might call it systemic differentiation to distinguish it
from two others, hierarchization or parallelism) takes all its importance in
the debate about the institutionalization of life course patterns as opposed
to actorial agency although it is more often invoked as important than
really taken into account.� Gender differences or more generally the social relationships between the
sexes,5 but also a number of other cases of strong social differentiation of
both a cultural and a structural kind that have strong incidences on the
Why Look at Life Courses in an Interdisciplinary Perspective? 5
level of individual identities (like ethnic or religious group membership,
etc.), that are often leveled out by static analyses or even ignored by gender-
blind or more generally ‘‘differentiation-blind’’ theoretical models.6
� Linked lives, the strong interdependencies between the life courses of re-
lated persons that remind us that the real object of most of the social
sciences are relations between actors and not just monadic individuals as
some current mainstream paradigms would have it.7
If we are to bridge successfully all these gaps, with their thematical,
conceptual and methodological implications, one may object that we have
to become scientific geniuses. As social scientists we know that individual
mastery is not all, that groups are frequently wiser than the ‘‘addition’’ of
their individual members, so teamwork is certainly a way out of this prob-
lem. Working integratively in the sense just mentioned calls for teamwork.
Among the gaps that work on life courses intrinsically needs to bridge
figures also the one between scientific disciplines, above all in the social
sciences (including, of course, psychology). The arguments for inter-
disciplinarity in this area may not be fundamentally different from those
applying to other themes – as soon as we get interested in a real world
problem and not only an epistemological slice of it, we cannot be satisfied by
adopting one disciplinary perspective only. Most often, interdisciplinarity is
claimed for problem-oriented or ‘‘applied’’ research. If the research objec-
tive is to resolve a problem, disciplinary purity and coherence is of no
interest, we mobilize all promising resources whatever their disciplinary
origin and theoretical ramifications. This may not be the primary orienta-
tion of life course research. But there is another important and sufficient
motive for interdisciplinarity in this area that we have already encountered:
the holist perspective on individuals and their development in social context
which cannot be contented by any single aspect alone, be it sectoral or
disciplinary. It is no accident that interdisciplinarity is frequently called for
in this area. When formulating his ‘‘life course paradigm’’, Elder (1995)
included interdisciplinarity explicitly, but the overriding research praxis
remains monodisciplinary.8
A further question is then: what should we mean by interdisciplinarity?
We usually use the term in a very general sense, meaning just going beyond
the limits of any one of our habitual disciplines (monodisciplinarity). In
order to be more precise, we may join an emergent distinction between
� multidisciplinarity (or pluridisciplinarity), meaning the cooperation of
several disciplines that work on complementary aspects of a common
overall question,
RENE LEVY AND THE PAVIE TEAM6
� interdisciplinarity, meaning the integrated cooperation of disciplines on
the basis of common concepts, and� transdisciplinarity, meaning the scientific work on the basis of common
concepts that do not belong to any specific discipline and that may also
include the points of view of extra-scientific actors concerned by the
question under study.9
There is no general, absolute value hierarchy to be established between
these possibilities, ‘‘trans’’ is not necessarily better than ‘‘multi’’ and there
may very well be problems whose ‘‘mono’’ study is quite adequate. The
proper level between mono- and transdisciplinarity must be identified for
each research enterprise, and the optimum may very well be an evolving mix
and not a one-level solution fixed once and forever. But what is certainly
necessary is the capacity of people and teams to work on these different
levels. In order for such an enterprise to work, a more or less clearly stated
common goal is of course essential, but other conditions are of equal im-
portance: the real desire of the participants to work together, their ac-
knowledgment of the potential of mutual learning rather than competition,
even less the pursuit of disciplinary hegemony, and a fundamental working
knowledge allowing for an understanding and appreciation of the concep-
tual and methodological panoply of neighboring disciplines.
There is a huge gap between the almost unisonous choir of public voices
asking for interdisciplinarity, and scientific everyday practice that is needed
to discover the real difficulties of such an endeavor to begin with. A realistic
approach must probably start from the principle with a view to interdisci-
plinary convergence or at least cooperation, differences should not be si-
lenced, but discussed and integrated. Well-placed protagonists assure us that
this is no easy business – witness, as one example, Mayer (2002) who coolly
states on the encounter of life course sociology and life-span psychology (in
which he has been engaged himself): ‘‘In retrospect, despite all the strong
mutual recognition and reinforcement, surprisingly little convergence and
integration has actually occurred’’. There is definitely still some way to make;
the present volume is meant to bring us a step ahead on this way.
DIFFICULTIES OF INTERDISCIPLINARITY
As long as not many social scientists themselves have a really interdisciplinary
education and working profile, interdisciplinary work is bound to be team-
work. Along with more current organizational questions, interdisciplinary
Why Look at Life Courses in an Interdisciplinary Perspective? 7
teams have to confront quite real problems in their everyday functioning,
even if these may often seem somewhat trivial when looked at from a the-
oretical vantage point. These practical difficulties need to be taken seriously
in order to overcome them. Let us briefly look into some of them. But an
assured rootedness in their own disciplines of scholars working together in an
interdisciplinary team is most probably a precondition for successful inter-
disciplinarity, even though this may appear as a paradox.
A first kind of difficulty to be clarified and surpassed on the way to
interdisciplinary teamwork is in the area of vocabulary. Our disciplines use
some identical terms with quite different meanings (e.g., ‘‘norm’’ has an
entirely cultural, obligational ring in sociology, but the objectivist meaning
of statistical prevalence – or even biological ‘‘normality’’ – in developmental
psychology; or the different semantic extension of the notion of ‘‘micro’’
and ‘‘macro’’ between the same two disciplines). The other way round, we
sometimes use different terms for the same things (e.g., in the methodolog-
ical area). One example seems to pose few problems: the more sociological
or demographical term of life course seems to overlap very largely with the
more psychological one of life span. We therefore consider them in this
chapter as synonyms. Another, probably minor, example is different names
used in different disciplines for the same methods (e.g., survival analysis,
event history analysis, or, in French, ‘‘demographical analysis of bio-
graphies’’10 in the current terminology of demography all designate the same
or at least highly similar analytical methods; the same is true for multilevel
models – random effect models – mixed effect models and also hierarchical
models).
Other differences, that cannot be ruled out by mere lexical agreements,
are related to the typical frames of attention of the disciplines. Mainstream
(or traditional) developmental psychology focuses largely on very early (and,
to a lesser extent, very old) years of life while the bulk of sociological work
concerns the adult years (mostly without spelling out explicitly this horizon).
At first sight, social psychology does not seem to be very specific in this
respect; in fact, it shows only rarely an outspoken awareness of and interest
in processes of growing older or in the life course. But there is a hidden and
rarely acknowledged specificity inherent in much research in this discipline:
a large fraction of empirical research in social psychology, at least about
attitudes and representations, is done with the most accessible category of
experimental subjects, i.e., with students, who represent quite a specific age
group as well as a set of rather specific locations in social stratification.
Little is known about the impact of this double background specificity on
the results and scientific knowledge in this discipline. Two transitions have
RENE LEVY AND THE PAVIE TEAM8
attracted the attention of social demographers, the first from adolescence to
adulthood and the second the end of life (mortality).
There are, of course, many conceptual differences that may constitute as
many difficulties for mutual understanding and cooperation. Let us only
mention one example: the weight that different social science disciplines
usually attribute to biology vs. environment (or nature vs. nurture in a more
traditional formulation). While psychology has largely put aside earlier ap-
prehensions with respect to close contact with biology, sociology and even
more political sciences or social and cultural anthropology have up to now
maintained a greater distance; demography is hard to locate on this di-
mension, but would certainly have little hesitations to be situated near psy-
chology, and potentially even could stretch its interests as far as to reach not
only psychology or sociology, but also biology in the sense that it is cur-
rently interested in biologically structured events like birth and death. This
basic opposition has a close relative in the difference between internal and
external explanations of the ‘‘socialization vs. institutional channeling’’ va-
riety (e.g., concerning the probable impact of age-normative conceptions vs.
context or environmental conditions). Confronted to such differences in
basic assumptions that characterize the often implicit scientific cultures of
different disciplines, interdisciplinary collaboration needs above all the re-
nouncement of rigorous a priori stands and a pragmatic curiosity for the
other disciplines’ way of looking at things. Another important and poten-
tially dividing difference concerns the distinction between factual and rep-
resentational aspects that may be highlighted by the far-reaching attachment
to statistical description by demography on the one hand and the heavy
emphasis of most of social psychology on individual and collective repre-
sentations, on the other. If we push this line of thinking further, there are of
course situations where it may become very difficult to find common ground
between disciplines, like psychology and demography, because they are
normally interested in many different topics – but then, we are also leaving
the array where interdisciplinarity really makes sense.
PATHWAYS TO INTERDISCIPLINARITY
Beyond such unavoidable practical questions and their solutions, whose
importance must not be underestimated, we should also think about more
theoretically grounded pathways to real interdisciplinarity. In this section,
we wish to propose three such routes for practical exploration: direct the-
oretical linkages between concepts of different disciplines, identification of
Why Look at Life Courses in an Interdisciplinary Perspective? 9
formal characteristics of life courses that may acquire similar meanings for
different disciplines and specific substantial themes that may link different
disciplinary outlooks due to their transversal relevance. Our presentation will
necessarily remain abstract, with only limited illustrations, as our aim cannot
go beyond sketching some promising directions; their exploration is yet to
come, but hopefully, this outline can provoke some fruitful discussions that
help us progress in this sense. We feel moreover that systematic scrutiny of
the extant literature would produce a wealth of examples that have already
arisen out of research practice. Our proposal is probably little more than a
possible systematization of thoughts already present in our field.
THEORETICAL LINKAGES
A first possible pathway to the construction of interdisciplinarity in life
course research consists of creating theoretical linkages between concepts of
different disciplines. If this is not an explicit part of our everyday practice
when doing research, it is often implicitly present. If we take the example of
the relationship between sociology and social psychology, we may say that
many – perhaps all – sociological hypotheses located on the micro-
interactional level and concerning behavior of individuals and groups are
based on implicit assumptions of social–psychological mechanisms relating
the (sociological) concepts used. The most trivial such mechanism is some
kind of social rationality, be it along the lines of rational choice theory or
according to more complex paradigms. Less trivial examples are social
comparison, social learning, the pursuit of social recognition, reciprocity
and advantage, or conformity to social norms. Another, quite different
mechanism of that kind would be the one operating the transformation of
internal into external attribution in the course of social interactions leading
the implied actors to realize that each of them is not alone to experience a
given problem but that there is a social category of people with whom they
share the problem in question (classical example: the formation of the labor
movement, and social movements more generally), and the personally ‘‘al-
leviating’’ consequence of such processes of collectivization of subjectively
felt problems. Many hypotheses about social behavior, e.g., concerning bio-
graphical decisions with a view to social mobility, currently assume the
functioning of such mechanisms without normally making them explicit,
even less testing them. Similar relationships may exist or be developed
between other disciplines.
RENE LEVY AND THE PAVIE TEAM10
A slightly different example is the formulation of direct interdisciplinary
hypotheses, i.e., hypotheses linking dimensions ‘‘belonging’’ to different
disciplines. Stating more explicitly Baltes’ (1987, 1997) thesis of contextu-
ality in psychological development directly leads to such hypotheses, for
instance in postulating how structural locations of individuals (in the larger
social structure, in their family, in one of their peer groups) explain elements
of their identity formation, ways of psychological or interactional styles, etc.
FORMAL ASPECTS OF LIFE COURSES:
TRAJECTORIES, STAGES,
TRANSITIONS AND EVENTS
We know from research that life course transitions (such as entering or
leaving school, entering professional life or parenthood, marriage, divorce,
retirement, moving into nursing homes in old age) have a great significance
for individual identities and for the connection between people and insti-
tutions. The stages and transitions of life courses are significant in terms of
cognitive abilities, representations of the self, relations with significant oth-
ers and with the institutional and the societal order. Four current concepts
are directly related to any process perspective on the life course: trajectory,
stage, transition and event; they are central for the understanding of human
action, relatively common to various disciplines, and may therefore consti-
tute another inroad to facilitating the construction of interdisciplinary
cooperation.
Trajectory
The meaning of trajectory can be defined as ‘‘model of stability and long-
term changes’’ (George, 1993) or ‘‘sequence of profiles of insertion’’ par-
ticipation (Levy, 2001). In this sense, the notion of trajectory is mainly used
to describe the movements or developments occurring during the whole
span of life, i.e., all that takes place between the two ultimate life boundaries
– or, in our context, transitions – that are birth and death. If this explication
remains especially close to sociological use, one finds a similar one in social
psychology where trajectories are conceived of as all the movements of an
individual in the social space (see among others Viaud, 1999, p. 80) and in
social demography with the notion of biography (Courgeau & Lelievre,
Why Look at Life Courses in an Interdisciplinary Perspective? 11
1989, p. 56).11 On a more formal level, we may view trajectories as com-
posed by a sequence of transitions (change) and stages (stability).
Cognitive-developmental psychology proposes an in-depth analysis of the
models of change and stability in the long run (life span). The idea of the
integration of a lifelong trajectory results from the reformulation of the
concept of development. The principle of a definite succession of stages in a
classical structuralist perspective is completed by that of a sequence of
compromises and settlements (growth, maintenance/resilience, regulation)
arising from heterogeneous resources (biological, social, cultural). This,
considered under the angle of individual development (or ontogeny) and
cognitive capacities in particular, contributes to shaping an individual
trajectory.
Baltes and Schaie (1976) discuss cohort effects on cognitive performance.
Such differences were often confounded with age effects. These results jus-
tify the distinction of typical trajectories that correlate strongly with insti-
tutional influences and societal transformations while these may in turn
affect cognitive performance. Baltes (1987) discussed the relative regularity
of processes of change during the first years of life. Baltes interpreted these
results by stating that the social fabric is more solid, substantial and ho-
mogeneous during childhood than during adulthood, the latter showing less
regularity due to a greater variability of situational factors. Moreover, one
could add that the more relevant influences of biological nature during
childhood (maturation, etc.) imply more regularity in that section of the
developmental trajectory.
The shift to a social-psychological and sociological perspective on this
question happens in the first place by considering the principle of ‘‘linked
lives’’ formulated by Elder (1974): trajectories of the individual members of
the same primary group (family, friends, workmates) are interdependent.
For instance and in principle, it is necessary to take into account the tra-
jectory of the father and the mother of any individual to understand the
logic underlying that individual’s trajectory. In sociological research, this
principle is widely applied, notably in research on strategies of social mo-
bility, socialization, domestic functioning, etc. The same principle could
bring important results in interdisciplinary research.
In order to build a more complete model, it is advisable to add institu-
tional, historical and geographical dimensions (Kohli, 1985; Elder, 1985;
Mayer & Schoepflin, 1989; Heinz, 1992) to these individual and relational
factors. Ideally, the ambition of such a global model would be to capture
and interpret the interdependences between these various developmental
dynamics. In practice, however, one often prefers to distinguish specific
RENE LEVY AND THE PAVIE TEAM12
trajectories (marital, professional, health-related, residential) within the
global trajectory.12 This raises two important questions: on the one hand we
need to know how to take account of the mutual influences between these
particular trajectories instead of studying them separately. On the other, is it
after all possible and/or desirable to merge the separate trajectories into a
unique global trajectory?
In the perspective of modeling multidimensional trajectories, Abbott
(1992), a historian and sociologist working on both the empirical and the-
oretical level of trajectories, proposes a ‘‘narrative’’ approach where the
notion of narration is substituted to that of causality. According to Abbott,
one of the major problems of research concerning trajectories lies in the fact
that one is mostly confronted with a variety of sequences of events occurring
at different speeds. These multiple temporal horizons constitute a theoretical
barrier that prevents researchers from raising formalized stories beyond the
‘‘simplistic’’ analyses of stage process (stories which develop relatively in-
dependently) and sequences of rational actions. The possibility of matching
sequences of events occurring at various speeds might prove very useful for
life course research. The study of life courses might require the integration
of seemingly quite heterogeneous dimensions in a unique interpretational
model. Ontogenesis does not simply follow a linear succession of ordered
stages. Thus trajectories result from the interaction of various factors of
different kinds, such as biological, psychological, relational and institutional
ones, all probably occurring at different paces and with different impacts on
the life course. This fact raises important theoretical and methodological
questions and surely requires an interdisciplinary collaboration.
Stage
In sociology, demography, social psychology, psychology as well as in other
disciplines, the notion of stage refers to a life period of variable length,
characterized by a relative stability and often something like ‘‘balance’’. At
the very least and on the most abstract level, a stage can be defined as a
‘‘stable state between two transitions.’’ This definition is rather descriptive
and may constitute an empirical cut-out of reality based on the criteria
defined beforehand. Various terms (episode, state, phase, etc.) are based on
a similar logic or similar conceptions across the disciplines. At times, the
duration of a stage offers a better insight.
Jean Piaget has greatly contributed to the definition of the concept of
stage, especially in psychology. He defines stages as relatively long periods
Why Look at Life Courses in an Interdisciplinary Perspective? 13
resting on the same underlying structures and equilibrium states.13 In
cognitive developmental psychology, neofunctionalists (or neo-Piagetians)
and contextualists have reconsidered this concept by means of notions such
as ‘‘equifinality’’ (Kruglanski, 1996) and ‘‘testing the limits’’ (Baltes,
Lindenberger, & Staudinger, in press). Neostructuralists have maintained
certain Piagetian principles and added the following modifications: (1) the
redefinition of cognitive structures (other structures than Piaget’s ‘‘logico-
mathematical’’ ones), not necessarily seen as universal; moreover, cultural
and linguistic factors are also considered; (2) a stage is now defined by an
upper cognitive limit, rather than by certain behaviors in different situa-
tions; the upper limit moves during ontogenesis, hence, maturation and
aging are of chief importance; (3) individual differences and characteristic
patterns of individual development are considered; and (4) it is assumed that
affective and social development are similar to cognitive development.
Moreover, the notion of stage has been extended and applied to adult life
(Labouvie-Vief, 1980, 1982; Edelstein & Noam, 1982; Pascual-Leone, 1983;
Riegel, 1976; but see already Erikson, 1950).
In social psychology the concept of stage is used to refer to bases of
retrospective biographical memories (McAdams, 1993, 1999). However, in
many applications of research, the Piagetian conception of stage has been
maintained. Moral development, as conceived by Kohlberg and his follow-
ers (Turiel, 1983; Gilligan & Attanucci, 1988), has kept the universal de-
terministic flavor. However, this conception has been criticized for
theoretical and methodological reasons (Emler, 1999; Tostain, 1999).
In social demography, the concepts of episode or phase are usually used to
refer to well-defined life periods (e.g., employment phases and phases of
professional inactivity for women). Sometimes, the term of age is used to
designate a broader phase of the life course (e.g., adult age).14 However, the
application of the term of stage in a sense close to the Piagetian tradition
seems to be recent (De Bruijn, 1999); its universalistic and deterministic
features are mostly avoided. Instead, emphasis is put on human agency.
Furthermore, the deterministic feature of a stage is opposed to the stochastic
character attached to events.
In sociology, the concept of stage refers to a stable state of various dura-
tions. The social constraints and normative expectations typical of each
stage are generally emphasized. Historically, in a macro-sociological per-
spective, models directly relying on the conception of stage (Comte, Marx,
Spencer, Parsons) have been strongly criticized because of their evolutionist,
linear and deterministic features. In a micro-sociological perspective, the
notion of stage has been applied to the description of family development
RENE LEVY AND THE PAVIE TEAM14
and, of course, to individual life courses where it is sometimes used to
designate age-groups (e.g., adolescents, middle-aged, etc.), sometimes also
categories defined by a specific institutional participation (like preschoolers,
employed, retired, etc.). Close terms such as period are often used to avoid
the restrictions of the more assumption-loaded term of stage.
Transition
Across disciplines, the concept of transition refers to the idea of change:
change from one state or situation to another, from one life period to an-
other, from one status or role to another. In this vein and by inverting the
above definition of stage, we might define a transition as the (short) period
of change relating consecutive stages. More precisely and less circularly,
transitions are moments within a particular trajectory characterized by ac-
celerated changes, compared to the relative stability of stages. Examples of
transitions are quite different according to the type of change considered:
change from a limited to a broadened capacity of information processing
(possibly leading from one type of reasoning to another), from single life to
marriage, from adolescence to adulthood, from the status of an employee to
that of a manager. Three characteristics attributed to the concept of tran-
sition seem to be rather consensual across the different approaches: (1) a
transition always refers to its outcome, a novel status of relative stability (in
psychology often characterized by a higher equilibrium or stability, but not
in social demography or sociology); the different approaches, depending on
their objects of study, their objectives and their theoretical assumptions, do
or do not have precise criteria allowing an evaluation of a transition in terms
of the developmental progress it conveys; (2) transitions are processes more
or less clearly limited in time (although they may have long-term conse-
quences); and (3) most often, the concept of transition is applied to the
changes in an individual life course, but it appears also in notions of social
transition, collective transition or even demographical transition.
Defining a phenomenon as transitional cannot be done in an absolute way
since by definition, a transition relates a state ‘‘before’’ to a state ‘‘after’’.
The use of the term depends on the extension and on the specific processes
implied by the transition and may sometimes be a specific way of looking at
an object rather than this object’s inherent characteristics. For instance,
adolescence can either be qualified as a transitional stage, in particular in a
perspective of development from childhood to late adulthood (Durkin,
Why Look at Life Courses in an Interdisciplinary Perspective? 15
1995), or as a life stage in itself, with its own developmental tasks, at the end
of which the transition to adulthood takes place (Erikson, 1950).
Three related concepts, each referring to a particular kind of transition,
are: (1) ‘‘revolution’’ (Mounoud, 1982), which refers to a discrete transition,
to the emergence of a new ‘‘structure d’ensemble’’ and thus does not apply
to the acquisition of a new isolated capacity; (2) ‘‘turning point’’ (Gotlib &
Wheaton, 1997), which refers to a transition (or perhaps more frequently to
an event) that implies a change in the orientation of a trajectory and not just
a mere confirmation of this trajectory by a transition that fits into a general
pattern; and (3) ‘‘social mobility’’, upward, downward or horizontal, which
is only related to transitions that affect the social status of a person, mostly
implying an increase or a reduction of his economic, social or cultural cap-
ital (in the case of vertical mobility).
A number of theorists have attempted to apply to the study of individual
transitions the mathematical ‘‘catastrophe theory’’ (Thom, 1975) and the
concept of ‘‘bifurcation’’ (e.g., van der Maas & Molenaar, 1992). Such a
modeling approach has the advantage of combining quantitative with qual-
itative changes. For instance, a series of small quantitative changes can
eventually lead to an abrupt transition, resulting in what proves to be
qualitatively very different from the preceding periods.
Although commonalities of the concept of transition exist across disci-
plines, each discipline differentiates its approaches to and uses of the notion
of transitions. Social demography and sociology share a focus on the study of
the effects of various parameters – socio-demographical (e.g., fertility, life
expectancies, migration, social and geographical mobility) or institutional
(compulsory school, labor market, retirement system, etc.) on the occur-
rence, timing and variability of transitions (Hogan, 1981; Rindfuss,
Swicegood, & Rosenfeld, 1987; Settersten & Mayer, 1997). In social psy-
chology, the same transitions are most often conceived of as independent
variables whose effects are studied (e.g., in studies interested in the effects of
particular transitions on social representations (Viaud, 1999), and self-
concept (Hagestad & Neugarten, 1985; Havighurst, 1972; Kling, Ryff, &
Essex, 1997)). Developmental cognitive psychologists are more interested in
explaining the mechanisms and explanatory factors of transitions charac-
terizing the individual cognitive development (de Ribaupierre, 1989). In a
Piagetian perspective, these mechanisms are described as depending on
processes of maturation, cognitive conflict, equilibration and the interaction
with the social environment. Here, transitions are usually conceived as de-
pendent variables and at a more microscopic level than in sociology or social
demography.
RENE LEVY AND THE PAVIE TEAM16
The idiosyncrasies of the various disciplines have not hindered interdis-
ciplinary work. For instance, some researchers in social psychology are
interested in life choice orientations (which directly influence normative
transitions) through the analysis of the impact of values or of social iden-
tities on transitions considered as dependent variables (Cinnirella, 1998).
Furthermore, acknowledging the role of the social environment and the
individual behavior on cognitive transitions requires that the disciplines
interact and provide complementary contributions to the study of the life
course. In this regard, Doise and Mugny (1997) illustrate the facilitating role
of social interactions and social marking in the resolution of Piagetian tasks.
Indeed, this line of research provides insights into the influence of social
dynamics on the transition from one cognitive stage to another, and it also
shows how these dynamics can be studied with an interdisciplinary ap-
proach. Another possible link between social psychology, sociology and
social demography is their common interest as to how transitions are reg-
ulated by normative representations as well as by life experiences and their
interpretation on the one hand, and on the other how these are shaped by
changes in their socio-historical background (Neugarten, Moore, & Lowe,
1965; Elder, 1974).
Event
According to a general definition, an event is ‘‘what happens at a given time
in a given place’’ (Encyclopaedia Universalis, 1999). This notion is ambig-
uous since some events are characterized by their singularity and unexpect-
edness, others by their regularity and expectedness. This ambiguity holds for
several events of the life course. For instance, life events (such as a second
marriage) can be considered as unique at the level of individuals. They ‘‘are
particular moments in a particular time and place, complete with particular
characters, actions, thoughts and feelings’’ (McAdams, 1993, p. 258). But
they can also be characterized by their repetition at the level of a social
group or a population. In this case, the event could be defined as a ‘‘fact
which concerns an individual and which affects the structure of populations
and its development’’ (Pressat, 1979, p. 68). Scientific disciplines could be
distinguished depending on the emphasis they give either to the singularity
or to the expectedness of an event. In this way, one can oppose, e.g., socio-
demographical and socio-psychological approaches.15
Social demography emphasizes regularity (i.e., repetition). This is achieved
by: (1) describing the event (e.g., marriage, birth, death); (2) considering this
Why Look at Life Courses in an Interdisciplinary Perspective? 17
event as part of a collective phenomenon (e.g., nuptiality, natality, mortal-
ity, mobility and migration); (3) using quantitative methods for this de-
scription; and (4) eventually predicting the event with the tools of
demographic projection. The meaning of the event is defined on the basis
of its statistic regularity. For instance, in the frequently cited paper titled
‘‘The changing meaning of cohabitation and marriage’’ of Manting (1996),
the meaning of cohabitation or marriage in the Netherlands is not derived
from Dutch values and opinions about these events, but from the devel-
opment of partnerships starting as cohabiting unions and marriages of co-
habiting couples. Entry into cohabitation was rare during the 70 s, and
cohabiting couples rarely married. However, this kind of union became
more and more frequent during the 80s and 90s, as cohabiting couples
married more often. According to Manting, the role of cohabitation evolved
from a substitute to marriage to a preliminary period before marriage.
Social psychology insists on the singularity of events. This is achieved by
assuming that a person defines an event and then by focusing the research
on the possible disturbance brought about by the event on the individual
identity. The event represents a mark between a ‘‘before’’ and an ‘‘after.’’
Because the event is defined by the individual (or the group), social psy-
chology insists on its subjective character. For example, the concept of
attribution captures the idea that various degrees of causal relations could
be established between behaviors and ‘‘reinforcements’’ (e.g., the interpre-
tation of an event as positive or negative; Deschamps & Beauvois, 1996).
Attribution leads to the notion of allocation (i.e., how persons can explain
events in their everyday life). Here, a distinction is made between ‘‘internal
and external causalities.’’ Internal causality means that an individual at-
tributes the event to him- or herself and external causality that he or she
names an external, environmental cause for the event. The current hypo-
thesis is that the type of attribution performed by the individuals has major
consequences on their future behaviors.
Sociology and social demography have a certain extent of common ground
as far as the definition of events is concerned, especially of events perceived
by the individuals as intrinsically motivated (e.g., marriage). Furthermore,
sociology also insists on the expectedness of the event (normative life events
vs. stressful events). In the case of stressful events, the interest lies especially
on their ‘‘consequences’’ for the life courses.
In cognitive psychology, the event is mainly apprehended as a stimulus or
a novel phenomenon or a disturbance the individual has to react to, and is
thus usually defined by the experimenter (i.e., simulated and manipulated as
an independent variable). Typically, the research is not interested in the
RENE LEVY AND THE PAVIE TEAM18
event itself, but in the individual reaction to the event understood as a
stimulus.
The meaning of the term ‘‘disturbing event’’ is sometimes only slightly
different in each discipline. In social demography, the succession of events is
addressed by most researchers. The questions circle around the effects of a
perturbing event on the occurrence of other events (e.g., the effect of mi-
gration on marriage; Courgeau & Lelievre, 1989). In this perspective, dis-
turbing events are similar to the ‘‘turning points’’ in social psychology or
sociology. These events coincide with trajectory changes (‘‘bifurcation’’ as
described in sociology by de Coninck and Godard (1989) is a closely related
concept). An example of a turning point (or event) is a disabling illness or
accident and its eventual consequences in the form of a deep and global
reorientation of life. Events occurring in a ‘‘coherent’’ way within the overall
orientation of a trajectory can be opposed to turning points. In sociology
and social psychology, disturbing events are unexpected and often con-
sidered as ‘‘stressful’’, at least potentially. These events can be directly re-
lated to persons (e.g., illness) or to external events (e.g., death of a
significant relative, parents’ divorce, collective dismissal). In this last case,
the interest lies in the amount of stress caused by the event depending on its
social context and meaning.
In the present context, a special mention should be made of the research field
of life event research (Dohrenwend & Dohrenwend, 1982; Brown & Harris,
1989), an area of study that rests on the basic hypothesis that unforeseen events
in the life course can alter its further progress and have more or less important
consequences on the persons who experience them. It is multidisciplinary since
at least social psychiatry, social psychology and sociology are being mobilized
by life-event researchers, and it represents one extreme, ‘‘empiricist’’ option of
interdisciplinarity in that it is often practiced with particularly little theoretical
preconceptions, at least concerning the nature of significant events and of the
mechanisms that relate them to their possible effects on the life course. Even
though it seems to be rarely integrated by more outspoken life course or life-
span-oriented research, some results of this line of research confirm basic tenets
of life course research precisely because of its atheoretical conception, such as
the finding that the positive or negative value of life events cannot be found in
themselves, but depend strongly on their interpretation by the concerned actors
and their environment, that this evaluation depends on the coping resources
concerned actors can mobilize, and that the abruptness of their occurrence
may be as consequential as their qualitative noxiousness (which reminds one of
Durkheim’s notion of ‘‘happy crises’’ that may create, according to his
hypothesis, as much anomy as unhappy ones).
Why Look at Life Courses in an Interdisciplinary Perspective? 19
A disturbing or stressful event may act as a turning point. In a rigorous
conceptual perspective, we may even ask whether the notion of events
should be located on the same level as the other three terms, trajectory, stage
and transition. We might also consider events to be potential transitions (of
positive or negative social value) depending on the possibilities of life course
actors to cope with them. Finally, what an event is may vary as a function of
the fine-grainedness we choose for our interrogation: divorce can be con-
sidered as an event, unique and momentary, e.g., in a demographic analysis.
But it can as well be studied as a rather long-term process in which the
formal act of legalized separation is only one relatively late step, and maybe
not even the most consequential one.
TRANSVERSAL SUBSTANTIVE THEMES: CONTENTS
OF THIS VOLUME
A third line of interdisciplinary collaboration can be found in transversal
themes or research dimensions that appear in several disciplines, even
though they are typically approached differently by them. In a way, the
formal aspects treated in the previous section – especially transitions and
events – have also very substantive meanings; they can therefore also be
subsumed under this heading and may be considered as first examples.
Examples of other such dimensions are, in an arbitrary and imperfect order,
gender differentiation, subjective vs. objective aspects of the life course and
their relationships (‘‘subjective’’ including questions of experience, repre-
sentations, identity, projects, cultural models, etc.), agency vs. structure in
and about life courses, linked lives, interinstitutional links forming life
course regimes. We may also think of thematic fields such as gerontology,
which, by definition, is inter- or at least multidisciplinary. Various academic
departments (especially human development and family studies) as well as
centers of gerontology typically count among their member scholars from
different disciplines, such as demography, sociology, psychology, biology,
kinesiology, nurse practice, health policy, communication science, social
work, educational sciences, ecology, counseling, anthropology, philosophy,
theology, etc.16 It is, however, certainly not sufficient to have different dis-
ciplines attached to the same organizational unit for interdisciplinarity to
develop, even if this can become a favorable condition. Direct collaboration
is a much more promising situation, especially in projects explicitly designed
to be interdisciplinary from the outset.
RENE LEVY AND THE PAVIE TEAM20
We have chosen to organize this volume around four such themes: agency
and structure, transitions, biographical reconstruction and methodological
innovations. The first three themes are substantive and take up some of the
transversal lines we have sketched earlier in this essay. The disciplinary
origins of their authors reflect that they are not yet equally treated in the
disciplines we solicited for this volume.
Agency and Structure
Agency and structure has become an important issue in sociological debates
since the 1980s, perhaps less so in other disciplines. The first chapter takes
up this theme, locating it clearly in the substantive area of life course
research.
Settersten and Gannon focus on the joint impact of social structure and
human agency, in proposing a life course model of agency within structure.
They consider how individuals actively create their own lives and maximize
their own development within parameters set by their social worlds (some of
which may constrain them, and some of which may enable them), and how
individuals interact with, and even make proactive attempts to alter, those
worlds. Several examples from three different life periods are provided:
childhood and adolescence, early adulthood through midlife, and old age.
The tension between the theses of standardization or destandardization of
life courses is reconceptualized as not to be resolved by one thesis winning
out over the other, but as being acknowledged to be complementary proc-
esses, loosely associated with agentic and structural dimensions of the life
course.
Marshall tackles the topic of structure and agency first by reviewing the
literature for explicit and implicit definitions of these two crucial terms and
by revisiting, in an autobiographic return, some earlier results of his com-
parative study of aging communities. He comes up with a theoretical model
containing both aspects and recovers Clausen’s concept of planful compe-
tence as one straightforward way to conceptualize life course agency. Like
other contributors, he reminds us of important complexity that must not go
unnoticed in life course research, especially a necessary differentiation of the
concept of identity, and expresses hope for a rejuvenating theoretical input
by European life course researchers.
Luscher looks into a specific and often neglected aspect of social relations,
especially intergenerational relations, tending to rehabilitate ambivalence as
a basic psychological ingredient for sociological analysis. This is a timely
Why Look at Life Courses in an Interdisciplinary Perspective? 21
hint at the fact that the links between the lives life course analysis is bound
to take into account are not just arrows on network charts, but real social
relations with their affective depth and complexity. In fact, the concept of
linked lives is extremely important, but still awaits theoretical and empirical
development, not only concerning the type of people that can lead linked
lives (like successive generations or partners in a couple), but also referring
to the nature of their relationship (besides the emotional valence, their
strength is also a variable that merits more consideration: are linked lives
based only on strong ties or may weak ties also play this role?).
Transitions
While life course research makes a strong argument for considering the
whole life span instead of limiting its interest to specific parts of it (especially
single phases), specific aspects of life courses may be more fertile for re-
search than others. Transitions are particularly revealing of life course dy-
namics on several accounts and are at the center of four contributions.
Although, for the time being, they seem to be principally investigated by
sociologists, they look promising for psychology and social psychology as
well.
Mortimer and her co-authors attempt to disentangle the respective influ-
ences of social structures and individual agency, in focusing on the issue of
educational attainment in the transition to adulthood. Some scholars high-
light cultural values, normative timetables, stratification processes and in-
stitutional career lines as determinants of the contents and pacing of role
changes through the life course. Others, in contrast, emphasize the exercise
of human agency as a central causal force in shaping the life course, in-
cluding the expression of values and identities, self-regulative processes,
decision-making and striving to achieve personal objectives through goal
selection, strategic planning and action. Mortimer and colleagues focus on
the interactions existing between these two dimensions. Using data from the
Youth Development Study, a 15-year panel study of work experience and
the transition to adulthood, they show that early goals and values are linked
to work behavior during high school, which in turn has predictive power
with respect to subsequent trajectories of work, schooling and educational
attainment. The study presents promising ways for opening the ‘‘black box’’
of the processes and mechanisms, both structural and agentic, underlying
life paths of transition to adulthood.
RENE LEVY AND THE PAVIE TEAM22
With an autobiographical twist, Furstenberg takes up the thesis of many
life course researchers that chronological and sequential ‘‘order’’ is impor-
tant, that deviation from this order (like being out of schedule) constitutes a
problem and is bound to provoke undesirable outcomes. Reviewing recent
analyses, he shows that results seeming to confirm this somewhat structuro-
deterministic idea are in fact less than convincing and remain open to
contrary, more optimistic and ‘‘agentic’’ interpretations; according to his
argument, ‘‘life course deviance’’ may as well lead to growth in life course
managing capacities. A major role in determining whether non-normative
trajectories, be they out-of-time or out-of-order, become problematic for
later stages in life seems to be reserved to social and psychological resources
and thus index the structural environment of ‘‘life course passengers’’.
Looking into The Secret of Transitions, Bird and Kruger develop a strong
argument about the danger of substantive blinders brought along by meth-
odological and conceptual reductions, for example, in the case of the study
of life course transitions by help of event-history analysis. They advise us
not to forget the complexity of life courses in the double sense that they are
made up of several parallel and related trajectories and that transitions
belonging to different trajectories may interact without necessarily being
synchronized. Moreover, even in one trajectory, transitions are often multi-
layered and last longer than their treatment as ‘‘events’’ would suggest. On
the basis of empirical examples concerning sex-differentiated life courses,
they propose a timely typology of transitions based on their socio-dynamic
features that can help avoid excessive reduction of social complexity on the
conceptual and interpretive level.
Biographical Reconstruction
Life course and biographical analysis (in the sociological, not the demo-
graphic use of the term as we pointed out before) have long been considered
to be strangers ignoring each other – if not hostile antagonists – especially in
some European research traditions. Heirs of the fundamental debate be-
tween the ‘‘two cultures’’ in the social sciences, life course analysis is often
purported to be positivist, explanatory and quantitative; biographical anal-
ysis constructivist, interpretive and qualitative. While this division still ex-
ists, it has been questioned by more recent pragmatic positions, preferring
mixed methodologies (Tashakkori & Teddlie, 1998) and quali–quanti tri-
angulation (Erzberger & Prein, 1997) to methodological fundamentalism.
Even though this debate has probably raged more in sociology than in most
Why Look at Life Courses in an Interdisciplinary Perspective? 23
other social science disciplines, it is highly relevant for all of them, especially
in the life course area. The contributions concerning biographical recon-
struction are instructive in this respect: the first relates identity changes
directly to life course transitions, the second approaches directly and em-
pirically the crucial theme of biographical memorization and the third is an
example of interpretive work in a discipline that has heretofore privileged
quantitative approaches. Biographical accounts have become a field more
currently shared at least by social psychologists and sociologists.
In a chapter relating this part of the volume to the previous one, Emler
presents in a social psychological perspective how identity shifts take place
during life transitions. A first section is devoted to the presentation of the
classical view of cognitive development (Piaget, Kohlberg) as abrupt qual-
itative changes resulting in a succession of qualitatively different stages.
Then social identity is defined on the basis of self-categorization theory,
taking into account the relationships between social identity and social net-
works. One central issue of this chapter is to understand if identity devel-
opment is a gradual process or if it corresponds to a stage model with
sudden and discrete changes. Based on different examples in the areas of
political identity and changes in personal relationships, it opens new direc-
tions of thought and research in the study of shifts of identity across life
transitions.
Perrig-Chiello and Perrig bring together different important psychology
fields in the study of personality across the life span: autobiographical
memory, well-being and personality traits. Psychologists have developed
different theories on how the individuals construct their autobiographical
memory. This chapter presents empirical evidence showing that personality
traits like neuroticism and extraversion have a consistent relationship with
recollection of autobiographical episodes and well-being. Results also in-
dicate that the impact of personality traits may be more important than
positive or negative life events. These results raise important questions
about how individuals adapt and reconstruct life events across their life
course, and how subjective experiences, and in particular personality traits,
may have objective effects on life trajectories.
McAdams is a personality psychologist who has developed a narrative
approach to identity across the life span – a highly welcome complement to
the personality trait approach. Building on the idea that identity is con-
structed in a continuous process across the life course, he shows that life
stories are developed by individuals following distinct ‘‘story lines’’. For
example, he shows that life stories may be narratively organized as
contamination (good to bad) or redemption (bad to good) sequences.
RENE LEVY AND THE PAVIE TEAM24
The various antecedents and consequences of these types of life stories are
described on the basis of empirical examples. A promising development of
this approach is the possible link between these life stories and the cultural
context in which they are expressed and constructed.
Methodological Innovation
Methods of analysis have a peculiar role in our thematic field, especially so
in quantitative methods.17 They are not very numerous and well-known and
they are based on technical and also conceptual implications that have to be
scrutinized all the more seriously for their implicit assumptions and basic
logics for the interpretation of results as they are practically all imported
from more or less distant fields or disciplines (this is probably the most
multidisciplinary sector in life course research!). Moreover, the danger of
one method dominating the field and the substantive outlook of researchers
merits particular attention (one obvious candidate for becoming a ‘‘he-
gemonic method’’ in quantitative life course research is event history anal-
ysis, with optimal matching a new-coming junior competitor that remains at
the margins of the arena for the time being). Therefore, we find it useful to
contribute to methodological diversification with a set of promising and
innovative contributions presenting methods for exploration and analysis
that have not yet made decisive inroads into this area. The three chapters of
this volume nicely complement each other, largely due to the different, but
sometimes converging and highly complementary, research fields of the
authors. Francesco Billari’s applications are mainly in demography, espe-
cially in the research fields of fertility and transition to adulthood. Michel
Oris, a demographic historian, and Gilbert Ritschard, an econometrician,
have been collaborating for several years on projects requiring both an acute
historical sense for the analyses of demographical issues and advanced an-
alytical tools, which statistically address the theoretical questions of interest.
The work of Jack McArdle, a psychologist, has been focused on age-
sensitive methods for psychological and educational measurement, and lon-
gitudinal data analysis. Together, the four authors provide an exceptional
overview of novel life course methodologies of great potential and invite us
to think about future developments. Each chapter briefly summarizes the
state of the art of its research field before heading into the most recent
advances and promising extensions.
In his contribution, Billari compares the two major approaches of
quantitative methods used in life course research, the event-oriented and the
Why Look at Life Courses in an Interdisciplinary Perspective? 25
holistic approach. The former quantitative approach is largely represented
by the freshly popular analytical techniques of event history (or survival)
analysis and also newer techniques stemming from program evaluation work.
The latter approach is mostly represented by techniques used to analyze any
sequence of events and trajectories over the whole life course, and the prime
among these is optimal matching analysis. Both approaches are used in turns
to answer different but complementary life course research questions. The
event-oriented approach is often used in a causal perspective of analysis, and
the author describes all different kinds of causal factors susceptible to affect
the occurrence of an event. The holistic approach is adopted to analyze
different factors influencing trajectories. Billari then invites the reader to
consider that the two perspectives are couched in the two statistical cultures
of data vs. holistic modeling discussed in the very influential paper of
Breiman (2001). The data modeling culture naturally pairs up with the
event-oriented approach, the algorithmic modeling culture with the holistic
approach. Billari provides several examples to clarify his arguments and to
illustrate how the two traditions and approaches complement each other in
addressing life course phenomena.
The chapter by Ritschard and Oris discusses how the fields of historical
demography and contemporary demography promisingly joined forces, es-
pecially under the life course research paradigm. They discuss three ana-
lytical approaches and illustrate them by relevant examples. First, event
history analyses are again taken up, but this time in relation to very recent
advances allowing to account for shared heterogeneity or frailty in the
presence of multiple groups. The combination of event history analysis with
multilevel modeling will undoubtedly witness much success in several re-
search fields. The second set of statistical techniques presented by the authors
is that of Markov transition models used to analyze state sequences, or in-
dividual trajectories across quantitatively and possibly qualitatively different
phases in the life course. The categorical changes can be dependent of ob-
served or latent covariates and differ with respect to their temporal expres-
sion, or lags. Again, this technique is progressing at a quick rate and the range
of its applications is growing accordingly. The third set of analyses consists in
longitudinal data mining techniques based on an induction tree approach.
While more exploratory in nature than the previous two techniques, induction
trees ideally describe the change trajectories of complex multivariate systems,
in which the elements are ordered by importance and predictability power. As
illustrated by the authors, this newer set of techniques may complement the
other two and add much insight into life course research objects, over and
above what was already learnt with more classical techniques.
RENE LEVY AND THE PAVIE TEAM26
McArdle accompanies the reader in exploring a unique data set of cog-
nitive measures obtained on seven occasions from 1931 to 1998 on the same
individuals. The data are first described and observed with respect to their
presence as well as absence (i.e., incomplete data). Then individual and
group change characteristics are determined and possible predictors of
change are tested. Latent growth curve models, a particular application of
structural equation modeling, are used to this end, which are known in the
educational and biostatistical literature under different names (hierarchical
linear modeling, multilevel modeling, random or mixed effects modeling, etc.).
Group differences in developmental trajectories are then explored, on an a
priori level (via traditional multigroup structural equation modeling) as well
as on an a posteriori level (with more recent latent mixture models). Finally,
McArdle introduces an advanced class of structural equation models based
on previous work in econometrics, biostatistics and behavior genetics to
explore dynamics of change. In other words, we are now invited to examine
the data in a non-static perspective, going from description to explanation of
change phenomena. The advanced latent difference score models introduced
by McArdle represent a very promising set of analytical tools capable of
addressing classical as well as innovative theories of change in several
research fields.
Finally, in their afterthought chapter, the editors bundle together the
novel impulses for fostering interdisciplinary life course research that can be
gleaned from the contributions to this volume and suggest directions for
future development.
NOTES
1. The Pavie Team, as it existed when the interdisciplinary venture that producedthis volume was launched, comprised a founding group of six professors and eightresearchers and research assistants. They all participated in the elaboration of aworking document that was the basis of an international colloquium held in October2003 and that has been largely integrated into this introduction. In order to under-score the team character of this work, we list their names in strict alphabetical order:Jean-Claude Deschamps, Guy Elcheroth, Yannic Forney, Jacques-AntoineGauthier, Paolo Ghisletta, Jean Kellerhals, Christian Lalive d’Epinay, Jean-MarieLe Goff, Rene Levy, Anik de Ribaupierre, Claudine Sauvain-Dugerdil, Dario Spini,Manuel Tettamanti, and Eric Widmer.2. Coming from two – at times – different scientific traditions, the terms life course
and life span share so much meaning at many levels (conceptual and methodological)
Why Look at Life Courses in an Interdisciplinary Perspective? 27
that we can safely focus on the similarities between the two terms and consider themas basically synonymous. Instead of using them interchangeably, we rather settle onthe use of life course for the sake of simplicity (for a discussion of these two terms,along with the less adequate one of life cycle, see Settersten (2003a) who takes thesame terminological stand as we do).3. An entirely parallel argument could be made for psychology and probably
other disciplines of the social sciences.4. One can lead lengthy discussions about the proper use of terms like macro-
scopic or microscopic – what is micro for a sociologist may be very macro for apsychologist, what is micro for a psychologist may be way underneath the systemiclevel for which other social sciences have analytical resources. Instead of doing this,let us just illustrate the notion of encompassing system levels by citing three examplesin ascending order: different psychological systems (like memory, emotion, etc.) –interindividual interactions – relations between social institutions (e.g., in organizingsubsequent or simultaneous stages of the life course).5. This is an uneasy translation of the standard French expression ‘‘rapports
sociaux de sexe’’.6. It goes without saying that a similar argument could be made for all classical
sociological variables supposed to capture some aspect of social differentiation, firstof all ‘‘class’’ or position in social hiearchies, but several of these are socially less‘‘thick’’. We do, of course, not mean to relegate them to a secondary position foranalysis, but rather feel that in current research, there is a danger of too exclusive aconcentration on them.7. The most general paradigm of this sort in sociology is rational choice theory or
‘‘methodological individualism’’. But there are many more specialized areas wherepredominant approaches at least implicitely adopt an individualist stance, for in-stance in the form of the status attainment paradigm in mobility research. In psy-chology, the same problem exists but in a more fundamental sense, considerations of‘‘contextuality’’ being more intrinsically out of theoretical bounds than, say, insociology, social psychology or political science (but see Bronfenbrenner’s ecologicalmodel (Bronfenbrenner, 1979), Baltes’ contextualism (Baltes, Lindenberger, &Staudinger, in press), or even Erikson’s epigenetic model (1950).8. There are some special research fields where interdisciplinarity is already pract-
iced in a more or less regular way, such as gerontology. But even here, collaborationis often more of a coordinated parallelism than real cooperation. Moreover, in thiscase, research is not systematically placed in a full life course framework.9. Some authors suggest that the vocabulary of truly transdisciplinary analysis
should be based outside the boundaries of scientific terminology. This may be usefulin a highly problem-oriented research, but should not be seen as a general rule.Systems theory provides a scientific vocabulary for transdisciplinary conceptual-ization which is, to the least, not less promising than everyday or other non-scientificlanguage.10. The current French expression is ‘‘analyse demographique des biographies’’
and has nothing to do with the qualitative study of biographical reconstruction thatsociologists might have in mind.11. We should note that in some contexts, as here, ‘‘biography’’ is practically
synonymous of ‘‘life course’’ whereas in others, the distinction of these two concepts
RENE LEVY AND THE PAVIE TEAM28
is almost as important as that of sex and gender in gender studies, with ‘‘biography’’meaning the narrated, principally self-accounted history of an individual, and ‘‘lifecourse’’ the factual unfolding of life as it can be registered by an outside observer(e.g., the Research Committee Nr. 38 ‘‘Biography and Society’’ of the InternationalSociological Association is clearly interested in biographical research in the sense ofthis distinction; cf. Rosenthal & Kottig, 2004).12. One can distinguish the interdependences between processes belonging to dif-
ferent spheres in the life of an individual, or between processes which concern anindividual and his significant others, or a process which concerns an individual and atransforming context.13. In most disciplines, the concept of stage, even if conceived as an element of a
continuous process, is not inherently linked to notions of finality and universalityalthough these notions have often been intimately associated with the definition ofstage. This misconception has been strongly criticized, especially by anthropologists.An important example of such a debate about implicit and scientifically unfoundeddirectionality assumptions on a supraindividual level has concerned modernizationtheory.14. For some researchers interested in fertility issues, the reproductive age (career)
might be divided in several phases or stages, each of them characterized by a specificknowledge, a childbearing motivation, and a style of decision making (Forrest,1988): (1) before first sexual intercourse; (2) from first sexual intercourse to marriage;(3) from marriage to first birth; (4) first birth to desired family size and (5) fromfamily completion to mother’s menopause.15. We do not explicitly integrate the specific area of life-event research that has
developed mainly in social medicine and psychiatry, but would like to mention it asan interesting source of empirical findings that strongly underscore the relevance ofspecific and especially of singular events for subsequent periods of life. At the sametime, the diversity and sometimes even contradictory characters of these findingspoint to the necessity of explicitly taking into account conceptual tools of variousdisciplines, particularly with respect to the non-reductible complementarity of ob-jective and subjective features (i.e., their personal meanings) of such events if we areto understand their consequences.16. This is often possible because faculty members hold joint or courtesy ap-
pointments. Hence, they belong to distinct departments and only occasionally col-laborate on common projects. This sort of institutional structure, althoughrepresenting progress, is not the ideal environment for fostering common researchgrounds. The sharing of information is undoubtedly facilitated by centers of thiskind. However, interdisciplinarity, as we claimed before, aspires to deeper linkagesamong the disciplines. This is reflected by the lack of Ph.D. programs in gerontology.To our knowledge, various postgraduate diplomas exist in gerontology, but only fewPh.D. titles; the same seems to hold for life course studies.17. Given their greater openness, qualitative methods do not seem to present the
same potential of substantively orienting research and of blinding it toward aphenomena for which the methods we use have no outright provision. Even though,a similar danger could be related to an exclusive use of qualitative methods, espe-cially if accompanied by a conceptual reduction of the range of research questions tothose qualitative methods can answer.
Why Look at Life Courses in an Interdisciplinary Perspective? 29
REFERENCES
Abbott, A. (1992). From causes to events: Notes on narrative positivism. Sociological Methods
& Research, 20, 428–455.
Baltes, P. B. (1987). Theoretical propositions of life-span developmental psychology: On the
dynamics between growth and decline. Developmental Psychology, 23, 611–626.
Baltes, P. B. (1997). On the incomplete architecture of human ontogeny: Selection, optimiza-
tion, and compensation as foundation of developmental theory. American Psychologist,
52, 366–380.
Baltes, P. B., Lindenberger, U., & Staudinger, U. M. (in press). Life-span theory in develop-
mental psychology. In: R. M. Lerner (Ed.), Handbook of child psychology, Theoretical
models of human development, (Vol. 1, pp. 1029–1143). New York: Wiley.
Baltes, P. B., & Schaie, K. W. (1976). On the plasticity of intelligence in adulthood and old age:
Where Horn and Donaldson fail. American Psychologist, 31(10), 720–725.
Breiman, L. (2001). Statistical modeling: The two cultures (with comments and rejoinder).
Statistical Science, 16, 199–231.
Bronfenbrenner, U. (1979). The ecology of human development. Cambridge: Harvard University
Press.
Brown, G. W., & Harris, T. O. (Eds) (1989). Life events and illness. New York: Guilford.
Cinnirella, M. (1998). Exploring temporal aspects of social identity: The concept of possible
social identities. European Journal of Social Psychology, 28, 227–248.
Courgeau, D., & Lelievre, E. (1989). L’approche biographique en demographie. Revue franc-aise
de sociologie, 31(1), 55–74.
De Bruijn, B. J. (1999). Foundations of demographic theory. Choice, process, context.
Amsterdam: NethurD Publications.
de Coninck, F., & Godard, F. (1989). Les formes temporelles de la causalite. Revue franc-aise de
sociologie, 31(1), 23–53.
de Ribaupierre, A. (Ed.) (1989). Transition mechanisms in child development. Cambridge:
Cambridge University Press.
Deschamps, J.-C., & Beauvois, J.-L. (Eds) (1996). Des attitudes aux attributions – Sur la con-
struction de la realite sociale (Vol. II). Grenoble: Presses universitaires de Grenoble.
Dohrenwend, B. S., & Dohrenwend, B. P. (Eds) (1982). Stressful life events and their contexts.
New York: Neale Watson.
Doise, W., & Mugny, G. (1997). Psychologie sociale & developpement cognitif. Paris: Armand
Colin.
Durkin, K. (1995). Developmental social psychology from infancy to old age. Malden: Blackwell.
Edelstein, W., & Noam, G. (1982). Regulatory structures of the self and ‘‘postformal’’ stages in
adulthood. Human Development, 6, 407–422.
Elder, G. H. (1974). Children of the great depression. Chicago: University of Chicago Press.
Elder, G. H. (Ed.) (1985). Life course dynamics: Trajectories and transitions, 1968–1980. Ithaca:
Cornell University Press.
Elder, G. H. (1995). The life course paradigm: Social change and individual development. In:
P. Moen, G. H. Elder & K. Luscher (Eds), Examining lives in context: Perspectives on the
ecology of human development (pp. 101–139). Washington: APA Press.
Emler, N. (1999). What does principled versus conventional moral reasoning convey to others
about the politics and psychology of the reasoning? European Journal of Social Psy-
chology, 29(4), 455–468.
RENE LEVY AND THE PAVIE TEAM30
Encyclopaedia Universalis Multimedia version 8. (1999). Paris: Encyclopaedia Universalis.
Erikson, E. H. (1950). Childhood and society. New York: Norton.
Erzberger, C., & Prein, G. (1997). Triangulation: Validity and empirically based hypothesis
construction. Quality and Quantity, 31(2), 141–154.
Forrest, J. D. (1988). Contraceptive needs through stages of women’s reproductive lives. Con-
temporary Obstetrics Gynaecology, 32, 12–22.
George, L. K. (1993). Sociological perspectives on life transitions. Annual Review of Sociology,
19, 353–373.
Gilligan, C., & Attanucci, J. (1988). Two moral orientations: Gender differences and similar-
ities. Merril-Palmer Quarterly, 34(3), 223–237.
Gotlib, I. H., & Wheaton, B. (Eds) (1997). Stress and adversity over the life course. Cambridge:
Cambridge University Press.
Hagestad, G. O., & Neugarten, B. L. (1985). Age and the life course. In: E. Shanas (Ed.),
Handbook of aging and the social sciences, (Vol. 2, pp. 35–61). New York: Van Nostrand
Reinhold.
Havighurst, R. H. (1972). Developmental tasks and education (3rd ed.). New York: David
McKay.
Heinz, W. (Ed.) (1992). Institutions and gatekeeping in the life course. Weinheim: Deutscher
Studienverlag.
Hogan, D. P. (1981). Transitions and social change: The early lives of American men. New York:
Academic Press.
Kling, K. C., Ryff, C. D., & Essex, M. J. (1997). Adaptative changes in the self-concept during a
life transition. Personality and Social Psychology Bulletin, 23, 981–990.
Kohli, M. (1985). Die Institutionalisierung des Lebenslaufs. Kolner Zeitschrift fur Soziologie
und Sozialpsychologie, 37, 1–29.
Kruglanski, A. W. (1996). Goals as knowledge structures. In: J. A. Bargh (Ed.), The psychology
of action: Linking cognition and motivation to behavior (pp. 599–618). New York:
Guilford.
Labouvie-Vief, G. (1980). Beyond formal operations: Uses and limits of pure logic in life-span
development. Human Development, 23, 141–161.
Labouvie-Vief, G. (1982). Growth and aging in life span perspective. Human Development,
25(1), 65–79.
Levy, R. (2001). Regard sociologique sur les parcours de vie. In: P. Domininice (Ed.), Regards
pluriels sur l’approche biographique : Entre discipline et indiscipline (pp. 1–20). FAPSE,
UNIGE: Cahiers de la section des sciences de l’education, No 95.
Manting, D. (1996). The changing meaning of cohabitation and marriage. European Socio-
logical Review, 12, 53–65.
Mayer, K. U. (2002). The sociology of the life course and life-span psychology: Diverging or
converging pathways? In: U. M. Staudinger & U. Lindenberger (Eds), Understanding
human development: Lifespan psychology in exchange with other disciplines. Dordrecht:
Kluwer.
Mayer, K. U., & Schoepflin, U. (1989). The state and the life course. Annual Review of
Sociology, 15, 187–209.
McAdams, D. P. (1993). The stories we live by: Personal myths and the making of the self.
New York: Guilford.
McAdams, D. P. (1999). Personal narratives and the life story. In: O. John (Ed.), Handbook of
personality: Theory and research. (2nd ed., pp. 478–500), New York: Guilford.
Why Look at Life Courses in an Interdisciplinary Perspective? 31
Mortimer, J. T., & Shanahan, M. J. (Eds) (2003). Handbook of the life course. New York:
Kluwer/Plenum.
Mounoud, P. (1982). Revolutionary periods in early development. In: T. G. Bever (Ed.), Re-
gression in mental development (pp. 119–131). Hillsdale, NJ: Erlbaum.
Neugarten, B. L., Moore, J. W., & Lowe, J. C. (1965). Age norms, age constraints, and adult
socialization. American Journal of Sociology, 70(6), 710–717.
Pascual-Leone, J. (1983). Growing into human maturity: Toward a metasubjective theory of
adulthood stages. In: O. G. Brim Jr. (Ed.), Life-span development and behaviour, (Vol. 5,
pp. 117–156). New York: Academic Press.
Pressat, R. (1979). Dictionnaire de demographie. Paris: PUF.
Riegel, K. F. (1976). The dialectics of human development. American Psychologist, 31, 689–700.
Rindfuss, R. R., Swicegood, C. G., & Rosenfeld, R. A. (1987). Disorder in the life course: How
common and does it matter? American Sociological Review, 49, 359–372.
Rosenthal, G., & Kottig, M. (2004). Biography and society– Reading list. Internal paper of ISA
RC 38 ‘‘Biography and Society’’.
Settersten, R. A. (1999). Lives in time and place. The problems and promises of developmental
science. Amityville: Baywood.
Settersten, R. A. (2003a). Propositions and controversies in life-course scholarship. In:
R. A. Settersten (Ed.), Invitation to the life course: Toward new understandings of later life
(pp. 15–45). Amityville: Baywood (Chapter 1).
Settersten, R. A. (2003b). Rethinking social policy: Lessons of a life-course perspective. In:
R. A. Settersten (Ed.), Invitation to the life course: Toward new understandings of later life
(pp. 191–222). Amityville: Baywood (Chapter 7).
Settersten, R. A., & Mayer, K. U. (1997). The measurement of age, age structuring, and the life
course. Annual Review of Sociology, 23, 233–261.
Staudinger, U. M., & Lindenberger, U. (Eds) (2003). Understanding human development. Di-
alogues with lifespan psychology. Boston: Kluwer.
Tashakkori, A., & Teddlie, C. (1998). Mixed methodology. Combining qualitative and quanti-
tative approaches. Thousand Oaks: Sage.
Thom, R. (1975). Structural stability and morphogenesis: An outline of a general theory of
models. Reading, MA: Benjamin-Cummings Publishing.
Tostain, M. (1999). Psychologie, morale et culture. L’evolution de la morale de l’enfant a l’age
adulte. Grenoble: Presses Universitaires de Grenoble.
Turiel, E. (1983). The development of social knowledge: Morality and convention. Cambridge:
Cambridge University Press.
van der Maas, H. L., & Molenaar, P. C. (1992). Stagewise cognitive development: An appli-
cation of catastrophe theory. Psychological Review, 99, 395–417.
Viaud, J. (1999). Principes organisateurs et representations sociales de l’economie: genese et
dynamique. Revue Internationale de Psychologie Sociale, 12(2), 79–105.
Wohlwill, J. F. (1970). The age variable in psychological research. Psychological Review, 77,
49–64.
RENE LEVY AND THE PAVIE TEAM32
STRUCTURE, AGENCY, AND
THE SPACE BETWEEN:
ON THE CHALLENGES AND
CONTRADICTIONS OF A BLENDED
VIEW OF THE LIFE COURSE
Richard A. Settersten, Jr. and Lynn Gannon
The field of life course studies has at its core two propositions for which
there is an inherent tension: one emphasizing that the life course is the
product of social forces (broadly construed as ‘‘social structure’’), and the
other emphasizing individual capacities and effort (broadly construed as
‘‘human agency’’). A wide array of perspectives on structure and agency can
be found in the literature. At one extreme are models of structure without
agency. More common in the discipline of sociology and in European
scholarship, these models take the life course to be largely constrained, if not
determined, by the characteristics of and processes in social settings, and by
the locations of individuals within those settings. Politically, these models
can be viewed as problematic, at least if they are carried too far, because
they ‘‘externalize’’ blame and leave little or no room for personal respon-
sibility.
At the other extreme are models of agency without structure. More com-
mon in the discipline of psychology and in North American scholarship,
Towards an Interdisciplinary Perspective on the Life Course
Advances in Life Course Research, Volume 10, 35–55
Copyright r 2005 by Elsevier Ltd.
All rights of reproduction in any form reserved
ISSN: 1040-2608/doi:10.1016/S1040-2608(05)10001-X
35
these models take the life course to be largely fostered, if not determined by,
individual decisions and actions. Life, as the adage goes, is largely what one
makes of it. While few of these latter models take the life course to be
completely devoid of social obstacles and barriers, these models often
downplay the effects of social forces and assume that good planning and
hard work go a long way in overcoming barriers. In the political sphere,
these models can be dangerous, at least if they are carried to an extreme,
because they blame people for negative outcomes and suggest that individ-
uals need not be extended support from the state because their problems and
circumstances are their own doing.
In both cases, these models have led to the question ‘‘What matters more,
structure or agency?’’ and to discussions of agency versus structure. While
these two prevalent perspectives offer different and important lenses for
understanding human development, a third and underdeveloped set of
models instead seem necessary to advance interdisciplinary life course re-
search. These are blended models of agency within structure, which explicitly
seek to understand how individuals set goals, take action, and create mean-
ings within – and often despite – the parameters of social settings, and even
how individuals may change those parameters through their own actions.
These models bridge over-structured and under-structured views of the life
course by asking how the characteristics of and processes in social settings
interact with the characteristics, capacities, and actions of individuals to
jointly affect life trajectories and outcomes. They involve incorporating in-
teractions with a wide range of social settings both near to and far from
individuals – from proximal settings of everyday life such as families, peer
groups and friendships, neighborhoods, schools, or work organizations, to
more distal settings such as the labor market, the state and its policies, and
historical events and periods of social change.
Modes of agency within structure bring significant challenges because
they demand that boundaries between disciplines be crossed, especially be-
tween life-span psychology and life course sociology; that the concepts and
measures of ‘‘structure’’ and ‘‘agency’’ be clarified; and that more sophis-
ticated theories and methods be developed to frame and analyze them.
Models of agency within structure also demand a critical evaluation of the
unique nature and effects of structure–agency dynamics within and across
distinct life periods. The tensions between structure and agency need not be
resolved as much as capitalized upon to build new social theories and re-
search on specific life periods and on the life course as a whole.
Finally, there is mounting and conflicting evidence that the life course has
become both more standardized (with regularity in life course patterns being
RICHARD A. SETTERSTEN, JR. AND LYNN GANNON36
driven by the increased ‘‘institutionalization’’ through norms, laws, and
social policies) and de-standardized (or ‘‘individualized,’’ with variability in
life course experiences being driven by greater freedom to ‘‘live a life of one’s
own’’). The chapter ends with some thoughts on how these contradictions
may be explained by more explicitly tending to structure–agency dynamics.
STRUCTURE, AGENCY, AND THE SPACE BETWEEN
Structure
The concept of social structure is difficult to grasp and make precise. The
things indexed by the notion of social structure are recognized as the central
sociological contribution to the study of lives, yet there is little agreement
about what exactly structure is (see also Alwin, 1995). Indeed, sociologists
not only find it difficult to define structure adequately, but they often cannot
do so without using the word ‘‘structure’’ or a variant of it (see also Sewell,
1992). Conventional approaches treat structure as a powerful set of stable
top-down forces that impinge upon individuals and cannot be (easily) al-
tered. As an example, consider Alwin’s (1995, p. 218) definition of social
structure as a set of ‘‘opportunities and constraints within networks of roles,
relationships, and communication patterns, which are relatively patterned
and persisting (emphasis added). These opportunities and constraints may,
at one extreme, refer to ‘‘large, organic institutional structures, such as
bureaucracies, which structure and orient human activities,’’ or they may, at
the other extreme, refer to a set of ‘‘dyadic norms negotiated between two
individuals for the purposes of social exchange.’’ This definition of social
structure is typical in that it emphasizes stability, but it is unusual in that it
acknowledges both constraints and opportunities, and both macro- and
micro forces.
The emphasis on stability, however, has led to the neglect of dynamic
aspects of structure. Few approaches have emphasized the fact that people
have the ability to change structures (that is, that the relationship between
people and structures is reciprocal) and that both social structures and
human lives and the connection between them, are dynamic (an important
exception to this trend has been the age stratification framework long ad-
vocated by Riley and colleagues; e.g., Riley & Riley, 1999). On the surface,
the notion of ‘‘dynamic’’ structures seems to negate the very concept of
structure, which has stability at its core. But in a fast-paced world of rapid
social change, it may be transformations in social structure, and the po-
Structure, Agency, and the Space Between 37
tentially disruptive effects they have on human lives, that pose new chal-
lenges for growing up and growing older today.
Sociologists of ‘‘social stratification’’ have emphasized the fact that the
social structure of a society can be described through key dimensions of
social organization: age (or cohort), race (or ethnicity), sex (or gender), and
social class (education, occupation, income, or some combination thereof).
Some of these are ‘‘ascribed’’ statuses into which individuals are born or
over which individuals have little control. Others are ‘‘achieved’’ statuses
that are largely the result of performance, effort, or things over which in-
dividuals presumably have some control. Much social thought related to
stratification has often been tied to the question of who gets what in society.
In most complex societies, certain individuals and groups hold dispropor-
tionate shares of social resources (e.g., property, power, prestige), and so-
ciologists have traditionally been interested in how and why the distribution
of resources varies as a function of the dimensions noted above.
Similarly, both life course sociology and life-span psychology continue to
be dominated by what Bronfenbrenner (1988) once called ‘‘personal at-
tribute,’’ ‘‘social address,’’ and ‘‘social niche’’ models. Personal attribute
models group and compare individuals based on biological or physical fea-
tures (e.g., age, sex, body type). Social address models group and compare
individuals by geographical or social group (e.g., urban or rural, social class,
race or ethnic group). While social address models focus on geographical or
social dimensions, they nonetheless often rely on personal attributes as a
means for grouping individuals. Social niche models group and compare
individuals based on intersections between multiple statuses (e.g., poor,
young, unmarried mothers versus other groups).
All of these models can be described as class-theoretical models because
they assume that the characteristics of individuals serve as important proxies
for ‘‘social structure’’ and individuals’ experiences in it, that they index some
aspect of inequality, and that the phenomena under study are somehow
explained by the categories themselves. In reality, however, these variables
provide little or no direct information about the characteristics of or proc-
esses in the social worlds that individuals inhabit, nor about the experiences
of individuals in those worlds. When investigators find significant differ-
ences between classes, they are then faced with the challenge of having to
explain these differences – and it is with respect to explanation that class-
theoretical models are not sufficient. Because these models are based on the
characteristics of individuals, they do not tap what most sociologists and
ecologically minded psychologists think about as social structure. A big and
problematic leap of faith is required when we assume that individuals who
RICHARD A. SETTERSTEN, JR. AND LYNN GANNON38
share a set of personal characteristics also share similar social worlds and
experiences. At best, such a leap leads to suspect interpretations; at worst, it
results in misleading and false ones. This is especially true when class-
theoretical approaches are used to explore what Bronfenbrenner calls field-
theoretical questions – that is, when we want to understand the ecology of
human development.
Because ‘‘social structure’’ is such an elusive concept, it is not surprising
that its measurement and modeling are so difficult. Important strides must
be made in conceptualizing social settings, measuring their characteristics
and the processes that occur in them, and analyzing additive and multipli-
cative effects across settings. This includes attention to the proximal and
distal settings noted earlier. Of course, the more distal the environment, the
more difficult it is to describe, and the harder it is to trace the processes and
mechanisms through which it affects specific developmental outcomes. Yet
even the measurement of seemingly simple personal characteristics on which
investigators have long relied can quickly become complicated. Indeed, the
measurement of race and ethnicity, socioeconomic status, sex and gender,
and age and cohort are all highly controversial practices in the social sci-
ences (see Oakes & Rossi, 2003; Passel, 2001; Settersten, 1999).
Agency
Matters of agency are as complicated as matters of structure. While the
‘‘structure–agency’’ debate has long been central to the discipline of soci-
ology, the ‘‘structure’’ side has been emphasized more than the ‘‘agency’’
side – except perhaps at the boundary between sociological social psychol-
ogy and psychological social psychology. In psychology, in contrast, the
bundle of concepts associated with the notion of agency is recognized as one
of its central contributions to the study of lives. Despite this, there is little
agreement about what exactly agency is, how it matters theoretically, or how
it should be measured. A wide range of concepts – such as ‘‘self-efficacy,’’
‘‘self-determination,’’ ‘‘locus of control,’’ ‘‘effort,’’ ‘‘mindfulness,’’ ‘‘re-
sourcefulness,’’ ‘‘mastery,’’ and ‘‘autonomy’’ – is often used to index, or
used interchangeably with, ‘‘agency.’’
New depictions of individuals as active and self-aware are especially the
result of a growing ‘‘constructivist’’ view of human development, in which
individuals are seen as the primary architects of their own lives – making
their own decisions, creating their own opportunities, and generating their
own meanings. This reflects a growing recognition of the need to integrate
Structure, Agency, and the Space Between 39
action perspectives in the field of human development, a trend which is also
mirrored in the discipline of sociology (Giele, 2002). These views, for ex-
ample, have transformed old views of children and childhood, in which the
self was treated as something given, and have instead promoted new views
of the self as something created and reflexive (e.g., James, Jenks, & Prout,
1998).
Constructivist views are equally important for understanding adulthood.
Activities of self-regulation and intentional self-development typically be-
come more differentiated and concrete in the transition to adulthood, as
independence and autonomy gain in importance and as personal goals and
‘‘identity projects’’ are expected to become clearer (Brandtstadter, 1998).
Personal goals and identity projects also undergo a process of continual
revision and readjustment throughout adult life, including advanced old age
(e.g., Freund & Smith, 1999). They are not only shaped by the input of
important others, but are dependent on the plans of other intimates (e.g.,
spouse or partner, children) with whom one’s life is interwoven (that is, lives
must be jointly negotiated and coordinated). They are also shaped by a
larger system of cultural norms, as representations of the ‘‘normal’’ or ‘‘ex-
pectable’’ life course are incorporated into and presumably guide personal
plans. Cultural scripts for life in modern societies seem unclear, and whether
these scripts actually affect personal plans seems even less certain (Settersten,
2003). The pervasive focus on personal control and agency in developmental
science corresponds with the growing emphasis on internal motivation,
planning, decision-making, and open and flexible pathways in societies
(Diewald, 2001).
An important agency-based concept in sociology has been Clausen’s
(1993) notion of ‘‘planful competence.’’ Clausen envisioned the life course as
largely the result of personal choice, and positive trajectories as largely the
result of planful competence. Clausen argued that planful competence is
characterized by three things: dependability, intellectual involvement, and
self-confidence. Based on Q-sort methods, Clausen’s measurement of these
components has been the source of some debate. But as a concept, planful
competence is simple and intuitive: It means knowing your strengths, lim-
itations, and interests, and knowing what options are available and how to
take advantage of them. It means being able to assess the actions and feel-
ings of others, and to take these into account when interacting with others.
Most importantly, it means having goals and the self-confidence to carry
them out, coupled with a high degree of flexibility and openness to new
experiences.
RICHARD A. SETTERSTEN, JR. AND LYNN GANNON40
Clausen’s work is important theoretically because it yields interesting in-
sights, especially when combined with recent theorizing on cumulative ad-
vantage and disadvantage over the life course (e.g., Dannefer, 2003).
Clausen suggested that planful competence can be discerned by the ado-
lescent years (but that it develops naturally as adults mature), and that it has
powerful effects on subsequent life. Adolescents who are planfully compe-
tent make good decisions and have successes early in life, the benefits of
which cumulate and lead to further good decisions and successes over the
life course.
If planful competence is a skill that is at least partially learned, interesting
interventions might be designed to help instill it in children and young
adults. This also raises the question of whether it is ever too late to acquire
planful competence, for even individuals in midlife or old age might have
much to gain from it, despite their limited time horizons. Of course, even the
most planfully competent individuals may not be able to act on opportu-
nities if they do not have the social resources to express these competencies
or if other social barriers prevent them from doing so. This is a reminder
that forms of agency are constrained by both barriers in the social world and
personal resources and capacities. Models of agency are, in part, models of
rationality. They take humans to be capable of computing probabilities and
joint distributions, and to have substantial knowledge of their environ-
ments. But these models must be ‘‘faithful to the actual cognitive capacities
of human beings’’ – to real limits in ‘‘knowledge, attention, memory, and
other resources’’ (Gigerenzer, 2003, p. 425).
Attention to agency is important because the life course is to some extent
a personal construction. But the life course also entails selective social
processes that sift and sort people into and out of various settings, and open
or close opportunities depending on the characteristics of people and the
contexts that surround them. These settings and opportunities are not en-
tirely their own doing but are systematically allocated by social forces re-
lated to race, sex, age, socio-economic status, and other factors. It is for this
reason that joint models of agency and structure, to which we now turn,
represent such important ventures for future scholarship.
Agency within Structure
Advances in life course research require greater attention to the joint impact
of social structure and human agency. Models of agency within structure
consider how individuals actively create their own lives and maximize their
Structure, Agency, and the Space Between 41
own development within parameters set by their social worlds (some of
which may constrain them, and some of which may enable them), and how
individuals interact with and make proactive attempts to alter those worlds.
These models require a stronger partnership between life-span psychology
and lifecourse sociology (see Diewald, 2001; Mayer, 2003; Settersten,
2005a). Sociologists, for example, have often lost sight of the person and
overlooked the roles of personality traits and characteristics, motivations,
and action as life course determinants. From the vantage point of sociology,
however, psychological factors are often viewed as being outside its realm,
given that the purpose of sociology is often understood to be the explana-
tion of the social by the social, a tradition dating back to Durkheim (1895/
1964). Or, if psychological factors are viewed as relevant, they are not taken
to have independent causal significance if they are partially created through
social processes.
Psychologists, by contrast, have often neglected the powerful social and
historical forces that promote or limit development. Given the emphasis on
matters of agency in psychology, a working assumption would seem to be
that individuals are capable of altering structures, or at least are reciprocally
affected by them. Yet modern psychology has focused on family and
interpersonal relationships to the exclusion of more distal settings. In
addition, perceptions, beliefs, and goals are often viewed as separate and
independent from structural opportunities and constraints, which they are
not, for social forces may already be present in psychological states, as noted
above.
The life course is the result of social institutions, culture, and history and
the result of decision-making, action, control, and personality (Diewald,
2001). New models of the life course must capture a more complete range of
these factors. The latter factors, however, traditionally the domain of psy-
chology, will become increasingly important as the strength of institutions
and norms weakens, and as individuals have greater latitude to select or
develop their own life scripts. This freedom results in both new possibilities
and risks, as the experimental nature of ‘‘do-it-yourself’’ biographies makes
them more fragile (Beck, 2000). When individuals choose or find themselves
on pathways that are not widely shared by others or reinforced in institu-
tions or policies, they may lose important sources of support and struggle
with institutional barriers built on more dominant models.
Research in life course sociology and life-span psychology is more com-
patible than it seems at the first sight. Indeed, the more fine-grained work of
life-span psychology, which seeks to unearth the quality and quantity of
‘‘intra-individual plasticity,’’ can offer an empirical base to reinforce
RICHARD A. SETTERSTEN, JR. AND LYNN GANNON42
sociological research by decomposing macro patterns into their more prox-
imal and micro-genetic causes (Settersten, 2005a). Together, the two fields
offer tremendous potentials for moving bodies of knowledge from one dis-
cipline to another, and for providing mutually supportive explanations from
different levels of aggregation and with different views on causality. The
concepts of structure and agency, and the dynamics between them, can be
more actively combined in the design and execution of collaborative inter-
disciplinary work (see also Sibeon, 1999).
Several models offer excellent starting points for improving knowledge of
agency–structure dynamics, especially if they are developed in more contex-
tual ways. For example, the ‘‘developmental regulation’’ models of Heck-
hausen (2003) give extensive attention to the selective and compensatory
control strategies of individuals, but relatively limited attention, at least em-
pirically, to larger socio-structural forces that constrain or enable develop-
mental potential. Great possibilities also exist in extending the models of
‘‘selective optimization and compensation’’ of Baltes and colleagues (for ap-
plications to intellectual functioning, see Baltes, Staudinger, & Lindenberger,
1999), and those of ‘‘assimilative’’ and ‘‘accommodative’’ coping developed
by Brandtstadter (e.g., Brandtstadter & Rothermund, 2003), to better incor-
porate socio-structural forces.
Modes of agency within structure cannot be general, but must be under-
stood within particular domains of functioning and particular environ-
ments. They must also consider agency not only as an individual
phenomenon, but as a collective one – as when many individuals who dare
to make innovative life decisions end up creating new options for others, or
when entire groups of people acting in concert attempt to forge new patterns
through social change and collective movements. These models must also
explore how the nature and balance of agency and structure may change
within and between individuals as they move through different periods of
life. This includes a need to assess individuals’ subjective understandings of
their own capacities and resources, as well as those that exist in the settings
around them, for these understandings affect how life is interpreted and
projected forward. It is important to note, however, that life periods them-
selves are important elements of social structure. They are a constant part of
societies (though their boundaries and content may change); they are widely
recognized and have shared meanings; they are reflected and reinforced in
law, policies, and institutions; and they are given practical forms in everyday
life. We now briefly explore concerns related to agency and structure in
three periods: childhood and adolescence; early adulthood through midlife;
and old age.
Structure, Agency, and the Space Between 43
STRUCTURE, AGENCY, AND
SPECIFIC LIFE PERIODS
Childhood and Adolescence
Four immediate social settings have been central to understanding the de-
velopment of children and adolescents: families, schools, neighborhoods,
and peer groups and friends (for a review, see Cook, Herman, Phillips, &
Settersten, 2002). This literature also contains debates about the relative
importance of these four settings on child development, the changing
strength and nature of their influence as children become adolescents (es-
pecially what the growing influence of peers and friends, and the emergence
of romantic relationships and sexual awakening, mean for the other set-
tings), and their additive and multiplicative effects. Given the significance of
these four settings and the natural links between them, research on this
period of life has examined person-context interactions more than any other
period. Family characteristics and processes are probably best articulated,
followed by schools, neighborhoods, and peer groups.
Much of the focus has been on the negative forms of these settings – the
problems that ‘‘bad’’ neighborhoods, schools, peer groups, or families pose
for development, and how these settings may be improved to promote pos-
itive outcomes (or at least minimize negative ones). The interesting question
becomes how and why some kids in bad environments manage to do well,
despite these disadvantages, while many others do not (e.g., Furstenberg,
Cook, Eccles, Elder, & Sameroff, 1999). There has been much less attention
to the equally interesting flip-side of these dynamics: how and why some
kids in good environments do not do well, despite their advantages.
Given the emphasis on settings of poor quality, views of ‘‘structure’’ often
carry negative connotations, while those of ‘‘agency’’ carry positive ones.
Here, the agency of children and adolescents, often in combination with that
of adults, is seen as being critical to overcoming the challenges of these
environments – and the lack of agency, in contrast, is seen as devastating.
For example, it is popular belief that success in school, ensured through
hard work and effort, is the key to a better life. In reality, of course, success
in school is more complicated than this. Hard work and effort alone are
often not enough to guarantee success, especially in poor schools and if
coupled with other settings of poor quality. Nevertheless, a core component
of agency in any of these settings relates to how well children are able to
activate whatever resources exist in their environments. This includes build-
ing positive relationships with adults in those settings who have the ability to
RICHARD A. SETTERSTEN, JR. AND LYNN GANNON44
take action on behalf the child, or who can protect or promote a child who
might be at risk in some way.
The renegotiation of family relationships as children become adolescents
and young adults is also an important aspect of agency during this period.
Indeed, typical ‘‘acting out’’ in the family or other settings may even be
considered an exertion of agency against structure. Many adolescents also
begin to develop attachments to the world of work through part-
employment and volunteer activities, and become politically and civically
aware. These, too, offer important opportunities to actively develop and
express one’s self.
Of course, agency can also be self-destructive. For example, when parent-
ing strategies are too permissive, when neighborhoods have too little social
cohesion, or when schools are disorganized, children may have too much
autonomy – which can lead to negative outcomes. There is therefore a tricky
balance between structure and agency in promoting positive outcomes for
children and adolescents. The literature suggests the strong influence of
structural forces during this period, expressed through these four settings,
which are created and controlled by adults.
The presence and power of these settings takes place against, and must be
responsive to, the child’s growing need for independence and autonomy.
Much literature, especially in North America, leaves one with the impres-
sion that children and adolescents, as dependents, are not yet capable of
making their own decisions. The emerging field of childhood studies in
Europe is altering this view. This field has been heavily influenced by the
United Nations (1989) Convention on the Rights of the Child. It takes
seriously the notion that children are competent social actors who can – and
should be granted opportunities to – participate in decisions that affect
them, define the directions and processes of their own development, and
participate in the social world (e.g., James, Jenks, & Prout, 1998). This field
has produced innovative studies of how children inhabit and negotiate social
settings, including a range of both public and private spaces with varying
levels of adult control and supervision, and with varying levels of direct
involvement of children in designing and creating those spaces (see, for
example, Philo, 2000).
Early Adulthood through Midlife
Dramatic changes in transitions to adulthood have given rise to a host of
questions about whether current generations of young people are more
Structure, Agency, and the Space Between 45
dependent on parents, less interested in growing up, and more wary of
making commitments and taking action for themselves and with others. As
movement into adulthood has become more prolonged and complex, par-
ents are now providing extensive support to children well into their 30 s
(Settersten, Furstenberg, & Rumbaut, 2005). The types and levels of in-
vestments needed to launch even well-positioned young people only exac-
erbate the needs of vulnerable populations who have little or no family
support to back them (Foster & Gifford, 2005; Schoeni & Ross, 2005).
Despite the often extraordinary actions required of families to see their
children into adulthood today, young people nonetheless express an under-
standing that they are responsible for their life directions, and that ulti-
mately their own determination and efforts are necessary to achieve their
goals (Furstenberg, Kennedy, McLoyd, Rumbaut, & Settersten, 2004).
While young people today may reach legal adulthood at the age of 18 or 21,
at least in most Western societies, they are often not adults psychologically,
socially, or economically, until much later. A brand new challenge to un-
derstanding this period is how individuals develop a sense of autonomy
amidst increasingly long periods of dependence on others, without strong or
clear scripts to guide them, and when the institutions are based on models of
early adulthood that no longer reflect the realities of the modern world.
Modes of political expression also emerge as concerns during the early
adult years, especially as legal rights of adulthood and responsibilities of
citizenship are granted (Settersten, 2005b). There is convergent evidence that
young adults who wrestle with social issues and participate in civic matters
are more likely to be engaged citizens throughout life (Flanagan, 2004).
With these shifts come new experiences with collective agency, especially
through youth movements, which often coalesce around challenging the
status quo and the power to ‘‘make history.’’ Unlike children, adults can
more easily mobilize and advocate on their own behalf. In the most extreme
of social movements – revolutions – ‘‘structure’’ is something not to be
reshaped but to be toppled (e.g., Blechler, 2000).
The growing autonomy of youth begins an emphasis on both the ability
and need to set developmental goals, harness one’s resources, and exert
control over the environment. Attention to these processes have propelled
agency-related matters to center stage in literature on adult development, as
they are necessary for the successful performance and management of mul-
tiple adult roles. With longer and more certain lives, experiences are po-
tentially more predictable and controllable, and careful planning may be
increasingly necessary in a world with seemingly unlimited opportunities
and time. As suggested in the previous section, adults not only strive for
RICHARD A. SETTERSTEN, JR. AND LYNN GANNON46
agency in their own lives, but they also hold positions and take actions that
directly affect the agency of others, especially children. Recent research,
however, has stressed the significant need to understand the ecology of
adulthood – how adults shape and are themselves shaped by the social
spaces in which they exist (for illustrations, see Gecas, 2003; Settersten &
Owens, 2002). New theories and research in these directions are necessary to
compensate for the heavy emphasis on individual capacities and resources in
research on adult development and aging.
In midlife, individuals must confront many new challenges, including
changes or stressors related to physical health, memory, personality, emo-
tional development, adaptation and resilience, work and retirement, and
family relationships (for overviews, see Lachman, 2001). These challenges,
like those of old age, bring new concerns about how to maintain forms
or levels of agency in the face of growing personal constraints, whether real
or anticipated. Yet some of this literature has also emphasized the new
potentials of midlife and the chance to reclaim aspects of the self that were
lost or put aside in early adulthood or to develop the self in whole new ways.
Old Age
Like earlier periods of adulthood, scholarship on old age continues to em-
phasize agency through planning, goal-setting, decision-making, even
through the end of life. What is unique to old age is that these matters
are now assumed to be heavily conditioned by losses in physical, cognitive,
psychological, and social capacities, and by increased dependence on others.
The changing configuration of capacities in old age constrains the degree to
which old people have or can express agency. Indeed, maintaining control
over one’s life is understood to be a (if not the) central task of old age, and
fears about losing control preoccupy many old people. Many of the dom-
inant models in life-span psychology noted earlier (e.g., Baltes, Heckhausen,
Brandtstadter, and their colleagues) have guided this literature. In the light
of reduced capacities and a limited time horizon, opportunities for action –
and those selectively chosen – become especially meaningful to old people
(see also Lang & Carstensen, 2002).
The strong emphasis on individual capacities and characteristics in old
age has resulted in a surprisingly acontextual view of this period – what
Hagestad and Dannefer (2001) call the ‘‘microfication’’ of gerontology –
with the exceptions of the vast bodies of literature on health care environ-
ments, transitions to assisted living or nursing homes, giving and receiving
Structure, Agency, and the Space Between 47
care, and old age policies (for overviews, see Binstock & George, 2001). In
these cases, agency is a significant concern as old people interface with the
family, institutions, and the state, and as they jointly walk the blurry and
shifting line between dependence and independence. End-of-life issues also
result in a widespread interest in spirituality and religion in old age. In some
cases, this may signal the relinquishment of one’s own control over life and
the placement of control in another entity, or it may signal a shift away from
the self and toward generative actions and ideas meant to improve the lives
of others or humanity.
Gerontology has in the last few decades seriously challenged the belief
that old age is a bleak dark period of great losses. Apart from the areas
noted above, gerontologists seem committed to promoting positive images
of old people and combating negative stereotypes about aging and old age,
especially in promoting models of ‘‘successful’’ aging (e.g., Rowe & Kahn,
1998). These models have helped produce portraits of active elders enrolled
in school, involved in neighborhoods, productive at work, engaged in pol-
itics, and enjoying leisure. All of these spheres offer important opportunities
for expressing agency, especially for those who have health and wealth. The
lengthening period of old age, like that of early adulthood, should prompt
interest in how old people forge pathways through the final decades of life
without strong or clear scripts to guide them, how they navigate institutions
that have no or outdated scaffolding to support them, and how they manage
to retain a sense of autonomy against potentially long periods of dependence
on others.
We must also ask whether we do ourselves and old people a disservice
when we downplay the very real hardships encountered in old age. There is
no document parallel to the U.N. Convention on the Rights of the Child,
mentioned earlier, that enforces special rights for old people. But like chil-
dren, many old people around the world do have special concerns related to
independence, participation, care, self-fulfillment, and dignity, all of which
were recently outlined in the United Nations (1999) Principles for Older
Persons as part of the International Year of Older Persons.
Finally, there is growing empirical support for the notion that the ‘‘ar-
chitecture’’ of the life span becomes increasingly ‘‘incomplete,’’ and that the
relative influence of biology and culture changes, over the course of
adulthood, with cultural influences diminishing and biological influences
increasing over time (Baltes, 1997; Li, 2003). The intersections between
socio-cultural and bio-genetic forces in different life periods offer many
possibilities for exploring structure–agency dynamics, especially during
childhood and old age (see Settersten, 2005c). In this section, we have briefly
RICHARD A. SETTERSTEN, JR. AND LYNN GANNON48
considered some of the ways in which the nature and relative balance of
agency and structure may differ across particular life periods. We now turn
to a few observations on the life course as a whole.
STRUCTURE, AGENCY, AND
THE LARGER LIFE COURSE
There is mounting and conflicting evidence that the life course has become
both more standardized (with regularity in life course patterns being driven
by the increased ‘‘institutionalization’’ brought about by norms, laws, and
social policies) and de-standardized (or ‘‘individualized,’’ with variability in
life course experiences being driven by the greater choices and control in-
dividuals have over their lives). Much contemporary life course research
emphasizes the latter, with widespread belief that a wide range of macro-
and micro-level factors in the last few decades have resulted in life courses
that are less conventional, patterned, and predictable, and more risky in
private and public spheres alike (Mayer, 2004). This seems especially true of
American research, but it is also increasingly true of Western European
research.
Forms of individual agency are assumed to be central to the emergence of
pluralistic life courses. Extending Prout (2005), this diversity is ‘‘locally
constructed’’ through repeated interactions between the self and others in
immediate environments. The fragility of everyday life in the modern world
demands that individuals focus constantly on maintaining and repairing
themselves and social relationships. Only rarely do individuals realize that
the contour of their lives and nature of their experiences may be shared by
many others, wrapped up in larger patterns produced by resources and
constraints of settings well beyond their immediate environments – what
Mills (1959) called the ‘‘sociological imagination.’’ Even individual agency,
as the centerpiece of these models, is often glossed over and taken to be an
essential characteristic that requires no explanation.
The standardization thesis, which suggests the opposite, rests especially
on the strength of welfare states and social policies to regulate particular
transitions (e.g., marriage) or the structure and content of life periods. This
thesis has mostly been based on Western European research, though inter-
national findings are highly variable and heavily conditioned by the type of
welfare-state ‘‘regime’’ and its benefits (for an overview, see Esping-
Andersen, 2002; Mayer, 2001). For example, at one extreme are ‘‘Liberal
Structure, Agency, and the Space Between 49
Market States,’’ such as the United States or United Kingdom, which pro-
vide only temporary and limited support under specific circumstances. At
the other extreme are ‘‘Scandinavian Social Democratic Welfare States,’’
which provide high degrees of social protection and support across life. In
between are ‘‘Continental Conservative Welfare States’’ such as Germany,
and ‘‘Southern European Welfare States’’ such as Italy.
These views pay special attention to the significance of the state, as a
distal force, in determining the structure and content of the life course.
Again extending Prout (2005), these views are short-sighted because they
take nations and welfare-state regimes to be ‘‘stable and bounded entities.’’
Descriptions of welfare state regimes are important in that these regimes are
frames for understanding how the life course is organized in particular so-
cieties, and are helpful for explaining cross-national variability in life course
patterns. But they reveal little about the dynamic nature of boundaries
within and between societies, or of exchanges across these boundaries. More
importantly, these regimes ‘‘homogenize’’ forms of the life course within
societies because they describe life course patterns more than they explain
how the patterns are produced or maintained. They assume that large-scale
patterns ‘‘trickle down’’ and explain the action of individual and collective
agents, or the options from among which they must choose, rather than seek
to understand how individual and collective activities ‘‘percolate up’’ to
explain large-scale patterns. Evidence of standardization or institutionali-
zation should also not be interpreted to mean that the decisions and actions
of individuals do not matter. Indeed, one could argue that whatever deci-
sions and actions individuals are able to make or take will become even
more precious under conditions of standardization or institutionalization.
The evidence for the two theses also differs depending on whether the
historical view is narrow or wide, and on the phenomena of interest. For
example, while the timing of many American life course transitions became
more uniform over the course of the twentieth century, especially mid-
century, their sequencing simultaneously became more diverse. This is es-
pecially true of transitions typically associated with entry into adulthood
(Shanahan, 2000). Yet for American women – and contrary to contemporary
discussions of the emergence of ‘‘non-traditional’’ family patterns – the
timing and sequencing of family transitions has been high, since at least the
early decades of the twentieth century (Wu & Li, 2005).
This also serves as a reminder that views of institutionalization must span
and be differentiated across multiple domains. Kruger and Levy’s (2001)
distinctions between three types of ‘‘institutional framing’’ will yield fresh
insights into points of positive and negative synergy between institutions,
RICHARD A. SETTERSTEN, JR. AND LYNN GANNON50
and into the ‘‘not so visible nexus’’ between men and women. Sequential
institutionalization structures particular life periods and prompts movement
from one period to another. The traditional, but dissolving, lock-step
organization of men’s lives – with education early in life, continuous em-
ployment in the middle, and retirement at the end – symbolizes this type of
institutionalization. Simultaneous institutionalization refers to the attach-
ment of individuals to multiple organizational forms within particular pe-
riods. Popular discussions of the difficulties of balancing work and family
demands speak to the complexities of simultaneous institutionalization
during the early and middle adult years. Adjacent institutionalization relates
to the constraints that other institutions bring for managing work and
family life, such as those posed by schools, public administration, trans-
portation services, or businesses. Kruger and Levy also remind us that much
institutionalization of the life course is not intended or direct, but rather
unintended and secondary. These dimensions of institutionalization also
warrant greater attention in life course research.
The tension between standardization and de-standardization need not be
resolved, with one thesis winning out over the other, as much as actively
seized to more creatively theorize the life course. Indeed, the evidence for
each thesis need not be incompatible and can be simultaneously true, de-
pending on the level of analysis and the target domain or phenomenon
under study. Joint attention to these matters – to the tensions between
structure and agency, to evidence for standardization or de-standardization,
and to the possible connections between them – will advance interdiscipli-
nary research. The most obvious connections to be explored are how (a)
structural factors produce standardization and uniformity (at least for sub-
groups exposed to common forces), and (b) forms of agency produce
de-standardization and variability. But consideration should also be given
to the ‘‘off-diagonals’’ – to how (c) forms of agency might produce stand-
ardization and uniformity (such as when cohort differences in attitudes and
values prompt new and widespread decisions about marriage or parenting),
and (d) forms of structure might produce de-standardization and variability
(such as when disorganization in, or poorly coordinated connections be-
tween, educational institutions and the labor market results in disjointed or
incoherent experiences).
Models of agency within structure, described earlier, are central to un-
derstanding how the life course is partly the result of active and free choices,
partly created within a fixed set of possibilities and partly imposed from
outside – all of which come with consequences, some good and some bad,
for individuals, depending on how far their paths stray from others or
Structure, Agency, and the Space Between 51
deviate from those assumed in social institutions and policies. These models
will demand clearer definitions and more precise measurement of ‘‘agency’’
and ‘‘structure,’’ their characteristics and processes, and their sources and
determinants.
A central challenge here is that the lives of individuals and successive
cohorts have changed rapidly, but the assumptions that underlie social
institutions and policies are often based on outdated models of life (see also
Settersten, 2005b). These mismatches may bring serious risks for the func-
tioning of individuals and societies, and there is significant need to re-
architect social institutions and policies so that they better meet the changing
needs and realities of individuals and societies. New commitments must si-
multaneously improve and make more flexible the institutions through which
individuals move (including endorsing or permitting a wider range of paths),
as well as improve and make more flexible the connections between them.
New commitments must also strengthen the skills and resources of individ-
uals so that they can better navigate the life course, for these capacities are
vital to ensuring positive outcomes amidst the rapid social dramatic change
and great uncertainty of the contemporary world.
REFERENCES
Alwin, D. (1995). Taking time seriously: Studying social change, social structure, and human
lives. In: P. Moen, G. H. Elder Jr. & K. Luscher (Eds), Examining lives in context:
Perspectives on the ecology of human development (pp. 211–262). Washington, DC:
American Psychological Association.
Baltes, P. B. (1997). On the incomplete architecture of human ontogeny: Selection, optimiza-
tion, and compensation as foundation of developmental theory. American Psychologist,
52, 366–380.
Baltes, P. B., Staudinger, U. M., & Lindenberger, U. (1999). Life-span psychology: Theory and
application to intellectual functioning. Annual Review of Psychology, 50, 471–507.
Beck, U. (2000). Living your own life in a runaway world: Individualisation, globalisation, and
politics. In: W. Hutton & A. Giddens (Eds), Global capitalism (pp. 164–174). New York:
The New Press.
Binstock, R. H., & George, L. K. (Eds) (2001). Handbook of aging and the social sciences.
San Diego: Academic Press.
Blechler, M. (2000). Structure and agency, intellectual ‘‘nationalism,’’ and method: Tang Tsou’s
contributions to China studies and social science. Modern China, 26(2), 239–247.
Brandtstadter, J. (1998). Action perspectives on human development. In: R. M. Lerner (Ed.),
Handbook of child psychology Vol. 1. Theoretical models of human development (5th ed.,
pp. 807–863). New York: Wiley.
RICHARD A. SETTERSTEN, JR. AND LYNN GANNON52
Brandtstadter, J., & Rothermund, K. (2003). Intentionality and time in human development
and aging: Compensation and goal adjustment in changing developmental contexts. In:
U. M. Staudinger & U. Lindenberger (Eds), Understanding human development: Dia-
logues with lifespan psychology (pp. 105–124). Norwell, MA: Kluwer Academic
Publishers.
Bronfenbrenner, U. (1988). Interacting systems in human development: Research paradigms,
present and future. In: N. Bolger, A. Caspi, G. Downey & M. Moorehouse (Eds),
Persons in context: Developmental processes (pp. 25–49). New York: Cambridge
University Press.
Clausen, J. A. (1993). American lives: Looking back at the children of the great depression.
New York: Free Press.
Cook, T. D., Herman, M. R., Phillips, M., & Settersten, R. A., Jr. (2002). Some ways in which
neighborhoods, nuclear families, friendship groups, and schools jointly affect changes in
early adolescent development. Child Development, 73(4), 1283–1309.
Dannefer, D. (2003). Cumulative advantage/disadvantage and the life course: Cross fertilizing
age and social science theory. Journals of Gerontology: Social Sciences, 58B(6),
S327–S337.
Diewald, M. (2001). Unitary social science for causal understanding: Experiences and prospects
of life course research. Canadian Studies in Population, 28(2), 219–248.
Durkheim, E. (1895/1964). The rules of sociological method (8th ed.). New York: Free Press.
Esping-Andersen, G. (2002). Towards the good society, once again? In: G. Esping-Andersen
(with D. Gallie, A. Hemerijck & J. Myles), (Ed.) Why we need a new welfare state.
Oxford: Oxford University Press.
Flanagan, C. A. (2004). Volunteerism, leadership, political socialization, and civic engagement.
In: R. M. Lerner & L. Steinberg (Eds),Handbook of adolescent psychology (pp. 721–746).
New York: Wiley.
Foster, E. M., & Gifford, E. J. (2005). The transition to adulthood for youth leaving public
systems: Challenges to policies and research. In: R. A. Settersten Jr., F. F. Furstenberg
Jr. & R. G. Rumbaut (Eds), On the frontier of adulthood: Theory, research, and public
policy (pp. 501–533). Chicago: University of Chicago Press.
Freund, A., & Smith, J. (1999). Content and function of self-definition in old age and very old
age. Journals of Gerontology: Psychological Sciences, 54B, P55–P67.
Furstenberg, F. F., Jr., Cook, T. D., Eccles, J., Elder, G. H., Jr., & Sameroff, A. (1999).
Managing to make it: Urban families and adolescent success. Chicago: University of
Chicago Press.
Furstenberg, F. F., Jr., Kennedy, S., McLoyd, V., Rumbaut, R., & Settersten, R. A., Jr. (2004).
Growing up is harder to do. Contexts, 3(3), 33–41.
Gecas, V. (2003). Self-agency and the life course. In: J. Mortimer & M. Shanahan (Eds),
Handbook of the life-course (pp. 369–388). New York: Kluwer Academic/Plenum
Publishers.
Giele, J. (2002). Life careers and the theory of action. In: R. A. Settersten Jr. & T. Owens (Eds),
Advances in life-course research: New frontiers in socialization (pp. 65–88). London:
Elsevier Science, Ltd.
Gigerenzer, G. (2003). The adaptive toolbox and lifespan development: Common questions? In:
U. M. Staudinger & U. Lindenberger (Eds), Understanding human development: Dia-
logues with lifespan psychology (pp. 423–436). Norwell, MA: Kluwer Academic
Publishers.
Structure, Agency, and the Space Between 53
Hagestad, G. O., & Dannefer, D. (2001). Concepts and theories of aging: Beyond microfication
in social science approaches. In: R. Binstock & L. George (Eds), Handbook of aging and
the social sciences (5th ed., pp. 3–21). San Diego, CA: Academic Press.
Heckhausen, J. (2003). The future of lifespan developmental psychology: Perspectives from
control theory. In: U. M. Staudinger & U. Lindenberger (Eds), Understanding human
development: Dialogues with lifespan psychology (pp. 383–400). Norwell, MA: Kluwer
Academic Publishers.
James, A., Jenks, C., & Prout, A. (1998). Theorising childhood. Cambridge: Polity Press.
Kruger, H., & Levy, R. (2001). Linking life courses, work, and the family: Theorizing a not so
visible nexus between women and men. Canadian Journal of Sociology, 26(2), 145–166.
Lachman, M. E. (Ed.) (2001). Handbook of midlife development. New York: Wiley.
Lang, F. R., & Carstensen, L. L. (2002). Time counts: Future time perspective, goals, and social
relationships. Psychology and Aging, 17(1), 125–139.
Li, S.-C. (2003). Biocultural orchestration of developmental plasticity across levels: The inter-
play of biology and culture in shaping the mind and behavior across the life span.
Psychological Bulletin, 129(2), 171–194.
Mayer, K. U. (2001). The paradox of global social change and national path dependencies: Life
course patterns in advanced societies. In: A. Woodward & M. Kohli (Eds), Inclusions
and exclusions in European societies. New York: Routledge.
Mayer, K. U. (2003). The sociology of the life course and lifespan psychology: Diverging or
converging pathways? In: U. M. Staudinger & U. Lindenberger (Eds), Understanding
human development: Dialogues with lifespan psychology (pp. 463–481). Norwell, MA:
Kluwer Academic Publishers.
Mayer, K. U. (2004). Whose lives? How history, societies, and institutions define and shape life
courses. Research in Human Development, 1(3), 161–187.
Mills, C. W. (1959). The sociological imagination. New York: Oxford University Press.
Oakes, J. M., & Rossi, P. H. (2003). The measurement of SES in health research: Current
practice and steps toward a new approach. Social Science and Medicine, 56(4), 769–784.
Passel, J. S. (2001). Censuses: Demographic issues. In: N. J. Smelser & P. B. Baltes (Eds),
International encyclopedia of the social and behavioral sciences (pp. 1599–1605). Oxford:
Elsevier.
Philo, C. (2000). The cornerstones of my world: Editorial introduction to special issue on spaces
of childhood. Childhood, 7(3), 243–256.
Prout, A. (2005). The future of childhood. London: Routledge Falmer.
Riley, M. W., & Riley, J. W., Jr. (1999). Sociological research on age: Legacy and challenge.
Ageing and Society, 19(1), 123–132.
Rowe, J., & Kahn, R. (1998). Successful aging. New York: Pantheon.
Schoeni, R., & Ross, K. (2005). Material assistance from families during the transition to
adulthood. In: R. A. Settersten Jr., F. F. Furstenberg Jr. & R. G. Rumbaut (Eds), On the
frontier of adulthood: Theory, research, and public policy (pp. 396–416). Chicago:
University of Chicago Press.
Settersten, R. A., Jr. (1999). Lives in time and place: The problems and promises of developmental
science. Amityville, NY: Baywood.
Settersten, R. A., Jr. (2003). Age structuring and the rhythm of the life course. In: J. Mortimer
& M. Shanahan (Eds), Handbook of the life course (pp. 81–98). New York: Kluwer
Academic/Plenum Publishers.
RICHARD A. SETTERSTEN, JR. AND LYNN GANNON54
Settersten, R. A., Jr. (2005a). Toward a stronger partnership between life-course sociology and
life-span psychology. Research in Human Development, 2(1–2), 25–41.
Settersten, R. A., Jr. (2005b). Social policy and the transition to adulthood: Toward stronger
institutions and individual capacities. In: R. A. Settersten Jr., F. F. Furstenberg Jr. &
R. G. Rumbaut (Eds), On the frontier of adulthood: Theory, research, and public policy
(pp. 534–560). Chicago: University of Chicago Press.
Settersten, R. A., Jr. (2005c). Linking the two ends of life: What gerontology can learn from
childhood studies. Journal of Gerontology, 60(4), S173–S180.
Settersten, R. A., Jr., Furstenberg, F. F., Jr., & Rumbaut, R. G. (Eds) (2005). On the frontier of
adulthood: Theory, research, and public policy. Chicago: University of Chicago Press.
Settersten, R. A., Jr., & Owens, T. (Eds) (2002). New frontiers in socialization. London: Elsevier.
Sewell, W. H., Jr. (1992). A theory of structure: Duality, agency, and transformation. American
Journal of Sociology, 98, 1–29.
Shanahan, M. J. (2000). Pathways to adulthood in changing societies: Variability and mech-
anisms in life course perspective. Annual Review of Sociology, 26, 667–692.
Sibeon, R. (1999). Agency, structure and social chance as cross-disciplinary concepts. Politics,
19(3), 139–144.
United Nations. (1989). Convention on the rights of the child. New York: Author. http://
www.uniceff.org.crc
United Nations. (1999). Principles for older persons. New York: Author. http://www.un.org/sea/
socdev/iyop/iyoppop.htm
Wu, L., & Li, A. (2005). Historical roots of family diversity: Marital and childbearing tra-
jectories of American women. In: R. A. Settersten Jr., F. F. Furstenberg Jr. &
R. G. Rumbaut (Eds), On the frontier of adulthood: Theory, research, and public policy
(pp. 110–149). Chicago: University of Chicago Press.
Structure, Agency, and the Space Between 55
AGENCY, EVENTS, AND
STRUCTURE AT THE END OF THE
LIFE COURSE$
Victor W. Marshall
My purpose is to address a number of related conceptual issues that deal
with the basic, conceptual building blocks of the life course perspective.
These building blocks have been described, over time and from different
perspectives, in different languages. There is the language of the psychology
of human development, with its stages, crises, and developmental transfor-
mations. There are different languages within sociology, describing careers,
status passages, events, transitions, trajectories, and so forth. Many of these
terms appear to be synonyms, different words for the same underlying con-
cept. There is also at least one case in life course scholarship where one word
is used with many different underlying concepts: that is the word ‘agency’,
$This paper draws in part on an earlier paper, ‘‘Agency, Structure, and the Life Course in the
Era of Reflexive Modernization’’ presented at the American Sociological Association meetings,
Washington, DC, August 2000. I have had the pleasure to discuss many of these ideas with my
colleague, Glen Elder at UNC; with Margaret Mueller, our former student at UNC; with two of
my former students and continuing colleagues, Philippa Clarke (now at Duke University) and
Julie McMullin (now at The University of Western Ontario). I am indebted to them for their
insightfulness, intellectual honesty and collegial support. I also thank the students in my grad-
uate seminar in current issues in sociological theory, Department of Sociology, University of
North Carolina, with whom many of these ideas were discussed.
Towards an Interdisciplinary Perspective on the Life Course
Advances in Life Course Research, Volume 10, 57–91
Copyright r 2005 by Elsevier Ltd.
All rights of reproduction in any form reserved
ISSN: 1040-2608/doi:10.1016/S1040-2608(05)10002-1
57
which takes on many meanings in different disciplines, but also within the
same discipline and indeed within the writings of the same author.
I will address three questions with reference to my own early research on
the sociology and social psychology of aging and dying. Thus, the paper re-
interprets work that I did, for the most part prior to the emergence and
formalization of the life course paradigm, in order to address conceptual
and methodological issues of today in life course terminology. The first
question that I shall address is conceptual that asks whether there exists, in a
given society, values for human development and the life course, that is, a
shared view of what is a good life course. I will examine this with reference
to the concept of the ‘good death’. The second, methodological, question
asks how the different concepts (such as trajectory, stage, transition, and
event) are usually operationalized, and with what consequences. I will ex-
amine this with respect to the comparison of ‘objective’ with ‘subjective’
data about the dying trajectory. The third, results, question asks to what
extent changes in status result in the redefinition of identity. I will address
this question by referring to the distinction between self and identity, and to
the action of the aging and dying individual, in managing identity.
I am thus revisiting my early scholarship undertaken before the life course
perspective came to be institutionalized in the social sciences. I am person-
ally curious to see how the life course perspective can shed new light on my
scholarship of 30 years ago; and I hope that this exercise will help to clarify
some of the concepts of the life course that are currently in vogue.
Older individuals face objective changes in their life situation and their
sense of themselves in time. Events and the entire past and anticipated life
course have taken on a new meaning in an interplay between what is in-
creasingly referred to as agency and social structure. I find that terminology
problematic and, before turning to examine these three research questions in
light of my data on aging and dying, I will deal at some length with the
concept of agency and the agency–social structure linkage in sociology.
1. AGENCY
Every social theory, whether implicitly or explicitly, has to deal with two
things – energy and direction. Without energy there would be no action. One
metaphor for this is heating a house. You need to have a furnace to produce
heat. But to understand the heating of a house you also have to account for
direction – how and where that warm air moves from the furnace, through
ducts, to the various levels and rooms of the house (this metaphor rests on a
VICTOR W. MARSHALL58
more North American than European concept of how heating systems
work). Turning to social theory, the two easiest examples would be
psychoanalytic thought and symbolic interactionism. For Freud and his
followers, the id is the furnace, it is the source of energy. But that energy
needs channeling, and the complexities of psychoanalysis provide an expla-
nation for how this energy is channeled into socially acceptable behavior (or
not). In symbolic interactionism, the ‘I’, acting in the dialectic of the self
process, is a necessary hypothetical construct if the perspective is to move
beyond stasis to process. One variety of symbolic interactionism, the Iowa
School, downplays the ‘I’ to focus on the ‘me’ (which is akin to the duct
work in the furnace metaphor in steering behavior). In emphasizing the me,
it loses the sense of process or dynamism that is such an important con-
tribution of the symbolic interactionist perspective.
Agency lies somewhere in the area of energy. The term has increasingly
popped up in the sociology of the life course, but rarely is it defined or
explored in great detail. Even more rarely is agency measured. Is agency just
a new name for something old, such as the concept of action or the concept
of voluntarism? The term agency is used in conflicting ways, but always has
something to do with choice. Dannefer and Uhlenberg decry the way choice
is handled in life course research and theorizing:
In the study of action, choice is a problem to be analyzed, not an accomplishment to be
asserted (Dannefer, 1999). Given the problematic epistemological and ontological status
of ‘‘choice’’ in the wider social science literatureyits remarkably unproblematic ap-
pearance in life course theory cannot be defended. What is almost always measured in
such discussions is behavior, and it is simply presumed that behavior is based on choice
(Dannefer & Uhlenberg, 1999, p. 312).
The fuzzy and conflicting treatment of choice and agency has also been
recognized as a major problem for more general sociological theory. Thus,
Meyer and Jepperson (2000, p. 101) charge that ‘‘ythere is more abstract
metatheory about ‘actors’ and their ‘agency’ than substantive arguments
about the topic.’’ It is a concept more often invoked than measured.
1.1. Agency as a Principle of the Life Course
Glen Elder has been more specific than anyone else in the aging and life
course domain to define this concept. For Elder, agency is one of five
defining principles of the life course. Here is the principle: ‘‘Individuals con-
struct their own life course through the choices and actions they take within
the opportunities and constraints of history and social circumstances’’
Agency, Events, and Structure at the End of the Life Course 59
(Elder & Johnson, 2003). Settersten (2003, p. 39) lists the same as ‘‘an
‘emerging proposition’ of the life course perspective: The life course is a
partial product of human action. It is actively created within the confines of
the social worlds in which individuals exist. Life course scholarship therefore
promotes models of human agency within structure.’’ But this principle has a
long history in developmental psychology and life course sociology. I would
therefore like to review a bit of that history.
1.2. Agency as Production of a Life
Of many arbitrary starting points,1 I will begin with a largely psychological
publication in the human development area. In the Foreword to Lerner and
Busch-Rossnagel’s (1981a) important edited volume, Individuals as Produc-
ers of their Development: A Life-Span Perspective, sociologist Orville Brim
notes that ‘‘The idea that organisms act to create environments to elicit
responses from themselves is not new.’’ However, Brim argues that the 1981
volume is the first to treat the idea broadly and in the context of the theory
of life-span development. While the term, ‘agency’, does not seem to be used
in the volume, the idea of agency is there. As Brim (1981, pp. xv–xvi) puts it:
Behind this idea, to be sure, is the view that the organism is dynamic, powered by
curiosity, growth, expansion, and a drive toward mastery over itself and its world; and
also by the development during the first two years of life of a sense of self as a distinctive
being, and the construction of images of future selves that are different from what one is
now. Behind the idea is also the view that organisms are open to change, are much more
malleable than heretofore thought, and that the consequences of early experience and
biological endowment are transformed by later experience.
The Lerner and Busch-Rossnagel volume lays a broad foundation, from
psychology, evolutionary biology, and anthropology, for a view that indi-
viduals are not only produced by, but also produce their world. This work
suggests that the capacity for organisms to produce their own world varies
both ontogenetically and phylogenetically. In terms of individual human
lives and ontogenetic development, the capacity to act on the world and to
have greater ‘plasticity’ of function increases with human development, and
this capacity can be described in terms of physiological changes in the brain
and, more broadly, the entire organism (Lerner & Busch-Rossnagel,
1981b).2 I will turn to the sociological foundations of this same notion
later, but in the life course perspective, which is explicitly interdisciplinary,
this is an important line of theoretical reasoning related to the concept of
agency, and one that continues to the present (e.g., Diehl, 1999).
VICTOR W. MARSHALL60
Without using the term agency, in his Foreword to the volume Brim raises
the possibility that agency might be considered a variable as contrasted with
simply an aspect of human nature – the ability to choose: ‘‘Throughout this
volume the subject is humanity and the concern is with the species rather
than with individual differences. Certainly there may be individual differ-
ences in degree of mastery of, and re-creation of, the environment because of
individually endowed or acquired differences, but the idea refers to the main
thrust of the human animal, not an occasional remarkable human being’’
(Brim, 1981, p. xvi).
1.3. Agency as Environmental Proactivity or Adaptation
Another line of research in psychology, at times related to the exploration of
Erikson’s developmental framework, either explicitly uses the concept of
agency or uses related terms, which are themselves conceptualized in terms
of agency, and which considers agency to be a variable, and measurable,
property of individuals. Lawton (1989), whose earlier work on person–
environment fit and the ‘environmental docility hypothesis’ viewed the
individual as largely reactive to environmental limitations and pressures,
subsequently introduced the hypothesis of ‘environmental proactivity’, to
recognize ‘‘action and agency – the person’s competence as a determinant of
environment’’ (Lawton, 1989, p. 140).
Building on Lawton’s conceptualization, Kahana and Kahana (1996)
used the term agency in a paper promoting the concept of ‘proactive ad-
aptation’. Here is what they say:
Dependency models of aging have emphasized the propensity of older adults to be
passive respondents to environmental influencesy. However, there has been a small but
growing group of gerontological researchers, who have recognized that older persons
can play significant proactive roles and behave in ways that draw upon and can generate
resources in their environment (Lawton, 1989). This orientation parallels theoretical
developments in the broader field of sociology (Giddens, 1983) that increasingly rec-
ognize the role of agency, reflecting progress, intentionality, and responsibility in the
actions of human beings.
1.4. Agency as Masculine Trait
The term agency has been used by psychologists investigating the concept of
‘ego strength’, and studying the relationship between agency and Eriksonian
‘generativity’. Gutmann (1965) noted that the psychological concept of ‘ego
Agency, Events, and Structure at the End of the Life Course 61
strength’ had decidedly masculine properties. Building on Gutman, Bakan
distinguished between ‘agency’ and ‘communion’, with the former referring
to ‘‘the existence of an organism as an individual’’ and the latter to ‘‘the
participation of an individual in some larger organism of which the indi-
vidual is a part. Agency manifests itself in self-protection, self-assertion, self-
expansiony’’ (Bakan, 1966, p. 15). The life course sociologist Alice Rossi
(1980, p. 9) uses the term ‘affiliation’ instead of ‘communion’, in order to
‘‘avoid any theological association with the concept’’, but retains Bakan’s
concept of agency.
In a more recent paper, psychologists measured agency as both a psy-
chological trait and a characterization of behavior (Ackerman, Zuroff, &
Moskowitz, 2000). Although an independent definition of the construct,
agency, was not given, the conceptualization relates it to ‘masculine’ as
opposed to ‘feminine’ traits and behavior. The self-report measure of agency
is based on the measurement of psychological traits through Likert scales to
assess ‘self-assertive-instrumental traits’ versus ‘interpersonal-expressive
traits’. These generate ‘masculinity’ and ‘femininity’ scores that are con-
sidered to represent agency (masculine) or communion (feminine). An
alternative adjective-rating scale contrasts ‘love’ and ‘status’, the latter
measure a conceptual domain ‘‘from dominance to submission and is
thought to represent agency’’ (Ackerman et al., 2000, p. 30). Behaviorally,
self-described behaviors were coded on four dimensions: dominance and
submission, and agreeableness and quarrelsomeness. Dominance minus
submissiveness is considered to represent agency, while agreeableness minus
quarrelsomeness is considered to represent ‘communion’. Agency, measured
in these ways, is implicated in predicting generativity (as conceptualized by
Erikson), and this relationship, the authors speculate, is because ‘‘creating is
a primarily agentic form of generativity’’, such that ‘‘highly agentic indi-
viduals manifest generative concern focused on creatingy’’ (Ackerman
et al., 2000, p. 37).3 In this psychological formulation, it is clear that agency
is considered to be a variable personality trait of the individual.
The issue of measurement of agency, an issue that is boldly addressed in
the article by Ackerman et al., cannot be dismissed. For example, George
(1999) who is sympathetic to an ‘agentic’ perspective on the self in relation
to society, points out that ‘‘Researchers typically assume that behavior ob-
served in natural settings is triggered by self-perceptions and self-related
motives, but that view remains more assumption or interpretation than
documented fact’’ (p. 47). George is here arguing that the aspect of agency
asserted by many, often qualitative, researchers is nothing but the invoca-
tion of an unmeasured, hypothetical construct. The construct is inferred
VICTOR W. MARSHALL62
from the behavior that it is presumed to cause. If Ackerman’s measurement
approach is problematic (as it seems to me), this may be more in its
execution than its intent.
1.5. Agency as Making Possible ‘Loose Coupling’ (Agency as
‘Unexplained Variance’)
Let us return to Glen Elder. Elder (1985, p. 42) has noted that ‘‘An agentic
concept of individuals in shaping their own trajectory has been a central
principle of the life course framework.’’ Indeed, Elder has made it so. He
notes that conceptions of human agency have characterized ‘life-history’
studies at least since the classic study of the Polish peasant in Europe and
America, by Thomas and Znaniecki (1918), Elder (1997), and Elder and
Johnson (2003). Elder sees agency as important in selection processes:4
‘‘Within the constraints of their world, competent people are planful and
make choices among alternatives that form and can recast their life course’’
(Elder, 1997, pp. 964–965). Agency refers to choices among available
options, and Elder relates this to his concept of ‘loose coupling’: ‘‘Loose
coupling reflects the agency of people even in constrained situations as well
as their accomplishments in rewriting their journeys in the course of aging’’
(Elder, 1997, p. 965). Constraint is accounted for, in Elder’s formulation of
the life course perspective, by ‘‘The principle of historical time and place:
The life course of individuals is embedded in and shaped by the historical
times and places they experience over their lifetime’’ (Elder & Johnson,
2003, p. 62). Elder’s own work has emphasized the impact of the depression
and WW II as structuring choices. Agency can only manifest itself through
choice, and choice is possible only if there are alternatives.
In Elder’s usage, agency makes possible ‘loose coupling’, which in turn is
a ‘principle’ of life course theory that makes room for departures from
structure. In a way, agency functions in this theoretical perspective in the
same way that ‘unexplained variance’ functions in statistical models: if be-
havior is not patterned structurally, then it must reflect resistance to struc-
ture. As Elder and O’Rand (1995) put it, ‘‘Loose coupling reflects the agency
of individuals even in constrained situations as well as their achievements in
rewriting past journeys in the course of aging’’ (p. 456). The relationship of
agency to structure is complex for Elder. Agency is presumably necessary if
individuals are to operate either within or outside the boundaries of social
structure. As Elder and O’Rand (1995, p. 457) argue, ‘‘Age grades and loose
coupling exemplify two sides of the adult life course – its social regulation
Agency, Events, and Structure at the End of the Life Course 63
and the actor’s behavior within conventional boundaries, and even outside
of them’’.
1.6. Agency as Overcoming Resistance
While Elder seems to argue that humans always exercise agency (from which
we might infer agency is a property of human nature), the way in which
Elder describes agency sometimes leaves the implication that it is manifested
only at critical turning points of the life course, or in resistance to the
established social order. I will first deal with the latter: agency as overcom-
ing resistance. Thus, Elder (1997) argues that loose coupling reflects the
antithesis of age grading, and that the agency involved in loose coupling
exemplifies ‘‘the actor’s initiatives and interpretations that press for indi-
viduality and deviations from convention’’ (p. 965). This would imply that
agency is not exemplified if someone follows convention, complies with
social norms or social control pressures, or leads a life in conformity with
social institutions. This is a much narrower view of agency than, for ex-
ample, Giddens’ arguments. The same view is found in Heinz (1996, p. 57),
who notes that while biography is constructed with guidance from ‘‘insti-
tutional standards and the unequal distribution of resources for building
continuous life paths’’, ‘‘Such limitations must not be seen as fateful con-
straints, they can be overcome by human agency.’’5
In setting agency as against social structure, Elder and others adopt a
specific, and questionable, stance about the link between the individual and
social structures. This point has been vigorously made by Dannefer and
Uhlenberg, and can also be pursued through Giddens, Gubrium and others.
Dannefer (1999) links agency in a fundamental way to the processes of
world construction without which there would be no social system. As he
puts it (p. 73), ‘‘yhuman behavior is purposeful; it is not guided by instincts
but by intentions (Weber, 1978).’’ The interaction of intentional actors co-
produces the social system at both micro and macro levels and the self-hood
or social (human) nature of the actors themselves (Dannefer, 1999).6
1.7. Agency as Evidenced in Life Transitions
Moreover, at times in Elder’s writings, agency is indicated not in the smooth
flow or the routines of everyday life, but in life’s more dramatic moments,
moments of transition rather than of continuity, even if the notion of
VICTOR W. MARSHALL64
resistance is not implied. This theme runs through Elder’s work. Thus, ‘‘No
idea better illustrates the contemporary link between social context and the
agency of the individual than the concept of life transition, which defines the
problem as a change in states – social and psychological. Adults bring a
history of life experiences to each transition, interpret the new circumstances
in terms of this legacy, and work out adaptations that can alter their life
course. When transitions disrupt habitual patterns of behavior, they provide
options for new directions in life, a turning point’’ (Elder & O’Rand, 1995,
p. 456). This notion that agency is found only at transition points or, as
Giddens (1991a, p. 113) calls, them, ‘‘fateful moments,’’7 or the related
notion that it is found only in resistance to social barriers, is at odds with the
notion that agency is a property of human nature. Three different notions of
agency are found, I believe, in Elder’s work (capacity, resistance, transition),
leading to the question of whether different terms should be used to more
clearly indicate these three meanings.
1.8. Agency as Responsibility
As the last in this long list of usages of agency, let us turn to the notion of
‘being an agent’ for someone or something. While this is perhaps the most
commonsensical, lay-usage meaning of agency, it is little found in life course
studies. Rather, it is considered to be a sub-theory of rational choice theory
(Kiser, 1999) that draws on Max Weber’s sociology and the ‘new institu-
tionalism’ in economics. According to Kiser (1999, p. 146), ‘‘An agency
relation is one in which a ‘principal’ delegates authority to an ‘agent’ to
perform some service for the principal’’. This notion can, however, be re-
lated to life course theorizing. Meyer and Jepperson (2000) have recently
provided a detailed theoretical examination of the social construction of
agency, in which they argue that the modern actor is a cultural construction
and is ‘‘yan authorized agent for various interests (including those of the
self)’’ (p. 101). For them, agency is ‘‘legitimated representation of some
legitimated principal, which may be an individual, an actual or potential
organization, a nation-state, or abstract principlesythe concept ‘agency’
draws attention to the devolution of external authority, and to the external
legitimation and chartering of activity’’ (Meyer & Jepperson, 2000, note 2).
I can act as an agent for someone else or someone else can act as my agent.
But I can also be constructed as having authority to act for myself, to be my
own agent, to act on behalf of myself. This may be formulated in terms of
the I–me distinction found in G. H. Mead (as the authors acknowledge) and
Agency, Events, and Structure at the End of the Life Course 65
symbolic interactionism (p. 111). In life course parlance, Elder and O’Rand
(1995, p. 465) note that ‘‘ypeople function as agents of their own life course
and development’’. In terms of the sociology of the risk society and its
implications for the life course, as in so much European life course theo-
rizing (see Marshall & Mueller, 2003), a move to the risk society may be
characterized by an increasing delegation of authority for life course man-
agement from society to the individual.
Moreover, Meyer and Jepperson (2000) maintain, in the modern world I
am more and more called upon to assume such agency, in large measure
because of a decline in the attribution of agency to spiritual forces: ‘‘Some
agency is built into modern pictures of the agentic authority and respon-
sibility of the state and other organizations; much devolves to the modern
individual, who is empowered with more and more godlike authority and
vision’’ (p. 105). This notion of agency therefore extends to ‘being respon-
sible for your self’, in which case it points to the reflexivity of the self –
taking responsibility for the self as an object.
In summary, I have reviewed many ways in which the concept of agency,
either directly so named or denoted by a closely related term, has enriched
life course and aging studies. In a very general way, agency has been seen as
the production of a life. The agent is the producer; human development, the
lived life, the narrative, is produced by agency. In more specific formulations
along the same lines, agency has been viewed in terms of environmental
proactivity or adaptation. This recognizes that people not only react but act
and, in acting, produce their biographical selves.
Agency has also, in a few studies by psychologists, been seen as a mas-
culine trait – doing rather than being, taking charge, making things happen.
This is clearly not logically consistent with the previously discussed notions,
which can all be grouped as making an assertion about the nature of being
human. Rather, it postulates that agency is a variable property of humans,
albeit one much more often found among men than women.
In the most widely accepted theoretical statements about the life course, by
Elder and those who have adopted his language, agency has been seen as the
force making possible ‘loose coupling’, thereby somehow accounting for the
fact that all people do not follow standardized, institutionalized life courses.
But I question whether this adds anything to an explanation. Is it not the
same as depicting ‘unexplained variance’ in multivariate models? One can
draw a ‘causal arrow’ to show where variance is unexplained, but the arrow
emanates from something unexplained. If you push that kind of argument
too far, you end up with unexplained variance as the ‘uncaused cause’, which
St. Thomas Aquinas argued was proof for the existence of God.
VICTOR W. MARSHALL66
While the above usages of agency are quite general, some uses are highly
specific, referring to agency only if the individual is overcoming or resisting
social structural barriers, or only if the behavior of the individual is with
respect to a major or important life transition. In such formulations, people
would not be ‘agentic’ most of the time, but only at critical points in their
lives.
And, finally, I noted the usage that suggests agency refers to culturally
legitimated responsibility to act – on behalf of others, of organizations or
ideas, or of one’s own self. This is quite a general conceptualization of
agency, but more restricted than the first ones, which would view the ca-
pacity to act as agency, regardless of its cultural or social legitimation. This
usage of the concept of agency also treats it as a variable, and the authors
I drew on specifically trace a historical shift of agency from spiritual forces
to institutions and to individuals.
2. MAKING SENSE OF AGENCY: TOWARD
TERMINOLOGICAL CLARITY
To summarize, I have noted that the term agency has many meanings in the
sociology and the psychology of aging and the life course. I find it generally
useful to use specific terms for specific meanings. It may be necessary to
distinguish between four different sorts of things: (1) the human capacity to
make a choice, that is, to be intentional; (2) the resources within the indi-
vidual or at the command of the individual that can be brought to bear in
intentional or agentic behavior; (3) behavior of individuals that reflects
intention; and (4) the social and physical structuring of choices. As I turn to
elaborate on these, I note that each of these four constructs has a meta-
phorical place in the analogy of the furnace that I developed earlier.
By the first, I mean that a social theorist may propose that a capacity to
exercise choice is a fundamental aspect of human nature. In terms of the
metaphor, this corresponds to the physical fact of the furnace. Without the
furnace there is no capacity to heat the house. This is not to deny that this
capacity develops in the human over time. Such a capacity would include
awareness and thus requires cognitive capacities to give identity to objects
and events in the world. This can be seen as agency in this first sense and it
can be viewed as a developmental capacity of (virtually) all humans.8
I recommend using the term agency in this way to refer to the human
capacity (as aspect of what it is to be human), to act intentionally, planfully,
Agency, Events, and Structure at the End of the Life Course 67
and reflexively and in a temporal or biographical mode. Except in explicit
developmental work, most life course researchers would not need to treat or
measure this as a variable. Rather, it is an assumption about human nature,
of the order, ‘‘all human beings have free will’’.
By the second construct, the resources that can be brought to bear in
agentic behavior, I refer to both personal capacities of the individual and
resources at his or her command. Metaphorically, this corresponds to the fuel
for the furnace. Some furnaces might have more, or better, fuel than others.
Personal capacities might be intelligence or stage of cognitive development,
learned skills or abilities, knowledge, or physical strength and talent. Re-
sources might be economic (wealth), social (social capital, networks, con-
tractual or informal social ties, and alliances). I do not think it is helpful to
use the term agency to describe this variable property, which is better called
resources or perhaps assets. Resources can be measured, although with care
to include those ‘within’ the individual (e.g., education), and external to the
individual but at his or her ‘command’ (e.g., social capital). Clausen calls
‘‘planful competence’’ the ability to make informed, rational decisions, and
set realistic short- and long-term goals (Clausen, 1991). I would consider
‘planful competence’ to be a resource, rather than a capacity. Clausen (1991,
1993) saw planfulness as indicated by self-confidence, intellectual investment,
and dependability. Shanahan, Elder, and Miech (1997, p. 59), drawing on
Clausen’s concept of planfulness, defined ‘planful competence’ as ‘‘The extent
to which an individual is aware of his or her knowledge, abilities, and in-
terests; pursues goals; and reflects on important decisions’’. They found
‘‘reasonable approximations’’ for these dimensions in the Stanford–Terman
data they used in parents’ assessments of these traits in their children. These
are not basic traits of human nature, but variables in the ability to do the
things that humans do. There is no reason why measurement of these di-
mensions of planful competence, as a resource, could not be further devel-
oped. Presumably, more objective measures would include not only the
awareness of an individual of his or her knowledge, abilities and interests, but
the actual existence of these things regardless of awareness of them. One can
have an ability without knowing it, and this might have resource conse-
quences even if not at the level of awareness.
By the third construct, the actual behavior that reflects intention, I refer
to what has been called ‘action’ or ‘voluntaristic action’ or ‘social action’ by
a host of sociologists who rarely used the term ‘agency’. I think the term
‘action’ is appropriate here, although I have a trained incapacity, as a so-
ciologist, to conceive of action as other than social (see, Campbell, 1996).
This term is itself manifestly used in various contradictory ways, as Colin
VICTOR W. MARSHALL68
Campbell (1996), Derek Layder (1994), and others have shown, but I think
it will be useful if specified in this way. In terms of the furnace metaphor,
this construct would be captured by adjusting or programming the ther-
mostat and opening and closing duct openings and valves.
By the fourth construct, the social and physical structuring of choices,
I refer to the de facto structure of opportunities or life chances that is open
within the range of action of the actor. Metaphorically, the construct
corresponds to the duct work, the degree of insulation, the leaky windows,
and other structural aspects that pose barriers to effective home heating –
regardless of how good the furnace, the fuel, and the thermostat. The con-
cept of ‘life chances’ (Dahrendorf, 1979; Weber, 1978) is useful here, so long
as it is recognized that life chances are not static but emanate from social
processes.
It is useful to emphasize the co-constitution of self and society, through
which action creates social structure just as social structure constrains, or
opens up, possibilities for choice and, thereby, ‘structures’ action. A large
number of life course theorists have enjoined their colleagues to avoid the-
orizing ‘‘agency without structure’’ or ‘‘structure without agency’’ (George,
1999; Ryff, Marshall, & Clarke, 1999; McMullin & Marshall, 1999;
Marshall, 1995, 1996; Settersten, 1999, p. 223). The broader challenge is to
theoretically address the linkages between agency and structure or self and
society (Campbell, 1996; Layder, 1994; Ryff & Marshall, 1999).9 Moreover, if
agency is difficult to define, so too is social structure (Alwin, 1995).10 I turn
next to that conceptual issue.
3. DEFINING SOCIAL STRUCTURE
Despite its centrality to the very core of sociology as a discipline, a plethora
of definitions exists for the term and its conceptualization varies greatly.
I relate a conceptualization of social structure directly to the aforemen-
tioned Weberian concept of life chances, and I draw very heavily on Sewell
for this. If it is assumed that human beings have agency (the capacity for
choice) and that they manifest agency through their actions, then the di-
lemma is to find a way to define social structure in a way that recognizes
agency and action without sacrificing the notion of social constraint. Berger
and Luckmann (1966, p. 18) approach the action-structure problem by
outlining three dialectical processes in which ‘‘subjective meanings become
objective facticities’’. The first of these is ‘externalization’, the process
through which individuals create their social worlds through their action.
Agency, Events, and Structure at the End of the Life Course 69
The second is ‘objectivation’, through which once created, these social con-
structions take on a reality of their own, independent of their constructors.
The third process, ‘internalization’ refers to the process by which the ‘‘ob-
jectivated social world is retrojected into consciousness in the course of
socialization’’ (Berger & Luckmann, 1966, p. 61). As Smelser (1997a, b) has
noted, this concept does not fully capture the ways in which reality con-
strains action. Rather, objectivated reality ‘acts back’ upon individuals
through a number of ‘reality maintenance’ devices, often making use of
language and ritual, and with institutional and physical traces as well (see
McMullin & Marshall, 1999). It is the latter – ‘institutional and physical
traces’ – that moves the conceptualization somewhat beyond Giddens’
(1991b) concept of the duality of action and structure. There is something
more enduring than the instantiation of reproduced social structure through
action, and this ‘something’ can take a number of forms, including legal
arrangements, rules and regulations, and the capacity to mobilize resourc-
es.11 For example, an army is more than the activities of its personnel. It
includes its authority and reporting structure and, not the least, its person-
nel, weapons, and supplies. It is true, as Giddens (1984) argues, that these
other things come to be through human action over time, but they are ‘real’
and constraining of action all the same, and independent of the action that
constitutes them and certainly more than ‘‘memory traces, the organic basis
of human knowledgeability, and as instantiated in action’’ (p. 377). As
Sewell (1992, p. 3) remarks, ‘‘ythe notion of structure does denominate,
however problematically, something very important about social relations:
the tendency of patterns of relations to be reproduced, even when actors
engaging in the relations are unaware of the patterns or do not desire their
reproduction’’.
While Berger and Luckmann are helpful in emphasizing ‘objectivated’
social reality as one moment in the dialectic, I would emphasize an insight
from one of their intellectual sources, Alfred Schutz (1964), which is that
people are not fully free to fashion their own lives because they are born into
a world of predecessors. Archer (1995) suggests that we have to analytically
stretch out the three modes of a dialectic similar to that of Berger and
Luckmann’s in order to recognize that there are objective (material) con-
ditions that already exist prior to the initiation of action.12 For her the
modes are structure, interaction, and structural elaboration. For Sewell,
ideas of social structure are more than the internalized ‘rules’ of Giddens
(and perhaps Berger & Luckmann), but they are socially shared ideas, which
he prefers to call ‘schemas’, and which I would consider to be cultural; and
for Sewell (1992, p. 13), ‘‘Sets of schemas and resources may properly be
VICTOR W. MARSHALL70
said to constitute structures only when they mutually imply and sustain each
other over time.’’ And in this framework, which I endorse, as Sewell states
it, ‘‘Agencyyis the actor’s capacity to reinterpret and mobilize an array of
resources in terms of cultural schemas other than those that initially con-
stituted the array’’ of resources available to them. Agency is then the human
capacity to make something new, to contribute to innovative social
production of the world.13
4. AGENCY AND STRUCTURE IN THE LAST
CHAPTERS OF LIFE
At this point, I will use these constructs to briefly summarize the findings of
my research on the social psychology of aging and dying, and relate the
discussion to the specific research questions mentioned at the beginning.
4.1. My Early Research in Aging and Dying
I will briefly outline my research in the sociology of aging and dying, which
began with my doctoral dissertation work. Georg Simmel has set out a basis
for my research questions early in the last century
ywe are, from birth on, beings that will die. We are this, of course, in different ways.
The manner in which we conceive this nature of ours and its final effect, and in which we
react to this conception, varies greatly. So does the way in which this element of our
existence is interwoven with its other elements (Simmel, 1950).
I set out, in my dissertation, to find some answers to the questions implied in
Simmel’s statement, to wit:
How is it that people die in different ways? How is it that people conceive their nature
and its ‘final effect’? How do people react to their conception of death? In what way is
the individual’s conception of his mortality interwoven with other elements of his life?
This study focuses on the aging, and attempts to assess the implications of the fact that
the aging are also dying. It is, therefore, not a study of death or dying per se, but rather
of living under the growing realization of impending death (Marshall, 1972, p. 2).
In contemporary life course terms, I was interested in the final trajectory
before the transition to dying, and to investigate this trajectory I would have
to identify the transition that led into this trajectory. We are all dying of
course, but my interest was in that period in which the individual became
more highly aware of this fact, and I used the term ‘awareness of finitude’ to
Agency, Events, and Structure at the End of the Life Course 71
describe it. I pursued this research in a field study using participant obser-
vation and semi-structured interviews in a middle-class retirement commu-
nity and a nursing home that had a clientele of a much lower social class
background.14
I was not using terms that predominate in current life course usage.
Leonard D. Cain Jr. (1964) had already published the first major statement
about the life course in an important essay, ‘‘Life course and social struc-
ture’’, in which he aimed ‘‘to identify, isolate, and systematize a life course,
or age status, frame of referencey’’, as well as to ‘‘contribute to the ad-
vancement of a sociology of age status’’ (Cain, 1964, p. 273). But I did not
cite it. He had been influenced by Anselm Strauss, and I did draw on
Strauss, and his concept (with Glaser) of status passage, which is a direct
steal from the career concept of Everett Hughes (1971). But it was only in
developing publications from my dissertation that I moved deeply into using
the career and status passage concepts (Marshall, 1978–1979, 1980).
This was before much of the current life course terminology had been re-
introduced by Elder (1975) and Riley (1979). After more than a decade of
‘chewing on’ my data, I developed a metaphor to describe the main line of
my theorizing and published this in a book, Last Chapters: A Sociology of
Aging and Dying (Marshall, 1980). In this metaphor, I saw aging individuals
as if they were about to author the last few chapters of their lives. With the
inevitability of death approaching, I saw them looking at their own bio-
graphies, and motivated to develop a stance that their lives made sense. (See,
McAdams, this volume, for a general approach to narrative and the quest to
make sense of one’s biography. My metaphor is based on similar psycho-
logical assumptions about motivation, but the elements of the metaphor are
specific to the situation of heightening awareness of finitude.) This meant
making sense of the earlier chapters of the autobiography, but also of the
end of the biography in death itself. Moreover, I saw people in such cir-
cumstances as wanting to assume responsibility for their lives as a whole.
These metaphorical elements stood for the theoretical dimensions of recon-
structing and legitimating one’s past biography, developing a legitimation
for the self as dying, and assuming control, or responsibility for the life as
lived and as ending.
4.2. Is there a Shared Conception of what is a Good Life Course?
I turn now to the first question that I shall address. This conceptual question
asks whether there exist in a given society values for human development
VICTOR W. MARSHALL72
and the life course, that is, a shared view of what is a good life course. I will
examine this with reference to the concept of the ‘good death’. At one level,
this is a question of culture; at another, it is a question of the individual.
I drew on the anthropologist, David Counts (1976) for a cultural conception
of ‘the good death’. The Kaliai of Western New Britain hold a cultural
conception of the ‘good death’ that is quite similar to the process of aging
and dying that is held to be ideal in the disengagement theory of aging, a
North American model (Cumming & Henry, 1961). As in many cultures, the
Kaliai see death not as the end of all experience, but as a transition between
different life states – a transition that takes time and is aided by funerary
rituals. Later life is seen as a progressive severing of ties and putting one’s
house in order. When the Kaliai sense that death is coming they try to ward
it off, to give time for bringing social relationships to a close. In the idealized
good death, the dying person ‘‘called all his kinsmen to gather around him,
disposed of his possessions after repaying the obligations owed by him and
forgiving any obligations of others to him, and then informed those gath-
ered that it was time for him to die’’ (Counts, 1976–1977, p. 370).15 There
are no mortuary rites for a good death – all is appropriate and nothing needs
to be ‘made right’ through ritual intervention. In Kaliai, however, there are
virtually no ‘good death’s, which means that almost everyone has a funeral.
The good death in Kaliai, as the good death in terms of the disengage-
ment theory of aging, is one in which the individual times withdrawal from
society to coincide with actual biological decline and cessation. The indi-
vidual ceases to be a social being at the same time that he or she ceases to be
a living biological being.16 The ‘good death’ has been little studied in west-
ern societies but in the terminology of the life course perspective, a good
death would mean a correspondence between objective and subjective as-
pects of the career (Marshall, 1978–1979) in which, as with other transitions,
rituals may be required in order to bring about this concordance in the
biographical experience of individuals and in the social definition of the
transition by members of that individual’s social group.
It is not just that cultural ideals or schemas of the good death influence
people’s action; their social and collective actions develop such cultural
ideals or schemas. In the retirement village (or congregate living complex for
older people)17 that I studied, the residents, or at least a leadership group
among them, recognized that they were all in the final trajectory of their
lives, one that leads to death. I drew on Berger and Luckmann’s concept of
the legitimating function of symbolic universes for this analysis. As they
noted in their classic book, The Social Construction of Reality (Berger &
Luckmann, 1966, p. 101), ‘‘A strategic legitimating function of symbolic
Agency, Events, and Structure at the End of the Life Course 73
universes for individual biography is the ‘location’ of death. The experience
of the death of others and, subsequently, the anticipation of one’s own death
posit the marginal situation par excellence for the individual’’. In this re-
tirement village (pseudonym Glen Brae), I gathered persuasive evidence that
the residents were highly accepting of the frequency of death in this com-
munity of old people, and also of their own impending death (Marshall,
1975a, b). However, in a nursing home (pseudonym, St. Joseph’s Home)
that I studied at the same time, there was much less evidence of the ac-
ceptance of death (Marshall, 1975b).
In the retirement village, the administration had not made plans for the
management of dying and death as a community event, but by the time the
community had been in existence for only one year, the residents had begun
to organize as a community of the dying. An editorial in the newsletter
published by the ‘residents’ forum’ is worth quoting at length. It noted that
life was taking on a new meaning at Glen Brae,
And new responsibilities are ours too. Fifteen deaths have occurred to date, which was
the predicted actuarial rate. The rate will increase as we grow older. With 100 new
residents arriving next year, it is forecast that we can expect a death amongst us as
frequently as one every two weeks. This is a sober thought.
Our responsibility, therefore, involves a point of view, a determination. Either Glen Brae
will turn into a place shrouded in a funeral parlor atmosphere of tears and perpetual
sadness, or it will play its intended role – the best place to be when crises occur. It is
suggested that each of us look toward the future and be prepared, that we respect the
faith of others, the wishes of the survivor, and above all else that we reduce to a
minimum the prolongation of sorrow, the discussion of pain, loss, tragedy. It is up to us,
not management, to make Glen Brae the haven we desire.
Deaths at Glen Brae are acknowledged by a brief bulletin-board notice and
a name-only listing in the newspaper. Funerals are held off-site. There is an
infirmary at the facility and many resident deaths occur there, thus reducing
the visibility of death. At congregate mealtimes, the hostess sometimes di-
rected sympathetic residents to the tables of recently bereaved residents. In a
number of ways, then, the residents developed a culture internal to their
community, which gave them a set of guidelines and procedures to handle
the frequent occurrence of death among them.
In contrast, at St. Joseph’s Home, where death was an equally frequent
visitor, I found no resident-developed cultural beliefs or organizational
procedures that suggested they had themselves taken control of this final
stage of their life course. This was consistent with a general organizational
contrast between it and the retirement village (Marshall, 1975b). At Glen
Brae there was little administrative intervention into the lives of residents;
VICTOR W. MARSHALL74
St. Joseph’s much more strongly represented a total institution in Goffman’s
(1961) terminology, where the whole round of life was organized by the
administration in ways that limited spontaneity and control by the residents
over their lives. Time was more regulated. For example, residents lined up
outside the dining hall and entered when a bell was rung. At St. Joseph’s the
term ‘dying’ seems to be reserved for the very last stages. As one nursing
aide noted, ‘‘When they start dying, they don’t last for but a day. Some of
them take longer’’. However, the trajectory toward death begins earlier and
is marked by physical movement from a dormitory or private room on an
upper-level floor to an infirmary on the main level. Then, if or when the
prognosis of ‘dying’ is made, the resident is moved to a special room, which
is referred to by staff and residents as ‘the dying room’, but also known to
both as ‘St. Peter’s Room’. In a Catholic institution this clearly signifies the
last stage prior to death and, ironically, this is also Room 13.
Residents are acutely aware of this trajectory. One resident moved to
St. Peter’s room with a ‘dying’ prognosis returned on his own accord to his
old room in the middle of the night. Another resident, not ‘dying’ but placed
in the room when it was vacant because he had been ‘acting out’ and dis-
turbing roommates, also abruptly left the room when he realized where he
was. Almost all of the 15–20 deaths per year were commemorated by fu-
nerals at St. Joseph’s chapel, with an open coffin and at which the priest
deliberately delivered the funeral oration in a booming voice that could be
heard throughout the home. The death of residents is also more visible in the
nursing home than the retirement community because residents eat every
meal at fixed dining hall places, marked by a name card. In my observation,
these name cards, and the empty place, remained for some time following a
resident’s death.
It might be said that people move to the retirement community, but have
organized a collective way to live with the fact that they are a community of
the dying as well as the living; but that the nursing home is a place where
people go to die, where an important aspect of daily life centers around the
fact that people die there, and where death receives considerable ritual
treatment. In some senses, judging from a staff-resident interaction patterns
and the low level of visitation of residents by family members or friends,
many of the nursing home residents were ‘socially dead’, to use Kalish’s
(1966) terminology:
Social death occurs when an individual is thought of as dead and treated as dead,
although he remains medically and legally alive. Any given person may be socially dead
to one individual, to many individuals, or to virtually everyone, and perhaps to himself
as well.
Agency, Events, and Structure at the End of the Life Course 75
Referring back to my earlier treatment of agency as delegation – authorizing
parties to act on behalf of something or someone, including themselves –
agency is not delegated to residents of St. Joseph’s home. Decisions are
made for them more than by them. Kalish elsewhere notes that ‘‘The self-
perceived socially dead individual has accepted the notion that he is ‘as good
as dead’, or that he is, for all practical purposes, dead’’: (Kalish, 1966) Such
characterizations reflect the social construction of the life course, which
creates life course trajectory categories through which some segments of the
aging population will pass. The contrast between these two social settings
shows the differentiation of the life course and the fact that its organiza-
tional properties and biographical experience by individuals can vary
greatly.
In summary on this point, the research question was, Is there a shared
conception of what is a good life course? I approached this very narrowly in
terms of the concept of the good death. There is cultural and subcultural
variability in concepts of how people should end their life course in death.
This can be seen cross-culturally and within the subcultures of two res-
idential facilities for the aged in the same geographical area, but which
differed by social class, religion, and many other characteristics. But the
comparison also says something about agency and social structure. The
social organization and administration of the two communities reflected
extreme positions with respect to structural barriers to social action. The
two community populations differed greatly in both human and social cap-
ital (e.g., in education, socioeconomic status, and viable family and friend-
ship networks), and these differences were reflected in the administrative
stances of the administrations of the two communities. At Glen Brae, social
action by residents was encouraged by administrative principles and prac-
tices; at St. Joseph’s home, it was not. It would have been much more
difficult to organize around the social construction of the last trajectory of
the life course in the latter milieu than in the former. Formulated much
more abstractly and extending from this example to other aspects of the life
course, my answer to the research question is three-fold.
First, let me return to the general conceptual discussion in the first half of
this paper. The issue is framed as one of culture, which I take to mean the
beliefs and values shared by the members of a collectivity. Cultures are
schemas developed by social groups and communities. Together with re-
sources, they constitute social realities, which have their own existence over
and above those of the individuals who came together to construct them,
and this (rules and resources) is social structure, with its potential to either
liberate or constrain social action.
VICTOR W. MARSHALL76
Second, there is cultural variability in conceptions of the good life course;
this variability may lead to conflicts between constituencies – societies,
communities, interest groups – with different idealized views of the life
course.
Third, social structures, such as the organizational features of the two
settings I studied, can have a broader impact in either constraining or fa-
cilitating social action. In a longer time frame, some social structures rec-
ognize the agency of societal or group members, and this can produce new
social institutions and structures (shared meanings, schema, and values plus
patterns of social behavior and accompanying resources) that meet member
needs. Other social structures place agency in the hands of a minority such
as the administrators at St. Joseph’s home, and fail to reward social action
by members. In terms of this research question, it is not simply a case of the
existence or not of conceptions of the good life course, but rather it is also a
case of the existence of structural opportunities or constraints on both the
social construction of such life courses and the realization of these life
courses in individual biographies.
4.3. Operationalizing the Life Course
The second, methodological, question asks how the different concepts (such
as trajectory, stage, transition, and event) are usually operationalized, and
with what consequences. I will examine this with respect to the comparison
of ‘objective’ with ‘subjective’ data about the dying trajectory. The previous
discussion has shown how the objective dying trajectory was differentially
constructed in different societies and in the two settings in the United States
that I had studied. Here I want to first introduce the notion of differential
objective and subjective life courses among those near to death, and then to
describe a situation in which objective and subjective realities conflict, with
intriguing social consequences.
Just as the changing nature of work and retirement has dramatically
altered the objective nature of the working life course, as well as its bio-
graphical or subjective experience through the adult years and the transition
to retirement (Marshall, Heinz, Kruger, & Verma, 2001), changes in life
expectancy and life expectancy in good health have dramatically altered
health transitions and the transition from life to death – both objectively
and subjectively. Objectively, people are living longer and could expect to do
so. I thought it would be interesting to ask people how long they expected to
Agency, Events, and Structure at the End of the Life Course 77
live. The findings point to the subjectivity of such responses and to the
importance of social psychological comparison processes.
I asked respondents at Glen Brae how old they would live to be in three
different ways, with varying success. Of 79 respondents, only 31 gave a
definitive answer to a direct question, ‘‘How old do you think you will live to
be’’; 60 placed a mark on a line to indicate where they saw themselves
between birth and death; and 50 answered a four-alternative question,
‘‘Which one of these would you say about your own future?’’, ranging from
‘‘I shall be around for some time yet; more than ten years’’ to ‘‘The end may
be any time now’’. All indicators correlated quite strongly with age, but
there was a lot more going on in their estimates of what I called ‘‘awareness
of finitude’’ (Marshall, 1975c). Focusing on the third, fixed alternative
measure, responses to it could be better predicted in relation to a variable
describing whether the respondent was currently younger than the age at
death of both parents, one parent or none. That relationship was also much
stronger for men than women. Another strong predictor was the number of
dead brothers or sisters. Qualitative data also suggested that awareness of
finitude is higher (i.e., anticipated life expectancy is shorter) with the death
of friends. Awareness of finitude is thus a partial function of age, but it is
also influenced by a number of social comparison processes, and the
conviction (which is of course supported by data) that ‘longevity runs in
families’.
This said, it is also the case that many Glen Brae respondents found
themselves in the position of having lived beyond the age they had antic-
ipated. Here the objective life course and the subjective life course were at
odds. In the particular living circumstances of this retirement community,
this created problems for many residents – problems made more complex
because the individual residents were not the only ones with a stake in the
accuracy of these estimates. Glen Brae was one of the early examples of a
retirement community organized on the ‘life care principle’. On entry, a
resident would pay a founder’s fee that guaranteed his or her right to remain
in the retirement community for life, paying rents that would be set by the
corporation. Since the founder’s fee was quite significant, a rational person
would not move to this community unless he or she expected to live long
enough to make this a good investment. In addition, the resident had to be
satisfied that, after making the substantial founder’s fee payment on entry,
he or she had enough assets to generate a revenue base to pay the monthly
rental fees as long as they lived. On the other hand, the administration of
Glen Brae used the founder’s fees to amortize the large mortgage on the
new, costly physical plant, while counting on monthly rental fees to meet
VICTOR W. MARSHALL78
operating expenses. Once all units in Glen Brae were occupied, the only way
for the corporation to generate funds to pay the mortgage was through new
founder’s fees, but these could be generated only when a resident died and
the apartment became vacant. Thus, in an abstract sense, two parties –
resident and administration – were making bets based on life-expectancy
estimates.
I examined this situation from the perspective of game theory, and found
that both parties had made significant errors in the life estimates. As noted
earlier, residents made subjective estimates of their life expectancy based on
their age and other social comparisons. They mostly neglected to take into
account general cohort increases in life expectancy, and cohort differences in
socioeconomic status, which correlate with health and life expectancy. In
other words, they were not good life course theorists in that they ignored
cohort factors. (I did not characterize this in life course terms at the time,
but it is a good example of one of the most basic of life course principles).
The corporation, on the other hand, used age alone. Rather, it relied on the
insurance company that held the mortgage to estimate turnover of the
apartment units, and that company did so based on life-expectancy tables
for its insurance policy holders. However, it did not take into account two
things: first that the typical resident of Glen Brae is of a higher social class,
and therefore likely to live longer than the typical life insurance policy-
holder; and second, that the person moving to the retirement community
may be thought of as acting in a different strategic or decision-making mode
than the typical insurance policyholder. The former may be seen to be
making a very strategic bet to live long enough to make this a good in-
vestment, and people who think they will not live very long may very well
opt not to pay that expensive founder’s fee.
To cast this situation as a game is to go well beyond the data, reconstructing
actions of the administration and residents in terms of imputed rationality.
However, the exercise calls attention to a few points related to the research
question I have been discussing. That question, which is put to us as a meth-
odological question, asks, how are different concepts in the life course opera-
tionalized in different disciplines, and with what consequences? As with the first
question (and the next), I am addressing this very narrowly with respect to two
concepts – trajectory, and that troublesome conceptual relationship between
agency and social structure.
The methodology here is principally in my crude attempt to measure and
to understand the anticipated final trajectory or, more precisely, its length or
duration; i.e., awareness of finitude. In other aspects of my research on
aging and dying, I used awareness of finitude as an independent variable, to
Agency, Events, and Structure at the End of the Life Course 79
assess as best I could its causal relationship to other phenomena, such as
concerns about death and dying, time perspective, and the extent to which
the individual is preoccupied with self and identity issues. These investiga-
tions, which I cannot go into here, and those just reported, which try to
explain how people come to make estimates of anticipated life expectancy,
all help us to understand the concept of awareness of finitude (through
construct validity). In terms of the life course perspective per se, there has
been less work done on the anticipated life course than on the experienced
life course (see Markus & Nurius, 1986; Ryff, 1991; Mueller, 2002). The
general points I would make with respect to this methodological question
are:
First, the distinction between objective and subjective aspects of trajec-
tories should be maintained and investigated, and discrepancies between the
two are likely to be theoretically interesting – as work on Neugarten’s notion
of ‘social clocks’ attests (Neugarten & Hagestad, 1976; Ryff, 1991). Antic-
ipated and realized careers or trajectories are ideally investigated with lon-
gitudinal data (Mueller, 2002), and there are great concerns that
retrospective data will be biased through self-serving reconstruction. How-
ever, reconstructions of the past as well as changing constructions of the
future are core materials in studies of the life course.
Second, and more generally, there can be multiple objective and subjec-
tive life courses ‘in play’ at the same time, and objective life courses can
include largely biological aspects of the life course including physiological
changes such as puberty, menopause, physiological declines, and finally
death.18
Third, the investigations I have described, and particularly the use
of game theory, raises questions about limits to Clausen’s (1991) concept of
planful competence, discussed earlier under agency. As noted earlier,
‘planful competence’ is the ability to make informed, rational decisions
and set realistic short- and long-term goals. I suggested earlier that this can
be seen as agency in this first sense of the metaphor – the existence of
the furnace, and it can be viewed as a developmental capacity of (virtually)
all humans. However, using game theoretical analysis is a methodological
trick to show the difference between capacity and outcome or, in my
preferred terminology, between agency and action. Life courses are far
from ‘rational’ in any pure sense, and related to this departure from ra-
tionality is the fact that people live their life courses socially, negotiating
their way through life with other actors also exercising planful com-
petence, but with often-conflicting definitions of the situation, goals, and
resources.
VICTOR W. MARSHALL80
4.4. Status Changes and Changes in Identity
The third, results question asks to what extent changes in status result in the
redefinition of identity. I will address this question by referring to the dis-
tinction between self and identity, and to the action of the aging and dying
individual, in managing identity. The question begs for a definition of
identity, which is a term used in many different ways in the social sciences.19
In my work, identity is that sense of sameness and continuity in the
organization of one’s selves over time. Following Goffman (1963) I also
distinguish between ego of felt identity, imputed or social identity, and
personal or presented identity. The former refers to the individual’s own
sense of selves over time, the second to the views of others, and the third to
the presented self (which as Goffman so well shows, does not always cor-
respond closely to the former and is a way in which people seek to influence
others to change the identity they impute to them).
At the social or collective level, I described above how the residents at
Glen Brae sought explicitly to shape their social identity. The residents had,
in fact, changed their social status when they moved to the retirement com-
munity, which, as a ‘life-care community’ they anticipated would be their
collective home until each died. But they recognized options in how they
were to view themselves and collectively committed to an option that fo-
cused on themselves as living, rather than as dying. There are good social
psychological concepts to deal with this phenomenon, of which the concept
of role distance is as insightful as it is difficult to measure (Goffman, 1961).
Role distance is setting a distance between one’s self and a specific role. It is
a way of remaining more than the specific role one is playing at the time. At
Glen Brae, residents collectively took the position, ‘I know I am dying but
my whole being is not wrapped up in that. I am many other things than a
dying person’.
At a more individual level, I was interested in my early research in how
the recognition of the self as dying leads to changes in ego or felt identity.
This is the metaphor of the individual in the last chapters of life (Marshall,
1980). Long ago, the hermeneutic scholar Wilhelm Dilthey described the
selectivity of autobiographical reflection as follows:
The person who seeks the connective threads in the history of his life has already, from
different points of view, created a coherence in that lifey. He has created it by ex-
periencing values and realizing purposes in his life, making plans for it, seeing his past in
terms of development and his future as the shaping of his lifey. He has in his memory,
singled out and accentuated the moments which he experienced as significant; others he
has allowed to sink into forgetfulness (Dilthey, 1962, p. 86).
Agency, Events, and Structure at the End of the Life Course 81
To summarize my findings briefly, I found a relationship between awareness
of finitude, and reminiscence. While scholars such as Erikson (1959) and
Butler (1963) had focused on awareness of finitude as initiating focused or
intense individual reminiscence and reconstruction of one’s identity, I found
a strong relationship with social reminiscence as well. Asked, ‘‘Do you often
talk about things that have happened in your past life with anybody else?,
36% of respondents estimating more than ten years to live, 32% of those
estimating five to ten years, but fully 62% of those estimating less than five
years said they do so ‘at least once a week’, rather than less frequently
(Marshall, 1975c, p. 125). Higher social reminiscence was associated with a
greater intensity of reminiscence (they reported more turning points when
asked to describe their lives), with reporting turning points that were coded
as reflecting internal rather than external locus of control (things they did
rather than things that happened to them), and with a greater likelihood of
reporting satisfaction with their life as a whole (summarized in Marshall,
1980, pp. 115–119).
It is impossible to sort out causation from these cross-sectional data. Only
a longitudinal study could establish a clear temporal relationship among
changes in awareness of finitude, changes in the extent to which the indi-
vidual engages in individual or social reminiscence, and changes in the con-
tent of the remembered biography as well as satisfaction with the biography
as a whole. My cross-sectional data let me talk about differences but not
changes. My interpretation that social reminiscence assists the individual in
the reconstruction of biography leans more strongly on theory than on my
data. Moreover, in attending to social in addition to individual reminis-
cence, it contrasts with more individualistic, psychological approaches to
biographical narrative, such as those described in McAdam, this volume.
Let me return to the third question, about results, is it possible to dem-
onstrate that a change in status implies a redefinition of the identities of the
persons involved? From my limited study of three decades ago, I infer the
following answer to this question:
First, life course scholars should devote theoretical and methodological
attention to the complexity of identity. If a person has as many selves as
there are groups about whose opinion he or she cares, then a concept of
identity is needed to deal with how the individual prioritizes, situates,
manages, and organizes these many selves so as to give some sense of
sameness and continuity over time. We need first to distinguish self from
identity.
Second, we need to distinguish identity as experienced by the individual
from identity as attributed to the individual by various groups of audiences.
VICTOR W. MARSHALL82
Third, it is easier to describe differences in identity than to explain
changes in identity, and we would all benefit from a sustained attempt to
theorize the mechanisms of identity change. This task calls for multidisci-
plinary skills because these processes, my work suggests, take place at bi-
ological, psychological, and social levels, and the intersection or interaction
of these levels.
5. AGENCY, EVENTS, AND STRUCTURE AT THE END
OF THE LIFE COURSE
I want to quickly conclude with some general remarks based on my visit to
earlier research that I conducted more than three decades ago. I have tried
not to reconstruct the language of this work, while at the same time showing
its relevance to contemporary issues of the life course.
Perhaps, the major conclusion of this exercise is that it reminds us of the
long chains of scholarship that have contributed to what we now think of as
the life course perspective; and perhaps in addition we can be reminded that
the life course perspective of today is highly diverse. I have addressed both
these issues elsewhere (Marshall & Mueller, 2003), focusing on the similar-
ities and differences between North American and European approaches to
the life course.
In some senses I came to the ‘life course perspective’ late, but in other
senses I came to it early. Many theoretical strands come together to form
what is now the life course approach, and I had made use of some of these
strands. Some of them, grounded in symbolic interactionism, gave us con-
cepts like career and status passage that gained currency in parts of Europe,
but were largely supplanted in North America by the terminology subse-
quently introduced by Riley, Elder, and their associates. Status passages and
careers became transitions and trajectories in most American scholarship,
and increasingly sophisticated statistical techniques were applied to analyses
that paid more attention to complex descriptions of the life course than to
how people actually made their way – as individuals and in collective action
– through life courses that they were also helping to construct. The concept
‘agency’ came to be a catch-all phrase that most of the time simply paid
respect to the human capacity for voluntaristic action, and told the reader
that the author was not a complete determinist (despite using deterministic
statistical methods). Agency often meant, ‘unexplained variance’ – if the
Agency, Events, and Structure at the End of the Life Course 83
model did not explain much variance, then this must mean that people were
making innovative choices or struggling against structural barriers.
Revisiting my early, ‘pre-life course perspective’ scholarship, I did not feel
uncomfortable with the symbolic interactional and phenomenological sources
and language I had used. Instead, I think this approach to life course the-
orizing made it easier than the more orthodox language that predominates
today, to address the negotiated character of life course construction and
transitions. I hope that the Europeans will reinvigorate international life course
theorizing by continuing to make good use of the life course–concepts they
initially borrowed from Chicago school sociology.
Revisiting my early work produced a stronger sense than I would have
anticipated of the importance of bringing together objective and subjective
aspects of the life course. When cultural ideas of the life course are com-
bined with resource allocation mechanisms and resources, we have social
structure, which is objective at the macro level. Individuals then create,
experience, and endure their life courses in relation to these objective di-
mensions of social structure. As they do so they may change the social
structure or they may reinforce it, by submitting to its constraints or cap-
italizing on the opportunities it provides (these two together constitute life
chances). Their careers or life courses may turn out to be some jointly
produced product of objective, structural constraints, and opportunities, the
assets and resources they bring to the choices they make, and the micro-level
interactions they have with others who, like them, are trying to create,
experience, and endure their own life courses. A given individual’s vision of
his or her life course may be realistic or not in terms of objective or struc-
tural life course conditions. It may coincide, conflict, or synergize with that
of other members of his or her social group or collectivity and this might in
turn influence the nature of the attained life course. In any case, people will
experience their own objective life courses, the dimensions of which can be
measured and analyzed in life course research. But the full nature of their
life experiences includes a subjective life course as well, one with anticipat-
ions and plans, but also memories and reconstructions or reinterpretations
of the lives they have led.
NOTES
1. Another beginning point might well be the work of the late Klaus Riegel (e.g.,Riegel, 1975, 1976; see Lerner & Busch-Rossnagel, 1981b for a brief discussion of hiscontributions in this area).
VICTOR W. MARSHALL84
2. The work of Piaget is grounded in, and empirically supports, such a notion,with the individual seen as actively seeking novelty so that, through accommodation,schemata will change so as to permit assimilation, enabling increasing mastery ofone’s environment.3. It may be instructive to examine creativity across the life course in light of
agency and developmental theory. See Kastenbaum (1992).4. On this point he is criticized by Dannefer (1999).5. At this point Heinz is in fact drawing on Giddens (1984), but elsewhere in
Giddens, e.g., 1991b) the notion of agency is much more broadly drawn. I discussthis below.6. This conceptualization seems to include agency as an aspect of human nature.
However, in the same article, Dannefer treats agency as a variable. Elaborating on acase described by Gubrium, Holstein, and Buckholdt (1994), of a child labeled as lowin ability and placed in the lowest of three ability groups, Dannefer shows howlabeling creates a self-fulfilling prophecy that keeps this child in his or her place. Hereis how Dannefer interprets this situation:
Even when the outcome is positiveywhat is being described here is a social system
process in which the labeled individual participates in a subordinate manner. Thus, the
agentic force of the individuals in question, including the positively labeled individual is,
in principle, no more in evidence in the case of positive than negative labeling. This raises
the question of whether and how genuine human agency might be augmented – a
question that, in any domain, is ultimately political. (Dannefer, 1999, p. 78).
7. Giddens (1991a, p. 112) contrasts ‘fateful moments’ with daily routines of life.‘‘Fateful moments are those when individuals are called on to take decisions that areparticularly consequential for their ambitions or more generally for their future lives.Fateful moments are highly consequential for a person’s destiny’’.8. I limit the statement as this is not the place to consider whether persons born
with limited or no cognitive capacity are fully human.9. This task is related to, but quite distinct from, the conceptual and method-
ological challenges of linking ‘micro’ and ‘macro’ levels of analysis; where challengesare addressed absent the notion of individual agency. See, for example, Collins(1992); Heinz (1996); Marshall (1995) and O’Rand and Campbell (1999).10. Alwin (1995, p. 217) observes of a panel that tried to define it: ‘‘There was little
theoretical consensus on what structure was and a confusion of meanings, but virtuallyeveryone agreed that the conceptual apparatus conveyed by the concept of socialstructure was what made that sociological contribution to the study of individual livesa possibility’’. The same cannot be said of agency or even of action, as there aresociologists, such as some structuralists, who argue such concepts are not needed.11. Going beyond Giddens’ social structure as virtual ‘rules’ and ‘resources’,
Sewell (1992, p. 8) ‘‘would in fact argue that publicly fixed codifications of rules areactual rather than virtual and should be regarded as resources rather than as rules inGiddens’ sense’’.12. While she postulates this view as against Giddens, Berger and Luckmann,
there are many similarities to both. As for possible differences, Archer sees a con-stant possibility for emergence, from unanticipated consequences of social action,
Agency, Events, and Structure at the End of the Life Course 85
but also from the innovative nature of individuals. The similarity to Schutz is ap-parent in the following: ‘‘The activity-dependence of structures is in no way com-promised by the argument that a given structure was issued in by a particulargeneration/cohort of actors as an unintended yet emergent consequence of theiractivities, whilst it then necessarily pre-existed their successors. This is the humancondition, to be born into a social context (of language, beliefs, and organization)which was not of our making: agential power is always restricted to re-making,whether this be reproducing or transforming our social inheritance.’’ (Archer, 1995,p. 72). Because Giddens’ structuration theory denies the existence of emergence, andalso (p. 87) because Giddens is interested in a short time frame, Archer argues thatstructuration theory has ‘‘an inability to examine the interplay between structure andagency over longer temporal tracts because the two presuppose one another so close-ly’’. Archer’s desire to make time an actual component in the theory may be comparedto the importance of time in most life course perspective approaches, such as Elder.13. I cannot do justice to Sewell’s full argument here, including his use of
Bourdieu, but I think he has given the most satisfactory conceptualization to theagency-structure discussion in current sociological theorizing. In my own thinking Iwould draw further on Berger and Luckmann and Archer (whom Sewell does notresource).14. Data were collected through participant observation, interviewing, and ar-
chival methods as my doctoral dissertation research. I was in the field in these twosettings over the period 1969–1970. The bulk of my data were based on fieldwork andsemi-structured interviews in the retirement community, where I completed at leastone lengthy focused interview with 105 of just under 400 residents. Of these,I interviewed 68 respondents three times, and 92 respondents twice. In the nursinghome, I conducted fieldwork over a 3-month period, mostly using participantobservation, but also unsystematic focused interviews.15. The Limbu of Nepal (Jones, 1974) and the LoDagaa of West Africa (Goody,
1962, pp. 208–209) have a similar conception of the good death and Meyerhoff(1978) describes the dying of an elderly Jewish American in a similar light (Marshall,1980, pp. 33–34, 151–152).16. Hertz (1960, p. 76) notes that in many parts of the world, deaths of children,
strangers and slaves ‘‘arouse no emotion, occasion no ritual’’, and with respect to theaged, he says, ‘‘In various Australian tribes, old people who, because of their greatage, are incapable of taking part in the totemic ceremonies, who have lost theiraptitude for sacred functions, are buried immediately after deathy. This is so be-cause, due to the weakening of their faculties, they have ceased to participate in sociallife; their death merely consecrates an exclusion from society which has in fact alreadybeen completed, and which every one has had time to get used to’’ (Hertz, 1960, p. 85).17. Current terminology in the USA for such a facility would be ‘‘Continuing
Care Retirement Community’’.18. I am not able to address the importance of catastrophe and diversity in life
courses. Many things simply cannot be predicted and these range from major eventssuch as wars, terrorist attacks and natural disasters to individual-level events such asmotor vehicle accidents and heart attacks. Linda George (2003) in a recent essay on lifecourse perspectives in the area of aging and health, notes that health traumas oftenlead to divergence in life course pathways. She also reminds us of the tremendous
VICTOR W. MARSHALL86
divergence of pathways in the absence of trauma, as noted, for example, in the classicstudy by Rindfuss, Swicegood, and Rosenfeld (1987).19. Often the terms self and identity are used interchangeably. I argue that it is
important to distinguish the two. Self, drawing on symbolic interactionism, is theprocess of reflection and cognitive orientation in which the I responds to the self – asobject (the me) based on both internal cues and external feedback from others.Fol-lowing James, I have as many selves as there are groups of people about whom Icare. These many selves develop in relations with others in role relationships andindividual selves become situated in social relationships with (types of) others.However, some selves are more important than others, and moreover the socialscientist needs to be able to account for the individual’s ability to select and presentselves.
REFERENCES
Ackerman, S., Zuroff, D. C., & Moskowitz, D. S. (2000). Generativity in midlife and young
adults: Links to agency, communion, and subjective well-being. International Journal of
Aging and Human Development, 50(1), 17–41.
Alwin, D. F. (1995). Taking time seriously: Studying social change, social structure, and human
lives. In: P. Moen, G. H. Elder Jr. & K. Luscher (Eds), Examining lives in context.
Washington, DC: American Psychological Association.
Archer, M. S. (1995). Realist social theory: The morphogenic approach. Cambridge: Cambridge
University Press.
Bakan, D. (1966). The duality of human existence. Chicago: Rand McNally & Co.
Berger, P. L., & Luckmann, T. (1966). The social construction of reality. New York: Doubleday.
Brim, O. G. (1981). Foreword. Introduction to R. M. Lerner & N. A. Busch-Nagel (Eds),
Individuals as producers of their development: A life-span perspective (pp. xv–xvii).
New York: Academic Press.
Butler, R. N. (1963). The life review: An interpretation of reminiscence in the aged. Psychiatry:
Journal for the Study of Interpersonal Processes, 26, 65–76.
Cain, L. D., Jr. (1964). Life course and social structure. In: R. E. L. Faris (Ed.), Handbook of
modern sociology (pp. 273–309). Chicago: Rand McNally.
Campbell, C. (1996). The myth of social action. Cambridge: Cambridge University Press.
Clausen, J. S. (1991). Adolescent competence and the shaping of the life course. American
Journal of Sociology, 96, 805–852.
Clausen, J. S. (1993). American lives: Looking back at children of the great depression.
New York: Free Press.
Collins, R. (1992). The romanticism of agency/structure versus the analysis of micro/macro.
Current Sociology, 40, 77–97.
Counts, D. (1976–1977). The good death in Kaliai: Preparation for death in Western New
Britain. Omega, 7(4), 367–372.
Cumming, E., & Henry, W. E. (1961). Growing old: The process of disengagement. New York:
Basic Books.
Dahrendorf, R. (1979). Life chances. Chicago: The University of Chicago Press.
Agency, Events, and Structure at the End of the Life Course 87
Dannefer, D. (1999). Neoteny, naturalization, and other constituents of human development.
In: C. D. Ryff & V. W. Marshall (Eds), The self and society in aging processes
(pp. 67–93). New York: Springer (Chapter 3).
Dannefer, D., & Uhlenberg, P. (1999). Paths of the life course: A typology. In: V. L. Bengtson &
K. W. Schaie (Eds), Handbook of theories of aging (pp. 306–326). New York: Springer
(Chapter 17).
Diehl, M. (1999). Self-development in adulthood and aging: The role of critical life events. In:
C. Ryff & V. W. Marshall (Eds), The self and society in aging processes (pp. 150–183).
New York: Springer.
Dilthey, W. (1962). In: H. P. Rickman (Ed.), Pattern and meaning in history. New York: Harper
Torchbooks.
Elder, G. H., Jr. (1975). Age differentiation and the life course. Annual Review of Sociology, 1,
165–190.
Elder, G. H., Jr. (1985). Perspectives on the life course. In: G. H. Elder (Ed.), Life course
dynamics: Trajectories and transitions, 1968–1980 (pp. 23–49). Ithaca: Cornell University
Press (Chapter 1).
Elder, G. H., Jr. (1997). The life course and human development. In: R. M. Lerner (Ed.),
Handbook of child psychology: Theoretical models of human development (Vol.1,
pp. 939–991). New York: Wiley (Chapter 1).
Elder, G. H., Jr., & Johnson, K. M. (2003). The life course and aging: Challenges, lessons, and
new directions. In: R. A. Settersten Jr. (Ed.), Invitation to the life course: Towards new
understandings of later life (pp. 49–81). Amityville, NY: Baywood Publishing Company.
Elder, G. H., Jr., & O’Rand, A. (1995). Adult lives in a changing society. In: K. S. Cook, G. A.
Fine & J. S. House (Eds), Sociological perspectives on social psychology (pp. 452–475).
Boston: Allyn & Bacon (Chapter 17).
Erikson, E. (1959). Identity and the life cycle. Psychological Issues, 1(1).
George, L. K. (1999). Social perspectives on the self in later life. In: C. D. Ryff &
V. W. Marshall (Eds), The self and society in aging processes (pp. 42–66). New York:
Springer (Chapter 2).
George, L. K. (2003). What life-course perspectives offer the study of aging and health. In:
R. A. Settersten Jr. (Ed.), Invitation to the life-course: Toward new understandings of later
life (pp. 161–188). Amityville NY: Baywood.
Giddens, A. (1983). Profiles and critiques in social theory. Berkeley, CA: University of California
Press.
Giddens, A. (1984). The constitution of society: Outline of the theory of structuration. Berkeley
and Los Angeles: University of California Press.
Giddens, A. (1991a). Modernity and self-identity: Self and society in the late modern age.
Stanford: Stanford University Press.
Giddens, A. (1991b). New rules of sociological method (2nd ed.). Stanford, CA: Stanford Uni-
versity Press.
Goffman, I. (1961). Role distance. In: I. Goffman (Ed.), Encounters (pp. 85–152). Indianapolis:
Bobbs-Merrill.
Goffman, I. (1963). Stigma. Harmondsworth: Penguin.
Goody, J. (1962). Death, property and the ancestors. Palo Alto, CA: Stanford University Press.
Gubrium, J. F., Holstein, J. A., & Buckholdt, D. R. (1994). Constructing the life course. Dix
Hills, NY: General Hall.
VICTOR W. MARSHALL88
Gutmann, D. (1965). Women and the conception of ego strength. Merrill-Palmer Quarterly, 11,
229–240.
Heinz, W. R. (1996). Status passages as micro–macro linkages in life course research. In:
A. Weymann & W. R. Heinz (Eds), Society and biography: Interrelationships between
social structure, institutions and the life course (pp. 51–65). Weinheim: Deutscher Studien
Verlag.
Hertz, R. (1960). In: R. C. Needham (Trans.), Death and the right hand. Aberdeen: Cohen and
West.
Hughes, E. C. (1971). The sociological eye: Selected papers. Chicago: Aldine, Atherton.
Jones, R. L. (1974). Religious symbolism in Limbu death-by-violence. Omega, 5(3), 257–266.
Kahana, E., & Kahana, B. (1996). Conceptual and empirical advances in understanding aging
well through proactive adaptation. In: V. L. Bengtson (Ed.), Adulthood and aging: Re-
search on continuities and discontinuities (pp. 18–40). New York: Springer (Chapter 2).
Kalish, R. A. (1966). A continuum of subjectively perceived death. The Gerontologist, 6(2),
73–76.
Kastenbaum, R. (1992). The creative process: A life-span approach. In: T. R. Cole, D. D. Van
Tassel & R. Kastenbaum (Eds), Handbook of the humanities and aging (pp. 285–306).
New York: Springer.
Kiser, E. (1999). Comparing varieties of agency theory in economics, political science, and
sociology: An illustration from state policy implementation. Sociological Theory, 17(2),
146–170.
Lawton, M. P. (1989). Environmental proactivity and affect in older people. In: S. Spacapan &
S. Askamp (Eds), The social psychology of aging (pp. 135–163). Newbury Park: Sage.
Layder, D. (1994). Understanding social theory. London: Sage.
Lerner, R. M., & Busch-Rossnagel, N. A. (Eds) (1981a). Individuals as producers of their de-
velopment: A life-span perspective. New York: Academic Press.
Lerner, R. M., & Busch-Rossnagel, N. A. (1981b). Individuals as producers of their own
development: Conceptual and empirical bases. In: R. M. Lerner & N. A. Busch-Nagel
(Eds), Individuals as producers of their development: A life-span perspective (pp. 1–36).
New York: Academic Press.
Markus, H. R., & Nurius, P. (1986). Possible selves. American Psychologist, 41, 954–969.
Marshall, V. W. (1972). Continued living and dying as problematical aspects of old age.
Unpublished doctoral dissertation in sociology, Princeton University.
Marshall, V. W. (1975a). Socialization for impending death in a retirement village. American
Journal of Sociology, 80, 1124–1144.
Marshall, V. W. (1975b). Organizational features of terminal status passage in residential fa-
cilities for the aged. Urban Life, 4(3), 349–368.
Marshall, V. W. (1975c). Age and awareness of finitude in developmental gerontology. Omega,
6(2), 113–129.
Marshall, V. W. (1978–1979). No exit: A symbolic interactionist perspective on aging. Inter-
national Journal of Aging and Human Development, 9(4), 345–358.
Marshall, V. W. (1980). Last chapters: A sociology of aging and dying. Monterey, CA: Brooks/
Cole.
Marshall, V. W. (1995). The micro–macro link in the sociology of aging. In: C. Hummel &
C. L. D’Epinay (Eds), Images of aging in western societies (pp. 337–371). Geneva: Centre
for Interdisciplinary Gerontology, University of Geneva.
Agency, Events, and Structure at the End of the Life Course 89
Marshall, V. W. (1996). The state of theory in aging and the social sciences. In: R. H. Binstock,
& L. George (Eds), Handbook of aging and the social sciences (4th ed., pp. 12–30).
San Diego, CA: Academic Press.
Marshall, V. W., Heinz, W. R., Kruger, H., & Verma, A. (Eds) (2001). Restructuring work and
the life course. Toronto: University of Toronto Press.
Marshall, V. W., & Mueller, M. M. (2003). Theoretical roots of the life-course perspective. In:
W. R. Heinz & V. W. Marshall (Eds), Social dynamics of the life course: Transitions,
institutions, and interrelations (pp. 3–32). New York: Aldine de Gruyter.
McMullin, J. A., & Marshall, V. W. (1999). Structure and agency in the retirement process:
A case study of Montreal garment workers. In: C. Ryff & V. W. Marshall (Eds), The self
and society in aging processes (pp. 305–338). New York: Springer.
Meyer, J. W., & Jepperson, R. L. (2000). The ‘‘actors’’ of modern society: The cultural con-
struction of social agency. Sociological Theory, 18(1), 100–120.
Meyerhoff, B. G. (1978). A symbol perfected in death: Continuity and ritual in the life and
death of an elderly Jew. In: B. G. Myerhoff & A. Simic (Eds), Life’s career–aging:
Cultural variations on growing old (pp. 163–205). Beverly Hills and London: Sage.
Mueller, M. M. (2002). Work, family and well-being over the life course: Continuities and dis-
continuities in the lives of American women. Doctoral dissertation in sociology, University
of North Carolina at Chapel Hill.
Neugarten, B. L., & Hagestad, G. O. (1976). Age and the life course. In: R. Binstock, E. Shanas,
with the assistance of associate editors, V. L. Bengtson, G. L. Maddox, & D. Wedderburn
(Eds), Handbook of aging and the social sciences (pp. 35–55). New York: Van Nostrand
Reinhold.
O’Rand, A., & Campbell, R. T. (1999). On reestablishing the phenomenon and specifying
ignorance: Theory development and research design in aging. In: V. L. Bengtson &
K. W. Schaie (Eds), Handboook of theories of aging (pp. 59–78). New York: Springer.
Riegel, K. R. (1975). Toward a dialectical theory of development. Human Development, 18,
50–64.
Riegel, K. R. (1976). The dialectics of human development. American Psychologist, 31, 689–700.
Riley, M. W. (1979). Introduction: life-course perspectives. In: M. W. Riley (Ed.), Aging from
birth to death. Interdisciplinary Perspectives (pp. 3–13). Boulder, CO: Westview Press for
American Association for the Advancement of Science.
Rindfuss, R. R., Swicegood, C., & Rosenfeld, R. A. (1987). Disorder in the life course: How
common and does it matter? American Sociological Review, 52, 785–801.
Rossi, A. S. (1980). Life-span theories and women’s lives. Signs, 6(1), 4–32.
Ryff, C. D. (1991). Possible selves in adulthood and old age: A tale of shifting horizons.
Psychology and Aging, 6, 286–295.
Ryff, C. D., & Marshall, V. W. (Eds) (1999). The self and society in aging processes. New York:
Springer.
Ryff, C. D., Marshall, V. W., & Clarke, P. (1999). Linking the self and society in social
gerontology: Crossing new territory via old questions. In: C. D. Ryff & V. W. Marshall
(Eds), The self and society in aging processes (pp. 3–41). New York: Springer.
Schutz, A. (1964). The problem of rationality in the social world. In: A. Brodersen (Ed.),
A Schutz, collected papers II: Studies in social theory (pp. 64–88). The Hague: Martinus
Nijhoff.
Settersten, R. A., Jr. (1999). Lives in time and place: The problems and promises of developmental
science. Amityville, NY: Baywood Publishing Company.
VICTOR W. MARSHALL90
Settersten, R. A., Jr. (2003). Propositions and controversies in life-course scholarship. In:
R. A. Settersten Jr. (Ed.), Invitation to the life course: Towards new understandings of
later life (pp. 15–45). Amityville, NY: Baywood Publishing Company.
Sewell, W. H., Jr. (1992). A theory of structure: Duality, agency, and transformation. American
Journal of Sociology, 98(1), 1–29.
Shanahan, M. J., Elder, G. H., Jr., & Miech, R. A. (1997). History and agency in men’s lives:
Pathways to achievement in cohort perspective. Sociology of Education, 70, 54–67.
Simmel, G. (1950). In: K. H. Wolff (Trans., Ed.), The sociology of Georg Simmel. New York:
The Free Press of Glencoe.
Smelser, N. J. (1997a). Problematics of sociology: The Georg Simmel lectures. Berkeley, CA:
University of California Press.
Smelser, N. J. (1997b). Social structure. In: N. J. Smelser (Ed.), Handbook of sociology.
Newbury Park, London & New Delhi: Sage.
Thomas, W. I., & Znaniecki, F. (1918). The Polish peasant in Europe and America, volumes 1–2.
Boston: Badger.
Weber, M. (1978). In: G. Roth & C. Wirth (Eds), Economy and society, Vol. I: An outline of
interpretive sociology. Berkeley: University of California Press.
Agency, Events, and Structure at the End of the Life Course 91
LOOKING AT AMBIVALENCES:
THE CONTRIBUTION OF A
‘‘NEW-OLD’’ VIEW OF
INTERGENERATIONAL
RELATIONS TO THE STUDY OF
THE LIFE COURSE
Kurt Luscher
1. INTRODUCTION
This chapter has its origins in the kind invitation to present, at the PaVie-
Colloquium, an idea that is receiving increasing attention in the study of
intergenerational relations. Its essence can be summarized in the following
hypothesis: Intergenerational relationships, especially among adult children
and their parents, imply the experience of ambivalences and, consequently,
require dealing with ambivalences.1 Thus, my point of departure does not
seem to be a major issue of life course research. However, at second glance,
one may recall that embeddedness in intergenerational relations is crucial
for personal development. Most human beings are conceived in and born
into familial contexts, and parent–child relationships – as diverse as they
may be – are in many ways important for the unfolding of personal abilities
Towards an Interdisciplinary Perspective on the Life Course
Advances in Life Course Research, Volume 10, 93–128
Copyright r 2005 by Elsevier Ltd.
All rights of reproduction in any form reserved
ISSN: 1040-2608/doi:10.1016/S1040-2608(05)10003-3
93
and the consciousness of the self. Youth is a formative phase in the life
course where intergenerational relationships are of importance, simply be-
cause their dominance may be challenged by other relationships, such as
those among siblings and peers. This is also true for early adulthood. Later,
through partnerships and marriages, and thus the acquisition of in-laws,
there is an increase in the number of elders with whom close and intimate
relationships become possible or are even expected and required. In mid-life,
nowadays, most adults belong to genealogical networks involving three or
even four generations. Later, obligations for the care of the very old may
emerge. The rules and the practice of inheritance once more accentuate the
social and material importance of intergenerational relationships and their
impact for the conduct of personal lives. In addition, the institution of
inheritance reminds us that any life course may also be comprehended as a
link in a chain of generations.
Indeed, the study of the life course may profit from taking into account
the interplay with the study of intergenerational relationships, and conse-
quently from recent developments in this field. To this obvious statement,
I would like to add two points. First, because of their omnipresence, in-
tergenerational relations are at the core of the processes of socialization and
of human sociability. This is why insights from the study of intergenera-
tional relationships are of foremost interest for the analysis of social rela-
tionships in general, be it with regard to what they have in common with
other relationships, or to where they differ from them, for instance from
market relationships. My second remark is meta-theoretical. Because of the
great relevance of intergenerational relations, their understanding is usually
bound to moral judgments. Such normative views often penetrate scholarly
descriptions. For instance, it is quite common to idealize intergenerational
relations – positively – with reference to the concept of solidarity, or to
deplore them – negatively – as a notorious source of conflict. As I will show,
a well-grounded theory of ambivalence allows us to overcome these biases,
because it simultaneously takes into account and analyzes both perspectives.
In this way, a high degree of social authenticity can be achieved, and re-
spective normative orientations can become a deliberate topic of analysis.
Moreover, we recall that general assumptions about human nature underlie
the concepts used in social science research, especially about such funda-
mental issues as the conduct of human lives and their social organization.
However, at this point I cannot present a comprehensive account of the
importance of ambivalence for the study of the life course. I must limit
myself to outlining the meanings of this concept as such, and I will present
the conceptual frame that I and other researchers have developed. Taking
KURT LUSCHER94
this as a point of reference, I will also illustrate the usefulness of this ap-
proach by presenting some exemplary results of recent research. I shall
concentrate on issues closely related to the study of the life course and of
connected lives, and I will suggest further applications in this field.
2. AMBIVALENCE IN INTERGENERATIONAL
RELATIONS: THE REDISCOVERY OF
AN OLD EXPERIENCE
The idea of drawing on the concept of ambivalence for the study of
intergenerational relationships has two sources. First, an awareness of the
usefulness of ambivalence as a theoretical concept arose from a critical eval-
uation of the existing literature on intergenerational relationships, which in
the 1990s was aptly characterized as data-rich and theory-poor (Luscher &
Pillemer, 1998). In particular, we criticized the dominance of the so-called
solidarity perspective, because it presents a picture of intergenerational re-
lationships that pays too much attention to positive aspects and too little
to the innately darker ones. The solidarity perspective arose in reaction to
Talcott Parsons’s (1942, 1949) portrayal of the nuclear family as isolated. It
holds that, to the contrary, extensive family solidarity does exist. (Shanas et
al., 1968; Littwak, 1965; Sussman, 1959). Since the early 1970s, Bengtson and
co-workers have continued to develop this approach in an influential series of
articles and books (cf. Roberts, Richards, & Bengtson, 1991; Bengtson
& Harootyan, 1994; Bengtson, Giarusso, Mabry, & Silverstein, 2002). The
solidarity perspective has also been adopted by other researchers in the Unit-
ed States (Rein, 1994; Rossi & Rossi, 1990) and serves as a reference point for
many European authors, although not without critical reservations (Attias-
Donfut, 1995; Bawin-Legros, Gauthier, & Strassen, 1995; Donati, 1995;
Finch & Mason, 1993; Szydlik, 2000). However, at the same time as scholars
in the solidarity tradition have emphasized mutual support and value con-
sensus, another line of research has focused on isolation, caregiver stress,
family problems, conflict and abuse (Marshall, Matthews, & Rosenthal,
1993). The image of weakened family ties and the abandonment of the
elderly continues to be widely held in popular opinion and in portrayals of
the family in contemporary fiction and theater. Thus, some scholars, as well
as the public at large, appear reluctant to accept that intergenerational re-
lationships include solidarity and are characterized by shared values and
reciprocal help. As Marshall et al. (1993, p. 47) have succinctly put it, ‘‘the
Looking at Ambivalences 95
substantive preoccupations in gerontology over the past 30 years point to a
love–hate relationship with the family.’’ In a somewhat different mode,
Lalive d0Epinay and Bickel (1994), summarizing their comprehensive
description of the aged and their familial networks in Switzerland, refer to
the tensions created between the potentials of family solidarity and the
limitations imposed by contemporary social conditions. In view of such
accounts, Karl Pillemer and I have proposed that the study of parent–child
relations in later life must move beyond a ‘‘love–hate relationship’’ (Luscher
& Pillemer, 1998). The vacillation between images of mistreatment and ne-
glect, on the one hand, and comforting images of solidarity, on the other,
are not two sides of an academic argument that will ultimately be resolved in
favor of one viewpoint.
Second, parallel to this theoretical evaluation, we conducted a research
project at the University of Konstanz on the reorganization of families after
divorce in later life, e.g. an important event in the life course (Luscher &
Pajung-Bilger, 1998). Data were collected in semi-structured interviews with
103 persons in 65 families. These interviews included questions about the
way all the subjects experience intergenerational relations. Our goal was to
distinguish different degrees of mutual solidarity in the aftermath of what
in many cases represents a ‘‘turning point’’ in the lives of the individuals
involved and their experience of intimate relationships. Yet, even a differ-
entiation in terms of everyday concerns, and by content and types of re-
lationships, did not yield conclusive results regarding the relevance of
solidarity. Family members reported both instances of support and of
neglect. This led us to search for a concept with which we could take into
account the existence of both solidarity and conflict in the process and the
understanding of intergenerational relations. The notion of ambivalence in
the everyday sense (being torn in two directions) was a first and natural
choice.
In the course of work along these lines, we became aware, however, that
references to the experience of ambivalence in social relationships, and es-
pecially in personal relationships, which involve dependency and intimacy,
have long been a topic of popular wisdom and of literary writings, even
before the term existed. Indeed, insights into what we call in modern lan-
guage ‘‘ambivalence’’ between parents and adult children can be traced back
to the beginnings of human society. In Greek mythology, some of the
greatest sagas depict what we now refer to as ambivalence. The best known
of these is the tragic drama of the relationship between Oedipus and his
father and mother. Reinharz (1986) gives an informative overview on ‘‘lov-
ing and hating one’s elders’’ as ‘‘twin themes in legend and literature.’’
KURT LUSCHER96
She refers, among other examples, to the tragedy of Uranus and his sons.
Hamlet as well, she tells us, can be read as a portrait of intergenerational
relations. Peter von Matt (1995) presents a comparable and very colorful
overview of the theme under the provocative title ‘‘Verkommene Sohne,
missratene Tochter’’ (Degenerate Sons, Misguided Daughters). He draws a
line from the biblical story of Absalom to the admonitory children’s book
Der Struwwelpeter (Shock-headed Peter – a classic of moralizing German
children’s literature) and recalls the complex relationships described in The-
odor Fontane’s Effie Briest and in Kafka’s tale ‘‘The Metamorphosis.’’ We
may add, as one more illustration certainly known to many readers, Philip
Roth’s novel American Pastoral as an example of ambivalence in recent
American literature.2 Furthermore, ambivalence can be seen as an ongoing
theme in the life-script or biography. Kierkegaard could serve as one of many
examples. An impressive study with ambivalence as a latent theme is Lee’s
(1998) study of generativity in the life course of the dancer Martha Graham.
In everyday life, ambivalences are often experienced, for example, in ne-
gotiations over caregiving. They can also be found by examining the overall
history of a given relationship. Seen this way, ambivalence is a conceptual
tool for evaluating specific situations, as well as for studying the develop-
ment and institutionalization of the self in the life course. This brief account
of recent approaches to the study of intergenerational relations (and given
the already-mentioned interplay: to the study of the life course) that draw
upon the idea of ambivalence illustrate why it is appropriate to speak of a
‘‘new–old perspective.’’ However, in order to become a useful tool for con-
temporary social research, a complete, detailed conceptualization is needed.
3. CONCEPTUALIZING AMBIVALENCE
3.1. Elements of a Comprehensive Definition
In the light of the foregoing, it seems reasonable to start with a brief look at
the original formulation of the term. As far as we know, ambivalence was
conceived and first introduced by the Swiss psychiatrist Eugen Bleuler
(1910) as one of four core symptoms of schizophrenia. Yet, soon thereafter
he argued that ambivalence is not merely a symptom of mental illness, but
can also be experienced and thus observed in everyday life. He distinguishes
between affective and cognitive ambivalence and points out that the two are
closely intertwined (Bleuler, 1914, p. 98). His text already contains a ref-
erence to ambivalence in intergenerational relationships (p. 103). Freud first
Looking at Ambivalences 97
used the concept in an article on the theory of transference (e.g. also with
regard to social relationships!). Later, he included it in his theory of the
Oedipus complex, as is concisely and clearly demonstrated in a short essay,
‘‘Some reflections on schoolboy psychology’’ (Freud, 1914). Freud thus ap-
plied ambivalence to the analysis of an exemplary intergenerational phe-
nomenon, as well as assigning it a role in the life course.
This is not the place for a more detailed history of the concept, its re-
ception and its adaptation in different scholarly discourses. Taking into
account the major contributions and arguments in the existing literature,3
I would list the following elements as constituents of a comprehensive un-
derstanding of ambivalence:
� The experience of diametrically opposed (polarized) structures and forces
in the dynamic fields of individual (and collective) actions and respective
relationships.� The insight that these experiences are relevant for the identities (selves) of
the actors (individuals, in certain contexts also collective actors). In other
words, the experience of ambivalence and the ability to cope with it can be
understood as an aspect of human agency.� The assumption that these polarizations will be interpreted as irreconcil-
able as long as the actors belong to a certain field of action (or situation)
and are concerned, in this context, with the reflection of these tasks. This
field of action can be brief, e.g. a turning point, or extend over a longer
period of time (for instance becoming a parent).4
� The assumption that the experience of ambivalences and the ways of
dealing or coping with them can be systematically connected with the
aspects of psychological functioning, of the logic of social relations and
social structures, including the regulation of social control and power.
In view of the background of the concept’s history and its acceptance in
the social sciences, I would like to propose the following definition: For
purposes of sociological research on intergenerational relations, it is useful
to speak of ambivalence when polarized simultaneous emotions, thoughts,
volitions, social relations and structures that are considered relevant for the
constitution of individual or collective identities are (or can be) interpreted
as temporarily or even permanently irreconcilable.
Taking this attempt at a comprehensive analytical definition as a refer-
ence point, we find, in scholarly texts, two different usages. First, the term
can serve as an interpretative (or explanatory) concept. This is, in fact,
its primary use in macro-sociological texts as, for instance, in the wide-
spread characterization of ‘‘post-modernity’’ as pervaded by ambivalence.
KURT LUSCHER98
References to social reality are confined to generalizations, based mostly on
highly aggregated, generalized data. Descriptions are sometimes presented
in the form of ‘‘ideal-types’’ or ‘‘model personalities’’ such as Bauman’s
(1997) proposed ‘‘tourist’’ or ‘‘player.’’ This usage is also common in re-
search reviews, for instance in Cohler’s text about young adults ‘‘coming
out’’ as gay or lesbian and their parents (see below). Second, the concept of
ambivalence may be used as a ‘‘research construct.’’ Here, the goal is to
apply the concept in research, such as in surveys, experiments, observations
and the analysis of documents. For this purpose, an explicit definition is
necessary – one that can serve as the reference point for formulating specific
hypotheses and constructing research instruments.
We can hypothesize that people must live with ambivalences and that they
can cope with them in more or less competent, productive ways. People can
even create ambivalences, as mentioned above with regard to the works of
creative writers and artists. Deliberately constructing ambivalences can also
be a strategy in social interaction. This possibility is another reason to view
ambivalences as both opportunities and as burdens. In this regard, the un-
derstanding of ambivalence suggested here differs from other usages where –
more or less explicitly – the term bears a negative connotation. This is true,
for instance, of the term’s usage in characterizing styles of attachment be-
tween mothers and children, as well as in other typologies.
Closeness and intimacy may reinforce or strengthen the susceptibility to
ambivalence. An important precondition of ambivalence is dependency
(Smelser, 1998), which begins with birth (or even during pregnancy), con-
tinues through childhood and youth into adulthood, and in many cases even
into the later phases of the life course. It manifests itself very early in the
needs for nurture, care, protection and education. Beyond these immediate
obligations, and in the course of fulfilling them, parents develop and acquire
specific information and particular knowledge about their individual child
as a person. This knowledge reinforces the parents’ power to control and to
discipline the child, not only while he or she is young, but also in later life
phases. Over the intergenerational life course, the direction of dependency
between children, parents and older or younger generations may become
more complicated – support and care are specific instances explored in this
book. Yet the authority of older persons, established early in life, may
persist as another source of ambivalence, even as situations arise that lead to
a potential or actual reversal of dependency. Cohler and Grunebaum’s
(1981, pp. 120ff., 197ff.) studies of the relationships of mothers and daugh-
ters in Italian immigrant families provide many convincing illustrations of
this process (see below). More generally, ambivalences in the past and the
Looking at Ambivalences 99
present may offer an interesting topic in the study of life reviews, both in
scholarly work (Staudinger, 1989) and in the curricula of courses offered on
practical gerontology.
The contemporary relevance of ambivalence can be deduced from a close
examination of the structural and cultural conditions of present Western
(postmodern) societies. On the macro-sociological level, population dynam-
ics have created a frame in which ambivalence easily emerges. The rise in life
expectancy, attributable to improved living conditions for increasingly large
segments of the population, was accompanied by a decrease in infant mor-
tality. As a child’s chances of survival increased, the possibility of seeing it as
an individual person also increased. A decrease in the birth rate was a logical
consequence. Childhood and youth soon came to be seen as specific phases
of the life course calling for their own institutions – for instance, public
schooling. The same observation can be made with respect to the other end
of the life course via the recognition of aging as a life stage calling for its
own institutions. The demarcation of different periods or segments of the
life course has led to a heightened consciousness of the importance of re-
lationships between age groups, or in other words, between generations.
This has been true especially in the realm of the family, and also in society as
a whole. The development of social welfare was another factor contributing
to this demarcation of life stages and of intergenerational relationships. In
many instances, structural conditions for both dependence and autonomy
were thereby created. Seen in this way, the concept of ambivalence is an-
other possibility to relate the analysis of the life course to the study of
contemporary society and the dynamic interplay of generations and their
cultural manifestations (see for example Edmunds & Turner, 2002a, b;
Blossfeld, this volume).
3.2. Proposal for a Research Module
The foregoing discussion represents a background for new applications in
research and respective operationalizations.5 The concern shared by the
study of intergenerational relations and life course analysis for the
development of personal identity (or the self) through interaction and in-
stitutionalization is a major point of reference and allows us to concom-
itantly pay attention to social relationships. This approach is compatible
with a two-dimensional view of personal identity, particularly with G.H.
Mead’s (1938) notion of the self as emerging from the interplay between ‘‘I’’
and ‘‘me,’’ where ‘‘I’’ refers to spontaneous subjectivity and ‘‘me’’ refers to
KURT LUSCHER100
generalized others or, more generally speaking, to the interplay between a
subjective and an institutional component of the self. Many interpersonal
models of personality explicitly refer to Mead. For example, Leary (who
developed a circumplex model that describes personality as located between
the poles of love vs. hate and dominance vs. submission) speaks of Mead as
a ‘‘creative watershed to which later theories of interpersonal relations can
trace their sources’’ (Leary, 1957, p. 101).
We can see in the juxtaposition between the subjective and the institu-
tional dimensions a primary condition for the experience of ambivalences. In
addition, within the module presented below, a secondary condition is sug-
gested by hypothesizing that both dimensions of an intergenerational re-
lationship, the subjective as well as the individual, can be influenced and
shaped by fundamental polarizations. Thus, the module is based on a
‘‘twofold’’ notion of ambivalence. This implies a departure from the eve-
ryday understanding of the term.
The ‘‘personal’’ or ‘‘subjective’’ dimension can be characterized as fol-
lows: Parents, children and the members of other involved generations share
a certain degree of similarity. While some of this similarity can be attributed
to biological inheritance, no inheritance is total, insofar as individual par-
ents and individual children are never genetically identical. Their similarity
is reinforced by the intimacy of interactive learning processes, which creates
a potential for closeness and subjective identification. At the same time, the
biological equipment of each organism is different. Sociologically speaking,
processes of maturation increase difference and diversity. Ultimately, chil-
dren develop different personal identities than their parents. In order to
create a schematic representation that can be used in different contexts, two
rather abstract labels are needed. To account for not only the socio-spatial,
but also for the socio-temporal aspects, we propose – for the subjective
component – the terms ‘‘convergence’’ and ‘‘divergence.’’ These two po-
larities can serve as umbrellas for a variety of attributes. Convergence
includes such relational attributes as loving, warm, solicitous, reliable
and close. Divergence is characterized as cool, easy-going, indifferent and
superficial.
For the structural–institutional component, we can conceive of a polar
opposition between a desire to preserve the traditional social forms or
structures of relationships and a desire for dramatic change. Neither is fully
realizable. For instance, although children may choose a way of organizing
their private lives that is vastly different from that customary in their family
of origin, some ties to childhood experiences may remain, even if only in
that they provide a negative background. As technical designations, taking
Looking at Ambivalences 101
into account again the socio-temporal as well as the socio-spatial aspects,
the terms ‘‘reproduction’’ and ‘‘innovation’’ appear useful to express the
idea of a dynamic polarization. Here, reproduction includes relational
attributes such as inflexible, restrictive and ‘‘stuck in a rut.’’ Innovation is
expressed by terms such as open to new experiences, changeable and so on.
We can represent these considerations in the form of a module (or
diagram). In this way, it is possible to analytically deduce four basic modes
of experiencing and dealing with intergenerational ambivalences. Referring
to empirical findings and their discussion, as well as to conceptual consid-
erations, we went through different phases of representation.6 We also took
into account criticisms that representation in the form of a circumplex-model
suggests a static typology, in other words, one where a certain way of
dealing with ambivalences is viewed as finite. Overcoming this limitation
is highly desirable in the field of life course studies. It seems likely that
individual modes of experiencing ambivalences and coping with them change
as people move through different contexts and segments of their lives.
In order to visualize the dynamics of development e.g. the possibility to
move from one type of experience and of coping to another, we suggest
using the geometric form of a spiral. As for characterizations of the modes
of ambivalence, the already-existing descriptions seem still useful. Thus, the
modified module (graphic representation) can be presented and commented
on in the following way (see also Luscher & Pajung-Bilger, 1998; Luscher &
Lettke, 2002, 2004; as well as Lang, 2004; Brannen, 2003):
1. Solidarity refers to reliable support, or the willingness of the generations
to provide each other with services of a not necessarily reimbursable sort.
This involves the exercise of authority, but not in the sense of a one-sided
exertion of influence and power. Rather, it is understood as represent-
ative action including empathy. The maxim of action can be character-
ized as to ‘‘preserve consensually.’’ The members of a family feel
committed to their traditions and get along with one another quite well.
Thus, ‘‘solidarity’’ is one possible mode of dealing with intergenerational
ambivalences, which in this case may be more covert than overt. (It
should be noted that this term implies a specific notion of solidarity and
that the term ‘‘loyalty’’ may also be appropriate for this dynamic.)
2. Where family members strive for emancipation, actions predominate that
support mutual emotional attachment (convergence) and openness to-
ward institutional change (innovation). Relationships between parents
and children are organized in such a way that the individual development
and personal unfolding of all family members is furthered without losing
KURT LUSCHER102
sight of their mutual interdependence. This general setting contains a
certain amount of direct, common purpose pursued by efforts to ‘‘mature
reciprocally.’’ Tensions can be discussed openly, and temporary practical
solutions can be continually negotiated.
3. Atomization takes into account that family cohesiveness is no longer as-
sured by institutional ties and the subjective experiences of relational
histories. The concept expresses the fragmentation of the family unit into
its smallest components, specifically individual family members who
‘‘separate conflictingly.’’ Apart from the unalterable fact that family
members are parents and children, they otherwise have very little in
common. Actions follow a line of conflicting separation, although an
awareness of generational bonds remains.
Subjective (personal) dimension: Convergence vs. Divergence
Institutional dimension: Reproduction vs. Innovation
Captivation Atomization
EmancipationSolidarity
To preserve
consensually
To mature
reciprocally
To conserve
reluctantly
To separate
conflictingly
Divergence
Reproduction Innovation
Convergence
Intergenerational Ambivalence: A research module
Looking at Ambivalences 103
4. Captivation designates cases where the family as an institution is invoked
to support the claims of one family member against another. A fragile
relationship of subordination and superiority thereby arises in which
moral claims and moral pressure are used to exert power. Usually one
generation, predominantly the parental, attempts – by invoking the in-
stitutional order – to assert claims on the other or to bind them by means
of moral appeals without, however, basing its claims on a sense of per-
sonal solidarity. The guiding maxim here is to ‘‘conserve reluctantly,’’
whereby family members may try to ‘‘instrumentalize’’ each other, not
respecting each other as subjects, but using each other as ‘‘means to an
end’’ or as objects.
I would like to underscore the heuristic character of the module. It is used
in an attempt to synthesize and visualize certain basic assumptions about
intergenerational ambivalence and to suggest a first set of labels for the poles
that characterize the dimension of simultaneously experienced juxta-
positions. It also suggests ways to see how the micro- and macro-systems
are embedded in a social ecology of action. The module, so far, emphasizes
the experience of ambivalences in relationships. Metaphorically, we can
evoke the image of a ‘‘dialogue with significant others.’’. Along this line, we
can think of other modi. Thus, we can comprehend the experience of am-
bivalences in the form of a ‘‘dialogue with oneself,’’ and furthermore as a
‘‘dialogue with generalized others,’’ namely as a quarrel with general nor-
mative (societal) expectations, or prescriptions.
As a general schematic representation, the module encourages further
differentiations and adaptations to specific research topics. Such specifica-
tions seem to be necessary, especially in applications to life course analysis.
Thus, I offer the foregoing conceptual ideas as a proposal to analytically
structure the field of research in terms of the concept of ambivalence, par-
ticularly in studying intergenerational relationships. Existing studies can be
characterized by the way, and to the extent that, they refer to elements of
this conceptualization, or use alternatives. The conceptualization represents
one of several possible approaches.
4. CURRENT STATUS OF RESEARCH
4.1. Methodological Preliminaries
Although this is not the place for a detailed methodological discussion (for
this see Lettke & Klein, 2004 and the literature discussed there), I will start
KURT LUSCHER104
with a brief comment on the possibilities to assess the experience of am-
bivalences and respective actions. In general, it seems more reasonable to
use qualitative methods. But we should not ignore the fact that they require
highly elaborate interpretative strategies in order to achieve inter-subjective
validity, especially when studying accounts given in everyday language and
experiences that are not always conscious. Beside the well-established re-
search techniques in the social sciences, advances may also be possible
through cooperation with literature studies. For instance, Zima (2002) pro-
vides a complex demonstration of ambivalence on the level of syntax, on one
hand, and on the level of semantics and content, on the other. In quan-
titative research, a major obstacle lies in the general orientation of many
scaling techniques, insofar as they strive for clarity, in an effort to strictly
avoid contradictions. In the available research on ambivalence, the follow-
ing approaches, techniques and methods are found:
1. Interview techniques addressing the awareness of ambivalence: Respond-
ents can be asked about their awareness of ambivalences in a more or less
direct way, by using the term itself or by presenting circumscriptions such
as ‘‘feeling torn in two directions.’’
2. Assessment of relationships with regard to covert ambivalence: Subjects can
be invited to characterize their relationships with polarized attributes pre-
sented separately, such as warm or loving for convergence, indifferent or
superficial for divergence. If the answers are contradictory, because both of
the two opposing attributes are simultaneously judged applicable, they can
be transformed into indicators of ambivalence. Currently, the most widely
used procedure is one proposed by Thompson, Zanna, and Griffin (1995).
3. Use of vignettes: Subjects are presented with situations in which they have
to make ambivalent choices.
In the following overview, I concentrate on contents. It is not meant to be
comprehensive, but rather illustrative. Its focus is on findings and studies,
mostly of a quantitative nature, which highlight aspects that may be espe-
cially relevant for transfer from the analysis of intergenerational relations to
life course research. The systematization is not a strict one, insofar as some
studies obviously concern different topics.
4.2. Assessment and Differentiation of Ambivalences
Ambivalences, formulated in direct or circumscribed ways, are part of eve-
ryday life and are therefore commonplace experiences for men and women,
Looking at Ambivalences 105
parents and (adult) children. This finding has frequently been confirmed.
For instance, an exploratory study by Pillemer and Suitor (2002, p. 609)
demonstrates, ‘‘that direct measures of ambivalence toward children can be
used effectively...and that ambivalent assessments of the relationship are
sufficiently widespread to be of scientific interest.’’ In another analysis of the
same data, concerning mothers’ general assessments of parent–child
relationships, Pillemer (2004, p. 128) concludes that the ‘‘data offer con-
vincing evidence that parental ambivalence regarding adult children is suf-
ficiently widespread to be of scientific interest.’’ Similar conclusions can be
drawn from studies by Connidis (2001), Jekeli (2002), Spangler (2002), and
Willson, Shuey, and Elder (2003) and others.
Coenen-Huther, Kellerhals, and von Allmen (1994) made a survey of the
relations among kin in a representative sample of families. They discovered
that a majority of relations, approximately 60%, were experienced and
judged positively. However, one third (36%) referred to ambivalences, and a
small minority (4%) judged their relationships negatively. More interest-
ingly, the intensity of dilemmas rose with the frequency of mutual help.
Ambivalent judgment that are considered important can be detected in
about half of the cases. The authors conclude: ‘‘Intensive solidarity is not
self-evident’’ (Coenen-Huther et al., 1994, p. 334). Reluctance is apparent,
especially in long-term relations.
The ongoing studies at Konstanz (Luscher & Lettke, 2004) confirm that if
one asks about them directly, using everyday expressions, experiences of
ambivalence turn out to be almost commonplace. A similar picture emerges
from data concerning the answers to contradictorily formulated statements
about relationships, such as, for instance, the following statement: ‘‘[Name
of other person] and I often get on each other’s nerves, but nevertheless we
feel very close and like each other very much.’’7
In addition, these studies yield a finding that is particularly relevant for
life course research: The experience of ambivalence is not judged, per se, as
negative. Of importance seems to be the level, the intensity and perhaps the
context of ambivalent experiences. In other words, dealing with ambiva-
lences may be understood as a challenge, hence in the context of the life
course as a ‘‘developmental task.’’ Here, a connection exists to the origins of
the concept and its elaboration in psychotherapy, where several authors see
the acceptance or the ‘‘tolerance of ambivalence’’ as a criterion of growth
and maturity and stipulate it as a goal of therapeutic efforts.
We also find the idea of an optimal level in the experience of ambivalence,
for example in a study by Mayer and Filipp (2004). This questionnaire study
explored middle-aged adults’ perceptions of their parents’ generativity and
KURT LUSCHER106
the interpersonal consequences of these perceptions. The subjects assessed
the typicality of behaviors indicating generativity for their mother or father
and evaluated the parent–child relationship on several measures (affection,
manifest and latent conflicts). Some of those relations were moderated by
adult children’s positive regard for parental advice. Affection was highest at
intermediate levels of perceived generativity, but was also linked with mod-
erate levels of manifest parent–child conflict. In the understanding of the
authors, these results ‘‘suggest to analyze effects of generativity under the
aspect of intergenerational ambivalence’’ (Mayer & Filipp, 2004, p. 166).8
The idea of an optimum level is useful to interpret nonlinear variations
and correlations as the expression of the interplay between contradictory
forces. Such a view encourages a secondary analysis of existing research.
Empirical research on kin networks shaping the life course suggests that the
effect of support networks on conjugal quality is curvilinear (Holman,
1981), i.e., extremely cohesive networks might be detrimental to conjugal
functioning. The interference model (Johnson & Milardo, 1984; Julien,
Markman, Leveille, Chartrand, & Begin, 1994) states that social networks
and conjugal relationships may actually compete. Developing relationships
create anxiety in social networks, because the time and energy devoted to
other relationships are thereby reduced. Thus, social network members may
try to hold or regain some influence on their ego by interfering with conjugal
relationships. In this perspective, strong networks may not buffer the effects
of conjugal conflict, but may actually increase them, because the emergence
of conjugal problems opens doors to further interference by network mem-
bers with a couple’s relationship. These examples also invite us to look at the
dynamics of conjugal relationships as a field of overt and covert ambivalent
feelings and behaviors.
In the Konstanz studies, as outlined in the conceptual part of this chapter,
we emphasize the analytical distinction between an institutional and a sub-
jective dimension of ambivalence. The data suggest evidence for the fruit-
fulness of this idea. In general, ambivalences on the institutional dimension
seem to be more pronounced than on the subjective dimension (Lettke &
Luscher, 2001, p. 527ff.). This is true for both parents and adult children, a
finding which suggests, in addition, that the so-called ‘‘generational stake’’
hypothesis is questionable with regard to ambivalences. Overall, then, am-
bivalent experiences seem commonplace, yet they differ in character. In
other words, the concept of ambivalence should be differentiated. This is an
idea that can be traced back to Bleuler, who distinguished ambivalences of
feelings, cognitions and volitions. Other authors also adopt this view in their
current work (see for instance, Lorenz-Meyer, 2004).
Looking at Ambivalences 107
Brannen (2003), in a small-scale study of four-generation families, pro-
vides a typology of intergenerational relations with respect to the transmis-
sion of material assets, childcare and elder care, sociability, emotional
support and values. It examines two a fortiori conditions that are thought to
shape intergenerational relations: (a) occupational status continuity/mobil-
ity and (b) geographical proximity/mobility. Four types of intergenerational
relations are generated by this examination: traditional solidaristic; differ-
entiated; incorporation of difference; and reparation in estrangement. The
authors look at families holistically and draw on the concept of ambivalence
to describe the forces which encourage family members to preserve family
patterns and divisive forces that lead them to strike out on their own. It
shows how, whatever the type of intergenerational pattern, each genera-
tional unit seeks to make its own particular mark.
4.3. Diversification of Contexts
In the wider horizon of a comparative study, Fingerman and Hay (2004,
p. 145ff.) ‘‘revealed that parents and their offspring do seem to experience
greater ambivalence toward one another than they experience in many other
social ties.’’ However, other relationships are also considered ambivalent, in
particular ties to romantic partners and ties to siblings. The authors’dis-
cussion hints at another topic of interest in the possible application of the
ambivalence perspective to the study of the life course. Since nearly all the
romantic partners of adults older than 20 in the Fingerman and Hay study
were spouses or cohabiting partners, they hypothesize that ‘‘proximity may
play a role in the experience of ambivalence with romantic partners and
siblings. When siblings grow up and no longer live in the same household,
there is a precipitous drop in the likelihood that they will be classified as
ambivalent; teenagers classified their ties to siblings as ambivalent, whereas
individuals in their 20 s did not. It may simply be the case that individuals
are more likely to experience ambivalence when they occupy the same life
space. This pattern regarding proximity was not the same for parents and
children, however. Adult children in their 20 s who do not reside in their
parents’ households were more likely to consider their ties to their parents
ambivalent than were teenagers who lived with their parents. Therefore,
ambivalence between parents and children may reflect different factors than
does ambivalence in other social ties’’ (ibid.). The conclusion that suggests
itself is plausible: The experience of ambivalences may change over the life
KURT LUSCHER108
course, but this is certainly only the starting point for a range of propo-
sitions still to be developed.
In an extension and follow-up of the survey done at Konstanz (see above)
using as far as appropriate the same instruments, interviews have been made
of two types of families facing specific tasks and difficulties. In one group,
an adult child suffers from schizophrenia, in the other group, an adult child
is on drugs. In both instances, the child was living in a clinical institution at
the time of the research. This design allows, among others, a comparison
between statements concerning the relationship to the sick child and to other
children in the same family. The data show, as hypothesized, a higher fre-
quency of ambivalence in the relationship with the sick child, and a lower
relationship quality. Surprisingly enough, there is no significant difference in
feelings of connectedness to the children in the families (Brand, 2004;
Rudorf, 2004; Burkhardt, 2005).
Taking into account additional findings, the general conclusion for this
study is the conclusion is justifiable that most parents distinguish among
their children in many ways, yet they feel close to and committed to all of
them. These results give rise to certain doubts and criticisms of the holistic
view of families propagated by some popular systemic approaches used in
family therapy. More generally, we may again observe that the usage of
indicators of ambivalence, i.e. the ambivalence perspective, promises an
understanding of families that reflects their internal dynamics and therefore
comes close to real life. The subjective attitudes and orientations of family
members are taken into account without neglecting the role of institution-
alized bonds.
The concern for parent–child relationships in exceptional families is also
reflected in studies of families with gay or lesbian children. A large body of
research is available; Cohler (2004) offers a comprehensive overview draw-
ing upon the interpretative power of the concept of ambivalence. Among the
many topics covered, of particular interest in the life course perspective is
the process of ‘‘coming out.’’ It is subject to several forms of ambivalence
and requires different strategies of coping, e.g. with regard to personal
sameness and difference, to traditional and new life styles. Parents may also
have the task of revealing their child’s sexual orientation to kin and friends.
On another level, a kind of institutional ambivalence may be implied in
the way legislation deals with homosexual partnerships. Should they be
treated as just another form of marriage, or should a special legal institution
be created (e.g. civil partnerships or civil unions, as is the case in most
European countries)? Do gays and lesbians themselves want to accept rules
derived from traditional marriage, especially with regard to the dissolution
Looking at Ambivalences 109
of the relationship? Quite to the point, the German author Lautmann (1996)
uses the notion of ‘‘ambivalences of the law.’’
Extending the horizon, it is easy to propose other family configurations as
breeding grounds for latent and manifest ambivalences. In single-parent
families, relationships with the absent father or mother and struggles for
custody may bear all the features of an enduring conflict, putting the child in
an ambivalent position. In the case of foster families, the child as well as
those who have institutional responsibilities for the arrangement, such as
social workers, may find themselves caught up in struggles between the
biological mother and the so-called social parents (or legal parents). Here
too, legal regulations and procedures may be relevant to the search for a
way of pragmatically coping with ambivalences. In Germany, this is the case
for the legal obligations of adult children to support their parents when they
are poor and need institutional care (Hoch & Luscher, 2002).
Divorce at all stages of marital and generational biographies may accen-
tuate, often over a longer period or for an entire lifetime, overt and covert
ambivalences. One is reminded of the proposal by Cherlin (1981) to view re-
marriage as an incomplete institution. In these and comparable cases a
specific and elaborate operationalization of the concept of ambivalence is
needed if one wants to go beyond simple plausibility. As a result of these
efforts, one can expect, as mentioned above, at least a higher level of au-
thenticity with regard to the diversities and the dramas of everyday life. One
should also strive for findings, which systematically illuminate the conse-
quences of different levels of awareness and of different strategies in dealing
with ambivalences. Practical interests may lie in the evaluation of thera-
peutic interventions that strive to heighten the awareness of ambivalences
and to establish specific ways of dealing with them.
4.4. Ambivalences at Turning Points and Transitions
The notion of turning points refers to phenomena, experiences and actions
where the awareness of ambivalences may be especially promising and
where the interplay between generations and the life course is quite perti-
nent. A turning point may be understood, metaphorically speaking, as an
interruption in a person’s development. It coincides with the necessity, or at
least the possibility, to reflect upon personal relationships and the commit-
ments they involve. Changes may be requested and importance attributed to
particular relationships, or persons may be asked to restructure their rela-
tionships. New commitments and obligations may emerge that compete with
KURT LUSCHER110
ongoing concerns and ties. In reality, ‘‘turning points’’ may extend a certain
period of orientation and search, hence it is also appropriate to speak of
transitions. They can be seen as fields of action entailing an accentuated
experience of ambivalences.
Perhaps the most obvious turning point at the intersection between in-
tergenerational relationships, the life course and the social context is the
transition to parenthood. This appears in many ways and, not surprisingly,
there is still no comprehensive theory of generative behavior and decision-
making. Several attempts, however, refer to the notion of ambivalence,
mostly using the word in an everyday meaning. More elaborate studies
along this line point out that decisions are reached only through a lengthy
process that takes the form of oscillations typical of ambivalences. A good
illustration is the phenomenon of late first motherhood (see Engstler &
Luscher, 1991).
The experience of ambivalences (as defined above) is bound to the self and
personal identity. In addition to their search for the subjective meaning of
motherhood, many women are confronted with or exposed to normative
expectations, traditional or progressive, by others who are close to them,
and also by society at large, as represented by subcultures such as religions
and ethnic groups, not to speak of economic pressures and the contempo-
rary organization of the labor market. This topic also illustrates what is
referred to above as the experience of ambivalence ‘‘in the dialogue with
generalized others.’’
An attempt to draw upon the concept of ambivalence and to further
explore its relevance for a typological differentiation of generative behaviors
is offered as part of in an analysis of the Swiss Family Survey (Le Goff,
Sauvain-Dugerdil, Rossier, & Coenen-Huther, 2005). Ambivalence is used
as an alternative to the notion of rational choice in discussing fertility be-
havior in low-fertility countries like Switzerland. It serves as a key concept
to distinguish between four main types of the fertility project: The familialist
subculture, either sequential or simultaneous articulation between labor
market participation and motherhood, and childlessness. Future trends are
discussed in the light of the pressure to change exerted by those women who
experience a high degree of ambivalence between their own life aspirations
and normative expectations, while also possessing high levels of personal
resources.
With regard to motherhood as such, a treatise by Parker, with the sug-
gestive title ‘‘Mother Love, Mother Hate,’’ written from a psychoanalytical
perspective, merits special attention. Parker (1995, p. 6) refers to Melanie
Klein, who ‘‘considered that ambivalence had a positive part to play in
Looking at Ambivalences 111
mental life as a safeguard against hate.’’ Parker adds: ‘‘I want to go further
and claim a specifically creative role for manageable maternal ambivalence.
I suggest that it is in the very anguish of maternal ambivalence itself that a
fruitfulness for mothers and children resides.’’ The major mechanism can be
described as follows: Given the fundamental dichotomy and the awareness
of love and hate, mothers are able even in desperate situations to reactivate
the forces of love. More generally, mothers search continuously, even under
difficult situations, for arrangements that serve the well-being of their chil-
dren. This fundamental ability to cope with ambivalence creatively can be
seen as a genuine cultural and social contribution of mothers to civilization.
Contributions like Parker’s make clear why – and also how – a focus on
ambivalence can be compatible with feminist thinking. This field is sensi-
tized to possible ambivalences in gender relations and to constructive, as
well as destructive, strategies for dealing with them.
Referring to a later phase in the life course, Pillemer and Suitor (2002)
focus on the tension between autonomy and dependence and find that a key
dilemma leading to intergenerational ambivalence is the conflict between the
norm of solidarity with children and the normative expectation that children
will develop independent lives in the case of the so-called ‘‘off-time tran-
sitions’’ – here in the lives of children. As a general finding, the authors
showed, ‘‘that adult children’s failure to achieve and maintain normative
adult statuses and financial independence, and mothers’ developmental
stage predict ambivalent assessments of the relationship. Regression anal-
yses supported these hypotheses and also revealed that the variables pre-
dicting ambivalence differed from those that predicted closeness and
interpersonal stress’’ (Pillemer & Suitor, 2002, p. 602). In particular, height-
ened ambivalence can be anticipated when adult children have not attained
(or maintained) adult statuses. When parents face such unexpected circum-
stances, they are likely to experience mixed emotions involving a desire to
protect and assist the child, as well as disappointment at the child’s situation
and self-doubt regarding parenting. This study, like the one mentioned be-
fore, makes explicit use of the concept of ambivalence. It is not difficult to
imagine other turning points that display preconditions for the experience of
ambivalences, such as occupational choice, or – at the end of a professional
career – the period of retirement. Work in these areas would require – and
could stimulate – further efforts in the conceptualization of ambivalence.
The example suggests viewing non-normative (or even deviant) behavior
as a cause of ambivalence. From a theoretical point of view, there may be a
linkage with the analysis of stigma, such as that of Goffman (1963). In-
terestingly enough, although the latter does not use the term ambivalence,
KURT LUSCHER112
he describes behaviors that can be interpreted as strategies for coping with
ambivalences.
If attention is directed toward specific features of the life course, trauma is
certainly an experience that can generate ambivalences in several ways. Under
the impact of personal and structural violence, the self is threatened, and this
may remain so for a long time, or even lifelong. Thus, the traumatic expe-
rience becomes part of the personality. On one hand, it is so subjective that it
cannot be shared with others, but on the other hand there may be a strong
desire to share one’s experiences, not least of all in the hope of receiving
therapeutic support. This holds true for personal traumatic experiences such
as child abuse. Traumas can also be collective, as in the case of wars. The
Holocaust is a unique case of the experience of collective trauma for which an
extensive body of literature exists (see for example Ludewig-Kedmi, 2001,
2004). The twofold experience of ambivalence in connection with personal
and collective trauma is concisely summarized by Smelser (2004, p. 53) in the
following passage:
One of the peculiarities that have been noticed in connection with acute psychological
traumas is a very strong dual tendency: to avoid and to reliveyAt the ideational level
one main defense is some form of amnesia (numbing, emotional paralysis)y, actual
forgetting, denial, difficulty in recalling, or unwillingness to contemplate or dwell on the
traumatic event. At the same time, the trauma has a way of intruding itself into the mind,
in the form of unwanted thoughts, nightmares and flashbacks. These apparently an-
tagonistic tendencies have presented themselves to some as a paradoxy At the behavi-
oral level, the same double tendency has been observed: A compulsive tendency to avoid
situations that resemble the traumatic scene or remind the victim of it, but at the same
time an equally strong compulsion to repeat the trauma or to relive some aspect of ity
When seeking an analogy at the socio-cultural level, we discover such dual tendencies –
mass forgetting and collective campaigns on the part of groups to downplay or ‘put
behind us’, if not actually to deny a cultural trauma on the one hand, and a compulsive
preoccupation with the event, as well as group efforts to keep it in the public con-
sciousness as a reminder that ‘we must remember’, or ‘lest we forget’, on the other.
A memorial to an eventyhas both reactionsy[we can speak of] the compulsion to
remember and the compulsion to forget.
4.5. Ambivalences Concerning Specific Fields of Action
4.5.1. Caring
The experience of ambivalences may be greater in tasks where tensions and
contradictions cumulate. This is certainly the case in caring. For caregivers,
and in reference to the subjective component of relationships, sympathy and
antipathy are at play, and many caring activities include intimate behaviors
Looking at Ambivalences 113
that may be embarrassing. From an institutional perspective, normative
expectations may exist which juxtapose the commitment of a woman as the
daughter of elderly parents with the duties of husbands and wives. Men, too,
may be burdened in this way, but caring is still considered a primarily female
obligation. These traditional gender ideologies may add to the pressures and
thereby further the likelihood of ambivalences. Seen from the point of view
of the care-receivers, ambivalent feelings and attitudes may exist as well,
since they realize the tensions between insight into apparent dependency and
the wish for independency.
This is the topic of a monograph by Cohler and Grunebaum (1981) which
is cited here as an example of a study that analyzes the phenomenon of
ambivalence without using the concept itself. The authors focus on mother–
adult daughter relationships in four families of Italian Americans. Their
point of departure is the ‘‘paradox in contemporary society where, on the
one hand, it is believed that adults will strive to become both psychologically
and economically autonomous and self-reliant, while, on the other, findings
from systematic investigations of family life show that dependence across
the generations is the typical mode of intergenerational relations, including
the interdependence of very old parents on their middle-aged offspring’’
(ibid., p. 10). In the concrete case, for the mothers, the acceptance of the
daughters commitments are in conflict with the mothers desire to continue
to lead their own lives. The authors describe as an illustrative example the
relationships of one mother (Mrs. Scardoni) and her daughter (Mrs. Russo)
in the following way:
Mrs. Russo’s continuing emotional involvement with her mother is both a source of
support as well as a source of considerable discomfort and strain. Neither Mrs. Scardoni
nor Mrs. Russo can tolerate any disagreement or disharmony, for neither mother nor
daughter can admit to their own mixed feelings. On the one hand, Mrs. Russo is very
dependent on her mother for help with even the most minute aspects of her life, such as
recipes for supper or advice on her problems with her daughter or her husband. On the
other hand, she is afraid that her mother will forget about her if she does not maintain
continual contact. Burdened by her mother’s demand that she and her brother provide
Mrs. Scardoni with the identity that she had never achieved for herself and unable to
derive any sense of security or satisfaction from their relationship, Mrs. Russo feels
frustrated, resentful, and then guilty. Finally, she becomes so distraught that she can
only continue to function by swallowing large doses of the several ‘tranquilizers’ that her
family doctor has prescribed for her. (ibid., p. 120)
A similar study of fathers and sons in later life has been published by
Nydegger and Mitteness (1991). Their analysis also contains colorful de-
scriptions of ambivalences without using the term itself. There are certainly
more studies which contain an implicit and consequently not yet elaborated
KURT LUSCHER114
reference to the idea of ambivalence. It may be worthwhile to reanalyze
them in the light of the emergence of a theory of ambivalences.
In this regard, a secondary analysis of data from the Berlin Aging Study
(BASE) by Lang (2004) merits special attention. It also contains an explicit
connection to the life course approach. Data are available from responses by
adult children (mean age 54.4 years) to a mailed questionnaire on personal
networks and the quality of relationships with parents. Ultimately,
yfour distinct patterns of adult children’s relationship styles towards their parents were
identified based on indicators of support exchange, personal norms and affective
strength: close exchange, resilient giving, strained altruism, and detached distance. The
four relationship styles were associated with motivations for seeking contact with par-
ents and the inconsistency of relationship satisfaction with parents. Each of the four
relationship styles reflects an individual response to the challenges of the filial task in
midlife.
In the interpretation of the author (Lang, 2004, p. 199ff.),
yfour observed styles of adult children’s relationships with their older parents are most
consistent with the assumptions of the heuristic model of intergenerational ambivalence
(Luscher, 1998). According to this model, ambivalence is conceived as an implicit and
underlying structure that may be experienced within any intergenerational relationship.
For example, adult children may respond to ambivalence with detachment from their
parents, referred to as atomization. This response is well reflected in the detached-distant
relationship style of adult children’s attitudes towards their parents. Another prototyp-
ical response described in the heuristic model of ambivalence is captivation, which refers
to feelings of being obligated to take responsibility, while at the same time feeling
strained by such responsibility. This response pattern is well reflected in the strained-
altruistic relationship style of adult children. A third prototypical response to inter-
generational ambivalence according to the heuristic model of ambivalence is the
expression of normatively taking responsibility and close supportive exchanges with the
aged parent. This response pattern may be characterized as solidarity and is best re-
flected in the group of adult children who display a style of close exchange with their
parents characterized by strong emotional closeness and much supportive exchange with
parents. The relationship style of close exchange with parents comes closest to the
concept of family solidarity, at least with respect to the constructs of normative, func-
tional and affective solidarity. Adult children in this group were mostly satisfied with
their relationship to their parents and displayed the strongest level of consistency across
different ratings of satisfaction. A fourth prototypical response pattern refers to eman-
cipation, which involves a pragmatic attitude of keeping an affective distance to one’s
parent while at the same time giving what is needed. Again, this response pattern is
reflected in the group of adult children who displayed a style of resilient giving towards
their parents. Adult children of this group gave much support because they felt obliged
to do so, but also showed relative affective neutrality towards their parents.... Mani-
festations of personal ambivalence as indicated by the degree of inconsistency in ratings
of satisfaction with parents were differently distributed across the four relationship
styles. In particular, the strained-altruistic and the resilient-giving relationship styles
Looking at Ambivalences 115
were found to have the greatest potential for perceptions of ambivalence (i.e. incon-
sistency). Both styles were associated with a basic and strong attitude towards giving
support to one’s parents.
Lorenz-Meyer (2004) has explored, through narratives of young adults in
Germany, the generation of ambivalences and strategies of dealing with
them in relation to prospective parental care. In her own words, ‘‘the anal-
ysis shows that in contemporary Germany the (anticipated) transition of
parents requiring personal care is perceived as a structurally ambivalent
situation for many adult children that simultaneously values two opposing
courses of actions and leads to decisional ambivalence of children between
personally supporting their parents in old age and placing them in a nursing
home. Participants’ reflections on viable and consensual care arrangements
that can be interpreted as an attempt to deal with decisional ambivalence
involved a multifaceted process of taking stock of (a) the personal
relationship between parents and children, often in comparison with the
relationship between parents and siblings; (b) the living situation of older
parents; (c) the respondent’s own living situation; (d) past family care ar-
rangements; (e) cultural-normative guidelines; (f) care institutions; and
(g) expected commitments of other siblings (and partners).’’ (ibid., p. 246f).
The interviews also show, ‘‘that research participants interpreted ambiva-
lences not just in a biographical, but also in a socio-historic context. Par-
ticipants’ localization of intergenerational positions and relationships in
concrete historical conditions can serve to de-personalize and possibly mit-
igate personal ambivalences’’ (ibid., p. 248).
In the context her analysis, Lorenz-Meyer also focused on points of con-
nection and differentiation with the four strategies of dealing with ambiv-
alence identified in the Konstanz studies. ‘‘Displaying inaction and not
planning for parental care needs, for example, was not considered as con-
tradicting a solidaristic orientation (and could even be interpreted
as ‘‘emancipation’’ in the Konstanz typology, if previous familial care
arrangements were not reproduced and personal contact maintained).
This was a strategy of dealing with ambivalence that was used mainly by
men. The assumption that other siblings, usually a sister, would provide
co-residential care tended to facilitate inaction and mitigate decisional am-
bivalence. Conversely, it was exclusively women (with intermittent employ-
ment) who committed themselves to providing co-residential care (that can
be interpreted as ‘‘solidarity’’ or, if the initiation of alternative arrangements
had failed, as ‘‘captivation’’). Women were also the majority of those who
KURT LUSCHER116
explicitly anticipated accommodating the parent in a home while commit-
ting themselves to complementary emotional care (which can be interpreted
as ‘‘emancipation’’ if elder care had been provided in the family). For both
groups of women the perceived absence of care commitments from other
siblings increased decisional ambivalence. A crucial factor for planning res-
idential (rather than co-residential) care was the availability of material
resources to afford quality care among women (and some men) with more
continuous employment that thereby had a mitigating effect on decisional
ambivalence.’’ (ibid., p. 249).
Lorenz-Meyer distinguishes between multiple, personal and structural
ambivalences that underlie decisional ambivalence in the following way:
� Personal ambivalences refer to the simultaneity of opposing feelings
and orientations such as closeness and distance that came to the fore
when participants imagined co-residential living arrangements with their
parents.� Structural ambivalences refer to the simultaneity of opposing offerings,
directives or guidelines for action inherent in institutional structures, such
as state agencies or social policies.� The notion of multiple ambivalences refers to overlapping personal and
structural ambivalences that constitute multiple sources, rather than a
single cause for decisional ambivalence.
As part of the already-mentioned OASIS project, an extensive compar-
ative study on the care of the elderly and the role of family support systems,
complemented the traditional focus on solidarity with an analysis of am-
bivalences. The authors summarize the results of the quantitative and the
qualitative analysis as follows (Lowenstein & Ogg, 2003, p. 223):
Correspondence analysis of the ten questions relevant to inter-generational conflict,
ambivalence and solidarity resulted in categorizing parent–child relationships into four
distinct styles. Harmonious relationship styles were categorized, for example, by getting
along extremely well but with an acceptance that conflict and ambivalent feelings could
and did occur but without altering the essentially positive relationship experience. Dis-
tant family styles were conversely evidenced by emotional distancing, differences in view
and the experience of conflict and ambivalent feelings in a way which could or did have a
deleterious effect on family relationships. – In the qualitative data, dyads who expe-
rienced their relationships as effective and essentially harmonious tended to identify
ambivalence or conflict as a part of the process of their relationship. Transitions created
by changes in parental health for example, brought about the possibility of negotiating
or redefining roles and responsibilities without impinging on participants’ views of the
overall quality of the relationship.
Looking at Ambivalences 117
4.5.2. Inheritance
If one is searching for phenomena that seem in the light of experience to be
breeding grounds for ambivalences, inheritance is undoubtedly a major
candidate. Thus, we may use this topic as an illustration of how the new
orientation, namely the interest in ambivalences, sheds light, encourages,
stimulates new research interests, close to daily life, and also recalls the
importance of interdisciplinary cooperation. Certainly a core phenomenon
in the field of generations, inheritance has found surprisingly little attention
in the field of social science. This is also true for its relevance for patterns of
life courses, individual lives and personal ties.
The chapter by Plakans (2004) in Pillemer and Luscher (2004) is a good
starting point. The author recalls how important the regulations concerning
inheritance were in the past and how much they could influence the life
courses of the rich, including aristocrats, as well as peasants and artisans.
Major sources of ambivalences can be assumed on a structural level in the
juxtaposition of institutional rules and customs, and the desire of the dona-
tors to express their personal sympathies, or to reward a child (or another
person) for support and attention. Another conflict which most likely in-
duced everyday ambivalences has to be seen in the self-interest of the old in
their role as heads of households, as opposed to the desire of the young to
have a family of their own and to become autonomous. Ambivalences may
also be nourished by the rivalries among siblings.
To this Plakans offers concrete illustrations. Ehmer and Gutschner (2000)
confirm the overall fruitfulness of the concept of ambivalence for the study
of inheritance and more generally speaking for the social history of the
family and its implications for personal biographies. They see a major ad-
vantage or function of the concept in that it serves to deconstruct the ide-
alizations that have long dominated family rhetoric.
An attempt to include the concept of ambivalence in a study of present-
day processes of inheritance has been made by Lettke in the Konstanz
Inheritance Survey (Konstanzer Erbschafts Survey – KES), which is rep-
resentative of the German population age 40 and above, using the method
of telephone interviewing. His findings confirm, that about a third of the
subjects refer to ambivalences – a number which seems lower than one
would expect at first glance, and with regard to the usual socio-demographic
variables, those with lower levels of education show a significantly higher
rate of ambivalence. A more detailed analysis reveals that those who
have already received an inheritance are significantly more ambivalent,
which suggests that actual experience turns out to entail more difficulties
than anticipated. Strong correlations exist between the experience of
KURT LUSCHER118
ambivalences and the responses in terms of motivation. The following in-
stances appear to be of significant importance: the intention to reward those
who have provided care, who are especially sympathetic, by whom one
wants to be remembered and with whom one shares common convictions
and beliefs. Ambivalences also arise if a person wants to support children
who have a family of their own and those who are in need. More generally,
ambivalences seem to increase if the testator has reasons to deviate from the
rules stipulated by the law and by a general societal idea of equity. With
regard to the dimensions of the module suggested above, inheritance seems
to be a field of action where the tensions between the subjective or personal
and the institutional dimensions seem of particular relevance.
5. OUTLOOK
In this section, I offer some proposals for a greater rapprochement between
the study of intergenerational relationships and the study of the life course,
especially with regard to its institutional embeddedness. Such an orientation
refers back to the older issue of the interplay between biology and culture,
which is fundamental both to the concept of human development and
generational succession. A major focus is the understanding of personal
identity and the self.
In this connection, and also as an answer to recent calls for more there in
the field of generational studies, the concept of ambivalence is appropriate.
This is appropriate and attractive for at least three reasons. First, this con-
cept too is relevant for a deeper understanding of personal identity in a non-
metaphysical and non-normative way. Second, if used in the sense of Mead
(1938), identity development can be understood as advanced by ongoing
dialogues with oneself and with significant others. Third, such dialogues
imply the possible experiences of being torn in two opposed directions and
oscillating between them.
With regard to a life course perspective, ambivalences are presumed to
activate, or at least to stimulate, the human potential for action in social
structures. In other words, dealing with ambivalence requires ‘‘agency.’’
Thus, it is fruitful to view ambivalences as ‘‘neutral,’’ i.e. as possible pre-
conditions for acting. Research on ambivalence should therefore focus on
awareness and coping. We can hypothesize, first, that people cope with
ambivalence in more or less competent, productive, or even creative ways.
Second, the deliberate construction of ambivalences can be a strategy in
shaping and organizing social interactions. Third, the personal experience of
Looking at Ambivalences 119
ambivalences depends on aspects of interactions and social structures
and on the embeddedness of ambivalences in role models and collective
identities.
We can expect that ambivalence will be especially manifest at ‘‘turning
points’’ and that it will likewise be apparent throughout the biographical
histories of the relationships between parents and their children. Ultimately,
dealing with ambivalences can be conceptualized as a ‘‘meta-task’’ of the
personal and social organization of intergenerational relations (and other
kinds of social relations) over the life course (and vice versa).
In addition, we may hypothesize (beyond the existing frameworks) that
ambivalences, in a life course perspective, may be experienced in introspec-
tion (‘‘inner dialogues’’), as suggested by the idea of ‘‘life review’’ or ‘‘life
reflection’’ (Staudinger, 2001). They may be experienced (and have to be
dealt with) in social relationships (‘‘dialogues with others’’). Finally, one
may consider the impact of generational politics (and politics in general) as
creating conditions that can generate ambivalences (‘‘dialogues with gen-
eralized others’’). Recalling the frame of reference presented at the PaVie
Colloquium Lausanne, we may ask where the experience of ambivalences
can be expected to occur and where we may discover specific strategies of
coping. I offer the following overview:
Topics Perspective of Subjects Perspective of
Researcher/Structures
Trajectory Life reviews Socio-biological
foundations
Conflict nature/nurture
Stage Reproductive behavior Stages of development
Transitions Leaving home Developmental tasks
Retirement Generativity (Erikson)
Events/tasks/
roles
Caring
Grandparenthood
Inheritance (multiple sense)
Trauma
Newly introduced concepts also engage us to adopt a new perspective in
examining existing theories and their interconnections. In this regard, fur-
ther explorations within the field of intergenerational relationships, as well
KURT LUSCHER120
as the field of life course analysis, may well be undertaken in regard, for
instance, to Erikson’s well-known theory of identity. His schema of eight
stages in the development of identity can certainly be read as a sequence of
dilemmas with ambivalent qualities. However, Erikson’s theory would have
to be linked systematically to descriptions of conduct, social relationships
and roles and their possible relevance for the emergence of ambivalence.
Another bridge can be built to the theory of generativity. In a recent, very
concise summary of its substance by McAdams and Logan (2004), at least
the second proposition points to a logical structure of the concept which
comes close to ambivalence: ‘‘Generativity may spring from desires that are
both selfless and selfish’’ (p. 18).
If we want to strive for a closer integration, we should be aware that the
focus here has been on relationships. This focus may be welcome in studies
of the life course. The linkages between lives merit greater attention. Quite
obviously, this draws attention to the dynamics of interpersonal relation-
ships. The idea of ambivalence, as obvious as it may be in the case of
intergenerational relationships, can certainly be enlightening for other per-
sonal relationships, such those between partners or husband and wives,
siblings and even friends and comrades. Their dynamics over a life course
may be quite meaningful and consequential.
Despite the importance of the concept in relationships, a self-critical ob-
servation may well be appropriate. At the present stage of the development
of the ambivalence perspective, concern with the consequences of ambiv-
alent experiences is unexplored. The distinction of four different modes in
dealing with ambivalences may well be a first step. Yet, more work is need-
ed. As one direction, I would like to offer the following argumentation. The
experience of ambivalences – it has been said – should be seen as relevant for
the development of the self or personal identity. (This connection is also
useful to distinguish ambivalences from trivial experiences of tensions and
choices in daily life). Within the framework of a theory of social action, the
reference to the self or personal identity implies a close connection to the
concept of agency, insofar as it may be understood as the locus of action and
of control (see also the chapter by Marshall, in this volume). Therefore, we
should pay greater attention, on one side, to the extent to which the pos-
sibility (or the inability) to control one’s own behavior with regard to others,
and therefore to shape relationships, is a source of ambivalences, and in
what ways the mastering of ambivalences goes together with the exercise of
power. Preliminary considerations along this line have been presented
by Connidis and McMullin (2002). Other behavioral consequences of
dealing with ambivalences may be considered as well. In short, a closer
Looking at Ambivalences 121
interconnection between the study of ambivalences, agency and social con-
trol seems desirable and promising.
I would like to finish on a more general note by referring to Smelser (1998,
p. 13). Exploring the deeper meanings of the ambivalent, he states that, ‘‘we
are dealing with a fundamental existential dilemma in the human condition.
It is communicated in various dichotomies – freedom versus constraint,
independence versus dependence, autonomy versus dependence, maturity
versus infancy, and more – but ever the dichotomy, the dilemma appears to
be insoluble.’’ In a time, when professional and even economic interests play
a major role in the enterprise called ‘‘social science,’’ a reminder of some
basic humanistic issues may well be appropriate – not least with regard to
the question of how we organize and can organize our lives over the life
course and master – as chances and as burdens – the ambivalences occurring
in social relationships.
NOTES
1. Note the deliberate formulation as a general heuristic hypothesis: It is notsuggested that intergenerational relationships are per se ambivalent, or that theyalways require dealing with ambivalences, but several reasons and observations, asshown below, speak for the assumption that this may often be the case.2. From a systematic point of view, this reference to literature implies an impor-
tant insight: Insofar as fictional works are or can be seen as constructions of im-agined worlds, one also may see the ambivalences as deliberately constructed. Thismay be done on the assumption that these ambivalences are also experienced byreaders or viewers in their personal lives. The deliberate creation of ambivalences isalso used as a technique in certain psychotherapeutic methods. e.g the so-called‘‘paradoxical intervention’’3. For a more detailed, yet still preliminary overview see Luscher, 2004. Some
references to the role of the concept in different discourses can be found in Smelser(1998). The reception of the concept in psychotherapy is outlined in Otscheret (1988)and Knellessen (1978).4. Thus, we could also say that the concept of ambivalence refers to ‘‘decision-
making as a process’’ for which the metaphor of ‘‘oscillation’’ seems quite appro-priate, or as an alternative image, a ‘‘tug-of-war.’’5. Due to spatial limitations, I will focus only on the broad outlines. For a full
presentation, see Luscher & Lettke, 2004. The research instruments developedat Konstanz, in partial cooperation with Karl Pillemer, are available in English andGerman under: http://www.uni-konstanz.de/FuF/SozWiss/fg-soz/ag-fam/famsoz-i.html6. Here, a note about the methodological status of a diagram may be in order.
Following an idea by Bogen & Thurlemann (2003), diagrams represent a uniquecategory of ‘‘text,’’ which stems from the combination of words and graphics. Due toa certain degree of ambiguity and of openness, this kind of representation encouragesfurther interpretations and can thus serve as a means to develop new ideas and evenhypotheses.
KURT LUSCHER122
7. This example is taken from the questionnaires used in the Konstanz studies, seefootnote 5.8. There is a parallel to the finding of Pyke and Bengston (1996). In a qualitative
research project on family elder-care, they coined the concept of ‘‘overcare,’’ definedas care exceeding recipients’ actual needs which thus may have negative conse-quences, both relational and developmental. Close-knit networks may not alwaysfacilitate parent–child relationships, especially when the expectations of parents andother network members about the child are inconsistent (Belsky, 1984), or whennetwork members are perceived by parents as competitors rather than as supportersin the parenting process (Robertson, Elder, Skinner, & Conger, 1991). It might wellbe worthwhile to reanalyze these studies in the light of the emerging theory ofintergenerational ambivalence.
ACKNOWLEDGMENT
I would like to thank James Stuart Brice for support in stylistic and editorial
matters, and Denise Ruttinger for general assistance.
REFERENCES
Attias-Donfut, C. (1995). Le double circuit des transmissions [The twofold circle of transmis-
sions]. In: C. Attias-Donfut (Ed.), Les solidarites entre generations. Veillesse, familles,
etat [Solidarities between generations. Aging, families and the state] (pp. 41–81). Paris:
Edition Nathan.
Bauman, Z. (1997). Flaneure, Spieler und Touristen. Essays zu postmodernen Lebensformen
[Flaneur, gamblers and tourists. Essays about postmodern life-forms]. Hamburg:
Hamburg Edition.
Bawin-Legros, B., Gauthier, A., & Strassen, J.-F. (1995). Les limites de l’entreaide intergenera-
tionelle [The limits of intergenerational support]. In: C. Attias-Donfut (Ed.), Les sol-
idarites entre generations. Vieillesse, familles, etat [Solidarities between generations]
(pp. 117–130). Paris: Nathan.
Belsky, J. (1984). The determinants of parenting. A process model. Child Development, 55,
83–96.
Bengtson, V. L., & Harootyan, R. A. (1994). Generational linkages and implications for public
policy. In: K. Kronebusch, L. Lawton, M. Schlesinger, M. Silverstein & R. E. Vorek
(Eds), Intergenerational linkages (pp. 210–233). New York: Springer.
Bengtson, V. L., Giarusso, R., Mabry, B., & Silverstein, M. (2002). Solidarity, conflict, and
ambivalence. Complementary or competing perspectives on intergenerational relation-
ships? Journal of Marriage and the Family, 64, 568–576.
Bleuler, E. (1910). Zur Theorie des schizophrenen Negativismus [About the theory of
schicophrenic negativsm]. Psychiatrisch-Neurologische Wochenschrift, 18, 171–176; 19,
184–187; 20, 189–191; 21, 195–198.
Bleuler, E. (1914). Die Ambivalenz [Ambivalence]. In: U. Zurich (Ed.), Festgabe zur Einweihung
der Neubauten (pp. 95–106). Zurich: Schulthess & Co.
Looking at Ambivalences 123
Bogen, S., & Thurlemann, F. (2003). Jenseits der Opposition von Text und Bild. Uberlegungen
zur Theorie des Diagramms und des Diagrammatischen [Beyond the opposition of text
and image. Consideration for a theory of the diagram]. In: A. Patschovsky (Ed.), Die
Bildwelt der Diagramme Joachims von Fiore [The images of the diagrams of Joachim von
Fiore] (pp. 1–22). Ostfildern: Thorbecke.
Brand, C. (2004). Generationenbeziehungen in Familien mit psychisch Kranken. Diplomarbeit
[Intergenerational relationships between familys with mentally disordered persons].
Konstanz: Universitat Konstanz.
Brannen, J. (2003). Towards a typology of intergenerational relations. Continuities and change
in families. Sociological Research Online, 8, 1–17.
Burkhardt, A. (2005). Generationenambivalenzen in Familien mit einem psychisch kranken
erwachsenen Kind. Dissertation [Intergenerational ambivalences between familys with a
mentally disordered adult child]. Konstanz: Universitat Konstanz.
Cherlin, A. J. (1981).Marriage, divorce, remarriage. Cambridge, MA: Harvard University Press.
Coenen-Huther, J., Kellerhals, J., & von Allmen, M. (1994). Les reseaux de solidarite dans la
famille [Networks of solidarity in the family]. Lausanne: Realites Sociales.
Cohler, B. J., & Grunebaum, H. U. (1981). Mothers, grandmothers and daughters. Personality
and childcare in three-generation families. New York: Wiley.
Cohler, B. J. (2004). The experience of ambivalence within the family. Young adults ‘‘coming
out’’ gay or lesbian and their parents. In: K. Pillemer & K. Luscher (Eds), Inter-
generational ambivalences. New perspectives on parent–child relations in later life
(pp. 255–284). Oxford: Elsevier Science Ltd.
Connidis, I. A. (2001). The intergenerational ties of gay and lesbian adults and step kin.
A conceptual discussion. Paper presented at the 54th meeting of the Gerontological Society
of America. Chicago, IL: Gerontological Society of America.
Connidis, I. A., & McMullin, J. A. (2002). Sociological ambivalence and family ties. A critical
perspective. Journal of Marriage and Family, 64, 558–567.
Donati, P. (1995). Quarto rapporto sulla famiglia in Italia [Fourth report on the family in Italy].
Cinisello Balsamo: Edizione San Paolo.
Edmunds, J., & Turner, B. S. (Eds) (2002a). Generational consciousness, narrative and politics.
Lanham, Boulder, New York, Oxford: Rowman & Littlefield Publishers.
Edmunds, J., & Turner, B. S. (2002b). Generations, culture and society. Buckingham: Open
University Press.
Ehmer, J., & Gutschner, P. (2000). Das Alter im Spiegel der Generationen. Historische und
sozialwissenschaftliche Beitrage [Aging in the mirror of generations. Historical and social
science contributions]. Wien: Bohlau.
Engstler, H., & Luscher, K. (1991). Spate erste Mutterschaft. Ein neues biographisches Muster
der Familiengrundung? [Late first motherhood. A new biographical pattern of family
formation?]. Zeitschrift fur Bevolkerungswissenschaft, 17(4), 433–460.
Finch, J., & Mason, J. (1993). Negotiating family responsibilities. London: Routledge.
Fingerman, K. L., & Hay, E. (2004). Intergenerational ambivalence in the context of the
larger social network. In: K. Pillemer & K. Luscher (Eds), Intergenerational ambivalences.
New perspectives on parent–child relations in later life (pp. 133–150). Oxford: Elsevier
Science Ltd.
Freud, S. (1914/1974). Some reflections on schoolboy psychology. In: J. Strachey & A. Freud
(Eds), The standard edition of the complete psychological works of Sigmund Freud,
(Vol. 13, pp. 241–244). London: Hogarth Press.
KURT LUSCHER124
Goffman, I. (1963). Stigma. Notes on the management of spoiled identities. Englewood Cliff:
Prentice-Hall.
Hoch, H., & Luscher, K. (2002). Familie im Recht [The Family in the law. A socio-ecological
approach]. Konstanz: UVK-Verlagsgesellschaft.
Holman, T. B. (1981). The influence of community involvement on marital quality. Journal of
Marriage and Family, 43, 143–149.
Jekeli, I. (2002). Ambivalenz und Ambivalenztoleranz [Ambivalence and tolerance of ambiva-
lence]. Osnabruck: Der Andere Verlag.
Johnson, M. J., & Milardo, R. (1984). Network interference in pair relationships. A social
psychological recasting of Slater’s theory of social regression. Journal of Mariage and
Family, 46, 893–899.
Julien, D., Markman, H. J., Leveille, S., Chartrand, E., & Begin, J. (1994). Networks’ support
and interference with regard to marriage. Disclosure of marital problems to confidants.
Journal of Marriage and Family, 56, 16–31.
Knellessen, O. (1978). Ambivalenz und Doppelbindung. Eine Untersuchung des psychoanalytis-
chen Ambivalenzbegriffes [Ambivalence and double bind. Analysis of the psychoanalytic
concept of ambivalence]. Salzburg: Universitat Salzburg.
Lalive d’Epinay, C., & Bickel, J.-F. (1994). Personnes agees et reseau familial [Old people and
familar networking]. Medecine psychosomatique, (1/94), 12–23.
Lang, F. R. (2004). The filial task in midlife. Ambivalence and the quality of adult children’s
relationships with their old-aged parents. In: K. Pillemer & K. Luscher (Eds), Inter-
generational ambivalences. New perspectives on parent–child relations in later life
(pp. 183–206). New York, Oxford: Elsevier.
Lautmann, R. (1996). Ambivalenzen der Verrechtlichung. Die gleichgeschlechtlichen Partners-
chaften im Gesetzgebungsverfahren [Ambivalences of the law. Gay and lesbian
partnership in the legislative process]. Zeitschrift fur Frauenforschung, 14(4),
121–128.
Leary, T. (1957). Interpersonal diagnosis of personality. New York: The Roland Press Company.
Lee, S. (1998). Generativity and the life course of Martha Graham. In: D. P. McAdams &
E. de St. Aubin (Eds), Generativity and adult development. How and why we care for the
next generation (pp. 429–448). Washington, DC: American Psychological Association.
Le Goff, J.-M., Sauvain-Dugerdil, C., Rossier, C., & Coenen-Huther, J. (Eds) (2005).Maternite
et parcours de vie. L’enfant a-t-il toujours une place dans les projets des femmes en Suisse?
[Maternity and the life course. Is there still a place for the child in the life-projects of Swiss
women?] Bern: Peter Lang.
Lettke, F., & Luscher, K. (2001). Wie ambivalent ‘‘sind’’ familiale Generationenbeziehungen?
[How ambivalent ‘‘are’’ familiar intergenerational relationships?]. In: J. Allmendinger
(Ed.), Gute Gesellschaft? Verhandlungen des 30. Kongresses der Deutschen Gesellschaft fur
Soziologie [Good society? Negotiations of the 30th congress of the German Society for
Sociology] (pp. 519–540). Opladen: Leske+Budrich.
Lettke, F., & Klein, D. (2004). Methodological issues in assessing ambivalences in intergene-
rational relations. In: K. Pillemer & K. Luscher (Eds), Intergenerational ambivalences.
New perspectives on parent–child relations in later life (pp. 85–113). Oxford: Elsevier
Science Ltd.
Littwak, E. (1965). Extended kin relations in a democratic industrial society. In: E. Shanas &
G. Streib (Eds), Social structure and the family (pp. 290–323). Englewood Cliff,
NJ: Prentice-Hall.
Looking at Ambivalences 125
Lorenz-Meyer, D. (2004). The ambivalences of parental care among young German adults. In:
K. Pillemer & K. Luscher (Eds), Intergenerational ambivalences. New perspectives on
parent–child relations in later life (pp. 225–252). Oxford: Elsevier Science Ltd.
Lowenstein, A., & Ogg, J. (2003). OASIS. Old age and autonomy. The role of service systems and
intergenerational family solidarity. Final report. Haifa: University Mimeo.
Ludewig-Kedmi, R. (2001). Opfer und Tater zugleich? Moraldilemmata judischer Funkt-
ionshaftlinge in der Shoah [Victim and suspect at once? Moral dilemmas of jewish prisoners
in the Shoah]. GieXen: Psychosozial-Verlag.
Ludewig-Kedmi, R. (2004). Ambivalenz im Umgang mit der Shoa. Psychologische Perspektiven
von Erzahlgeboten und Erzahlverboten [Ambivalence in exposure with the Shoa. Psy-
chological perspectives of narrative commandments and narrative interdictions]. In:
B. Bannasch & A. Hammer (Eds), Verbot der Bilder. Gebot der Erinnerung. Mediale
Inszenierungen der Schoah [Ban of images. Commandment of memory. Medial staging of
the Shoa] (pp. 1–18). Frankfurt am Main: Campus.
Luscher, K. (1998). A heuristic model for the study of intergenerational ambivalence. Konstanz:
Forschungschwerpunkt ‘‘Gesellschaft und Familie’’, Universitat Konstanz.
Luscher, K. (2004). Conceptualizing and uncovering intergenerational ambivalence. In:
K. Pillemer & K. Luscher (Eds), Intergenerational ambivalences. New perspectives on
parent–child relations in later life (pp. 23–62). Oxford: Elsevier Science Ltd.
Luscher, K., & Lettke, F. (2002). L’ambivalence, une cle pour l’analyse des relations int-
ergenerationelles [Ambivalence, a key to the study of intergenerational relationships].
Retraite et Societe, 35, 140–169.
Luscher, K., & Lettke, F. (2004). Intergenerational ambivalence. Methods, measures, and
results of the Konstanz Study. In: K. Pillemer & K. Luscher (Eds), Intergenerational
ambivalences. New perspectives on parent-child relations in later life (pp. 153–179).
Oxford: Elsevier Science Ltd.
Luscher, K., & Pajung-Bilger, B. (1998). Forcierte Ambivalenzen. Ehescheidung als Heraus-
forderung an die Generationenbeziehungen unter Erwachsenen [Forced ambivalences. Di-
vorce as challenge to intergenerational relationships between adults]. Konstanz:
Universitats-Verlag.
Luscher, K., & Pillemer, K. (1998). Intergenerational ambivalence. A new approach to the
study of parent–child relations in later life. Journal of Marriage and the Family, 60(2),
413–425.
Marshall, V. W., Matthews, S. H., & Rosenthal, C. J. (1993). Elusiveness of family life.
A challenge for the sociology of aging. In: G. L. Pladdox & M. P. Lawton (Eds), Annual
Review of Gerontology and Geriatrics (pp. 39–72). New York: Springer.
Mayer, A.-K., & Filipp, S.-H. (2004). Perzipierte Generativitat alterer Menschen und die Qua-
litat der Eltern-Kind-Beziehung [Perceived Generativity of old people and the quality of
the parent-child-relationship]. Zeitschrift fur Soziologie der Erziehung und Sozialisation,
24, 166–181.
McAdams, D. P., & Logan, R. L. (2004). What is generativity? In: E. de St. Aubin,
D. P. McAdams & T.-C. Kim (Eds), The generative society. Caring for future generations
(pp. 15–31). Washington, DC: American Psychological Association.
Mead, G. H. (1938). The philosophy of the act. Chicago: University of Chicago.
Nydegger, C. N., & Mitteness, L. S. (1991). Fathers and their adult sons and daughters. In:
S. K. Pfeifer & M. B. Sussman (Eds), Families. Intergenerational and generational con-
nections (pp. 249–265). London: The Haworth Press.
KURT LUSCHER126
Otscheret, E. (1988). Ambivalenz. Geschichte und Interpretation der menschlichen Zwiespaltigkeit
[History and interpretation of human ambivalence]. Heidelberg: Roland Asanger.
Parker, R. (1995). Mother love, mother hate. The power of maternal ambivalence. New York:
Basic Books.
Parsons, T. (1942). Age and sex in the social structure of the United States. American Soci-
ological Review, 7(7), 604–616.
Parsons, T. (1949). The social structure of the family. In: R. N. Anshen (Ed.), The family. It0s
function and destiny (pp. 173–201). New York: Harper.
Pillemer, K. (2004). Can’t live with ‘em, can’t life without ‘em. Parents’ ambivalence about their
adult children. In: K. Pillemer & K. Luscher (Eds), Intergenerational ambivalences. New
perspectives on parent–child relations in later life (pp. 115–132). Oxford: Elsevier Science Ltd.
Pillemer, K., & Luscher, K. (Eds). (2004). Intergenerational ambivalences. New perspectives on
parent–child relations in later life. Amsterdam: Elsevier.
Pillemer, K., & Suitor, J. J. (2002). Explaining mothers’ ambivalence toward their adult
children. Journal of Marriage and the Family, 64(3), 602–613.
Plakans, A. (2004). Intergenerational ambivalences in the past. A social-historiacal assessment.
In: K. Pillemer & K. Luscher (Eds), Intergenerational ambivalences. New perspectives on
parent–child relations in later life (pp. 63–82). Oxford: Elsevier Science Ltd.
Pyke, K. D., & Bengtson, V. L. (1996). Caring more or less. Individualistic and collectivist
systems of family eldercare. Journal of Marriage and Family, 58, 379–392.
Rein, M. (1994). Solidarity between generations. A five-country study of the social process of
aging. Wien: Institut fur Hohere Studien.
Reinharz, S. (1986). Loving and hating one’s elders. Twin themes in legend and literature. In:
K. Pillemer & R. S. Wolf (Eds), Elder abuse. Conflict in the family (pp. 25–48). Dover,
MA.: Auburn House.
Roberts, R. E. L., Richards, L. N., & Bengtson, V. L. (1991). Intergenerational solidarity
in families. Untangling the ties that bind. Marriage and Family Review, 16, 11–46.
Robertson, E. B., Elder, G. H., Skinner, M. L., & Conger, R. D. (1991). The costs and benefits
of social support in families. Journal of Marriage and Family, 53, 403–416.
Rossi, A. S., & Rossi, P. H. (1990). Of human bonding. Parent–child relations across the life
course. New York: De Gruyter.
Rudorf, S. (2004). Generationenambivalenzen in Familien mit einem substanzabhangigen
erwachsenen Kind. Diplomarbeit [Intergenerational ambivalences in families with a drug
addicted adult child]. Konstanz: Universitat Konstanz.
Shanas, E., Townsend, P., Weddeburn, D., Friis, H., Milhoj, P., & Stehouwer, J. (1968). Old
people in three industrial societies. New York: Atherton Press.
Smelser, N. J. (1998). The rational and the ambivalent in the social sciences. American
Sociological Review, 63(1), 1–16.
Smelser, N. J. (2004). Psychological trauma and cultural trauma. In: J. C. Alexander,
R. Eyerman, B. Giesen, N. J. Smelser & P. Sztompka (Eds), Cultural trauma and col-
lective identity (pp. 31–59). Berkley, Los Angeles, London: University of California
Press.
Spangler, D. (2002). Ambivalenzen in intergenerationalen Beziehungen. Hochaltrige Mutter und
deren Tochter. Diplomarbeit [Ambivalences in intergenerational relationships. Old mothers
and their daughters]. Berlin: Technische Universitat Berlin.
Staudinger, U. (1989). The study of live review. An approach to the investigation of intellectual
development across the life span. Berlin: Max-Planck-Institut fur Bildungsforschung.
Looking at Ambivalences 127
Staudinger, U. M. (2001). Life reflection. A social-cognitive analysis of life review. Review of
General Psychology, 5, 148–160.
Sussman, M. B. (1959). The isolated nuclear family. Fact or fiction. Social Problems, 6,
333–347.
Szydlik, M. (Ed.) (2000). Lebenslange Solidaritat? Generationenbeziehungen zwischen
erwachsenen Kindern und Eltern [Lifelong solidarity? Intergenerational relationships be-
tween adult children and parents] (Vol. 2). Opladen: Leske+Budrich.
Thompson, M. M., Zanna, M. P., & Griffin, D. W. (1995). Let’s not be indifferent about
(attitudinal) ambivalence. In: R. E. Petty & J. A. Krosnick (Eds), Attitude strength.
Antecedents and consequences (pp. 361–386). Mahwah, NJ: Lawrence Erlbaum.
von Matt, P. (1995). Verkommene Sohne. MiX ratene Tochter. Familiendesaster in der Literatur
[Degenerated sons. Failed doughters. Family disasters in literature]. Munchen: Hanser.
Willson, A. E., Shuey, K. M., & Elder, G. H. J. (2003). Ambivalence in the relationship of adult
children to aging parents and in-laws. Journal of Marriage and Family, 65(4), 1055–1072.
Zima, P. V. (2002). L’ambivalence romanesque. Proust, Kafka, Musil [Romanesque ambivalence.
Proust, Kafka, Musil]. Paris: L’Harmattan.
KURT LUSCHER128
AGENCY AND STRUCTURE IN
EDUCATIONAL ATTAINMENT
AND THE TRANSITION
TO ADULTHOOD$
Jeylan T. Mortimer, Jeremy Staff and Jennifer C. Lee
INTRODUCTION
A macrostructural perspective highlights cultural values, normative timeta-
bles, stratification processes, and institutional career lines as determinants of
the content and pacing of role enactment through the life course, and the
acquisition of corresponding role identities. A social psychological perspec-
tive, in contrast, emphasizes the exercise of human agency as a central causal
force in shaping the life course, including the expression of values and
identities, self-regulative processes, decision-making, and striving to achieve
personal objectives through goal selection, strategic planning and action.
According to the macrostructural perspective, the life course consists of
sets of interrelated trajectories (of education, work, family, community
participation, and so forth) with an individual’s location in each at any
$This research was supported by a grant ‘‘Work Experience and Mental Health: A Panel Study
of Youth,’’ from the National Institute of Chile Health and Human Development (HD44138)
and the National Institute of Mental Health (MH42843).
Towards an Interdisciplinary Perspective on the Life Course
Advances in Life Course Research, Volume 10, 131–153
Copyright r 2005 by Elsevier Ltd.
All rights of reproduction in any form reserved
ISSN: 1040-2608/doi:10.1016/S1040-2608(05)10004-5
131
given stage heavily dependent on positioning in prior phases. For example,
in the US context the socioeconomic status of the family of origin affects
children’s schooling in the very early years (placement in reading groups,
marks received, grade retention), which in turn influences assignment to
middle school and high school ‘‘tracks,’’ the opportunity to take college
preparatory and ‘‘advance placement’’ courses, and consequently, the like-
lihood of entering a selective college, or even any institution of higher ed-
ucation (Entwisle, Alexander, & Olson, 2003). The educational trajectory is,
in turn, closely linked to the trajectory of occupational achievement.
A macrostructural perspective strongly informs sociological models of
intergenerational occupational mobility. In these models, social psycholog-
ical processes involving significant others’ influence and aspirations are
considered essentially as intervening variables, mediating the relations be-
tween socioeconomic origins and destinations. At the extreme, expressed
aspirations and goals for the future may be interpreted as signifying mere
recognition of likely outcomes, given initial starting points, rather than
psychological constructs that have their own motivating and determining
force. The psychological constructs may predict subsequent outcomes, but
for structural reasons alone.
The second social psychological approach, while acknowledging that so-
cial location yields diverse and unequal opportunities and constraints at all
stages of the life course, emphasizes the differential capacities of individuals
to effectively exercise agency or initiative. Important individual differences
enable some to take fuller advantage than others of the opportunities linked
to their structural positions, or to more effectively overcome the constraints
associated with disadvantage. Social scientists who take this point of view
focus on psychological processes that lead to the crystallization of goal
formation and channel individual action in preferred directions. Gecas
(1991) argues that the self-concept itself is a motivating force because per-
sons strive to protect and to enhance their self-esteem, sense of efficacy, and
authenticity, and to validate their valued character identities. Individuals
express their identities most notably through role selection and perform-
ance. Personal constructs that implicate future action, such as ‘‘the possible
self’’ (Markus & Nurius, 1986), ‘‘self-efficacy’’ (Bandura, 1997), ‘‘planful
competence’’ (Clausen, 1993), ‘‘locus of control’’ (Rotter, 1966), and ‘‘au-
thenticity’’ (Gecas, 1991) are of central interest, since these are thought to
guide strategic action. Processes of ‘‘goal selection,’’ ‘‘compensation,’’ ‘‘op-
timization,’’ ‘‘assimilation,’’ ‘‘accommodation,’’ planning, and related proc-
esses (see Wiese, Freund, & Baltes, 2000; Brandstaedter, 1998; Heckhausen,
1999) may be more or less adequate in bringing about the more favorable
JEYLAN T. MORTIMER ET AL.132
life course outcomes. Individual orientations and actions are presumed to
have significant consequences for life course trajectories of socioeconomic
attainment, family formation and stability, psychological well-being, and
physical health. Agentic strategies may be more effective among those in
more favorable structural locations, but individual differences in these ca-
pacities are expected to influence life outcomes, net of the advantages or
disadvantages associated with structural position. According to this point of
view, agentic decision-making and behavior are especially important in late
adolescence and early adulthood because the points of entry to socio-
economic trajectories are established at this time.
These two fundamental paradigms, emphasized by macrosociologists and
social psychologists, are surely not mutually exclusive, since structural and
agentic processes operate in tandem. In fact, their coexistence and their
codetermination of social processes make interdisciplinary collaboration
essential. Greater mutual attention is especially called for between the ‘‘so-
cial structure and personality’’ branch of social psychology, concerned with
the manifestations of social location in enduring individual psychological
orientations and behavior; and the subareas of sociology concerned with the
‘‘micro–macro link.’’ The strong reliance on the rational choice paradigm
among investigators interested in micro–macro analysis has them closer to
economics and political science than to social psychology (House &
Mortimer, 1990). Integration of social psychological and sociological frame-
works is especially well suited to the increasingly prominent life course
perspective (Mortimer & Shanahan, 2003). To fully comprehend the pro-
gression of individual lives, we must not only study the changing historical
context and the variable and shifting structures of institutional
trajectories and the ways these impinge on individual actors (the family
life cycle, occupational careers, organizational career lines, etc.). We must
also consider the ways persons subjectively orient themselves to the oppor-
tunities and constraints that they encounter in particular times and places,
and the strategic actions they devise to reach their goals in particular struc-
tural contexts.
However, in the face of persistent evidence that socioeconomic origin
constrains life chances intra- and inter-generationally, it could be argued
that the expression of agency makes little difference. Do adolescents’ as-
pirations in fact exert a meaningful influence upon their attainments? Or,
alternatively, do their aspirations merely signify recognition, some-
times resignation, with respect to probable future destinations, given well-
understood consequences of placement in class structures and school tracks?
This question, in general, has perplexed scholars of stratification and the life
Agency and Structure in Educational Attainment 133
course for several decades (see, for example, Roberts, 1968, for an early
formulation). Whereas associations between psychological dimensions (val-
ues, efficacy, and identity) at one point in time and social locations at a later
date may be interpreted as providing support for the social psychological
paradigm, the alternative structural interpretation of such relations is like-
wise tenable.
It must be acknowledged at the outset that answers to these questions
must be conditional, and especially, historically specific. That is, the relative
importance of structural and social psychological determination of the life
course is likely to vary across time and place. Shanahan, Elder, and Miech’s
(1997) study illustrates this point very well. The Terman gifted men who
came of age during the Great Depression, facing a quite limited employment
market, showed strong proclivity toward postsecondary enrollment irre-
spective of their prior aspirations. But for a subsequent cohort, entering the
labor force in circumstances of expanding economic opportunity, early ed-
ucational values and aspirations were quite consequential for decisions to
pursue postsecondary schooling. Educational aspirations were thus more
predictive of higher educational attainment for the later cohort who en-
countered the postwar economic boom as they entered adulthood. Thus, the
capacity of personal dispositions and goals to influence subsequent attain-
ment depended on historically variable opportunities for choice.
Our examination of the process of attainment is especially strategic in the
contemporary US context where institutional bridges between school and
work are notably undeveloped (Mortimer & Kruger, 2000). High school
education is focused on college preparation, with little emphasis on voca-
tional certification, apprenticeship programs, or other structural connec-
tions between schools and employers. Young people develop their own
strategies in positioning themselves for the labor market. When the time
comes to find a full-time job, they must rely largely on their own resources,
friends, and relatives. Of course, those who come from more advantaged
families have greater resources at their disposal, which help them to com-
plete college and compete successfully for highly coveted professional and
managerial jobs. But even for these youths, pathways to college completion
and high levels of occupational attainment are by no means transparent or
certain. Youths’ efforts in school and their work behavior reveal two major
strategies of human capital acquisition, the first, more successful one,
through education; the second, constituting the more rapid route to full-
time employment, through early part-time work (Mortimer, 2003).
Investigations of the process of attainment have generally taken the
strategy of linking personal dispositions to outcomes measured at a much
JEYLAN T. MORTIMER ET AL.134
later date (see Mortimer, 1996, for a review). For example, researchers have
predicted the features of workers’ jobs by their occupational reward values,
articulated as long as a decade prior (Mortimer & Lorence, 1979; Lindsay &
Knox, 1984). In the classic status attainment model, educational and oc-
cupational aspirations measured during the senior year of high school pre-
dict first jobs and subsequent jobs, attainments that often occur many years
later. Clausen (1993) reports that adolescents who were more ‘‘planfully
competent’’ had more satisfying occupational careers, more stable marriag-
es, and generally more gratifying experiences throughout their lives. Given
this general research strategy, the processes linking the personal orientat-
ions, measured at Point A, and the outcomes, measured at Point B, are
largely unknown.
In this paper, we argue that attention could fruitfully be directed to what
is inside this ‘‘black box,’’ to the processes, mechanisms, and behaviors
between Points A and B. We contend that researchers should examine the
actions that may express agency, actions that connect earlier psychological
orientations, such as goals and values, and subsequent attainments in the life
course. What are the immediate, day-to-day behaviors, provisional goals,
and shorter-term outcomes that link the widely spaced orientations and
outcomes that are the subject of most prior investigations? Moreover, we
argue that psychological orientations provide clearer links to agentic proc-
esses, and are more likely to be predictive of the attainments and conditions
of life that are of interest to social scientists, if they are domain specific. In
support of this claim, we have presented evidence elsewhere that economic
efficacy has stronger predictive power with respect to educational plans and
postsecondary school attendance than global efficacy (Grabowski, Call, &
Mortimer, 2001).
We draw on findings from the Youth Development Study (YDS), a
15-year prospective study of work experience and the transition to adult-
hood, to illustrate our approach. These highlight the interrelations of ed-
ucational orientations, patterns of employment during high school, and
educational attainment; as well as the ways adolescent future orientations
influence youth’s multifaceted pathways of transition to adulthood.
THE YOUTH DEVELOPMENT STUDY
The Youth Development Study (YDS) is the only long-term longitudinal
study designed to address the processes of structure, agency, development,
and attainment that surround early work experiences in adolescence.
Agency and Structure in Educational Attainment 135
The study began with a randomly chosen community sample of ninth grad-
ers enrolled in the St. Paul Public School District in Minnesota;1 1,010
students agreed to participate, and 1,000 filled out surveys in the first data
collection in 1988. United States Census data indicate that this site is quite
comparable to the nation as a whole with respect to economic and social
indicators (Mortimer, 2003).
Most of the analyses reported here use data collected over the years from
1988, when the respondents were in the first year of high school (ninth
grade), mostly at the age of 14 and 15; through 2000, when they were 26–27
years old. This panel study is still in process; the most recently collected data
were obtained in 2005. Extensive tracking and follow-up procedures during
each wave of data collection ensured that students who dropped out of
school after ninth grade, or moved to another school district, continued to
be followed. Data for the post-high school period, from 1992 to 2002, were
collected via mailed surveys. The surveys contained large batteries of ques-
tions focused on work experiences in several domains – in the family, school,
and neighborhood, as well as in formal work establishments. Extensive se-
ries of questions addressed the respondents’ psychological engagement, val-
ues, goals, and future orientations with respect to education, work, family,
and community domains, as well as key activities in each of these spheres.
As such, this study is well suited to examine the interrelations of structure
and agency as youths make their transition to adulthood.
Seventy-six percent of the initial panel remained through the 2000 data
collection. The retained sample was somewhat more advantaged than the
initial sample in terms of socioeconomic background and family compo-
sition (favoring the two-parent family), and retention was more likely for
females than males. The distributions of demographic characteristics across
the initial 1988 and 2000 samples, however, were substantially the same.
EDUCATIONAL ORIENTATIONS AS PREDICTORS
OF EARLY WORK INVESTMENT
The YDS enables assessment of lengthy pathways of attainment from
orientations toward schooling at the time of entry to high school, to patterns
of work investment during high school, to young adult educational and
occupational trajectories, and to 4-year college degree receipt.
Let us first consider adolescent educational achievement and engagement
at the age of 14 and 15, as predictive of very early labor force involvement
JEYLAN T. MORTIMER ET AL.136
during the succeeding 3 years of high school. Most prior studies of ado-
lescent employment rely on the number of hours worked at a given time of
observation, or over a period of months or years. Our more nuanced meas-
ure attempts to capture more long-term strategic involvement in the labor
force, as it includes employment duration, measured in months, as well as
intensity, or the average hours of work during the full period of employ-
ment. Each was dichotomized: months, so as to distinguish near-continuous
employment from much shorter investment in work; and hours, so as to
separate those who on average work more than 20 hours per week from
those who work more moderately, limiting their hours to 20 or fewer. The
20-hour mark is widely considered the point at which employment becomes
‘‘excessive’’ for in-school American youth, interfering with other activities
and increasing the likelihood of problem behaviors (Committee on the
Health and Safety Implications of Child Labor, 1998).
Cross-classifying the two dimensions yields five work patterns during the
10–12th grades of high school: (1) the most invested, high duration/high
intensity workers; (2) sporadic, low duration/high intensity workers; (3)
occasional, low duration/low intensity workers; (4) steady, high duration/
low intensity workers; and (5) non-workers. Table 1 shows the distribution
Table 1. Patterns of Labor Force Participation During High School
by Gender.
Percentage Distribution Mean Months of
Work
Mean Hours of
Work
Total Boys Girls Boys Girls Boys Girls
Not working 7.0 9.9 4.6 0.0 0.0 0 0
Occasional
Low duration-low
intensity
23.7 23.2 24.1 9.8 11.7 578 650
Sporadic
Low duration-high
intensity
18.4 23.2 14.3 10.4 11.8 1,216 1,376
Steady
High duration-low
intensity
24.9 18.2 30.6 22.0 22.0 1,263 1,328
Most Invested
High duration-high
intensity
26.0 25.6 26.4 21.9 22.2 2,678 2,587
Total 100.0 100.0 100.0
N 887 406 481
Agency and Structure in Educational Attainment 137
of these patterns of work investment. Occasional, steady, and most invested
workers are about one quarter each of the total; sporadic workers constitute
18%. The fact that only 7% of the panel reported no work experience at all
demonstrates the high prevalence of part-time work experience, while school
is in session, in the American context. Combining school and work is a near-
universal experience in adolescent life. There are some notable gender dif-
ferences; boys are more likely than girls to be sporadic workers and to be
non-workers; girls are more likely than boys to be steady workers. For boys
and girls in the same investment category, however, mean months and hours
of work are quite similar. That is, high duration workers are employed
approximately 22 of the 24 months of observation and low duration workers
accrue only about half that number.
From an agentic perspective, adolescent work may be conceived as in-
strumental action directed to enhancing future prospects for socioeconomic
attainment (and other goals). Those who expect that they will accrue human
capital mainly through their efforts in the educational system would likely
downplay employment during high school. But those who think that they
have limited chances for academic success, especially with respect to higher
education, would likely seek alternative means of positioning themselves for
adult full-time work. Disinterest and poor performance in school, as well as
structural disadvantage (i.e., low socioeconomic status of origin) would
likely foster attempts to acquire human capital through early intensive em-
ployment. If educational achievement has rather low priority in the ado-
lescent peer culture, as Coleman (1961) suggested, one might also expect
those who are more highly oriented to their peers to be less engaged in
school and more attracted to work. School misconduct constitutes another
indication of disengagement and difficulty in the educational realm.
In prior investigations, we found that adolescents are not randomly se-
lected (or allocated) to the work investment patterns (Mortimer, 2003;
Mortimer, Oesterle, & Staff, 2003). For example, high levels of ‘‘educational
promise’’ in the ninth grade increased the likelihood that youth would pur-
sue steady or occasional patterns of employment, rather than more intensive
work investment, during the following 24 months of high school. Educa-
tional promise referred to the ninth graders’ level of engagement and success
in school: it included indicators of grade point average, academic self-esteem
(the perception of self as intelligent, a good reader, and high in school
ability), educational plans for the future, and intrinsic motivation toward
schoolwork. High-promise youths scored above the median on at least three
of these measures. Youths who pursued the ‘‘steady’’ work pattern during
high school were particularly advantaged vis-a-vis the educational realm.
JEYLAN T. MORTIMER ET AL.138
They had high educational promise in the ninth grade, were unlikely to get
in trouble at school, and had relatively low levels of engagement with their
peers. Occasional workers were similarly characterized by high educational
promise and low school misconduct.
What is it about steady and occasional work that appears to attract young
people who are more strongly engaged in the educational enterprise, who
consider their classes more meaningful and interesting to them, and whose
early efforts in the educational realm (as indicated by grade point average)
are more successful? Surely, the ‘‘steady’’ pattern of working is the one that
is the most ‘‘balanced,’’ as it involves high continuity of employment while
at the same time, because hours of work are limited, permits simultaneous
investment in homework and other school activities. Interestingly, youth
who pursue this pattern are especially likely to say that they had sought
employment ‘‘to save for my future education.’’ Our further analyses show
that such ‘‘balanced’’ workers pursue multifaceted time use strategies
throughout high school, indicating substantial investment in paid work,
schoolwork, extracurricular activities, the peer group, and the family
(Mortimer, 2003; Shanahan & Flaherty, 2001).
Further indicative of agentic action, prior educational engagement not
only influences the pattern of investment in work; it also is linked to the
quality of work that is subsequently obtained. The features of their jobs
suggest that teenagers with limited proclivity toward school seek human
capital acquisition through employment. For example, low grade point av-
erage in the ninth grade is associated with greater advancement opportunity,
higher earnings, more learning opportunities, and a perception that work
confers greater status amongst one’s peers. Those youths who had lower
academic self-esteem early on had higher earnings in their high school jobs
and perceived greater work-derived status amongst their peers (Mortimer,
2003, p. 132).
Clearly, a set of trade-offs between school and work is revealed. Young
people who are more successful in school invest in work in a manner that
yields less in the way of extrinsic and intrinsic occupational rewards, but
enables greater investment in school and the capacity to pursue a well-
rounded, multifaceted, and achievement-oriented adolescent life style. In
contrast, youths who perform more poorly in this domain use their time so
as to obtain more learning opportunities in the workplace, greater imme-
diate economic reward, and higher status in the eyes of their peers (likely
comprised of others who have similar orientations to school and work as
they do themselves). Assessment of work investment and work quality after
the ninth grade tells much about how the expression of academic interests,
Agency and Structure in Educational Attainment 139
achievements, and goals upon entry to high school affect immediately en-
suing work-related behaviors that have quite divergent consequences for
socioeconomic attainments during the post-high school years.
WORK INVESTMENT AND TRAJECTORIES OF
SOCIOECONOMIC ACHIEVEMENT
A growth curve analysis applied to annual survey data collected in the years
following high school (1991–2000) has shown that youths who pursued less
intensive, steady and occasional work patterns during high school make
greater investments in higher education in subsequent years. That is, they
accrue more months of higher education during the first year following high
school, with background and socioeconomic origin controlled, and they
manifest fewer months of full-time work (Mortimer et al., 2003, p. 451).
High intensity work patterns were associated with lower initial levels of
educational investment, and also with greater investment in full-time work.
Furthermore, patterns of growth in months of education and months of
full-time work for the work investment groups confirm the merit of thinking
of these behavioral patterns in terms of strategic action. Low-intensity high
school workers, starting from a higher initial educational investment level,
have steeper losses in months of education during the 9 years following high
school. They exhibit more pronounced increases in full-time work. In con-
trast, high-intensity workers during high school are more attached to the
labor force during the first year after high school, and manifest a less steep
increase in months of full-time work thereafter. Lower intensity (steady and
occasional) workers start off with fewer mean months of full-time work. By
1998, however, they have caught up to the more intensively employed stu-
dents during high school. Their gains in full-time work are steeper than
those of their more intensively employed high school peers.
To what extent do these distinct patterns of time use following high
school-investment in higher education vs. full-time work – exhibited by the
teenage work investment groups-yield differential achievement? We can now
link early educational orientations, high school work patterns, and educa-
tional attainment. We have examined the youth’s capacity to achieve a
prominent educational goal, the BA (bachelor of arts or sciences) degree
(Mortimer, 2003; Mortimer et al., 2003). Consistent with classical models of
status attainment, we found that early school performance and aspirations
(indicated by the educational promise variable in the ninth grade), as well as
JEYLAN T. MORTIMER ET AL.140
parents’ education, were linked with the receipt of a BA degree or higher in
young adulthood.
However, employment in a ‘‘steady’’ pattern of work during high school,
in comparison to more intensive involvement, also conferred a net advan-
tage in educational attainment in young adulthood, independent of socio-
economic background, educational promise, school misconduct, and other
predictors. Based on the most recent wave of data collection (2002), Fig. 1
displays the percentage of respondents that currently have obtained at least
a BA degree based on their prior engagement in work during the high school
period. Forty-four percent of young people in the steady work pattern
during high school had earned a BA degree by 2002, which was much higher
than the 17% of the ‘‘most invested’’ and 12% of the ‘‘sporadic’’ workers,
but also higher than that of the ‘‘occasional’’ and non-workers (33%).
Although the high school patterns of work are influenced by prior socio-
economic background and school performance, defined by gender, race,
parental education, and early educational promise, even when these and
0%
5%
10%
15%
20%
25%
30%
35%
40%
45%
50%
No Work Steady Occasional Most Invested Sporadic Total
Fig. 1. Receipt of 4-Year BA/BS degree (2002) by High School Work
Investment Categories.
Agency and Structure in Educational Attainment 141
other relevant factors are controlled, teenage work patterns are found to
have independent influence on postsecondary educational achievement. The
‘‘steady’’ work pattern is found to be particularly salutary. Why do steady
workers during high school have the advantage in postsecondary educa-
tional attainment? We find evidence that subsequent work patterns are
critical. More recent investigation by YDS researchers (Mortimer & Staff,
2004) demonstrates that the work patterns the young people pursue in the
years immediately following the high school period appear to mediate the
relationship between the ‘‘steady’’ work pattern in high school and early
adult educational attainment. In Fig. 2(a) and (b), we show the average
months of higher education, full- and part-time work in the years right after
high school for the ‘‘steady’’ and ‘‘most invested’’ workers.
As shown in Fig. 2(a), the ‘‘steady’’ high school workers averaged nearly
identical months of postsecondary school attendance and part-time work
(between 7–8 months) in the 4 years immediately following the high school
period. Approximately 4 years after the end of high school, months of
school attendance and part-time work began to decline, while months of
full-time work increased rapidly. Fig. 2(b) highlights a very different pattern
of work and school involvement during the transition to young adulthood
for the ‘‘most invested’’ high school workers. For these youths, investment
in part-time work and school is rapidly replaced with increasing involve-
ment in full-time work. Despite the predictive power of work involvement
during high school with respect to educational attainment in young adult-
hood, we find that the inclusion of subsequent work patterns in this key
period of postsecondary education investment renders prior associations
insignificant (Mortimer & Staff, 2004). Especially for the ‘‘steady’’ workers,
the effective combination of working and studying is a familiar pattern by
the time they enter college. Since most college students have to maintain
employment to at least partially assume their educational (and/or living)
expenses, the time-use strategies these students learned earlier – of balancing
school and work – may enable them to continue combining school and work
roles without compromising their educational goals.
We thus find evidence of agency in the selection to work patterns, in
patterns of educational exploration and investment during high school, and
in the determination of future educational and work trajectories. Our strat-
egy of assessing domain-specific achievement-relevant indicators (e.g., eco-
nomic self-efficacy, educational promise) and achievement-relevant
behaviors closely following (i.e., patterns of labor force investment during
high school, steps taken to go to college) have yielded high returns with
respect to the understanding of trajectories of school enrollment and
JEYLAN T. MORTIMER ET AL.142
0
(a)
(b)
1
2
3
4
5
6
7
8
9
10
19 20 21 22 23 24 25 26
Age
School
Pt-work
Ft-work
0
1
2
3
4
5
6
7
8
9
10
19 20 21 22 23 24 25 26
Age
School
Pt-work
Ft-work
Fig. 2. Average Months of Higher Education, Ft-Work, and Pt-Work in the Years
Following the High School Period for the (a)’’Steady’’ High School Workers, and
(b) ‘‘Most Invested’’ High School Workers.
Agency and Structure in Educational Attainment 143
full-time employment (Grabowski et al., 2001), and eventual BA receipt.
While we cannot fully rule out structural explanations, our analyses con-
sistently show that agency-relevant psychological precursors have signifi-
cant effects on immediate achievement-relevant behaviors, and that these
behavioral patterns are associated with educational attainment, net of
several indicators of socioeconomic origin.
AGENCY AND PATHWAYS OF TRANSITION
TO ADULTHOOD
The collaborative work of the YDS team is now moving to a more general,
and also more complex objective – to assess how diverse family and
achievement-related attitudes and plans, measured in high school, influence
successive configurations of roles in the succeeding years that constitute
latent pathways of transition to adulthood (see Macmillan & Eliason, 2003).
These pathways encompass not only education and work roles, but also
markers of adult family formation (leaving the parental home, marriage,
and parenthood). Thus, our earlier treatment of attitudes toward school as
precursors of work investment is expanding to consider further agentic
motivations and expectations related to the family and intimate relations.
Our earlier focus on trajectories of educational attainment and employment
is now also supplemented by consideration of multifaceted, interrelated
trajectories of educational, work, and family behavior, constituting diver-
gent pathways of transition to adulthood.
We use a two-staged latent class analysis set forth by Macmillan and
Eliason (2003; see also Clogg, 1995) to empirically investigate the simulta-
neous interplay among these social roles in the transition to young adult-
hood. This method is especially germane to a life course perspective, in that
it enables the identification of role configurations (and how they change or
remain stable over time) and the life paths that link these role configurations
through stages of the life course.2
Fig. 3(a)–(c) depicts the role trajectories within three general life paths.
Life path I (probability in sampled population ¼ 0.165), shown in Fig. 3(a),
characterizes a precocious or hastened transition to adulthood. During
adolescence, individuals in life path I have a relatively high probability of
parenthood (approaching 20%), as well as relatively low probabilities
of residing with parents (80%) and attending school (80%) in comparison
to those individuals in life paths II and III. An accelerated transition to
JEYLAN T. MORTIMER ET AL.144
0.00
(a) (b)
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
Ages 14-18 Ages 18-19 Ages 20-21 Ages 23-24 Ages 25-26
Age
Ex
pec
ted
Ro
le P
rob
ab
ilit
y
In school MarriedChildren Lives with parents
Steady, full-
time work
In school MarriedChildren Lives with parents
Steady, full-
time work
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
Ages 14-18 Ages 18-19 Ages 20-21 Ages 23-24 Ages 25-26
Age
Ex
pec
ted
Ro
le P
rob
ab
ilit
y
In school MarriedChildren Lives with parents
Steady, full-
time work
0.00
(c)
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
Ages 14-18 Ages 18-19 Ages 20-21 Ages 23-24 Ages 25-26
Age
Ex
pec
ted
Ro
le P
rob
ab
ilit
y
Fig. 3. Latent Life-Path Probabilities for (a) Precocious Transition to Adulthood, (b) Delayed Transition to Adulthood and
(c) Multifaceted Transition to Adulthood.
Agency
andStru
cture
inEducatio
nalAtta
inment
145
adulthood continues in the years immediately following high school, as the
probability of school attendance declines sharply, while the probability of
steady, full-time work grows rapidly.3
Life path II (probability in sampled population ¼ 0.576), depicted in
Fig. 3(b), indicates a delayed transition to adulthood. During adolescence,
individuals in this pathway have a very low probability of parenthood.
Residence in the parental home (for at least part of the year) and school
attendance are prolonged in the years immediately following high school.
Family formation is also delayed (the probability of parenthood and mar-
riage are both under 30% at ages 25–26). Interestingly, sporadic, full-time
work is favored between the ages of 18 and 21 (not shown), perhaps in-
dicating summer work that is compatible with higher education, followed by
a rapid increase in the probability of steady, full-time work.
Finally, life path III (probability in sampled population ¼ 0.259), shown
in Fig. 3(c), suggests a ‘‘multifaceted’’ transition to young adulthood that is
neither rushed nor delayed. Similar to life path II, respondents have a high
probability of residing with parents and attending school. The work patterns
of individuals in life path III are also similar to those in life path II, as there
is a rapid increase in steady, full-time work in the years immediately fol-
lowing high school. However, the probability of having a child rises much
faster at ages 23–24 than in life path II (50% at ages 23–24, compared to just
9% for life path II); marriage exhibits a similar pattern. In addition, the
probability of residing with parents also declines more rapidly in early
young adulthood for this group.
To assess the impact of structural placement and agency on the prob-
abilities of following these three life paths, we regress individuals’ life path
probabilities on family background factors – race, gender, parents’ highest
educational level, family income, and family composition – and future plans
and goals for education, work, and family. As shown in Table 2, measures
of family background and socioeconomic status have a significant impact on
one’s life-path probabilities. In model 1, females and those whose parents
have lower levels of education and family income have higher propensities
for following a precocious life path. Males and adolescents whose parents
have higher educations and income have higher propensities for the slower,
delayed entrance into family and work roles, manifested in life path II, than
those with lower socioeconomic origins. Parental education is the only
significant socioeconomic predictor of the probability of following life path
III. Adolescents whose parents have higher levels of education have lower
propensities toward this multifaceted path. These findings highlight the sig-
nificance of structural advantage in the transition to adulthood.
JEYLAN T. MORTIMER ET AL.146
Table 2. OLS Regression Coefficients of Latent Life Path Probabilities (n ¼ 498).
Precocious (Life Path I) Delayed (Life Path II) Multi-faceted (Life Path III)
Model 1 Model 2 Model 1 Model 2 Model 1 Model 2
B (s.e.) B (s.e.) B (s.e.) B (s.e.) B (s.e.) B (s.e.)
White (vs. non-white) �0.634 (0.34) �0.752 (0.33)� 0.474 (0.40) 0.531 (0.40) �0.086 (0.21) �0.047 (0.21)
Male (vs. female) �0.538 (0.23)� �0.552 (0.23)� 0.946 (0.28)�� 0.966 (0.28)�� 0.007 (0.14) 0.011 (0.15)
Parental education �0.418 (0.08)��� �0.285 (0.08)�� 0.318 (0.10)�� 0.207 (0.10)� �0.132 (0.05)�� �0.135 (0.05)�
Family income �0.287 (0.06)��� �0.241 (0.06)��� 0.351 (0.07)��� 0.320 (0.07)��� �0.045 (0.04) �0.050 (0.04)
Intact family (vs. non-intact) �0.211 (0.30) �0.042 (0.29) �0.009 (0.35) �0.134 (0.35) �0.019 (0.18) �0.028 (0.18)
Marriage importance – – 0.166 (0.20) – – �0.166 (0.25) – – 0.030 (0.13)
Parenthood importance – – �0.247 (0.19) – – 0.144 (0.23) – – 0.045 (0.12)
Family self-efficacy – – �0.090 (0.16) – – 0.158 (0.19) – – �0.087 (0.10)
Career importance – – 0.113 (0.20) – – �0.098 (0.25) – – �0.057 (0.13)
Occupational aspiration SEI – – �0.020 (0.01)�� – – 0.022 (0.01)� – – �0.001 (0.00)
Occupational designation – – 1.468 (0.49)�� – – �1.731 (0.60)�� – – 0.086 (0.31)
Economic self-efficacy – – �0.131 (0.06)� – – 0.102 (0.07) – – �0.010 (0.04)
High educational promise (vs. low) – – �1.100 (0.25)��� – – 0.848 (0.30)�� – – 0.023 (0.16)
Intercept 0.752 (0.46) �0.513 1.14 �3.814 (0.55)��� �2.463 (1.38) �0.940 (0.28)� �0.792 (0.73)
R2 0.19 0.26 0.15 0.19 0.03 0.04
�po0.05.��po0.01.���po0.001.
Agency
andStru
cture
inEducatio
nalAtta
inment
147
Of particular relevance given our present concerns, the results show that
agency and planfulness are also important in predicting subsequent life
trajectories. According to our analyses, agentic orientations toward work
and education is what matters, not expectations and values with respect to
family life. Model 2 shows that the ninth graders with higher educational
promise and those with higher occupational aspirations have lower
propensities for the precocious path. Economic efficacy also reduces the
likelihood of precocious adulthood. Adolescents who specify what their
career might be in the future, indicating greater vocational crystallization,
have significantly higher probabilities of following this precocious life path.
High educational promise and high occupational aspirations upon entry
to high school predict higher probabilities of being on a path characterized
by ‘‘delayed’’ transitions into adulthood, which is not surprising given the
great educational investment and late exit from school that characterize this
path. It is interesting to note that adolescents who had greater occupational
uncertainty (indicated by failure to answer the question about their future
work plans) also have a greater propensity for path II. For those who do
know what occupation they hope to have in the future, however, the higher
their aspirations, the more likely they will follow this pathway. This pattern
of findings suggests that delayed entrance into adulthood can signify vo-
cational indecision, in which youth uncertainty leads to longer time spent in
school and later family formation. Alternatively, for those who have for-
mulated high aspirations at an early stage (age 14–15), delayed markers of
adulthood may result from a more focused postponement of full-time work
and family roles in order to attain the amount of education necessary for
high career aspirations. The longer incumbency of preadult roles and un-
certainty about future goals are often discussed in tandem when addressing
the contemporary transition to adulthood, as both foster extended explo-
ration and enable the postponement of adult role commitments. They do
not, however, always coincide. Moreover, delayed timing, when combined
with uncertainty, is likely to have a very different socioeconomic outcome
than delayed timing without uncertainty. With clear goals, young people can
maximally utilize the elongated transition to adulthood to develop their
human capital and otherwise groom themselves for future attainments.
Without clear goals, the lengthy transition to adulthood can be one of
‘‘floundering,’’ insecurity, and reduced achievement.
Attesting to the significance of adolescent agency, the addition of an
individual’s plans and expectations for family, career, and school in model
2 explains 7% and 4% more of the variance in the first two life path
propensities, respectively, than the measures of family background alone.
JEYLAN T. MORTIMER ET AL.148
In addition, we find evidence that while the other socioeconomic predictors
remain just as strong, the impact of parents’ education on these two life-path
propensities is partially mediated by individual agency. These results suggest
that while social position does in fact play a part in ‘‘routing’’ individuals
into certain life trajectories and impacts the timing of onset of adult roles,
one’s goals and expectations for these roles also have independent effects on
his or her early life course trajectories.4 But unlike the other two pathways,
Table 2 also suggests that an individual’s propensity for a multifaceted life
path is not affected by the importance he or she places on family, education,
or career.
DISCUSSION
Our assessment of domain-specific indicators of agency and immediate ac-
tions taken in the pursuit of goals is shown to have payoff with respect to
understanding the process of socioeconomic attainment during adolescence
and the transition to adulthood in contemporary urban America. We there-
by obtain a glimpse of what may be inside the ‘‘black box’’ intervening
between the expression of goals and values in middle adolescence and life
course outcomes in early adulthood. Our study is, however, not without
limitations. We have confined ourselves to the study of a single panel, lo-
cated in a particular national setting, in one brief period of historical time.
Given its origin in the upper Midwest region of the US, this panel is not well
suited to the assessment of racial and ethnic differences. Moreover, we have
little information about structural location other than socioeconomic origin.
The St. Paul schools, from which the students were selected, do not have
visible ‘‘tracks’’ that may constitute institutionally located ‘‘ladders’’ of
achievement. In general, the study does not speak to structural placement
within the high school that could address the kinds of macrostructural ar-
guments we alluded to at the beginning of this paper.
Notwithstanding these limitations, our work does suggest distinct path-
ways through the early life course, implicating agency, achievement-relevant
educational and work behaviors, and early adult attainments. Moreover,
our analyses suggest that agency is intricately connected to adolescent em-
ployment patterns. A central contribution of our work is the finding that
adolescent employment is neither uniform nor randomly distributed during
high school. In fact, there are five distinct patterns of work investment that
are linked to social background, early educational promise, the quality of
work experiences, and early adult socioeconomic attainments.
Agency and Structure in Educational Attainment 149
Might the work investment patterns – steady, occasional, sporadic, most
invested, and none – be considered a kind of ‘‘emergent’’ structure? They
would surely be recognized as structural dimensions or patterns, as ‘‘insti-
tutional career lines,’’ if they were assigned by educational or other au-
thorities, or if they were clearly recognized and validated by their
connections to the school. However, unlike the German system of appren-
ticeship, no central authority decides how much students should work while
they are attending school. Instead, these work-investment patterns emerge
from the youth’s actions in the context of the naturally occurring, ‘‘free’’
labor market (Mortimer & Kruger, 2000). They arise from the multiple
decisions and actions of employers, parents, counselors, students them-
selves, and other ‘‘significant others’’ who persuade, advise or influence
them.
To the extent that social structures that regulate and channel behavior are
emergent from the very agentic behaviors of actors themselves, our attempts
to discover distinct influences of agency and structure become increasingly
complicated, perhaps futile, as all are part of an interrelated and finely
articulated process of life course emergence. However, our analyses show
how early goals and values are linked to work behaviors during high school,
which in turn have predictive power with respect to subsequent trajectories
of work, schooling, and educational attainment. Our most recent, and at
this time least fully developed analyses, take this strategy one step further –
to consider how a variety of agentic orientations affect life pathways
characterized by configurations of school, work, and family roles. We thus
explore how early family, school, and work orientations and goals affect life
paths of transition to adulthood. The preliminary analyses presented here
indicate that expectancies and values surrounding future educational and
occupational role enactment must be taken into account to fully understand
propensities for earlier or delayed timing of key markers of adulthood.
We have come a long way from early models of attainment that simply
linked adolescent educational aspirations with adult educational and occu-
pational attainments. In expanding the range of agentic orientations under
consideration, and addressing the intervening behaviors that enable goals
and values to be actualized, our understanding of agency and the transition
to adulthood is enhanced. While structure – here considered by way of social
background location – is certainly also important, we find considerable
evidence that the psychological constructs exert independent effects.
These investigations could fruitfully be extended to other realms – for
example, to trajectories of family stability, civic involvement, mental health,
and physical well-being. We trust that these reflections and analyses will
JEYLAN T. MORTIMER ET AL.150
prod others to explore how structural location, psychological indicators of
agency, and agentic behaviors act in concert to produce contemporary life
course pathways.
NOTES
1. To assess whether participants (64% of those invited) differed from non-participants along common demographic characteristics, a probit analysis of thedecision to participate was conducted, including information from the 1980 Censusreported at the tract level to characterize the neighborhoods of all eligible families.Boys and older students (some of whom were retained in grade) were less likely toparticipate than girls and those who were the same age as most of their classmates.(Information about the gender and age of invitees was obtained from schoolrecords.) The racial, family structure, and socioeconomic composition of students’neighborhoods (indicated by median household income, receipt of public assistance,and educational and occupational status) were not related to participation (Finch,Shanahan, Mortimer, & Ryu, 1991).2. In the first stage, latent class modeling techniques are used to estimate the
unobserved latent role configurations of the markers at each age range (14–18,18–19, 20–21, 23–24, and 25–26), based on the cross-classifications of observed rolesat each. The second stage of the analysis generates a latent class model of the age-specific latent class configurations developed in the first stage. Latent life pathmodels are estimated using cross-classifications of unobserved, or latent, role con-figurations. These cross-classifications are derived from a ‘‘realization’’ of the latentrole configuration transition table (Macmillan & Eliason, 2003).3. The steady ‘‘full-time’’ workers averaged near continuous employment during
the sophomore, junior, and senior high school year at more than 20 h per week. Inthe years immediately following the high school period, steady full-time work in-dicates those respondents averaging 35 or more hours of paid work per week during11 or more months of the year.4. To see if these patterns differ by gender, we included interaction terms between
gender and each psychological measure in the model (not shown). These interactioneffects were not significant.
REFERENCES
Bandura, A. (1997). Self-efficacy: The exercise of control. New York: Freeman.
Brandstaedter, J. (1998). Action perspectives on human development. In: R. M. Lerner (Ed.),
Handbook of child psychology, 5th ed., Vol. I: Theoretical models of human development
(pp. 807–863). New York: Wiley.
Clausen, J. A. (1993). American lives: Looking back at the children of the great depression.
New York: Free Press.
Agency and Structure in Educational Attainment 151
Clogg, C. (1995). Latent class models. In: G. Arminger, C. Clogg & M. Sobel (Eds), Handbook
of statistical modeling for the social and behavioral sciences (pp. 311–359). New York:
Plenum Press.
Coleman, J. S. (1961). The adolescent society: The social life of the teenagers and its impact on
education. New York: Free Press.
Committee on the Health and Safety Implications of Child Labor. (1998). Protecting youth at
work. Washington, DC: National Academy Press.
Entwisle, D. R., Alexander, K. L., & Olson, L. S. (2003). The first grade transition in life course
perspective. In: J. T. Mortimer & M. Shanahan (Eds), Handbook of the life course
(pp. 229–250). New York: Kluwer Academic/Plenum.
Finch, M. D., Shanahan, M. J., Mortimer, J. T., & Ryu, S. (1991). Work experience and control
orientation in adolescence. American Sociological Review, 56, 597–611.
Gecas, V. (1991). The self-concept as a basis for a theory of motivation. In: J. A. Howard &
P. L. Callero (Eds), The self-society dynamic: Cognition, emotion and action
(pp. 171–188). New York: Cambridge University Press.
Grabowski, L. J., Call, K. T., & Mortimer, J. T. (2001). Global and economic self-efficacy in the
educational attainment process. Social Psychology Quarterly, 64, 164–179.
Heckhausen, J. (1999). Developmental regulation in adulthood: Age-normative and sociostructural
constraints as adaptive challenges. Cambridge and New York: Cambridge University Press.
House, J. S., & Mortimer, J. T. (1990). Social structure and the individual: Emerging themes
and new directions. Social Psychology Quarterly, 53, 71–80.
Lindsay, P., & Knox, W. E. (1984). Continuity and change in work values among young adults:
A longitudinal study. American Journal of Sociology, 89, 918–931.
Macmillan, R., & Eliason, S. R. (2003). Characterizing the life course as role configurations and
pathways: A latent structure approach. In: J. T. Mortimer & M. Shanahan (Eds),
Handbook of the life course (pp. 529–554). New York: Kluwer Academic/Plenum.
Markus, H., & Nurius, P. (1986). Possible selves. American Psychologist, 41, 954–969.
Mortimer, J. T. (1996). Social psychological aspects of achievement. In: A. C. Kerckhoff (Ed.),
Generating social stratification: Toward a new research agenda (pp. 17–36). Boulder:
Westview.
Mortimer, J. T. (2003). Working and growing up in America. Cambridge, MA: Harvard
University Press.
Mortimer, J. T., & Lorence, J. (1979). Work experience and occupational value socialization:
A longitudinal study. American Journal of Sociology, 84, 1361–1385.
Mortimer, J. T., & Kruger, H. (2000). Pathways from school to work in Germany and the United
States. In: M. T. Hallinan (Ed.), Handbook of the sociology of education (pp. 475–498).
New York: Kluwer Academic/Plenum.
Mortimer, J. T., & Shanahan, M. J. (2003). Handbook of the life course. New York: Kluwer
Academic/Plenum.
Mortimer, J. T., Staff, J., & Oesterle, S. (2003). Strategic patterns of adolescent work and early
socioeconomic attainment. In: J. T. Mortimer & M. Shanahan (Eds), Handbook of the
life course (pp. 437–460). New York: Kluwer Academic/Plenum.
Mortimer, J. T., & Staff, J. (2004). Trajectories of educational and occupational attainment in
adolescence and the transition to adulthood. Paper presented at the biennial meeting of
the Society for Research on Adolescence, March 11–14, Baltimore.
Roberts, K. (1968). The entry into employment: An approach to a general theory. The
Sociological Review, 16, 165–184.
JEYLAN T. MORTIMER ET AL.152
Rotter, J. B. (1966). Generalized expectancies for internal vs. external control of reinforcement.
Psychological Monographs, 80, 1–28.
Shanahan, M. J., Elder, G. H., Jr., & Miech, R. A. (1997). History and agency in men’s lives:
Pathways to achievement in cohort perspective. Sociology of Education, 70, 54–67.
Shanahan, M. J., & Flaherty, B. P. (2001). Dynamic patterns of time use in adolescence. Child
Development, 72, 385–401.
Wiese, B. S., Freund, A. M., & Baltes, P. B. (2000). Selection, optimization, and compensation:
An action-related approach to work and partnership. Journal of Vocational Behavior, 57,
273–300.
Agency and Structure in Educational Attainment 153
NON-NORMATIVE LIFE COURSE
TRANSITIONS: REFLECTIONS ON
THE SIGNIFICANCE OF
DEMOGRAPHIC EVENTS ON LIVES
Frank F. Furstenberg
INTRODUCTION
A prominent tenet of life course research holds that ill-timed and unex-
pected events, non-normative transitions, and disorderly status sequences
have profound and lasting consequences for an individual’s success in later
life. Simultaneously rooted both in early demographic and sociological
studies undertaken in the middle part of the last century, life course theorists
have long argued that social timetables regulate status passages, creating
social support and expectations, and generating resources and rewards to
those who observe culturally mandated schedules (see, for example, Elder,
1998; Neugarten, Moore, & Lowe, 1965; Rindfuss, Swicegood, & Rosenfeld,
1987).
In their classic treatise on aging and social stratification, Riley, Johnson,
and Foner (1972) described how role transitions were socially orchestrated
through age norms that govern movement of populations into and through
social roles. In kinship-regulated societies, these rules determine the struc-
ture of generations and social reproduction. In modern societies, theorists
Towards an Interdisciplinary Perspective on the Life Course
Advances in Life Course Research, Volume 10, 155–172
Copyright r 2005 by Elsevier Ltd.
All rights of reproduction in any form reserved
ISSN: 1040-2608/doi:10.1016/S1040-2608(05)10005-7
155
such as Parsons (1954), Merton (1968), and Ryder (1965) explained that
social timetables are critical for determining and coordinating the basic in-
stitutions of society: the family, educational systems, and the labor market.
It is easy to understand how the assumption that unscheduled life events
have a singular power to cast a permanent shadow over the life course
became pervasive in the sociological and social demographic literature
grounded in a life course perspective. Many theoretical arguments appear to
support the notion that planning, timing, and orderly sequencing of role
transitions increase the probability of successful attainment.
First, in the middle of the past century, sociologists who worked in the
tradition of symbolic interaction emphasized that preparation for making a
status or changes was crucial to an actor’s ability to negotiate transitions.
The ideas of role preparation and anticipatory socialization, popular in the
sociological literature of the 1950s and 1960s, advanced the proposition that
successful acquisition of social roles required training, practice, and self-
definition (Becker & Strauss, 1956). It logically followed that correct timing
was an important feature of successful passage through a series of inter-
related status transitions. Social demographers, interested in the transition
to adulthood, began to examine the impact of timing of events on long-term
success in status attainment and being off-time conferred social risk
(Clausen, 1972; Marini, 1984; Winsborough, 1979).
With the work of Neugarten et al. (1965), social psychologists and soc-
iologists also began to explore how norms about timing operated to regulate
when and under what circumstances individuals would make significant
status transitions (Hagestad, 1990; Settersten, 1998). A line of research be-
gan to look into how institutional gatekeepers – family, schools, employers,
and legal authorities – helped to set and reenforce the social timetable by
allocating resources and rewards for staying on schedule and penalties for
violating age norms. The costs of either being early or late have been as-
sessed in a host of journal articles. As Elder (1984) ingenuously argued in
Children of the Great Depression, historical events such as wars or economic
downturns, outside the control of individuals, can drastically revise the op-
portunities for role acquisition and hence shape the course of lives. It is
evident in Elder’s work that he embraced the general assumption that timing
– fortuitous or ill – affects the structure of opportunities at both an indi-
vidual and social level, influencing the life chances of actors and cohorts in a
given society.
Another related line of theory about the life course argued that the se-
quencing of life events similarly had important consequences for success in
both the labor market and the family. Again, this proposition rests on the
FRANK F. FURSTENBERG156
assumption that when individuals follow predictable and socially organized
pathways, they are more likely to encounter success. Educational completion
prior to employment or marriage prior to first birth compared to the opposite
order are demographic examples that have received widespread attention by
demographers, economists, and sociologists. In a well-known article pub-
lished in 1961, Harold Wilensky (1961) showed how orderly as opposed to
disorderly careers were associated with positive consequences for a range of
economic outcomes as well as for psychological adjustment in later life.
Thus, there is a long history in the social sciences of believing that non-
normative or unscheduled transitions confer social disadvantage. My own
work was deeply grounded in the life course perspective, and on how the
timing and sequencing of transitions altered prospects for successful incor-
poration into adult roles (Furstenberg, 1976; Modell, Furstenberg, &
Hershberg, 1976). Of course this set of ideas, as Matilda Riley and her
coworkers observed several decades ago, were equally applicable to the
passage from adulthood to retirement.
TEENAGE CHILDBEARING AS
A NON-NORMATIVE EVENT
At the same time that Riley, Elder, Wilensky, and other social scientists were
setting out a life course perspective, I was beginning what has turned out to
be a life-long study of the social consequences of teenage childbearing. My
project in Baltimore, initiated in the mid-1960s, began more by happen-
stance than a deliberate intention to test a basic proposition of life course
theorists. However, I soon came to see the applicability of this theoretical
framework for examining the question of how and why premature parent-
hood occurs and its lasting implications for success in later life for young
parents and their children.
In my first book on the topic, I grounded my research in the idea that
early childbearing was a prime example of a violation of the normative
schedule for family formation. My point of departure was the oft-quoted
observation by the social demographer, Arthur Campbell (1968), who spec-
ulated that:
The girl who has an illegitimate child at the age of 16 suddenly has 90 percent of her life’s
script written for her.
In the decades that followed a huge literature has accumulated in the U.S.,
especially, but in Western Europe more generally, on the impact of early
Non-Normative Life Course Transitions 157
childbearing on the later life success of teenage parents and their children
(see, for example, Hayes, 1987; Brown & Eisenberg, 1995; Maynard, 1997).
It is beyond the scope of this essay to review this literature in any detail or
even recount the findings of my own 30-year longitudinal study of a pop-
ulation of teenage mothers who gave birth to a child while in their mid-
teens. Initially, my findings, like the work of others, appear to supply strong
support for the conclusion advanced by Campbell without the benefit of
much empirical evidence. But over time, with more sophisticated techniques
of data analysis, there is growing consensus among researchers in the field
that early childbearing did not produce the kinds of effects that life course
theorists would have predicted. While policy makers, the public at large, and
even teenage mothers themselves, still tend to believe that early childbearing
has deleterious, long-term effects on educational and occupational success
as well as mental and social health – both for the mother and the offspring –
the findings in the literature show that support for this conclusion is, at best,
ambiguous (Furstenberg, 2003).
Long-term, longitudinal studies reveal, contrary to expectations, that, at
most, parenthood during adolescence has only modest long-term impacts on
teenage mothers and their children when compared to mothers who delay
their first births and their children when the two populations are matched in
other respects. Thus, rather than affirming that non-normative demographic
transitions like teenage childbearing profoundly shape the course of later life
– that demography is destiny – the evidence seems to point in quite the
opposite direction. Non-normative events produce diverse consequences,
conferring disadvantage on some but seeming to perturb little or not at all
the pathways of others. I will return to the reasons why this is so shortly
after discussing the other example from another line of research that I have
engaged in, the impact of marital dissolution and remarriage for children’s
development.
MARITAL DISRUPTION AS
A NON-NORMATIVE EVENT
Long before the formulation of the life course perspective, social critics and
cultural commentators in the early part of the 20th century had reached the
conclusion that marital disruption has devastating effects on children’s life
chances. Their literature argued that divorce, as a life event or transition,
when viewed from the perspective of children, has many of the same prop-
erties as early childbearing. It thrusts children into a family situation which
FRANK F. FURSTENBERG158
is usually unplanned and for which they are typically unprepared. It de-
prives them of resources in the form of parental time and money, and it
restricts their ability to acquire social skills from the absent parent. More-
over, since divorce is frequently followed by remarriage or cohabitation
(and sometimes further disruption), it exposes children to a succession of
transitions that are often accompanied by geographic moves, changes in
household routines, and shifting rules and regulations (Furstenberg &
Cherlin, 1991; McLanahan & Sandefur, 1996). In short, there is abundant
reason to believe that divorce as a life event fundamentally compromises
children’s immediate and long-range welfare.
In recent years, the social and personal costs of divorce for children have
been the topic of a considerable body of research. As in the case of early
childbearing, the popular impression of what this research has shown di-
verges from what many professional researchers have concluded, though it
can hardly be said that consensus exists. A few researchers, most notably in
the U.S., Wallerstein and Blakeslee (1989) and Waite and Gallagher (2000),
have concluded that divorce is every bit as detrimental to children’s life
chances as has been long believed. However, a growing number of re-
searchers in the field have reached the conclusion that, surprisingly, marital
disruption is relatively weakly linked to later life outcomes (Moynihan,
Smeeding, & Rainwater, 2004). Like many of my peers, I would be inclined
to say that the long-term impacts of divorce are modest to moderate in the
aggregate, which is to say that divorce confers risk for some but certainly
not for most children whose parents’ marriages break up during their
childhood years (Furstenberg & Cherlin, 1991; see also Cherlin, 1999).
Now, if I am correct in my characterization of the evidence from studies
carried out over the past several decades, we might reach the conclusion, at
least based on these two prime examples, that the consequences of non-
normative events outcomes in later life are surprisingly minor in light of life
course theory. In the remainder of this paper, I will first try to explain why
theorists overstated the adverse impact of ill-timed and out-of-sequence
events. Then I will try to revise the theory in ways that are more consistent
with the evidence. Doing so suggests an agenda for future research, a topic
that I develop in the final part of the paper.
WHEN THEORY AND EVIDENCE COLLIDE
Looking back at the two case examples that I have mentioned, they share
some prominent features. Both could be characterized as social problems
Non-Normative Life Course Transitions 159
where researchers have been working against a backdrop of heated public
discourse and have joined the public in treating evidence in a cavalier fash-
ion: the theory seemed so compelling that the evidence collected was un-
critically evaluated. Even if Arthur Campbell, a respected social
demographer, had engaged in hyperbole in assessing the impact of early
childbearing on the subsequent life course of young mothers, surely could
any reasonable person doubt that teenage parenthood does not irreparably
compromise the life chances of young mothers and their offspring? And,
even if Judith Wallerstein was a self-declared advocate for reducing the toll
that divorce takes on children, is it reasonable to question her conclusion
that divorce produces lasting damage for those who experience the loss of a
parent and grow up in a single-parent household or, worse yet, experience a
cascade of family change (Ahrons, 2004)?
And, indeed the findings in each of these separate lines of research initially
seemed to conform to conventional wisdom. In the 1970s and early 1980s,
study after study showed that teenage mothers had extraordinarily high
rates of school dropout, reliance on welfare, subsequent fertility, unstable
marriages, and so on. Similarly, Wallerstein’s conclusions, based on a small
clinical study in the 1970s, found widespread support from other research
showing a powerful association between divorce and a range of adverse
outcomes for children. However, both areas of research – as has been true in
many studies of unscheduled and ill-timed events in the life course – have
suffered from potentially fatal methodological problems.
Economists have long recognized the problem of unobserved differences
or endogamy, the inability to identify causality between an event and its
consequences. In the 1980s, some sociologists too, began to come to grips
with the issue (Lieberson, 1985). Simply put, teenage childbearing, divorce,
and other such events, do not occur randomly within a population. To the
contrary, these events occurred selectively to individuals or couples who
differ in a wide variety of ways that distinguish them from their counterparts
who are able to avoid the event. I need not dwell on this point because it has
become so widely understood by social scientists that few fail to appreciate
the burden it places on reaching causal conclusions about the links between
events and their consequences.
Over the past couple of decades, a wide variety of methodological strat-
egies have been employed to tease out causality in assessing the influence of
timing and sequencing in the life course. Controlling for prior group dif-
ferences has become standard, but it is widely known that such an approach
inevitably over-estimates casual effects by omitting unmeasured differences.
Fixed effects models in longitudinal research partially address this problem.
FRANK F. FURSTENBERG160
Similarly, researchers have used statistical methods for distinguishing
between exogenous and endogamous influences. Finally, researchers have
devised clever quasi-experimental approaches to identify true casual links.
None of these approaches, by itself, has completely solved the problem of
selection, but together they have demonstrated that a healthy share of causal
influence of ill-timed or out-of-sequence events is social selection rather than
true cause.
Specifically, much and probably most of the presumed effect of teenage
childbearing on later life outcomes appears to be spurious (Maynard, 1997).
That is, if we could postpone the births of teenagers by 5 years, it would
have little effect on their life chances because most would not end up going
to college merely because their first birth occurred later. Or, when they
eventually became pregnant, they would not be married or otherwise ca-
pable of supporting their child as a single parent. Paraphrasing Campbell’s
assertion, we might say that if a pregnant women in the US is poor, black,
and has been a low-achiever in school, 70 percent of her life script may be
written for her – whether or not she has a child as a teenager or in her early
20 s. Early childbearing usually does not help young women to succeed, but
neither does it cause them to fail, having the crushing impact on their lives
that Campbell, and many other social scientists, have hypothesized.
The story about the consequences of divorce for children appears to be
similar but not yet so clearly documented. In the first place, the evidence on
the detrimental effects of divorce on children cannot be so easily disregarded
in part because researchers have not yet found as many ways to rule out the
influence of unobserved differences. Divorce is the outcome of a process that
typically involves protracted parental conflict, economic uncertainty, and
unresolved sexual and interpersonal problems. Moreover, it frequently oc-
curs in unions formed by couples with prior emotional, social, and economic
problems. Sorting out how much the divorce, per se, adds to or complicates
the circumstances of children who are already vulnerable and in conflict-
ridden households, is a challenging problem. Still, there is little doubt that
many of the presumed effects of divorce are in fact results of differences that
occur prior to the event itself such as poor parenting, marital strife, eco-
nomic and social circumstances, and the like.
If my evaluation of the literature on these two examples is correct, it raises
a huge issue for life course researchers than has not been adequately ad-
dressed: why are the effects of non-normative events not as large or perhaps
as long lasting as life course theory has predicted? In the second part of this
paper, I will try to answer this question. In doing so, I will suggest mod-
ifying, if not abandoning, the overly simple model that assumed that the
Non-Normative Life Course Transitions 161
timing, scheduling, and sequencing of life course transitions invariably
produce powerful and persistent consequences for the course of later life.
In its place, I argue for a more intricate theory of when and why non-
normative transitions have lasting effects or are of little consequence over
the long term.
COMPLICATING THE PICTURE: THE MODERATION
OF LIFE EVENTS OVER TIME
The problem posed requires an understanding of why non-normative events
do not produce the large and lasting impact that theory predicts? Answering
this question requires a close examination of how individuals, families, and
social systems react to the occurrence of ill-timed and poorly sequenced
transitions such as teenage childbearing or marital disruption. Accordingly,
I will draw examples from findings of my own research and the work of
others to illustrate why and how our theory needs to be revised to take
account of the findings of life course research.
Several different reasons might explain the seemingly anomalous findings
to which I have referred: First, it is possible that these events only affect a
small number of the highly vulnerable individuals and that most individuals
are relatively robust and impervious to non-normative transitions. This has
sometimes been referred to in the literature as sources of resiliency in de-
velopmental psychology.
A second possibility, which could be considered a variant of the first, is
that there is an inevitable distribution of responses to the occurrence of non-
normative events in any given population: some individuals react adversely
while others may actually benefit from the unexpected transition; still others
may have little or no response. In other words, a series of conditions and
circumstances at the individual and social level moderate the impact of non-
normative transitions.
Finally, we might imagine that individuals are widely affected by the event
but in time these responses fade much as ripples in a pond diminish with
distance from the initial splash. Over time, individuals can take actions to
repair or recover from the initial response to a non-normative event, off-
setting any potential damage done. In fact, these explanations are not en-
tirely mutually exclusive and may all have a part in explaining why the
impact of non-normative transitions is not as large as many believed. Let us
consider each in turn.
FRANK F. FURSTENBERG162
Back in the 1960s, when the first interviews were conducted in the
Baltimore study, the then pregnant teenagers and their mothers were asked
how they initially reacted when they learned that they were pregnant.
Recalling the event, the majority of parents replied that they were extremely
angry and upset and their daughters, similarly, reported that they were
frightened and distressed. A typical response by the pregnant teens went
something like this:
‘‘I was shocked and sad. I knew my mother would be furious. I was really
scared of what she would do when she found out.’’ However, when then
asked how they now felt (because the first interview occurred several months
after they learned that they were pregnant), most replied that they had
become reconciled to, if not excited by, the prospect of parenthood in the
interim. The fact is that many of the parents, who were angry at first,
softened their reactions and most of their daughters, fearful of their parents’
reactions, learned that they would be assisted and supported by their
families.
By this time, a number of parents summed up their change of heart by
saying of their daughters’ circumstances, ‘‘Everyone is entitled to make one
mistake.’’ The pregnant teens, in turn, had begun to feel as though they
could recover from their misstep by making a greater commitment to their
future and the future of their child. Young women, who had or were about
to enter marriage, were especially positive by the time of the interview be-
cause they believed, as many pregnant teens did at the time, that marriage
would put them back on course. However, even the teens that did not
contemplate matrimony became more sanguine about their prospects of
rectifying their situation. Many learned that they could stay in school, re-
ceive special services, or would receive help from the baby’s father. On both
psychological and sociological levels, most families mobilized their resources
to respond to the impending challenge by a redefinition of the situation.
I would not want to claim that this redefinition neutralized the potentially
adverse effects of early childbearing, but it does suggest that actors reacted
neither passively nor uniformly to early childbearing. Many were able to
mobilize in response to becoming pregnant in ways that might mitigate some
of the impact. And, apart from their ability to redefine the situation, the
young mothers and their parents began to explore a range of different pos-
sible adaptations to becoming pregnant, including marriage, returning to
school, and the realignment of family roles, to mention but a few.
True, the unplanned pregnancy forced a series of unanticipated changes in
the life course of the teenagers, but the changes were not always negative.
Some adolescents claimed at the time as well as in subsequent interviews
Non-Normative Life Course Transitions 163
that the pregnancy was a turning point in their lives. It made them grow up
faster, take themselves more seriously, and realize that they had to take
seriously their new responsibilities. In many cases, they reported improved
relations with their parents, greater engagement in school, and an enhanced
sense of self (see Furstenberg, 1976; Furstenberg, Brook-Gunn, & Morgan,
1987). As an indication of these changes, the interviewer who watched the
young mothers from one wave to the next, often commented how much and
how quickly they had grown up after they became parents.
Of course, considerable variation occurred in the years following the first
birth of the teen mothers in both their reactions to parenthood and the level
of support provided by families and the surrounding community. Among
those who hastily married, some entered seemingly viable unions while
others found less suitable or compatible partners. Similarly, for those who
remained single, some were able to find their way back to school and even-
tually into the labor force while others were not. Young mothers who found
their way to a special school for pregnant teenagers were especially likely to
complete high school in the five years following the birth of the first child.
The young mothers used public and private support differently. Some, who
received public assistance, made good use of it by returning to school; others
devoted themselves to caring for their children until they were of school-age
at which point they either went back to school or found employment; still
others were unable to mobilize to gain economic independence in the years
that followed the birth of their first child.
I, along with many other researchers, have devoted considerable attention
to explaining the sources of these varied life trajectories of teen mothers
after the birth of their first children. Clearly, part of the source of variation
lies in individual attributes of the young mothers themselves. Physical and
mental health, cognitive skills, attractiveness, motivation, flexibility, and a
host of other attributes, no doubt, figured into their ability to respond to
their changed circumstances. In the language of developmental psycho-
logists, some mothers were more vulnerable and others more resilient in the
face of a new challenge.
Families too, possess different levels of resources and support to offer to
the young mothers in the way of time, money, guidance, social contacts, and
the like. They differed in family-based social capital, the ability to mobilize
support for their children and cultural capital, and knowledge about the
social world. The families, and the young mothers, were also differently
embedded in neighborhoods and communities, and thus had differing access
to resources to cope with premature parenthood. My collaborators and I
have shown what an important part access to services, programs, mentoring,
FRANK F. FURSTENBERG164
and the like played in determining successful adjustment in the immediate
years following the birth of a first child.
These conditions all affected the likelihood of other related transitions
such as marriage, return to school and graduation, and most notably, the
birth of a second child. The probability of a second birth was much greater
for those who married than those who remained single and differed among
the single women depending on whether they returned to school or not.
While information on individual level psychological resources was limited,
clearly, some women were more prone to having subsequent unplanned
births than were others, a potent predictor of long-term success. In time,
some women curtailed their fertility by sterilization and others became in-
fertile, but higher order births proved to delay or deter the ability to gain
economic independence and invest in children already born. In some cases,
they led to ruptures with families, who were initially willing to offer assist-
ance but whose patience or resources depleted when demands persisted or,
worse yet, increased.
The longitudinal evidence suggests that there is no single path to recovery
nor is there any point from which recovery becomes irretrievable. To make
matters even more complicated, new events such as the formation of a new
partnership, the death of a parent, or a changing employment situation
continually perturb trajectories for better and for worse. At each interview –
and there have been seven in all – we discovered that the circumstances of
some mothers would change direction. Moreover, sometimes, gains or losses
might occur in economic fortunes without affecting other domains of suc-
cess such as interpersonal relationships, mental and physical health, or self-
definitions. The correlation of different indicators of success was only
modest, complicating our ability to tell a single story about successful and
unsuccessful pathways. In short, it is not easy to point to a single set of
strategies that invariably paid off in later life success. Women who married,
for example, did not necessarily fare better than those who remained single
in part because the marriages were so brittle and were often accompanied by
higher rates of fertility. Conversely, seeking public assistance did not prove
to be a poor strategy if women made use of welfare to get more schooling.
To discriminate successful from unsuccessful trajectories, we have to under-
stand how the strands of the life course are interwoven through a series or
chain of small decisions, looking at the intentions of the young mothers as
they faced certain choices or encountered different possible pathways. An
ongoing and continuous feedback took place between life decisions, how
they worked out in reality, and how situations were subsequently defined
and interpreted by the young mothers, their parents, their partners, and
Non-Normative Life Course Transitions 165
their children. Without seeing the interplay between the intentions, deci-
sions, and immediate consequences, it is difficult to make sense of the chain
of strategies that evolved over time to understand how they worked to
produce particular outcomes. Many of our analytic techniques are insen-
sitive to this complex process and most of the data that we collect are not
sufficient to carry out an analysis that splices together qualitative accounts
(motives and interpretations) and quantitative evidence about decisions and
their consequences. Doing so properly requires new ways of examining
longitudinal information and new techniques for blending qualitative and
quantitative information.
Even with the crude analytical techniques that I have employed in the
Baltimore Study, it is possible to see the emergence of different patterns of
success at mid-life. First, women who married early and were able to forge
viable and rewarding unions clearly did better than anyone else in the study,
but they were only a tiny minority (under 10 percent of the sample) of the
young mothers in the study. Case studies of these women showed that they
held different attitudes about marriage, came from families with more stable
unions, and possessed a greater array of interpersonal skills. They also
married men, often the fathers of their first child, who possessed similar
attributes. Thus, marriage confers few benefits unless those who marry
possess the resources, attitudes, and skills to make it work out. Lacking
some or all of these qualities, most entered a succession of unions that did
not work out well and indeed left the young mothers and their offspring
worse off than many of their never-married counterparts. Some women
rapidly took themselves out of the mating game, choosing to pair with men
in short-term sexual relationships or they eschewed men entirely. When
these women did so to invest in their children, their offspring frequently did
well. Alternatively, when they withdrew from men without economic or
family support, the children often fared poorly.
Second, women who were able to manage to control their fertility did
much better in later life, as did their children. Fertility control, like mar-
riage, required motives, resources, and opportunities. For example, some
women curtailed their childbearing by staying out of unions or avoiding
relationships with men. Others sought sterilization after they had a first or
second child. Only a small number were able to use contraception success-
fully. To understand the implementation of fertility control, we must look at
several different possible routes that the young mothers took as well as
considering the fact that some were simply more fecund than were others. In
sum, women achieve low fertility in many different ways or fail to do so for
several different reasons, such as their inability to manage relationships with
FRANK F. FURSTENBERG166
men, their poor skills at using contraception, or their religion beliefs that
prevented them from seeking abortion.
Third, work success, not surprisingly, came to those who returned to
school and gained credentials, but other women, who were not terribly
successful in gaining schooling, managed to find particular economic niches.
Again, we concluded that there was a strong link between educational suc-
cess and economic independence, but the pathways to educational success
and economic independence varied quite a bit among the women in the
study. Some returned to school immediately, others went to work and then
back to school, and still others only went back to school later in life.
Although we do not have enough cases to test the efficacy of these different
strategies, it appears that they all seemed to produce positive effects.
Unquestionably, the women who had the cognitive, psychological, and so-
cial support to gain more schooling and develop labor market skills fared
reasonably well compared to those who failed to build human capital.
Complications only grow when we try to link the success of the mothers
to their offspring. Without a doubt, links exist but they are rarely straight-
forward. Success in employment, for example, was associated with better
outcomes for children, but the correlation was relatively weak. Mothers’
actions on their own behalf sometimes led to improvement in their children’s
life chances, but very often they did not because the mothers’ investments in
themselves came at the cost of time and involvement in their children. Some
of the less successful mothers devoted themselves to their children, putting
their hopes in their children’s, rather than their own, careers. This strategy
occasionally paid off, at least if we believe some of the accounts of children
who succeeded despite growing up while on public assistance. Many of the
children when they reached early adulthood reported that their mothers had
held up their own experiences as teen parents as a negative example to their
children. Their mantra became: ‘‘Don’t follow in my footsteps.’’
Working with data from the Baltimore Study, as I have for almost four
decades, has taught me to appreciate how much actors construct, interpret,
and make meaning of their actions in ways that have powerful consequences
for both themselves and other immediate members of their families. Most of
the teen mothers that I followed over time felt that they had become preg-
nant too early and entered parenthood too soon. Almost all wanted their
children to take a different pathway. It is no easy matter, however, to
explain why some were able to follow their mothers’ wishes while others
became second- (or third-) generation teen mothers. The ability of parents to
translate their desires to their children differs greatly depending on their
parenting skills, their resources to keep their children on track, the support
Non-Normative Life Course Transitions 167
that they receive by others in their household and community, and of
course, the children’s individual and social circumstances. It is one thing to
say that lives are inextricably linked across the generations, it is quite an-
other to show exactly how such linkages play out within families.
One might conjecture that variability in outcome is greater in the pop-
ulations most susceptible to non-normative events. No doubt, low-income
and marginalized populations are more vulnerable to these events for both
structural and perhaps psychological reasons. However, I would argue that
the broad contours of what I have reported about the difficulty of predicting
trajectories of teen mothers and their offspring apply more broadly to less
vulnerable populations. John Clausen (1972) used the term loose coupling to
refer to the varied pathways that can be taken when events occur in the life
course. Not all researchers who have studied divorce have appreciated
Clausen’s observation that casual links between powerful events and their
consequences can sometimes be quite faint. Many of the same reasons why
teen parenthood does not inevitably have adverse consequences for parents
and children apply as well to divorce.
Divorce, like teenage childbearing, may have positive as well as negative
effects. The most obvious of these is that it often curtails chronic conflict
between parents or leads to the removal of a dysfunctional parent from the
household. Children may bond more strongly with the residential parent (or
even the non-residential parent) when the couple is no longer living together.
They may have opportunities to acquire new competencies by assuming
greater responsibilities in the household. Of course, none of these possibil-
ities is without potential costs, but researchers often ignore them in the
literature on divorce, which assumes that the consequences for children are
invariably bad (for an exception, see Barber & Eccles, 1992).
Quite apart from the potential benefits that divorce might confer to chil-
dren, researchers have shown that children in the same family react quite
different to marital disruption (Hetherington & Kelly, 2002). Depending on
their personal interpretations of the reasons for the divorce, their relation-
ships to the residential and non-residential parents, the reactions of siblings,
and a host of other conditions that moderate the impact of marital disso-
lution, children’s reactions to the event vary enormously. Moreover, parents
differ in how they manage the divorce, their capabilities for handing lone
parenthood, their economic circumstances, and the amount of support that
they can garner from others. All of these conditions affect the way that
divorce impinges on a child’s life both immediately and in the long term.
The immediate reaction of children to the divorce of their parents does
not necessarily signal how they may adapt to the divorce over time, as events
FRANK F. FURSTENBERG168
that follow the divorce in both the parents’ and children’s lives can over-
shadow the significance of the marital breakup and change its meaning for
children.
Suppose, for example, that upon the breakup of parental marriage, chil-
dren encounter a series of new and unstable family circumstances. Compare
their circumstances to the opposite situation where the parents are able to
work out a collaborative pattern of parenting and are committed to re-
maining single. Alternatively, we can imagine that one or another remarries
and the stepparent proves to be a capable partner and parent for the child.
These different scenarios can lead to very different outcomes and, in turn,
influence the child’s definition of the impact of the divorce on their lives.
Just as in the case of teenage childbearing, a number of different con-
ditions can moderate the impact of divorce by either heightening or atten-
uating its long-term consequences for parents and children. The initial
‘‘definition of the situation’’ (that is, the way that it is first experienced) sets
in place a response trajectory, but evidence seems to suggest that the initial
response is not a reliable gauge of the long-term consequences. And, in any
event, more than we frequently recognize, early responses are the product of
pre-existing individual differences and social conditions often linked to the
nature of the transition; these pre-existing differences are partially or largely
responsible for the slope of the response trajectory. Thus, we must take great
care to take into account the fact that non-normative events often occur
selectively to individuals and families that are already vulnerable. Assessing
even their initial impact has proved to be a challenge precisely because it
requires sorting out endogenous and exogenous casual factors.
Subsequent, events may often reinforce the impact of non-normative
transitions, but they often have just the opposite impact: they may offset the
initial adverse reaction because individuals continue to respond, react, and
adjust to their life situations. In this paper, I have attempted to explain why
and how this can happen. Let me conclude by summarizing some of the
most important conditions that can redirect response trajectories over time.
1. Individuals increase their motivation and commitment in response to a
non-normative event. In other words, they mobilize to counteract the
predictable negative impact of a non-normative transition.
2. Others in their social networks, especially family members and friends
may increase support and resources to help them recover from the event.
3. Social agents intervene providing services and access to opportunities
that might not otherwise have occurred. Policies can be designed to re-
duce the potentially adverse effect of a non-normative transition.
Non-Normative Life Course Transitions 169
4. The experience of the event increases previously undeveloped personal
and social capacities. Non-normative transitions can, and often do, pro-
mote development.
5. Subsequent life events offset and compensate for initial disadvantages.
Over time, other events and transitions may reverse the fortunes of those
who encounter non-normative transitions. This can happen merely from
good fortune or because of one or more of the other reasons listed above.
CONCLUSION
I have suggested that theorists describing the impact of non-normative
events – ill-timed, unscheduled and out-of-sequence transitions – have sub-
scribed to a simplistic and overly deterministic model of how the life course
evolves over time. They have not given enough attention to the role of actors
in defining and moderating the impact of such transitions nor have they
taken adequate account of how a wide range of conditions moderate the
impact of non-normative events on subsequent life pathways. Drawing from
both my own research and the work of others, I have identified a series of
important mechanisms that often moderate the impact of non-normative
transitions. Consideration of these mechanisms may help to explain why
events such as teenage childbearing and divorce do not cast as long or as
dark a shadow in the subsequent life course as is often believed.
I have also contended that future research on the life course must take
fuller account of how circumstances are defined and responded to by actors
as well as examine objective markers of success and failure. Failing to do so
means that we ignore the process by which individuals experience the event
and hence the ways that they seek out and discover possible corrective
actions. The management of life course events is an under-investigated topic
that brings together personal agency, social support, and opportunity struc-
tures in a common theoretical framework that may improve our under-
standing of how life courses are constructed and how individual
development occurs in response to both normative and non-normative
transitions.
REFERENCES
Ahrons, C. (2004). We’re still family: what grown children have to say about their parents’
divorce. New York: HarperCollins.
FRANK F. FURSTENBERG170
Barber, B. L., & Eccles, J. S. (1992). Long-term influence of divorce and single parenting on
adolescent family and work-related values, behaviors, and aspirations. Psychological
Bulletin, 111, 108–126.
Becker, H. S., & Strauss, A. L. (1956). Careers, personality, and adult socialization. American
Journal of Sociology, 62(3), 253–263.
Brown, S. S., & Eisenberg, L. (Eds) (1995). The best intentions: Unintended pregnancy and the
well-being of children and families. Washington, DC: National Institute of Medicine.
Campbell, A. A. (1968). The role of family planning in the reduction of poverty. Journal of
Marriage and the Family, 30(2), 236–245.
Cherlin, A. J. (1999). Going to extremes: Family structure, children’s well-being and social sci-
ence. Presidential address to the Population Association of America, New York, March
26.
Clausen, J. A. (1972). The life course of individuals. In: M. W. Riley, M. Johnson & A. Foner
(Eds), Aging and society: A sociology of stratification (pp. 457–514). New York: Russell
Sage Foundation.
Elder, G. H., Jr. (1984). The children of the great depression: Social change in life experience.
Chicago: University of Chicago Press.
Elder, G. H., Jr. (1998). The life course and human development. In: R. M. Lerner (Ed.),
Handbook of child psychology, Volume 1: Theoretical models of human development (5th
ed., pp. 939–991). New York: Wiley.
Furstenberg, F. F. (1976). Unplanned parenthood: The social consequences of teenage childbear-
ing. New York: The Free Press.
Furstenberg, F. F. (2003). Teenage childbearing as a public issue and a private concern. Amer-
ican Review of Sociology, 29, 23–39.
Furstenberg, F. F., Brooks-Gunn, J., & Morgan, S. P. (1987). Adolescent mothers in later life.
New York: Cambridge University Press.
Furstenberg, F. F., & Cherlin, A. J. (1991). Divided families: What happens to children when
parents part. Cambridge, MA: Harvard University Press.
Hagestad, G. (1990). Social perspectives on the life course. In: R. H. Binstock & L. K. George
(Eds), Handbook of aging and the social sciences (3rd ed., pp. 151–168). New York:
Academic Press.
Hayes, C. D. (1987). Risking the future: Adolescent sexuality, pregnancy, and childbearing (Vol. I).
Washington, DC: National Academy Press.
Hetherington, E. M., & Kelly, J. (2002). For better or for worse: Divorce reconsidered.
New York: W. W. Norton & Co.
Lieberson, S. (1985). Making it count. Berkeley: University of California Press.
Marini, M. (1984). Age and sequencing norms in the transition to adulthood. Social Forces, 63,
483–507.
Maynard, R. A. (Ed.) (1997). Kids having kids: Economic costs and social consequences of teen
pregnancy. Washington, DC: The Urban Institute Press.
McLanahan, S., & Sandefur, G. (1996). Growing up with a single parent: What hurts, what helps.
Cambridge, MA: Harvard University Press.
Merton, R. K. (1968). Social theory and social structure. New York: The Free Press.
Modell, J., Furstenberg, F., & Hershberg, T. (1976). Social change and transitions to adulthood
in historical perspective. Journal of Family History, 1, 7–32.
Moynihan, D. P., Smeeding, T. M., & Rainwater, L. (Eds) (2004). The future of the family.
New York: Russell Sage Foundation.
Non-Normative Life Course Transitions 171
Neugarten, B., Moore, J. W., & Lowe, J. C. (1965). Age norms, age constraints and adult
socialization. American Journal of Sociology, 70, 710–717.
Parsons, T. (1954). Essays in sociological theory. New York: The Free Press.
Riley, M., Johnson, M., & Foner, A. (1972). Aging and society: A sociology of stratification (Vol. 3).
New York: Russell Sage Foundation.
Rindfuss, R. R., Swicegood, C. G., & Rosenfeld, R. A. (1987). Disorder in the life course: How
common and does it matter? American Sociological Review, 52, 785–801.
Ryder, N. B. (1965). The cohort as a concept in the study of social change. American Soci-
ological Review, 30, 843–861.
Settersten, R. A., Jr. (1998). A time to leave home and a time never to return? Age constraints
around the living arrangements of young adults. Social Forces, 76(4), 1373–1400.
Waite, L., & Gallagher, M. (2000). The case for marriage: Why married people are happier,
healthier, and better off financially. New York: Doubleday.
Wallerstein, J. S., & Blakeslee, S. (1989). Second chances: Men, women, and children a decade
after divorce. New York: Ticknor & Fields.
Wilensky, H. (1961). Orderly careers and social participation. American Sociological Review, 26,
521–539.
Winsborough, H. (1979). Changes in the transition to adulthood. In: M. W. Riley (Ed.), Aging
from birth to death: interdisciplinary perspective (pp. 137–152). Boulder, CO: Westview
Press.
FRANK F. FURSTENBERG172
THE SECRET OF TRANSITIONS:
THE INTERPLAY OF COMPLEXITY
AND REDUCTION IN LIFE
COURSE ANALYSIS
Katherine Bird and Helga Kruger
1. DEFINING THE ISSUE
In a sociological perspective, transitions focus our attention on a segment of
the life course in which a biographically significant change of social posi-
tioning occurs. This apparently simple statement masks the complications
involved, which stem from framing the concept within and between disci-
plines, within and between methodologies, and within and between life
course assumptions.
Two recent articles (Elder, Johnson, & Crosnoe, 2003; Marshall &
Mueller, 2003) outline the development of the life course perspective from its
humble beginnings to an established discipline. Analogous to the continual
development of overall life course theories, the perception and conceptu-
alisation of transitions has also diversified. The endeavour has certainly not
been completed, since complexity arises on many fronts: on the macro-level,
status passages refer to institutional resources and guidelines for attaining a
new state (Heinz, 1996), while, on the micro-level, status passages are per-
sonally conceptualised and modified by biographical actors. Deciphering the
Towards an Interdisciplinary Perspective on the Life Course
Advances in Life Course Research, Volume 10, 173–194
Copyright r 2005 by Elsevier Ltd.
All rights of reproduction in any form reserved
ISSN: 1040-2608/doi:10.1016/S1040-2608(05)10006-9
173
occurrence, timing and variability of transitions (cf. Hogan, 1981; Rindfuss,
Swicegood, & Rosenfeld, 1987) reveals nation-specific diversity, and so does
the explanation of how transitions link institutions and actors by defining
timetables, and entry or exit markers. And when Elder (1998, p. 958) es-
tablishes ‘‘transitions are always embedded in trajectories that give them
distinctive form and meaning’’, he draws attention to another level of com-
plexity: the arbitrariness of their interpretation in both the interdependen-
cies of status accumulation and the different perspectives related to internal
(biographical and personal) success stories and external evaluations.
Given the extent of these problems within an emerging discipline, it is no
wonder that more often than not in the literature on transitions we en-
counter pragmatic strategies associated with specific theoretical objectives.
Examples include:
1. To draw attention to the course and prolongation of transition processes
before the subsequent status is reached, which calls to mind the contro-
versies about individualisation, such as the loosening of institutional ef-
fects on biographies (Beck, Giddens, & Lash, 1996), or just the expansion
of periods of personal uncertainty about status passages from one insti-
tutional grip to the next (Heinz, 1999; Blossfeld, 1985).
2. To consider the biographical timing of events, an endeavour that has been
greatly enriched by approaches to the interlacing of biological and social
clocks as internalised external gatekeepers for windows of opportunity,
and calls for the investigation of multiple temporal dimensions as frames
of reference (Settersten, 2003; Settersten & Mayer, 1997; Mortimer,
Oesterle, & Kruger, 2004).
3. To focus on the often neglected socio-personal framing of transitions and
stages by the standardising forces of transitions in one’s partner’s life
which afflict your own life (Moen, 2003a; O’Rand & Farkas, 2002;
Sørensen, 2004). Hagestad’s conceptualisation of durations and transi-
tions (Hagestad, 1992, see also Hagestad & Neugarten, 1985) deciphers
even ‘countertransitions’ in order to underline the occurrence of status
changes arising from transitions on the part of socially relevant associates,
such as being transported into grandmotherhood by a daughter’s giving
birth, or being deprived of your marital status by your partner’s death.
Reduction of transitions occurs by means of theoretical assumptions and
the correspondent definitions of transition markers as well as by means of
the postulated empirical abundance of the segment under scrutiny. On the
basis of the theoretical approaches outlined above we can identify ‘inline
KATHERINE BIRD AND HELGA KRUGER174
transitions’; for example, the analysis of transitions with respect to one life
course domain (see example 1 above), in contrast to ‘competing transitions’;
for example, the analysis of context norms and biographical opportunities
(see example 2) – very often relevant in female decisions in favour of or
against motherhood – and ‘coupled transitions’; for example, transitions in
linked lives (see example 3).
Reduction occurs methodologically by means of the data collection and
analysis tools employed. In this context, it is important to point out that
from the event history analysis perspective – and this applies easily to de-
mographers as well – a transition is minimised by defining it as the occur-
rence of a precise event, such as leaving school. Thus the terms ‘transition’
and ‘event’ become synonyms. The other extreme is to maximise transitions
into changes over a specific duration in life, such as the transition from
childhood to adolescence, which opens the door for broader develop-
mental perspectives and associated scholars. Between the two extremes are
transitions of much shorter duration, such as from company manager to
being unemployed, or from secretary to housewife. And although Elder
points out that every transition can be broken down into ‘‘a succession of
mini-transitions or choice-points’’ (1998, p. 958), we suggest moving away
from analysing the ordering of transitions in a linear fashion (cf. Sackmann
& Wingens, 2001) and to reconsidering complexity instead.
This endeavour leads us to concentrate on transitions within and between
domains of social participation. Given that the individual’s simultaneous
participation in other social domains will both affect the transition under
study and be affected by it; we wish to advance Levy’s very early suggestion
(Levy, 1977, further developed in 1997) to conceive of the life course as a
movement through varying participation profiles. The fact that pathways
are layered within life domains introduces personal and structural contin-
gency into transitions, which can easily be overlooked. By seeking to reduce
complexity, we might accentuate the education, employment and retirement
history, and simultaneously neglect transitions within the dynamics of fam-
ily life, or within the health and illness history of a life. However, each of
these histories are subjected to normative and institutional transition pro-
grammes that themselves follow a logic specific to that domain of life, and
which ‘dissect’ the life course into relatively separate but still interwoven
patterns of behaviour. Even in the case of apparently discrete transitions,
the analysis might have to take into account the social interplay of change
and continuity with respect to other life domains, such as, working while
still at school (Mortimer, 2003), or, more formally, undergoing vocational
The Secret of Transitions 175
training as dual participation (Kruger, 1999), or being at work while retired
(Moen, 2003b).
Individuals experience transitions as governed by the different logics of
distinct life domains, and it is the interplay of life domains within transitions
that attracted our interest. Against this background, we propose that there
are ‘interlacing transitions’ that require attention. If we assume that a tran-
sition in one life domain is interlaced with simultaneous participation proc-
esses in other life domains, some of these abide by socially defined criteria of
mutual exclusion producing contradictory forces that push or pull in dif-
ferent directions. We argue that we need to look deeper and further in order
to disentangle the various strands of structure and agency that underlie a
specific life course arrangement.
Introducing the category ‘gender’ facilitates the analysis of interlacing
transition processes. We have therefore decided to focus on women’s
lives. We will first outline the theoretical basis as their dual integration into
society via the labour market and the family to provide a starting point for
unravelling the dynamics of two strands of the life course, which, in coun-
tries with a strong life course regime like Germany, produce gender in-
equality (Kruger, 2003). The empirical data presented in the third section
offer a new look at the classical transition from employment to homemaking
on the birth of a child. The change of perspective reveals mainstream status
subsumption processes, which mask the different continuity demands of
distinct life domains. The second set of empirical findings focuses on the
example of official and actual marital status to examine the dissonance
between institutional status definitions and the reality of lone motherhood.
In the final section, we draw some conclusions that challenge sociologists to
reconsider the social structuration of life as well as the adequacy of our
analytical tools for capturing the hierarchical dependency of transitions
between different social domains.
2. LAYERED LIFE COURSE PATTERNS:
THEORETICAL ADVANCES ON THE FEMALE CASE
The simultaneous relevance of more than one life course domain in women’s
lives was highlighted 18 years ago by Regina Becker-Schmidt, who coined
the phrase ‘dual integration into society’ (doppelte Vergesellschaftung,
Becker-Schmidt, 1987). She was referring to the contradictions inherent in
women’s socialisation and participation in society as both a future worker
KATHERINE BIRD AND HELGA KRUGER176
and as a caregiver. Much has been said about the objective problems of
seeking to combine these two tasks, especially in terms of timetables: for
kindergarten, school, paid work and so forth. The qualitative research con-
ducted by Becker-Schmidt and her colleagues also revealed the subjective
problems associated with attempting to solve this balancing act: the women
studied were ambivalent towards both types of social participation. They
could neither say that paid work was the most important thing in their lives
nor were their families. To quote the title of the book: ‘‘One is not enough –
both are too much’’ (translation of the original title: ‘‘Eines ist zu wenig –
beides ist zu viel’’, Becker-Schmidt, Knapp, & Schmidt, 1984).
The significance of these findings was to underline how separate fields of
social participation are simultaneously relevant in individual life courses,
but framed in a contradictory manner. Although Becker-Schmidt and others
have made an important contribution to understanding the complexity of
women’s lives, their ‘dual integration into society’ and the ensuing ambiv-
alence is, in most life course approaches, ignored in favour of conflict-
reducing simplicity.
Entanglements are eliminated by defining one possible life domain as
dominant for each sex and thus blending out all others. The ‘economic
theory of the family’ suggests that the primary function of the family is best
achieved by a gendered division of labour between a male breadwinner and
a female caregiver (for a concise summary of this approach see Blossfeld &
Drobnic, 2001).1 Catherine Hakim (2001) has developed the ‘preference
theory’, which turned the conflicting demands of paid and family work into
an orientational concept of a ‘‘personal work-lifestyle preference’’ (Hakim,
2001, p. 16).2 Following in the early life course tradition of comparing the
timing of events across birth cohorts, Karl Ulrich Mayer indirectly sup-
ported this view by stating in 1998 that on the basis of indicators such as age
at marriage or at first birth, or the family dependent phasing of employment,
female life courses were ‘‘highly standardised and relatively homogenous,
but – with the exception of marriage – not institutionalised and differen-
tiated’’ (Mayer, 1998, p. 447).
The reduction of structural ambivalence in women’s processes of societal
integration to ‘rational choice’, ‘preferences’, or the statement of a lack of
institutionalisation and differentiation in female life courses per se prompts
the question of whether these presumptions primarily result from a theo-
retical and methodological artefact relating to the underestimation of the
interlacing of transitional regulations in the two life domains of the labour
market and the family. To examine this question further we will explore
some empirical findings concerning two family related transitional segments
The Secret of Transitions 177
in female life courses: the transition to motherhood and the transition to
lone parenthood.
3. EMPIRICAL FINDINGS
3.1. Interlacing Transition I: A Baby Break
The event ‘giving birth’ undoubtedly marks a turning point in patterns of
female employment and family life, but is usually reduced to just the oc-
currence of a series of labour market exits and re-entries.3 If we move away
from a construction of the life course as a sequence of discrete transitions
towards a conceptualisation of interlacing transitions involving changing
participation profiles in the juxtaposed layers of life domains, we uncover
societal transition programmes that seek to subsume specific life course
domains under their own particular logic. In searching for the strands rel-
evant to transitions around motherhood, we therefore decided to integrate –
in addition to labour market exits and re-entries – the instruments developed
by the state for shaping labour market ‘timeouts’ on the occurrence of the
‘event’ of giving birth, and thereby incorporated the ‘negative space’ occu-
pied by leave-taking rules into the analysis.
This broader view allows us to consider three historical forces with op-
posing or contradictory objectives that may act out their effects on ‘pref-
erences’. These are (a) The wide-scale expansion of part-time jobs that was
intended to pull mothers back into paid employment, whereas (b) occupa-
tional career structures push them out. In addition, the introduction and
extension of extended periods of (c) maternity and parental leave will pull
and push them for sequenced time spaces into one domain or the other.
To investigate the effects of the three life course structuring principles we
collected life course data on 2,130 West German women from three ap-
prenticeship-completion cohorts: 1960, 1970 and 1980.4 At these dates the
women were aged between 18 and 20 years and had just completed training
in one of the top 10 occupations for women in Germany.5
During the window of observation for these three cohorts, three types of
maternity or parental leave were in force. Initially, new mothers were for-
bidden from working shortly before and after a birth under the motherhood
protection law (Mutterschutz of 8 weeks). In 1979, a motherhood leave
(Mutterschaftsurlaub, literally ‘motherhood vacation’) of 6 months was in-
troduced. The extension beyond 8 weeks was optional, as was the longer
‘child-rearing leave’ (Erziehungsurlaub, literally ‘child-rearing vacation’)
KATHERINE BIRD AND HELGA KRUGER178
introduced in 1986. Initially lasting for 10 months, child-rearing leave was
extended in a stepwise fashion to 3 years in 1992 and could be taken by
mothers or fathers. With the introduction of this last measure, new oppor-
tunities for configuring the duration of leave and changing gender-specific
participation were established.
For the women studied, the median duration between completing training
and the birth of the first child varied between 5 and 8.5 years, depending on
occupation and cohort. To capture the effects of the different leave regu-
lations, the mothers were regrouped into ‘motherhood cohorts’ according to
which type of leave they could take when they gave birth: the motherhood
protection cohort, the motherhood leave cohort and the child-rearing leave
cohort.
During the period covered by the study, nearly no change was observed in
the integrational capacity of the labour market segments open to people
trained in the 10 occupations investigated. However, the expansion of part-
time working did cause considerable change, which was accompanied by
shifts in the cultural contexts of women’s lives. For the oldest women in the
sample, the dominant norm was that of the stay-at-home wife and mother
(as Hradil put it in 1992: The ‘Golden Times of Normality in Family Life’.
See also Born, Kruger, & Lorenz-Mayer, 1996). Gradually, this norm
transformed into acceptance of employment during motherhood, especially
part-time, accompanied by the official recognition in the state’s statistics
of the true extent of mothers’ paid work assisting in the family business
(Willms-Herget, 1985). The then empirically proven younger cohorts’ visibly
higher employment participation fitted well into individualisation theories
that highlighted how, as old ties that anchor females firmly into a specific
position in family life have dissolved, diversity and discontinuity both within
individual life courses and between gendered life courses in general have
increased (Beck & Beck-Gernsheim, 1994).
In spite of these well-known interpretations, by looking only at women’s
actual leave-taking behaviour,6 we found that with the new options for
lengthening and sharing parental leave, female life course patterns became
more, not less, standardised – and they followed occupational career pat-
terns, not part-time ones. The proportion of mothers who stopped working
(that is, took leave or quit their job) within 6 months of giving birth varied
according to both motherhood cohort and to occupational career patterns,
as shown in Fig. 1.
The ‘stopped working’ percentage shows that the motherhood protection
cohort (black bars) not only did not stop work as much as normatively
expected – normative frames would lead us to expect nearer to 80% at least
The Secret of Transitions 179
(see Wurzbacher, 1951; Myrdal & Klein, 1956/1968; Pross, 1976) – but in
addition, they did so in line with the specific opportunity structures asso-
ciated with the occupation trained for. The inclusion of the other two
motherhood cohorts highlights how – as the length of leave available was
extended – an increasing proportion of women stopped working when they
had a child, but again in the same occupationally specific manner.
It should be noted that in the younger cohorts, leave taking was optional
and not compulsory. But we can see that not just the birth itself but the
state’s framing of this event has encouraged women to realise this option at
this particular time. The frequency of the transition from paid employment
to family work within 6 months of giving birth has in fact increased – in
spite of the assumption of the younger cohorts’ higher commitment to paid
work.
Additional event history analyses of our data on the rate of transition
from employment to a family break on the birth of a child confirmed that
changing leave entitlements have significantly (at the 1% level) increased the
likelihood of a labour market exit for mothers in the younger cohort,
whereas those not entitled to extended leave were least likely to make this
transition (Bird, 2004). Other data also support this finding. Beckmann and
Kurtz (2001), for example, report that 84% of West German women took
advantage of longer leave regulations.
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
80.0
90.0
100.0
Nurse Bank Office Ind. office Hairdresser Dr. asst Wholesale Spec. sales Retail Hotel
Occupation trained for
Pro
port
ion
of
moth
ers
stop
pin
g w
ork
(%
)
Mothers' protection Motherhood leave Child-rearing leave
Fig. 1. Percentage of Mothers Stopping Work within 6 Months of Giving Birth to
their First Child by First Occupation and Motherhood Cohort.
KATHERINE BIRD AND HELGA KRUGER180
These findings contrast with Mayer’s (1998) assumptions about the lack
of institutional effects on transitions in women’s lives, as we clearly discern
two equally powerful institutional programmes at work – leave regulations
and occupational specifics – which contradict each other, but both also
disprove the effects of an assumed lower or higher commitment to paid
employment. How can we conceptualise these facts?
(a) From status definitions to ambiguity. An event within one strand of the
life course (such as giving birth) can be a catalyst for a labour market exit,
but is also interwoven with a second strand – the state’s regulation of
maternity leave. But the second strand (state regulations) used to remain
hidden or its effects were underestimated, since many of these exits are
masked if only the event of giving birth is related to the mother’s labour
market status. In Germany, the institutionalised participation in the labour
market does not formally change while on leave, since leave-takers still
enjoy the status of being employed.7 If the resulting status subsumption is
analysed with a method that permits only one activity at a time (such as
event history analysis), the conclusions drawn will be inaccurate. If, how-
ever, we consider the mothers’ factual participation profiles rather than their
formal employment status, in the younger cohorts we actually find an in-
creased proportion of stay-at-home mothers dedicated to family work and
not to paid employment.
(b) From events to echo effects. The state’s regulation of parental leave
produces a time frame for the transition back to employment, which impacts
on the layered female life trajectory. With the exception of the small per-
centage of women who did not take advantage of prolonged exit oppor-
tunities, our data revealed the high degree of conformity between length of
leave available and length of leave taken (Bird, 2004).8 These findings raise
questions for both psychologists and sociologists concerning the meaning of
a longer break for the early period of motherhood.
We can assume that an extended absence from the labour market affects
the mothers’ perception and memory of the early period of motherhood. It
may provide the feeling of having now shared enough time with the baby, so
that the mothers can return to paid work without a bad conscience, or such
an absence could lead to increased insecurity about a return to the public
sphere in general, or influence their confidence in their ability to perform well
at work. Research is also required into how mothers personally anticipate
reconciling work and the family as the child grows older (see the timing of
debates on rising ‘overprotection’ in mothering, such as Schutze, 1986).
Qualitative data collection could provide illuminating answers to these
questions if the women in the oldest cohort who took a 3-year break under
The Secret of Transitions 181
the motherhood protection law were compared with the youngest mothers
who took a 3-year child-rearing leave. More generally, the factual masking
of the interlacing transitions between paid and family work has meant that a
whole range of important research questions have been neglected.
(c) From labour market to occupational structure. In all three cohorts the
mothers’ behaviour follows an occupational pattern. Hence, a third strand in
this transition is, undoubtedly, the first occupation trained for.9 In the anal-
ysis illustrated here, the hairdressers stopped work far less often than, for
example, the bank or office employees. Another interesting finding on the life
course structuring effect of the first occupation was that the fertility rates also
varied between the women who trained for different occupations. The ex-
planation for these phenomena is to be found in the market rules (labour
supply) and normality assumptions (including age norms) for the labour
force in the German service sector (for more details see Born et al., 1996).10
Occupational specifics are reflected in whether or not women had children
and whether they stopped working when they did so. Their occupations
negate the rules of cultural change in the acceptance of mothers’ employ-
ment. Over the different cohorts, women’s behaviour is dominated by the
same occupational patterns, and so are the effects of the motherhood tran-
sitions on the long-term female employment outcomes (Bird & Gottschall,
2003). All these findings are undoubtedly evidence of the impact of insti-
tutionalisation programmes in female life courses and equally refute the
importance of preferences in Hakim’s sense.
In summary, our analysis highlights that it is important to expand the
depth of focus – in this case to discover the overlapping of the state’s
mediating regulations with the peculiarities of the occupation trained for.
This implies overcoming or questioning official employment status defini-
tions and identifying the interwoven strands of a ‘setting in change’, also
including psychological processes, and to investigate their influence on fur-
ther transitions around labour market exits and entries.
The remarks on the need to question official status definitions that reduce
complexity also holds for our second empirical example, although it refers
to another transition in a setting in change.
3.2. Interlacing Transition II: The overlapping realities of
marital status and lone motherhood
The following section focuses on motherhood and marital status, which are
the classical transitions for describing change in female life courses across
KATHERINE BIRD AND HELGA KRUGER182
time (see, for examples, Huinink, 1991; Tolke, 1989; Mayer, 1998). In
Germany per official social welfare law (and transportation into everyday
life knowledge as well as into life course research), the legal status ‘married’
excludes the co-status of lone parenthood, which being single, divorced or
widowed would admit.
In order to investigate the category ‘marital status’ in its relevance for
exits or entries into lone motherhood, our research project took a two-fold
approach. In the standardised questionnaire, we first asked the respondents
for dates of marriage, divorce, widowhood, and several pages later we of-
fered a life-history calendar which opened up the chance to mark time slots
in simultaneously occupied fields in their lives (see Note 4). We compared
the durations of lone motherhood with the data on marital status
(Erzberger, 2001a). A total of 261 women provided apparently contradic-
tory information on their marriage and lone motherhood histories. An
analysis of these cases identified four combinations of both (illustrated in
Fig. 2), all consisting of somewhat confusing participational patterns be-
tween the beginning and end of periods of being married and being a lone
parent (i.e., solo).
The chart is a symbolic representation of the arrangements of layered
periods found in the data; the length of the blocks symbolises the overlap of
statuses in question rather than the actual durations in each. The vertical
lines represent the event of official change in marital status. The example
Type 1
Married
Solo
Type 2
Married
Solo
Type 3
Married
Solo
Type 4
Married
Solo
Fig. 2. Correspondence of Marital Status and Lone Motherhood.
The Secret of Transitions 183
illustrates vividly the lack of synchronicity in interwoven strands of the life
course and poses a challenge for the classification of these phenomena:
In Type 1, the phase as a lone mother starts before the marriage ended,
and ends before a new marriage is recorded. This pattern was found for 114
or 44% of the lone mothers. The average time spent as a lone mother while
still married and not married again was 2 years and 2 months. If we follow
the traditional logic of transitions, the question arises of which event ‘causes’
lone parenthood and which one ends it. The law would say: from the date of
divorce until the date of a new marriage. This clearly does not reflect the
empirical reality of women’s lives. If we consider lone parenthood to start
with the separation of the partners, then we are still confronted with the
question of when does separation start: When the partner moves out, when
the decision to move out is made, or when the first irreconcilable differences
emerge? And then, the same questions arise with respect to the event of
ending lone motherhood.
Type 2 is even more confusing. The phase of lone motherhood occurs
during the marriage. Such a pattern even occurs fairly often; it was observed
for 72 mothers (28% of the sample). How does this situation arise? One of
the project members suggested that the husband might be absent for a time
because of his work, perhaps as a captain on a ship. Another member
suggested that he could have been in prison. A third proposed that a mar-
ried woman separates from her husband and becomes a lone mother, then
later she moves in with a new partner and so ceases to be a lone parent, but
is still married to the first man.
The interpretation of the two remaining types (marriage as an unsynchro-
nised interlude in lone motherhood or continuing lone motherhood after
marriage) also results in unsatisfactory explanations that we do not need to
expand on here. We maintain that it is impossible to decide which of the many
plausible explanation is in fact the correct one, and whether only one expla-
nation would be valid for all the women in a group. The important lesson to
learn from this example is that if we use marital status as a proxy for lone
parenthood, then analyses based on this assumption will be inaccurate.
Similar to our empirical findings on the interlacing transition to moth-
erhood, other data from our previous qualitative studies within the Bremen
Life Course Centre revealed many more important transitions in women’s
lives, which again are swept over by the dominance of analytical conven-
tions. Examples include:
� Women who left paid employment in order to reduce stress factors and
so better their chances of becoming pregnant, but it did not happen. In
KATHERINE BIRD AND HELGA KRUGER184
this case, a transition had been made (employment to homemaking) but
the anticipated event (birth) had not occurred.� Women who returned to employment to be financially independent in
anticipation of a divorce, but it did not happen.� Again others who left paid employment, because their child was doing
badly at school, but no visible event emerges.
Evidently, transitions occur in many cases, but they often cannot be asso-
ciated with a visibly corresponding explanatory event, which influences the
perception of transitions in sociological research. If the analysis of transitions is
minimised to the study of events, then we end up shooting wide of the mark.
The empirical examples presented here have sought to clarify what is
meant by the term ‘interlacing transition’. In the first example, status sub-
sumption masked empirical reality. In the second example, using an official
definition for an event masked its occurrence or non-occurrence. Both ex-
amples show that the reduction of complexity in female life courses can
easily seduce researchers into superficial analyses and to working with mis-
leading causal assumptions about the logic of transitions within institutional
constraints. Therefore, instead of seeking to reduce complexity we have
focused on areas where complexity has been reduced to such an extent that a
distorted perception of reality has emerged.
4. DISCUSSION: INTERLACING TRANSITIONS –
WEAVING THE STRANDS OF A LIFE COURSE
Transitions involve structural interventions in the shaping of the life course
and precipitate reorganisation processes affecting the self within layered
frames of social order. Many authors have made statements with respect to
the timing and sequencing of transitions, to their relation to trajectories, and
to gatekeeping practices of institutions that provide standardisations and
normality assumptions about which transition is expected, which regulations
are to be followed, and how goals are personally and socially combined (for
an overview see Mortimer & Shanahan, 2003; Heinz & Marshall, 2003).
Although it is an indispensable goal of sociological theory to reduce the
complexity of individual lives to a manageable explanation of human life
course patterns, simplification should not be taken so far that the distinguishing
features of lives lived simultaneously in several social fields are eclipsed.
To draw attention to the interweaving of different strands of social partici-
pation we introduced the term ‘interlacing transition’. The reintroduction
The Secret of Transitions 185
of the necessary intricacy has implications for a multi-dimensional study of
transitions. We have to take into account:
1. Theories shape awareness and define timeframes. Transitions are most
appropriate for probing the multi-layered facets of the life course. But it
depends on our basic assumptions about life course patterns whether we
consider it worthwhile to devote sufficient attention to the turning points
that we discover and permit a large enough window of observation for
sounding out their depths. As the empirical examples showed:
� Regarding the birth of a child as the core of an interlacing transition
reveals the regulations and expectations for behaviour to be found in the
working environment (occupational specifics), the legal regulation of paid
employment (leave regulations) and social norms (cultural role models for
motherhood). These deeper layers of the transition are often overlooked
because the theoretical treatment of simultaneously relevant participation
domains has been underdeveloped.� A foreshortened perception of the transition to lone motherhood that only
focuses on official status is not able to reveal the variation in real expe-
rience found if we consider this to be an interlacing transition. Enlarging
the window of observation beyond the moment of change in official status
aids the identification of the topology of the transition and may indicate
social change.
Other researchers, primarily in North America, are following a similar
strategy. Phyllis Moen (2003b) and her associates (Han & Moen, 2001) are
studying retirement as an interlacing transition. The strands that are rel-
evant to the timing and form of retirement include career pathways and the
opportunities and restrictions presented by living with a partner. As Moen
(2003b, p. 269) pointed out, ‘‘[the transition to retirement] is a process
embedded in a number of overlapping contexts’’.
2. Interlacing transitions can reframe contemporary discourse. Searching for
and following the strands of an interlacing transition is a good guide for
sociological life course research to expose the contradictory forces of the dif-
ferent societal programmes that can point the life course in different directions.
The ambiguity and contradictions inherent in the female life course offer
ample opportunities for identifying the interweaving of different social
domains. The lack of a theoretical framework for managing the contradic-
tory forces produced by interlacing transitions has meant that, at least in
KATHERINE BIRD AND HELGA KRUGER186
Germany, the analysis of male life courses might have been oversimplified.
The increase in more flexible working practices and the emerging ideal of the
‘‘new father’’ indicate that the contradictory strands in women’s lives are
increasingly finding their counterpart in men’s lives. A new framework for
analysing the issue of reconciling work and the family, for example, may
facilitate its perception as a problem that involves mothers and fathers
equally, along with their employers and the legal opportunities and restric-
tions imposed by the state.
3. The role of the state. The perception of interlacing transitions underlines
Mayer and Muller’s (1986) early proposal that, if we are to fathom the
significance of transitions for the shape of the life course, then – at least in
countries with a high degree of life course standardisation – we need an
analysis that incorporates the role of the state. They elaborated the effects of
state-run social insurance introduced to protect the individual worker
against the risks of work-related accidents, illness, unemployment and old
age. Other researchers have focused on the state’s intervention in the life
course by structuring the education, employment and retirement trajectories
(Kohli, 1985; Blossfeld, 1985; Heinz, 1999). The state’s power to anchor
status definitions in legislation and in formal distinctions (in certificates and
documents) should not be underestimated. But this is not the end of the
story. We can expand on these previous analyses to conceptualise the state
as an important intervening power in shaping interlaced transitions in the
family and employment trajectories.
On the other hand, the second empirical example on lone motherhood
highlighted the inadequacy of relying solely on official status definitions
to determine when and whether co-transitions in family life occur.
Even though the state can frame a specific transition (both normatively
and administratively), we should not, without reason, assume conformity
of individual action. As Elder et al. (2003, p. 8) pointed out ‘‘Individuals
choose the paths they follow, yet choices are always constrained by the
opportunities structured by social institutions and culture’’. Too rigid a
focus on the dimensions of state action can mask the processual character of
transitions or blend out turning points that should be considered transitions,
but have not yet found their appropriate place in theories of the life course.
4. Re-framing the instruments of transition analysis. We can only capture the
contingent and multi-layered nature of interlacing transitions if we use em-
pirical tools that are more sensitive to complexity.
The Secret of Transitions 187
The collection of quantitative data needs to be reconsidered with respect
to the extent to which a questionnaire permits the recording of synchronicity
within different layers of the life course. Analogously, the analysis of these
data needs to include opportunities for reversing the practice of subsuming
simultaneous status positions under one dominant one.
In classic event history analysis, using large data sets such as the German
Socio-Economic Panel (SOEP), the state space is defined by the researcher
as consisting of mutually exclusive statuses. Discrete transitions from one
state to another can be analysed, but not the attainment of the target state
without leaving the origin state. Although such methods of data reduction
certainly facilitate its analysis, we would argue that this approach not only
masks multi-layered activities, but also defines partial transitions as of sec-
ondary importance for the timing and ordering of events.11 In order to
overcome the temptation to reductionism inherent in this method, addi-
tional methods of data analysis are necessary to assess the long-term dis-
tribution of non-discrete participation profiles. Explorative techniques such
as optimal matching of large-scale data sets (Abbott, 1995; Erzberger,
2001b; Aisenbery, 2000) are extremely useful in mapping out the actual
sequence of activities during a specific timeframe and thus also offer a
means to compare the layeredness of life sequences with simple event pat-
terns (Erzberger & Kluge, 2000; Erzberger, 2001b). Even within an event
history analysis, exploratory analyses can facilitate the development of a
parametric model (Wu, 2003).
In interpreting the results of either explorative analyses or parametric
modelling, however, the limitations of a quantitative approach become ap-
parent. The relationship between quantitative and qualitative research is
more than a simple either-or (cf. Kluge & Kelle, 2001). Without the con-
tribution of our previous qualitative research in uncovering the variety and
steering forces in female life courses we would have missed the importance
of interlacing transitions in women’s lives, and the discovery of new ques-
tions to be studied. In addition, it would have been much harder to con-
struct an instrument for quantitative research that did not force women to
subsume the layers of their lives into just one, but to expand on events and
durations in juxtaposed time spans.
5. Transitions: Definitions in progress. We believe it to be self-evident that
the conceptualisation of the term ‘transition’ should not lead to a search for
a formal definition, but is an ongoing process being pushed forward by the
interaction of theoretical and methodological advances. As we learn more
KATHERINE BIRD AND HELGA KRUGER188
about how to study and understand lives, we automatically move away from
the simple guillotine-like perception of transitions.
Theories of the life course based on a reduction of complexity have cer-
tainly been useful in mapping out a rough plan. However, they can lead to
the ‘disappearance’ of specific transitions resulting in an oversimplification
that could explain the suggestion that the female life course is ‘‘not insti-
tutionalised and differentiated’’ (Mayer, 1998, p. 224), or Hakim’s attribu-
tion of women’s decision-making to wide-reaching preferences for either
employment, or family work (or both just part-time).
We can agree with Glen Elder (1985, p. 31) who prefers the term ‘‘tran-
sition’’ to ‘‘life event’’ because changes in the life course are not just sudden
events but are embedded in interceding, preceding and succeeding processes.
However, Elder’s reasoning stops just short of the mark because it does not
consider how an interlacing transition implies simultaneous changes of so-
cial positioning in other layers of the life course. An event in one strand may
have repercussions for the others, which themselves are continuously push-
ing or pulling the life course in specific directions.
The proposal to perceive transitions as the interlacing of transformation
processes in participation patterns and thinking in terms of status config-
urations rather than status changes will, hopefully, lead to more precise
empirical and theoretical approaches that will consign oversimplifications to
the past.
NOTES
1. In a recent edited volume comparing transitions between paid and family workin an international context, the authors referred to the economic theory of the familyas ‘‘the most highly developed and influential theory in this field so far available’’(Blossfeld & Drobnic, 2001, p. 16).2. For an extensive critique of this theory see, for example, Crompton and Harris
(1999).3. This can lead to neglecting the partner’s labour market opportunities, an over-
sight that is being corrected by, for example, studies on couples’ careers (Blossfeld &Drobnic, 2001; Elder, 1996; Marx-Ferree, 1997); on the pathways into retirement(Moen, 2003b; Han & Moen, 2001); or simply the husband’s income which, as ourstudy of older women revealed, provided the best argument in hindering women’sintentions to leave or return to paid employment (Born et al., 1996).4. The research was conducted within the Special Research Program ‘Status Pas-
sages and Risks in the Life Course’ at the Bremen Life Course Centre and wasfinanced by the Deutsche Forschungsgemeinschaft. The survey was restricted toWest Germany because the structure and normative context of the labour market for
The Secret of Transitions 189
women in East Germany was too different during the historical period investigatedto allow a meaningful cohort comparison.The data were gathered at the end of 1997 with a life-history calendar that was
designed to find out what the women actually did, rather than their official status atany particular point in time. For this reason we asked them to record all theiractivities at different times, rather than making them decide which was their principalactivity. By asking about different types and hours of employment, motherhoodleave and family breaks, and time spent supporting and caring for dependent familymembers, it was possible to gather information on parallel layers of activity overtime. Interestingly, there were many women who stated that their family activitycontinued long after they had returned to work, confirming the lasting importance ofdual social participation, while at the same time highlighting how fallacious it is toask women to only name their principal activity.5. The occupations were nurse, qualified bank employee, trained office employee,
industrial office employee, hairdresser, doctor’s clerical and medical assistant, officeemployee in a wholesale and/or export firm, specialist sales assistant (in a baker’s orbutcher’s), general retail sales assistant and hotel receptionist. In all three cohorts,between 65% and 76% of all female apprentices trained for one of these occupations.6. Men who have made real use of the new opportunity to take child-rearing leave
do not require further consideration here, because between 1986 and 2001 no morethan 2% of eligible fathers took leave (Schneider & Rost, 1998; BMFSFJ, 2002).7. Women (or men) who take child-rearing leave are now entitled to 3 years of
leave, or even longer if another child is born before leave ends. These women stillhave an employment contract and the guarantee of a job to return to. So, formally,these women are counted as employed in statistics used to calculate the employmentrate. Although the Statistisches Bundesamt has since 1996 made a distinction betweenpeople who are employed and those who are on temporary leave (including child-rearing leave), this distinction is rarely included in calculations of the female em-ployment rate. If no account is taken of temporary leave, then 48% of mothers withat least one child under three are in paid employment. If those on leave are excluded,the employment rate sinks to 30% (all figures taken from Beckmann, 2003, p. 6).8. Other research (Buchel & SpieX, 2002; Ondrich, SpieX, & Yang, 2002;
Beckmann & Engelbrech, 2001) has shown that highly qualified women and thosewho grew up in East Germany were more likely than other women to return to workbefore the end of their leave entitlement.9. Not only did the women train in this occupation, they also worked in it. After
completing training, 85% of the women worked full-time in this occupation.10. The normality assumptions can have particularly dramatic effects. In the
1960s a norm of celibacy for nurses was still prevalent, which at least partiallyaccounts for the high proportion of nurses without children in this cohort (33%)compared to the subsequent one (19%). Similarly, Cremer (1984) showed how theexpectation that hairdressers are young and attractive made it increasingly difficultfor mature women to remain in the occupation.11. Analyses of large data sets with sophisticated methods run the danger of de-
generating into what Hartmut Esser termed ‘variable sociology’ (Esser, 1996). Althoughthe variation between the variables may be statistically explained, the processes in thelife courses that generated them are reduced to possibly misleading causal assumptions.
KATHERINE BIRD AND HELGA KRUGER190
REFERENCES
Abbott, A. (1995). Sequence analysis: New methods for old ideas. Annual Review of Sociology,
21, 93–113.
Aisenbery, S. (2000). Optimal matching analysis. Opladen: Leske + Budrich.
Beck, U., & Beck-Gernsheim, E. (Eds) (1994). Riskante Freiheiten. Individualisierung in mod-
ernen Gesellschaften. Frankfurt am Main: Edition Suhrkamp.
Beck, U., Giddens, A., & Lash, S. (1996). Reflexive Modernisierung. Eine Kontroverse. Frankfurt:
Suhrkamp.
Becker-Schmidt, R. (1987). Die doppelte Vergesellschaftung – die doppelte Unterdruckung:
Besonderheiten der Frauenforschung in den Sozialwissenschaften. In: L. Unterkircher &
I. Wagner (Eds), Die andere Halfte der Gesellschaf, (Vol. 1, pp. 9–25). Wein: Verlag des
Osterreichischen Gewerkschaftsbundes.
Becker-Schmidt, R., Knapp, G.-A., & Schmidt, B. (1984). Eines ist zuwenig – beides ist zu viel.
Erfahrungen von Arbeiterfrauen zwischen Familie und Fabrik. Bonn: Verlag Neue
Gesellschaft.
Beckmann, P. (2003). Die Beschaftigungsquote: – (k)ein guter Indikator fur die Erwerb-
statigkeit von Frauen? IAB-Kurzbericht, 11/2003. Retrieved August 1, 2003 from http://
doku.iab.de/grauepap/2003/kb1103_langfassung.pdf
Beckmann, P., & Engelbrech, G. (2001). Die schwierige Balance: Frauen zwischen Beruf und
Familie. Personalfuhrung, 6, 118–129.
Beckmann, P., & Kurtz, B. (2001). Die Betreuung der Kinder ist der Schlussel. IAB-Kurzbericht
10. Nurnberg: Bundesanstalt fur Arbeit.
Bird, K. (2004). Reconciling work and the family: The impact of parental leave policies and
occupation on the female life course. Frankfurt: Peter Lang.
Bird, K., & Gottschall, K. (2003). Erosion of the male-breadwinner model? Female labor-
market participation and family-leave policies in Germany. In: H. Gottfried & L. Reese
(Eds), Equity in the workplace: Gendering workplace policy analysis (pp. 281–308).
Lanham, MD: Lexington Books.
Blossfeld, H.-P. (1985). Bildungsexpansion und Berufschancen. Empirische Analysen zur Lage der
Berufsanfanger in der Bundesrepublik. Frankfurt/New York: Campus-Verlag.
Blossfeld, H.-P., & Drobnic, S. (2001). Theoretical perspectives on couples’ careers. In:
H. P. Blossfeld & S. Drobnic (Eds), Careers of couples in contemporary society. From
male breadwinner to dual earner families (pp. 16–50). Oxford: Oxford University Press.
Born, C., Kruger, H., & Lorenz-Mayer, D. (1996). Der unentdeckte Wandel. Annaherung an das
Verhaltnis von Struktur und Norm im weiblichen Lebenslauf. Berlin: edition sigma.
Buchel, F., & SpieX, K. (2002). Form der Kinderbetreuung und Arbeitsmarktverhalten von
Muttern in West- und Ostdeutschland. Stuttgart: Kohlhammer.
Bundesministerium fur Familie, Senioren, Frauen und Jugend (BMFSFJ). (2002).
Bundesstatistik Erziehungsgeld 2001. Retrieved April 14, 2004 from www.bmfsfj.de/
RedaktionBMFSFJ/Abteilung2/Pdf-Anlagen/PRM-24020-Statistik-zum-Bundeserzie-
hungs.pdf
Cremer, C. (1984). Schonheit wird zur Pflicht. Friseurin – Beruflichkeit auf Zeit. In: C. Mayer,
H. Kruger, U. Rabe-Kleberg & I. Schutte (Eds), Madchen und Frauen. Beruf und Biog-
raphie (pp. 85–98). Munchen: DJI Verlag.
Crompton, R., & Harris, F. (1999). Employment, careers and families: The significance of
choice and constraint in women’s lives. In: R. Crompton (Ed.), Restructuring gender
The Secret of Transitions 191
relations and employment. The decline of the male breadwinner (pp. 128–149). Oxford:
Oxford University Press.
Elder, G. H. (1985). Perspectives on the life course. In: G. H. Elder (Ed.), Life course dynamics.
Trajectories and transitions, 1968–1980 (pp. 23–49). Ithaca: Cornell University Press.
Elder, G. H. (1996). The life course paradigm: Social change and individual development. In:
K. Luscher (Ed.), Examining lives in context: Perspectives on the ecology of human de-
velopment (pp. 101–139). Washington, DC: APA Press.
Elder, G. H. (1998).The life course and human development. In: R. M. Lerner, (Ed.),Handbook
of child psychology. Theoretical models of human development (Vol.1, pp. 939–991).
New York: Wiley.
Elder, G. H., Johnson, K. M., & Crosnoe, R. (2003). The emergence and development of life
course theory. In: J. T. Mortimer & M. J. Shanahan (Eds), Handbook of the life course
(pp. 3–19). New York: Kluwer Academic & Plenum Publishers.
Erzberger, C. (2001a). Uber die Notwendigkeit qualitativer Forschung: Das Beispiel der All-
einerziehungsdaten in quantitativen Daten. In: S. Kluge & U. Kelle (Eds), Methoden-
innovation in der Lebenslaufforschung. Integration qualitativer und quantitativer Verfahren
in der Lebenslauf- und Biographieforschung (pp. 169–188). Weinheim/Munchen: Juventa.
Erzberger, C. (2001b). Sequenzmusteranalyse als fallorientierte Analysestrategie. In: R. Sackmann
&M.Wingens (Eds), Strukturen des Lebenslaufs: Ubergang–Sequenz–Verlauf (pp. 135–162).
Weinheim: Juventa.
Erzberger, C., & Kluge, S. (2000). Reprasentativitat qualitativer Untersuchungen: Lebensverlauf-
muster als Ressource fur Auswahlentscheidungen. In: W.R. Heinz (Ed.), Ubergange. In-
dividualisierung, Flexibilisierung und Institutionalisierung des Lebensverlaufs. Zeitschrift
fur Soziologie der Erziehung und Sozialisation. 3. Beiheft, 298–313.
Esser, H. (1996). What is wrong with ‘‘variable sociology’’? European Sociological Review,
12(2), 159–166.
Hagestad, G. O. (1992). Assigning rights and duties: Age, duration, and gender in social in-
stitutions. In: W. R. Heinz (Ed.), Institutions and gatekeeping in the life course (Vol. 3,
pp. 261–279). Weinheim: Deutscher Studien-Verlag.
Hagestad, G. O., & Neugarten, B. L. (1985). Age and the life course. In: R. H. Binstock &
E. Shanas (Eds), Handbook of aging and the social sciences (pp. 35–61). New York:
Von Nostrand Reinhold Company.
Hakim, C. (2001). Work-lifestyle choices in the 21st century: Preference theory. Oxford: Oxford
University Press.
Han, S.-K., & Moen, P. (2001). Coupled careers: Pathways through work and marriage in the
United States. In: H.-P. Blossfeld & S. Drobnic (Eds), Careers of couples in contemporary
society. From male breadwinner to dual-earner families (pp. 201–231). New York: Oxford
University.
Heinz, W. R. (1996). Status passages as micro-macro linkages in life course research. In:
A. Weymann & W. R. Heinz (Eds), Society and biography. Interrelationships between
social structure, institutions and the life course (pp. 51–65). Weinheim: Deutscher Studien
Verlag.
Heinz, W. R. (1999). Job-entry patterns in a life-course perspective. In: W. R. Heinz (Ed.), From
education to work: Cross-national perspectives (pp. 214–234). New York: Cambridge
University Press.
Heinz, W. R., & Marshall, V. W. (Eds) (2003). Social dynamics of the life course. Transitions,
institutions and interrelations. New York: Aldine de Gruyter.
KATHERINE BIRD AND HELGA KRUGER192
Hogan, D. P. (1981). Transitions and social change: The early lives of American men. New York:
Academic Press.
Hradil, S. (1992). Die ‘‘objektive’’ und die ‘‘subjektive’’ Modernisierung. Der Wandel der
westdeutschen Sozialstruktur und die Wiedervereinigung. Aus Politik und Zeitgeschichte.
Beilage zur Wochenzeitung Das Parlament, B-29-30/92, 3–14.
Huinink, J. (1991). Familienentwicklung in der Bundesrepublik Deutschland. In: K. U. Mayer,
J. Allmendinger & J. Huinink (Eds), Vom Regen in die Traufe: Frauen zwischen Beruf und
Familie (pp. 289–317). Frankfurt/New York: Campus.
Kluge, S., & Kelle, U. (Eds) (2001). Methodeninnovation in der Lebenslaufforschung. Integration
qualitativer und quantitativer Verfahren in der Lebenslauf- und Biographieforschung.
Weinheim/Munchen: Juventa.
Kohli, M. (1985). Die Institutionalisierung des Lebenslaufs. Kolner Zeitschrift fur Soziologie
und Sozialpsychologie, 37, 1–29.
Kruger, H. (1999). Gender and skills. Distributive ramifications of the German skill system. In:
P. D. Culpepper & D. Finegold (Eds), The German skills machine. Sustaining comparative
advantage in a global economy (pp. 189–227). New York, Oxford: Berghahn Books.
Kruger, H. (2003). The life-course regime: Ambiguities between interrelatedness and individ-
ualization. In: W. R. Heinz & V. W. Marshall (Eds), Social dynamics of the life course.
Transitions, institutions and interrelations (pp. 33–56). New York: Aldine de Gruyter.
Levy, R. (1977). Der Lebenslauf als Statusbiographie. Die weibliche Normalbiographie in ma-
krosoziologischer Perspektive. Stuttgart: Ferdinand Enke.
Levy, R. (1997). Status passages as critical life course transitions. A theoretical sketch. In:
W. R. Walter (Ed.), Theoretical advances in life course research (pp. 74–96). Weinheim:
Deutscher Studien Verlag.
Marshall, V. W., & Mueller, M. M. (2003). Theoretical roots of the life-course perspective. In:
W. R. Heinz & V. W. Marshall (Eds), Social dynamics of the life course. Transitions,
institutions and interrelations (pp. 3–32). New York: Aldine de Gruyter.
Marx-Ferree, M. (1997). Gender, conflict and change: Family roles in biographical perspectives.
In: W.R. Heinz (Ed.), Theoretical advances in life-course research (2nd ed., pp. 123–137),
Weinheim: Deutscher Studien Verlag.
Mayer, K. U. (1998). Lebensverlauf. In: B. Schafers & W. Zapf (Eds), Handworterbuch zur
Gesellschaft Deutschlands (pp. 438–451). Opladen: Leske + Budrich.
Mayer, K. U., & Muller, W. (1986). The state and the structure of the life course. In: A. B. Sørensen,
F. E. Weinert & L. R. Sherrod (Eds), Human development and the life course: Multi-
discliplinary persepectives (pp. 217–245). Hillsdale: Lawrence Erlbaum Associates.
Moen, P. (2003a). Linked lives. Dual careers, gender and the contingent life course. In:
W. R. Heinz & V. W. Marshall (Eds), Social dynamics of the life course. Transitions,
institutions and interrelations (pp. 237–258). New York: Aldine de Gruyter.
Moen, P. (2003b). Midcourse: Navigating retirement and a new life stage. In: J. T. Mortimer &
M. J. Shanahan (Eds), Handbook of the life course (pp. 269–291). New York: Kluwer
Academic & Plenum Publishers.
Mortimer, J. T. (2003). Working and growing up in America. Cambridge: Harvard University
Press.
Mortimer, J. T., & Shanahan, M. J. (Eds) (2003). Handbook of the life course. New York:
Kluwer Academic & Plenum Publishers.
Mortimer, J.T., Oesterle, S., & Kruger, H. (2004). The transition to adulthood: Age norms,
institutional structures, and the timing of educational completion and parenthood. In:
The Secret of Transitions 193
R. Macmillan (Ed.), Advances in life course research (Vol. 9). The structure of the life
course: Standardized? individualized? differentiated?, Amsterdam: Elsevier Science.
Myrdal, A., & Klein, V. (1968). Women’s two roles. Home and work. London: Routledge &
Kegan Paul Ltd.
Ondrich, J., SpieX, K., & Yang, Q. (2002). The effects of maternity leave on women’s pay in
Germany 1984–1994. DIW Discussion Paper, 289. Berlin: Deutsches Institut fur Wirt-
schaftsforschung.
O’Rand, A. M., & Farkas, J. I. (2002). Joint retirement among US dual earner couples in the
1990s: Family and market influences on labor exit patterns. International Journal of
Sociology, 32, 11–29.
Pross, H. (1976). Die Wirklichkeit der Hausfrau. Reinbek bei Hamburg: Rowohlt.
Rindfuss, R. R., Swicegood, C. G., & Rosenfeld, R. A. (1987). Disorder in the life course: How
common and does it matter? American Sociological Review, 52, 785–801.
Sackmann, R., & Wingens, M. (2001). Theoretische Konzepte im Lebenslauf: Ubergang, Se-
quenz und Verlauf. In: R. Sackmann & M. Wingens (Eds), Strukturen des Lebenlaufs.
Ubergang – Sequenz – Verlauf (pp. 17–48). Weinheim/Munchen: Juventa.
Schneider, N. F., & Rost, H. (1998). Vom Wandel keine Spur – warum ist Erziehungsurlaub
weiblich? In: M. Oechsle & B. Geissler (Eds), Die ungleiche Gleichheit. Junge Frauen und
der Wandel im Geschlechterverhaltnis (pp. 217–236). Opladen: Leske + Budrich.
Schutze, Y. (1986). Die gute Mutter. Zur Geschichte des normativen Musters ‘‘Mutterliebe’’.
Bielefeld: Kleine.
Settersten, R. A., Jr. (2003). Age structuring and the rhythm of the life course. In:
J. T. Mortimer & M. J. Shanahan (Eds), Handbook of the life course (pp. 81–98).
New York: Kluwer Academic & Plenum Publishers.
Settersten, R. A., Jr. & Mayer, K. U. (1997). The measurement of age, age structuring and the
life course. Annual Review of Sociolology, 23, 233–261
Sørensen, A. (2004). Economic relations between women and men: New realities and the re-
interpretation of dependence. In: J. Z. Giele & E. Holst (Eds), Changing life patterns in
Western industrial societies (pp. 281–297). Amsterdam: Elsevier.
Tolke, A. (1989). Lebensverlaufe von Frauen. Familiare Ereignisse, Ausbildungs- und
Erwerbsverhalten. Munchen: DJI Verlag.
Willms-Herget, A. (1985). Frauenarbeit. Zur Integration von Frauen auf dem Arbeitsmarkt.
Frankfurt/New York: Campus.
Wu, L. L. (2003). Event history models and life course analysis. In: J. T. Mortimer &
M. J. Shanahan (Eds), Handbook of the life course (pp. 477–502). New York: Kluwer
Academic & Plenum Publishers.
Wurzbacher, G. (1951). Leitbilder gegenwartigen deutschen Familienlebens. Stuttgart: Enke
Verlag.
KATHERINE BIRD AND HELGA KRUGER194
LIFE COURSE TRANSITIONS AND
SOCIAL IDENTITY CHANGE
Nicholas Emler
The thesis to be developed in this chapter is that transitions in the life course
correspond to a particular kind of category change, namely shifts in social
identity. It is therefore built upon a hitherto largely unrepresented discipline
in the study of the life course, social psychology. The utility of a perspective
that places social identity at the centre of the life course is that it draws
attention to the kinds of events and circumstances associated with, and to
some degree responsible for, life course transitions; it helps make sense of
the consequences or effects of transitions; it directs attention to questions
previously relatively neglected in the life course literature and it has the
capacity to link the approaches of different disciplines to the analysis of the
life course.
The chapter will unfold as follows. First, the traditional view of transition
within developmental psychology will be briefly set out as a counterpoint to
the view recommended here. The concept of social identity will be elabo-
rated, noting the manner in which it is currently constructed in social
psychology, particularly in Social Identity Theory (SIT) and Self-
Categorisation Theory, and then emphasising what is believed to be the
key quality of social identity, the manner in which it connects individuals to
their social worlds. Discussion will then turn to the psychological elements
and processes in identity formation, taking the example of political identity.
Towards an Interdisciplinary Perspective on the Life Course
Advances in Life Course Research, Volume 10, 197–215
Copyright r 2005 by Elsevier Ltd.
All rights of reproduction in any form reserved
ISSN: 1040-2608/doi:10.1016/S1040-2608(05)10007-0
197
Finally, the chapter considers transition processes and associated changes in
social relations.
PROTOTYPICAL TRANSITIONS: THE LESSONS
FROM COGNITIVE DEVELOPMENT
The butterfly provides one of the more extreme and clear-cut examples of a
life cycle marked by transitions. Periods spent successively as a caterpillar, a
chrysalis and a butterfly are separated by quite rapid changes between these
states. This life cycle would seem therefore to capture the essential features
of transitions, namely that they are rapid, qualitative changes intervening
between longer periods of relative stability. Human life cycles contain noth-
ing resembling the dramatic physical changes that result in butterflies. Un-
questionably, however, they do involve changes, and at least some of the
changes appear to be qualitative. But are these changes also transitions? An
influential prototype for this kind of change has been the Piagetian model of
cognitive development (e.g., Piaget, 1952).
Until Piaget’s work found a wider audience in the Anglophone world (cf.
Flavell, 1963), two models dominated thinking about psychological devel-
opment. One had its roots in psychoanalysis, a perspective that represented
personality development as a sequence of qualitatively distinct stages. The
other originated in learning theory and regarded development quite differ-
ently. From the perspective of this model, change was considered to be
quantitative, not qualitative, and was taken to be gradual. Piaget’s ideas
were clearly closer to psychoanalysis in one respect; he described develop-
mental change as a sequence of qualitatively distinct stages, not as gradual
quantitative change. But in other important respects it was quite distinct.
Piaget provided a very detailed formal description of each stage, and his
descriptions appeared to be empirically verifiable (Flavell, 1963).
The power and persuasiveness of Piaget’s account lent popularity to other
stage models of psychological development, including models that addressed
the themes previously within the province of psychoanalysis. Examples in-
cluded Marcia’s (1966) reworking of Erikson’s (1950) stage theory of iden-
tity development and Loevinger’s (1976) analysis of ego development. But
one of the most influential stage models to follow the route pioneered by
Piaget was Kohlberg’s (1976) theory of moral development. It drew closely
on Piaget’s cognitive developmental theory in numerous respects. It adopted
his proposal that cognitive functioning takes the form of structural wholes,
NICHOLAS EMLER198
integrated systems of thinking or generalised strategies for relating to the
world, that each possess a degree of internal equilibrium. It also regarded
the general cognitive stages identified by Piaget as the foundations for cor-
responding general stages of moral reasoning. But of particular interest here
is the account Kohlberg provided of stage transitions.
Stage transitions, Kohlberg proposed, are qualitative changes in cognitive
structure precipitated by cognitive conflict or disequilibrium. The conflict
might be largely internal in origin – for example, mutually contradictory
conclusions generated by the same reasoning system – or external in origin –
for example, arising from difficulties of assimilating new experiences within
existing cognitive structures or of solving newly encountered problems
within existing frameworks. The critical feature of the transition points or
stage changes for Kohlberg was that they were inherently unstable and
unable to endure without resolution. The resolution would be the construc-
tion of a new equilibrium.
Kohlberg’s theory has been criticised on numerous counts. The criticisms
have included challenges to most of his central claims about stages in de-
velopment of moral reasoning, namely the claims that these stages are or-
dered in a progressive, universal, invariant and irreversible sequence. I am
less concerned here with the force of these challenges than with the appro-
priateness of his neo-Piagetian concept of transition as applied to develop-
mental change. It has turned out that empirically it is difficult if not
impossible to pin down a transition point in the sense of a relatively limited
period during which one form of cognitive functioning is being replaced by
another. Moreover, this proves to be the case for the kinds of cognitive
changes first described by Piaget, for example the change from pre-
operational to concretely operational thinking as well as the developmental
shifts in moral reasoning anticipated by Kohlberg, for example the change
from pre-conventional to conventional moral reasoning. Finding clear cases
of short, unstable transitional phases in the data has proved to be difficult.
In contrast, with respect to moral reasoning, Kohlberg’s own longitudinal
data (Colby & Kohlberg, 1987) indicate that what should in theory be a
transitional condition, one in which forms of moral reasoning correspond-
ing to adjacent stages are used concurrently, is actually both more common
and more enduring than the stable states – those using just one form of
moral reasoning or the other – between which it is supposed to intervene
only briefly.
In the case of both Piaget’s and Kohlberg’s stages the changes involved do
appear to be qualitative; in each case there is a category change in cognitive
functioning. But in the concept of developmental stage both Piaget and
Life Course Transitions and Social Identity Change 199
Kohlberg combine the proposals that psychological development involves
category changes and that category changes have the characteristics of
transitions; they are relatively abrupt shifts intervening between periods of
relative stability and continuity. This model of stages does not appear to be
verifiable in the cognitive domain. Are there other kinds of category change
during the life course that do have the properties to qualify them as tran-
sitions? A promising class of candidates would appear to be provided by
changes in social identity.
SOCIAL IDENTITY
Over the past 30 years social psychology’s thinking about social identity has
been dominated by a particular point of view, at least within Europe. This is
the view first articulated in detail by Henri Tajfel (1978) as Social Identity
Theory (SIT). Tajfel’s core idea, now widely accepted in social psychology
(see Brown, 1986), is that in essence people’s subjective sense of who they are
is to an important degree determined by the manner in which they define
themselves socially. More specifically, what matters is which social catego-
ries form part of their self-definition.
This argument grew out of research on prejudice and its connection to
stereotyping. Tajfel saw that stereotyping is a process of social categorisa-
tion; it involves assigning other people to categories and regarding members
of a common category as if they were qualitatively distinct from people
outside the category. In other words, categories have boundaries and at the
boundaries there are discontinuities. Tajfel (1969) drew on his work in cog-
nitive psychology that predicted cognitive and perceptual effects of catego-
risation; the boundaries can be artificially created by perceiving greater
discontinuities across imposed category boundaries and greater similarities
within categories than exist in reality – processes also referred to as, re-
spectively, contrast and assimilation, or sharpening and levelling. Just as
there is no objective discontinuity in the electromagnetic spectrum between
the areas conventionally labelled ‘green’ and ‘blue’ so also there is no ob-
jective discontinuity between people who are categorised, respectively, as
‘black’ and ‘white’. But social categorisation can only operate as an effective
way of organising and simplifying experience of the social world by per-
ceiving it as if this was the case.
Thinking about the link between stereotyping as social categorisation on
the one hand and prejudice on the other, led inevitably to thinking about
self-categorisation. It was recognised that various ways of categorising
NICHOLAS EMLER200
social objects – other people – tend to correspond, such that those who are
categorised together descriptively are also categorised together evaluatively
or emotionally – one feels similarly about them – and categorised together
behaviourally – one treats them similarly (Doise, 1978). But why should one
feel negative about particular categories and positive about others, and why
should one discriminate against the former and favour the latter? One very
strong reason for making the evaluative/emotional and behavioural dis-
tinctions in a particular direction is that any way of categorising the social
world that is comprehensive or exhaustive must include a category to which
the self belongs. And it would be natural to regard this category as better
than others, to feel more positive about it than others, to regard it as more
worthy than others and, when the opportunity arises, to treat it more fa-
vourably than others.
It soon became apparent that matters were not this simple. Among
the problems are that people do not invariably feel positive about the
social categories they inhabit (Giles & Powesland, 1975). Nor is it clear, as
was originally suggested, that in emphasising both the categorical distinc-
tiveness and superiority of their own categories people are primarily driven
by the desire to see themselves positively, to defend their own self-esteem
(Rubin & Hewstone, 1998). But for the present discussion this is less sig-
nificant than the fact that self-categorisations and social identifications of
the self became of concern in social psychology. In effect, the focus of
theorising and research shifted to recognition that social categorisation
shapes how people think about themselves and not just how they think
about others.
The significance of self-categorisations has been explored in detail by one
of Tajfel’s students, John Turner, and systematised as Self-Categorisation
Theory (SCT; Turner, 1987). SCT incorporates several key ideas. One is that
any individual potentially has access to multiple self-categorisations, such
that from situation to situation the salience and therefore the influence of
any one of these can vary. Another is that self-categorisation has a wide
range of consequences. The particular category that is salient at any moment
will shape affiliation and communication patterns, sensitivity to sources of
influence, choices and judgments, accounts given and opinions expressed.
Perhaps the most interesting propositions of SCT, however, are that such
categorisations are organised hierarchically and that self-categorisations at
different hierarchical levels are mutually exclusive. One may categorise
oneself at the highest level of abstraction, perhaps as a living entity and in
this respect like all other members of this category, at the lowest level that is
to say as a unique individual distinct from all others, or at some level in
Life Course Transitions and Social Identity Change 201
between. The levels in between are occupied by ‘‘social categories’’ – Italian,
catholic, socialist, female, student, chemist, etc. – but Turner’s point is that
one cannot simultaneously think and behave in terms of categories at dif-
ferent levels of the hierarchy. And most attention has been given to the
personal and social levels of categorisation as, necessarily, alternative ways
of thinking about the self.
The origins, particularly of SIT, in the study of racism and prejudice, have
been reflected in a research focus on ethnic and national identities. However,
both SIT and SCT have aspired to be generic accounts of social identity, in
principle applicable to social identity as such and therefore to any potential
form of social identity. The minimal group paradigm (Tajfel, Flament,
Billig, & Bundy, 1971) reflected this general aspiration. This paradigm cre-
ated conditions in which all meaning is stripped away from social categories
except the one quality they require for them to provide social identities,
namely that each is a categorisation shared by two or more people (they
divide a population into mutually exclusive categories). But SIT, and also
SCT have stimulated research in which a wide variety of real social identities
have been studied, including identities based on work-groups (Hennessy &
West, 1999), occupations (Skevington, 1981; Marson, 2001), enrolment in
different university courses (Reicher, 1984), support for different political
parties (Abrams, 1994), and allegiance to different football clubs (Platow
et al., 1999). Nonetheless, these different kinds of identities have more often
been studied as opportunities to validate general principles by demonstrat-
ing replication across different cases. They have less often been examined to
identify significant differences among kinds of social identities. I shall argue
that there are some such differences and that they are important, partic-
ularly from a life course perspective.
Finally, something should be said here about the relation between the
concepts of social identity and social role. In particular, there is a question
as to whether the former concept adds anything not already provided by the
traditional sociological analysis of roles. According to one view, social cat-
egories and roles are actually alternative bases for structuring interactions
(Mitchell, 1969). I suggest, however, that they emphasise different facets of
the ways people relate to one another. The role concept, with its roots in
drama, focuses on the manner in which the responsibility for tasks is dis-
tributed among positions within a shared structure. The emphasis therefore
is upon integration of different activities. This approach draws our attention
to the manner in which roles relate to and complement other roles – role sets
being constituted by organisational structures; human organisations have,
indeed, been defined as a system of roles (Katz & Kahn, 1978).
NICHOLAS EMLER202
There are three obvious ways in which a social identity perspective differs.
First, it emphasises differentiation rather than integration; its focus, as pre-
viously noted, is on the processes that reinforce similarities within categories
and differences between categories, on boundaries rather than on links.
Second, it emphasises who people are seen to be and feel themselves to be in
contrast to the role perspective emphasis on the tasks people perform. In
these two respects the two concepts are somewhat complementary. Third,
however, social identity appears to be a more inclusive concept. All social
roles create categories of people and thus social identities, but many cate-
gories do not entail roles in a shared social organisational structure. Think,
for example, of social identities based on religion, nationality or ethnicity. As
to whether social identity or social role is the more useful concept in studying
the life course; the answer I think is that both have a contribution to make.
This chapter focuses on the hitherto more neglected contribution of the two.
The paradigm case in the social psychology of social identity remains
ethnic identity, an identity that is normally relatively fixed over the life
course. And perhaps for this reason, analysis of change in social identity –
which naturally arises as a central question for a life course approach – has
not been extensive within these theoretical traditions. The principle excep-
tion has been Tajfel’s own examination of social mobility (Tajfel, 1978). But
mobility was proposed as a defensive manoeuvre, undertaken only rarely
and then to protect self-worth by escaping from a negative identity. The
absence of compelling evidence that membership of low prestige social cat-
egories is associated with lower self-esteem (Emler, 2001) argues against this
motivation. Within the life course frame social identity change is normative,
not exceptional. The challenge is to explain the process of change.
CORE SOCIAL IDENTITIES
Self-categorisation theory emphasises the situational specificity of social
identification, which is to say categories and category contrasts that are
momentarily salient for the individual social actor. Indeed, Turner (1987)
notes that social categorisations can be spontaneous or emergent construc-
tions within a particular situation, the so-called minimal group being just
such a situation; social identities are therefore not confined to classifications
that are culturally provided or built into the social structure of a society.
From the perspective of life course analysis, however, social identities that
both dominate or characterise entire periods of life and that are widely
recognised and rooted within the culture are of particular interest.
Life Course Transitions and Social Identity Change 203
But is it really the case that some social identities are more dominant or
central, chronically more salient? One kind of research that might decide this
would sample individuals’ identity salience across time, much in the way
that experience sampling methods have been used to determine the prev-
alence of different mood states (cf. Czikszentmihalyi & Larson, 1987), but
such research is yet to be done. Indeed, it is more generally true that to date
there has been almost no longitudinal research focused specifically on social
identity, so that the evidence available to us is cross-sectional – comparisons
of people at different ages or life stages. There are, however, other kinds of
indications that some social identities are reliably more central or significant
in people’s lives. One such indication comes from research on the strength of
different kinds of identification in the world of work. Marson (2001) showed
that of the three kinds of identification in this domain, respectively with a
work team, with an employing organisation and with an occupation or
profession, the third consistently emerged as by far the strongest across a
variety of populations.
Second, as Fiske (1998) points out, some categorisations are more reg-
ularly important because they are visually highly salient and therefore so-
cially functional; they are immediately available to structure social
interactions. Fiske’s top three categories on this basis are race, gender and
age. However, other categorisations can be and often are made chronically
visually salient by manipulating appearance. This can be done through
conventionalised styles of grooming and adornment and even through bod-
ily movement, as Mauss (1935) observed many years ago. But these options
are usually combined with an even more visually powerful marker, dress.
The alternative categories with which young people have identified – hippy,
skinhead, goth, punk and so on – are signalled by combinations of dress
codes and bodily adornments, as are many other significant social catego-
risations based on ethnicity, religion and even class and occupation. The
point is that such visual markers of social identity lend cross-situational
durability, and therefore importance, to these identities because they cannot
rapidly be changed. And the rapid changes in appearance that are possible –
shaving the head, acquiring a tattoo – are only significant if they cannot be
equally rapidly reversed.
Another source of evidence comes from research on the manner in which
connections are anticipated in networks of acquaintances. Milgram’s (1967)
studies of the ‘‘small-world’’ phenomenon led to a research showing that
people are surprisingly good at working out how they can efficiently reach
others they do not yet know through the intermediary of personal ac-
quaintances (Klineberg, 2000). It appears that this facility is based on the
NICHOLAS EMLER204
fact that acquaintances have multiple social identities; they can be simul-
taneously classified in several different ways. However, when people
undertake this task they typically use only a small number of these poten-
tial classifications, and usually only two of them. Those most commonly
used are based, respectively, on occupation and geography (Killworth &
Barnard, 1978).
This last example underlines another feature of the social identities that is
likely to be most important over the life course: these are the principle
categories applied to people by their own acquaintances. Social identities are
therefore social in two distinct respects. As categories they are socially de-
fined and the most relevant categories will be widely recognised within a
culture. As applied to particular individuals, however, these identities are
validated by others. A person may aspire to a particular identity but unless
this aspiration is acknowledged by at least some significant others it remains
no more than a private desire. Social identities are viable bases for relating
to others only if at some level they have been negotiated with and accepted
by a relevant audience, an argument we have made on the basis of our
studies of delinquent and non-delinquent identities (Emler & Reicher, 1995);
social identities assume reality and become consequential when they are
performed before an audience and when the audience confirms the authen-
ticity of the performance (cf. also Emler, 1990).
If identities reflect how others define us, they might nonetheless be re-
lationship-specific. This was the essence of William James’s famous decla-
ration that we have as many selves as people who know us, the implication
being that identities are entirely relationship-specific. Though at some level
this might be true, the degree to which it applies is probably trivial. It would
be very odd if people who know you well did not agree whether you were a
mother, a Catholic, of Irish descent, a civil servant and so on. It might be
countered that these are not the essence of the self that others know, but the
response is that they and other such categories do constitute the social self.
There will be, in addition, a history to each relationship that is its own and a
judgment by the other on this basis as to one’s personality. But such judg-
ments are also not, as a matter of evidence, relationship-specific; our ac-
quaintances tend to agree about our personality (Kenrick & Stringfield,
1980), and the better they know us the more they agree (Kenny, 1994). Some
measure of agreement is additionally generated by the fact that our ac-
quaintances also know each other and share with each other their views of
us.
Up to this point the importance of social identities to everyday life can be
summarised as follows: our social identities shape the character and content
Life Course Transitions and Social Identity Change 205
of our relations with others. Perhaps therefore an indicator of the centrality
of an identity is the scope of such effects – how many details of our re-
lationships are affected, and with how many of the others we know and
interact with.
Social identities are likely to shape two further details of our connections
to the social world: who we know and who we encounter on a daily basis. In
the first case, the effect is that we will be personally acquainted with more
people who share a social identity with us than will people who do not share
the social identity with us. University students will know more members of
this category (and mothers will know more mothers, socialists more social-
ists, Glaswegians more Glaswegians, delinquents more delinquents, married
people more married people) than will people who are not university
students etc. This, of course is the basis for the ‘‘small-world’’ facility de-
scribed by Killworth and Bernard (1978).
Additionally, for students and for members of some other categories,
many and perhaps the majority of their daily encounters are with other
members of the same category. But although these two features – who else
one knows and who else one encounters most regularly – are sometimes
strongly related, they are not the same. Take the extreme case in which a
priest or a doctor is one of a kind in a small, rural community. In each case
their daily encounters will be strongly shaped by their vocation, many of the
priest’s encounters are with other community residents as his parishioners,
and the doctor’s with residents as patients. Other members of their own
respective vocational categories will not dominate their array of daily social
contacts. But both will have personal acquaintances who do belong to their
own social category at a higher rate than will other members of their com-
munity; the doctor will know more doctors than will his or her patients, the
priest will know more priests than will his parishioners. In other words, part
of what it means to belong to a social category is that one will be personally
acquainted with other members of the category at a higher rate than the
base rate for the population as a whole.
In a series of studies we have collected data on people’s patterns of rou-
tine social contacts (e.g., Emler, 1990; Emler & McNamara, 1996). These
data have been derived from a variety of groups, including adolescents in
high school, university students, young people in employment, and adults
working as managers, in effect groups of people belonging to different social
categories. The data indicated that various features of contact patterns are
associated with category membership. This point is illustrated in Fig. 1 with
data (from Emler, 2000) on the social contact patterns of young people in
this case all at the same point in the life course but occupying one of four
NICHOLAS EMLER206
distinct categories. These data do not give us the social identities of the
contacts, but it is a fair guess that many are members of the same social
categories as the respondent – university students in the case of the first
group, vocational college students in the case of the second, work colleagues
in the case of the third. The guess is supported by the common locations for
encounters in each group, respectively, on the university campus, at the
college and in the workplace. Something else that these data reveal is that
alternative social category memberships can be associated with dramatically
different rates of social contact;for example, compare the unemployed group
with others.
To summarise, social identities connect individual to social worlds in the
following respects. First, individuals are identified by significant others with
particular social categories and on the basis of this others have particular
expectations about how an individual will and should behave, about who
else that person knows and about how that person should be treated. In
other words, social identities structure interactions. Second, social identities
are reflected in the categories of people an individual is acquainted with.
This has consequences for the resources to which they have access and their
value to others. Third, social identities are linked to patterns of routine
social contact; often, but not always, people will tend to have more regular
contact with other members of the same social category than will those who
0
2
4
6
8
10
12
14
16
18
20
Close Fr Other Fr Cas Aqu Formal Stranger
Univ
Coll
Employ
Unemp
Fig. 1. Contact Patterns: Number of Different People Encountered in
Each Relationship Category (Over 7 days).
Life Course Transitions and Social Identity Change 207
do not belong to that category. This has consequences for, among other
things, the range of ideas, opinions and influences to which individuals are
most exposed.
CHANGING SOCIAL IDENTITY
If the above summary is appropriate, then it has important implications for
transitions in social identity. In effect, changes in social identity require
changes in social relationships. Moreover, the foregoing analysis implies two
kinds of change in social relationships. The first is change in how you are
defined, and therefore treated by the people who already know you – the
substance and character of your interactions with them. The second is
change in who you know and with whom you regularly associate.
Superficially, it might appear that changes of these latter kinds would
not always and invariably occur. Think of a child who grows to adulthood,
gets married and becomes a parent in a small agricultural community. At
each of these transitions there might be small adjustments in the patterns
of daily contact, for example more with the new spouse and less with
other opposite sex peers – but there will be little change in who this indi-
vidual knows. Childhood friends and playmates are likely to go through
similar transitions at similar ages, and so become adult neighbours and
acquaintances.
The problem with this scenario is that entirely closed human communities
in which the only turnover in membership arises from births and deaths do
not and never have existed. On the other hand, there is undoubtedly var-
iation across time and societies in the extent of turnover in community
membership and thus variation in the continuity of any individual’s ac-
quaintances (Slater, 1968). Moreover, to the extent that emotional ties and
regular contacts persist with the same people across transitions in social
identity these would appear to provide obstacles to the process of successful
transition. Because social identities derive stability from the continuity of
relations with others, this very continuity is also a source of inertia. If the
people who know you are in the habit of thinking of you in certain terms,
treating you in particular ways, and having specific expectations about you,
they will not readily or easily switch to thinking, behaving and evaluating in
quite different ways. The mother who has difficulty adjusting to the change
in status of her beloved son from child to adult is only a more extreme
example of a commonplace phenomenon. It may be that the various rituals
of transition – wedding services, baptisms, graduation ceremonies,
NICHOLAS EMLER208
21st birthday parties, coronations, retirement celebrations – are social de-
vices to counteract such inertia. In these rituals, communities gather and
publicly acknowledge a movement of one or more of their number from one
social category to another.
Later, I will reconsider these transition rituals together with the contem-
porary significance of changes in acquaintance and contact patterns. But if a
change in social identity is a change in the way the individual is connected to
the social world, there are nonetheless other, more internal and psycholog-
ical aspects to the process of social identity change. This can be illustrated
with the case of political identity.
THE FORMATION OF POLITICAL IDENTITY
The political life of societies tends to be played out as a competition between
groups of people who have adopted differing political identities. For a long
time political psychology was dominated by questions about the determi-
nants of this choice, interpreted as one of positioning along a left–right
ideological dimension: why do some people end up aligned with the political
left and others with the political centre or right? However, it gradually
emerged that an additional and perhaps more significant question existed:
why do people become politically aligned at all? This question was pro-
moted by the observation that many adults appear to have no political
leanings (Converse, 1964). Subsequently, evidence for a wide variety of
variations in political involvement emerged, including variations in the ex-
tent of political participation, interest in politics, opinion stability, political
knowledge, attention to political events and so on. Research in which these
dimensions of variability have been concurrently assessed has indicated that
they are strongly interrelated and are apparently manifestations of a com-
mon latent variable. In one such study, Nie, Junn and Stehlik-Barry (1996)
labelled this variable ‘political engagement’.
Another interpretation is possible, however. Identity formation is a proc-
ess in which different elements are related in a causal sequence. In the case of
political identity, the following sequence might be proposed (cf. Emler,
2002). First, one needs to have some interest in politics, and this is a nec-
essary condition for taking a second step, namely, paying attention to po-
litical information and events. Such attentiveness allows the individual to
accumulate knowledge of the political world. This individual can then begin
to form judgments or opinions on the basis of this knowledge, and
Life Course Transitions and Social Identity Change 209
subsequently to organise these opinions into coherent political positions.
Finally, an established political position provides the foundation for ide-
ologically consistent political action.
We have made a preliminary test (Emler, Romney, & Bynner, 2004) of a
sequential model of political identity formation by drawing on data from a
large sample of 16–20 -year-olds (Bynner, Romney & Emler, 2003). This test
supports a slightly modified sequence, which begins with political interest
leading to attentiveness. This later, in turn, precedes responsiveness defined
as a positive attitude towards politics. Responsiveness in its turn leads to
action; in the data set action was assessed in terms of attending political
meetings, taking part in demonstrations, etc. The final three steps in the
sequence are from action to voting intention, and thence to opinionation,
this last step reflecting the extent to which individuals have clear opinions on
political questions. This was an imperfect test given that it was not based on
data collected specifically to test the model. In particular, there was no
measure of political knowledge. However, the relevance of this example is to
make the point that a developmental model of identity formation or change
is testable, and can provide a basis for assessing progress towards a new
identity.
I would also argue that the four kinds of psychological elements that
appear to be involved in the construction of a political identity, namely,
motivational, cognitive, attitudinal and behavioural elements, will be found
in all significant social identities, and particularly in those that dominate
and define phases in the life course. This tells us that social identity change is
a process requiring time because it requires a sequence of interconnected
psychological changes.
Having considered the psychological elements of identity formation and
change, it is important to recognise that the psychological changes are
also socially supported. In the case of political identity, motivation or in-
terest will be encouraged or discouraged by acquaintances. Palmonari,
Pombeni, & Kirchler (1992) give the example of young people belonging to
friendship groups in which interest in politics violates group norms. Friends
and acquaintances may be sources of political information but are likely to
play an even more important part in opinion formation and attitude or-
ganisation, providing opportunities to rehearse positions and check their
social meaning (Emler, 1990). Finally, others will be role models for, and
mobilisers to, action. Progress in the formation of political identity, there-
fore, will depend on access to others willing and able to provide these sup-
porting roles. Parallel social processes are likely to be at work in other kinds
of social identity change.
NICHOLAS EMLER210
SOCIAL IDENTITY CHANGE AND
THE PROCESS OF TRANSITION
There are features of the change in political identity, from unaligned to
aligned, that would seem to make it quite an atypical example of the identity
changes characterising the life course. In particular, there is no abrupt
transition in identity. On the contrary, the change is characteristically grad-
ual. But is it really the case that the life course is periodically punctuated by
rapid shifts in social identity? This brings us back to the concept of tran-
sition as a change that is abrupt as well as qualitative.
Major life course transitions are often represented as specific events –
puberty, graduation from university, the first day at work, a wedding (or
divorce), birth of a first child, retirement day. But, with the exception of
childbirth, there are common examples of more extended transitions cor-
responding to each of these cases. It is now common for many couples
to move gradually towards shared domestic arrangements long before
marriage and in many cases the relationship is never formalised as a legal
marriage. The move into work is frequently drawn out through work ex-
perience and periods of training or apprenticeship. At the other end of their
working lives, it is not uncommon for people to go part-time long before
they finally cease working for an income. In Finland, a high proportion of
students never terminate their studies with a completed qualification; instead
they become progressively more part-time as students, spending corre-
spondingly more time working for an income, and so pass gradually rather
than abruptly from the status of a student to that of a worker.
The most conspicuous case of protracted transition, however, is that from
childhood to adulthood, a process to which the very term ‘‘extended tran-
sition’’ was first regularly applied. In contemporary societies there is no
specific event to mark this transition. Instead, there is a progressive accu-
mulation of legal rights at different ages, together with gradual financial and
domestic independence from parents, which may not be completed until the
late 20s or even beyond (Jones & Wallace, 1992).
These various extended transitions have both advantages and disadvan-
tages. In so far as new social identities require new knowledge, attitudes and
behavioural repertoires, the extended time scale provides more opportunity
to develop and integrate these. The major disadvantage is in achieving social
recognition of the new identity.
To see how this disadvantage might be overcome, let us return to the
process by which new social identities are established between the individuals
who are to adopt them and the social worlds in which they dwell. How is this
Life Course Transitions and Social Identity Change 211
social world led to accept an individual’s movement into a new identity? As
already noted, rituals of transition provide devices for securing collective and
public consent to new identities. But, as already noted, both the prevalence
and the significance of many of these appear to be on the decline. This
decline can be linked to a concurrent rise in resort to two other options,
either to change one’s appearance or to change the people one knows.
Formal changes in appearance are, like rituals of transition, currently
becoming less commonplace. In Britain most children used to go to school
in uniforms, and then changed or abandoned uniforms as their age-based
identities changed, but these practices are in decline. Manner of dress also
signifies certain religious identities, and thus their adoption and abandon-
ment. Practitioners of a few vocations still signal their identities by routinely
wearing the appropriate apparel. But occupation-linked uniforms now more
often signal a role only temporarily occupied, a time- and place-specific
performance. On the other hand, one might speculate that when teenagers
are drawn to sub-cultures with highly distinctive dress codes this will be
filling the gap left by the transition rituals that would in former times have
marked a change in identity from child to adult.
What of the other option, changing the people one knows? This is most
readily accomplished by a radical alteration in the patterns of daily contact.
In effect, we interact with the people whose paths cross ours in the course of
our regular daily activities (Emler, 2000). This might seem a trivial point but
it is quite fundamental. Social contact is generated primarily by routines
that take people into particular places to do particular things at particular
times. Contact patterns are therefore altered by changes in these routines.
Birth of a first child usually forces a substantial shift in a woman’s routine
(but also usually has rather less effect on the father’s routine).
The change produced by childbirth in a mother’s routine arises from the
fact that she must commit her time in a new way, and spend more of it in a
particular setting. This, namely relocation in physical space, is the principle
way in which routines, more generally, are altered. There are many examples
of this effect – going to school for the first time, moving from elementary to
high school (and, still a common pattern in Britain going to a residential or
a boarding school), leaving home to go to university, leaving the parental
home to set up one’s own, starting a new job, doing military service, moving
on retirement to a warmer climate. Each of these physical displacements has
the effect of altering patterns of routine social contact, and therefore each
potentially supports a shift in social identity.
Equally evident is that there are cultural variations in common practice
that may reflect the problems a particular economy poses for identity
NICHOLAS EMLER212
change. Children growing up on a Kibbutz have been expected to work on
the Kibbutz as adults but have also been expected at the end of their ed-
ucation to leave for a year and travel (and during this period also find a
spouse). And there are changes in practice over time. In Britain, the tra-
dition of leaving home to attend university is in the decline as rising costs
force more students to attend a local university so that they can continue
living at home. But this trend has been accompanied by another that pro-
duces a break in routine and a discontinuity in relationships, the ‘‘gap year’’
in which travelling abroad intervenes between high school and university.
It is also evident that physical displacement is not a purely modern so-
lution. Some tribal societies have created discontinuity of association in
their otherwise settled communities by physically segregating the sexes for a
time at puberty (Goode, 1959). In the 19th century it was the fashion for
young Englishmen of aristocratic background to depart on a European
tour, and in the 20th century for the females to be sent off to ‘‘finishing
schools’’ in Switzerland. More generally, leaving home at the end of child-
hood to find fame and fortune as well as to find an adult identity, has long
been a theme of folklore and fairytales. Indeed, life as a journey is an almost
universal metaphor, but perhaps not so metaphorical; real journeys are
involved.
At the conclusion of this particular journey, I wish to argue that it has
shown that transitions in the life course as changes in social identity are
more likely to be gradual than abrupt, for two kinds of reasons. First,
identity change requires a number of psychological changes that take time to
build. Second, social identities derive resistance to change from the rela-
tionships with others in which they are embedded and through which they
are confirmed. But a common response to this inertia is some displacement
in physical space, supporting both the formation of new relationships on a
new basis and a return to the old ones from the vantage of a new category.
One hope is that the future will bring the kind of longitudinal evidence
needed to move these ideas about social identity change in the life course
from the realm of speculation to that of empirically grounded observation.
REFERENCES
Abrams, D. (1994). Political distinctiveness: An identity optimising approach. European Journal
of Social Psychology, 24, 357–365.
Brown, R. (1986). Social psychology: The second edition. New York: Free Press.
Life Course Transitions and Social Identity Change 213
Bynner, J., Romney, D., & Emler, N. (2003). Dimensions of political and related facets of
identity in late adolescence. Journal of Youth Studies, 6, 319–335.
Colby, A., & Kohlberg, L. (1987). The measurement of moral judgment, Vol. 1. Theoretical
foundations and research validation. New York: Cambridge University Press.
Converse, P. (1964). The nature of belief systems in mass publics. In: D. Apter (Ed.), Ideology
and discontent (pp. 206–261). New York: Free Press.
Czikszentmihalyi, M., & Larson, R. (1987). Validity and reliability of the experience sampling
method. Journal of Nervous and Mental Disease, 175, 526–536.
Doise, W. (1978). Groups and individuals: Explanations in social psychology. New York:
Cambridge University Press.
Emler, N. (1990). A social psychology of reputation. European Review of Social Psychology, 1,
171–193.
Emler, N. (2000). Social structures and individual lives: Effects of participation in the social
institutions of family, education and work. In: J. Bynner & R.-K. Silbereisen (Eds),
Adversity and challenge in the life course in England and the new Germany (pp. 62–84).
London: Macmillan.
Emler, N. (2001). Self esteem: The costs and causes of low self worth. New York: York
Publishing Services.
Emler, N. (2002). Morality and political orientations: An analysis of their relationship.
European Review of Social Psychology, 13, 259–291.
Emler, N., & McNamara, S. (1996). The social contact patterns of young people: Effects of
participation in the social institutions of family, education and work. In: H. Helve &
J. Bynner (Eds), Youth and life management: Research perspectives (pp. 121–139).
Helsinki: Helsinki University Press.
Emler, N., & Reicher, S. (1995). Adolescence and delinquency: The collective management of
reputation. Oxford: Blackwell.
Emler, N., Romney, D., & Bynner, J. (2004). The development of political engagement. EARA
9th Biennial Conference, Porto, May 2004.
Erikson, E. H. (1950). Childhood and society. New York: Norton.
Fiske, S. (1998). Stereotyping, prejudice and discrimination. In: D. Gilbert, S. T. Fiske &
G. Lindzey (Eds), The handbook of social psychology (4th ed., Vol. 2, pp. 357–411).
New York: McGraw-Hill.
Flavell, J. H. (1963). The developmental psychology of Jean Piaget. New York: Van Nostrand.
Giles, H., & Powesland, P. F. (1975). Speech style and social evaluation. London: Academic
Press.
Goode, W. (1959). The theoretical importance of love. American Sociological Review, 24, 38–47.
Hennessy, J., & West, M. A. (1999). Intergroup behaviour in organisations: A field test of social
identity theory. Small Group Research, 30, 361–382.
Jones, G., & Wallace, C. (1992). Youth, family and citizenship. Buckingham: Open University
Press.
Katz, D., & Kahn, R. (1978). The social psychology of organisations (2nd ed.). New York:
Wiley.
Kenny, D. A. (1994). Interpersonal perception: A social relations analysis. New York: Guilford
Press.
Kenrick, D. T., & Stringfield, D. O. (1980). Personality traits and the eye of the beholder:
Crossing some traditional philosophical boundaries in the search for consistency in all of
the people. Psychological Review, 87, 88–104.
NICHOLAS EMLER214
Killworth, P. D., & Bernard, H. R. (1978). The reverse small world experiment. Social Net-
works, 1, 159–192.
Klineberg, J. (2000). Navigation in a small world. Nature, 406, 845.
Kohlberg, L. (1976). Moral stages and moralization. In: T. Lickona (Ed.), Moral development
and behaviour: Theory, research and social issues. New York: Holt, Rinehart & Winston.
Loevinger, J. (1976). Ego development: Conceptions and theories. San Fransisco: Jossey-Bass.
Marcia, J. (1966). Development and validation of ego-identity status. Journal of Personality and
Social Psychology, 3, 551–558.
Marson, K. (2001). Work-based social identities. British Psychological Society Social Psychol-
ogy Section Annual Conference, Surrey, July 2001.
Mauss, M. (1935). Les techniques du corps. Journal de la Psychologie, 32, 271–293.
Milgram, S. (1967). The small-world problem. Psychology Today, 1, 60–67.
Mitchell, J. C. (1969). Social networks in urban situations. Manchester: Manchester University
Press.
Nie, N. H., Junn, J., & Stehlik-Barry, K. (1996). Education and democratic citizenship in
America. Chicago: University of Chicago Press.
Palmonari, A., Pombeni, M.L., & Kirchler, E. (1992). Evolution of the self-concept in ado-
lescence and social categorisation processes. In: W. Stroebe & M. Hewstone (Eds),
European review of social psychology (Vol. 3, pp. 285–308). Oxford: Wiley.
Piaget, J. (1952). The origins of intelligence in children. London: Routledge.
Platow, M., Durante, M., Williams, N., Garrett, M., Walshe, J., Cinotta, S., Lianos, G., &
Barutchu, A. (1999). The contribution of sport fan social identity to the production of
prosocial behaviour. Group Dynamics, 3, 161–169.
Reicher, S. D. (1984). Social influence in the crowd: Attidudinal and behavioural effects of de-
individuation in conditions of high and low group salience. British Journal of Social
Psychology, 23, 341–350.
Rubin, M., & Hewstone, M. (1998). Social identity theory’s self-esteem hypothesis: A review
and some suggestions for clarification. Personality and Social Psychology Review, 2,
40–62.
Skevington, S. (1981). Intergroup relations and nursing. European Journal of Social Psychology,
11, 43–59.
Slater, P. E. (1968). Some social consequences of temporary systems. In: W. Bennis &
P. E. Slater (Eds), The temporary society. New York: Harper & Row.
Tajfel, H. (1969). Cognitive aspects of prejudice. Journal of Social Issues, 25, 79–97.
Tajfel, H. (1978). Differentiation between social groups: Studies in the social psychology of
intergroup relations. London: Academic Press.
Tajfel, H., Flament, C., Billig, M. G., & Bundy, R. F. (1971). Social categorization and in-
tergroup behaviour. European Journal of Social Psychology, 1, 149–177.
Turner, J. C. (1987). Rediscovering the social group: A self-categorization theory. Oxford:
Blackwell.
Life Course Transitions and Social Identity Change 215
THE IMPACT OF PERSONALITY
AND LIVING CONTEXT
ON REMEMBERING
BIOGRAPHICAL TRANSITIONS
Pasqualina Perrig-Chiello and Walter J. Perrig
1. BIOGRAPHICAL EXPERIENCES AND THEIR
IMPORTANCE FOR THE REGULATION OF
WELL-BEING
The question, as to what extent past experiences have an impact on current
well-being and future anticipations, has provided an important, but con-
troversial, research topic since the very beginning of psychology as a dis-
cipline in its own right. Some investigators have emphasized the importance
of past biographical experiences, claiming that, at any given time, individ-
uals are very much the product of their own life history quite apart from
external, situational demands, opportunities and barriers (Freud, 1917;
Erikson, 1959; McAdams, 1993). Others have pointed to a lack of reliable
empirical evidence on this matter, owing to the largely retrospective char-
acter of most of the data (Rutter, 1996).
Recent advances in life-span developmental psychology have meanwhile
brought increasing empirical insight and rigor to this crucial research ques-
tion. It has been suggested that individual lives can be characterized as a
Towards an Interdisciplinary Perspective on the Life Course
Advances in Life Course Research, Volume 10, 217–235
Copyright r 2005 by Elsevier Ltd.
All rights of reproduction in any form reserved
ISSN: 1040-2608/doi:10.1016/S1040-2608(05)10008-2
217
series of interrelated transitions. Transitions define points in the human life
cycle when roles are transformed, redefined and left behind for new roles
(e.g. starting school, experiencing puberty, starting work, leaving home,
getting married, having children, retirement, etc.). These changes involve
and give shape and direction to various aspects of a person’s life (Mercer,
Nichols, & Doyle, 1989; Sugarman, 2001).
It has been argued that early transitions can have lasting effects on sub-
sequent transitions, even after many years and decades have passed (Elder,
1998). Some studies have shown a wide range of effects in later life following
the experience of specific transitions in childhood (Wadsworth, Maclean,
Kuh, & Rodgers, 1990; Wertlieb, 1997) and adolescence. For example,
psychological functioning during puberty and adolescence was significantly
correlated with that in old age (Vaillant, 1990; Clausen, 1991). Ryff and
Heidrich (1997) showed that reported normative early life events and tran-
sitions were significant predictors for multiple aspects of present and future
well-being across different age groups.
Such findings lead to questions about the underlying mechanisms re-
sponsible for such a relationship. For example, can a biographical transition
directly affect, or even give rise to, a specific developmental outcome in later
life course? Could there also be a third, mediating variable behind the ob-
served correlation? Or, is it the case that some kind of backward inference,
that is, for example recollected (or even reconstructed) mental projections
from the present into the past can account for the data? This chapter is
devoted to these questions and mainly elaborates on those memory proc-
esses that are related to autobiographical experience and its role on reg-
ulation of well-being (see also McAdams contribution in this volume).
2. TRANSITIONS AND LIFE EVENTS: OBJECTIVE
FACTS VERSUS SUBJECTIVE RECONSTRUCTIONS
Despite the persuasiveness of the relationship between earlier transitions
and later developmental outcomes, the theoretical basis of these findings is
far beyond clear. Because these findings are mainly based on retrospective
reports, the question cannot easily be answered, whether a past transition
per se has a long-term effect on later life, or whether it is rather the sub-
jective post-hoc interpretation that matters (Rutter, 1996).
From a memory theoretical point of view, subjective reports of the
past are, like all autobiographical memories, the products of personal
PASQUALINA PERRIG-CHIELLO AND WALTER J. PERRIG218
reconstruction. For example, if a relationship is found between the remem-
bered past and later life events, it could well be that the memorized emo-
tional quality of the previous events accounts for later well-being and not
the event per se (Schacter, 1996; Schwarz & Strack, 1999; Keyes & Ryff,
1999). In memory research there is abundant evidence that reports of the
past are not objective records of events but constructions that are biased by
multiple sources. Remembering embraces memory traces, amalgamated
with actual perceived information and contextual constraints. These are all
subject to current decision-making processes that can be very subtle, re-
sulting in statements with varying accuracy and giving rise to subjective
(rather than objective) truth values (Tulving, 1983; Johnson, Hashtroudi, &
Lindsay, 1993; Rubin & Berntsen, 2003). Furthermore, recalling biograph-
ical episodes not only includes re-experiencing the past, but also the factual
retrieval of autobiographical knowledge that e.g. might have been reported
by others.
There is good evidence in the literature for a remember/know distinction
in memory (Conway, 1987; Gardiner & Java, 1990; Rubin, 1986) even down
at the neuropsychological level (Wheeler, Stuss, & Tulving, 1997). Also, it is
generally accepted that autobiographical knowledge is stored and organized
within a hierarchical framework (Conway & Rubin, 1993). A lifetime period
is the most general, most abstract and most inclusive type of knowledge, and
denotes typically units of years. General events (at the next lower hierar-
chical level) represent common or frequently experienced episodes that are
amalgamated and consolidated in memory, giving rise to a kind of basic, but
personalized, knowledge store (Bartlett, 1932/1993). Specific, single-event
knowledge (at the lowest hierarchical level) is particularly salient, and gives
rise to unique memory traces that appear to be stored separately. In sub-
jective reports of past life transitions, it will never be easy to separate out
what is actually re-experienced (episodic memory) from what is generally
known – and expected (semantic memory) in any of the three types of
autobiographical memory. Neither it is clear, whether memory reports rep-
resent real facts from the past or just constructed inferences induced by the
retrieval setting. Thus, in exploring past-event reports the distinction be-
tween the more cognitive-oriented analyses of autobiographical reports and
the more experiential-oriented analyses of reminiscence is an important one
(Bluck & Alea, 2002). Memorized life transitions can thus be considered as
products of a kind of motivated forgetting or remembering serving indi-
vidual psychodynamic functions (such as equilibration of self-identity).
There are ample reports in the literature of work describing phenomena
related to constructed, reconstructed and distorted memories in general
The Impact of Personality and Living Context 219
(Ayers & Reder, 1998; Brainerd, Reyna, Wright, & Mojardin, 2003) and in
autobiographical memory research in particular (Brewer, 1988; Conway &
Rubin, 1993). There are impressive demonstrations of how misleading
information or faint feelings of familiarity can bias memory reports
(Whittlesea, 2002). Certainly, the most dramatic demonstrations of false
memories describe how participants in certain experimental studies can be-
gin to ‘‘remember’’ painful autobiographical events that were simply sug-
gested to them and which never really happened (Loftus, 2003). Even so, the
precise constructive mechanisms underlying false or distorted reports of past
events are not yet completely understood and continue to be investigated
(Conway, 1996; Bluck, 2003; Bluck & Alea, 2002).
Bartlett’s sensible analysis of remembering and his theoretical conclusions
(Bartlett, 1932/1993) help to better understand these processes. In his view,
the remembering of autobiographical events is influenced by ‘‘schematic’’
knowledge. Bartlett’s ‘‘schema’’ refers to the activation of relevant past
experience in such a way that it affects the interpretation of any presently
incoming sensory information. Although this usefully ensures the updating
of general knowledge held in memory, and helps us to respond in situations
where there is inadequate information to hand, it can sometimes lead us
astray through faulty expectations. For example, when a subject starts to
remember a complex situation, the first thing that comes to mind is often a
general impression of the whole, which is an ‘‘attitude’’ toward it. By ‘‘at-
titude’’, Bartlett referred to a complex psychological state or process, which
is very largely a matter of feeling or affect. He characterized this state as a
collection of subjective phenomena such as ‘‘yby doubt, hesitation, sur-
prise, astonishment, confidence, dislike, repulsion and so on’’ (Bartlett,
1932/1993, p. 207). Recall of the situation or event is thus a reconstruction,
made largely on the basis of this current cognitive–emotional attitude. The
general effect is that of a justification of the attitude: ‘‘This and this and this
must have occurred, in order that my present state should be what it is’’
(Bartlett, 1932/1993, p. 202).
If the topic of remembering is a biographical transition, the relevant
‘‘schema’’ for self-related knowledge will be activated and may control the
process of remembering. Here, the actual schema held by a person about
her/himself (self-concept) can be considered as his/her hierarchically organ-
ized, autobiographical knowledge that has developed over time and has
been very largely determined by immediately preceding or actual experi-
ences. Therefore, the current schemata or themes of the self allows for
constancy as well as change. Obviously, there is enormous stability in the
preservation and continuity of one’s self-identity and of personality, that is,
PASQUALINA PERRIG-CHIELLO AND WALTER J. PERRIG220
the complex of all the attributes that characterize a unique individual over
time. But there is also instability in the construction of one’s autobiograph-
ical memories, which are dependent on the situational living contexts of
different retrieval episodes. Predictions from this theoretical perspective can
be applied to (1) the correlation between remembered life events and actual
well-being, (2) the correlation between self-identity/personality and memory
performance and (3) the influence of particular living contexts on remem-
bering the past.
Based on the arguments set out above, we would expect firstly that any
studies describing past transitions and actual life situations by the means of
subjective reports should find significant relationships between the two. The
reason for this is twofold: On one side autobiographical schematic knowl-
edge contributes to the constancy and stability of self-identity over time
(Bluck, 2003), on the other side the constructive processes involved in ex-
periencing the actual situation as well as remembering the past are self-
serving in the regulation of the actual well-being. Actual well-being and
stability in self-identity is based on conditions of the present, but also on the
past, that serves as reference or justification for the actual situation. Such
self-serving regulatory processes do operate subtly and smoothly during
stable periods of life, but might be extreme in their effects in specific living
contexts (e.g. critical life situations or in therapeutic settings).
Our further predictions relate to the role of personality and the situational
living context in reports of past and present life events and circumstances.
By personality, we refer to the complex of attributes that characterize a
unique individual over time, which includes autobiographical knowledge,
self-perception and self-identity.
From the constancy necessary in the accumulation and integration of a
stable self, we can predict secondly that individual differences in personality
correlate with differences in type or style of cognitive as well as of emotional
processes. Because differences of personality can be measured reliably by
questionnaires, personality inventories can be used as a direct test of per-
sonality. From this it follows that we should also be able to find relation-
ships between life-transition reports and personality measures. Indeed, such
are the results of the studies we report below.
Thirdly, quite apart from personality variables, we should also be able to
demonstrate correlations between the current living context and retrospec-
tive life-transition reports. The importance of shedding light into these
processes has been pointed out by different authors (Baltes, Lindenberger, &
Staudinger, 1998; Bluck, 2003). And again, we are able to present empirical
evidence below for this prediction.
The Impact of Personality and Living Context 221
3. RELATIONSHIP BETWEEN BIOGRAPHICAL
RECOLLECTION, PERSONALITY, LIVING CONTEXT
AND WELL-BEING: THREE ILLUSTRATIVE STUDIES
ON MIDDLE AND OLD AGE
In this contribution we want to give illustrative examples on how autobi-
ographical reports of past life transitions or life events and actual well-being
are related, and how personality and specific living contexts, such as going
through a particular biographical transition like marital separation or dis-
missal from work may have an impact on autobiographical recollection. The
findings reported here are based on data from two research programs:
(1) from the ‘‘Basle Interdisciplinary Study on Ageing’’ (IDA-Study1)
focusing on old persons (Perrig–Chiello, Perrig, Stahelin, Krebs, & Ehrsam,
1996) and (2) from the study ‘‘Transitions and Life Perspectives in Middle
Age’’2 focusing persons in middle age (Perrig-Chiello, Hopflinger, Kaiser, &
Sturzenegger, 1999).
Based on the theoretical rationale presented above, we analyzed the re-
lationships between the emotional valence of autobiographical memories,
actual well-being, personality and actual living context.
We hypothesize that:
(a) There is a positive relationship between the emotional valence of ret-
rospective recollected autobiographical memories and current well-
being.
(b) This relationship is due on the one hand to inherent personality var-
iables and on the other to specific living-context factors, which mediate
the effects.
3.1. The Impact of Personality on Biographical Reports and Well-Being in
Middle and Old Age (Study 1)
3.1.1. Evidence from Research on Old Age
The interview data reported here are original and were collected through the
Basel Interdisciplinary Study on Ageing (IDA-Study) (Perrig-Chiello et al.,
1996). The aim of the IDA-Study was the examination of the cross-sectional
and longitudinal changes of health, well-being and autonomy in old age, and
their determinants. The project was specifically designed to document the
availability of physical, psychological and social resources, and to
investigate their impact on these three outcome variables (health, well-being
PASQUALINA PERRIG-CHIELLO AND WALTER J. PERRIG222
and autonomy). The project is a follow-up of a longitudinal study that
started in 1960 and takes into account the data collected in 1993. Participants
in the IDA-study had to be 65 years or older at the time of interview. They
were selected by random sampling from the longitudinal pool, which still
comprised 3,768 people in 1993. For the IDA-study, 442 healthy persons
aged 65–94 (309 males and 133 females, mean age: 74.95 years) agreed to
participate. Psychological assessment included:
(a) Life event inventory: A list of thematically ordered life events was pre-
sented to the participants (own health, housing condition, financial
condition, family concerns, activities, etc.) (see the appendix). For each
of these domains they were asked to indicate whether and how they had
experienced specific events within the last 10 years.
(b) Personality: In order to assess personality traits we used the two main
subscales ‘‘Extraversion’’ and ‘‘Neuroticism’’ from the Freiburger
Personlichkeits-Inventar (Fahrenberg, Hampel, & Selg, 1970, 1984) (be-
ing the most used personality inventory in German-speaking countries).
(c) Psychological well-being was assessed by means of a 9-item-scale which
taps the dimensions ‘‘satisfaction with own past’’, ‘‘purpose of life’’ and
‘‘mastery’’. This instrument meets all psychometric standards (see
Perrig-Chiello, 1996, 1997).
In order to answer the specific question about the relationship between
the number of the reported positive and negative life events, personality
variables and current well-being, we first performed correlational analyses.
Results show that the reported number of positive life events is negatively
correlated with neuroticism (R ¼ �0:132; po0.05) and positively with psy-
chological well-being (R ¼ 0:118; po0.05), whereas negative life events are
positively correlated with neuroticism (R ¼ 0:123; po0.05). In a second
step, we calculated multiple-regression analyses in order to determine the
predictive power of the reported number of positive and negative life events
on actual psychological well-being. Since in our data we had an overrep-
resentation of men, we controlled also for gender. Results show that the
number of reported negative and positive life events is indeed significant
predictor of psychological well-being, independently of gender. However, as
soon as neuroticism is introduced as additional predictor, the number of
reported life events loses its predictive power in favor of neuroticism (see
Table 1).
Even though these effects are not very large, they are reliable. Further-
more, the same phenomenon was demonstrated in another study with a
middle-aged sample, as described in the following section.
The Impact of Personality and Living Context 223
3.1.2. Evidence from Research on Middle Age
The findings reported here stem from an interdisciplinary longitudinal study
entitled ‘‘Transitions and Life Perspectives in Middle Age’’ (Perrig-Chiello
et al., 1999). The study investigated the impact of past transitions on current
well-being and on anticipation of old age. A total of 268 middle-aged per-
sons (197 women; 71 men) participated in the study (mean age ¼ 47:2years). This sample is a subsample of a larger survey study (N ¼ 1015) and
can be considered as being representative of a healthy middle-aged urban
population in Switzerland. Participants in the study completed two ques-
tionnaires (psychological well-being and personality) and were given an in-
depth interview on biographical transitions (timing and emotional valence
of transitions) (Perrig-Chiello & Hopflinger, 2001). The crucial variables
were assessed as follows:
(a) Reported biographical transitions: A list of age-normed and frequent
non-age-normed transitions was presented and participants had to in-
dicate the age at which the specific transition occurred, as well as the
emotional valence of this event as experienced at that time. The list
referred to transitions across the whole life span (from school entry,
puberty, first love through anticipation of retirement and transition to
old age).
(b) Psychological well-being was assessed by means of the same nine-item
test as in the IDA-Study mentioned previously.
(c) Personality was assessed by means of the NEO-Five Factor Inventory
(Costa & McCrae, 1985).
Table 1. Life Events and Neuroticism: Which Measure is Better as
Predicting Current Well-Being?.
Predictors of Well-Being Standard b (p)
Model 1 Number of negative life events �0.11 (0.03)
Number of positive life events 0.16 (0.001)
Gender �0.05 (n.s.)
N ¼ 301 R ¼ 0:17; R2 ¼ 0:03; p ¼ 0:004
Model 2 Neuroticism �0.256 (o0.0001)
Number of negative life events �0.02 (n.s.)
Number of positive life events 0.087 (n.s.)
Gender �0.02 (n.s.)
N ¼ 301 R ¼ 0:28; R2 ¼ 0:08; po0.0001
PASQUALINA PERRIG-CHIELLO AND WALTER J. PERRIG224
Regression analyses were performed to evaluate the predictive power of
emotional valence of experienced transitions along with personality varia-
bles on present psychological well-being. Age group and gender, emotional
valence of past transitions as well as personality served as predictor var-
iables. We performed hierarchical regression analyses; age group and gender
were entered as block 1 (R2 ¼ 0:05), emotional valence of transitions as
block 2 (R2 ¼ 0:13) and personality variables as block 3 (R2 ¼ 0:42). Theanalyses revealed that, in particular, emotional valence of one transition,
namely puberty, was a significant predictor of actual psychological well-
being (b ¼ 0:18; po0.05). However, as soon as personality variables were
entered, psychological well-being was best predicted by neuroticism
(b ¼ �0:53; po0.01) and by conscientiousness (b ¼ 0:18; po0.05). By
entering these two variables, the predictive power of the remembered emo-
tional valence of past transitions disappeared: only low scores on neurot-
icism and high scores on conscientiousness were associated with high
psychological well-being (see also Perrig-Chiello & Perren, 2005).
On the one hand, our results suggest that recollection of past transitions,
especially the transition to adulthood, is significantly related to current well-
being. In fact, it has been suggested that adolescence can be seen as an
anchor point from which we begin the story of our adult lives and that the
memories from this period help us define who we are later in life (Fitzgerald,
1988; Fitzgerald & Shifley-Grove, 1999; Josselson, 1987; Rybash, 1999). On
the other , however, our results also showed that the predictive power of the
remembered emotional valence of this specific transition vanishes as soon as
personality variables are considered as predictors for current psychological
well-being. These results indicate that the reconstruction of one’s own past,
as well as the perception of one’s present state may both be driven by
personality factors.
3.2. Personality Predicts not only Biographical Memories but also Episodic
Recollection (Study 2)
As described in the two precedent studies, personality factors (especially
neuroticism) can be significant and robust predictors for biographical
recollection. Now, one could conclude that biographical reports are mere
products of motivated recollection and are consequently biased and unre-
liable. Furthermore, a memory bias of this kind might stand in contrast to
more cognitive forms of episodic recollection. The latter kind of remem-
bering might be considered to be free of influence of personality factors, and
The Impact of Personality and Living Context 225
hence, trust worthier. However, this assumption may be erroneous, and it is
reasonable to assume that personality as the individual bundle of complex
features of an integrated stable self is determining cognitive processes like
remembering in general as well. The relationship between objective memory
performance and personality factors is, at best, rather obscure. For example,
only a few investigations to date have looked at the relation between per-
sonality and episodic memory (Perrig-Chiello, Perrig, & Stahelin, 2000;
Meier, Perrig-Chiello, & Perrig, 2002), and there is a growing con-
viction that personality variables are interwoven with memory outcomes
(Cavanaugh & Green, 1990). Gold and Arbuckle (1990) proposed a model
of personality–cognition relations in old age. Recurring to the ‘‘big 5’’ cen-
tral measures (introversion–extraversion, neuroticism, openness, agreeable-
ness and conscientiousness), they hypothesized that most of these traits have
both direct and indirect effects on memory. Furthermore, they argue that, in
later life, senescent changes in the nervous system, coupled with reductions
in externally driven cognitive demands, can result in increased influences of
personality on cognitive functioning. These effects are viewed not as direct
determinants of specific skills or processes, but rather as having a more
general facilitative or detrimental effect on cognition. Neuroticism, in par-
ticular, has been associated with negative outcome expectations and with
poorer cognitive functioning (Gold & Arbuckle, 1990). Results from these
studies suggest that this approach could be a very promising way to study
interindividual differences in episodic memory.
Within the IDA-Study we had the opportunity to examine the impact of
personality on episodic memory performance in a sample of 287 healthy
adults aged 68–95 years. The purpose was to examine the contribution of
extraversion and neuroticism to the explanation of episodic memory var-
iability in old age. To assess personality, we used the two main subscales
‘‘Extraversion’’ and ‘‘Neuroticism’’ from the Freiburger Personlichkeits-
Inventar (Fahrenberg et al., 1970). For episodic memory assessment we used
a computerized test (Perrig et al., 1994). This test allows to assess different
memory dimensions (working memory, recognition and free recall). We ex-
pected higher extraversion and lower neuroticism to be related to better
memory performance. Consistent with our expectations, extraversion was
associated with higher (b ¼ 0:16; po0.01), and neuroticism was associated
with lower levels of episodic memory performance (b ¼ �0:18; po0.01) (for
more details see Meier, Perrig-Chiello, & Perrig, 2002).
In conclusion, and in line with previous research, our results indicate that
the relationship between memory and personality is not very strong, but
consistent (Arbuckle, Gold, Andres, Schwartzman, & Chaikelson, 1992;
PASQUALINA PERRIG-CHIELLO AND WALTER J. PERRIG226
Hultsch, Hertzog, Small, & Dixon, 1999). Even though the relationship
between personality and memory has not been studied extensively, this re-
lationship exists and deserves to be studied more fully. Overall, we conclude,
that both autobiographical and episodic memory recollection, in general,
can be predicted by personality variables. In other words, we believe that
recollection – whether episodic or semantic – are all in some way influenced
by personality factors. But this explains only a part of the variance of the
phenomenon. In order to gain a better appreciation of the determining
factors of biographical recollection, other predictors, such as actual living
context, should be taken into account. The influence of such factors is de-
scribed in the following study.
3.3. The Importance of Living Context on Biographical
Recollection (Study 3)
With respect to the results gathered so far and considering the fact that the
living context is thought to play an important role in the prediction of
biographical and episodic memory outcomes, we took the opportunity to
test this possibility using extreme group comparisons.3 We compared two
groups of middle-aged women living in two different contexts. One group
consisted of an ‘‘average’’ sample of middle-aged women (92 women from
study 2, mean age: 52 years); the other group was a subclinical sample of
women of the same age (45 women, mean age 51.56 years). The latter were
all new recruits, receiving a psychotherapeutic treatment at the time of data
collection in various outpatient clinics, for problems related to a specific
critical life course transition, namely the loss of a partner (separation/
divorce, N ¼ 35; death of spouse, N ¼ 6; combination of loss of partner and
dismissal from work, N ¼ 5). Common to all these women were symptoms
of grief due to the loss of a partner. Assessment instruments and procedure
were the same as described in Section 3.1.2.
We expected that the problem-loaded context of the subclinical group
(being in a difficult partnership transition) would lead to an overall pes-
simistic view of past biographical events. In accordance with empirical re-
sults from studies of mood-dependent information processing (having
persons with negative mood reporting more negative events (Perrig & Perrig-
Chiello, 1988; Eich & Macaulay, 2000). Together with what had been stated
in the introduction part about the function of reconstructive memories and
motivated remembering, we hypothesized that the subclinical group would
report more negative autobiographical events, and that the reported
The Impact of Personality and Living Context 227
emotional valence of their recollection would be overall more pessimistic
than that of the control group.
According to our expectations, the subclinical group did indeed report
having experienced significantly more negative transitions than the control
group (see Table 2). The emotional valence of the reported family transi-
tions, e.g. was significantly more negative in the subclinical group compared
to the controls. In particular, transitions concerning the own offspring were
rated much more negatively by the subclinical compared to the control
group (e.g. birth of the first child and departure of the last child). However,
this phenomenon and interpretation of pessimistic biographical recollection
by the subclinical group has to be restricted, due to one exception. This
exception concerns the remembered partnership translations namely, ‘‘First
love’’: Women in the subclinical group undergoing a difficult partnership-
related transition at the present time reported a significantly higher
Table 2. Reported Emotional Valence of Passed Transitions:
Comparison Subclinical Sample versus Average Group.
Emotional Valence of
Reported Transitions
Subclinical
Group
(N ¼ 45)
Average
Group
(N ¼ 92)
Z Df t p
Mean/SD Mean/SD
Number of negative
transitions
3.8/2.01 2.91/2.01 0.89 135 2.43 0.017
Emotion. valence of
transitions (total)
6.37/1.16 6.68/1.16 �0.30 134 �1.435 0.154
Emotion valence of
family transitions
6.05/1.12 6.77/2.02 �0.72 129 �2.216 0.028
Puberty 5.04/2.43 4.87/2.38 0.18 134 0.404 0.687
First job 7.18/2.81 7.8/2.42 �0.62 133 �1.333 0.185
First love 8.87/1.71 8.08/2.34 0.79 132 2.001 0.047
Departure from home 7.47/2.92 7.33/2.80 0.14 132 0.271 0.787
Marriage 7.82/2.4 7.87/2.43 �0.04 125 �0.097 0.923
Pregnancy 1 7.93/2.74 8.18/2.71 �0.25 111 �0.480 0.632
Birth 1 5.76/3.60 7.51/2.87 �1.75 111 �2.836 0.005
Separation 4.23/3.59 3.40/2.99 0.83 54 0.945 0.349
Departure of last
child
3.38/2.68 5.84/2.79 �2.47 33 �2.651 0.012
Menopause 4.63/2.94 5.64/2.03 �1.01 94 �1.917 0.058
Death of mother 2.69/2.39 3.04/1.78 �0.35 40 �0.545 0.589
Death of father 2.7/2.38 3.23/2.02 �0.54 87 �1.130 0.261
PASQUALINA PERRIG-CHIELLO AND WALTER J. PERRIG228
(positive) emotional valence of this romantic biographical event in the past
compared to the control group (see Table 2).
How can we explain this seemingly contradictory result? The large ma-
jority of women in the subclinical group were in the process of mourning
due to separation, divorce or death from/of their spouse. According to the
interview data, in most of the cases, the lost spouse actually was their first
love. Furthermore, this first love was remembered as being highly positive,
and compared to other people’s first love, ‘‘much better’’ (social compar-
ison: 6.42 versus 5.32, p o0.01). Could this be a compensating mechanism –
a self-serving biased recollection – aimed at reducing self-doubts concerning
this new broken long-term relation? Or, for the women to give a plausible
socially accepted explanation for their excessively strong, psychological
reaction to that loss?
In conclusion, we can say that there is a difference between knowing and
remembering our past. The current living context not only determines per-
ception of the actual situation but systematically bias reports on earlier life
events – for good or bad. Remembering our past goes beyond reporting
autobiographical facts, and implies reconstructive recollection and interpre-
tation, motivated by the perception, that is the interpretation or meaning of
the actual situation. Again, we would emphasize that these processes need
not to be deliberate or accompanied by conscious insight. Especially in
periods of change, growth, when new challenges arise, biographical memory
seems to be tuned to operate selectively (Fitzgerald & Shifley-Grove, 1999).
Distortions and self-serving biases are inevitably associated with this
recollective process.
4. DISCUSSION
The aim of this contribution was to demonstrate possible psychological
sources of biased remembering of autobiographical events and their relation
to well-being based on data from two research programs on middle and old
age. Our results suggest that the recollection of passed life events is asso-
ciated with personality variables (especially neuroticism and extraversion).
Current well-being is related to the emotional valence of biographical rec-
ollection. This latter relationship, however, seems to be mediated by the
personality variables, because it disappears as soon as personality variables
are introduced into the regression analysis, at least for episodic memory
performance, in general. Our finding fits well with the theoretical framework
of constructive processes in perception and memory (Bartlett, 1932/1993;
The Impact of Personality and Living Context 229
Schacter, 1996). Personality ‘‘schemata’’ and personality ‘‘attitudes’’, or, in
other words, the organized set of autobiographical knowledge that consti-
tutes our self-identity, relates past and present well-being in a consequent
meaningful way.
Finally, from a developmental perspective, several authors have also un-
derlined the importance of subjective reconstruction of life for current and
future well-being. For example, Erikson (1959) had stated that the ability to
really see one’s self requires a continuous perspective, both in retrospect and
prospect. It mandates linking the presently understood past and the antic-
ipated future with the experiences present in the individual. In a similar vein,
for McAdams (1993) life stories bind together events in time, organizing
present reality by connecting past and future. These stories are less about
facts and more about meaning. In the subjective telling of the past, the past
is constructed. Butler (1963) has postulated the adaptive function of life
review: Achievement of integrity in such a life review promotes successful
aging. Life review has the function of equilibrating the sense of self-worth,
coherence and of reconciliation of one’s past and present. In this sense, the
remembered emotional valence of past transitions should not (and probably
cannot) be measured exclusively in terms of true/not true, but rather in
terms of preserving a sense of coherence, of continuity of the ‘‘self’’. Ryff
(1991) has suggested that comparisons of the self in the present with the self
in the past constitute an important source for equilibrating actual well-being
for adults (see also Suls & Mullen, 1982). And finally, Keyes and Ryff (1999)
have pointed out that this is how people construe their experiences, not just
experience per se, that matters. The meaning people attach to some events
reflects a concern for maintaining a favorable image.
Our results also provide additional empirical evidence that not only auto-
biographical recollection is influenced by personality, but episodic memory
in general. From our findings presented here, one might assume that per-
sonality predicts better episodic recollection in memory tests, while the ac-
tual living context is more predictive in autobiographical reports. Hence, in
order to predict subjective biographical recollection more accurately, actual
well-being and living context have to be taken into account. In fact, our
results on extreme group comparison show that retrospective recollection and
evaluation of past transitions reflect not only conscious phenomenological re-
experiencing of real-life events, but also a person’s motivation of recon-
structing his/her life in order to gain a sense of coherence and control. In this
way, autobiographical reports can be considered to be the result – to a sub-
stantial degree – of an internal dynamics of the psyche, and as such, they may
not be available to conscious introspection. According to Schacter (1996), the
PASQUALINA PERRIG-CHIELLO AND WALTER J. PERRIG230
complex mixture of personal knowledge that we retain about our past is
woven together to form our life stories. These are the biographies of self that
provide narrative continuity between past and future – a set of memories that
form the core of personal identity. On how exactly personality does influence
the emergence of actual living conditions, or on how exactly personality
relates to cognitive processes – or vice versa – must remain open here, but
should be a most important research topic for the future. To unwrap this
complex package of hidden causalities provides a formidable challenge for
psychological investigations and promises large steps forward in understand-
ing better the human being.
NOTES
1. National Research Program 32 ‘‘Ageing’’, Swiss National Science Foundation.2. Swiss Priority Program ‘‘Switzerland toward the Future’’.3. Original data based on a master thesis (Rusca, 2003), performed within the
research program ‘‘Transitions and Life Perspectives in Middle Age’’ under super-vision of the first author.
ACKNOWLEDGMENTS
This research was supported by grants by the Swiss National Science Foun-
dation (No. 4032-035642 and No. 5004-058457). We wish to express our
gratefulness to Dr. Josephine Cock, University of Berne, for her helpful
comments on an earlier draft.
REFERENCES
Arbuckle, T. Y., Gold, D. P., Andres, D., Schwartzman, A., & Chaikelson, J. (1992). The role
of psychosocial context, age, and intelligence in memory performance of older men.
Psychology and Aging, 7, 25–36.
Ayers, M. S., & Reder, L. M. (1998). A theoretical review of the misinformation effect: Pre-
dictions from an activation-based memory model. Psychonomic Bulletin and Review,
5(1), 1–21.
Baltes, P. B., Lindenberger, U., & Staudinger, U. M. (1998). Life-span theory in developmental
psychology. In: W. Damon & R. M. Lerner (Eds), Handbook of child psychology: The-
oretical models of human development (Vol. 1, pp. 321–356). New York: Wiley.
Bartlett, F. C. (1932/1993). Remembering. Cambridge: Cambridge University Press.
The Impact of Personality and Living Context 231
Bluck, S. (2003). Autobiographical memory. Exploring its function in everyday life. Memory,
11, 113–123.
Bluck, S., & Alea, N. (2002). Exploring the functions of autobiographical memory: Why do I
remember the automn? In: J. D. Webster & B. K. Haight (Eds), Critical advances in
reminiscence: From theory to application (pp. 61–75). New York: Springer.
Butler, R. N. (1963). The life review: An interpretation of reminiscence in the aged. Psychiatry,
26, 65–76.
Brainerd, C. J., Reyna, V. F., Wright, R., & Mojardin, A. H. (2003). Recollection rejection:
False-memory editing in children and adults. Psychological Review, 110(4), 762–784.
Brewer, W. F. (1988). Memory for randomly sampled autobiographical events. In: U. Neisser &
E. Winograd (Eds), Remembering reconsidered: Ecological and traditional approaches to
the study of memory. New York: Cambridge University Press.
Cavanaugh, J. C., & Green, E. E. (1990). I believe, therefore I can: Self-efficacy beliefs in
memory aging. In: E. A. Lovelace (Ed.), Aging and cognition. Mental processes, self-
awareness and interventions. Amsterdam: North-Holland.
Clausen, J. S. (1991). Adolescent competence and the shaping of the life course. American
Journal of Sociology, 96, 805–842.
Conway, M. A. (1987). Verifying autobiographical facts. Cognition, 26(1), 39–58.
Conway, M. A. (1996). Autobiographical knowledge and autobiographical memories. In:
D. C. Rubin (Ed.), Remembering our past. Studies in autobiographical memory
(pp. 67–94). New York: Cambridge University Press.
Conway, M. A., & Rubin, D. C. (1993). The structure of autobiographical memory. In:
A. E. Collins, S. E. Gathercole, M. A. Conway & P. E. M. Morris (Eds), Theories of
memory (pp. 103–137). Hove, Sussex, UK: Erlbaum.
Costa, P. T., & McCrae, R. R. (1985). Manual for the NEO personality inventory. Odessa, FL:
Psychological Assessment Resources, Inc.
Eich, E., & Macaulay, D. (2000). Are real moods required to reveal mood-congruent and mood-
dependent memory? Memory & Cognition, 20, 277–290.
Elder, G. H. (1998). The life course as developmental theory. Child Development, 69, 1–12.
Erikson, E. H. (1959). Identity and the life cycle. New York: International University Press.
Fahrenberg, J., Hampel, R., & Selg, H. (1970, 1984). Freiburger Personlichkeits-Inventar.
Gottingen: Hogrefe.
Fitzgerald, J. M. (1988). Vivid memories and the reminiscence phenomenon: The role of a self
narrative. Human Development, 31, 261–273.
Fitzgerald, J. M., & Shifley-Grove, S. S. (1999). Memory and affect: Autobiographical memory
distribution and availability in normal and recently detoxified alcoholics. Journal of
Adult Development, 6, 11–19.
Freud, S. (1917).Vorlesungen zur Einfuhrung in die Psychoanalyse. Reprint, Frankfurt: Fischer, 1981.
Gardiner, J. M., & Java, R. I. (1990). Recollective experience in word and nonword recognition.
Memory & Cognition, 18, 23–30.
Gold, D. P., & Arbuckle, T. Y. (1990). Interactions between personality and cognition and their
implications for theories of aging. In: E. A. Lovelace (Ed.), Aging and cognition.
Amsterdam: North-Holland.
Hultsch, D. F., Hertzog, C., Small, B. J., & Dixon, R. A. (1999). Use it or lose it: Engaged
lifestyle as a buffer of cognitive decline in aging? Psychology and Aging, 14, 245–263.
Johnson, M. K., Hashtroudi, S., & Lindsay, D. S. (1993). Source monitoring. Psychological
Bulletin, 114, 3–28.
PASQUALINA PERRIG-CHIELLO AND WALTER J. PERRIG232
Josselson, R. (1987). Finding herself: Pathways to identity development in women. San Francisco:
Jossey-Bass.
Keyes, C. L. M., & Ryff, C. D. (1999). Psychological well-being in midlife. In: S. L. Willis &
J. D. Reid (Eds), Life in the middle. Psychological and social development in middle age
(pp. 161–182). San Diego: Academic Press.
Loftus, E. F. (2003). Make-believe memories. American Psychologist, 58(11), 867–873.
McAdams, D. P. (1993). The stories we live by: Personal myths and the making of the self.
New York: Morrow.
Meier, B., Perrig-Chiello, P., & Perrig, W. (2002). The impact of personality on memory per-
formance in elderly. Aging, Neuropsychology, and Cognition, 9(2), 135–144.
Mercer, R. T., Nichols, E. G., & Doyle, G. C. (1989). Transitions in a woman’s life. Major life
events in developmental context. New York: Springer.
Perrig, W. J., & Perrig-Chiello, P. (1988). Mood congruity effects in absence of mood. Memory
& Cognition, 16, 102–109.
Perrig, W., Kling, V., Meier, B., Hofer, D., Perrig-Chiello, P., Ruch, M., & Serafin, D. (1994).
Computerunterstutzter Gedachtnis-Funktions-Test (GFT). Version 2.0 Manual und Dis-
kette. Institut fur Psychologie, Universitat Basel.
Perrig-Chiello, P. (1996). Wie konnen wir Wohlbefinden erfassen? In: H. W. Heiss, F. Huber,
B. Peter & H. B. Stahelin (Eds), Wohlbefinden im Alter – geriatrische und gerontologische
Strategien (pp. 7–17). Reinach: Roche Pharma.
Perrig-Chiello, P. (1997). Ressourcen des Wohlbefindens im Alter. Weinheim: Juventa.
Perrig-Chiello, P., & Hopflinger, F. (2001). Zwischen den Generationen – Frauen und Manner im
mittleren Lebensalter. Zurich: Seismo-Verlag.
Perrig-Chiello, P., Hopflinger, F., Kaiser, A., & Sturzenegger, M. (1999). Psychosoziale Aspekte
der Lebensbedingungen von Frauen und Mannern im mittleren Lebensalter. Zeitschrift
fur Familienforschung, 11, 5–27.
Perrig-Chiello, P., & Perren, S. (2005). Biographical transitions from a midlife perspective.
Journal of Adult Development, 12(3), 169–181.
Perrig-Chiello, P., Perrig, W. J., & Stahelin, H. B. (2000). Differential aspects of memory self-
evaluation in old and very old people. Ageing & Mental Health, 4(2), 130–135.
Perrig-Chiello, P., Perrig, W. J., Stahelin, H. B., Krebs, E., & Ehrsam, R. (1996). Autonomie,
Wohlbefinden und Gesundheit im Alter: Eine interdisziplinare Altersstudie (IDA).
Zeitschrift fur Gerontologie und Geriatrie, 29, 95–109.
Rubin, D. C. (1986). Autobiographical memory. Cambridge: Cambridge University Press.
Rubin, D. C., & Berntsen, D. (2003). Life scripts help to maintain autobiographical mem-
ories of highly positive, but not highly negative, events. Memory & Cognition, 31(1),
1–14.
Rusca, E. (2003). Soziobiographische Ressourcen und Wohlbefinden von Frauen im mittleren
Lebensalter in einer als kritisch erlebten Transition. Lizentiatsarbeit, Institut fur Psycho-
logie, Universitat Bern.
Rutter, M. (1996). Transitions and turning points in developmental psychopathology: As ap-
plied to the age span between childhood and mid-adulthood. International Journal of
Behavioral Development, 19, 603–626.
Rybash, J. M. (1999). Aging and autobiographical memory: The long and bumpy road. Journal
of Adult Development, 6, 1–10.
Ryff, C. D. (1991). Possible selves in adulthood and old age: A tale of shifting horizons.
Psychology and Aging, 6, 286–295.
The Impact of Personality and Living Context 233
Ryff, C. D., & Heidrich, S. M. (1997). Experience and well-being: Explorations on domains of
life and how they matter. International Journal of Behavioral Development, 20, 193–206.
Schacter, D. L. (1996). Searching for memory. The brain, the mind, and the past. New York:
Basic Books.
Schwarz, N., & Strack, F. (1999). Reports of subjective well-being: Judgmental processes and
their methodological implications. In: D. Kahneman, E. Diener & N. Schwarz (Eds),
Well-being. The foundations of hedonistic psychology (pp. 61–85). New York: Russell
Sage Foundation.
Sugarman, L. (2001). Life-span development. Frameworks, accounts and strategies (2nd ed.). East
Sussex: Psychology Press.
Suls, J., & Mullen, B. (1982). From the cradle to the grave: Comparison and self-evaluation
across the life-span. In: J. Suls, (Ed.), Psychological perspectives on the self (Vol. 1,
pp. 97–125). Hillsdale, NJ: Erlbaum.
Tulving, E. (1983). Elements of episodic memory. Oxford: Oxford University Press.
Vaillant, G. E. (1990). Avoiding negative life outcomes: Evidence from a forty-five year study.
In: P. B. Baltes & M. M. Baltes (Eds), Successful aging: Perspectives from the behavioural
sciences (pp. 323–358). New York: Cambridge University Press.
Wadsworth, M. E. J., Maclean, M., Kuh, D., & Rodgers, B. (1990). Children of divorced and
separated parents. Family Practice, 7, 104–109.
Wertlieb, D. (1997). Children whose parents divorce: Life trajectories and turning points. In:
I. H. Gotlib & B. Wheaton (Eds), Stress and adversity of the life course (pp. 179–197).
Cambridge: Cambridge University Press.
Wheeler, M. A., Stuss, D. T., & Tulving, E. (1997). Toward a theory of episodic memory: The
frontal lobes and autonoetic consciousness. Psychological Bulletin, 21(3), 331–354.
Whittlesea, B. W. A. (2002). False memory and the discrepancy-attribution hypothesis: The
prototype-familiarity illusion. Journal of Experimental Psychology: General, 131(1),
96–115.
APPENDIX. LIFE EVENT INVENTORY
Basle Interdisciplinary Study on Aging (IDA-Study)
Life Event Inventory
Which important events, positive as well as negative, come to mind if you
think back the last ten years?
I will provide some topics/themes. Please tell me in each case which event(s)
come to mind and how you experience those events at present.
PASQUALINA PERRIG-CHIELLO AND WALTER J. PERRIG234
positive negative
Employment, work .....................(i.e. retirement or retirement of one's ... spouse/partner, change of job)..............
2
22
1
11
0
00
ambivalent
Illness and Health ................................
(i.e. injury, illness concerning oneself,
or one's spouse/partner, children).........
2
2
2
1
1
1
0
0
0
Residence ..............................................
(move, relocation i.e. move to another .
town, renovation) ...................................
2
2
2
1
1
1
0
0
0
Death ......................................................
(spouse/partner, children, other family .
members) ................................................
2
2
2
1
1
1
0
0
0
Partnership and Familial situation ...
(separation, divorce, ..............................
grandparenthood) ...................................
2
2
2
1
1
1
0
0
0
..................................................................
..................................................................
Financial situation ...............................2
2
2
1
1
1
0
0
0
Recreation, entertainment ..................
(travel, social activities).........................
2
2
2
1
1
1
0
0
0
The Impact of Personality and Living Context 235
STUDYING LIVES IN TIME:
A NARRATIVE APPROACH
Dan P. McAdams
Social scientists conceive of the human life course in many different ways.
Life-span developmental theorists like Erikson (1963) and Levinson (1978)
imagine a sequence of stages through which the individual passes on the
journey of life. Each period or season offers its own distinctive challenges,
and each individual confronts those challenges in a predetermined and pre-
dictable order. More sociologically informed theorists invoke social roles,
generational cohorts, and historical events in their conceptions of human
lives in time, suggesting that development is rather too contingent and con-
textual to conform to any predetermined sequence (Dannefer, 1984; Elder,
1995). The person moves through time along a dynamic trajectory that
results from the complex interplay between human agency and the manifold
forces of social structure. A third approach in the social sciences argues for
the primacy of personal traits in shaping the life course (McCrae & Costa,
1990; Roberts & Pomerantz, 2004). Human beings differ with respect to
broad psychological tendencies that are surprisingly consistent, stable, and
pervasive over developmental time. While contexts may change, personal
dispositions continue to exert their steady and predictable influence.
Conceptions of the life course that emphasize developmental stages, con-
tingent trajectories, and personal traits all have considerable value in mak-
ing sense of how individual lives evolve over time. Stage theories spell out
common developmental demands that people face in many different cultures
Towards an Interdisciplinary Perspective on the Life Course
Advances in Life Course Research, Volume 10, 237–258
Copyright r 2005 by Elsevier Ltd.
All rights of reproduction in any form reserved
ISSN: 1040-2608/doi:10.1016/S1040-2608(05)10009-4
237
and contexts. Theories emphasizing context and contingency in the life
course pay careful attention to economic, social, cultural, and historical
forces that impact lives in variable ways, and how individuals react to and
sometimes resist those forces. Trait conceptions underscore the importance
of human individuality, and how people construct lives that are consistent
with their internal dispositions. For all their value, however, these three
major approaches to the life course pay only passing attention to the sub-
jective meanings of lives. How do individuals themselves make sense of their
lives in time? What meanings do persons construct for themselves as they
move through developmental stages, confront the vagaries of off-time and
on-time challenges in the life course, and act in accord with their internal
dispositions?
The current chapter affirms a fourth approach to the life course, one that
emphasizes the meanings that people make of their own lives in time. The
chapter’s thesis is that those meanings often take the form of stories (Cohler,
1982; McAdams, 1985; Polkinghorne, 1988). People construe their own lives
as evolving stories that aim to reconstruct the past and imagine the future in
meaningful and coherent ways. Narrative approaches to the life course see
persons as storytellers, see lives as stories told, and see the life course as a
psychosocial construction reflecting both personal inclinations and the nar-
rative conventions and traditions that prevail in a given society.
PERSONALITY, NARRATIVE, AND
THE LIFE COURSE
The last two decades have witnessed an upsurge of interest among scholars
and social scientists in the role of stories in social life. Narrative approaches
have had a major impact in sociology (Holstein & Gubrium, 2000), crim-
inology (Maruna, 2001), gerontology (Birren, Kenyon, Ruth, Shroots, &
Svendson, 1996), organizational studies (Gabriel, 2000), educational re-
search (Casey, 1996; Ely, 2003), theology and moral philosophy (MacIntyre,
1984), developmental psychology (Fivush & Haden, 2003), social psycho-
logy (Gergen, 1991; Murray & Holmes, 1994), clinical psychology (Angus &
McLeod, 2004; Lieblich, McAdams, & Josselson, 2004), and the cognitive
sciences (Conway & Pleydell-Pearce, 2000; Schank & Abelson, 1995).
Bruner (1990) and Sarbin (1986) deem narrative to be a new and powerful
root metaphor for the social sciences. Josselson and Lieblich (1993) describe
the narrative study of lives as a loosely organized, interdisciplinary endeavor
DAN P. McADAMS238
aimed at telling and analyzing the stories people construct to make sense of
social life in time, with some special emphasis given over to narrative ac-
counts, such as those told by women and people of color, that have his-
torically been suppressed or marginalized (see also Franz & Stewart, 1994;
Rosenwald & Ochberg, 1992). The many different approaches today in
social science incorporating narrative range from those employing story-
telling methodologies to examine social behavior (e.g., Baumeister, 1994;
Woike, 1995) to those promoting narrative theories about social life itself,
viewing lives as ongoing stories that are told, revised, and retold in culture
(e.g., Gregg, 1991; Hermans, 1996).
The particular narrative approach to be described in the current chapter
evolved within the field of personality psychology (McAdams, 1985), though
its origins were strongly influenced by the approaches in life-span develop-
mental psychology (Erikson, 1963), life course sociology (Bertaux, 1981),
Sartrean philosophy (Charme, 1984), and studies of narrative and bio-
graphy in the humanities (Edel, 1978; Langbaum, 1982; Ricoeur, 1984).
Personality psychology views itself as the scientific study of the whole per-
son (Allport, 1937; McAdams, 1997). The aim of personality psychology is
to provide a scientifically credible account of human individuality – how the
individual person is (a) like all other persons, (b) like some other person, and
(c) like no other person (Kluckhohn & Murray, 1953). As such, personality
psychology considers both species-typical characteristics of human nature
and individual differences between people. Research in the field, however,
tends to focus mainly on individual differences. What are the most impor-
tant individual differences in psychological and social life? Current thinking
in personality psychology tends to focus on three different classes or levels
of individual differences – dispositional traits, characteristic adaptations,
and integrative life stories (Hooker, 2002; Hooker & McAdams, 2003;
McAdams, 1995, 2006a; Sheldon, 2004). From the standpoint of personality
psychology, then, life narrative holds a particularly prominent position
within an evolving constellation of characteristics – including traits and
adaptations – that specify how a person is similar to and different from
other persons.
Within the constellation of human individuality, the most recognizable
and longitudinally consistent aspect of the person is his or her dispositional
traits (Matthews, Deary, & Whiteman, 2003; McCrae & Costa, 1990).
Dispositional traits are those broad, stylistic, and linear dimensions of hu-
man individuality, bearing names such as ‘‘extraversion/introversion’’ and
‘‘depressiveness,’’ that account for general consistencies in behavior,
thought, and feeling across situations and over time. While traits were once
Studying Lives in Time: A Narrative Approach 239
the object of considerable derision among those psychologists who empha-
size the role of situations in shaping behavior (e.g., Mischel, 1968), traits
today are the coin of the realm in personality research. The scientific viability
of the trait concept comes from many sources. First, a consistent body of
research shows that trait attributions based on careful observations reflect
real differences in the behavior and personalities of the people about whom
the attributions are being made (Funder & Colvin, 1991). Second, research
consistently shows that scores on self-report trait scales predict general con-
sistencies in behavior when sampled across different situations and over time
(Kenrick & Funder, 1988) and predict important outcomes such as well-
being, health, and even longevity (Friedman et al., 1993; Matthews et al.,
2003). Third, individual differences in self-report trait scores show consid-
erable stability over developmental time, especially during the adult years
(Costa & McCrae, 1994; Roberts & Del Vecchio, 2000). Fourth, twin studies
suggest that traits are at least moderately heritable, with as much as 50–60%
of the variance in trait scores being attributed to genetic differences between
people (Bouchard, Lykken, McGue, Segal, & Tellegen, 1990). Fifth, 30 years
of factor-analytic studies have resulted in an emerging consensus in the field
of personality regarding the organization of the trait universe. Many per-
sonality psychologists today subscribe to the Big Five model of traits,
grouping the many possible traits that might be invoked in describing human
individuality into the five general categories of extraversion/introversion,
neuroticism, conscientiousness, agreeableness, and openness to experience
(Goldberg, 1993; John & Srivastava, 1999; McCrae & Costa, 1990).
Showing considerable stability over time, the Big Five traits provide
something akin to a dispositional signature for human lives. A highly
extraverted person tends to express high levels of energy and social interest,
enjoys being with people, and seeks opportunities for social stimulation to a
greater extent, on average, than does a highly introverted (low extraversion)
person. If a person scores high on an extraversion scale at, say, age 30, he or
she is likely to score fairly high again 20 years down the road, at age 50. If
the person is low on extraversion at Time 1, chances are good that he or she
will score low again at Time 2. In other words, people tend to hold their
relative position in the normal distribution of scores for any given trait. Of
course, this differential continuity in traits is not perfect. People can and do
move around in the distribution over time (Roberts & Pomerantz, 2004).
But the general trend is toward relative stability. Traits are well designed to
show the interindividual stability of personality over the long haul. But
this stability is of a very general kind, referring to a signature style of
behavior, thought, and feeling. The style may be apparent very early on in
DAN P. McADAMS240
an interpersonal interaction, even within the first minute or two (Ambady,
Hallahan, & Rosenthal, 1995). But as one gets to know a person better, one
begins to see aspects of human individuality that appear to go beyond traits,
aspects that are also subject to considerable change over the life course
(McAdams, 1994). Personality begins with traits, but it does not end there.
At a second level of human individuality, personality psychologists speak
of personal motives, goals, values, interests, strategies, schemas, stage-
specific concerns, and conditional patterns of behavior that are situated in
time, place, and/or social roles (Cantor, 1990; Emmons, 1986; Freund &
Baltes, 2000; Little, 1999; Mischel & Shoda, 1995; Schwartz, 1994; Winter,
John, Stewart, Klohnen, & Duncan, 1998). These many different aspects of
human individuality may be provisionally grouped under the label charac-
teristic adaptations. If dispositional traits sketch an outline of human in-
dividuality, characteristic adaptations fill in some of the details. Though
sometimes loosely related to dispositional traits, people’s goals, plans, sche-
mas, coping strategies, and the like express the highly specific and person-
alized ways in which they have adapted to the variegated demands of social
life. Unlike traits, many characteristic adaptations are expected to change
substantially over the life course, as the circumstances of life change and as
new developmental issues arise. For example, longitudinal and cross-
sectional studies show that adults’ concern for the next generation – what
Erikson (1963) called the developmental task of generativity – rise and fall
over the adult life course, reaching something of a peak in the middle-adult
years (McAdams, 2001; McAdams, de St. Aubin, & Logan, 1993; Rossi,
2001; Vaillant & Milofsky, 1980).
But traits and adaptations do not provide the full picture. People differ
from each other in ways that go beyond their dispositional signature and
their characteristic motivational and developmental concerns, and this
becomes especially apparent as people move into and through their adult
years. At a third level of human individuality, then, reside integrative life
narratives (Hooker & McAdams, 2003; McAdams, 1995, 1996). These are
the internalized and evolving stories of the self that people construct to
make sense of their lives in time. These stories, or narrative identities (Singer,
2004), combine a person’s selective reconstruction of the past with his or her
imagined anticipation of the personal future to produce a story with plot
and characters to explain who I am, how I came to be, and where my life is
going in the future. As Giddens (1991) writes, ‘‘a person’s identity is not
to be found in behavior, nor – important though this is – in the reactions
of others, but in the capacity to keep a particular narrative going’’ (p. 54).
Narrative identities provide life with some degree of unity and purpose.
Studying Lives in Time: A Narrative Approach 241
As people develop over the life course, they rewrite their own self-defining
life stories, albeit in typically implicit and unconscious ways. If, then, dis-
positional traits sketch the outline and characteristic adaptations fill in some
of the details of human individuality, integrative life stories speak to what a
person’s life means in the overall, set in subjective time. Personality itself,
then, is a unique patterning of traits, adaptations, and stories evolving in a
complex interpersonal, social, cultural, and historical context.
THE DEVELOPMENT OF THE STORY
Stories are fundamentally about the vicissitudes of human intention organ-
ized in time (Bruner, 1990; Ricoeur, 1984). The developmental origins of
storytelling, therefore, may be traced back to infancy. By the end of the first
year of life, human infants appear to have a rudimentary sense of human
intentionality, as indexed by their preference for observing and imitating
goal-directed rather than random behaviors (Tomasello, 2000). By age 2,
children have developed what Howe and Courage (1997) call an auto-
biographical self. They begin to construe the world from the standpoint of a
subjective, appropriating self (what William James called ‘‘the I’’) who nar-
rates and remembers personal events as episodes that happen to ‘‘me.’’
Parents typically assist children in the encoding, recollection, and telling of
personal memories, encouraging their effort to narrate their experiences and
acquainting them with the conventions for good storytelling (Fivush &
Haden, 2003). By the time the children have reached their fourth birthday,
they typically understand that stories contain characters who seek to enact
their intentions, characters who want things and try to get them over time.
With few exceptions, young children come to understand that people have
minds, that minds contain desires and thoughts, and that people act upon
those desires and thoughts in goal-directed ways (Wellman, 1993). By age 5,
they are able to tell stories about the self that express this implicit under-
standing of human minds and human intentionality. By the age of 5, they
have developed a story schema, or a set of expectations regarding what a
story should look and sound like (Mandler, 1984). A story should be set in a
particular time and place. It should involve a motivated character who sets
out to get or do something. The character should eventually run into ob-
stacles in the goal-directed action, which will lead to reactions and changes
and ultimately some kind of a resolution. When stories do not conform to
the schema, children find them odd, and they may seek to reformulate
noncanonical narratives so that they fit the schema better.
DAN P. McADAMS242
Although children can tell stories about their own experiences by age 5,
they do not have, nor are they really working on, narrative identities. Chil-
dren do not have the developmental maturity, the social experiences, nor the
cognitive skills to see their own lives as stories evolving over time, stories for
which they are both the author and the protagonist. Beginning with Erikson
(1963), many developmental theorists have argued that people do not con-
struct full-fledged identities until adolescence or young adulthood (Arnett,
2000). McAdams (1985) suggests that what Erikson viewed as the challenge
of identity formation in adolescence and young adulthood is largely about
formulating a story for one’s life – what Sartre called a true novel (Charme,
1984) – that selectively reconstructs the past and imagines the future as an
integrated temporal whole, to provide life with meaning and purpose and
situate the person’s imagined life trajectory within a recognizable societal
niche. As Hankiss (1981) suggested, the challenge of identity involves
‘‘mythologically re-arranging’’ one’s life history into a coherent narrative of
self (p. 203). Similarly, Kohli (1981) suggested that the creation of a life
story presupposes an advanced facility with autobiographical forms: ‘‘The
autobiographical form presupposes a developed individuality, a self-
conscious ‘I’ being able to grasp itself as the organizer of its own life
history’’ (p. 64). A number of theorists contend that the full utilization of
the autobiographical form typically awaits late adolescence.
For example, Habermas and Bluck (2000) argue that the ability to con-
strue one’s life as an evolving narrative of the self that integrates the re-
constructed past and imagined future requires mastery of four different
cognitive skills. First, the individual must be able to order events into tem-
poral sequences, what Habermas and Bluck (2000) refer to as temporal
coherence. Second, the individual must be able to assimilate his or her life to
what society deems to be the typical course of life, showing what Habermas
and Bluck (2000) call autobiographical coherence. Third, Habermas and
Bluck identify causal coherence as the ability to link different auto-
biographical events into causal sequences to explain how a particular aspect
of the self came to be. A teenager may explain her distrust of authority by
telling how she used to trust her father completely, but how she gradually
lost confidence in him through a series of events wherein he disappointed
her. Finally, life storytelling requires what Habermas and Bluck call the-
matic coherence – the inductive derivation of themes or principles from a
series of personal events. A man may come to conclude that ‘‘I am some-
body who follows my deepest dreams’’ by noting how in many different
events in the past he defied convention and acted in accord with his most
cherished beliefs. Habermas and Bluck (2000) review research, showing that
Studying Lives in Time: A Narrative Approach 243
facility with causal and thematic coherence tends not to appear in self-
narration until the late teenage years.
Equipped with a full complement of cognitive skills necessary for life
narration and encouraged by their psychosocial environments to figure out
who they are and how they are to fit into the adult world, older adolescents
and young adults – what Arnett (2000) terms emerging adults – typically seek
to arrange their lives into narrative identities. The move toward narrative
identity in emerging adulthood is especially common in modern societies,
wherein emerging adults are strongly urged to build a life that both ex-
presses their individualized tastes and proclivities and conforms in some
manner to the exigencies of a rapidly changing economic and cultural world
(Gergen, 1991; Giddens, 1991; McAdams, 1996). The development of nar-
rative identity in the emerging adult years occurs gradually, unevenly, and in
the context of close personal relationships (McAdams, 1993; Thorne, 2000).
Emerging adults try out different stories for their lives, imagine and tell past
events and future goals in different ways, with different friends and family
members, for different purposes and effects, in a social context wherein
other emerging adults are doing pretty much the same thing. Stories are
shared and refined. Stories are revised to reflect changing priorities, new
insights, transformed goals and beliefs. Eventually, many people settle on a
few key story lines, accentuate a set of key events from the past and down-
play many others, imagine a circumscribed set of alternative and yet more-
or-less realistic scenarios for the future, and gradually give their lives a
narrative shape.
Emerging adulthood is prime time for the development of narrative iden-
tity. But the process of constructing one’s life as a narrative does not end in
one’s 20s. People continue to work on their life stories across the bulk of the
adult life span (Birren et al., 1996; Cohler, 1982; Hooker & McAdams, 2003;
McAdams, 1993). The midlife years, for instance, may be occasioned by
considerable identity work for many modern adults. Life-stage theorists
have written about how the realization that one’s life is now more than half
over can bring to the psychosocial fore concerns about loss and mortality
and can stimulate the actualization of long-suppressed tendencies, such as
traditionally masculine tendencies in women and feminine tendencies among
men (Gutmann, 1987; Levinson, 1978). Life course theories emphasize
changing social roles and relationships in the midlife years and shifting
contingencies in the social ecology of everyday life (Elder, 1995). Theorists
of many different stripes tend to agree that midlife can be a period wherein
considerable revision of one’s life story is likely to occur. Even as one’s traits
continue to express continuity in selfhood, people change their narrative
DAN P. McADAMS244
understandings of the past, present, and future in response to a wide range
of expected and unexpected life events and transitions, from off-time
setbacks like the premature death of a spouse to on-time markers like the
children leaving home to go to college, one’s 40th birthday, or the transition
from work to retirement.
Changes in narrative identity constitute real personality change – just as
real as changes in dispositional traits, but different. Changes in narrative
identity are typically the kinds of changes that people allude to when they
maintain that they have ‘‘matured’’ with ‘‘experience,’’ that they are ‘‘very
different people’’ today than they once were (McAdams, 1994). They are the
kinds of changes, furthermore, that therapists and counselors often try to
effect in their clients (Angus & McLeod, 2004). Conceptions of personality
that focus exclusively on traits, therefore, underestimate the nature and
extent of development over the span of life. Stories describe change and
development in life, and life stories themselves change and develop over the
life course.
STUDYING LIFE STORIES
Social scientists have developed many different methodologies for studying
narrative identity. Some researchers prefer in-depth interviews while others
collect short written accounts of important life scenes. Some researchers
engage in a deep, hermeneutical process for each case while others employ
content-analysis systems to many cases in an effort to quantify results.
Within personality psychology, narrative researchers have gathered in-depth
case data from interviews and short written accounts through open-ended
questionnaires (McAdams, 1999; Singer, 2004). Most personality research-
ers, furthermore, have tried to quantify their data in order to test hypotheses
about life stories in select samples of individuals. For example, research has
examined the relations between aspects or features of life stories on the one
hand and traits (McAdams et al., 2004), social motives (Woike, 1995), val-
ues and personal ideologies (de St. Aubin, 1996), coping with stress (King,
Scollon, Ramsey, & Williams, 2000), mental health and well-being (Bauer &
McAdams, 2004), drug and alcohol abuse (Singer, 1997), and patterns of
interpersonal relationships (Thorne, 2000) on the other.
One standard procedure used in a number of studies is McAdams’s (1993)
Life Story Interview. The interviewer begins by asking the participant to
think about his or her life as if it were a book and to provide a brief outline
of that book. Then the interviewer zeroes in on particular scenes or mo-
ments that stand out in the story – high points, low points, and turning
Studying Lives in Time: A Narrative Approach 245
points, for example. For each of these scenes, the participant describes in
detail what happened in the event, who was there, what he or she was
thinking and feeling in the scene, and what he or she thinks the scene may
say about who the person is, was, or will be. Many of these key scenes
qualify for what Singer (1995) has termed self-defining memories – vivid,
affectively charged events from the past that express key hopes, fears, and
conflicts in life. The interviewer then moves to the future and asks the
participant to imagine what the next chapter in his or her life story might be
like, focusing on dreams and fears regarding the future. Toward the end of
the interview, the participant considers more abstract and philosophical
questions regarding the life story, articulating those religious and/or social
beliefs and values that may be important in the narrative and commenting
on what message or moral the story seems to express.
Depending on the purpose of the investigation, accounts obtained from
the Life Story Interview, and similar procedures, may be analyzed according
to many different systems, some of which emphasize structural features such
as narrative coherence and complexity and others of which emphasize story
content. With respect to content, a number of researchers have examined
Bakan’s (1966) distinction between agency and communion in life-narrative
accounts (e.g., Hermans, 1996; McAdams, 1985; McAdams, Hoffman,
Mansfield, & Day, 1996; Woike, 1995). Agency refers to characters’ strivings
to expand, assert, control, or defend the self, and is manifested in themes
related to achievement, power, status, and self-mastery. By contrast, com-
munion refers to characters’ strivings to connect the self with others, man-
ifested through themes of friendship and love, interpersonal
communication, caring for and helping others, and feeling a sense of com-
munity. In their various manifestations, agency and communion appear to
be highly salient themes in life stories and common dimensions upon which
stories can be said to differ in meaningful ways.
Another thematic distinction that has enjoyed currency in recent research
is that between contamination sequences and redemption sequences in life
narrative (McAdams & Bowman, 2001; McAdams, Reynolds, Lewis,
Patten, & Bowman, 2001). In a contamination sequence, an affectively
positive scene is rather suddenly transformed into something negative. The
main character’s initial state of strong joy or excitement suddenly gives way
to fear, sadness, shame, guilt, or some other negative emotional reaction.
The scene’s opening goodness is contaminated, ruined by what follows.
Most everybody can recall contamination sequence from their lives. But
some individuals use this narrative form in making sense of their lives more
often than do others. Research suggests that adults who express more
DAN P. McADAMS246
examples of contamination in their life narrative accounts tend to score
much lower on measures of psychological well-being and higher on meas-
ures of depression compared to individuals who show fewer contamination
sequences in life stories (McAdams et al., 2001). In a study of 74 midlife
adults, self-report measures of self-esteem, life satisfaction, and sense of life
coherence were negatively correlated with, and depression positively corre-
lated with, scores on contamination sequences derived from content analysis
of the life narrative interviews. Contamination sequence scores were also
significantly associated with a narrative measure of depressogenic attribut-
ional style – the tendency to attribute negative life events to internal, stable,
and global causes (Peterson & Seligman, 1984). But contamination sequence
scores from these same life stories proved to be a significantly strong pre-
dictor of depression than was depressogenic attributional style (Adler,
Kissel, & McAdams, in press).
In a redemption sequence, a bad or affectively negative life-story scene
turns good – the bad is salvaged or redeemed by the positive outcome that
ensues (McAdams et al., 2001). Perhaps not surprisingly, the redemptive
form of life storytelling tends to be positively associated with a number of
positive psychosocial features – such as life satisfaction among midlife
American adults and among college students (McAdams et al., 2001). In one
study, researchers coded written accounts of 10 key scenes in life narratives
from 125 college students. Redemption imagery in the stories was positively
associated with numerous self-report measures of well-being, such as life
satisfaction and purpose in life. The researchers also coded the overall
emotional quality of the stories, from positive to negative. The results
showed that redemptive stories are not exactly the same thing as happy,
positive stories about the self. Redemption sequences were only weakly
correlated with overall emotional positivity of the life stories. Furthermore,
redemption sequences proved to be substantially stronger predictors of well-
being than did the overall emotional tone of the narrative accounts. The
results suggest that it is not so much that happy people tell happy stories
about their lives. Instead, happy people tend to tell life stories filled with
episodes in which suffering is redeemed by positive outcomes.
Redemption sequences are at the center of a collection of narrative themes
that tend to characterize the life stories told by highly generative American
adults. In a series of studies, McAdams and colleagues (Mansfield & McAdams,
1996; McAdams & Bowman, 2001; McAdams, Diamond, de St. Aubin, &
Mansfield, 1997; McAdams, Ruetzel, & Foley, 1986; McAdams et al., 2001;
Van de Water & McAdams, 1989) collected life-narrative accounts from
midlife adults who score especially high on objective indices of generativity.
Studying Lives in Time: A Narrative Approach 247
Generativity is an adult’s concern for and commitment to promoting the
well-being of future generations. A growing body of research shows that
adults who score high on generativity measures tend to be more deeply
invested in a wide range of endeavors aimed at improving the world around
them – from parenting to volunteer work to voting (McAdams, 2001;
McAdams & de St. Aubin, 1992; Rossi, 2001). Compared to their coun-
terparts scoring lower in generativity, moreover, highly generative adults
tend to construct life stories that feature a wide range of variations on the
theme of redemption. In addition, their stories often feature these related
themes: (1) a childhood sense of feeling special or advantaged; (2) an early
sensitivity to the suffering or oppression of others; (3) the consolidation of a
simple but compelling personal ideology in adolescence and the commit-
ment to that ideology through the adult years; (4) tension between agentic
and communal strivings in adulthood; and (5) anticipating growth and fru-
ition for the future. Taken together, these themes comprise what McAdams
(2006b) calls the redemptive self. Table 1 summarizes the central themes of
the redemptive self.
The redemptive self represents a particular kind of narrative identity that
serves well the psychosocial needs of many highly generative men and
women in their midlife years. The story reinforces their commitments to
promoting the well-being of future generations. Believing that they were
once the beneficiaries of early advantages in life, they feel some obligation to
give back for the blessings they received. Their stories tell them that early on
they were chosen or advantaged in some way in a world where many other
people suffer. The contrast between their sense of personal blessing and their
early awareness of the suffering of others may set up in their story a moral
challenge and a felt obligation to be of some good use to others (Colby &
Damon, 1992). The beliefs they consolidated in adolescence serve as their
steady guides. In that generativity is often hard work, they manage to per-
severe in part because their story tells them that bad things are often over-
come, suffering gives way to redemption. They look to the future with hope,
even as they realize that the world is a very dangerous place. The hero in the
story is the dogged, if sometimes guileless, protagonist who remains upbeat
and focused on making a positive difference in his or her family, neighbor-
hood, church, community, society – even against the odds.
McAdams (2006b) also suggests that the redemptive self, for better and
for worse, may represent a characteristically American way of narrating a
caring and productive life at midlife. As highly generative American adults
shape their lives into redemptive narratives, they implicitly draw upon an
optimistic and highly individualistic understanding of the life course
DAN P. McADAMS248
that celebrates personal redemption through the discourses of atonement
(religion), emancipation (politics), recovery (medicine), self-actualization
(psychology), and upward social mobility (economics). These discourses
have their origins in such canonical American texts as the spiritual
autobiographies of the New England Puritans, Benjamin Franklin’s auto-
biography, Emerson’s 19th-century lectures and essays on ‘‘self-reliance,’’
Horatio Alger stories, the Gettysburg Address, and the powerful narratives
written by escaped slaves before the American Civil War. Variations on
the same themes run through 20th-century American autobiography and
fiction, American television and movies, and the burgeoning literature
of American self-help. Translated into myths of national identity, these
themes are reflected in such quintessentially American notions as ‘‘the cho-
Table 1. The Redemptive Self: Six Themes Characterizing Life Stories
Constructed by Highly Generative American Adults.
1. Early advantage As a young child, the story’s protagonist enjoys a special
advantage or blessing that singles him or her out in the family
or vis-a-vis peers. From an early age onward, the protagonist
feels that he or she is special in a positive way.
2. Suffering of others Early in the story, the protagonist witnesses the suffering or
misfortune of other people and feels sympathy or empathy
for them. Objects of the protagonist’s concern might include
the sick, dying, disabled, mentally ill, economically
disadvantaged, or any of a number of other groups or
individuals that might require special care or help.
3. Ideological
steadfastness
By adolescence, the protagonist has established a clear and
coherent belief system that governs his or her life. The belief
system, often rooted in religion, remains relatively stable and
steadfast over time. Once the belief system is established, the
protagonist does not experience profound ideological doubt,
uncertainty, or crisis.
4. Redemption
sequences
Bad or affectively negative life events are immediately followed
by good or affectively positive outcomes. The bad scene is
redeemed, salvaged, and made better by what follows.
5. Power vs. love As an adult, the protagonist repeatedly finds that strong agentic
desires to distinguish the self by having a positive impact on
the world repeatedly conflict with equally strong communal
desires to form loving relationships and be accepted by others
as an equal.
6. Prosocial future In looking to the future chapters of the life story, the
protagonist sets goals that aim to benefit society in general or
its institutions.
Source: McAdams (2006b) and McAdams et al. (1997).
Studying Lives in Time: A Narrative Approach 249
sen people,’’ ‘‘manifest destiny,’’ and ‘‘the American dream.’’ The life stories
constructed by highly generative American adults today employ rich met-
aphors and ways of thinking about identity that Americans have both
cherished and contested, found both inspiring and problematic, for over 300
years – from the Puritan landing in New England in 1630 to the most recent
episodes of The Oprah Winfrey Show.
NARRATIVE, CULTURE, AND THE LIFE COURSE
Research on the redemptive self suggests that life stories may say as much
about the culture and society wherein the narrator lives as they do about the
narrator’s life itself. Many social scientists see life stories as cultural texts
(Rosenwald & Ochberg, 1992; Shotter & Gergen, 1989). Life stories mirror
the culture wherein the story is made and told. Stories live in culture. They
are born, they grow, they proliferate, and they eventually die according to the
norms, rules, and traditions that prevail in a given society, according to the
society’s implicit understandings of what counts as a tellable story, a tellable
life. As Rosenwald (1992) puts it, ‘‘when people tell life stories, they do so in
accordance with the models of intelligibility specific to the culture’’ (p. 265).
Habermas and Bluck (2000) contend that before a person can formulate a
convincing life story, he or she must become acquainted with the culture’s
concept of biography. In modern Western cultures, Denzin (1989) and
McAdams (1996) suggest, biographies are expected to begin in the family, to
involve growth and expansion in the early years, to trace later problems back
to earlier conflicts, to incorporate epiphanies and turning points that mark
change in the protagonist’s quest, and to be couched in the discourse of
progress versus decline. But other societies tell about lives in different ways
and have different views of what constitutes a good story to tell (Gregg, 1991).
Even within a given society, furthermore, different stories compete for
dominance and acceptance. Feminists such as Heilbrun (1988) argue that in
Western societies many women ‘‘have been deprived of the narratives, or the
texts, plots, or examples, by which they might assume power over – take
control over – their lives’’ (p. 17). It is painfully clear that life stories echo
gender and class constructions in society and reflect, in one way or another,
prevailing patterns of hegemony in the economic, political, and cultural
contexts wherein human lives are situated. Power elites in society privilege
certain life stories over others, and therefore a number of narrative re-
searchers and clinicians seek to give voice and expression to forms of life
DAN P. McADAMS250
narrative that have traditionally been suppressed or marginalized (Franz &
Stewart, 1994; White & Epston, 1990).
At the same time, it is clear that people, even those whose lives do not
affirm a society’s dominant narratives, do not simply acquiesce to prevailing
cultural norms and standards with respect to the narrative construction of
lives. Human agency confronts the instrumental and expressive givens in a
particular social ecology and seeks to appropriate what it can, picking and
choosing among different narrative options, making a life and a life story
that may even defy the status quo. Gjerde (2004) argues that the relation
between self and culture is complex and often contested. People sometimes
resist the norms to construct individual life patterns that defy cultural con-
vention. Gjerde (2004) writes: ‘‘Culture can be said to exist as contested
representations situated in public domains or institutions in which power is
both exercised and resisted’’ (p. 146). Consequently, life stories represent
something of a resultant compromise between human agency and social
structure. Narrative identity is a psychosocial construction – a joint product
of individual and society.
Life stories draw on the stories that people learn as active participants in
culture – stories about childhood, adolescence, adulthood, and aging.
Stories capture and elaborate metaphors and images that are especially
resonant in a given culture. Stories distinguish between what culture glorifies
as good characters and vilifies as bad characters, and they present the many
varieties who fall in between. Stories depict full and fragmented lives that
are exciting, frightening, infuriating, enlightening, admirable, heroic, digni-
fied, ignoble, disgusting, wise, foolish, and boring. Stories teach people how
to live and what their lives may mean. Stories also spell out expected stages
and trajectories in the life course. From a narrative perspective, human lives
do not follow a natural stage-like progression; they do not conform readily
to social roles and norms; and they do not follow the predictable trends
ascribed to a given dispositional trait. Instead, life stories follow whatever
form the individual and society manage to work out. In the best cases, that
form meets what the individual feels to be his or her inner needs and
aspirations while making a productive contribution within a specified psy-
chosocial niche. In cases that are less ideal but probably more common,
individuals struggle to find life-narrative forms that more-or-less see them
through a difficult life terrain, amidst personal setbacks, failures, frustra-
tions, and a demanding and stubbornly uncooperative world. People do the
best they can in an effort to make sense of their lives in time, even under
trying circumstances.
Studying Lives in Time: A Narrative Approach 251
Culture, then, provides each person with an extensive menu of stories
about how to live, and each person chooses from the menu (McAdams,
2006b). Because different people within a given culture have different
experiences and opportunities, no two people get exactly the same menu.
Furthermore, a person cannot eat everything on the menu, so narrative
choices spell out a person’s relationship to culture. When the food comes
from the kitchen, people doctor it to their own tastes. They add pepper and
salt; they mix things up and throw some things away; they nibble from
somebody else’s plate; they may even send the order back and ask to see the
menu again. This is to say that individuals select and appropriate in the
making of narrative identity. They choose from competing stories, rejecting
many others, and they modify the stories they choose to fit their own unique
life, guided by the unique circumstances of their social, political, and eco-
nomic worlds, by their family backgrounds and educational experiences,
and by their dispositional traits and characteristic adaptations. A person
constructs a narrative identity by appropriating stories from culture. Self
and culture come to terms with each other through the narrative.
In conclusion, a narrative approach to understanding lives in time com-
plements prevailing models of the human life course that posit either (1)
developmental stages (e.g., Levinson, 1978), (2) life trajectories (e.g., Elder,
1995), or (3) personality traits (e.g., McCrae & Costa, 1990), while calling
into question the extent to which any model of the life course is itself
something of a narrative convention. In describing the life course in terms of
predictable passages and transitions, contextualized life trajectories, or sta-
ble dispositional traits, social scientists are prioritizing certain stories about
life over others. Yet, the imposition of one kind of narrative frame over
others is probably an inescapable feature of theory-making in the social
sciences. Indeed, some general stories about the life course may have more
scientific validity than others. Data can be garnered to support many dif-
ferent models of the life course, including those outlining developmental
stages, contextualized life trajectories, and stable dispositional traits. One
advantage, however, of narrative models for the life course is that they
prioritize the individual’s own construction of lives in time, rather
than projecting the construction offered by theorists themselves. Of course,
narrative approaches themselves are a projection of theorists’ preferences.
In prioritizing the story over other forms of human expression, narrative
approaches assume that storytelling is a natural, even universal, human
tendency, and that it represents a dominant way in which people make sense
of their lives in time (Bruner, 1990). Reasonable people may wish to take
exception to that assumption.
DAN P. McADAMS252
Narrative approaches focus attention on the meanings that people make
as they live their lives over time and try, over time, to make sense of them.
And narrative approaches show that those meanings are more than either
the idiosyncratic tales of self-contained individuals or the passive recitations
of society’s dominant narrative forms. Instead, life stories represent the
creative, contested, and constantly evolving interplay between a storytelling
agent and a complexly structured and storied world. As lives play out over
time, stories ultimately emerge, narrated in fits and starts by a compromised
but determined narrator who doubles as the protagonist of the tale. The
author may feel that it is his own story to tell. But even in the most heroic
tales, authorship is always joint – a project shared by the narrator himself
and the world wherein his story is told.
REFERENCES
Adler, J., Kissel, E., & McAdams, D. P. (in press). Emerging from the CAVE: Attributions in
life narratives and their relations to psychological well-being. Cognitive Therapy and
Behavior.
Allport, G. W. (1937). Personality: A psychological interpretation. New York: Holt, Rinehart &
Winston.
Ambady, G., Hallahan, M., & Rosenthal, R. (1995). On judging and being judged accurately in
zero-acquaintance situations. Journal of Personality and Social Psychology, 69, 518–529.
Angus, L. E., & McLeod, J. (Eds) (2004). Handbook of narrative and psychotherapy. London:
Sage.
Arnett, J. J. (2000). Emerging adulthood: A theory of development from the late teens through
the twenties. American Psychologist, 55, 469–480.
Bakan, D. (1966). The duality of human existence. Boston: Beacon Press.
Bauer, J. J., & McAdams, D. P. (2004). Growth goals, maturity, and well-being. Developmental
Psychology, 40, 114–127.
Baumeister, R. F. (1994). Introduction to symposium: Stories as samples. Personality and Social
Psychology Bulletin, 20, 649.
Bertaux, D. (Ed.) (1981). Biography and society. Beverly Hills, CA: Sage.
Birren, J., Kenyon, G., Ruth, J., Shroots, J. J. G., & Svendson, J. (Eds) (1996). Aging and
biography. New York: Springer.
Bouchard, T. J., Jr., Lykken, D. T., McGue, M., Segal, N. L., & Tellegen, A. (1990). Sources of
human psychological differences: The Minnesota study of twins reared apart. Science,
250, 223–228.
Bruner, J. S. (1990). Acts of meaning. Cambridge, MA: Harvard University Press.
Cantor, N. (1990). From thought to behavior: ‘‘Having’’ and ‘‘doing’’ in the study of person-
ality and cognition. American Psychologist, 45, 735–750.
Casey, K. (1996). The new narrative research in education. In: M. W. Apple (Ed.), Review of
research in education (Vol. 21, pp. 211–253). Washington, DC: American Educational
Research Association.
Studying Lives in Time: A Narrative Approach 253
Charme, S. (1984). Meaning and myth in the study of lives: A Satrean perspective. Philadelphia,
PA: University of Pennsylvania Press.
Cohler, B. J. (1982). Personal narrative and the life course. In: P. Baltes, & O. G. Brim (Eds),
Lifespan development and behavior (Vol. 4, pp. 205–241). New York: Academic Press.
Colby, A., & Damon, W. (1992). Some do care: Contemporary lives of moral commitment.
New York: The Free Press.
Conway, M. A., & Pleydell-Pearce, C. W. (2000). The construction of autobiographical mem-
ories in the self-memory system. Psychological Review, 107, 261–288.
Costa, P. T., Jr., & McCrae, R. R. (1994). Set like plaster? Evidence for the stability of adult
personality. In: T. F. Heatherton & &J. L. Weinberger (Eds), Can personality change?
(pp. 21–40). Washington, DC: APA Books.
Dannefer, D. (1984). Adult development and social theory: A paradigmatic reappraisal.
American Sociological Review, 49, 100–116.
de St. Aubin, E. (1996). Personal ideology polarity: Its emotional foundation and its mani-
festation in individual value systems, religiosity, political orientation, and assumptions
concerning human nature. Journal of Personality and Social Psychology, 71, 152–165.
Denzin, N. (1989). Interpretive biography. Newbury Park, CA: Sage.
Edel, L. (1978). Biography: A manifesto. Biography, 1, 1–3.
Elder, G. H., Jr. (1995). The life course paradigm: Social change and individual development.
In: P. Moen, G. H. Elder Jr. & K. Luscher (Eds), Examining lives in context
(pp. 101–139). Washington, DC: APA Books.
Ely, M. (2003). Braiding essence: Learning what I thought I already knew about teaching
qualitative research. In: R. Josselson, A. Lieblich & D. P. McAdams (Eds), Up close and
personal: The teaching and learning of narrative research (pp. 215–238). Washington, DC:
APA Books.
Emmons, R. A. (1986). Personal strivings: An approach to personality and subjective well-
being. Journal of Personality and Social Psychology, 51, 1058–1068.
Erikson, E. H. (1963). Childhood and society (2nd ed.). New York: Norton.
Fivush, R., & Haden, C. (Eds) (2003). Autobiographical memory and the construction of a
narrative self. Mahwah, NJ: Erlbaum.
Franz, C., & Stewart, A. J. (Eds) (1994). Women creating lives. Boulder, CO: Westview.
Freund, A. M., & Baltes, P. B. (2000). The orchestration of selection, optimization, and
compensation: An action-theoretical conceptualization of a theory of developmental
regulation. In: W. J. Perrig & A. Grob (Eds), Control of human behavior, mental proc-
esses, and consciousness (pp. 35–58). Mahwah, NJ: Erlbaum.
Friedman, H. S., Tucker, J. S., Tomlinson-Keasy, C., Schwartz, J. E., Wingard, D. L., & Criqui,
M. H. (1993). Does childhood personality predict longevity? Journal of Personality and
Social Psychology, 65, 176–185.
Funder, D., & Colvin, C. R. (1991). Explorations in behavioral consistency: Properties of per-
sons, situations, and behaviors. Journal of Personality and Social Psychology, 60, 773–794.
Gabriel, Y. (2000). Storytelling in organizations: Facts, fictions, and fantasies. New York:
Oxford University Press.
Gergen, K. J. (1991). The saturated self. New York: Basic Books.
Giddens, A. (1991). Modernity and self-identity: Self and society in the late modern age.
Stanford, CA: Stanford University Press.
Gjerde, P. F. (2004). Culture, power, and experience: Toward a person-centered cultural
psychology. Human Development, 47, 138–157.
DAN P. McADAMS254
Goldberg, L. R. (1993). The structure of phenotypic personality traits. American Psychologist,
48, 26–34.
Gregg, G. (1991). Self-representation: Life-narrative studies in identity and ideology. New York:
Greenwood Press.
Gutmann, D. (1987). Reclaimed powers. Evanston, IL: Northwestern University Press.
Habermas, T., & Bluck, S. (2000). Getting a life: The emergence of the life story in adolescence.
Psychological Bulletin, 126, 748–769.
Hankiss, A. (1981). On the mythological rearranging of one’s life history. In: D. Bertaux (Ed.),
Biography and society (pp. 203–209). Beverly Hills, CA: Sage.
Heilbrun, C. (1988). Writing a woman’s life. New York: Norton.
Hermans, H. J. M. (1996). Voicing the self: From information processing to dialogical inter-
change. Psychological Bulletin, 119, 31–50.
Holstein, J. A., & Gubrium, J. F. (2000). The self we live by: Narrative identity in a postmodern
world. New York: Oxford University Press.
Hooker, K. (2002). New directions for research in personality and aging: A comprehensive
model linking levels, structures, and processes. Journal of Research in Personality, 36,
318–334.
Hooker, K., & McAdams, D. P. (2003). Personality reconsidered: A new agenda for aging
research. Journal of Gerontology: Psychological Sciences, 58B, P296–P304.
Howe, M. L., & Courage, M. L. (1997). The emergence and early development of auto-
biographical memory. Psychological Review, 104, 499–523.
John, O. P., & Srivastava, S. (1999). The big five trait taxonomy: History, measurement, and
theoretical perspectives. In: L. Pervin & O. P. John (Eds), Handbook of personality:
Theory and research (2nd ed., pp. 102–138). New York: Guilford.
Josselson, R., & Lieblich, A. (Eds) (1993). The narrative study of lives. Thousand Oaks, CA:
Sage.
Kenrick, D. T., & Funder, D. C. (1988). Profiting from controversy: Lessons from the person-
situation debate. American Psychologist, 43, 23–34.
King, L., Scollon, C. K., Ramsey, C. M., &Williams, T. (2000). Stories of life transition: Happy
endings, subjective well-being, and ego development in parents of children with Down
Syndrome. Journal of Research in Personality, 34, 509–536.
Kluckhohn, C., & Murray, H. A. (1953). Personality formation: The determinants. In:
C. Kluckhohn, H. A. Murray & D. M. Schneider (Eds), Personality in nature, society,
and culture (pp. 53–67). New York: Alfred A. Knopf.
Kohli, M. (1981). Biography: Account, text, method. In: D. Bertaux (Ed.), Biography and
society. Beverly Hills, CA: Sage.
Langbaum, J. (1982). The mysteries of identity: A theme in modern literature. Chicago:
University of Chicago Press.
Levinson, D. J. (1978). The seasons of a man’s life. New York: Alfred A. Knopf.
Lieblich, A., McAdams, D. P., & Josselson, R. (Eds) (2004). Healing plots: The narrative basis
of psychotherapy. Washington, DC: APA Books.
Little, B. R. (1999). Personality and motivation: Personal action and the conative evolution. In:
L. Pervin, & O. P. John (Eds), Handbook of personality: Theory and research (2nd ed.,
pp. 501–524). New York: Guilford.
MacIntyre, A. (1984). After virtue. Notre Dame, IN: University of Notre Dame Press.
Mandler, J. M. (1984). Stories, scripts, and scenes: Aspects of schema theory. Hillsdale, NJ:
Erlbaum.
Studying Lives in Time: A Narrative Approach 255
Mansfield, E., & McAdams, D. P. (1996). Generativity and themes of agency and communion
in adult autobiography. Personality and Social Psychology Bulletin, 22, 721–731.
Maruna, S. (2001). Making good: How ex-convicts reform and rebuild their lives. Washington,
DC: APA Books.
Matthews, G., Deary, I. J., & Whiteman, M. C. (2003). Personality traits (2nd ed.). Cambridge:
Cambridge University Press.
McAdams, D. P. (1985). Power, intimacy, and the life story: Personological inquiries into iden-
tity. New York: Guilford.
McAdams, D. P. (1993). The stories we live by. New York: Guilford.
McAdams, D. P. (1994). Can personality change? Levels of stability and growth in personality
across the life span. In: T. F. Heatherton & J. L. Weinberger (Eds), Can personality
change? (pp. 299–313). Washington, DC: APA Books.
McAdams, D. P. (1995). What do we know when we know a person? Journal of Personality, 63,
365–396.
McAdams, D. P. (1996). Personality, modernity, and the storied self: A contemporary frame-
work for studying persons. Psychological Inquiry, 7, 295–321.
McAdams, D. P. (1997). A conceptual history of personality psychology. In: R. Hogan,
J. Johnson & S. Briggs (Eds), Handbook of personality psychology (pp. 3–39). San Diego,
CA: Academic Press.
McAdams, D. P. (1999). Personal narratives and the life story. In: L. Pervin & O. P. John (Eds),
Handbook of personality: Theory and research (2nd ed., pp. 478–500). New York:
Guilford.
McAdams, D. P. (2001). Generativity in midlife. In: M. Lachman (Ed.), Handbook of midlife
development (pp. 395–443). New York: Wiley.
McAdams, D. P. (2006a). The person: A new introduction to personality psychology (4th ed.).
New York: Wiley.
McAdams, D. P. (2006b). The redemptive self: Stories Americans live by. New York: Oxford
University Press.
McAdams, D. P., Anyidoho, N. A., Brown, C., Huang, Y. T., Kaplan, B., & Machado, M. A.
(2004). Traits and stories: Links between dispositional and narrative features of per-
sonality. Journal of Personality, 72, 761–784.
McAdams, D. P., & Bowman, P. J., Jr. (2001). Narrating life’s turning points: Redemption and
contamination. In: D. P. McAdams, R. Josselson & A. Lieblich (Eds), Turns in the road:
Narrative studies of lives in transition (pp. 3–34). Washington, DC: APA Books.
McAdams, D. P., & de St. Aubin, E. (1992). A theory of generativity and its assessment through
self-report, behavioral acts, and narrative themes in autobiography. Journal of Person-
ality and Social Psychology, 62, 1003–1015.
McAdams, D. P., de St. Aubin, E., & Logan, R. L. (1993). Generativity among young, midlife,
and older adults. Psychology and Aging, 8, 221–230.
McAdams, D. P., Diamond, A., de St. Aubin, E., & Mansfield, E. D. (1997). Stories of com-
mitment: The psychosocial construction of generative lives. Journal of Personality and
Social Psychology, 72, 678–694.
McAdams, D. P., Hoffman, B. J., Mansfield, E., & Day, R. (1996). Themes of agency and
communion in significant autobiographical scenes. Journal of Personality, 64, 339–378.
McAdams, D. P., Reynolds, J., Lewis, M., Patten, A., & Bowman, P. J., Jr. (2001). When bad
things turn good and good things turn bad: Sequences of redemption and contamination
DAN P. McADAMS256
in life narrative, and their relation to psychosocial adaptation in midlife adults and in
students. Personality and Social Psychology Bulletin, 27, 472–483.
McAdams, D. P., Ruetzel, K., & Foley, J. M. (1986). Complexity and generativity at midlife:
Relations among social motives, ego development, and adults’ plans for the future.
Journal of Personality and Social Psychology, 50, 800–807.
McCrae, R. R., & Costa, P. T., Jr. (1990). Personality in adulthood. New York: Guilford.
Mischel, W. (1968). Personality and assessment. New York: Wiley.
Mischel, W., & Shoda, Y. (1995). A cognitive-affective-system theory of personality: Recon-
ceptualizing situations, dispositions, dynamics, and invariance in personality structure.
Psychological Review, 102, 246–268.
Murray, S. L., & Holmes, J. (1994). Storytelling in close relationships: The construction of
confidence. Personality and Social Psychology Bulletin, 20, 650–663.
Peterson, C., & Seligman, M. E. P. (1984). Causal explanations as a risk factor for depression:
Theory and evidence. Psychological Review, 91, 347–374.
Polkinghorne, D. (1988). Narrative knowing and the human sciences. Albany, NY: SUNY
Press.
Ricoeur, P. (1984). Time and narrative. Chicago: University of Chicago Press.
Roberts, B. W., & Del Vecchio, W. (2000). The rank-order consistency of personality from
childhood to old age: A quantitative review of longitudinal studies. Psychological Bul-
letin, 126, 3–25.
Roberts, B. W., & Pomerantz, E. (2004). On traits, situations, and their integration: A de-
velopmental perspective. Personality and Social Psychology Review, 8, 402–416.
Rosenwald, G. (1992). Conclusion: Reflections on narrative self-understanding. In: G. Rosenwald
& R. L. Ochberg (Eds), Storied lives: The cultural politics of self-understanding
(pp. 265–289). New Haven, CT: Yale University Press.
Rosenwald, G. C., & Ochberg, R. L. (Eds) (1992). Storied lives: The cultural politics of self-
understanding. New Haven, CT: Yale University Press.
Rossi, A. (Ed.) (2001). Caring and doing for others. Chicago: University of Chicago Press.
Sarbin, T. (1986). The narrative as a root metaphor for psychology. In: T. Sarbin (Ed.),
Narrative psychology: The storied nature of human conduct (pp. 3–21). New York:
Praeger.
Schank, R., & Abelson, R. (1995). Knowledge and memory: The real story. In: R. S. Wyer
(Ed.), Advances in social cognition (Vol. 8, pp. 1–86). Hillsdale, NJ: Erlbaum.
Schwartz, S. H. (1994). Are there universal aspects in the structure and content of human
values? Journal of Social Issues, 50, 19–46.
Sheldon, K. M. (2004). The psychology of optimal being: An integrated, multi-level perspective.
Mahwah, NJ: Erlbaum.
Shotter, J., & Gergen, K. J. (Eds) (1989). Texts of identity. London: Sage.
Singer, J. A. (1995). Seeing one’s self: Locating narrative memory in a framework of person-
ality. Journal of Personality, 63, 429–457.
Singer, J. A. (1997). Message in a bottle. New York: The Free Press.
Singer, J. A. (2004). Narrative identity and meaning making across the adult lifespan: An
introduction. Journal of Personality, 72, 437–459.
Thorne, A. (2000). Personal memory telling and personality development. Personality and
Social Psychology Review, 4, 45–56.
Tomasello, M. (2000). Culture and cognitive development. Current Directions in Psychological
Science, 9, 37–40.
Studying Lives in Time: A Narrative Approach 257
Vaillant, G. E., & Milofsky, E. (1980). The natural history of male psychological health: IX.
Empirical evidence for Erikson’s model of the life cycle. American Journal of Psychiatry,
137, 1349–1359.
Van de Water, D., & McAdams, D. P. (1989). Generativity and Erikson’s ‘‘belief in the
species’’. Journal of Research in Personality, 23, 435–449.
Wellman, H. M. (1993). Early understanding of mind: The normal case. In: S. Baron-Cohen,
H. Tager-Flusberg & D. J. Cohen (Eds), Understanding other minds: Perspectives from
autism (pp. 10–39). New York: Oxford University Press.
White, M., & Epston, M. (1990). Narrative means to therapeutic ends. New York: Norton.
Winter, D. G., John, O. P., Stewart, A. J., Klohnen, E. C., & Duncan, L. E. (1998). Traits and
motives: Toward an integration of two traditions in personality research. Psychological
Review, 105, 230–250.
Woike, B. (1995). Most memorable experiences: Evidence for a link between implicit and
explicit motives and social-cognitive processes in everyday life. Journal of Personality and
Social Psychology, 68, 1081–1091.
DAN P. McADAMS258
LIFE COURSE ANALYSIS:
TWO (COMPLEMENTARY)
CULTURES? SOME REFLECTIONS
WITH EXAMPLES FROM THE
ANALYSIS OF THE TRANSITION
TO ADULTHOOD
Francesco C. Billari
1. LIFE COURSE ANALYSIS IN DEMOGRAPHY
The life course approach as an interdisciplinary program of study has been
under development since the mid-1970s (Mayer & Tuma, 1990). The idea of
studying the unfolding of individual lives within their local and broader
context has unavoidably brought life course scholars to emphasize
complexity rather than simplicity. In their end-of-millennium review, Giele
and Elder (1998) identify four chief elements as fundamentally shaping life
courses: individual development (human agency), history and culture
(location in time and place), social relations (linked lives). The intersection
of such elements constitutes the fourth chief element: the timing of lives. In
the life course approach, a set of interconnected trajectories lies at the heart
of the analysis; trajectories are themselves shaped by events.
Towards an Interdisciplinary Perspective on the Life Course
Advances in Life Course Research, Volume 10, 261–281
Copyright r 2005 by Elsevier Ltd.
All rights of reproduction in any form reserved
ISSN: 1040-2608/doi:10.1016/S1040-2608(05)10010-0
261
Demographers who are familiar with the life course approach are aware
that the key elements outlined by Giele and Elder are naturally linked to
notions used to define the coordinates of populations in demography. The
human agency concept lies behind the use of age as a privileged time axis.
Location in time and, to an extent, the idea of linked lives suggests cohorts
or groups of cohorts as basic descriptive units for comparison. History and
culture emphasize the importance of period and location in space. As events
lie at the very heart of the tradition of demography, it is not a surprise that
the life course approach has been particularly influential in demographic
research (Dykstra & van Wissen, 1999).
Large-scale surveys collecting individual-level retrospective demographic
history, starting from the World Fertility Survey (WFS) have given the
biggest impetus to collect in most demographic surveys data on the timing
of events in relevant trajectories (Hobcraft & Murphy, 1986). Elsewhere
(Billari, 2003), life course analysis has been defined as the statistical analysis
of data on the timing of events (when do events happen?), their sequencing
(in which order do events happen?), and their quantum (how many events
happen?). Ideally, life course analysis includes the possibility to analyze the
timing, sequencing, and quantum of events as depending on the elements
mentioned by Giele and Elder: individual-level human development, social
relations, location in time and place.
A progressively increasing complexity has shaped – as we shall see in more
detail in this paper – methods for the analysis of life courses (Billari, 2003).
While scholars who advocate a wider use of qualitative research see their
approach as a way to cope with the complexity of life courses, those who,
like myself, feel more comfortable with an emphasis on quantitative research
have more and more moved to complex methods as well. Complexity comes
at some costs, but it also brings some paybacks. We shall try to disentangle
costs and payback by looking at a specific field of study: the transition to
adulthood. The study of the transition to adulthood has been a primary
topic for life course scholars, and it has greatly benefited from advances in
the life course approach and in life course analysis. The need for method-
ological advances in this area comes from the peculiar ‘‘demographic den-
sity’’ of early adulthood: in advanced societies several events happen during
this period, and diversity between individual lives becomes importantly vis-
ible (Rindfuss, 1991). Processes that have usually been studied using a life
course approach include leaving the parental home, starting and/or dis-
solving a union, having a child, migration, job entry and exit, retirement.
In this paper I argue that there are two main approaches to life course
analysis (and in particular the analysis of demographic life courses), and
FRANCESCO C. BILLARI262
that these two approaches serve complementary, and equally necessary,
aims. The first is the event-based approach, based in general on event history
analysis; the latter set of techniques has become one of the basic ingredients
of advanced methodological courses in population sciences during the last
decades of the twentieth century. Connected to this approach, I review
developments from the econometric literature on program evaluation, which
aims at discovering causal relationships within life courses and appears still
to not have strongly influenced life course research. The event-based ap-
proach is mainly targeted at the explanation of life courses, more specifically
at the decomposition of the life course picture into the raw materials
(in particular, events) that constitute it, and in the search for the causes that
underlie the timing of these events. The second is the holistic approach,
mostly relying on sequence analysis, which has had a less significant impact
on life course research with respect to the event-based approach. This sec-
ond approach is mainly targeted at what we may define complex description
of life courses, or at the construction of what sociologists will define ‘‘ideal-
types’’ of trajectories. Looking at life courses as conceptual units, analyses
based on the holistic approach aim at gazing the whole picture, in a complex
way that somehow parallels what can be obtained using qualitative analysis.
Both approaches have been supported and have become widespread with
the availability of corresponding specific software packages.
The paper is structured as follows. In Section 2, I start from analyzing a
debate between two cultures in statistical analysis and argue that the main
approaches in life course analysis are related to these two cultures. In
Section 3, I shortly review the ‘‘raw material’’, event-based approach, while
in Section 4, I review the ‘‘whole picture’’, holistic approach. Further re-
flections on the complementarity between the two approaches, future re-
search directions, and needs for data collection are discussed in Section 5.
2. TWO CULTURES? A DEBATE IN STATISTICS AND
ITS IMPLICATIONS FOR LIFE COURSE ANALYSIS
Life course analysis inevitably and heavily relies on statistics. It is thus
useful to look at the debate within the statistical community to better grasp
present and future directions in life course analysis. The idea that there are
‘‘two cultures’’ in statistical modeling has been launched in a controversial
paper by Breiman (2001), published in ‘‘Statistical Science’’. The paper is
followed by comments of other statisticians and by a rejoinder of the author.
Life Course Analysis: Two (Complementary) Cultures? 263
The two cultures to which Breiman refers are: (1) a ‘‘data modeling culture’’,
which is the mainstream in statistics, which assumes that ‘‘data are gener-
ated by a given stochastic data model’’; (2) an ‘‘algorithmic modeling cul-
ture’’, clearly favored by Breiman, which treats the data generation
mechanism as unknown. According to Breiman, the focus on data models
has ‘‘led to irrelevant theory and questionable scientific conclusions’’, ‘‘kept
statisticians from using more suitable algorithmic models’’, and ‘‘prevented
statisticians from working on exciting new problems’’. On the other hand,
algorithmic issues point to the ‘‘multiplicity of good models’’ – illustrated by
reference to the Japanese movie ‘‘Rashomon’’ to the conflict between sim-
plicity and (predictive) accuracy – the Occam razor issue, to the dimen-
sionality problem ‘‘curse or blessing?’’ In particular, these points relate to
the issue that prediction may be improved when using simultaneously more
than one model of the same data, and that using more complex models
improves prediction at the expense of interpretability (the latter being
in practical applications, according to Breiman, is a kind of second-
order need), and to the importance of adding dimensionality to problems.
Breiman’s thought has clearly been shaped by his experience as a statistical
consultant for businesses; it is clear that in this type of experience data have
to be taken as ‘‘given’’.
The most important comment – in terms of ‘‘culture’’ confrontation – on
Breiman’s paper is the one by Cox, which is in turn criticized in Breiman’s
rejoinder. Cox argues that the starting point is not ‘‘data’’ but ‘‘an issue, a
question or a scientific hypothesis’’ and that real scientific applications are
targeted at unraveling causal links using statistics, not only in form of the
analysis of collected data, but also in terms of data collection design. Model-
based statistical techniques provide, in his view, the best opportunity to
illuminate causality. It is clear that Cox’s thoughts have been shaped by the
interaction with the scientific community in the medical and social sciences
rather than with business or sciences involving very complex data (for ex-
ample, astronomy or genomics).
We can draw some lessons and inspiration from the Breiman–Cox debate
for what follows. The event-based approach to life course analysis is strictly
connected to what Breiman defines the ‘‘data modeling culture’’ – not by
chance is Cox one of the founders of modern event history analysis (Cox,
1972). Also, connected to the arguments of Cox, and differently from stat-
isticians who specialize in data analysis, scientists interested in the study of
the life course do not necessarily have to take data as given, and sometimes
they can design a study in view of unraveling the cause(s) behind the timing
of a certain event. Scientific hypotheses may take the shape of a parametric
FRANCESCO C. BILLARI264
or semi-parametric statistical model, and statistical techniques may help us
in comparing the goodness-of-fit of such models. For instance, in the case of
models for the age at first marriage or first union, theory-based parametric
formulation of the hazard rate have been described and can possibly be
compared. As we shall see more in detail later in this paper, the statistical
significance of the impact of covariates on hazard rates can be tested using
standard approaches, and this has been extensively used in event history
analysis looking for ‘‘causal’’ approaches (see Blossfeld & Rohwer, 2002).
Nevertheless, the algorithmic culture defended by Breiman has also some-
thing to contribute in the field of life course analysis. First of all, it is not
always possible in this field to design a study in view of performing a specific
study. An important example is the study of archival data, collected for
administrative needs, and used by historical demographers. One can say
without doubt that analyses of archival data have contributed greatly to the
analysis of life courses. Second, as the life course is a complex set of in-
terrelated trajectories, disentangling causal links may not be the only aim;
more specifically, algorithmic models can contribute greatly to the analysis
of life courses as whole conceptual units. A very similar argument to the one
illustrated here is presented by Ritschard and Oris (in this volume), who
distinguish between a ‘‘statistical modeling’’ and a ‘‘data mining’’ approach
to the analysis of life course data in demography – it is not by chance that
they find this important distinction starting from issues raised in historical
demography, a field in which the use of archival data has allowed to reach
important scientific knowledge.
These considerations justify the view that is taken in the remainder of this
paper: both approaches are useful, in a complementary way. In other words,
would we only ‘‘adopt’’ one of the two approaches (or the underlying sta-
tistical culture), we will know less about life courses. They are of different
importance for different problems, and a review and reflection on these
situations is necessary. Let us consider them in turn, starting with the event-
based approach.
3. THE EVENT-BASED, OR ‘‘CAUSALITY’’, CULTURE
3.1. Event History Analysis: from the ‘‘Causal Approach’’ to
‘‘Multilevel and Multiprocess Modeling’’
The set of statistical techniques that is now broadly defined as event history
analysis has become since the 1980s as one of the principal toolkits of
Life Course Analysis: Two (Complementary) Cultures? 265
demography and life course research in general (Allison, 1984; Yamaguchi,
1991; Courgeau & Lelievre, 1992; Petersen, 1995; Blossfeld, Hamerle, &
Mayer, 1989; Blossfeld & Rohwer, 2002; Wu, 2003). More ambitiously, the
diffusion of this approach has been linked to a change of paradigm in
demography by Courgeau and Lelievre (1996), and as a solution to the
problem of search for causality in the social sciences (Eerola, 1994; Petersen,
1995; Bocquier, 1996; Blossfeld, 1996; Blossfeld & Rohwer, 2002).
Event history analysis generalizes life-table and standardization tech-
niques that have been extensively used in twentieth-century demography.
The statistical models of event history analysis usually aim to model indi-
vidual-level data collected from sample surveys or population registers.
Event history analysis focuses on the time-to-event as the dependent var-
iable, and since its early applications in demography it has been applied to
several types of events in the study of mortality, migration, family forma-
tion, and dissolution (Hobcraft & Murphy, 1986).
Besides the possibility to describe and compare life courses, the statistical
models of event history analysis have contributed to the explanation of life
course dynamics by linking time-to-event with explanatory variables
(covariates). Covariates can be external to a trajectory (as is the case for
macro-level variables, period effects), or internal to a trajectory (as is the
case of other trajectories of the life course that are potentially influencing the
time-to-event in a trajectory of primary interest). The difference between
external and internal covariates is relevant when one aims at giving a causal
interpretation of the impact of changes of the covariates on the hazard rate
of a given phenomenon.
External covariates can be classified into three main categories. First,
some covariates that are fixed during a life course or starting from a par-
ticular point of the life course, and are time-constant (gender, race, cohort,
age at marriage when studying time-to-divorce). Second, other covariates
have a temporal development that cannot be influenced by the process of
interest (age of an individual in the case of time-to-divorce). Third, some
covariates are located at an aggregate-level (‘‘macro-level’’) or proxy the
social dynamics (period, regional economic indicators, policy indicators). In
that sense, age-period-cohort models can be reconstructed at the individual
level. Multilevel event history models have been developed, taking advan-
tage of the improvement of software packages, at the end of the 1990s. They
allow the explicit modeling of the case of individuals who are aggregated in
household or regional clusters (see e.g. Barber, Murphy, Axinn, & Maples,
2000). As such covariates are ‘‘exogenous’’ – to use an econometric term – a
causal interpretation could be justified. This would in principle allow the
FRANCESCO C. BILLARI266
identification of causal effects in many types of life courses analyses. It is
however questionable that time-constant covariates such as race or gender
could be seen as causal factors in the timing of an event. The same is true
about age – in fact age is usually recognized as a proxy for other causal
factors that are unmeasured (Settersten & Mayer, 1997). Also in the case of
macro-level factors, it is often the case that they can be seen as ‘‘causing’’ the
timing of an event – nevertheless they can be considered triggering factors in
what could be defined a ‘‘situational mechanism’’ (Hedstrom & Swedberg,
1998), as they represent the situation (‘‘location in time and place’’ to use
Giele and Elder definition) in which a life course is unfolding.
Internal covariates usually refer to other trajectories of the same individ-
ual or of linked individuals, and their use allows researchers to study the
very complex interdependencies between trajectories that go to the very core
of the life course approach. Event history analysis may take into account
unobserved factors underlying these complex interdependencies, such as
unobserved value orientations or attitudes. There have been different ap-
proaches in the literature of the 1990s, with respect to the focus on the
relevance of time-constant unobserved factors for the analysis of parallel
and potentially interdependent trajectories (Wu, 2003). The so-called causal
approach (Blossfeld & Rohwer, 2002), assumes that all factors that are rel-
evant to the simultaneous analysis of several trajectories are observed and
included in the past history of the trajectories. This assumption, however, is
not completely explicit in the illustration of the proponents of the ‘‘causal
approach’’ or in papers using that approach.
Other modeling approaches, especially in the econometric literature,
allow for the inclusion in event history models of the impact of time-
constant unobserved factors. The most general applied approach to this
second type of strategy has been developed in the multilevel and multiprocess
modeling of life courses (Lillard, 1993; Lillard & Panis, 2000). These models
constitute in fact a generalization of the ‘‘causal approach’’, which has been
often neglected by scholars looking for the causal impact of changes in one
life course trajectory on other life course trajectories. The ‘‘multilevel and
multiprocess’’ modeling approach proposed by Lillard and Panis aims at the
simultaneous evaluation of the (1) impact of changing covariates (including
those related to other life course trajectories and thus to be considered as
internal covariates) on the hazard of experiencing an event; and (2) possible
common, time-constant, factors that influence a set of hazard rates. The first
aim is the same of the ‘‘causal approach’’, but it is reinforced by the fact that
in multilevel and multiprocess hazard models it is possible to control for the
common factors included in the second aim. In statistical terms, it is always
Life Course Analysis: Two (Complementary) Cultures? 267
possible to test whether the general multilevel and multiprocess hazard
model can be safely restricted to the model following the causal approach –
this is done by testing whether the correlation of the unobserved factors that
influence a set of processes is zero. In other words, the causal approach
hypothesizes that there are no common and unobserved factors that affect
the set of processes under study.
The importance of taking into account the potential role of common
factors affecting life course trajectories can be described by one example.
Le Goff (2002) analyzed the transition to first marriage and first birth
among cohabiting couples in France and West Germany, by using a Lillard-
type model with a system of hazard equations with possibly correlated un-
observed heterogeneity. This could be contrasted for instance to Blossfeld
and Mills (2001), who analyze the same problem without allowing for
potentially correlated unobserved factors affecting life course trajectories,
but who explicitly refer in the title of their paper to the ‘‘causal approach
to interrelated family events’’. We focus on Le Goff’s results for West
Germany, for which the positive correlation between factors affecting entry
into marriage and into parenthood is statistically significant (+0.5455). In
Table 1, we show the results of the multiprocess model (Model (a)) and by
comparison those of the model without unobserved heterogeneity and pos-
sible correlation between unobserved factors (Model (b)). We see that not
taking the correlation into account leads to important biases in the estima-
tion of the impact of getting married on the transition to first birth (the bias
is up to 148% and never lower than 41%). More specifically, the ‘‘causal’’
accelerating impact of marriage is exaggerated if one does not take into
Table 1. Impact of Getting Married on the Log-Hazard of Conception
for Cohabitants: Models Taking (a) and Not Taking (b) into Account the
Correlation between Factors Leading to Marriage and First Birth.
Model (a) Model (b)
At marriage After 1 year At marriage After 1 year
Cohort (1953–1961) 0.2935 0.4505 0.7268 0.9208
Cohort (1962–1974) 0.4836 0.866 0.9726 1.2233
Note: Other covariates are included in the model (see Le Goff, 2002, Appendix). The estimated
correlation between unobserved factors simultaneously affecting marriage and first birth (in
Model (a)) is 0.5455.
Source: Own elaboration on estimates by Le Goff (2002).
FRANCESCO C. BILLARI268
account that the propensity to marry and to have a child are highly inter-
related, and this is due to common factors that are often unobserved.
3.2. The (Causal) Impact of Events: Ideas from
the Program Evaluation Literature
Unraveling causal links controlling for observed and unobserved spurious
factors is the primary focus of another approach, which originates from the
– mostly econometric – literature focusing on program evaluation from ob-
servational data. In program evaluation, the main task is to estimate the
impact of participating in a certain program (usually, a labor market pro-
gram), the treatment, on a specific outcome. This estimate is used as a
support for policy decisions. Two key issues arise. First, to illuminate pol-
icies, one wants to isolate the causal impact of a certain program from other
effects that link the program with the outcome. Second, because for cost–
benefit evaluation the size of the causal impact, and not only its direction or
statistical significance, matters, the emphasis on minimizing estimation bias
is usually stronger than in traditional life course analysis.
The characteristic problem in program evaluation is that participation in
a program is voluntary and can be due to factors that are themselves cor-
related with outcomes. Thus, we face a similar situation when we want to
evaluate the impact on some outcome measure of life course events. Life
course events are subject to choice, and factors leading to choices may be
influencing the outcome as well. With respect to the transition to adulthood,
this scenario applies to several types of substantive problems, and some
studies have already used approaches originating from the evaluation
literature to study the transition to adulthood and its implications. First, we
may be interested in evaluating the causal impact of events (i.e. timing or
sequencing) in the transition to adulthood on the subsequent pathways to
adulthood; a typical question is whether teenage childbearing influence
subsequent educational or labor outcomes during early adulthood (see e.g.
Hotz, Mullin, & Sanders, 1997), or, as in the traditional program evaluation
literature, whether participation in a certain welfare program affects sub-
sequent labor outcomes (see Dehejia & Wahba, 1999). An alternative but
related issue is to study the long-term impact of choices during the transition
to adulthood on later years. Second, we may be interested in evaluating the
causal impact of events involving relevant others, in particular youths’ par-
ents, on the transition to adulthood; a typical question is whether parental
divorce has a causal impact on educational outcomes or family choices in
Life Course Analysis: Two (Complementary) Cultures? 269
the transition to adulthood (see Painter & Levine, 2000). Third, we may
want to study the causal impact of pathways to adulthood on relevant
others; a typical question is whether out-of-wedlock births cause problems
to children; or, whether the leaving home of a child, and in particular the
transition to an ‘‘empty nest’’ has a causal impact on parental outcomes, for
example happiness (Mazzuco, 2003).
Although in some issues (i.e. welfare programs) experimental designs are
also feasible, we shall focus only on observational, non-experimental stud-
ies. The methodological literature on the issue, although relatively recent
(the origins can be traced to Heckman & Robb, 1985), has become vast (for
a non-technical review see, for instance, Bryson, Dorsett, & Purdon, 2002).
The basic ‘‘evaluation problem’’ or the ‘‘fundamental problem of causal
inference’’ (Holland, 1986) is that to truly know the effect of a certain event
(i.e. the participation into a program), we must compare the observed out-
come of an individual who has experienced the event of interest with the
outcome that would have resulted had that person not experienced the
event. Equivalently, we must compare the observed outcome of an individ-
ual who has not experienced the event of interest with the outcome that
would have resulted had that person experienced the event. This ‘‘counter-
factual’’ outcome cannot be observed, and all approaches developed in the
program evaluation literature aim at providing an estimate of the counter-
factual and at using this in order to identify the causal effect of an event.
In addition, the impact of an event can be different across different
groups of a population. For these reasons one needs to distinguish what is
called the average effect of treatment (what impact would the event have on
a randomly drawn individual) and the average effect of treatment on the
treated (what impact has the event had on individuals who have experienced
the event). The average effect of treatment indicates the average benefit were
the event to become compulsory (this is for instance interesting in the case of
welfare programs). The average effect of treatment on the treated indicates
the benefit, or cost of a specific event and is thus the parameter of main
interest in our field. For instance, it is the answer to questions like ‘‘how
different a teen mother’s subsequent Ys would be if she postponed or fore-
went the birth’’ (Hotz et al., 1997, p. 578). Hotz and colleagues explain why
they focus on the average effect of treatment on the treated in their study of
the effects of teenage childbearing by using two reasons. First, a pragmatic
reason: ‘‘this causal effect is more readily identified from available data than
are the causal effects applicable to the full population of women’’ (p. 579).
Second, a substantive, policy-oriented, reason: ‘‘policies that seek to reduce
the rate of teenage childbearing will likely target women who, under the
FRANCESCO C. BILLARI270
status quo, would become teenage mothers. Knowledge of a1 is sufficient to
assess the potential consequences of eliminating teenage childbearing for
these women’’ (p. 579).
We thus want to estimate the effect of treatment by controlling for spu-
rious dependence, possibly including dependence due to unobserved factors.
To estimate treatment effects from observational studies, there are three
approaches in the literature. First, one can use data on highly related in-
dividuals, usually twins, and exploiting the variation between them. This
approach has been used for instance by Kohler, Skytthe and Christensen
(2001) to estimate the impact of age at first birth on completed fertility.
Twin data however are not widely collected and the results are not always
easy to generalize. Second, one can use an instrumental variables (IV) ap-
proach, looking for variables that are good predictors of the event of
interest but are not related to the outcome measure. This approach is hardly
feasible when dealing with life courses, and the most promising avenue uses
IV to estimate bounds of causal effects, as in the case of the analysis of the
effects of teenage childbearing by Hotz et al. (1997). The third approach is
the most promising for life course research, propensity scores matching,
originally proposed by Rosenbaum and Rubin (1983), with the matching of
treated and untreated individuals according to observed covariates sum-
marized in a ‘‘propensity score’’.
We examine more in depth propensity score matching. It is basically a
two-step approach (although steps are relatively complex and some iteration
may be required). In the first ‘‘parametric’’ step, the ‘‘propensity score’’ is
estimated from a set of (possibly abundant) covariates that are supposed to
affect the probability that the event of interest (treatment) is experienced.
These covariates may also influence the outcome of interest (the outcome
also possibly being the probability to experience another event) but the
outcome is not used in the estimation of the propensity score. For instance,
one can use a probit or logit model with the probability of experiencing a
parental divorce as a function of a set of youth and family characteristics
(Painter & Levine, 2000 use this as a robustness check for their results). In
the second ‘‘nonparametric’’ stage, individuals who experienced the event of
interest (treated) are matched to individuals who have not experienced the
event of interest (untreated), according to the propensity scores estimated in
the first step. Matching is ideally similar to what is done with controlled
experiments, but different procedures are possible (for a practical review, see
Becker & Ichino, 2002). For instance, each treated individual could be
matched to the nearest untreated individual (nearest-neighbor matching)
but if treated individuals are fewer that untreated it is also possible to match
Life Course Analysis: Two (Complementary) Cultures? 271
one treated case with more than one untreated case; in addition, untreated
cases may be matched with more that one treated case. It is however nec-
essary that treated and untreated have a large enough ‘‘common support’’ in
terms of comparability since treatment effects can be estimated only within
the common support. Using another approach, Rosenbaum and Rubin
(1984) propose to divide the treated and untreated group in a number of
groups (i.e. with the same score range), and matching occurs when the
groups of treated and untreated have similar covariates. After the matching,
outcomes are compared and tests of statistical significance of the difference
may be computed.
The use of matching on propensity score permits to reduce the bias due to
observed variables influencing both the probability of experiencing the event
of interest, and the outcome (Rosenbaum & Rubin, 1983). To control for
fixed unobserved characteristics influencing both the potential ‘‘cause’’ and
the outcome, it is necessary to use other techniques usually relying on lon-
gitudinal data. For instance, instead of the difference in outcomes between
treated and untreated one can estimate the difference-in-differences between
treated and untreated.
Let us recall the results of an example. Mazzuco (2003) uses propensity
score matching to evaluate the causal impact of the departure of the last co-
resident child on parental well-being, in terms of satisfaction and of self-
rated health. He uses data from the European Community Household Panel
and is thus able to use the difference-in-differences estimator: how much
does the transition into the empty nest stage affect a 1-year change of well-
being and self-rated health? Mazzuco compares a country characterized by
relatively early home-leaving (France) with a country characterized by ex-
ceptionally late home-leaving (Italy). We report his results in Table 2. They
show that the departure of the last child does not have a (causal) impact on
the well-being of fathers in both countries, and that it has opposite impacts
on mothers. French mothers have a higher satisfaction after the departure
of their last child, and their perceived health does not change. On the con-
trary, Italian mothers have a lower satisfaction and their self-rated health
worsens. For the substantive interpretation of results we refer to the paper
by Mazzuco.
The production of easy-to-use packages for estimating treatment effects
based on propensity score matching has and will have a crucial role in the
diffusion of the methodology. Recently, Becker and Ichino (2002) have
produced a set of Stata programs that allow building estimators based on
propensity score matching. This becomes also a good opportunity for
scholars interested in the life course.
FRANCESCO C. BILLARI272
4. THE ALGORITHMIC OR ‘‘HOLISTIC’’ CULTURE
Older notions of life course – for instance the concept of family life cycle –
were holistic, with more or less explicit reference to biological structures
(Settersten & Mayer, 1997). This has no longer been true in the life course
literature since the 1990s. Nevertheless, by focusing mainly on specific
events, with what Elder (1985) has called the ‘‘short-view in analytical scope’’
researchers may not grasp a unitary, holistic, perspective on life courses.
There are two main reasons to complement event-based analysis with a
holistic approach (Billari, 2001a). For the sake of simplicity, they may be
defined as the strong and the pragmatic perspective. From the strong per-
spective, present for instance in neo-classical economics, life courses are seen
subject to accurate inter-temporal planning, for instance as an outcome of
utility maximization. This has also led to study empirically long-term plans
in life courses and their consistency, as well as to critique on the realism of a
dynamic programming view of lives from experimental economics. For in-
stance, Keane and Wolpin (1997) in a paper on the career decisions of young
men estimate empirically a model that involves ‘‘the repeated numerical
solution of a discrete-choice, finite-horizon optimization problem, formu-
lated as a dynamic programming problem’’ (Keane & Wolpin, 1997, p. 476).
As life courses are assumed to be the (uncertain) outcome of planning, a
holistic perspective is thus hypothesized to be present in the behavior of
individuals themselves. In the social-demographic literature, the notion of
strategy has been emphasized. In the psychological literature, individual life
courses are assumed to be marked by internalized timetables. For theoretical
Table 2. The Comparative Impact of the Departure of the Last Child
(Transition into the ‘‘Empty-Nest’’) on (Co-Resident) Parents’
Satisfaction and Well-Being in France and Italy.
France Italy
Mothers Fathers Mothers Fathers
Satisfaction +0.407� 0.066 �0.568� �0.023
Self-Rated Health �0.005 0.059 �0.149� �0.013
Note: Estimates based on propensity score matching using data from the European Community
Household Panel (Mazzuco, 2003).�Statistically significant differences with po0.05. Satisfaction is measured with a sum of sat-
isfaction items ranging from 4 to 24, self-rated health ranges from 1 to 5.
Life Course Analysis: Two (Complementary) Cultures? 273
approaches that assume that the individuals look holistically at their own
lives, it is undoubtedly necessary to have tools that allow to follow the same
perspective, and to analyze the life course as a whole.
From the pragmatic perspective, the life course as a conceptual unit is
thought of as being a contingent result of subsequent events. Following this
viewpoint, researchers should focus principally on events when they wish to
explain individual behavior (as discussed earlier). A holistic view is still
useful as a way to describe and to summarize the timing, sequencing, and
quantum of life course events. Comparative research across countries or
regions or across cohorts is one of the examples where the description of life
courses as whole conceptual units might provide particular insights.
Since the 1990s, Andrew Abbott introduced the social science to sequence
analysis (see Abbott & Tsay, 2000). In sequence analysis, each life course or
trajectory in the life course is represented as a string of characters (also
numerical). This representation is identical to the one used to code DNA
molecules in the biological sciences. Different analytical strategies to analyze
data arranged as sequences have been proposed. Optimal matching analysis
(OMA) is a method originally created for the alignment of DNA sequences.
The goal of OMA is to compute a matrix of dissimilarities between pairs of
sequences (thus, of life courses). This matrix can be used as the input for any
kind of statistical analysis requiring proximity data (e.g. cluster analysis or
multidimensional scaling). Applications in the sociology of occupations
have appeared since the 1990s, and a review of the potential applications in
the field of demography is provided in Billari (2001a), while a demographic
application on the transition to adulthood is discussed in Billari (2001b).
OMA has also specific drawbacks (Wu, 2000). The definition of distance
between states has an important influence on results, and thus it has to be
based on theoretical grounds. In addition, it can be difficult to understand
which variables in the definition of the obtained groups are more relevant
than others. Alternative approaches to the analysis of life courses as a con-
ceptual unit have also been discussed, although they have not yet had a large
impact on the literature. These include the use of correspondence analysis
(van der Heijden, 1987, Chapter 8) and data mining techniques (Billari,
Prskawetz, & Furnkranz, 2000).
As an example of the use of the algorithmic culture we discuss the analysis
of sequencing in the transition to adulthood among U.S. youth presented by
Mouw (2005). Mouw applies a monothetic divisive algorithm (MDA) to
create clusters of life courses that are internally as homogeneous as possible
and externally as heterogeneous as possible. The study follows the approach
suggested by Billari and Piccarreta (2001; 2005) in an analysis of the
FRANCESCO C. BILLARI274
transition to adulthood in Italy. As usual for sequence analysis, the coding
of variables is a key step. Mouw defines a sequence of dummy variables
called P (1 if an individual has left the parental home, 0 otherwise), E (1 if
an individual has left education, 0 otherwise), L (1 if an individual has
started working, 0 otherwise), M (1 if an individual has married, 0 other-
wise), B (1 if an individual has already become a parent, 0 otherwise). An
example for the coding of an individual life course is reported in Table 3.
Sequences are then processed by the MDA by choosing the splitting variable
that maximizes within-group homogeneity and between-group heterogene-
ity, according to a Gini heterogeneity index. It also allows the use of an R2
type of statistic to choose the optimal number of splits. As common in
algorithmic approaches, a tree representation is useful. We reproduce
(Fig. 1) the tree for U.S. women, which indicates that the most discrim-
inating circumstance is whether they had become mother by age 28 (to the
right of the tree, that is about two thirds of the sample). Six groups of
women’s life courses are found. Group 1 is characterized by delayed tran-
sition to motherhood and marriage. Group 2 is characterized by delayed
transition to motherhood, but marriage taking place at least by age 32.
Group 3 is characterized by having had birth by age 28, without having been
married by that age and having not begun work by age 33. Group 4 differs
from Group 3 by the experience of work by age 33. Group 5 is characterized
by a birth between age 24 and 28, and marriage by age 29. Group 6 is
characterized by earlier fertility (transition to motherhood by age 24) and a
transition to marriage by age 29.
5. CONCLUDING REMARKS
From the review we have made of the existing literature we can safely
conclude that no single approach is the ‘‘solution’’ to life course analysis. On
the one side, there are different approaches that are useful in specific cir-
cumstances within what we have defined the event-based, or ‘‘causality’’
culture – we have mentioned event history analysis and approaches based on
program evaluation, although in some cases the difference between the two
approaches is negligible or non-existent. On the other, by definition the
‘‘algorithmic’’ or holistic culture sees a multiplicity of ‘‘good’’ models as a
reasonable outcome. The two cultures prove to be more fruitful vis-a-vis
each other when answering to different questions? To put it simply, on one
hand, when we would like to ask ‘‘what would happen to the risk of ex-
periencing a certain event ify?’’, we shall probably use models rooted in the
Life Course Analysis: Two (Complementary) Cultures? 275
Table 3. Coding of the Life Course of an Individual who has Left Home
at Age 23, Finished School at 25, Entered the Labor Force at 26, Got
Married at 30 and had His or Her First Child at Age 32.
22 23 24 25 26 27 28 29 30 31 32 33 34 35
P 0 1 1 1 1 1 1 1 1 1 1 1 1 1
E 0 0 0 1 1 1 1 1 1 1 1 1 1 1
L 0 0 0 0 1 1 1 1 1 1 1 1 1 1
M 0 0 0 0 0 0 0 0 1 1 1 1 1 1
B 0 0 0 0 0 0 0 0 0 0 1 1 1 1
Source: Mouw (2005).
All women
N=2,939
B28=1
N=1,968
B28=0N=971
M32=0N=465
Group 1 Group 2
M32=1N=506
L33=0 L33=1N=298
N=721 N=1,247
N=423 N=352B24=0
M29=0 M29=1
B24=1N=895
Group 3 Group 4 Group 5 Group 6
Fig. 1. Tree Representation of the Results of Applying a Monothetic Divisive
Algorithm to Data on the Transition of Adulthood of U.S. Women. B24 ¼ 1 and
B28 ¼ 1 indicate having become a mother respectively by age 24 and 28 (B24 ¼ 0
and B28 ¼ 0 indicate the opposite), M29 ¼ 1 and M32 ¼ 1 indicate having married
respectively by age 29 and 32 (M29 ¼ 0 and M32 ¼ 0 indicate the opposite),
L33 ¼ 1 indicates having worked by age 33 (L33 ¼ 0 indicates the opposite. Source:
Mouw (2005).
FRANCESCO C. BILLARI276
‘‘event-based’’ culture. On the other, when we would like to ask ‘‘what are
the key factors that differentiate life courses in their unity, and what are the
consequences of a broadly defined trajectory?’’, we shall probably use mod-
els rooted in the algorithmic culture.
However, also within the same study the two approaches can fruitfully
complement each other. For instance, Mouw (2005) uses the output of the
clustering procedure we illustrated in Section 5 as an input for a regression
analysis that is much more oriented towards causality, under the heading
‘‘Does the sequence matter?’’ Regression analyses show important differ-
ence in the risk of experiencing outcomes such as poverty at age 35, with
women in Group 3 with the worst outcome in a logit regression model (see
Table 4). Sequences are also found to influence subsequent happiness and
depression (women in Group 6 are the happiest and least likely to be de-
pressed). The same spirit of complementarity between the two cultures in-
spires the work of McVicar and Anyadike-Danes (2001) who use clusters
obtained starting from an optimal matching analysis of sequences to predict
successful and unsuccessful transitions from school to work.
It is necessary to mention some caveats in these concluding remarks. In
the ‘‘causality’’ culture we should refrain from the idea of finding the ‘‘true’’
impact of events in the life course, as is sometimes done in the econometric
literature – the impact of an event, if we go back to the conceptual frame-
work on the life course by Giele and Elder, depends on the role of relevant
others (‘‘linked lives’’) as well as on other contextual factors (‘‘location in
time and place’’). The search for causality in the life course is thus a never
ending story, and the most promising new analyses based on either
Table 4. The Effect of Sequences on Life Course Outcomes: Coefficients
Related to Group Membership as Reported in Fig. 1 in a Logit Model for
Poverty Status at Age 35 (other coefficients not reported here).
Coefficient Standard error
Group 1 (B28 ¼ 0; M32 ¼ 0) 0.810 0.549
Group 2 (B28 ¼ 0; M32 ¼ 1) �0.297 0.547
Group 3 (B28 ¼ 1; M29 ¼ 0; L33 ¼ 0) 1.352�� 0.255
Group 4 (B28 ¼ 1; M29 ¼ 0; L33 ¼ 1): reference
Group 5 (B28 ¼ 1; M29 ¼ 1; B24 ¼ 0) �0.411 0.432
Group 6 (B28 ¼ 1; M29 ¼ 1; B24 ¼ 1) �0.074 0.239
Source: Mouw (2005).��Statistically significant differences with po0.01.
Life Course Analysis: Two (Complementary) Cultures? 277
simultaneous hazard models or on other types of program evaluation take
what we may define a multiplicity of causal impacts for a multiplicity of
contexts as a strong point in the analysis. The caveat for the algorithmic
culture is possibly a similar one – not always what has been developed in
other fields can be directly translated to the analysis of life courses, and a
fecund interaction is fundamental in order to obtain optimal results.
No approach can work without appropriate software, and here it is worth
mentioning that some scholars have sometime worked hard to help other
scholars implementing new approaches. Transition Data Analysis (TDA)
(Rohwer & Potter, 2003), applied Maximum Likelihood (aML) (Lillard &
Panis, 2000) have been explicitly developed to study life courses using haz-
ard models, with aML allowing for the inclusion of correlated unobserved
heterogeneity between processes. Commercial packages have and will be
used, with Stata playing a prominent role in the program evaluation ap-
proach. For what concerns the algorithmic approach, TDA is also providing
a toolkit for sequence analysis, including optimal matching analysis, while a
set of non-standard programs written by researchers for the moment is used
for other applications.
To conclude, data are crucial ingredients for our analyses. Breiman de-
fined what we have put in parallel to the event-based or causality culture as
‘‘data modeling culture’’. The sort of data that have been collected since the
1970s in demography included detailed retrospective reconstruction of life
course trajectories. For a while this seemed to be the final answer to the
search for causality (see, for instance, the claims of Blossfeld & Rohwer,
2002). Nevertheless, by emphasizing the importance of unobserved heter-
ogeneity and of the bias introduced when not taking it into account we
acknowledge that more work has to be done. Prospective panel studies,
aiming at observing what is unobservable in a retrospective study (e.g.
intentions, value orientations and social capital) will provide to the causal
approach a much better equipment. On the other hand, the algorithmic
approach is to be exploited when data that are not directly collected for
research purposes – such as in the case of administrative registers – are to be
analyzed. Much more work is still to be done, in different directions, to
progress in life course analysis.
ACKNOWLEDGMENTS
The author gratefully acknowledges for comments and suggestions the ed-
itors of this volume, the members of the Working Group on Transitions to
FRANCESCO C. BILLARI278
Adulthood in Developed Societies of the International Union for the Sci-
entific Study of Population (IUSSP), and Stefano Mazzuco. A presentation
based on these ideas was given at a lecture organized at the Department of
Sociology of the University of Calgary by Prof. Anne H. Gauthier. Finan-
cial support is acknowledged to the Department of Sociology of the Uni-
versity of Calgary, Universita Bocconi and MIUR.
NOTE
1. This is the symbol for the average effect of treatment on the treated.
REFERENCES
Abbott, A., & Tsay, A. (2000). Sequence analysis and optimal matching methods in sociology.
Sociological Methods & Research, 29, 3–33.
Allison, P. D. (1984). Event history analysis. Regression for longitudinal event data. Newbury
Park, CA: Sage.
Barber, J. S., Murphy, S. A., Axinn, W. G., &Maples, J. (2000). Discrete-time multilevel hazard
analysis. Sociological Methodology, 30, 201–235.
Becker, S. O., & Ichino, A. (2002). Estimation of average treatment effects based on propensity
scores. The Stata Journal, 7, 1–19.
Billari, F. C. (2001a). Sequence analysis in demographic research and applications. Canadian
Studies in Population, 28, 439–458.
Billari, F. C. (2001b). The analysis of early life courses: Complex descriptions of the transition
to adulthood. Journal of Population Research, 18, 119–142.
Billari, F. C. (2003). Life course analysis. In: P. Demeny & G. McNicoll (Eds), Encyclopedia of
population (Revised edition, pp. 588–590). New York: Macmillan.
Billari, F. C., & Piccarreta, R. (2001). Life courses as sequences: An experiment in classification
via monothetic divisive algorithms. In: S. Borra, R. Rocchi, M. Vichi & M. Schader
(Eds), Advances in classification and data analysis (pp. 351–358). Berlin: Springer.
Billari, F. C., & Piccarreta, R. (2005). Analysing demographic life courses through sequence
analysis. Mathematical Population Studies, 12, 81–106.
Billari, F. C., Prskawetz, A., & Furnkranz, J. (2000). Timing, sequencing and quantum of life
course events: A machine learning approach, Working Paper WP-2000-010, Rostock:
Max Planck Institute for Demographic Research.
Blossfeld, H.-P. (1996). Macro-sociology, rational choice theory, and time. A theoretical per-
spective on the empirical analysis of social processes. European Sociological Review, 12,
181–206.
Blossfeld, H.-P., & Mills, M. (2001). A causal approach to interrelated family events: A cross-
national comparison of cohabitation, nonmarital conception, and marriage. Canadian
Studies in Population, 28, 409–437.
Blossfeld, H.-P., & Rohwer, G. (2002). Techniques of event history modeling (2nd ed.). Mahwah,
NJ: Lawrence Erlbaum.
Life Course Analysis: Two (Complementary) Cultures? 279
Blossfeld, H.-P., Hamerle, A., & Mayer, K. U. (1989). Event history analysis. Statistical theory
and application in the social sciences. Hillsdale, NJ: Lawrence Erlbaum Associates.
Bocquier, P. (1996). L’analyse des enquetes biographiques a l’aide du logiciel STATA. Paris:
CEPED.
Breiman, L. (2001). Statistical modeling: The two cultures. (with comments and rejoinder).
Statistical Science, 16, 199–231.
Bryson, A., Dorsett, R., & Purdon, S. (2002). The use of propensity score matching in the
evaluation of active labour market policies. Working Paper no. 4, Department of Work
and Pensions, London.
Courgeau, D., & Lelievre, E. (1992). Event history analysis in demography. Oxford: Clarendon
Press.
Courgeau, D., & Lelievre, E. (1996). Changement de paradigme en demographie. Population,
51, 645–653.
Cox, D. R. (1972). Regression models and life-tables. Journal of the Royal Statistical Society,
34, 187–220.
Dehejia, R. H., & Wahba, S. (1999). Causal effects in nonexperimental studies: Reevaluation of
the evaluation of training programs. Journal of the American Statistical Association, 94,
1053–1062.
Dykstra, P. A., & van Wissen, L. J. G. (1999). Introduction: The life course approach as an
interdisciplinary framework for population studies. In: L. J. G. van Wissen &
P. A. Dykstra (Eds), Population issues. An interdisciplinary focus (pp. 1–22). New York,
NY: Kluwer Academic/Plenum Publishers.
Eerola, M. (1994). Probabilistic causality in longitudinal studies. New York: Springer.
Elder, G. H. (1985). Life course dynamics trajectories and transitions, 1968–1980. Ithaca, NY:
Cornell University Press.
Giele, J. Z., & Elder, G. H., Jr. (Eds) (1998). Methods of life course research. Qualitative and
quantitative approaches. Thousand Oaks, CA: Sage.
Heckman, J., & Robb, R. (1985). Alternative methods for evaluating the impact of interven-
tions. In: J. Heckman & B. Singer (Eds), Longitudinal analysis of labor market data.
New York: Cambridge University Press.
Hedstrom, P., & Swedberg, R. (Eds) (1998). Social mechanisms. Cambridge: Cambridge
University Press.
Hobcraft, J., & Murphy, M. (1986). Demographic event history analysis: A selective review.
Population Index, 52, 3–27.
Holland, P. W. (1986). Statistics and causal inference. Journal of the American Statistical
Association, 81, 945–970.
Hotz, V. J., Mullin, C. H., & Sanders, S. G. (1997). Bounding causal effects using data from a
contaminated natural experiment: Analysing the effect of teenage childbearing. Review
of Economic Studies, 64, 575–603.
Keane, M. R., & Wolpin, K. I. (1997). The career decisions of young men. Journal of Political
Economy, 105, 473–522.
Kohler, H.-P., Skytthe, A., & Christensen, K. (2001). The age at first birth and completed
fertility reconsidered: Findings from a sample of identical twins. Working Paper WP
2001-006, Max Planck Institute for Demographic Research, Rostock.
Le Goff, J.-M. (2002). Cohabiting unions in France and West Germany: Transitions to first
birth and first marriage. Demographic Research, 7, 591–624.
FRANCESCO C. BILLARI280
Lillard, L. A. (1993). Simultaneous equations for hazards: Marriage duration and fertility
timing. Journal of Econometrics, 56, 189–217.
Lillard, L. A., & Panis, C. W. A. (2000). AML multilevel multiprocess statistical software, release
1.0. Los Angeles, CA: EconWare.
Mayer, K. U., & Tuma, N. B. (Eds) (1990). Event history analysis in life course research.
Madison, WI: University of Wisconsin Press.
Mazzuco, S. (2003). When a child leaves the nest. A comparative analysis of effects of children
departures from home on parents’ wellbeing. Mimeo, University of Padova.
McVicar, D., & Anyadike-Danes, M. (2001). Predicting successful and unsuccessful transitions
from school to work by using sequence methods. Journal of the Royal Statistical
Association, Series A, 165, 317–334.
Mouw, T. (2005). Sequences of early adult transitions: How variable are they, and does it
matter? In: R. A. Settersten Jr., F. F. Furstenberg Jr. & R. G. Rumbaut (Eds), On the
frontier of adulthood: Theory, research and public policy. Chicago: University of Chicago
Press.
Painter, G., & Levine, D. I. (2000). Family structure and youths’ outcomes. Which correlations
are causal? The Journal of Human Resources, 35, 524–549.
Petersen, T. (1995). Analysis of event histories. In: G. Arminger, C. C. Clogg & M. E. Sobel
(Eds), Handbook of statistical modeling for the social and behavioral sciences
(pp. 453–517). New York: Plenum Press.
Rindfuss, R. R. (1991). The young adult years: Diversity, structural change, and fertility.
Demography, 28, 493–512.
Rohwer, G., & Potter, U. (2003). TDA user’s manual. Bochum: Ruhr University Bochum.
Rosenbaum, P. R., & Rubin, D. B. (1983). The central role of the propensity score in ob-
servational studies for causal effects. Biometrika, 70, 41–55.
Rosenbaum, P. R., & Rubin, D. B. (1984). Reducing bias in observational studies using sub-
classification on the propensity score. Journal of the American Statistical Association, 79,
516–524.
Settersten, R. A., & Mayer, K. U. (1997). The measurement of age, age structuring and the life
course. Annual Review of Sociology, 23, 233–261.
van der Heijden, P. G. M. (1987). Correspondence analysis of longitudinal categorical data.
Leiden: DSWO Press.
Wu, L. L. (2000). Some comments on ‘‘sequence analysis and optimal matching methods in
sociology: Review and prospect’’. Sociological Methods & Research, 29, 41–64.
Wu, L. L. (2003). Event history models for life course analysis. In: J. Mortimer & M. Shanahan
(Eds), Handbook of the life course (pp. 477–502). New York: Plenum Press.
Yamaguchi, K. (1991). Event history analysis. Newbury Park, CA: Sage.
Life Course Analysis: Two (Complementary) Cultures? 281
LIFE COURSE DATA IN
DEMOGRAPHY AND SOCIAL
SCIENCES: STATISTICAL AND
DATA-MINING APPROACHES$
Gilbert Ritschard and Michel Oris
1. FROM DEMOGRAPHIC ANALYSIS TO
LIFE COURSE APPROACH
This chapter has essentially a methodological purpose. It discusses recent
advances in statistical event history analysis and Markov models and pro-
motes the use of tools from the developing field of data mining, with special
attention to the discovering of characteristic sequences and induction trees.
Before turning to these methodological aspects, we begin here by explaining
why demographers have been relatively reluctant to implement the life
course paradigm and methods, while the quantitative focus and the concepts
of demographic analysis a priori favored such implementation. A real in-
tellectual crisis has been needed before demographers integrated the neces-
sity to face up the challenge of shifting ‘‘from structure to process, from
macro to micro, from analysis to synthesis, from certainty to uncertainty’’
$The funding for this research is from Swiss National Science Foundation, projects 1114-68113
and 100012-105478.
Towards an Interdisciplinary Perspective on the Life Course
Advances in Life Course Research, Volume 10, 283–314
Copyright r 2005 by Elsevier Ltd.
All rights of reproduction in any form reserved
ISSN: 1040-2608/doi:10.1016/S1040-2608(05)10011-2
283
(Willekens, 1999, p. 26). This retrospective look also shows impressive
progresses to promote a real interdisciplinarity in population studies, knot-
ting the ties between demography and the social sciences.
Although demographic analysis has a long history (see Dupaquier &
Dupaquier, 1985), the methods still used today have essentially been elab-
orated between the mid-nineteenth and the mid-twentieth centuries in West-
ern societies that felt successively threatened by race degeneration, declining
birth rates, and ageing. The macro frame was that of the demographic
transition, i.e. the evolution from young populations with high fertility and
high (especially infant and child) mortality to ageing populations with net
reproduction below the threshold of generations’ renewal and a tremendous
increase in life expectancy at birth.
From a methodological point of view, a starting point in demographic
analysis has been the mortality table. It implies a dynamic perception of
population with entrances (births or in-migrations) and exits (deaths or out-
migrations), and the idea that other events (like migration) can censor a given
risk (like mortality). It rapidly constrained the conceptual distinction between
a generation – i.e. those who are born in the same year or period – and a
cohort – i.e. those who are experiencing an event (marriage for example) in
the same year or period. The mortality table also resulted not only in an
average age at death but also in a distribution of the risk along the life course,
providing a survival curve. Finally, the immediate comparison of these curves
among sexes, and even more among matrimonial statuses, revealed selection
processes, like those supporting the over-mortality of singles compared to
married people. Heterogeneity and differential frailty were not ignored. After
the generalization of mortality analysis (from mortality- to life tables), cer-
tainly no scientific discipline was better prepared for the life course methods
than demography. Nevertheless, the paradigms clearly diverged.
First, dealing with structures and flows, demography has been a science of
reconstruction and description of patterns and behaviors, through a well-
established quantitative methodology, and the conviction that higher the
number of observations, more accurate – and possibly useful – were the
results (for a typical example, see Vallin, 2001). Demography was a science of
the masses, growing or stagnating, young or old, but not of the individuals!
Second, the engagement of generations of scholars was largely motivated
by the central character of population issues and the location of demo-
graphy at a crossroad between economy, sociology, epidemiological studies,
territorial analysis, political sciences, and, more recently, cultural and
gender approaches. However, research and collaborations were in reality
highly segmented, with a clear tendency to specialization on a geographical
GILBERT RITSCHARD AND MICHEL ORIS284
and/or thematic basis (typically mortality, fertility, marriage and family
formation, or dissolution, migrations, structures, prospective). Among
many others, the last edition of the excellent Encyclopedia of Population
(Demeny & McNicoll, 2003) illustrates that propensity. Such segmentation
was clearly inscribed in the methods. In the estimations of mortality,
mobility was statistically treated as a censor but explicitly presented as a bias
for a ‘‘pure’’ analysis of mortality. Clearly, the approach consisted in stud-
ying a demographic behavior as independently as possible from the other
ones, without systemic perspectives.
Third, the demographic evolution made apparent the limits of the es-
tablished methodology. Both mortality and life tables can be calculated
longitudinally based on the observation of generations, or on cross-
sectional data, i.e. the observation of deaths by age classes at a given time
point – mixing thus generations with different history together. For the
simplest table, that of mortality, since expectation of life at birth now ex-
ceeds 80, the longitudinal approach implies a population reconstruction
from at least the 1920s, what is quite difficult, especially if we do not accept
the hypothesis of a null effect of migration. Inversely, all the statistical
offices of the developed countries collect the data for the calculation of
cross-sectional measures for a long time. However, can we accept the under-
lying hypothesis of continuity while the duration of life wins 1 year every
3–4 years and while we know that the – generational – age distribution of
those gains has drastically changed during the last decades? Similarly, while
in the context of the so-called ‘‘second demographic transition’’ there are so
many changes in the fertility calendar, do we have to constrain ourselves to
the observation of those generations who have finished their fertile life and
renounce to study the present with other indicators than those of the mo-
ment? What is today the rationale of detailing the access to marriage while
cohabitation is rising? How record an informal event like entrance in co-
habitation? Both data collection and analytical tools have been challenged
by recent changes in demographic behaviors and family dynamics (for a
more in-depth discussion of those issues, see Caselli, Vallin, & Wunsch,
2001).
A real intellectual crisis resulted from the hesitation about the status of
demography within the social sciences, as well as from the frustration
against segmentation and the deficiency of old methods. The conscience that
description, especially some quantification with a pretension of objectivity,
hid and diffused ideological visions about ‘‘good’’ or ‘‘optimal’’ populations
also grew (Greenhalgh, 1996; van de Kaa, 1996; Veron, 1993; Szreter, 1993;
Hogdson, 1988, 1991).
Life Course Data in Demography and Social Sciences 285
Among the many reactions, revisions, and re-examinations, new
approaches and new methods rapidly emerged. No significant use of the
life course statistical tools can be observed before the mid-1980s, while for
example Cox’s foundling paper is dated 1972. When they finally have been
integrated by demographers, the new methods found many uses. Probably
the most obvious progress they supported was to replace demography in its
family setting. Something that could seem very strange, but perfectly illus-
trates this assertion, is indeed the discovery, precisely in the 1980s, of an
almost complete absence of dialogue between demography and family
sociology. While family is the place where most of the demographic be-
haviors take place and, to some extent, are decided, ‘‘few textbooks on
population contain a chapter devoted to the demography of the family.
Where such chapter does exist, it is generally shorter and more superficial
than those that deal with fertility, mortality, nuptiality, and migration, or
with the dynamics of age structure’’ (Hohn, 1992, p. 3). In 1982, the In-
ternational Union for the Scientific Study of Population created an ad hoc
committee to develop its study, but even in 1992 the animators of this group
saw family demography as ‘‘a recent and relatively underdeveloped branch
of population studies’’ (Berquo & Xenos, 1992, p. 8).
Its development has been extraordinary in the last years. Francesco Billari
chapter in this volume provides a nice illustration of such a change, which is
part of a shift from macro to micro, from an emphasis on macro-economic
evolutions as the essential determinants of demographic ‘‘answers’’, to a
multi-causal – multivariate – approach of behaviors, a shift also from av-
erage results to a more detailed study of distributions. In a quantitative
discipline, major evolutions necessarily imply to take up technical challeng-
es. ‘‘The traditional demographic analysis of such events as births, mar-
riages, divorces, deaths, and migration has the advantage that number of
these events can be related to individuals in the same age group and can,
therefore, be measured more easily and included in models. The inclusion of
other family members in such analyses causes difficulties because they will
generally differ in age and sex, and complications are also introduced be-
cause they do not generally live together continuously’’ (Hohn, 1992, p. 3).
Although several attempts have been made to construct a ‘‘household
demography’’ (Van Imhoff, Kuijsten, Hooimeijer, & van Wissen, 1995), the
life course paradigm and its methodological individualism clearly imposed
themselves. Offering both concepts and statistical methods, it represents a
shift toward microanalysis of individual data and causal research that not
only deeply renews the discipline, but also provides the vocabulary for a new
interdisciplinarity, first within the social sciences, then beyond (Blossfeld &
GILBERT RITSCHARD AND MICHEL ORIS286
Rohwer, 2002; Dykstra & van Wissen, 1999). The first substantial gain has
been the study of multiple events, marriage and first birth, or moving and
starting a new job for instance, a kind of investigation that also raises the
issue of event sequencing and interactions that is typically treated with event
history analysis. If people have several careers that they must make com-
patible, their life transitions also reflect socioeconomic constraints, cultural
norms (about the ‘‘proper’’ age, sex, or behavior), as well as compromises
between several individual aspirations within or beyond the domestic unit.
Through researches in this huge area, family demography made for sure
tremendous progress during the last 20 years.
However, the shift has been so sudden that globally the complexity of
causalities remains too often underestimated (see Courgeau & Lelievre,
1993; Blossfeld & Rohwer, 2002; Bocquier, 1996; Alter, 1998; Billari, 2005),
as well as several technical traps. The problem is essentially that when
studying a population of individuals observed along the time, since each life,
the product of complex and multiple interactions is, as a matter of fact,
unique. Hence, interpreting and generalizing from samples require several
cautions. In the next section, we recall the main event history regression
models and discuss the question of heterogeneity. We cannot consider that
the elaboration of indicators at an individual level about household, family,
and community contexts is enough to deal with the more and more raised
issue of ‘‘linked’’ or ‘‘interdependent’’ lives (Hagestad, 2003). We show the
interest of robust estimates and shared frailty in that perspective. In the
same section, we also present the Markovian models that are particularly
useful for the study of transitions within a set of states (matrimonial or
social status, for example) periodically observed. In the interdisciplinary
perspective, which is one of the life courses, we consider it important to go
beyond the simple transitions typically studied in demography (from single
to married, from the first to a possible second child, from life to death, and
so on), and to investigate how, from a starting position, a destination is
selected among several possible. While family dynamics and life courses are
more and more open, such investigations are essential to deal with the
characterization of transitions as ‘‘normal’’ or ‘‘non-normal’’ without falling
again in the trap of ideological reading (see, for example, the discussion in
Oris & Poulain (2003) about the stigmatization of early home leaving).
Indeed, we assess more globally that there is a deficit of research on
trajectories between aggregate descriptions and causal analysis. Regression
models attempt to quantify how a factor, measured by an indicator, affects a
risk. However, such results tell us nothing about the calendar and no more
about the alternatives to this risk in life courses. It is essential to look
Life Course Data in Demography and Social Sciences 287
carefully at transitions in trajectories to target properly a causal analysis,
and this step is clearly too often superficial, if not absent. Several methods,
recently developed or recently made available in statistical packages, offer
opportunities to fill this gap. Among them, we promote in Section 3 some
highly flexible heuristic tools from the developing field of data mining, es-
pecially mining event-sequential association rules, and induction trees that
seem to us the more promising for life course data analysis.
2. STATISTICAL MODELING OF LIFE EVENTS
Life course data are longitudinal in their essence. Here, we focus on events,
an event being the change of state of some discrete variable, e.g. the marital
status, the number of children, the job, or the place of residence. Such data
are collected mainly in two ways: as a collection of time-stamped events or
as state sequences. In the former case, each individual is described by a
collection of time-stamped events, i.e. the realization of each event of in-
terest (e.g. being married, birth of a child, end of job, moving) is mentioned
together with the time at which it occurred. In the latter case, the life events
of each individual are represented by the sequence of states of the variables
of interest. Panel data are special cases of state sequences where the states
are observed at periodic time. The first kind of data is typically analyzed
with event history regression methods, while methods for state-sequence
analysis like Markov transition models are best suited for the latter. We
briefly discuss hereafter the scope and limits of these approaches.
2.1. Event History Regression Models
When we have time-stamped events, the question of interest is the duration
of the spell between two successive events, or somewhat equivalently, the
hazard rate h(t) for the next event to occur precisely after a duration t, i.e.
the conditional probability for the event to occur at t knowing that it did not
occur before t. Longitudinal-regression models focus on this aspect. They
express either the duration or the hazard rate as a function of covariates. It
is worth mentioning that these models are also known as survival models,
especially in area like biomedicine and engineering where the event of con-
cern is just death or breakdown.
There are continuous-time models and discrete-time forms. With contin-
uous time, the main formulations (see Blossfeld, Hamerle, & Mayer, 1989;
GILBERT RITSCHARD AND MICHEL ORIS288
Courgeau & Lelievre, 1993) are as a duration model or as a proportional-
hazards model. Duration models consider ln(T), the logarithm of the time to
the event, as a linear function of the explanatory factors. Proportional-
hazards models suppose that the ratio between the hazard for a given profile
(in terms of the covariates) and that for a reference baseline profile remains
constant over time and expresses the logarithm of this ratio (or proportion)
as a linear function of the covariates.
Duration models, also known as accelerated failure time models, assume
usually an exponential, Weibull, log-normal, log-logistic, or gamma distri-
bution for the duration T. The proportional-hazards model is compatible
with for instance, exponential, Weibull, and Gompertz duration distribu-
tions. It includes also the perhaps most widely used Cox (1972) semi-
parametric model that requires no assumptions on the form of the duration
distribution. Most statistical packages (SAS, S-Plus, Stata, R, TDA, etc.)
provide procedures for estimating such models. At least until version 13,
SPSS, however, offers only support for the Cox model.
Discrete-time models (see Allison, 1982; Yamaguchi, 1991) include the
proportional hazard-odds model, also owe to Cox (1972), the discrete pro-
portional-hazards model (Aranza–Ordaz, 1983), and the log-rate model
(Holford, 1980). In the first model, it is not just the hazard ratio, but the
ratio of the odds of the hazards that is supposed to be constant and having a
logarithm depending linearly on the covariates. The discrete proportional
hazards model expresses the log minus log of the complementary hazard as a
linear function of the covariates. The log-rate model on its side expresses the
log-hazard in terms of proper and interaction effects of categorical variables
and also possibly of their interactions with duration.
For the estimation of the proportional hazard-odds model, some as-
sumptions are usually required upon the baseline hazard-odds. Letting b0tbe the baseline log hazard-odds after a duration t, the most common as-
sumptions are that it remains constant with t, is linear in t (Gompertz), or is
linear in ln(t) (Weibull). With these assumptions, a proportional hazard-
odds model can, if we organize the data in a person–period form, simply be
estimated as a logistic regression. Hence, it can be estimated by any software
that proposes logistic regression. Likewise, a log-rate model can be esti-
mated with any log-linear model procedure that allows for weighted cell
frequencies. Indeed, the log-rate model is a log-linear model of the weighted
number of events occurring in a time interval, the weight being the inverse of
the population at risk in this interval. The fitting of a discrete proportional
hazards model requires the perhaps less frequently implemented procedures
for binary regression with a complementary log-log link.
Life Course Data in Demography and Social Sciences 289
A common issue with the time to event models is the handling of censored
data. Censored data occur when the observed start (left) and/or end (right)
time of a spell are not its actual start and end time. For instance, if we
observe job duration, some jobs may not be terminated at the time of the
survey and are hence right-censored. Though no event is recorded at the
end of the right-censored spells, these cases are taken into account by
entering the population at risk for job length lower or equal to the observed
duration.
Another issue is the handling of time-varying covariates. The solution is
quite straightforward in the discrete-time setting that works on person–time
data. For the continuous case, there are two major solutions: an ad hoc
extension of the Cox model that allows for discrete-time-varying covariate
and the episode-splitting approach (see Blossfeld & Rohwer, 2002 for de-
tails). Time-varying covariates offer a way to test and relax the somehow
strong proportionality assumption required by most hazard-rate models.
Indeed, this assumption implies the time independency of the ratio of haz-
ards of any two individuals, which clearly does not hold when the ratio
depends on a time-varying variable. It is common practice to check the
significance of the interaction of a supposed time-independent variable with
t or ln(t). A significant interaction would provide evidence against time
invariance (see Therneau & Grambsch, 2000 for other tests of proportional
hazards and more advanced developments of the Cox model).
This event-history modeling, especially the Cox proportional-hazards and
discrete-time proportional hazard-odds models, has become popular among
demographers. Together with other social science scientists, historical
demographers have to face issues like competing events (multiple destina-
tions), repeatable events, and interacting events. The first two can easily be
handled with a software like TDA (Rohwer & Potter, 2002) that supports
episodes defined by four parameters, namely the origin state, the start time,
the destination state, and the end time. The interaction between events,
marriage end, and first child, for instance, needs a simultaneous equation
approach that has been investigated by Lillard (1993), and is discussed more
in depth in Billari’s contribution to this volume.
2.1.1. Shared Heterogeneity and Multi-Level Modeling
A further issue of importance, shared heterogeneity, is concerned with the
sampling nature of the data. These are often clustered, i.e. the individual
data come from a selection of groups, parishes, or families for example. In
such cases, members of a same group share a same contextual framework
and it is then of primary importance to distinguish effects that hold at the
GILBERT RITSCHARD AND MICHEL ORIS290
group level from those that work at the individual level. A very concrete
example is the study of orphans’ survival after father’s death by Beekink,
van Poppel, and Liefbroer (1999) for a 19th-century Dutch provincial town.
In the event-history file, initially each orphan was considered as a single
individual while there were not individuals but groups of siblings that en-
tered in the population at risk because of a shared event – dad’s death – and
supported this experience while sharing the same household context. Taking
into account the interrelatedness of the observations changed the results!
Along the same line, both in contemporary and historical demography, the
issue of the death clustering at the family level is a growing concern (Alter,
Oris, & Brostrom, 2001). All those studies extend the original discussion of
‘‘the impact of heterogeneity in individual frailty on the dynamics of mor-
tality’’ by Vaupel, Manton, and Stallard (1979).
To explain this aspect, let us consider the case of a simple linear regression
of the number of children on the education level in the presence of three
clusters like those depicted in Fig. 1, where the clusters are, let us say, three
villages. A simple regression on the whole data set is a straight line with a
positive slope, indicating that the number of children increases with edu-
cation. This effect clearly holds at the aggregated village level, i.e. the higher
the average education level in a village, the higher the average number of
children. This aggregated effect results despite the regression is fitted on
individual data. A separate regression on each cluster exhibits a negative
y = 3.2 + 0.2 x
y = 6.2 - 0.8 x
y = 15.6 - 0.8 x y = 12.5 - 0.8 x
0
1
2
3
4
5
6
7
8
9
1 3 5 7 9 11 13 15
Education
Ch
ild
ren
Fig. 1. Multi-Level: A Simple Example with Three Clusters.
Life Course Data in Demography and Social Sciences 291
slope in each of the three villages, indicating a negative effect of education
on the number of children at the individual level. Indeed similar misleading
results may appear when event-history regressions are fitted on clustered
data as illustrated by the examples discussed by Beekink et al. (1999) and
Alter et al. (2001). What are the solutions?
Table 1 summarizes alternative formulations that can be adopted when
we are in the presence of G groups. For simplicity, we consider here re-
gression models with c covariates, generalization to more complex models
like event-history models being straightforward. Model m1 will capture ef-
fects at the group level. In models m2–m4, differences between groups are
introduced by means of additional parameters, an approach that is suitable
as long as G is not too large. Model m2 fits separate models on each clusters,
while in m3, the regressions are only seemingly independent, since the var-
iance of the error term is supposed to be the same in each group. Model m4
corresponds to the well-known case where, for each group, a specific effect is
introduced as a dummy variable. For a large number of groups, random
effect models1 m5 and m6 are best suited. In these models, the regression
coefficients are allowed to vary randomly from one group to another. In the
shared frailty formulation, only the constant is random, while the other
coefficients remain the same for all groups. The main advantage of these
random effect formulations is that their number of parameters is, as can be
shown from the last column, independent of the number of clusters. Ran-
dom effect models m5 and m6 may, therefore, have a much lower number of
parameters than models m2–m4 when G is large.
Even if we are interested in the aggregated effect, estimating them with
individual data, as with model m1, for example, requires some caution.
Table 1. Alternative Linear Models in Presence of G Clusters g.
Model Constant Effect of
Covariate
Variance of
Error Term
Number of
Parameters
m1 Average model Same Same Same 1+c+1
m2 Independent Group specific Group specific Group specific G(1+c+1)
m3 Seemingly
independent
Group specific Group specific Same G(1+c)+1
m4 Dummies Group specific Same Same G+c+1
m5 Random effects Random across
groups
Random across
groups
Same 2(1+c)+1
m6 Shared frailty Random across
groups
Same Same 2+c+1
GILBERT RITSCHARD AND MICHEL ORIS292
Indeed, the standard errors of the aggregated effects are derived from in-
dividual residuals, which may either over- or underestimate the between-
group discrepancy. For instance, in our example of Fig. 1, leaving out in
turn each of the three groups leads to great variations in the slope that
would be underestimated by the classical standard error. This aspect has
been investigated among others by Kish and Frankel (1974) and, for the
Cox model, for instance, by Lin and Wei (1989). In such settings, it is good
practice to use robust estimates of the variance of the regression coefficients.
Such robust estimates are usually obtained as grouped jackknife estimates,
i.e. by measuring the discrepancy of estimates obtained by leaving out suc-
cessively each of the G clusters, and can be expressed as sandwich estimates
(see Therneau & Grambsch, 2000, pp. 170–173).
Facilities for dealing with clusters are offered by several statistical systems,
Stata 8, S-Plus 6.2, and R 2.0 for instance. All the three mentioned programs
propose options to get robust standard errors. They also permit the intro-
duction of a shared frailty in parametric-hazard rate and Cox models. Com-
plete random effects are only available with discrete models that can be fitted
with logistic regression procedures. Indeed, logistic models are special cases
of generalized linear models (GLM).2 Hence, multilevel-logistic regression
is available whenever multilevel GLM is implemented. Barber, Murphy,
Axinn, and Maples (2000), for instance, show how to estimate a model with
several random effects with the HLM (Bryk, Raudenbush, & Congdon,
1996) and MLN (Goldstein et al., 1998) programs.
2.1.2. Illustration
To illustrate the scope of robust standard errors and shared frailty, we
consider a data set of 5,351 migrants collected from the 19th-century pop-
ulation registers of the Belgian commune of Sart (see Alter & Oris, 2000;
Alter et al., 2001, for a detailed description). This data set provides, among
others, information about the emigration date, the destination and the date
of return after emigration. Table 2 shows results of the fit of a continuous-
time Cox model. The hazard modeled is that of return after a time between 0
and 5 years, no return or return after 5 years being censored. We fitted a
basic model, i.e. without the cluster or frailty options, the same model but
requesting robust standard errors for the coefficients, and the model with a
gamma g(1/y,1/y) distributed frailty term shared by members of a same
family.3
The hazard ratios reported are just the exponential of the coefficients.
They indicate the hazard ratio for two profiles that differ by one unit of
the corresponding variable. For the frailty model this interpretation holds,
Life Course Data in Demography and Social Sciences 293
assuming the two profiles have the same frailty. For instance, according to
the basic model, the chances to return for a single are about one and a half
times the chances to return for a non-single. Likewise, the probability to
return is for a man about 3/4 of that for a woman. We checked on the basic
model that none of the time-covariate interactions is significant, which
comforts the proportionality assumption.
The coefficients are indeed the same for the basic and robust standard-
errors models. The significance of the coefficients differs, however, as can be
seen from the p-values. To be born in the Ardennes is significant at the 5%
level when we do not care about the cluster effect, while it is clearly not when
we control for it. This indicates that the seemingly significant birth-place
effect does not work at the family aggregated level. Likewise, we may notice
that, though the effect of the household economic ratio is significant among
families, its significance is not as clear as we would expect from the basic
model.
Let us now look at the results with a family shared frailty. First, we
may notice the highly significant variance of the random term, which
clearly indicates a between-families discrepancy. Two variables that looked
Table 2. Cox Model for Return within 5 Years after Emigration.
Coefficient Hazard Ratio p-Value (in %)
Basic Frailty Basic Frailty Basic Robust Frailty
Economic ratio 1.02 0.30 2.76 1.35 0.2 3.8 45.0
Man �0.28 �0.18 0.76 0.83 0.1 0.2 5.6
Single 0.40 0.52 1.49 1.68 1.2 1.2 0.3
Born in Ardennes 0.25 0.17 1.29 1.18 4.1 15.0 28.0
Age when leaving 0.01 0.00 1.01 1.00 12.0 17.0 62.0
To Ardennes Destination reference category
To rural �0.32 �0.60 0.73 0.55 5.7 14.0 0.2
To urban/indust �0.07 �0.23 0.93 0.79 50.0 68.0 6.8
To other �1.25 �1.25 0.29 0.29 0.0 0.0 0.0
Head or spouse of Parenthood reference category
Child of head 0.02 �0.25 1.02 0.78 89.0 90.0 19.0
Other parenthood 0.12 �0.27 1.13 0.76 54.0 56.0 26.0
No parenthood �0.50 �0.54 0.61 0.58 6.7 7.3 9.0
Standard deviationffiffiffi
yp
of family effect 1.75 0.0
Note: Sart 1812–1900, n ¼ 5,351.
GILBERT RITSCHARD AND MICHEL ORIS294
significant become non-significant, namely the gender (man) and the eco-
nomic ratio. This is not surprising for the latter, which is a typical family
contextual factor shared by members of the same family. Gender, on the
other hand, is clearly an individual characteristic. Its lack of significance in
the frailty model seems to indicate that the effect is not systematic within the
families. Its overall significance follows probably from differences among
male and female singles. A reverse phenomenon is observed for the rural
destination effect that becomes significantly different from the reference
Ardennes in the frailty model.
2.2. Markov Transition Models
In the presence of state sequences in panel data form, the natural question is
what are the transition probabilities from the states at time t�1 to the
possible states at time t, and how are these probabilities affected by indi-
vidual histories or contextual characteristics. Homogenous-Markov models
assume that these probabilities are independent of time t. In first-order
models, the transitions are supposed to depend only upon the state at t�1,
which means that the first lag summarizes the whole history of states at t�1
and before. Models of higher order k consider that the transitions depend on
k lags, i.e. on the states at t�k,y,t�1. Thus, basic Markov models state
that the transition probabilities remain constant over time and depend on a
limited, usually small, set of previous states.
Markov models of order k generate, when we are in the presence of
s states, sk transition distributions, i.e. a huge number of probabilities. They
may be approximated by mixture transition distribution (MTD) models
(Raftery & Tavare, 1994; Berchtold, 2001; Berchtold & Raftery, 2002) that
involve a much lower number of parameters, which renders the models
easier to interpret.
Other extensions of the Markov model include the hidden Markov model
(HMM) (see Rabiner, 1989; MacDonald & Zucchini, 1997) in which the
successive states of the observed variables are only indirectly linked through
an unobserved Markov chain and the double chain Markov model
(DCMM) (Paliwal, 1993; Berchtold, 1999, 2002), which states that the ob-
served states are outcomes of a Markov process randomly selected by a
hidden process. The use of hidden processes is a way to relax the usually
strong homogeneity assumption. For example, when studying social mo-
bility with data covering a whole century, it is hardly defendable to assume
that the same process works during the whole period (Lynch, 1998, p. 96).
Life Course Data in Demography and Social Sciences 295
Despite their interest, there has been only a limited use of Markov mod-
els, especially of non-homogenous ones, by historians and demographers. A
search on ‘‘Markov’’ within the famous Population Index database4 results
in only 28 hits among thousands of references. Moreover, most of those 28
hits refer to working papers or highly focused articles (with an emphasis on
the study of multistate population dynamics). The main reason for such a
limited use is that standard statistical packages offer only limited facilities to
fit such models. The available tools require a heavy coding task that dis-
courages most potential users. We can expect, however, that Markov
modeling will become much more popular with the recent release of March 2
(Berchtold & Berchtold, 2004). This software offers a friendly way to es-
timate Markov models without writing down any line of code.
2.2.1. Illustration
To illustrate the nature of knowledge we can expect from such an analysis,
we consider here the Blossfeld and Rohwer (2002) sample of 600 job ep-
isodes extracted from the German Life History Study. The episodes have
been classified into three job-length categories: (1) p3 years, (2)43 and
p10 years, and (3)410 years, and the data reorganized into 162 individual
sequences of 2–9 job episodes, dropping the cases with a single episode. The
question considered is how the present episode length depends upon those of
the preceding jobs. Notice that the job-length sequences considered here are
not panel data, which demonstrates that Markovian models are not re-
stricted to panel data. In this setting, the subscript t refers simply to the
position in the sequence rather to a specific time period.
The first- and second-order homogenous transition matrices are given in
Table 3. The same table also gives the distribution of the independence
model in which the transition probabilities stay the same irrespective of the
previous job length. Let us briefly illustrate how these tables should be read.
The independence distribution implies that the overall probability for a new
job to be a short one is 50%, while this probability is 35% for a medium job
and 15% for a long job. The first-order matrix indicates that the probability
that a new job started after a short one has 57% chances to be again a short
job. This probability falls to 43% after a job of medium length and to 20%
after a long job. From the second-order matrix, it follows, for instance, that
this same probability is 55% when the preceding short job was itself pre-
ceded by a short one, 60% when the preceding short job followed a medium
job and 100% when the preceding short job followed a long job. The last
column in the tables gives the half-length of a conservative 95% confidence
GILBERT RITSCHARD AND MICHEL ORIS296
interval for the probabilities in the concerned row. Hence, probabilities
smaller than this half-length should be considered as non-significant.
A glance at these tables leads to the following remarks. The first-order
matrix exhibits some differences in the transition probabilities after a short
(1), medium (2), or long (3) job. After a first job, the probability to start a
short job is significantly higher than to start a medium or long job, while
this is not the case after a medium or long job. The second-order matrix
does not provide evidence on the impact of the second lag job length. The
main differences concern the transition probabilities after long jobs (3),
which are mostly not statistically significant due to the low number of cases
concerned. This was confirmed by fitting an MTD model for which we
obtained a weight of 1 for the first lag and, hence, 0 for the second lag.
For relaxing the homogeneity assumption, we consider an HMM model
with a two-hidden-state process. Fitting this model, we get the distribution
of the initial state of the hidden variable, the transition matrix of the hidden
process, and the distributions of the transition to the job-length categories
associated to each of the two hidden states. These results are given in
Table 4. In addition, we get estimates (not shown here) of the most likely
sequence of hidden states associated to each observed sequence. Looking at
the cross tabulation below of these estimated hidden states with the ob-
served job length we see that the first hidden state is mainly associated to
Table 3. First and Second-Order Homogenous Markov Matrices.
Job Length at t Half Confidence
t–2 t–1 1 2 3 Interval
Independent 0.50 0.35 0.15 0.07
First Order 1 0.57 0.30 0.13 0.10
2 0.43 0.42 0.15 0.13
3 0.20 0.53 0.27 0.29
Second Order 1 1 0.55 0.30 0.15 0.11
2 1 0.60 0.30 0.10 0.20
3 1 1 0 0 0.65
1 2 0.37 0.45 0.18 0.18
2 2 0.50 0.41 0.09 0.20
3 2 0.45 0.33 0.22 0.38
1 3 0.33 0.17 0.50 0.46
2 3 0 0.87 0.13 0.40
3 3 1 0 0 1
Life Course Data in Demography and Social Sciences 297
short jobs and the second hidden state to medium and long jobs. This may
suggest considering only two types of jobs: p3 years and 43 years.
Hidden Observed
1 2 3
1 118 19 0
2 0 65 35
Table 5 summarizes goodness-of-fit statistics for our fitted models and for
the sake of comparison of the independence model. The shown statistics are
the number of independent parameters p, the deviance measured as minus
twice the log-likelihood5 (�2LogLik), the likelihood-ratio w2 statistics that
measures the improvement in �2LogLik over independence, its associated
degrees of freedom and its significance level, the pseudo R2 that gives the
relative improvement in �2LogLik and the Akaike (AIC) and Bayesian
(BIC) information criteria.6 These figures show that the fitted models do not
make much better than the independence model. We get the smallest
�2LogLik value for the second-order homogenous model, but at the cost of
11 additional independent parameters. The first-order homogenous model is
the only one that significantly improves the �2LogLik of the independence
model. It is also slightly better in terms of the AIC. However, no model
outperforms the independence model in terms of the BIC. These relatively
bad results are largely attributable here to the insufficient number of data
considered. This stresses a limitation of this Markov-modeling approach,
Table 4. Two State Hidden Markov Model.
Hidden State at Hidden State at t Half Confidence
t-1 t 1 2 Interval
Initial 0.56 0.44 0.11
1 0.78 0.22 0.12
2 0.53 0.47 0.19
Job Length at t
1 2 3
1 0.75 0.23 0.02 0.12
2 0.05 0.58 0.37 0.18
GILBERT RITSCHARD AND MICHEL ORIS298
namely the complexity of the models in terms of number of estimated pa-
rameters that requires a very large number of data.
3. MINING LONGITUDINAL LIFE COURSE DATA
Despite the last decade great boost in the use of data-mining tools for the
knowledge discovery from data (KDD) in fields ranging from genetics to
finance, from marketing to medical diagnosing, from text analysis to image
or speech recognition, such approaches have received only little attention
for extracting interesting knowledge from longitudinal data describing life
courses. An important exception is Blockeel, Furnkranz, Prskawetz, and
Billari (2001) who showed how mining frequent itemsets may be used to
detect temporal changes in event-sequences frequency from the Austrian
FFS data. In Billari, Furnkranz, and Prskawetz (2000), three of the same
authors also experienced an induction-tree approach for exploring differ-
ences in Austrian and Italian life-event sequences. We initiated ourselves
(Oris, Ritschard, & Berchtold, 2003) social-mobility analysis with induction
trees.
Data mining is mainly concerned with the characterization of interesting
pattern, either per se (unsupervised learning) or for a classification or pre-
diction purpose (supervised learning). Unlike the statistical-modeling ap-
proach, it makes no assumptions about an underlying process generating the
data and proceeds mainly heuristically.
Beside their non-parametric or assumption-free characteristic, data-
mining methods present also the advantage for our social demographic
framework to be able to handle sequences of the various family, education,
work, health, emotional, and other personal events that define a life course.
They seem in that regard promising tools for gaining knowledge about life
trajectories and should thus usefully complement the previously discussed
Table 5. Global Model Goodness-of-Fit Statistics.
Model m p �2LogLik w2 df Sig BIC AIC Pseudo R2
Independent 2 472.8 0 0 — 483.7 476.8 0
Homogenous order 1 6 462.6 10.2 4 0.04 495.4 474.6 0.022
Homogenous order 2 13 460.6 12.2 11 0.35 531.7 486.6 0.026
HMM 2 states 7 468.6 4.2 5 0.52 506.9 482.6 0.009
Note: Number of sequences ¼ 107, usable n ¼ 237.
Life Course Data in Demography and Social Sciences 299
statistical methods. Event-history models, for example, focus on the risk of a
given transition, but do not provide insights on trajectories. Markov mod-
els, on the other hand, attempt to characterize the stochastic process that
drives successive transitions between states. They provide in that sense some
synthetic information about trajectories. However, only trajectories between
states of a generally unique variable, social, or civil status, for example, can
be investigated this way. Markov models, even those allowing for covari-
ates, can hardly handle together the various life events. Furthermore,
Markov models remain quite rigid by assuming that the transition prob-
abilities do not depend upon the present time but only on a small limited
number (the order of the model) of previous states. From a substantial
standpoint, the hereafter discussed sequence-mining approach is best suited
to discover among the many possible trajectories, for example, from the
diversity of formations to the diversity of working lives, those that are
typical of real life courses of real persons and by contrast those that are
atypical.
Since data-mining methods are mainly assumption-free, exploring trajec-
tories with them may answer to the criticisms of the French sociologist
Pierre Bourdieu (1986) about the ‘‘biographical delusion’’. Bourdieu, in fact,
denounced the concept of ‘‘life cycle’’, and its emphasis on norms, norms
supposed to lend to ‘‘normal’’ trajectories. With the assumption-free mining
of longitudinal data, we precisely pass the boarder between the ‘‘causality’’
or ‘‘data-modeling culture’’ and what Breiman (2001) calls the ‘‘algorithmic
culture’’ (see Billari, this volume).
In the rest of this section, we shortly describe the mining of sequential
rules and the induction tree approach, focusing on the nature of knowledge
we may expect from such tools (for a more general introduction to data
mining, see Hand, Mannila, & Smyth, 2001; or Han & Kamber, 2001).
These books cover many more methods. The two tools discussed here are,
however, in our mind, the two more promising ones for longitudinal data.
3.1. Mining Event-Sequential Association Rules
Each life course can be seen as a sequence of life events: birth, important
disease, recovering from disease, starting school, ending school, first job,
first union, leaving home, first child, death of father, marriage, etc. Mining
sequential-association rules aims at determining the most typical sequences
or subsequences together with their frequencies, and at deriving association
rules like having experienced the subsequence first job, first union, first child,
GILBERT RITSCHARD AND MICHEL ORIS300
is most likely to be followed by a sequence marriage, second child. By
contrast, indeed, mining frequent sequences and rules also reveals atypical
life courses. Note that event sequences differ from state sequences as con-
sidered by Markov models or optimal matching. Nevertheless, sequence
mining could as well be applied to state sequences.
Technically, the mining of frequent-event sequences and sequential-
association rules is a special case of the mining of frequent itemsets and
association rules. In data mining, an itemset is a set of items that are selected
together and an association rule is just a rule that says that if A occurs then
B is very likely to occur too. The basic tuning parameters of the mining
process are the support and the confidence thresholds. The support is the
minimal frequency in the database for an item set to be selected, while the
confidence of the rule is the probability that the consequence occurs when
the premise is observed. These basic-selection criteria are complemented by
other additional measure of the interestingness of the rule, like the propor-
tion of the rule its counter examples. Most algorithms for seeking frequent-
itemsets and rules are variants of the well-known Apriori algorithm
(Agrawal & Srikant, 1994; Mannila, Toivonen, & Verkamo, 1994). A typ-
ical application consists in finding the items that are more often ordered
together by customers. Sequences that we consider here differ from general
itemsets in that order matters. Multiple algorithms adapted for sequences
have been proposed since the pioneering contributions by Agrawal and
Srikant (1995) and Mannila, Toivonen, and Verkamo (1997).
3.1.1. Illustration
We have not yet ourselves experienced a sequential rule mining analysis on
demographic data. For the sake of illustration, we report here the analysis
carried out by Blockeel et al. (2001). The data considered originated from
the 1995 Austrian Fertility and Family Survey (FFS). The events analyzed
are those of the partnership and fertility retrospective histories of 4,581
women and 1,539 men aged between 20 and 54 at the survey time. The
observed women and men were partitioned into 5 years cohorts and the
objective of the analysis was to discover frequent partnership and birth
event patterns that mostly varied among cohorts.
The mining was done by means of the Warmr process implemented in the
ACE Data-Mining System (Blockeel, Dehaspe, Ramon, & Struyf, 2004).
The search was not limited to simple sequences of strictly ordered events but
allowed for more complex patterns combining multiple subsequences. An
important pattern found was having a child after first union and having
Life Course Data in Demography and Social Sciences 301
both a marriage and a second child after this first birth, the marriage and
second child being not ordered. The seeking of such not strictly ordered
pattern requires indeed some filtering, namely the elimination of redundant
patterns. For example, completing the above mentioned pattern with the
additional condition of having a marriage after the first union would not
bring any new information and is therefore redundant. Also, the rules gen-
erated were restricted to premises refereeing to the cohort. Finally, only
patterns that exhibit a great discrepancy in the proportion of individuals
satisfying it in each cohort were retained.
Fig. 2 is an example of outcome provided by this analysis. It shows
the strong declining proportion of individuals who started their first union
when they married. The mining process found this pattern, i.e. date of first
union equals date of marriage, to be the one that exhibits the strongest
changes in frequency among cohorts. Indeed, many other patterns, some-
times more complicated, were also found to have great variability in their
frequency.
Fig. 2. Negative Trend in the Proportion of First Unions Starting at Marriage.
Source: Reproduced from Blockeel et al. (2001) with permission from the authors.
GILBERT RITSCHARD AND MICHEL ORIS302
3.2. Social Transition Analysis with Induction Trees
Let us now turn to induction trees and the insight they may provide on the
understanding of mobility. In mobility analysis, the focus is on how states at
previous time t�1, t�2,y, and possibly some additional covariates, influ-
ence the present state at t. This setting is very similar to that of Markovian
models. In contrast with this parametric-modeling approach, the tree in-
duction is, however, a non-parametric method. It provides a heuristic way to
catch how the previous states and covariates jointly influence the state at t.
Though we focus here on intergenerational social-mobility analysis, it is
worth mentioning that the scope of induction trees for life course analysis is
much broader. For instance, De Rose and Pallara (1997) used a tree
approach for segmenting time to marriage curves of Italian women; Billari
et al. (2000) used trees for analyzing differences in event sequences between
Austrians and Italians; and we can easily imagine many other applications.
Induction trees, i.e. decision trees induced from data, are basically su-
pervised classification tools (Quinlan, 1986). As pointed out in Ritschard
and Zighed (2003), they also convey powerful descriptive information. Their
learning principle is quite simple and they produce easily interpretable
results.
An induced tree defines rules for predicting the value of a response var-
iable from a set of potential predictors. The set of rules indeed characterizes
a partition of the cases, each rule defining a class. The prediction inside each
class of this partition is simply the modal-observed value when the response
is categorical and the mean observed value when it is quantitative. In the
quantitative case, the tree is called a regression tree (Breiman, Friedman,
Olshen, & Stone, 1984). Extension in this case includes model trees
(Malerba, Appice, Ceci, & Monopoli, 2002) and logistic model trees
(Landwehr, Hall, & Frank, 2003), which use a linear or logistic regression
for the prediction inside the classes of the partition. Tree algorithms have
also been proposed for predicting functions instead of values and those that
like RECPAM (Ciampi, Hogg, McKinney, & Thiffault, 1988) predict, for
instance, survival functions may be of special interest for life course analysis.
Here we consider only categorical responses, i.e. classification trees. The
easiest way to describe the tree induction principle is by looking at an
example. We begin therefore by describing the framework of the illustration
we will consider.
We use social family history data on intergenerational-social transition in
the 19th-century Geneva (Ryczkowska & Ritschard, 2004). The data were
collected from the marriage-registration acts that provide the profession of
Life Course Data in Demography and Social Sciences 303
the spouses as well as that of their parents. For 572 acts, it has been pos-
sible to find a match with the marriage of the father of one of the spouses.
For these cases, we have the profession of the married man, of his father
at the son’s marriage, of the matched father at his own marriage, and of
the grandfather at the matched father marriage. The professions were
grouped into three social statuses, namely low, high, and clock and watch-
makers who formed an important specific corporation in the 19th century
Geneva.
The variable we want to predict is the status of the son at his marriage,
which is clearly a categorical response, and we consider four potential pre-
dictors. The first three are status variables, namely the status of the father at
son’s marriage, the status of the matched father at his own marriage, and the
status of the grandfather at father’s marriage. The fourth predictor is the
birthplace that can take one of the 12 values: Geneva city (GEcity); Geneva
surrounding land (GEland); neighboring France (neighbF); Vaud (VD);
which is a neighboring region of Geneva; Neuchatel (NE), a further
French-speaking region also specialized in watch and clock making, other
French-speaking Switzerland (otherFrCH), German-speaking Switzerland
(GermanCH), Italian-speaking Switzerland (TI), France (F), Germany (D),
Italy (I), and other. The grown tree is shown in Fig. 3.
The tree-growing principle is as follows. First, all cases are grouped to-
gether in a root node (at the top of the tree) in which the distribution of the
response variable, the status of the married man for our analysis, is its
marginal distribution. The goal is to split this group in new nodes such that
the distribution of the response variable differs as much as possible from one
node to the other. The splitting is done iteratively using the categorical
values of the predictor selected at each step. At the first step, we seek the
predictor that best splits the root node and split the node according to the
values of this predictor. The process is then repeated at each new node until
a stopping rule is reached. Stopping rules typically concern the minimal
node size, the maximal number of levels or the statistical significance of the
improvement in the optimized criterion. In our study, we have retained the
CHAID method (Kass, 1980) that selects at each step the predictor that,
when it is cross tabulated with the response variable, generates the most
significant independence w2 statistics.7 CHAID also seeks the aggrega-
tion level of the categories of the predictors that generates the most sig-
nificant w2 and then splits indeed according to the optimally merged
categories. We generated the tree of Fig. 3 with Answer Tree 3.1 (SPSS,
2001) by setting the minimal node size to 15 and requiring a maximal sig-
nificance level of 5%.
GILBERT RITSCHARD AND MICHEL ORIS304
Fig. 3. Social Transition Tree with Birth Place Covariate.
Life
Course
Data
inDem
ographyandSocia
lScien
ces305
Alternative methods, among which CART proposed by Breiman et al.
(1984) and C4.5 due to Quinlan (1993) are among the best known, differ
mainly by the criteria used for selecting the split variable at each step.8
3.2.1. Knowledge Provided by the Tree
Looking at Fig. 3, we see that the first split is done according to the father
status at son’s marriage. This tells us that among the four attributes con-
sidered, the status of the father is the most discriminating. The status of the
married man depends, for instance, more on the father’s status than on his
birthplace. The distribution inside the nodes of the first level are just the
columns of the cross classification of the statuses of the father and the son.
We observe here that the clock makers form a much closed group with a
high probability for the son to become a clock maker when the father
himself is a clock maker while this probability is much lower for the three
other groups. A similar result holds for the high classes, while there are
evidences about social ascension possibilities when the father belongs to the
lower class.
Three of the four first-level nodes are split further. The only one that is
not split is that of married men whose father belongs to the clock and watch
makers. This node is thus a terminal leaf, which indicates that the status of
clock maker father conveys all the significant information for predicting the
status of the son. This is a consequence of the strong social reproduction
process inside the class of clock makers. The married men whose father was
dead are split according to the grandfather’s status, which means that the
grandfather’s status is more discriminating for this subgroup than the status
of the father at his own marriage. There is a strong tendency for the married
man to reproduce the grandfather’s status when the father is deceased. The
group defined by a high status of the father as well as that defined by a low
father’s status is split according to the birthplace. Both splits are binary.
They do not make use, however, of the same binary partition of birthplaces.
In both cases, i.e. with a father belonging to the low or high classes, the
men born in neighboring France, in German-speaking Switzerland, or in
Vaud have a relatively high probability to get only a low status. This is
also true for men born in French-speaking Switzerland outside Geneva and
Neuchatel when their father belongs to the lower class.
The additional levels show that when the high position of the father
results from a recent social ascension, i.e. ascension since the father’s
marriage (level 3) or from the position of the grandfather (level 4), the
reproduction of the father’s status by the married man is less strong. The
subtree that concerns the men whose father was dead, shows effects of
GILBERT RITSCHARD AND MICHEL ORIS306
the grandfather’s status very similar to those of the status of the father when
he is alive at the marriage.
3.2.2. Goodness-of-Fit of the Descriptive Tree
Classically, the quality of a tree is evaluated in terms of its classification
predictive quality, which is measured by the correct classification rate of the
tree. Recall that the classification is done by assigning to each case the most
frequent value in its leaf. For our tree, the correct classification rate is 57.6%.
This corresponds to a 42.4% error rate. At the root node, before introducing
any predictor, the correct classification rate is 44.4%, which gives an error
rate of 55.6%. Our tree allows thus a 24% ( ¼ (55.6-42.4)/55.6) reduction of
the error rate. These figures are, nevertheless, irrelevant in our case, since we
are not using the tree for classification purposes. We do not consider the
classification results. The descriptive knowledge considered follows directly
from the distributions inside the nodes. Hence, we consider the tree as a
probability tree rather than a classification tree. In Ritschard and Zighed
(2003), we have proposed indicators that better suit this descriptive point of
view. We can, for instance, measure with a likelihood-ratio w2 (G2) the di-
vergence between the distributions predicted by the tree (those in the leaves)
and those of the finest partition that our four predictors may generate.9 We
get 312.5 for 300 degrees of freedom, and its p-value is 29.8% indicating
apparently a good fit. Note that though the four predictors define theoret-
ically 576 different profiles, only the 163 actually observed are taken into
account. When these profiles are cross tabulated with the three statuses, we
get 489 cells. For 572 data, this gives an average of a bit more than one per
cell, which is insufficient to ensure the w2 distribution of G2. Hence, we
should not attach here too much confidence to the p-value (Table 6).
For comparison purposes, Table 6 reports the G2 statistic for a set of
nested trees, namely the independence tree corresponding to the root node
only, the tree expanded respectively one level only, two levels and three
levels, the fitted tree and the saturated tree that generates the finest partition.
Beside G2, its degrees of freedom and significance level, the table shows the
BIC and AIC information criteria and the adjusted pseudo R2. The latter
measures the percentage of reduction of the G2=df ratio as compared with
the independence tree. The BIC and the AIC are G2’s penalized for the
complexity.
We see that with less than three levels there is a lack of fit, the divergence
with the finest partition being significant at the 5% level. The difference in
G2s between two nested trees can also be compared with a w2 distribution
with degrees of freedom being the difference in these degrees for the two
Life Course Data in Demography and Social Sciences 307
models. Thus, the Level 3 tree differs by DG2 ¼ 15:1 and Ddf ¼ 4 from the
fitted model, which is clearly significant. Hence, the two splits leading to
Level 4 look jointly statistically significant. From the BIC point of view, the
Level 2 tree provides the best compromise between fit and complexity. Level
3 or 4 trees seem, however, preferable according to the interesting insight
brought by the additional levels and the significant divergence of Level 2
with the saturated model. The AIC, which is known, however, to under-
estimate the impact of complexity, selects here the fitted tree.
Trees look really promising thanks mainly to their ease of use and to their
visual outcome. When it comes to interpretation, one should be aware,
nevertheless, that trees may be instable in the sense that small changes in the
data could alter the structure especially splits and variables selected at
higher levels. It is then important to avoid growing too complex trees. Re-
laying on BIC or AIC criteria should help determining a somewhat robust
tree. Splits behind the optimal BIC or AIC will be less reliable and their
interpretation then requires more caution.
4. CONCLUSION
This paper stressed the scope and limits of various methods available for
analyzing life course data globally, and especially in demography. Demo-
graphers and historical demographers invented their own longitudinal-
analytical tools like the life tables or the family reconstitution, almost since
the birth of their discipline. However, everywhere but probably more in the
French-speaking areas, those sciences of the masses hesitated to take a step
further, while they were so close from the life course perspective and meth-
ods. Many academics are still living this transitiony. For adepts of highly
quantitative social sciences, we wanted to both introduce and illustrate
Table 6. Goodness-of-Fit of the Tree and Subtrees.
Tree G2 Df sig BIC AIC Pseudo R2
Independent 482.3 324 0.000 2319.6 812.3 0
Level 1 408.2 318 0.000 1493.9 750.2 0.14
Level 2 356.0 310 0.037 1492.5 714.0 0.23
Level 3 327.6 304 0.168 1502.2 697.6 0.28
Fitted 312.5 300 0.298 1512.5 690.5 0.30
Saturated 0 0 1 3104.7 978.0 1
GILBERT RITSCHARD AND MICHEL ORIS308
promising methodological perspectives without hiding the complexity of the
new approaches. At the same time, we did not elaborate this contribution
only for our disciplinary fellows, since one of the most important evolutions
is that the analytical techniques obviously lend us to neighboring disciplines
that share the same tools and explore similar concepts, giving to the inter-
disciplinary ambition a growing substance.
We have chosen to illustrate different approaches, and especially the
emerging data-mining techniques that should be able to provide original
additional insights on results provided by more classical statistical methods.
The discussion, however, is by no means exhaustive. Among the techni-
ques we did not discuss, optimal matching (Abbot & Forrest, 1986; Malo
& Munoz, 2003) deserves special attention. Optimal matching is, like
Markovian models, a state-sequence analysis tool. It is merely a data-mining
approach, since it proceeds heuristically. Unlike the mining of frequent se-
quences that does not care about the similarity between sequences, optimal
matching is concerned with the discovering of similarities between sequence
patterns. Optimal matching evaluates the proximity between two sequences
by seeking the minimal number of changes that can transform a sequence a
into a sequence b. Survival (Ciampi et al., 1988; Segal, 1988) and risk trees
(Leblanc & Crowley, 1992) developed in the field of biomedicine during the
first half of the 1990s would also merit further attention from historians and
demographers.
It is worth mentioning that the statistical and data-mining approaches are
not substitutes for one another. They are complementary, each method
bringing its own insight. The choice of a method will be dictated by the kind
of data available: spell durations, event sequences, state sequences, and
indeed the type of results expected: knowledge about probability of tran-
sitions, effects on these risks, characteristic trajectories, or life sequences.
Another important element for this choice, at least for the end user, is the
availability of user-friendly softwares and the level of expertise required to
run the method and interpret the results. Many softwares propose duration
or hazard models and/or classification trees. It is less obvious to find friendly
tools for Markovian models and the mining of sequential rules. March 2 is a
promising solution for Markovian models, while specialized softwares like
Clementine propose sequence mining tools (see http://www.kdnuggets.com
for a list of commercial and free data mining softwares). The use and in-
terpretation of hazard models is very similar to that of other regression-like
models, which renders them attractive. The interpretation of induction trees
is also very straightforward and looks therefore as a promising tool. Nev-
ertheless, the fine tuning of trees, which may be highly instable, requires
Life Course Data in Demography and Social Sciences 309
generally more care than hazard models. Mining frequent sequential pat-
terns also requires some experience to get interesting patterns. In any case,
the new highlights provided by these data-mining approaches are worth the
effort.
NOTES
1. Random effect models are also known as multilevel, hierarchical or mixed-effect models.2. GLM models (McCullagh & Nelder, 1989) cover a large number of parametric
models. They assume a distribution of the natural exponential family for the de-pendent variable and are, in their simpler form, simply characterized by a linkfunction that describes how the mean of the dependent variable is linked to the linearform of the explanatory variables. For example, we get the classical linear modelwith a Gaussian distribution and the identity link, the logistic model with a Bernoullidistribution and the logit link, and the log-linear model with a Poisson distributionand the log link.3. The estimations were obtained with S-plus 6.2. We suspect a bug in Stata 8 that
was not able to converge within 24 h for the frailty model while S-Plus provided theresults within 2 min.Formally, the estimated hazard model is h(t, x1,y, xp) ¼ ng h0(t) exp(b1x1+
y+bpxp), where h0(t) is the baseline hazard function and ng the shared frailty term.We estimated this model assuming a gamma g(1/y,1/y) distribution for the frailtyterm ng, for which we have E(ng) ¼ 1 and Var(ng) ¼ y.4. http://popindex.princeton.edu/5. The deviance -2LogLik may be seen as the distance between the predictions
generated by the model and the observed counts. Hence it is a measure of global fit.However, it cannot be used here to test the fit since we do not know its distribution.6. The AIC and BIC criteria are penalized forms of the –2LogLik that take ac-
count of the complexity, i.e. the number of estimated parameters. Among the two,the BIC is usually preferred since the AIC is known to insufficiently penalize com-plexity. The model with the minimal BIC offers the best compromise between fit andcomplexity.7. Significance is generally evaluated with a Bonferroni correction for taking ac-
count of the multiple test sequence that controls each split decision.8. CART maximizes the reduction in the Gini index also known as the quadratic
entropy. It generates only successive binary splits. C4.5 uses the gain ratio defined asthe reduction in Shannon’s entropy normalized by the entropy of the distributionamong the classes of the generated partition. Unlike the CHAID method, for whichthe significance of the w2 provides a natural validation for the split, CART and C4.5do not have such a natural split validation criteria. These methods complete there-fore the growing process with a post pruning round that, starting from the leaves,eliminates unreliable splits. Only splits that improve the predictive error rate areretained. There are also graph induction tools like SIPINA (Zighed & Rakotomalala,1996), which generalize trees by allowing the merge of nodes with similar insidedistribution.
GILBERT RITSCHARD AND MICHEL ORIS310
9. G2 is indeed the deviance -2LogLik. It measures how far the counts predicted bythe tree are from those observed for the finest possible partition. When the predictedcounts are not too small, it has an approximate w2 distribution and can be used fortesting the goodness-of-fit. Note that the w2 reported in Table 5 would correspondhere to the difference between the G2 of the tree and that of the root node (inde-pendence). We expect w2 to be large while G2 should be small.
REFERENCES
Abbot, A., & Forrest, J. (1986). Optimal matching methods for historical sequences. Journal of
Interdisciplinary History, 16, 471–494.
Agrawal, R., & Srikant, R. (1994). Fast algorithm for mining association rules in large data-
bases, In: Proceedings of the International Conference on Very Large Data Base
(VLDB’94), Santiago de Chile (pp. 487–499). San Mateo: Morgan-Kaufman.
Agrawal, R., & Srikant, R. (1995). Mining sequential patterns, In: Proceedings of the
International Conference on Data Engineering (ICDE), Taipei, Taiwan (pp. 487–499).
Taiwan: IEEE Computer Society.
Allison, P. D. (1982). Discrete-time methods for the analysis of event histories. In: S. Lienhardt
(Ed.), Sociological methodology (pp. 61–98). San Francisco: Jossey-Bass Publishers.
Alter, G. (1998). L’event history analysis en demographie historique: Difficultes et perspectives.
Annales de demographie historique, 2, 23–35.
Alter, G., & Oris, M. (2000). Mortality and economic stress: Individual and household re-
sponses in a nineteenth-century Belgian village. In: T. Bengtsson & O. Saito (Eds),
Population and economy: From hunger to modern economic growth (pp. 335–370). Oxford:
Oxford University Press.
Alter, G., Oris, M., & Brostrom, G. (2001). The family and mortality: A case study from rural
Belgium. Annales de demographie historique, 1, 11–31.
Aranza-Ordaz, F. J. (1983). An extension of the proportional hazards model for grouped data.
Biometrics, 39, 109–117.
Barber, J. S., Murphy, S. A., Axinn, W. G., &Maples, J. (2000). Discrete-time multilevel hazard
analysis. In: M. E. Sobel & M. P. Becker (Eds), Sociological methodology, (pp. 201–235).
New York: The American Sociological Association.
Beekink, E., van Poppel, F., & Liefbroer, A. C. (1999). Surviving the loss of the parent in a
nineteenth-century Dutch provincial town. Journal of Social History, 32, 614–670.
Berchtold, A. (1999). The double chain Markov model. Communications in Statistics: Theory
and Methods, 28(11), 2569–2589.
Berchtold, A. (2001). Estimation in the mixture transition distribution model. Journal of Time
Series Analysis, 22(4), 379–397.
Berchtold, A. (2002). High-order extensions of the double chain Markov model. Stochastic
Models, 18(2), 193–227.
Berchtold, A., & Berchtold, A. (2004). MARCH 2.01: Markovian model computation and
analysis. User’s guide, www.andreberchtold.com/march.html.
Berchtold, A., & Raftery, A. E. (2002). The mixture transition distribution model for high-order
Markov chains and non-Gaussian time series. Statistical Science, 17(3), 328–356.
Berquo, E., & Xenos, P. (1992). Editor’s introduction. In: E. Berquo & P. Xenos (Eds), Family
systems and cultural change (pp. 8–12). Oxford: Clarendon Press.
Life Course Data in Demography and Social Sciences 311
Billari, F. C., Furnkranz, J., & Prskawetz, A. (2000). Timing, sequencing, and quantum of life
course events: A machine learning approach. Working Paper no. 010, Max-Plank-Institute
for Demographic Research, Rostock.
Billari, F. C. (2005). Life course analysis. Two cultures? Some reflections with examples from
the analysis of the transition to adulthood. In: R. Levy, P. Ghisletta, J.-M, Le Goff,
D. Spini & E. Widmer (Eds), Towards an interdisciplinary perspective on the life course
(pp. 267–286), Advances in Life Course Research, Vol. 10. Amsterdam: Elsevier.
Blockeel, H., Dehaspe, L., Ramon, J., & Struyf, J. (2004). The ACE data mining system. User’s
manual. Katholieke Universiteit Leuven, Leuven.
Blockeel, H., Furnkranz, J., Prskawetz, A., & Billari, F. (2001). Detecting temporal change in
event sequences: An application to demographic data. In: L. D. Raedt & A. Siebes (Eds),
Principles of data mining and knowledge discovery: 5th European conference, PKDD 2001
(Vol. 2168 of LNCS, pp. 29–41). Freiburg in Brisgau: Springer.
Blossfeld, H.-P., Hamerle, A., & Mayer, K. U. (1989). Event history analysis, statistical theory
and application in the social sciences. Hillsdale, NJ: Lawrence Erlbaum.
Blossfeld, H.-P., & Rohwer, G. (2002). Techniques of event history modeling, new approaches to
causal analysis (2nd ed.). Mahwah, NJ: Lawrence Erlbaum.
Bocquier, P. (1996). L’analyse des enquetes biographiques a l’aide du logiciel STATA. Paris:
Centre franc-ais sur la population et le developpement.
Bourdieu, P. (1986). L’illusion biographique. Actes de la Recherche en Sciences sociales, 62–63,
69–72.
Breiman, L. (2001). Satistical modeling: The two cultures (with discussion). Statistical Science,
16(3), 199–231.
Breiman, L., Friedman, J. H., Olshen, R. A., & Stone, C. J. (1984). Classification and regression
trees. New York: Chapman & Hall.
Bryk, A., Raudenbush, S. W., & Congdon, R. (1996). HLM: Hierarchical linear and nonlinear
modeling with the HLM/2 l and HLM/3 l programs. Chicago: Scientific Software Inter-
national.
Caselli, G., Vallin, J., & Wunsch, G. (Eds), (2001). Demographie: analyse et synthese, vol. I.
La dynamique des populations. Paris: Institut national d’etudes demographiques.
Ciampi, A., Hogg, S. A., McKinney, S., & Thiffault, J. (1988). RECPAM: A computer program
for recursive partitioning and amalgamation for censored survival data and other sit-
uations frequently occurring in biostatistics I. Methods and program features. Computer
Methods and Programs in Biomedicine, 26(3), 239–256.
Courgeau, D., & Lelievre, E. (1993). Event history analysis in demography. Oxford: Clarendon
Press.
Cox, D. R. (1972). Regression models and life-tables. Journal of the Royal Statistical Society,
Series B, 34(2), 187–220.
Demeny, P., & McNicoll, G. (Eds) (2003). Encyclopedia of population (Vol. 2). New York:
McMillan.
De Rose, A., & Pallara, A. (1997). Survival trees: An alternative non-parametric multivariate
technique for life history analysis. European Journal of Population, 13, 223–241.
Dupaquier, J., & Dupaquier, M. (1985). Histoire de la demographie. Paris: Perrin.
Dykstra, P., & van Wissen, L. J. G. (1999). Introduction: The life course approach as
an interdisciplinary framework for population studies. In: L. J. G. van Wissen &
P. Dykstra (Eds), Population issues: An interdisciplinary focus (pp. 1–22). New York:
Plenum Press.
GILBERT RITSCHARD AND MICHEL ORIS312
Goldstein, H., Rasbash, J., Plewis, I., Draper, D., Browne, W., Yang, M., Woodhouse, G., &
Haely, M. (1998). A user guide to MLwiN. Technical Report, Multilevels Models Project,
London.
Greenhalgh, S. (1996). The social construction of population science: An intellectual, institu-
tional and political history of twentieth-century demography. Comparative Studies in
Society and History, 38(1), 26–66.
Hagestad, G. O. (2003). Interdependent lives and relationships in changing times: A life course
view of families and aging. In: R. A. H. Settersten (Ed.), Toward new understanding of
later life (pp. 135–160). Amytiville, New York: Baywood Publishing.
Hand, D. J., Mannila, H., & Smyth, P. (2001). Principles of data mining (adaptive computation
and machine learning). Cambridge, MA: MIT Press.
Han, J., & Kamber, M. (2001). Data mining: Concept and techniques. San Francisco: Morgan
Kaufmann.
Hogdson, D. (1988). Demography as social science and policy science. Population and Devel-
opment Review, 9(1), 541–569.
Hogdson, D. (1991). The ideological origins of the population association of America. Pop-
ulation and Development Review, 17(1), 1–34.
Hohn, C. (1992). The IUSSP programme in family demography. In: E. Berquo & P. Xenos
(Eds), Family systems and cultural change, (pp. 3–7). Oxford: Clarendon Press.
Holford, T. R. (1980). The analysis of rates and survivorship using log-linear models. Biomet-
rics, 65, 159–165.
Kass, G. V. (1980). An exploratory technique for investigating large quantities of categorical
data. Applied Statistics, 29(2), 119–127.
Kish, L., & Frankel, M. R. (1974). Inference from complex samples (with discussion). Journal of
the Royal Statistical Society, Series B, 36, 1–37.
Landwehr, N., Hall, M., & Frank, E. (2003). Logistic model trees. In: N. Lavrac,
D. Gamberger, L. Todorovski & H. Blockeel (Eds), Machine learning: ECML 2003
(Vol. 2837 of LNAI, pp. 241–252). Berlin: Springer.
Leblanc, M., & Crowley, J. (1992). Relative risk trees for censored survival data. Biometrics, 48,
411–425.
Lillard, L. A. (1993). Simultaneous equations for hazards: Marriage duration and fertility
timing. Journal of Econometrics, 56, 189–217.
Lin, D. Y., & Wei, L. J. (1989). The robust inference for the Cox proportional hazards model.
Journal of the American Statistical Association, 84, 1074–1078.
Lynch, K. A. (1998). Old and new research in historical patterns of social mobility. Historical
Methods, 31(3), 93–98.
MacDonald, I. L., & Zucchini, W. (1997). Hidden Markov and other models for discrete-valued
time series. London: Chapman & Hall.
Malerba, D., Appice, A., Ceci, M., & Monopoli, M. (2002). Trading-off local versus global
effects of regression nodes in model trees. In: M.-S. Hacid, Z. W. Ras, D. A. Zighed &
Y. Kodratoff (Eds), Foundations of intelligent systems, ISMIS 2002 (Vol. 2366 of LNAI,
pp. 393–402). Berlin: Springer.
Malo, M. A., & Munoz, F. (2003). Employment status mobility from a lifecycle perspective: A
sequence analysis of work-histories in the BHPS. Demographic Research, 9(7), 471–494.
Mannila, H., Toivonen, H., & Verkamo, A. I. (1994). Efficient algorithms for discovering
association rules. In: Proceedings of the AAAI’94 workshop knowledge discovery in
databases (KDD’94), Seattle, WA (pp. 181–192).
Life Course Data in Demography and Social Sciences 313
Mannila, H., Toivonen, H., & Verkamo, A. I. (1997). Discovery of frequent episodes in event
sequences. Data Mining and Knowledge Discovery, 1(3), 259–289.
McCullagh, P., & Nelder, J. A. (1989). Generalized linear models. London: Chapman & Hall.
Oris, M., & Poulain, M. (2003). Entre blocages spatiaux et nouvelles dynamiques familiales:
Quitter ses parents. In: Populations et defis urbains. Chaire Quetelet 1989 (pp. 313–337).
Louvain-la-Neuve: Academia-Bruylant/L’Harmattan.
Oris, M., Ritschard, G., & Berchtold, A. (2003). The use of Markov process and induction trees
for the study of intergenerational social mobility in nineteenth century Geneva. In:
Social science history association annual meeting, Baltimore.
Paliwal, K. K. (1993). Use of temporal correlation between successive frames in a hidden
Markov model based speech recognizer. Proceedings of the ICASSP, 2, 215–218.
Quinlan, J. R. (1986). Induction of decision trees. Machine Learning, 1, 81–106.
Quinlan, J. R. (1993). C4.5: Programs for machine learning. San Mateo: Morgan Kaufmann.
Rabiner, L. R. (1989). A tutorial on hidden Markov models and selected applications in speech
recognition. Proceedings of the IEEE, 77(2), 257–286.
Raftery, A. E., & Tavare, S. (1994). Estimation and modelling repeated patterns in high order
Markov chains with the mixture transition distribution model.Applied Statistics, 43, 179–199.
Ritschard, G., & Zighed, D. A. (2003). Goodness-of-fit measures for induction trees. In:
N. Zhong, Z. Ras, S. Tsumo & E. Suzuki (Eds), Foundations of intelligent systems,
ISMIS03 (Vol. 2871 of LNAI, pp. 57–64). Berlin: Springer.
Rohwer, G., & Potter, U. (2002). TDA user’s manual. Software, Ruhr-Universitat Bochum,
Fakultat fur Sozialwissenschaften, Bochum.
Ryczkowska, G., & Ritschard, G. (2004). Mobilites sociales et spatiales: Parcours intergenera-
tionnels d’apres les mariages genevois, 1830–1880. In: Proceedings of the Fifth European
social science history conference, Berlin.
Segal, M. R. (1988). Regression trees for censored data. Biometrics, 44, 35–47.
SPSS (Ed.). (2001). Answer tree 3.0 user’s guide. Chicago: SPSS Inc.
Szreter, S. (1993). The idea of demographic transition: A critical intellectual history. Population
and Development Review, 19, 659–701.
Therneau, T. M., & Grambsch, P. M. (2000). Modeling survival data. New York: Springer.
Vallin, J. (2001). Populations et individus. In: G. Caselli, J. Vallin & G. Wunsch (Eds),
Demographie: Analyse et synthese, La dynamique des populations (Vol. I, pp. 9–12). Paris:
Institut national d’etudes demographiques.
Van de Kaa, D. J. (1996). Anchored narratives: The story and findings of half a century of
research into the determinants of fertility. Population Studies, 50(3), 389–432.
Van Imhoff, E., Kuijsten, A., Hooimeijer, P., & van Wissen, L. J. G. (1995). Household de-
mography and household modeling. New York: Plenum Press.
Vaupel, J. W., Manton, K. G., & Stallard, E. (1979). The impact of heterogeneity in individual
frailty on the dynamics of mortality. Demography, 16, 439–454.
Veron, J. (1993). Arithmetique de l’homme: la demographie entre science et politique. Paris:
Editions du Seuil.
Willekens, F. J. (1999). The life course: Models and analysis. In: L. J. G. van Wissen &
P. Dykstra (Eds), Population issues: An interdisciplinary focus (pp. 23–51). New York:
Plenum Press.
Yamaguchi, K. (1991). Event history analysis. ASRM 28, Newbury Park, London: Sage.
Zighed, D. A., & Rakotomalala, R. (1996). SIPINA-W(c) for Windows. User’s guide, Lab-
oratory ERIC – University of Lyon 2, Lyon.
GILBERT RITSCHARD AND MICHEL ORIS314
FIVE STEPS IN LATENT CURVE
MODELING WITH LONGITUDINAL
LIFE-SPAN DATA
John J. McArdle
INTRODUCTION
During the past few decades, scientists from many disciplines have been
engaged in the study of human development over what is termed the ‘‘life-
span.’’ Of course, this term has early roots in the physical and biological
sciences. Life-span developmental research in psychology continues to pro-
mote the combination of (1) a paradigm shift in thinking about human
development as a process occurring over all ages in different ways, and (2)
the utility of a multivariate perspective on data collection and analysis (e.g.,
Baltes, Lindenberger, & Staudinger, 1998). The theoretical underpinnings
of life-span developmental psychology are outlined by Abeles (1987, p. 3):
Development is (1) a life-long process, (2)multi-dimensional, (3)multi-directional,
and (4) multi-determined. These kind of meta-theoretical principles have
become so embedded in the fiber of modern developmental research that
they are now hardly ever debated or even discussed. Instead, the current
issues focus on the best ways to formalize these ideas into practical multi-
variable research. The life-span research presented in this paper attempts to
add to this practice using contemporary statistical modeling analyses.
Towards an Interdisciplinary Perspective on the Life Course
Advances in Life Course Research, Volume 10, 315–357
Copyright r 2005 by Elsevier Ltd.
All rights of reproduction in any form reserved
ISSN: 1040-2608/doi:10.1016/S1040-2608(05)10012-4
315
In some respects, the life-span view of developmental psychology is rel-
atively new (for overview, see Baltes & Nesselroade, 1979). This area
emerged as a separate discipline in the early to mid-1970s with contribu-
tions of a few key researchers. The Department of Psychology at West
Virginia University was a fertile source of much important work – at one
time, Warner Schaie was the head of this department, and Paul Baltes and
John Nesselroade were working together in it, first as assistant professors,
and then as associate professors. About a decade earlier, Schaie began his
longitudinal research by following the people from Paul Horst’s multivari-
ate study at the University of Washington (later termed the ‘‘Seattle
Longitudinal Study’’). Baltes, then a promising student of Ernst Boesch
(Saarbruecken), a Piaget Ph.D., was well-versed in the elegant theoreti-
cal work of Klaus Riegel, the founder of modern dialectic psychology.
Nesselroade was then a promising student of Raymond Cattell, the founder
of modern multivariate psychology and a major contributor to the fields of
cognition and personality.
This group stimulated debate and controversy in research on the sepa-
ration of Age (A) effects, Cohort (C) effects, and Time-Period (T) effects.
Schaie (1965) promoted a formal separation of ACT effects in and ANOVA
fashion using what he termed the ‘‘general developmental model.’’ Perhaps
surprisingly, Baltes was one of the first critics of the ACT approach (Baltes,
1968) – he suggested not to use Schaie’s model for explanatory purposes but
descriptive ones only. In this spirit, Baltes shifted the emphasis to analyses
of cohort-sequential and longitudinal sequences (see Schaie & Baltes, 1975).
Soon afterwards, Schaie’s work was debated vigorously by Cattell and his
students: e.g., ‘‘Rarely has such a Tower of Babel sprung up so quickly as in
this area of research.’’ (Cattell, 1970, p. 154; also Horn & Donaldson, 1976).
Of course, Cattell himself had investigated complex multivariate designs
before (e.g., Cattell, 1963, 1969), but it was Nesselroade who advocated the
inclusion of multivariate measurements, and he was soon active in the de-
bate as well (e.g., Baltes & Nesselroade, 1970).
As it turned out, these controversies on ACT methods were going on in
other areas of research, especially Sociology, and stimulated many others
and led to some of the most important contributions to life-span research.
There were immediate practical applications such as the ‘‘longitudinal and
cross-sectional sequences’’ in research by Baltes and Nesselroade (1972) on
West Virginia Junior High and High School students. These activities also
stimulated an important series of books on ‘‘life-span developmental
psychology’’ focusing on the integration of substance and method (e.g.,
Goulet & Baltes, 1970; Baltes, Reese, & Nesselroade, 1977; Baltes &
JOHN J. MCARDLE316
Nesselroade, 1979; Baltes, 1983). A large number of psychologists soon
followed the active lead of this small group from Morgantown, West
Virginia (afterwards at Penn State, now at the University of Virginia), and
the result became known as the ‘‘life-span movement in psychology’’ (see
Nesselroade, 2001).
The ACT controversy that stimulated this life-span research could be
considered just a natural extension of previous longitudinal considerations
(e.g., Wohlwill, 1973; Bell, 1954; McArdle & Bell, 2000). Also, subsequent
statistical studies proved that the general ACT problem was ‘‘not identified’’
and unique solutions were only possible using highly restricted models (for
review, see Horn & McArdle, 1980; McArdle & Anderson, 1990; Donaldson
& Horn, 1992). But, in retrospect, the survival and popularity of the life-
span movement can now be seen due in large part to a successful integration
of developmental theory and multivariate method. Clearly, the substantive
thinking was enhanced by the methodological thinking, and vice versa.
Most of these methodological debates concluded that the collection of
longitudinal data are a necessary ingredient for life-span research, and many
researchers have defined these issues in extensive detail, but most have
emphasized ‘‘the explanation of inter-individual differences (or similari-
ties) in intra-individual change patterns’’ (e.g., Wohlwill, 1973; Baltes &
Nesselroade, 1979). During the last decade, many methodologists have
added to the current knowledge base, and classic models for ‘‘growth-curve
analysis’’ seems to have been revived as an important research technique
(e.g., see Rogosa & Willett, 1985; McArdle & Epstein, 1987). In a general
sense, the term growth-curve analysis denotes the processes of describing,
testing hypotheses, and making scientific inferences about the growth and
change patterns in a wide range of time-related phenomena. More specif-
ically, growth curves are longitudinal data with special properties of repeated
measures. Thus, in both theory and practice, the term ‘‘growth curve’’ is not
limited to the phases of life, where the organism ‘‘grows,’’ but it can also be
used to describe and analyze phases of life, where the organism ‘‘declines.’’
The phenomena that are studied here both ‘‘grow and decline over age,’’ and
these comprehensive features lead to an unusual set of opportunities for
developmental analyses of data.
This paper describes a current application of one class of longitudinal
growth-curve analysis models – latent curve modeling (LCM) using struc-
tural equation modeling (SEM) techniques. The analytic techniques of this
chapter are presented in the next five sections. These are presented in a
sequence of difficulty, as ‘‘steps’’ of developmental data analysis. Step 1:
Describing the Observed and Unobserved Longitudinal Data. I consider some
Five Steps in Latent Curve Modeling with Longitudinal Life-Span Data 317
useful ways to summarize longitudinal data, including statistical informa-
tion from both the complete and incomplete cases. Step 2: Characterizing the
Developmental Shape of Both Individual and Groups. I try to describe both
the group and individual characteristics of the longitudinal data. We dem-
onstrate how the SEM approach is generally easy and flexible. Step 3: Ex-
amining the Predictors of Individual and Group Differences in Developmental
Shapes. We recognize that individual differences in growth may be the result
of combinations of other measured variables. I describe how SEM can be
used in a ‘‘multi-level’’ form, can be extended to include concepts from
latent path analysis, and can provide empirical evidence for hypotheses
about the precursors or correlates of individual longitudinal patterns. Step
4: Examining Group Differences in Developmental Shapes. Individual differ-
ences in growth may be the result of combinations of different patterns of
change for persons in different groups. To study these possibilities, I de-
scribe how SEM can be used in a ‘‘multi-group’’ form, can be extended to
include concepts from ‘‘latent mixture’’ models, and can provide empirical
evidence for hypotheses about the heterogeneity in longitudinal growth
patterns. Step 5: Studying Dynamic Determinants Among Variables Over
Time. I show how the time-dependent nature of the latent variables can be
represented in SEM and used to study ‘‘lead–lag’’ relations using simple
dynamic expressions.
As an illustration within all five steps, I present alternative analyses of
a longitudinal data collection on intellectual abilities over the life span
from the Bradway–McArdle Longitudinal Study (McArdle & Hamagami,
1996, 2004; McArdle, Hamagami, Meredith, & Bradway, 2001). In
Table 1, I outline the multiple phases of data collection on intellectual
abilities measured over a large part of the life span – at seven occasions
between ages 2 and 72. Some of the data collected appear in the plots of
Fig. 1 and will be described in the next section. Analyses of these life-
span data have been reported in more detail in other recent papers, so
here I mainly summarize these results and highlight the life-span features
of prior analyses. These illustrations are used to convey the main pre-
sumptions and techniques as well as the benefits and limitations involved
in using this approach in developmental research. Mathematical and sta-
tistical issues are not presented in detail, and formal equations are not
presented. This is an overview of the general data analysis methodology
to demonstrate the flexibility of these methods for life-span developmental
research.
JOHN J. MCARDLE318
Table 1. Overview of the Bradway–McArdle Longitudinal Data at
Seven Time Points (N ¼ 111).
Date Sample Size Age Range Measures Investigators
1931 212 2–7 SB RT+, QM
1941 138 12–17 SB+ KB
1956 111 27–32 SB, WAIS+ KB
1969 48 40–45 SB, WAIS+ JK+, KB
1984 54 55–60 WAIS, WJ-R+ JM+, KB
1992 51 61–66 WAIS, WJ-R+ JM+, KB
1998 32 67–72 WAIS, WJ-R+ JM+, KB
Notes:
1. Measurement batteries are the Stanford–Binet (SB), Wechsler Adult Intelligence Scale
(WAIS), and Woodcock–Johnson Revised (WJ-R). (+) indicates that additional demographic
and/or psychometric measurements were included.
2. Study Investigators were Robert L. Thorndike (RT), Quinn McNemar (QM), Katherine
P. Bradway (KB), Jon Kangas (JK), and John J. McArdle (JM).
Fig. 1. A Plot of Individual Growth Curves for the Bradway–McArdle Longitu-
dinal Data (N ¼ 111).
Five Steps in Latent Curve Modeling with Longitudinal Life-Span Data 319
STEP 1: DESCRIBING THE OBSERVED AND
UNOBSERVED LONGITUDINAL DATA
The first step in any useful data analysis is an adequate description of the
data. The collection of longitudinal data can be extremely difficult, espe-
cially over large portions of the life-span, so the unique aspects of these data
should be emphasized.
The Bradway–McArdle Longitudinal Study
The data examined here come from persons, who were first measured in
1931 when they were aged 2–7 as part of the larger standardization sample
of the Stanford–Binet test (N ¼ 212). Many of these persons who could
be found (N ¼ 138) were measured again about 10 years later by Katherine
P. Bradway as part of her doctoral dissertation in 1944. Many of these same
persons were measured twice more by Bradway as adults at average ages of
30 and 42 using the Wechsler Adult Intelligence Scales (WAIS) (N ¼ 111;
for further details, see Bradway & Thompson, 1962; Kangas & Bradway,
1971). In 1984 and 1992, at average ages 57 and 65, the current author,
working with Bradway, measured as many of these same subjects as we
could locate (see McArdle & Hamagami, 1996; McArdle et al., 2001). About
half (N ¼ 55) of the adolescents tested in 1944 were measured again in 1984
at ages 55–57 and in 1992 at ages ranging from 64 to 72 (McArdle et al.,
2001). The seventh testing of the same individuals in 1998 yielded retests on
most of the same people (N ¼ 32), and attrition from the study was com-
plicated by the advancing age of the persons (e.g., attrition due to fatigue,
illness, and death). Although many other researchers were involved in data
collection (see Table 1), we now term this whole collection the Bradway–
McArdle Longitudinal study.
Up to now, our published analyses have focused only the first six time
points of longitudinal data. Fig. 1 is a display of individual growth curve
data for this measure at each age-at-testing for N ¼ 111 individuals. The
y-axis indexes the scores on ‘‘Verbal-Knowledge’’ and the x-axis is an index
of the age-at-testing. The connected lines in this figure are graphic descrip-
tions of the pattern of Verbal-Knowledge scores for each of the individuals,
so each line is termed a growth curves or trajectory. This kind of plot allows
us to see some overall trends of rapid rises in early childhood and adoles-
cence to long periods of consistency at older ages. There are also a hint of
JOHN J. MCARDLE320
patterns of growth and change for some individuals, and this interesting
possibility requires a rigorous examination.
The analysis of growth and change in similar constructs spanning child-
hood to adulthood still represents a major challenge for research over the
life span. It is easy to connect the data for a person over many time points of
testing, but it is much harder to insure that these scores meet the measure-
ment requirements of ‘‘exactly repeated measures.’’ Growth-curve analyses
typically require: (a) the same entities are repeatedly observed, (b) the same
procedures of measurement and scaling of observations are used, and (c) the
timing of the observations is known. In prior psychometric research, we
have used both scale-level factor analysis and Item Response Theory (IRT)
models to create comparable measurements over time. Composite scores
from the early Stanford–Binet tests (at ages of 4, 14, 30, and 42), as well as
the WAIS tests (at average ages 30, 42, 56, and 64) were used to form a
Verbal-Knowledge (or Crystallized Intelligence gc) score. These scores are in
a comparable metric, because they were based on a Rasch-scaling of selected
items in each test (see Hamagami, 1998; McArdle & Nesselroade, 2003;
McArdle et al., 2004).
Describing the Observed Data
The sample sizes, means, standard deviations, and correlations of these
IRT-scaled measures over six occasions are listed in Table 2. The overall
subject participation shows a nearly continual loss of participants over the
60 years. This is not an unusual loss of participants, even for a much smaller
period of time. The means and standard deviations show the simple pattern
described earlier, with rapid rises in childhood and adolescence together
with very little growth or decline in adulthood. The correlations over time,
the unique statistical information of the longitudinal data, present a com-
plex pattern of results, some correlations suggesting high stability of indi-
vidual differences (e.g., r40:9) and others suggesting low long-term stability
(ro0.1).
Fig. 2 is a scatter-plot matrix of the raw data at all occasions. In this
figure, the frequency histograms are placed on the main diagonal and the
pair-wise scatter plots are placed on the off-diagonals. These scatter plots of
growth data provide another way of describing the relative mobility of
cognition in these people over time. The spread of points is rather narrow
when the time interval is shortest (i.e., at consecutive measurements), and
the spread of points is much wider when the observations are most further
Five Steps in Latent Curve Modeling with Longitudinal Life-Span Data 321
apart in time (i.e., from childhood at age 4 to adulthood at ages 42–65). The
correlation matrix is a simple summary of these features, but the plots are
uniquely informative about the patterns of the people.
Table 2. Observed and Unobserved Summary Statistics based on the
Longitudinal Data at Six Time Points (N ¼ 111;MLE-MAR in brackets;
Step 1, see Figs. 1 and 2).
(a) Observed means, standard deviations, and ranges
N Mean [MLE] SD [MLE] Min Max
Age 04 110 22.1 [22.2] 9.8 [9.6] 0.0 37.6
Age 14 111 72.2 [72.2] 9.3 [9.3] 40.6 90.2
Age 29 110 84.4 [84.4] 7.0 [6.9] 58.2 95.7
Age 42 49 87.9 [87.2] 6.7 [6.8] 69.2 100.0
Age 57 51 88.9 [87.7] 5.8 [7.0] 73.8 99.3
Age 65 51 88.1 [87.1] 7.1 [7.1] 68.7 98.6
(b)Observed and unobserved correlations (each entry includes pairwise r and [MLE-MAR r])
Age 04 Age 14 Age 29 Age 42 Age 57 Age 65
Age 04 1.000
Age 14 0.553 1.000
[0.558]
Age 29 0.233 0.680 1.000
[0.224] [0.679]
Age 42 0.194 0.377 0.812 1.000
[0.143] [0.412] [0.787]
Age 57 0.304 0.489 0.800 0.798 1.000
[0.211] [0.574] [0.858] [0.843]
Age 65 0.079 0.472 0.759 0.663 0.911 1.000
[0.171] [0.579] [0.791] [0.744] [0.946]
(c) Patterns of complete (x) and incomplete (o) data
1 2 3 4 5 6 7 8 9
Age 04 x x x x x x x x x
Age 14 x x x x x x x x x
Age 29 x x x x x x x x o
Age 42 o x o x o x o x x
Age 57 o x o o x x x o x
Age 65 o x x o o o x x x
Frequency 36 29 12 9 8 7 6 3 1
JOHN J. MCARDLE322
Results from Dealing with Incomplete Information
The summary information presented in Table 2 is not limited to only those
participants with complete data at all six time points of measurement. As is
usual, we can represent all available information about the observed means and
variances and the ‘‘pair-wise’’ correlations. To deal with this problem, I present
a description of the patterns of complete and incomplete data in Table 2c. In
these data, there are only nine different patterns of incomplete data (out of a
possible set of 70 patterns), and most of the persons are either measured at all
six times (n ¼ 29) or only the first three times (n ¼ 36). These incomplete data
patterns can be represented as the ‘‘percentage of data’’ or ‘‘coverage’’ for each
covariance of these scores – in some cases 100% of the participants are used but
in some cases only 29.7% of the information is available (at ages 42 and 65).
In Tables 2a and b, I also use brackets to list an ‘‘incomplete data’’
estimate of the sample means, standard deviations, and correlations. These
estimates are based on what is termed maximum likelihood estimation under
GC
_A
GE
04
GC
_A
GE
14
GC
_A
GE
29
GC
_A
GE
42
GC
_A
GE
57
GC_AGE04
GC
_A
GE
65
GC_AGE14 GC_AGE29 GC_AGE42 GC_AGE57 GC_AGE65
Fig. 2. Relationships among the Scores of the Six ‘‘General Knowledge’’ (Gc)
Variables Over Time.
Five Steps in Latent Curve Modeling with Longitudinal Life-Span Data 323
missing at random assumptions (MLE-MAR; Little, 1995; McArdle, 1994;
Cnaan, Laird, & Slasor, 1997). This approach allows us to examine these
summary statistics ‘‘as if all persons were measured at all occasions.’’ These
newly estimated statistics are fairly close to the pair-wise estimates and this
indicates these data meet the minimal conditions of ‘‘missing at random’’
(MAR; Little, 1995). Most importantly, these estimated statistics do not
suffer from some common statistical problems (local linear dependency),
and we can generally use all available information from every person – that
is, we do not need to select a subset of persons with complete data (n ¼ 29)
out of all those measured (N ¼ 111).
Dealing with incomplete data as unobserved but important scores can alter
inferences about growth and change. For example, from Table 2b, the pair-
wise correlation of General Knowledge from age 4 to 42 is only r ¼ 0:19; andfrom age 4 to 65 only r ¼ 0:08; and these are certainly not as high as com-
parable correlations recently reported elsewhere (e.g., r47 in Deary, Whalley,
Lemmon, Crawford, & Starr, 2000). It would be reasonable to critique our
lower over-time correlations as being underestimated due to the loss of par-
ticipants (i.e., selective attrition). This concern leads us to try to account for
the patterns of incomplete data. It is interesting that when we do use con-
temporary technique to account for attrition, we obtain similar results – the
MLE-MAR-estimated correlation of General Knowledge from age 4 to 42 is
only r ¼ 0:14; and from age 4 to 65 only r ¼ 0:17 (from Table 2b). This new
result implies these low correlations over time may be due to initial selection but
are probably not attributable to subsequent attrition. This description of un-
observed statistics is an essential part of all contemporary growth models.
STEP 2: CHARACTERIZING DEVELOPMENTAL
SHAPES FOR GROUPS AND INDIVIDUALS
The second step in a useful data analysis is the attempt to highlight the key
features of the data in terms of a model. In contemporary behavioral science
research, one common approach to growth-curve analysis is to write a tra-
jectory equation for each group and individual. Some mathematical and
statistical aspects of these kinds of models are described next.
Linear Growth Models for Repeated Measures
Most forms of growth-curve analyses require longitudinal data with
repeated measurements where we observe the variable (Y) at multiple
JOHN J. MCARDLE324
occasions (t ¼ 1 to T) on the same person (n ¼ 1 to N) and we can symbolize
the scores (as Y ½t�n). This model includes three unobserved or latent scores
representing: (1) the individual’s initial level (y0), (2) the slopes representing
the individual linear change over time (y1), and (3) the independent errors of
measurements (e[t]). To indicate the form of the systematic change, we use a
set of group coefficients or basis weights, which define the timing or shape of
the trajectory over time (e.g., Age½t� ¼ t; or as A½t� ¼ AgeðtÞ). It is typical to
estimate the fixed group means for intercept and slopes (m0, m1) and also the
implied random variance and covariance terms (s02, s1
2, s01) describing the
distribution of individual deviations (d0n, d1n) around those means. We also
usually assume that there is only one random error variance (s2e), and the
error terms are assumed to be normally distributed and presumably uncor-
related with all other components.
The path diagram of Fig. 3 is an exact translation of the necessary matrix
algebra of these models. These diagrams can be conceptually useful devices
for understanding the basic modeling concepts. They are also practically
y0
d0
ys
ds
1
α [1]=0.40.4
α [2]=5.0
a[3]=6.2
a[4]=6.46.4α [5]=6.5
α [6]=6.5
σ0=9.89.8 σs=1.51.5
µs=10.710.7
0s� � = 0.77= 0.77
µ0=17.917.9
Y1
e1
Y2
e2
V3 Y4 V5 Y6
σe=17.417.4
Y5Y[1]
e[1]
Y[2]
e[2]
Y[4]
e[5]
Y[6]
e[6]
Y[3] Y[5]
17.417.4 17.417.4 17.417.4 17.417.4 17.417.4
e[3] e[4]
Fig. 3. A Path Diagram of Numerical Results from a Latent Basis Growth Model
from the Bradway–McArdle Longitudinal Data.
Five Steps in Latent Curve Modeling with Longitudinal Life-Span Data 325
useful because they can be used to represent the input and output of any of
the SEM computer programs. These diagrams were originally used with
variations on multiple regression and factor analysis models, but contem-
porary work has shown how these diagrams can be used in the context of
growth and change (e.g., McArdle, 1986; McArdle & Epstein, 1987;
McArdle & Anderson, 1990; McArdle & Woodcock, 1997). In this path
diagram, the observed variables are drawn as squares, the unobserved var-
iables are drawn as circles, and the required constant is included as a tri-
angle. Model parameters representing ‘‘fixed’’ or ‘‘group’’ coefficients are
drawn as one-headed arrows while ‘‘random’’ or ‘‘individual’’ features are
drawn as two-headed arrows. In this case, the initial level and slopes are
often assumed to be random variables with ‘‘fixed’’ means (m0, m1) but
‘‘random’’ variances (s02, s1
2) and correlations (r01).
Basic Linear Growth Models
Some initial growth-curve modeling results for the Bradway–McArdle data
are presented in Table 3. In these longitudinal models, any change score (y1)
is assumed to be constant within an individual but it is not assumed to be the
same between individuals. We do not usually estimate the unobserved scores
but we do estimate several parameters, which characterize the key features
of these unobserved scores.
The first model labeled 3a is no-growth model fitted with only three pa-
rameters: An initial level mean (m0 ¼ 68:6), an initial standard deviation (s0o 0.01), and an error variance (s2e ¼ 740). The parameters also yields a
model likelihood (L2 ¼ �2276), which shows the no-growth baseline is a
poor fit compared to the totally unrestricted model ðw2 ¼ 1510; df ¼ 24).
This model is typically just used as a baseline against which to judge the fit
of more informative models.
The second linear growth model labeled 3b uses a fixed set of basis co-
efficients formed by taking A½t� ¼ ½Age½t�=10� or fixed values (e.g.,
A½t� ¼ ½0:4; 1:4; 2:9; 4:2; 5:7; 6:5�). This linear scaling is only one of many
that could be used, but it was chosen to permit a practical interpretation of
the slope parameters in terms of a per-decade change. In contrast to the no-
growth baseline, this linear growth model also has three more free param-
eters: a slope mean (m1), standard deviation (s1), and correlation (r01). This
model fitted yields a new likelihood (L2 ¼ �2092), which is a large distance
from the unrestricted model (w2 ¼ 1142 on df ¼ 3) but is an improvement
in fit over the previous baseline (M1 vs. M0: Dw2 ¼ 368 on Ddf ¼ 3).
JOHN J. MCARDLE326
The resulting means describe a function that starts low at age 4 (m0 ¼ 41:8)but increases between ages 4 and 65 (by m1 ¼ 9:5 per decade). The variance
estimates of the intercept and slope parameters are too small to interpret (sjo 0.02), but the error variance has been reduced (from s2efM0g ¼ 740 to
s2efM1g ¼ 345). Of course, the main problem is that simple straight-line
linear growth model does not fit these data very well.
Table 3. Selected Results from Five Latent Growth Models Fitted to
Longitudinal Data at Six Time Points (N ¼ 111; Step 2, see Figs. 3–5).
Parameters 3a (Level) 3b (Linear) 3c (Quadratic) 3d (Latent) 3e (Spline)
Fixed effects
Basis a[04] 1, 0 1, 0.4 1, 0.4, 120.42 1, 0.4 1, 0, 0
Basis a[14] 1, 0 1, 1.4 1, 1.4, 121.42 1, 5.0 (0.6) 1, 1, 0
Basis a[29] 1, 0 1, 2.9 1, 2.9, 122.92 1, 6.2 (0.07) 1, 1, 0.80
(0.03)
Basis a[42] 1, 0 1, 4.2 1, 4.2, 124.22 1, 6.4 (0.08) 1, 1, 0.98
(0.04)
Basis a[57] 1, 0 1, 5.7 1, 5.7, 125.72 1, 6.5 (0.08) 1, 1, 1.01
(0.04)
Basis a[65] 1, 0 1, 6.5 1, 6.5, 126.52 1, 6.5 1, 1, 1
Level m0 68.6 (1.2) 41.8 (1.4) 47.8 (?) 17.9 (1.0) 22.2 (0.94)
Slope m1 — 9.5 (0.40) 0.42 (?) 10.7 (0.19) 50.0 (0.86)
Accelerate ma — — 0.12 (?) — 15.1 (0.86)
Random effects
Error se2 740 (54) 345 (28) 41.3 (?) 17.4 (1.5) 8.3 (0.95)
Level s0 o0.01 (?) o0.01 (?) 22.3 (?) 9.8 (0.79) 9.4 (0.69)
Slope s1 — o0.01 (?) 29.3 (?) 1.5 (0.14) 8.1 (0.69)
Correlation
r01
— 40.99 (?) �0.77 (?) �0.77 (0.05) �0.50 (0.08)
Accelerate sa — — 15.9 (?) — 7.1 (0.66)
Correlation
r0a, rsa
— — 0.33 (?), �0.84
(?)
— �0.60 (0.08),
�0.04 (0.12)
Fit indices
Numbers of
parameters
3 6 10 10 13
Degree of
freedom
24 21 17 17 14
Log likelihood �2276 �2092 �1830 �1583 �1507
Likelihood
ratio (DLL)
1510 41142 4617 122 46
RMSEA 0.75 0.69 0.56 0.23 0.14
Note: ‘?’ freely estimated from data.
Five Steps in Latent Curve Modeling with Longitudinal Life-Span Data 327
Including Polynomial Nonlinearity in Growth
The data of Fig. 1 suggest some nonlinearity over age in the scores of almost
all persons. To deal with this complexity, Wishart (1938) introduced the use
of power polynomials to better fit the curvature apparent in growth data.
The individual growth curve (consisting of t ¼ 1; T occasions) is summa-
rized into a small set of linear orthogonal polynomial coefficients based on a
fixed power-series of time (A[t], A[t]2, A[t]3,y, A[t]p) describing the general
nonlinear shape of the growth curve. A second-order (quadratic) polyno-
mial growth model can be introduced using a new set of component scores
(e.g., yp) introduced to represent another level of change with a set of pow-
ered coefficients (1=pA½t�p) . This leads to an implied change model, where
the change is linear with time (i.e., acceleration). Of course, a model of
growth data might require this form of a second-order (quadratic), third-
order (cubic), or even higher-order polynomial model fitted to the data. In
all cases, additional variance and covariance terms are added to account for
individual differences in latent scores. This polynomial growth curve ap-
proach remains popular (e.g., Bryk & Raudenbush, 1992).
A quadratic polynomial model 3c was fitted next, including an additional
slope variable (acceleration) defined by a fixed basis ð12A½t�2Þ; one new mean,
one new deviation, and two additional correlations. This 10 parameter
model is still far worse than the unrestricted model (w24617 on df ¼ 17),
but is an improved fit compared to the linear model (M2 vs. M1: Dw2 ¼ 524
on Ddf ¼ 3). The error or unique variance has also been reduced substan-
tially (from s2efM2g ¼ 41:3) but problems arose in the variance–covariance
estimation (sj o 0.02), numerical convergence was not achieved, and this
flexible polynomial model seems to inappropriate for these data.
Nonlinearity Using Latent Basis Curves
A different alternative of the linear growth model was proposed by Rao
(1958) and Tucker (1958, 1966) in the form of summations of ‘‘latent
curves’’ (see Meredith & Tisak, 1990). The use of this latent growth curve
offers a relatively simple way to investigate the shape of a growth curve – we
allow the curve basis to take on a form based on the empirical data. In this
approach, we write a model for the same person at different occasions but
some of the basis coefficients (e.g., a[2], a[3], a[4], and a[5]) are free to be
estimated. That is, the actual ages of the persons are known but the basis
parameters are allowed to be freely estimated, and we end up with an op-
timal shape for the group curve and individual differences. In this model, the
JOHN J. MCARDLE328
A[t] is estimated just like a common factor loading and thus has mathe-
matical and statistical identification problems. There are many alternative
ways to estimate these parameters but, in general, there are many free pa-
rameters in the two-component growth model (p ¼ 6þ T � 1). The use of
an estimated basis has been termed a ‘‘meta-meter’’ or ‘‘latent time’’ scale
that can be plotted against the actual age curve for visual interpretation (see
McArdle & Epstein, 1987).
The fourth model fitted 3d was a latent basis growth model where some of
the loadings, A[t], are free to vary. For the purposes of estimation, we fixed
A½1� ¼ 0:4 (at age ¼ 4) and A½65� ¼ 6:5 (at age ¼ 65), but the four other
coefficients were estimated from the data. This results in a large improvement
in the model likelihood (L2 ¼ �1583), which is much closer to the unrestricted
model (w2 ¼ 122 on df ¼ 17), and substantially better than the nested baseline
model (Dw2 ¼ 1386 on Ddf ¼ 7), the nested linear model (Dw2 ¼ 1018 on
Ddf ¼ 4), and better than the non-nested quadratic model with the same
numbers of parameters (�1830 vs. �1583). The error variance has also been
reduced (s2efM3g ¼ 17:4). The interpretation of the model parameters is an
important part of the latent growth analysis, and these are displayed in Figs. 3
and 4. The estimated latent means are m0 ¼ 17:9 and m1 ¼ 10:7; their devi-
ations are s0 ¼ 9:8 and s1 ¼ 1:5; and the two latent factors have correlation
of r01 ¼ �0:77: These values were included on the path diagram of Fig. 3. The
estimated basis coefficients were A½t� ¼ ½0:4; 5:0; 6:2; 6:4; 6:5; 6:5�; and these
are repeated in Fig. 4 also (and discussed later).
Nonlinearity Using Linear Spline Models
Another way to deal with nonlinearity is by introducing the concepts of
connected lines or splines (Seber & Wild, 1989). As a simple example here, we
could write a model for the same person at different occasions where we
define some critical age (i.e., C ¼ 30 years), and estimate two slope scores –
y1 before the age C and y2 after age C. This implies the estimated parameters
for the means and covariances of the intercept are now ‘‘re-centered’’ at age
C, and the slopes represent changes before and after age C. These models
can be combined with the polynomial models, and the critical cutoff ages
(Cn) can be estimated as well and these concepts can create a potentially
informative look at individual segments of growth (e.g., Cudeck & du Toit,
2001; Hamagami & McArdle, 2001).
This leads us naturally to the idea for the final model 3e, which is a spline
model composed of two latent slopes – the first representing changes up
to age 14 and the second representing changes after age 14. The exact
Five Steps in Latent Curve Modeling with Longitudinal Life-Span Data 329
Solid line = Exponential growth
model
(no decline in later ages)
Dashed line = Latent Basis model
(allowing decline at any age)
Fig. 4. Latent Growth Score Expectations for the General Knowledge Scores. (a) Group Expectations from Two Models.
(b) Individual Expectations from the Final Model.
JOHN
J.M
CARDLE
330
parameterization we used yielded a starting level of 22.2 at age 4 with a big
average gain jump of 50 points (79.4) up to age 14, followed by a small
steady gain of 15.1 points (78.1) from 14 to 65. However, this two latent
splines should not be ignored, because it shows much better fit overall
(w2 ¼ 46 on df ¼ 14) and when compared to the nested one-slope model
(M6 vs. M4: Dw2 ¼ 76 on Ddf ¼ 3) is notable.
Model-Based Expectations about Growth
In the models just presented, the changes within a person is initially rep-
resented by the latent means and variance terms in the growth models. In
later interpretations, we can examine the relative size of these parameters
and make substantive interpretations about the group and individual dif-
ferences. These parameters also allow us to form the expected growth curve
charts for both the observed and true scores. Details of derivations are not
presented here, but it is still useful to consider some key properties of the
model (see McArdle, 1986, 1989; McArdle & Woodcock, 1997). These pa-
rameters can be combined to create expected means (my½t� ¼ m0 þ m1A½t�).
Here, we obtain the expected group curve at mean scores of m½t� ¼
½22:2; 72:1; 84:4; 86:6; 87:6; 87:7�: These are plotted in Fig. 5a. This is a
growth curve with a shape that rises quickly between ages 4 and 14, peaks at
age 42, and exhibits no declines by age 65. All coefficients can be interpreted
in terms of changes over decades, so the average 10-year change is 10.7
points, but a person who is 14 years old can be said to have the latent age of
a 50 year old (i.e., A½14� ¼ 5:0). The individual differences in this model are
seen in the variances for the level (s20 ¼ 9:52) and the slope (s21 ¼ 1:52) pa-rameters. If we consider 100% of the changes between ages 4 and 65 (the
observed period), we now say that 76% of all cognitive growth in knowledge
occurred by age 14, 95% by age 42, and so on. The highest estimated
coefficient never was higher than the fixed value at age 65 – this is important
because it means the Verbal-Knowledge ability does not decline within
people repeatedly measured over this span of ages. This is not the typical
finding with cross-sectional age comparisons (e.g., McArdle et al., 2002).
Goodness-of-Fit, Expectations and Residuals Over Time
The choice between using a latent basis, or a polynomial, or spline or other
nonlinear model can be substantively important. The initial idea from
Wishart (1938) was that the basic shape of each individual curve could be
captured with a small number of fixed parameters and random variance
Five Steps in Latent Curve Modeling with Longitudinal Life-Span Data 331
Fig. 5. Individual Level Latent Growth Score Parameters and Residuals. (a) Individual Estimates of Latent Levels and
Slopes. (b) Individual Residuals of the Final Growth Model.
JOHN
J.M
CARDLE
332
components. In some cases, a fixed basis polynomial model is more par-
simonious than a free basis model. However, the polynomial model also has:
(a) a relatively fixed curvature and (b) requires an additional estimation of
covariances among the new latent scores (yp; this is true even if orthogonal
polynomial coefficients are used). Thus, the polynomial model may add
more complexity (via parameters to be estimated) than is actually needed
and the latent basis model may prove to be more efficient and interpretable.
Although many recent textbooks overlook the latent basis model (e.g.,
Duncan, T. E., Duncan, S. C., Strycker, Li, & Alpert, 1998; Singer &
Willett, 2003), we always treat this as an empirical choice (McArdle & Bell,
2000; McArdle & Nesselroade, 2003).
We do not usually need to make any specific decision about these model
differences at this point. It is enough to say that the two-part linear slope
model fits best, it is far better than the others considered, and it is far more
interesting than simply adding unique variances or covariances to improve
fit. However, the interpretation of the two-slope model is very similar to
what we have already described in the one-slope model, which changes
direction. The one latent slope model may need to be reconsidered and the
complexity of individual differences should not be overlooked.
There are many other models that could fit to these data, and others have
done so (e.g., Cudeck & du Toit, 2001; Hamagami & McArdle, 2001). In
such an enterprise it may be important to do a more detailed analysis of the
curve at the individual level. In Fig. 5a, we plot individual estimates of the
latent parameters for the level and slope (These are Bayes-restricted esti-
mates based on the latent model 3d). In this plot, many of the individuals
have estimated levels and slopes in large cluster, but other individuals are
scattered with negative slopes. The variation apparent here will be examined
in the next section. The second plot 5b is a description of the residuals
between the individual expected curve (based on the estimated level and
slope) and the observed data. Here it appears that at least one individual is
poorly represented (i.e., the large positive residual at age 45), and there is
much more misfit at age 14 than at any other point in time. These are aspects
of model fitting that need to be improved with subsequent analyses.
STEP 3: MODELING PREDICTORS OF INDIVIDUAL
DIFFERENCES IN DEVELOPMENTAL SCORES
The group mean parameters estimated in the previous analyses allow us
to plot the group trajectory over time. Similarly the estimated variance
Five Steps in Latent Curve Modeling with Longitudinal Life-Span Data 333
parameters allow us to consider the size of the between group differences at
each age. However, no prior information obtained in model fitting tells us
about the sources of this common variance. To further explore the differences
between persons we need to expand the basic latent growth model to include
impacts on the latent parameters.
The Multilevel Growth Model
Let us assume a variable termed X indicates some measurable difference
between persons (e.g., sex, educational level, etc.). If we measure this var-
iable at one occasion, we might like to examine its influence in the context of
a growth model for Y[t]. One popular model is based on the use of ‘‘ad-
justed’’ growth parameters as popularly represented in the techniques of the
analysis of covariance. In growth curve terms this model is written with fixed
(group) coefficients (g) with some effect on the measured Y[t] scores at each
occasions, and the X is an independent observed (or assigned) predictor
variable. In this case, the growth parameters (m0:x;m1:x; s0:x;s1:x;s0;1:x) areconditional on the expected values of the measured X variable. The pa-
rameters of the changes may be defined by the resulting difference or dif-
ferential equation, but the reduction of error variance (sex) is often
considered as a way to understand the overall impacts.
The apparent complexity of the covariance model leads to a simpler and
increasingly popular way to add an external variable – we can write a mixed-
or multi-level model, where the X variable has a direct effect on the param-
eters of the growth curve. This means we have intercepts (v) and regression
slopes (g) for the effect of X on the two latent components of Y[t]. This
model is drawn as a path diagram in Fig. 6. This diagram is the same as
Fig. 3 except that here, we have included several X variables as predictors of
the levels and slope components. It may be useful to write this as a reduced
form of SEM, so it is now clear that the three unobserved residual terms are
not simply separable by standard linear regression or covariance equation
(see McArdle & Hamagami, 1996).
This path diagram gives the basic idea of external variable models, several
other more complex alternatives will be considered in later sections. In this
simple linear model, as in more complex models to follow, we can always add
predictors X for the intercepts and the slopes. In some areas of research these
models have been termed mixed-effects models (Laird & Ware, 1982; Littell,
Miliken, Stoup, & Wolfinger, 1996; Singer, 1999). In other areas of research
these same models have been termed random-coefficients or multi-level
JOHN J. MCARDLE334
models, or slopes as outcomes or hierarchical linear models (e.g., Bryk &
Raudenbush, 1992). In other SEM research these models were considered
using factor analysis terminology as latent growth models with incomplete
data and extension variables (e.g., McArdle & Epstein, 1987; McArdle &
Hamagami, 1992). Using any terminology, these models can be generically
represented by the parameters in the path diagram of Fig. 6, and this is a
common way to understand the between group differences in within group
changes. Once considered in this way, no new model fitting procedure is
required.
Results for Educational Influences on Latent Scores
A variety of additional variables have been measured in the Bradway–
McArdle collections, including demographic (e.g., gender, educational at-
tainment by age 30 and 56, etc.), self reported health behaviors (e.g., smok-
ing, drinking, physical exercise, etc.) and other problems (e.g., general
Y1
e1
Y2
e2
V3
e3
Y4
e4
V5
e5
y0
e0*
ys
es*
Y6
e6
1
=0=0.76.76
.95.95.98.98 1.01.0 =1=1
9.39.3 8.98.963.263.2
2.12.1
−.53−.53
.02.02 1.341.34
Educ
Person
21.921.9
Y3Y5
Y[1]
e[1]
Y[2]
e[2]
V3
e[3]
Y[4]
e[4]
V5
e[5]
Y[6]
e[6]
Y[3] Y[5]
17.417.4 17.417.4 17.417.4
Educ
Dad
Educ
Mom.08.08 .25.25
8.82
17.417.417.417.417.417.4
Fig. 6. Results from Latent Growth With Mixed-effects or Multi-level Predictors
Latent Path Predictors.
Five Steps in Latent Curve Modeling with Longitudinal Life-Span Data 335
health, illness, medical procedures, etc.), and personality measures (e.g., 16
PF factors). In the analyses presented here we consider four variables: (1)
Gender (effect coded as –0.5 for males and +0.5 for females), (2) the per-
son’s own educational level (at age 30 in years), (3) Father’s educational
level (in years), and (4) Mother’s educational level (in years).
We started with a baseline latent growth model with no predictors of the
levels and slopes but then added additional variables as predictors of these
levels and slopes (as in Fig. 6). In these mixed-models, we only fit the latent
basis curve model (with a 0–1 basis), and we added Gender as an effect-
coded variable (i.e., M ¼ �0:5 and F ¼ þ0:5), and the three indices of
educational attainment centered at 12 years.
Table 4 is a list of results including these four variables and the six in-
tellectual ability measures. The first model 4a used all four variables and has
a misfit (w2 ¼ 153 on df ¼ 33), which is a reasonable fit when compared to
Table 4. Selected Results from Two Latent Growth Models with Four
Extension Variables Fit to the Longitudinal Data at Six Occasions (Step
3, see Fig. 7).
Parameters 4a: {Latent} Level 4a: {Latent} Slope 4b: {Latent} Level 4b: {Latent} Slope
Fixed effects
Basis A[t] [ ¼ 0, 0.76, 0.95, 0.98, 1.00, ¼ 1] [ ¼ 0, 0.76, 0.95, 0.98, 1.00, ¼ 1]
Intercept g0 22.8 (1.5) [0] 62.6 (1.7) [0] 23.3 (1.5) [0] 62.6 (1.6) [0]
Regression from
Gender gg
0.73 (2.0) [0.04] 1.63 (2.0) [0.09] 0 0
Regression from
Educ_Pers gp
0.20 (0.04) [0.048] 1.32 (0.47) [0.32] �0.13 (0.44) [0.00] 1.44 (0.45) [0.35]
Regression from
Educ_Dad gd
0.95 (0.47) [0.32] �0.68 (0.48)
[�0.23]
0.58 (0.38) [0.19] �0.24 (0.39)
[�0.08]
Regression from
Educ_Mom gm
�0.78 (0.54)
[�0.23]
0.65 (0.55) [0.19] 0 0
Random effects
Residual sd 9.0 (0.75) 8.7 (0.82) 9.1 (0.75) 8.8 (0.82)
Error se2 17.4 (1.5) 17.5 (1.5)
Correlation rd0, ds �0.81 (0.04) �0.81 (0.04)
Fit indices
Numbers of
parameters
32 28
Degree of freedom 33 37
Likelihood ratio 153 159
RMSEA 0.18 0.17
Note: Latent growth with all four predictors set to zero yields L2 ¼ 195 on df ¼ 41.
JOHN J. MCARDLE336
free latent curve model (M7 vs. M4: Dw2 ¼ 32 on Ddf ¼ 16). This implies
that the patterning of the correlations of predictors and outcomes can be
estimated using two latent variables (see McArdle & Prescott, 1992). The
MLE parameters suggest the following interpretations: 0 – The latent basis
coefficients (A[t]) were unaffected by the inclusion of the predictors.
1 – There are no accurate (significant) differences between Males and
Females on either levels or slopes. 2 – The Person’s Educational level (at age
30) was not predictive of the initial cognitive level (at age 4) but was a
positive indicator of overall changes in adulthood (g ¼ 1:32½0:32�). 3 – The
educational level of the Father (at age 14) was a positive contributor to the
initial cognitive level (g ¼ 0:95½0:47�) but not to the adult slope. 4 – The
educational level of the Mother (at age 14) was not related to the initial level
or slope.
The second set of results 4b was based on a model, where all coefficients
for the Gender and Mothers’ Education were fixed at zero. This model fits
the data just about as well as before (M8 vs. M7: Dw2 ¼ 6 on Ddf ¼ 4), but
the results obtained now differs in one respect – the coefficient for the
Father’s Education on initial level is no longer accurate. This is probably
due to the fact that the Mother’s Education (a) no longer has a direct effect,
(b) it is positively correlated (r ¼ 0:62) with Father’s Education, and (c) the
previous ‘‘suppression’’ effect has been eliminated and the coefficient is not
as strong. In essence, the previous interpretation (4a) needed to consider the
negative coefficient of Mother’s Education, and reinterpret the overall im-
pact or do so in terms of the difference between parental educational levels.
Upto this point, it seems the only direct educational influence found here
is that of the Person’s Educational level on their own slope. However, the
model can be expanded to include more complex path representations, and
this leads to a comprehensive series of mixed-model path equations. As-
suming the three education variables are centered at grade 12, and we view
the person’s educational attainment as an outcome of the parent’s educa-
tional attainment, we estimated a simultaneous latent path model where the
Mother’s Education (0.25) and Father’s Education (0.08) had direct effects
on the Person’s Education, but only the Person’s Education had an effect on
the latent level (0.20) and slope (1.34). This resulted in a good overall fit
(w2 ¼ 162 on df ¼ 39; so Dw2 ¼ 3 on Ddf ¼ 2), and emphasizes the positive
but indirect effects of the Mother (0.025� 1.34) on the adult slopes of their
children. This model is drawn as an SEM path diagram in Fig. 6 to illustrate
how the basic principles of multiple regression and path analysis follow
directly to the latent variables of a growth model (see McArdle, 2001;
McArdle & Hamagami, 2003).
Five Steps in Latent Curve Modeling with Longitudinal Life-Span Data 337
STEP 4: STUDIES OF GROUP DIFFERENCES IN
DEVELOPMENTAL SHAPES
The next step in longitudinal data analyses deals with the investigation of
group differences in developmental shapes or trajectories. While this is a
natural question raised in any multiple group SEM context, it is often
overlooked in this literature on mixed-effects or multi-level models. This
contrast between the two approaches was an issue raised by McArdle and
Epstein (1987) and McArdle (1990), and we now explore some alternatives.
Group Differences from a Multiple Group Perspective
One initial representation of group differences uses a new set of estimated
parameters to summarize between groups. This idea is clearly represented by
coding a set of variables (X) to characterizing the group differences and
then examining the effect of this set (X) on the model parameters (see, e.g.,
Fig. 6). However, this method is limiting in a number of important ways.
For example, some reasonable forms of growth processes are not immedi-
ately possible to account for within the standard framework. For example,
different groups of people could have different ‘‘amplitude’’ or be in dif-
ferent ‘‘phases’’ in their characteristic growth pattern. These features of
growth are not separated within the basic level and shape parameters al-
though these features may be realistic features of development.
An SEM treatment of this kind of a model uses concepts derived from
multiple-group factor analysis (e.g., Joreskog & Sorbom, 1979; McArdle &
Cattell, 1994). In these kinds of models, each group (g ¼ 1toG) is assumed to
follow a latent growth model where the basis parameters (A½t�ðgÞ) are definedby the application. Fig. 7 is a path diagram representing these kinds of
multiple group growth models (from McArdle & Hamagami, 1992). Since
the groups need to be independent (each person can only be in one group),
this kind of grouping is most easily done for discrete categorical variables
(i.e., sex, but not educational level).
In a most fundamental form, the multiple group growth model permits
the examination of the presumed invariance of the latent basis functions
(i.e., A[t](1) ¼ A[t](2) ¼ y ¼ A[t](g) ¼ y ¼ A[t](G)). The rejection of this
model implies that each independent group has a different shape of growth.
This parametric model is not one that is easily represented using standard
mixed-effects or multi-level models (for details, see McArdle & Hamagami,
1996). If invariance is found, we can also examine the equality of the
JOHN J. MCARDLE338
σσe
Y[1]
e[1]
Y[2]
e[2]
V
e[3]
Y[4]
e[4] e[5]
y0
y0*
ys
ys*
Y[6]
e[6]
1
α1
α2
α3 α4 α5α6
σ0 σsµs
0,s
µ0
Y[3] Y[5]
σe σe σe σe σe
Group
ρ ρ
1
σe
Y[1]
e[1]
Y[2]
e[2]
V
e[3]
Y[4]
e[4] e[5]
y0
y0*
ys
ys*
Y[6]
e[6]
1
α1
α2
α3 α4 α5α6
σ0 σsµs
0,s
µ0
Y[3] Y[5]
σe σe σe σe σe
Group
2
Fig. 7. A Latent Growth Curve Path Model for Separate Groups.
Five
Step
sin
Laten
tCurve
Modelin
gwith
Longitu
dinalLife-S
panData
339
variances of the latent levels and slope (sðgÞ0 ¼ . . . s
ðGÞ0 and sðgÞs ¼ . . . sðGÞs ).
Further analyses could include the error deviations (sðgÞe ), the total slope
variance and covariances, and functions of all the other parameters. We may
bring back the typical mixed-effects group difference parameters when we
examine the invariance of the latent means for initial levels and slopes
(mðgÞ0 ¼ m
ðGÞ0 and m
ðgÞ1 ¼ m
ðGÞ1 ). Of course, these group differences in the fixed
effects can be coded in the same way as in the typical mixed-effects analyses.
However, in important ways, each of these multiple group hypotheses rep-
resents a different kind of nonlinearity than was possible to examine using
the mixed-effects approach.
Results for Group Differences in Cognitive Growth for Males and Females
The group differences due to sex, education, and complete versus incomplete
data were also studied using the multiple group growth curves approach. To
illustrate this kind of analysis here, Table 5 gives the initial longitudinal data
on the six intellectual ability measures separately for each Gender. In these
cases, the two groups are created so the unrestricted likelihood for these
data is based on two sets of mean and covariance matrices; one for males
and one for females.
The first model (5a) allows both groups to have completely different
latent growth curves. The model now includes 10 parameters for each group,
and the 20 MLEs are listed in the first two columns. This results in a fairly
reasonable fit to both data sets (w2 ¼ 148 on df ¼ 34). A few small differ-
ences in MLEs can be seen between the two groups, but the key difference
appears to be the smaller error variance for the females (Ffs2eg ¼ 13:9 but
Mfs2eg ¼ 21:0). The second model (5b) adds the restriction that the latent
basis coefficients, while free to vary, must be identical across males and
females. This model is similar in fit to the free model (w2 ¼ 152 on df ¼ 38;
Dw2 ¼ 4 on Ddf ¼ 4), and this indicates the shapes of the curves may be
considered to be parallel across both groups.
The third model (5c) adds the restriction that all latent basis parameters,
while free to vary, must be identical across males and females. This model is
similar in fit to the free model (w2 ¼ 169 on df ¼ 44) and various compar-
isons are clear. (a) This is slightly worse in fit than the previous parallel
shape model (w2 ¼ 17 on Ddf ¼ 6), indicating some of the latent means or
covariances are different – and it is the error variance here. (b) This is not
much worse in fit than the free group model (w2 ¼ 21 on Ddf ¼ 10), indi-
cating not much overall differences between groups. (c) This is a bit worse in
JOHN J. MCARDLE340
fit than the one-group model (w2 ¼ 48 on Ddf ¼ 10), even though the MLE
parameters are identical, and this indicates there may be some basic dif-
ferences between groups that are not captured by this growth model.
As suggested earlier, these multiple group growth models have been used
to compare the complete and incomplete subsets of these data. In all such
cases, the key results for (a) the complete data only and for (b) the complete
and incomplete data together, and the parameters are much the same (as in
Diggle & Kenward, 1994; Little, 1995; McArdle, 1994). As a statistical test
for parameter invariance over these groups we calculated the difference in
Table 5. Numerical Results from Multiple Group Latent Growth
Models Fitted to Male & Female Longitudinal Data (Step 4, see Fig. 8).
Growth Model
Parameters
5a: Latent
Growth for
Males (n ¼ 52)
5a: Latent
Growth for
Females (n ¼ 59)
5b: Loading
Invariance over
Both Groups
5c: Total
Invariance over
Both Groups
Fixed effects
Basis a[04] 1, 0 1, 0 1, 0 1, 0
Basis a[14] 1, 0.75 (0.02) 1, 0.77 (0.01) 1, 0.76 (0.01) 1, 0.76 (0.01)
Basis a[29] 1, 0.94 (0.02) 1, 0.96 (0.01) 1, 0.95 (0.01) 1, 0.95 (0.01)
Basis a[42] 1, 0.99 (0.02) 1, 0.98 (0.02) 1, 0.98 (0.01) 1, 0.98 (0.01)
Basis a[57] 1, 1.01 (0.02) 1, 1.00 (0.02) 1, 1.00 (0.01) 1, 1.00 (0.01)
Basis a[65] 1, 1 1, 1 1, 1 1, 1
Level m0 21.6 (1.4) 22.7 (1.4) 21.5(1.4), 22.8
(1.4)
22.2 (0.97)
Slope ms 65.0 (1.9) 65.9 (1.5) 64.8 (1.6), 66.0
(1.5)
65.5 (1.2)vsp:0.5
Random effects
Error se2 21.0 (2.7) 13.9 (1.6) 21.4 (2.8), 14.0
(1.7)
17.4 (1.5)
Level s0 8.7 (1.1) 9.8 (1.0) 8.7 (1.1),9.7 (1.0) 9.3 (0.75)
Slope ss 9.7 (1.3) 9.3 (1.1) 9.6 (1.2), 9.2
(1.1)
9.4 (0.84)
Correlation r0s �0.64 (0.05) �0.85 (0.05) �0.63
(0.11),�0.85
(0.05)
�0.74 (0.05)
Numbers of
parameters
20 16 10
Log likelihood �1571. �1573 �1582
Likelihood ratio 148 152 169
Degree of
freedom
34 38 44
RMSEA 0.24 0.23 0.23
Five Steps in Latent Curve Modeling with Longitudinal Life-Span Data 341
the likelihoods, and these differences are trivial (w2 ¼ 21 on Ddf ¼ 10). This
suggests that selective dropout or subject attrition can be considered ran-
dom with respect to the General Knowledge abilities measured here. This
result allowed us to combine the complete and incomplete data for a more
powerful analysis.
Mixture Models for Latent Groups
Another fundamental problem is the discrimination of models of multiple
curves for the same people from models of multiple groups with different
curves. It is possible for us to have, say, three clusters of people each with a
distinct growth curve but when we aggregate information over all the people
we end up with three multiple factors and mixed curves. This is the essence
of a latent grouping of people, and parallels the focus on a ‘‘person centered
approach’’ to multivariate data analysis (e.g., Block, 1971; Cattell & Birkett,
1980; Magnusson, 1995).
The recent series of models termed growth mixture models have been
developed for this purpose (Wedel & DeSarbo, 2002; Muthen & Muthen,
1995, 2002; Nagin, 1999). In these analyses, the distribution of the latent
parameters are assumed to come from a ‘‘mixture’’ of two or more over-
lapping distributions. Current techniques in mixture models have largely
been developed under the assumption of a small number of discrete or
probabilistic ‘‘classes of persons’’ based on mixtures of multivariate nor-
mals (e.g., two classes). More formally, we can write a model as a prob-
ability weighted sum of curves, where the probability of class membership
(Prob{cn}) is defined for the person in c ¼ 1 to C classes. In this kind of
growth mixture analysis we estimate the most likely threshold parameter
for each latent distribution (tp, for the pth parameter), while simultane-
ously estimate the separate model parameters for the resulting latent
groups.
The growth mixture models may be seen as a model-restricted fuzzy-set
cluster analysis – a multiple group model without exact knowledge of group
membership for each individual. The concept of an unknown or latent
grouping can be successively based on the logic of multiple group factorial
invariance – starting with equality of latent level means and variances, then
on the latent slope means and variances, then on both the level and slope,
then on the growth loadings, and so on. The resulting MLEs yield a like-
lihood which can be compared to the results obtained from a model with
one less class, so the mixture model distribution can be treated as a hy-
pothesis to be investigated. As in standard discriminant analysis, we can also
JOHN J. MCARDLE342
estimate the probability of assignment of individuals to each class in the
mixture. A variety of new computer programs have been developed for this
purpose (e.g., Mplus, by Muthen & Muthen, 2000).
Results from Latent Mixture Models
These latent growth mixture models were fit using all the longitudinal data
(Fig. 1) and some results are described in Fig. 8. In a first latent mixture
model, we estimated a two-class model with free parameters for both
groups. This model required 21 parameters, adding the one more than the
previous two-group model, and led to another likelihood (L2 ¼ �1544). We
need to recognize that the statistical basis of this comparison is still some-
what controversial, but if we consider the threshold as an implied parameter
in some previous models, we can get some sense of the gain in fit. The extra
parameter fitted was the threshold (t ¼ �0:55) – a z-score that suggests the
total group can be considered a mixture of two classes of different sizes,
n ¼ 72 and 39, with different growth patterns. By contrast to the one-class
model (w2 ¼ 78 on df ¼ 10) or even versus the unrestricted two-group model
(w2 ¼ 92 on df ¼ 1), two groups are better than one.
By inspection, we can see that the larger group has a growth pattern with
parameters that are very similar to the overall pattern described earlier (i.e.,
0
10
20
30
40
50
60
70
80
90
100
0 4 14 30 42 56 65 75
Model 1, Class 1
Model 1, Class 2
Age (Years)
0
10
20
30
40
50
60
70
80
90
100
0 4 14 30 42 56 65 75
Model 2, Class 1
Model 2, Class 2
Age (Years)
Same latent basis A[t],
Fit χ2=30, df=3,
Prob(Class1)=.92
Free latent basis A[t],
Fit χ2=34, df=12,
Prob(Class1)=.67
Fig. 8. Results from SEM Growth-Mixture Model with Two ‘‘Classes’’ of Growth
Curves. (Note: Model Includes Two Classes of LGMs and Probability of Group
Assignment (Results from Mplus).)
Five Steps in Latent Curve Modeling with Longitudinal Life-Span Data 343
M4). However, the smaller second group seems to have a higher initial basis
(a½t�ð1Þ ¼ 0:73 vs. a½t�ð2Þ ¼ 0:84), a much higher initial mean (mð1Þ0 ¼ 16:9 vs.
mð2Þ0 ¼ 31:4), lower slope changes (mð1Þ1 ¼ 71:1 vs. m
ð2Þ1 ¼ 55:7), and much low-
er variances (sð1Þ0 ¼ 7:1 vs. s
ð2Þ0 ¼ 1:9). This second group shows almost no
variance in slope (sð1Þ1 ¼ 7:1 vs. sð2Þ1 40:1), and this could reflect an estimation
problem, or even a problem of limitation of measurement. Assuming these
are not critical problems, the second smaller class appears to be an initially
higher functioning group which slowly moves toward the final adult levels.
In second latent mixture model, we allowed the possibility of two latent
classes (C ¼ 2) with different parameters for the latent means and variance
but assuming the same growth basis. This results in a more equal separation
of persons (t ¼ �0:07; so n ¼ 56 and 55) with same free basis coefficients
(A½t� ¼ ½¼ 0; 0:76; 0:95; 0:98; 1:00;¼ 1�), but with the same pattern of means
and variances as the first model. This model loses a bit in fit (L2 ¼ �1565; soM14 vs. M13: Dw2 ¼ 36 on Ddf ¼ 6), so it may not be a good idea to force
exactly the same shape on groups that start at different points. Either way,
this still implies that a small group of persons started much higher than
average score and had a smaller change over time but the two classes of
curves eventually do converge in adulthood.
STEP 5: STUDYING DYNAMIC DETERMINANTS
ACROSS MULTIPLE VARIABLES
In recent research, we have considered some ways to improve the clarity of
the basic dynamic change interpretations with conventional SEM analytic
techniques. This has led to the development of a set of alternative models,
based on classical principles of dynamic growth and change (including de-
cline), but represented in the form of latent difference scores (e.g., McArdle,
2001; McArdle & Nesselroade, 1994). This alternative representation makes
it relatively easy to represent a dynamic hypothesis about the change within
a variable, and about the time-ordered determination of one variable upon
another. An overview of this new approach is described here as the fifth step
in longitudinal data analysis.
Modeling Latent Difference Scores
The introduction of multiple variables at each longitudinal occasion of meas-
urement leads naturally to questions about time-dependent relationships
JOHN J. MCARDLE344
among growth. A classical SEM for multiple variables over time is based on a
latent variable cross-lagged regression model (see Cook & Campbell, 1979;
Rogosa, 1978). This model can be written for latent scores with over-time
auto-regressions (fy, fx) and cross-regressions (dyx, dx) for time-lagged pre-
dictors, but the standard applications of this model do not include systematic
growth components (i.e., individual slopes). For this reasons, recent SEM
analyses have examination of parallel growth curves, including the correlation
of various components (McArdle, 1988, 1990; Willett & Sayer, 1994). A
popular alternative used in multi-level and mixed-effects modeling is based on
the analysis of covariance with X[t] as time-varying covariates. In this model,
the regressions (d) are fixed (group) coefficients with the same effect on
Y[t] scores at all occasions. These last two models are easy to implement
using existing computer software (e.g., Sliwinski & Buschke, 1999; Sullivan,
Rosenbloom, Lim, & Pfefferman, 2000; Verbeke, Molenberghs, Krickeberg,
& Fienberg, 2000), but the typical applications are limited to a few basic
forms of dynamic hypotheses.
To expand our SEM for other dynamic concepts we start in a different
way. First, we assume we have a pair of observed scores (Y ½t� and Y ½t� 1�)
measured over a defined interval of time (Dt ¼ 1), and we write a model with
latent scores (y½t� and y½t� 1�), and corresponding errors of measurement
(e½t� and e½t� 1�), so the new latent variable Dy[t] is directly interpreted as a
latent difference score. This latent difference score (Dy½t�n) is not the same as
an observed difference score (DY ½t�n), because the latent score is considered
separate from the removal of the model-based error component. In this
latent difference score approach, we do not directly define the A[t] coeffi-
cients, but instead we directly define changes as an accumulation of the first
differences among latent variables.
This deceptively simple algebraic device allows us to generally define the
trajectory equation as an accumulation of the latent changes (Dy½t�) up to
time t based on any model of change. Using this approach, all of the previous
latent growth models can be drawn in terms of first differences, and some
new models can more easily emerge (as in McArdle & Nesselroade, 1994;
McArdle, 2001; McArdle & Hamagami, 2001). For example, we can write a
composite change expression model, where we permit both a systematic
constant change (a) and a systematic proportional change (b) over time.
This linear difference model lead to a nonlinear mixed-effects model tra-
jectory from a simple accumulation of first differences. It can also be fitted
using standard SEM software (e.g., LISREL, Mx, etc.)
To deal with multiple variables we can now write a bivariate dynamic
change score model, such as the model depicted in the path diagram of Fig. 9.
Five Steps in Latent Curve Modeling with Longitudinal Life-Span Data 345
V3
e4 e5
Y[3]
y[1] y[2]
Y[4]
y[4] y[5]
Y[6]
y[6]
Y[5]
e2ey[1] ey[2] ey[4] ey[6]
y[3]
∆y[2] ∆y[3] ∆y[4]∆y[5] ∆y[6]
Y[1] Y[6]
∆x[2]∆x[3] ∆x[4]
∆x[6]
e4 e5
x[1] x[2] x[4] x[5] x[6]x[3]
V3X[3] X[4] X[6]X[5]X[1] X[6]
e2ex[1] ex[2] ex[4] ex[6]
y0
ys
xs
x0
∆x[5]
y0
ys
xs
x0
K
µy0
σy0,ys
µys
µx0
µxs
σxs
σx0
σys
σy0
σx0,xs
σys,xs
σy0,xs
σx0,ys
σx0,y0
αy
αx
βy βy βy βy βy
βx βx βx βx βx
γxγx γx γx
γx
γy γy γy γy γy
σy
2
σx2
σy,x
*
*
*
*
Fig. 9. A Bivariate Latent Difference Score (BLDS) Path Model for Dynamic Hypotheses.
JOHN
J.M
CARDLE
346
We assume a dual change score model within each variable (parameters a and
b) but also permit a coupling hypothesis (parameters g) across different var-
iables. This model is used to estimate the time-dependent effect of latent x[t]
on Dy[t] (gyx) as well as coupling parameter representing the time-dependent
effect of latent y[t] on Dx[t] (gxy). This model subsumes all aspects of the
previous cross-lagged, correlated growth, and time-varying covariate models
as special cases and this is useful because results from these dynamic alter-
natives can be quite different (see McArdle & Hamagami, 2001). These latent
difference score models can lead to more complex nonlinear trajectory equa-
tions (e.g., non-homogeneous equations), but the use of latent difference
scores makes it practical to analyze a variety of dynamic models using stand-
ard SEM.
Results from Fitting Latent Difference Score Models
The latent difference score dynamic models were fitted (using Mx and
Mplus) to both the General Knowledge variable (now termed Verbal) and
also to a second set of Non-Verbal scores. We will only highlight the dy-
namic results here (but see McArdle & Hamagami, 2004). In order to fit the
dual change model the additive slope coefficient was fixed for identification
purposes (a ¼ 1) but the mean of the slopes was allowed to be free (m1). This
allowed estimation of: (a) auto-proportion effects (b ¼ �0:34;�1:38), (b)initial level means (m0 ¼ 14:8; 32:1) at Age ¼ 5, and (c) linear slope means
(m1 ¼ 26:0; 110:1) for each 5-year-period after (Age 4 5). The goodness-
of-fit of the dual change model was compared to every other nested alter-
native and these comparisons show the best fit was achieved using this
model (e.g., versus a linear growth, w2 ¼ 584; 439 on df ¼ 1).
Several alternative Verbal–Non-Verbal bivariate coupling models based
on Fig. 9 were fitted to the data. A first model included six dynamic co-
efficients (two each for a, b, g), four latent means (m), six latent deviations
(s), and six latent correlations (r). This model was fitted with N ¼ 111
individuals with at least one point of data, 498 individual data observations,
and yield one overall fit (�2logL ¼ 7118), which was different from a ran-
dom baseline (w2 ¼ 379 on df ¼ 16). The model parameters can be used to
form the expected values described in Fig. 9. These parameters are specific
to the time interval chosen (i.e., Dt ¼ 5) and any calculation of the other
information (e.g., explained latent variance) requires a specific interval of
age. However, these seemingly small differences can accumulate over longer
periods of time so the larger N[t] is expected to account for an increasing
variance in DV [t] over age.
Five Steps in Latent Curve Modeling with Longitudinal Life-Span Data 347
The fitting of a further sequence of alternative models is needed to in-
terpret the replicability of the coupling across the V [t] and N [t] variables.
Table 6 gives details about two specific models fit to examine whether one or
more of the coupling parameters (g) were different from zero. In the first
alternative model (6a), the parameter representing the effect of N[t] on DV[t]
was fixed to zero (gx ¼ 0), and this led to a notable loss of fit (w2 ¼ 123 on
df ¼ 1). The second alternative (6b) assumed no effect from V ½t� on
DN½t�ðgy ¼ 0Þ and this is a much smaller loss of fit (w2 ¼ 27 on df ¼ 1).
These results were examined in a number of results of additional ways, but
these comparisons suggest that the second model 6b was the most reason-
able representation of these longitudinal data. The resulting interpretation is
a dynamic process, where scores on Non-Verbal abilities have a tendency to
impact score changes on the Knowledge-Verbal scores, but there is no no-
table reverse effect.
The estimated model parameters are highly dependent on the scalings
used, but the trajectory expectations allow us to interpret the results in a
relatively ‘‘scale-free’’ form – Fig. 10 gives a summary of this state-space
plot as a vector field plot (for details, see Boker & McArdle, 1995; McArdle
et al., 2001). Any pair of coordinates is a starting point (y0, x0) and the
directional arrow is a display of the expected pair of 5-year changes (Dy, Dx)
from this point. The last two figures show an interesting dynamic property –
the change expectations of a dynamic model depend on the starting point.
From this perspective, we can also interpret the positive level–level corre-
lation (ry0;x0 ¼ 0:78), which describes the placement of the individuals in the
vector field, and the small slope–slope correlation (rys;xs ¼ �0:06), whichdescribes the location of the subsequent change scores for individuals in the
vector field. The resulting ‘‘flow’’ shows a dynamic process, where scores on
Non-Verbal abilities have a tendency to impact score changes on the Verbal
scores, but there is no notable reverse effect.
DISCUSSION
This chapter can only serve as an introduction to a general class of pro-
cedures that can be classified under the rubric of ‘‘latent growth curve
modeling techniques.’’ The brevity of this chapter cannot do justice to the
broad applicability of these techniques, and the five steps outlined here
represent just one way to organize some of the inherent complexities of this
topic. In this last section, I discuss some of the future directions and chal-
lenges for this kind of research.
JOHN J. MCARDLE348
Table 6. Results of Bivariate Latent Difference Score (BLDS) Dynamics Models Fitted to Two Variables
over Six Occasions (for details, see McArdle & Hamagami, 2004a, b; Step 5, see Figs. 9 and 10).
Model Parameters 6a (No Coupling from Non-Verbal) 6b (No Coupling from Verbal)
Knowledge Non-Verbal Knowledge Non-Verbal
(a) Fixed effects
Initial mean m0 14.8 32.1 15.5 31.8
Slope mean ms 26.1 42.4 0.01 46.7
Loading a ¼ 1 ¼ 1 ¼ 1 ¼ 1
Proportion b �0.34 �0.40 �0.77 �0.58
Coupling from knowledge gnk — ¼ 0 — 0.70
Coupling from non-verbal gkn �0.15 — ¼ 0 —
(b) Random effects
Error deviation se 7.04 10.5 5.85 10.8
Initial deviation s0 2.87 2.84 1.64 4.92
Slope deviation ss 4.29 5.52 8.82 6.34
Correlation r0s o0.99 0.40 o0.99 0.52
Correlation ry0, x0 o0.99 o0.99
Correlation rys, xs 0.63 0.39
Correlation ry0, xs 0.44 o0.99
Correlation rys, x0 0.85 �0.39
(c) Goodness-of-fit
�2log L 7241 7145
Parameters 19 19
Five
Step
sin
Laten
tCurve
Modelin
gwith
Longitu
dinalLife-S
panData
349
Formal models for the analysis of these kinds of growth curves have been
developed in many different substantive domains. Researchers have found
many creative ways to analyze average trends in growth data, including
classical Analysis of Variance techniques (e.g., Pothoff & Roy, 1964; Bock,
1975). These classical methods provide powerful and accurate tests of
‘‘group trends.’’ However, the introduction of individual differences in
change analyses led to a great deal of statistical controversy in model fitting.
Early work on these problems led to the polynomial growth models by
Wishart (1934), where an individual regression coefficient was used to de-
scribe a growth characteristic of the person (see Rogosa & Willett, 1985).
Other current techniques have roots in the important innovations by
Meredith & Tisak (1990), who showed how the ‘‘Tuckerized curve’’ models
0 10 20 30 40 50 60 70 80 90 1000
10
20
30
40
50
60
70
80
90
100
Non-V
erbal
Verbal
Fig. 10. Results from the Vector Field Representation of the BDLS for the Six-
occasion Longitudinal Data. (Note: Each Arrow Represents the Expected Change of
Direction in the Scores – the Derivatives – for Pairs of Initial Starting Abilities.)
JOHN J. MCARDLE350
(named in recognition of Tucker’s contributions) could be represented and
fitted using structural equation modeling based on restricted common fac-
tors. For these reasons, the term latent curve models (LCM) seems appro-
priate for any technique that describes the underlying growth in terms of
latent changes using the classical assumptions (e.g., independence of errors).
These innovative techniques were important because this made it possible to
represent a wide range of alternative growth and change models by adding
the benefits of the structural equation modeling techniques (Meredith &
Tisak, 1990; McArdle & Prescott, 1997; McArdle, 1986; McArdle & Epstein,
1987; McArdle & Anderson, 1990; McArdle & Hamagami, 1992, 2001).
During the last decade it has become possible to prove that many seem-
ingly different statistical models have identical properties and can yield
identical results. They are all based on fitting observed raw-score longitu-
dinal growth data to the same theoretical model using the same likelihood-
based techniques (as in Little & Rubin, 1987; McArdle, 1994; McArdle &
Bell, 2000). These latent growth models have since been expanded upon and
described by many others (McArdle & Woodcock, 1997; Willett & Sayer,
1994; Muthen & Curran, 1997; Metha & West, 2000). The contemporary
basis of latent curve analyses can also be found in the recent developments
of multi-level models (Goldstein, 1995; Bryk & Raudenbush, 1992) or mixed-
effects models (Laird & Ware, 1982; Singer, 1998). In important work by
Browne & du Toit (1991), classical nonlinear models were added as part of
this same framework (see Cudeck & du Toit, 2001; McArdle & Hamagami,
1996, 2001; Pinheiro & Bates, 2000). When these options are added to the
latent variable path analysis models of SEM (e.g., McArdle & Prescott,
1992), many limitations apparent in previous longitudinal research can be
overcome.
In a similar way, most of the models presented here can be estimated
using contemporary computer programs including: (1) the SAS and S-Plus
packages (Littell et al., 1996; Singer, 1998; Verbeke & Molenberghs, 2000;
Pinheiro & Bates, 2000), (2) general SEM programs such as LISREL
(McArdle & Epstein, 1987; Cudeck & duToit, 2001), Mx (Neale, Boker, Xie,
& Maes, 1999), Mplus (Muthen & Muthen, 2002), and AMOS (Arbuckle &
Wothke, 1999), and (3) specialized softwares such as MIXOR (Hedecker &
Gibbons, 1996) and MLn (Goldstein, 1995). The results for the classical
SEM for latent curves will be the same no matter what program is used (for
demonstration, see Ferrer, Hamagami, & McArdle, 2004). Some of the more
complex models used here (e.g., Table 6) now require specialized SEM
software (e.g., Mx or Mplus), but it is likely that these routines will become
a part of standard computer packages in the near future.
Five Steps in Latent Curve Modeling with Longitudinal Life-Span Data 351
At the same time, I need to recognize that these SEM developments
represent a limited class of longitudinal data analyses (e.g., Nesselroade &
Baltes, 1979; Collins & Sayer, 2001), and we should be on the lookout for
possible improvements. In the analyses presented here, I have also tried
to present some of our most up-to-date interpretations dealing with the
developmental-dynamic processes around aspects of human cognition over
the life-span. Indeed, some of the most difficult problems for future work
on latent curves will not involve SEM statistical analysis or computer
programming, but will be focused on the rather elusive meaning of the la-
tent model parameters themselves (Zeger & Harlow, 1987; McArdle &
Nesselroade, 2003). These substantive-methodological problems remain
among the most difficult dynamic challenges for our future work.
ACKNOWLEDGMENTS
The work described here has been supported since 1980 by the National
Institute on Aging (Grant#AG-07137). I am especially grateful to the work
of my close friend and colleague, Fumiaki Hamagami, and to my close
collaborations with Katherine P. Bradway and John L. Horn. This research
was also helped by the support of my many friends and colleagues, including
Steven Aggen, Paul Baltes, Steven Boker, Emilio Ferrer-Caja, Paolo
Ghisletta, Patty Hulick, Bill Meredith, John Nesselroade, Carol Prescott,
and Dick Woodcock. I also thank the PaVie team for excellent editorial
suggestions and assistance in the presentation and preparation of this man-
uscript. Reprints can be obtained from the author at the Jefferson Psycho-
metric Laboratory, P.O. Box 400400, Department of Psychology, University
of Virginia, Charlottesville, VA 22904, USA.
REFERENCES
Abeles, R. (Ed.) (1987). Life-span perspectives and social psychology. Hillsdale, NJ: Erlbaum.
Arbuckle, J. L., & Wothke, W. (1999). AMOS 4.0 user’s guide. Chicago: Smallwaters.
Baltes, P. B. (1968). Longitudinal and cross-sectional sequences in the study of age and gene-
rational effects. Human Development, 11, 145–171.
Baltes, P. B. (1983). Life-span developmental psychology: Observations on history and theory
revisited. In: R. M. Lerner (Ed.), Developmental psychology: Historical and philosophical
perspectives (pp. 79–111). Hillsdale, NJ: Erlbaum.
Baltes, P. B., Lindenberger, U., & Staudinger, U. M. (1998). Life-span theory in developmental
psychology. In: R. M. Lerner (Ed.), Handbook of child psychology: Vol. 1. Theoretical
models of human development (5th ed., pp. 1029–1143). New York: Wiley.
JOHN J. MCARDLE352
Baltes, P. B., & Nesselroade, J. R. (1970). Multivariate longitudinal and cross-sectional se-
quences for analyzing ontogenetic and generational change: A methodological note.
Developmental Psychology, 2, 63–168.
Baltes, P. B., & Nesselroade, J. R. (1972). Cultural change and adolescent personality devel-
opment: An application of longitudinal sequences. Developmental Psychology, 7,
244–256.
Baltes, P. B., & Nesselroade, J. R. (1979). History and rationale of longitudinal research. In:
J. R. Nesselroade & P. B. Baltes (Eds), Longitudinal research in the study of behavior and
development. New York: Academic Press.
Baltes, P. B., Reese, H. W., & Nesselroade, J. R. (Eds) (1977). Life-span development meth-
odology: Introduction to research methods. Monterey, CA: Brooks-Cole.
Bell, R. Q. (1954). An experimental test of the accelerated longitudinal approach. Child De-
velopment, 25, 281–286.
Block, J. (1971). Lives through time. Berkeley, CA: Bancroft.
Bock, R. D. (1975).Multivariate statistical methods in behavioral research. New York: McGraw-
Hill.
Boker, S. M., & McArdle, J. J. (1995). Statistical vector field analysis applied to mixed cross-
sectional and longitudinal data. Experimental Aging Research, 21, 77–93.
Bradway, K. P., & Thompson, C. W. (1962). Intelligence at adulthood: A 25 year follow-up.
Journal of Educational Psychology, 53(1), 1–14.
Browne, M., & du Toit, S. H. C. (1991). Models for learning data. In: L. Collins & J. L. Horn
(Eds), Best methods for the analysis of change (pp. 47–68). Washington, DC: APA Press.
Bryk, A. S., & Raudenbush, S. W. (1992). Hierarchical linear models: Applications and data
analysis methods. Newbury Park, CA: Sage.
Cattell, R. B. (1963). The structuring of change by P-technique and incremental R-technique.
In: C. W. Harris (Ed.), Problems in measuring change (pp. 167–198). Madison, WI: The
University of Wisconsin Press.
Cattell, R. B. (1969). Comparing factor trait and state scores across ages and cultures. Journal
of Gerontology, 24, 348–360.
Cattell, R. B. (1970). Separating endogenous, exogenous, ecogenic, and epogenic component
curves in developmental data. Developmental Psychology, 3(2), 151–162.
Cattell, R. B., & Birkett, H. (1980). Can P-technique diagnosis be practicably shortened? Some
proposals and a test of a 50 day abridgement. Multivariate Experimental Clinical Re-
search, 5, 1–16.
Cook, T. D., & Campbell, D. T. (1979). Quasi-experimentation: Design & analysis issues for field
settings. Boston, MS: Houghton Mifflin.
Cnaan, A., Laird, N. M., & Slasor, P. (1997). Using the general linear mixed model to analyse
unbalanced repeated measures and longitudinal data. Statistics in Medicine, 16, 2349–2380.
Collins, L., & Sayer, A. (Eds) (2001). New Methods for the Analysis of Change (pp. 137–176).
Washington, DC: APA Press.
Cudeck, R., & du Toit, S. H. C. (2001). Mixed-effects models in the study of individual dif-
ferences with repeated measures data. Multivariate Behavioral Research, 31, 371–403.
Deary, I. J., Whalley, L. J., Lemmon, H., Crawford, J. R., & Starr, J. M. (2000). The stability of
individual differences in mental ability from childhood to old age: Follow-up of the 1932
Scottish Mental Survey. Intelligence, 28, 49–55.
Diggle, P. J., & Kenward, M. G. (1994). Informative drop-out in longitudinal data analysis.
Applied Statistics, 43, 49–93.
Five Steps in Latent Curve Modeling with Longitudinal Life-Span Data 353
Donaldson, G., & Horn, J. L. (1992). Age, cohort, and time developmental muddles: Easy in
practice, hard in theory. Experimental Aging Research, 18, 213–222.
Duncan, T. E., Duncan, S. C., Strycker, L. A., Li, F., & Alpert, A. (1998). An introduction to latent
variable growth curve modeling: Concepts, issues, and applications. Mahwah, NJ: Erlbaum.
Ferrer, E., Hamagami, F., & McArdle, J. J. (2004). Modeling latent growth curves with in-
complete data using different types of structural equation modeling and multilevel soft-
ware. Structural Equation Modeling, 11, 452–483.
Goldstein, H. (1995). Multilevel statistical models (2nd Ed.). New York: Oxford Press.
Goulet, L. R., & Baltes, P. B. (Eds) (1970). Life-span psychology: Research and theory.
New York: Academic Press.
Hamagami, F. (1998). A developmental-based item factor analysis. In: J. J. McArdle &
R. W. Woodcock (Eds), Human abilities in theory and practice (pp. 231–246). Mahwah,
NJ: Erlbaum.
‘Hamagami, F., & McArdle, J. J. (2001). Advanced studies of individual differences linear dy-
namic models for longitudinal data analysis. In: G. A. Marcoulides & R. E. Schumacker
(Eds), New developments and techniques in structural equation modeling (pp. 203–246).
Mahwah, NJ: Lawrence Erlbaum Associates.
Hedecker, D., & Gibbons, R. (1996). MIXOR: A computer program for mixed-effects ordinal
regression analysis. Computer Methods and Programs in Biomedicine, 49, 157–176.
Horn, J. L., & Donaldson, G. (1976). On the myth of intellectual decline in adulthood.
American Psychologist, 31, 701–709.
Horn, J. L., & McArdle, J. J. (1980). Perspectives on mathematical/statistical model building
(MASMOB) in research on aging. In: L. W. Poon (Ed.), Aging in the 1980’s: Selected
contemporary issues in the psychology of aging (pp. 503–541). Washington, DC:
American Psychological Association.
Joreskog, K. G., & Sorbom, D. (1979). Advances in factor analysis and structural equation
models. Cambridge, UK: Abt Books.
Kangas, J., & Bradway, K. P. (1971). Intelligence at middle age: A thirty-eight year follow-up.
Developmental Psychology, 5(2), 333–337.
Laird, N. M., & Ware, J. H. (1982). Random effects models for longitudinal data. Biometrics,
38, 963–974.
Littell, R. C., Miliken, G. A., Stoup, W. W., & Wolfinger, R. D. (1996). SAS system for mixed
models. Cary, NC: SAS institute.
Little, R. J. A. (1995). Modeling the dropout mechanism in repeated-measures studies. Journal
of the American Statistical Association, 90, 1112–1121.
Little, R. T. A., & Rubin, D. B. (1987). Statistical analysis with missing data. New York: Wiley.
Magnusson, D. (1995). Individual development: A holistic, integrated model. In: P. Moen,
G. H. J. Elder & K. Luescher (Eds), Examining lives in context (pp. 19–60). Washington,
DC: American Psychological Associations.
McArdle, J. J. (1986). Latent variable growth within behavior genetic models. Behavior Ge-
netics, 16(1), 163–200.
McArdle, J. J. (1988). Dynamic but structural equation modeling of repeated measures data. In:
J. R. Nesselroade & R. B. Cattell (Eds), Handbook of multivariate experimental psy-
chology (pp. 561–614). New York: Plenum Press.
McArdle, J. J. (1989). Structural modeling experiments using multiple growth functions. In:
P. Ackerman, R. Kanfer & R. Cudeck (Eds), Learning and individual differences: Abilities,
motivation, and methodology (pp. 71–117). Hillsdale, NJ: Erlbaum.
JOHN J. MCARDLE354
McArdle, J. J. (1990). Principle versus principals of structural factor analyses. Multivariate
Behavioral Research, 25, 81–87.
McArdle, J. J. (1994). Structural factor analysis experiments with incomplete data. Multivariate
Behavioral Research, 29(4), 409–454.
McArdle, J. J. (2001). A latent difference score approach to longitudinal dynamic structural
analyses. In: R. Cudeck, S. du Toit & D. Sorbom (Eds), Structural equation mode-
ling: Present and future (pp. 342–380). Lincolnwood, IL: Scientific Software Inter-
national.
McArdle, J. J., & Anderson, E. (1990). Latent variable growth models for research on aging. In:
J. E. Birren & K. W. Schaie (Eds), The handbook of the psychology of aging (pp. 21–43).
New York: Plenum Press.
McArdle, J. J., & Bell, R. Q. (2000). Recent trends in modeling longitudinal data by latent
growth curve methods. In: T. D. Little, K. U. Schnabel & J. Baumert (Eds), Modeling
longitudinal and multiple-group data: Practical issues, applied approaches, and scientific
examples (pp. 69–107). Mahwah, NJ: Erlbaum.
McArdle, J. J., & Cattell, R. B. (1994). Structural equation models of factorial invariance in
parallel proportional profiles and oblique confactor problems. Multivariate Behavioral
Research, 29, 63–113.
McArdle, J. J., & Epstein, D. B. (1987). Latent growth curves within developmental structural
equation models. Child Development, 58(1), 110–133.
McArdle, J. J., Ferrer-Caja, E., Hamagami, F., & Woodcock, R. W. (2002). Comparative
longitudinal structural analyses of the growth and decline of multiple intellectual abilities
over the life span. Developmental Psychology, 38, 115–142.
McArdle, J. J., & Hamagami, E. (1992). Modeling incomplete longitudinal and cross-sectional
data using latent growth structural models. Experimental Aging Research, 18(3), 145–166.
McArdle, J. J., & Hamagami, F. (1996). Multilevel models from a multiple group structural
equation perspective. In: G. Marcoulides & R. Schumacker (Eds), Advanced structural
equation modeling techniques (pp. 89–124). Hillsdale, NJ: Erlbaum.
McArdle, J. J., & Hamagami, F. (2001). Linear dynamic analyses of incomplete longitudinal data.
In: L. Collins & A. Sayer (Eds), New methods for the analysis of change (pp. 137–176).
Washington, DC: APA Press.
McArdle, J. J., & Hamagami, F. (2003). Structural equation models for evaluating dynamic
concepts within longitudinal twin analyses. Behavior Genetics, 33(3), 137–159.
McArdle, J. J., & Hamagami, F. (2004). Methods for dynamic change hypotheses. In: K. van
Montfort, J. Oud & A. Satorra (Eds), Recent developments in structural equation models
(pp. 295–335). Dordrecht, The Netherlands: Kluwer Academic Publisher.
McArdle, J. J., Hamagami, F., Jones, K., Jolesz, F., Kikinis, R., Spiro, A., et al. (2004).
Structural modeling of dynamic changes in memory and brain structure using longi-
tudinal data from the Normative Aging Study. Journals of Gerontology: Psychological
Sciences, 59B, P294–P304.
McArdle, J. J., Hamagami, F., Meredith, W., & Bradway, K. P. (2001). Modeling the dynamic
hypotheses of Gf-Gc theory using longitudinal life-span data. Learning and Individual
Differences, 12(2000), 53–79.
McArdle, J. J., & Nesselroade, J. R. (1994). Using multivariate data to structure developmental
change. In: S. H. Cohen & H. W. Reese (Eds), Life-span developmental psychology:
Methodological contributions (pp. 223–267). Hillsdale, NJ: Lawrence Erlbaum
Associates.
Five Steps in Latent Curve Modeling with Longitudinal Life-Span Data 355
McArdle, J. J., & Nesselroade, J. R. (2003). Growth curve analyses in contemporary psycho-
logical research. In: J. Schinka & W. Velicer (Eds), Comprehensive handbook of psycho-
logy, volume two: Research methods in psychology (pp. 447–480). New York: Pergamon
Press.
McArdle, J. J., & Prescott, C. A. (1992). Age-based construct validation using structural equa-
tion models. Experimental Aging Research, 18(3), 87–115.
McArdle, J. J., & Prescott, C. A. (1997). Contemporary models for the biometric genetic
analysis of intellectual abilities. In: D. P. Flanagan, J. L. Genshaft & P. L. Harrison
(Eds), Contemporary intellectual assessment: Theories, tests, and issues (pp. 403–436).
New York, NY: Guilford Publications.
McArdle, J. J., & Woodcock, J. R. (1997). Expanding test-rest designs to include developmental
time-lag components. Psychological Methods, 2(4), 403–435.
Meredith, W., & Tisak, J. (1990). Latent curve analysis. Psychometrika, 55, 107–122.
Metha, P. D., & West, S. G. (2000). Putting the individual back into individual growth curves.
Psychological Methods, 5(1), 23–43.
Muthen, B. O., & Curran, P. (1997). General longitudinal modeling of individual differences in
experimental designs: A latent variable framework for analysis and power estimation.
Psychological Methods, 2, 371–402.
Muthen, B. O., & Muthen, L. K. (1995). Tailoring psychometric techniques for epidemiological
and clinical applications. Technical Report.
Muthen, B. O., & Muthen, L. K. (2000). Integrating person-centered and variable-centered
analyses: Growth mixture modeling with latent trajectory classes. Alcoholism: Clinical
and Experimental Research, 24, 882–891.
Muthen, L. K., & Muthen, B. O. (2002). Mplus, the comprehensive modeling program for applied
researchers users guide. Los Angeles, CA: Muthen & Muthen.
Nagin, D. (1999). Analyzing developmental trajectories: Semi-parametric. Group-based ap-
proach. Psychological Methods, 4, 139–177.
Neale, M. C., Boker, S. M., Xie, G., & Maes, H. H. (1999). Mx statistical modeling (5th ed.).
Unpublished program manual, Virginia Institute of Psychiatric and Behavioral Genetics,
Medical College of Virginia, Virginia Commonwealth University, Richmond, VA.
Nesselroade, J. R. (2001). Paul B. Baltes: Gerontology with a life-span perspective. Contem-
porary Gerontology, 7(3), 75–77.
Nesselroade, J. R., & Baltes, P. B. (Eds) (1979). Longitudinal research in the study of behavior
and development. New York: Academic Press.
Pinheiro, J. C., & Bates, D. M. (2000). Mixed-effects models in S and S-PLUS. New York:
Springer.
Pothoff, R. F., & Roy, S. N. (1964). A generalized multivariate analysis model useful especially
for growth curve problems. Biometrics, 51, 313–326.
Rao, C. R. (1958). Some statistical methods for the comparison of growth curves. Biometrics,
14, 1–17.
Rogosa, D. (1978). Causal models in longitudinal research: Rationale, formulation, and in-
terpretation. In: J. R. Nesselroade & P. B. Baltes (Eds), Longitudinal research in human
development: Design and analysis (pp. 263–302). New York, NY: Academic Press.
Rogosa, D., & Willett, J. B. (1985). Understanding correlates of change by modeling individual
differences in growth. Psychometrika, 50, 203–228.
Schaie, K. W. (1965). A general model for the study of developmental problems. Psychological
Bulletin, 64, 92–107.
JOHN J. MCARDLE356
Schaie, K. W., & Baltes, P. B. (1975). On sequential strategies in developmental research:
Description or explanation? Human Development, 18, 384–390.
Seber, G. A. F., & Wild, C. J. (1989). Nonlinear models. New York: Wiley.
Singer, J. D. (1998). Using SAS PROC MIXED to fit multilevel models, hierarchical models,
and individual growth models. Journal of Educational and Behavioral Statistics, 24(4),
323–355.
Singer, J. D., & Willett, J. (2003). Applied longitudinal data analysis. New York: Oxford Uni-
versity Press.
Sliwinski, M. J., & Buschke, H. (1999). Cross-sectional and longitudinal relationships among
age, cognition, and processing speed. Psychology and Aging, 14, 18–33.
Sullivan, E. V., Rosenbloom, M. J., Lim, K. O., & Pfefferman, A. (2000). Longitudinal changes
in cognition, gait, balance in abstinent and relapsed alcoholic men: Relationships to
changes in brain structure. Neuropsychology, 14(2), 178–188.
Tucker, L. R. (1958). Determination of parameters of a functional relation by factor analysis.
Psychometrika, 23, 19–23.
Tucker, L. R. (1966). Learning theory and multivariate experiment: Illustration by determi-
nation of generalized learning curves. In: R. B. Cattell (Ed.), Handbook of multivariate
experimental psychology. Chicago, IL: Rand McNally.
Verbeke, G., & Molenberghs, G. (2000). Linear mixed models for longitudinal data. New York,
NY: Springer.
Verbeke, G., Molenberghs, G., Krickeberg, K., & Fienberg, S. (Eds) (2000). Linear mixed
models for longitudinal data. New York: Springer Verlag.
Wohlwill, J. F. (1973). The study of behavioral development. New York, NY: Academic Press.
Willett, J. B., & Sayer, A. G. (1994). Using covariance structure analysis to detect correlates and
predictors of individual change over time. Psychological Bulletin, 116, 363–381.
Wishart, J. (1934). Bibliography of agricultural statistics 1931–1933. Journal of the Royal Sta-
tistical Society, 1, 94–106.
Wishart, J. (1938). Growth rate determinations in nutrition studies with the bacon pig, and their
analyses. Biometrika, 30, 16–28.
Zeger, S. L., & Harlow, S. D. (1987). Mathematical models from laws of growth to tools for
biologic analysis: Fifty years of growth. Growth, 51, 1–21.
Five Steps in Latent Curve Modeling with Longitudinal Life-Span Data 357
INCITATIONS FOR
INTERDISCIPLINARITY IN LIFE
COURSE RESEARCH
Rene Levy, Paolo Ghisletta, Jean-Marie Le Goff,
Dario Spini and Eric Widmer
HOUSING THE HARVEST
Having gone through this volume, a critical reader might come to the con-
clusion that interdisciplinarity can be found more easily between the con-
tributions than within them (even though several of them address it directly,
e.g., Settersten & Gannon; Mortimer et al.).1 However, the contributors
share the common belief that studying humans’ unfolding lives in a web of
complicated interactions within their changing contexts requires the adop-
tion of an interdisciplinary research paradigm. To be sure, the life-span/life
course research traditions stemming from disciplines such as sociology,
psychology, social psychology and demography certainly have allowed
scholars to answer some key questions germane to this field (Baltes,
Lindenberger & Staudinger, in press; Elder, 1998). The empirical evidence
accumulated over the years contributed heavily to the validity of the en-
terprise represented by life course research (Baltes, Reese, & Nesselroade,
1977). The development toward interdisciplinarity needs, however, not only
solid disciplinary foundations and the shared wish to cooperate, but also
Towards an Interdisciplinary Perspective on the Life Course
Advances in Life Course Research, Volume 10, 361–391
Copyright r 2005 by Elsevier Ltd.
All rights of reproduction in any form reserved
ISSN: 1040-2608/doi:10.1016/S1040-2608(05)10013-6
361
hard and time-consuming work in interdisciplinary groups to progress con-
cretely in this direction, possibly along the three lines sketched out in our
introduction: constructing theoretical bridges between disciplinary ap-
proaches, building on common concepts that help describe and analyze
life courses, and working on transversal substantive themes.
In this final chapter, the editors take up the four thematic groups of
contributions, agency and structure, transitions, biographical reconstruction
and methodological innovations, in order to scan them for elements that
seem instrumental for further building up interdisciplinarity in life course
research.
INDIVIDUAL AGENCY WITHIN SOCIAL
STRUCTURE, AND STRUCTURAL AGENCY
Defining agency as the capacity to act intentionally, planfully and reflexively
in a temporal and biographical mode (Marshall, Settersten & Gannon)
points to the fact that the enactment of agency is, among other things, a
cognitive and emotional process, informed by the social environment, and
unfolding through time.2 Following Marshall’s argument, agency should be
distinguished from various related concepts, such as useful resources for
action, social action itself, intentions that motivate behavior, the social and
physical structuring of choices and unexplained variance.
Structure, on the other hand, is usually defined as the set of social con-
straints and opportunities within which individual agency plays out. Con-
tributions to this volume show that structures and agency form a system of
interrelated elements rather than a chain of distinct factors with a clear
causal ordering, for instance, from the macrosocial down to psychobio-
logical levels, or inversely, from cognitive skills upward to social structures.
We conclude from these contributions that social structures have agency of
their own, that agency has structures of its own, and that both should be
jointly looked at when studying life course issues.
To start with the first of these two combinations: social structures, in
some sense, have agency, too. States, as well as firms and other institutions
whose functioning has direct implications for individual life courses, are
shaped by human beings, be they simple members, clients or institutional
decision-makers, with purposes, goals, and agency. The latter, institutional
entrepreneurs in the sense of Eisenstadt (1968), put up programs to instan-
tiate their views about the life course of their citizens, patients or employees.
RENE LEVY ET AL.362
The agency of institutional entrepreneurs is transformed into structures that
both bound and orient the agency of other actors. A promising line of
inquiry in life course research concerns the way in which various institu-
tional actors participate in the structuring of the life course in democratic
societies. European sociology, especially in Germany (for instance, Mayer &
Muller, 1986; Mayer & Schoepflin, 1989) has much emphasized the impor-
tance of the modern State (with all its agencies, including the education
system and the law) and the development of the market economy in this
regard, leading to trends of standardization, individualization and sexual
typification of individual trajectories (Kohli, 1985, 1986; Kruger & Levy,
2001). Social historians show that these changes responded in many cases to
instrumental thinking and action of political leaders, such as Bismarck in
Germany, Lord Beveridge in the United Kingdom, or Roosevelt with the
New Deal (Elder, 1974). Structural changes therefore are often led by in-
stitutional agency, a fairly complex and constrained process that does not
only depend, in democratic societies, on a few institutional entrepreneurs,
but also on various political forces, including social movements – that can
themselves be analyzed in a life course perspective. Following Berger and
Luckmann’s (1966) line of thought, institutional agency may then be ob-
jectivated, i.e., become a constraining social fact, which may later be in-
teriorized by individuals who incorporate it when making life course plans.
On the other hand, agency has structures of its own, both psychological
and social. Narratives usually have a temporal, causal and thematic coher-
ence. Goals and motives are correlated with psychological traits (McAdams).
Life insight of individuals, that is, how to act for one’s own good and the
good of others, is deeply intertwined with personality and cognition abilities.
Agency as a cognitive process is therefore strongly shaped by psychological
characteristics, such as resources of various kinds, as studies of the effects of
personality on memories show (Perrig-Chiello & Perrig). This capacity to act
or react, which developmental psychologists sometimes refer to as resilience
or coping whereas sociologists may rather speak of planful competence (as
Settersten and Gannon remind us), displays various complementary strat-
egies, such as the selection–optimization–compensation or SOC strategies
found in relation with the aging process (Baltes & Baltes, 1990).
Agency seen as a cognitive process matters a lot for social behavior and
social structures throughout the life course. For instance, as Mortimer et al.
show, adolescent goals are transformed into part-time work habits, which
later unfold in distinct ways of entering the job market. Work on the effect
of early childbearing provides similar results. Contrary to social determin-
istic approaches, like the statement that ‘‘The girl who has an illegitimate
Incitations for Interdisciplinarity in Life Course Research 363
child at the age of 16 suddenly has 90 percent of her life’s script written for
her’’ (Campbell, 1968), Furstenberg points to the fact that non-normative
events produce diverse consequences over the life course depending on
mental health, cognitive skills, motivation and flexibility of individuals.
Settersten and Gannon provide examples from various phases of the life
course (childhood and adolescence, early adulthood, old age, etc.) in which
social constraints are dealt with in creative ways by individuals. Thus agen-
cy, as a cognitive process, makes a difference for individual lives.
Agency, however, is not only constrained or enhanced by psychological
and macrosocial factors, but also by microsocial factors unfolding through
individual time. Rather than considering agency as a black box, not to be
opened in empirical analyses, some sociologists propose to see it as path
dependent in a specific sense: narratives, goals, motivations and cognitive
abilities are shaped to a large extent by the actual trajectories themselves
(Abbott, 1992). In this regard, the long-time experience of social psychol-
ogists in dealing with identity formation in social contexts is particularly
useful.
In a second perspective (Emler), narratives, life wisdom and goals are not
constructed by isolated actors: they emerge in connection with narratives,
life wisdom and goals of significant others, in social networks in which
various agencies shape each other. Agency is not only an individual phe-
nomenon, but also a collective one (Settersten and Gannon): studying co-
agency between connected people (especially in gendered relationships) is
addressing the central issue of linked lives (Elder, 1996) in a novel and
interdisciplinary way.
More generally, a fruitful option for dealing with the agency–structure
debate suggested by this volume is to consider both of them as variables in
interaction rather than as main effects to be statistically controlled for. The
concept of ambivalence (Luscher) is especially helpful in this respect.
Luscher suggests four basic ways of experiencing and dealing with inter-
generational ambivalence within family relationships, with specific cou-
plings of agency (the subjective dimension of the model) and structure (its
institutional dimension): solidarity, emancipation, captivation and atomi-
zation. At one end of the spectrum, solidarity is defined as a situation in
which family members feel subjectively committed to the maintenance of
institutionalized patterns of help and relationships, while, at the other end of
the spectrum, atomization corresponds to cases in which family cohesiveness
is neither subjectively nor institutionally assured. According to Luscher, we
may expect the various types of interaction between agency and structure to
be useful for the understanding of turning points in the life course, such as
RENE LEVY ET AL.364
the transition to parenthood, which are associated with both new con-
straints and new roles. Specific life stages, such as old age with its increasing
tension between autonomy and dependency, may be especially conducive to
ambivalence. This may be particularly true if we agree with Settersten and
Gannon’s statement that the life course has become at a time more stand-
ardized and more destandardized in recent decades, a situation that may
lead to increased ambivalence throughout the life course.
Because of the emerging acknowledgement of the intricacy of the rela-
tionship between agency and social structure as variables, there is a growing
need for interdisciplinary approaches that may help to capture the interplay
between agency as a cognitive process with its own set of psychological and
microsocial constraints, and social structures, which are partly the result of
institutional agency. This implies a stronger partnership between life-span
psychology, social psychology, social demography and life course sociology.
As suggested by several contributions, sociologists and social demographers
often overlook the role of personality traits and identity narratives when
explaining individual trajectories, while psychologists, developmental or
social, often disregard contextual factors, in particular those stemming from
the historical context in which individuals are embedded. The contributions
to this volume suggest some ways in which sociologists may unfold the
richness and intricacies of the ‘‘homo psychologicus’’ and include its ele-
ments in their explanatory models of the life course. Likewise, it suggests
how life-span psychologists may include more sophisticated and refined
conceptualizations and measures of the social contexts in their research.
TRANSITIONS AS SOCIAL AND
PSYCHOLOGICAL ‘‘ANALYZERS’’
Life courses go through and are marked by transitions. Transitions and the
stages between them define each other mutually, not only in the formal sense
of stages being bounded by transitions and transitions being inserted be-
tween stages, but also in a more substantive sense: most transitions are being
prepared, anticipated and influenced in many ways during the phases pre-
ceding them, and influence in turn the phases that follow. Transitions may
be ‘‘normative’’ in psychological language or, as sociologists would prefer to
say, modal or statistically predominant, they may also be socially considered
to be desirable, possibly indispensable, or they may be non-normative or
even deviant with respect to social norms, as in the case of the teenage
Incitations for Interdisciplinarity in Life Course Research 365
mothers and divorced women discussed by Furstenberg. Some transitions
belong to the ‘‘normal’’ course of peoples’ lives and contribute to their
regular ‘‘progress’’, while others may throw a life off its rails and become
turning points, biographical accidents provoking life course bifurcations.
Some transitions are relatively smooth, making little ‘‘unrest’’ for the per-
sons directly or indirectly concerned, others, even if ‘‘normative’’ in the
above sense, may have considerable and partly non-anticipated conse-
quences (such as the transition from a cohabiting couple to a family with a
first child; Widmer, Levy, Pollien, Hammer, & Gauthier, 2003b; Widmer,
Kellerhals, & Levy, 2005; Moen & Han, 2001).
Even if it is somewhat reductionist, life course transitions may be seen as
moments of important change, in contrast to the periods between them that
are rather marked by stability, especially if we think of social–psychological
and psychological correlates. These should not be underestimated because,
as we know, individuals are not only socialized actors with ‘‘internal’’ cul-
tural competencies, they are also field- or context dependent in many re-
spects (identity, self-esteem, etc.), even if these dependencies may themselves
vary within (see, for example, Kernis & Goldman, 2003; Nesselroade, 1991)
and between individuals, and also across life phases. For this reason, tran-
sitions are likely to be in many respects more critical for the person than
stable periods.3 As a person’s situation changes, he or she finds him- or
herself in new conditions that are different from the preceding ones, not only
in terms of being more or less advantageous, more or less restrictive for
spontaneous initiatives, etc., but also in terms of their quality or nature. In a
sociological perspective, this is reason enough to postulate a series of con-
sequences that are not only of a sociological order, but concern also the
levels of social-psychological and psychological analysis. The following four
paragraphs are an attempt at formulating interdisciplinary hypotheses along
this line.
A first, highly probable corollary concerns identity, for oneself as well as
for others, as several contributors remind us (especially Marshall and
Emler): after a transition, a person will typically assume other roles, respond
to other expectancies, have other rights and duties and interact with other
people on other terms than before. According to the number and impor-
tance of the role shifts implied in a transition, this change will be more
unsettling or even dramatic, or less. It will affect the person’s self-image, the
representations others have of him or her and also her or his actively per-
formed self-presentation to others. Identity may appear as less settled during
a transition and more put into question, open to various directions and
influences, also to voluntaristic self-influence.
RENE LEVY ET AL.366
Second, taking up one of Kohli’s (1985, 1986) theses, we may ask our-
selves whether on the individual level transitions are also moments of
heightened ‘‘biographization’’, i.e., moments in which the transiting life-
passengers have an increased sense of their being themselves the construc-
tors of their life. It is, of course, also easy to figure out the contrary:
transitions that rather restrict opportunities of action are more apt at elic-
iting sentiments of being passively channeled rather than actively piloting
one’s life. This means at any rate the question of agency will be of particular
relevance in transitions, as particularly well highlighted by Furstenberg’s
comments on the variability of the outcomes he finds for non-normative
transitions, but also by the analyses of Mortimer et al. Do we know enough
about variations of persons’ control beliefs across major and minor bio-
graphical transitions, and about their consequences?
A third, more general hypothetical corollary can be seen as an extension
of the first one: it is more than plausible that throughout the life span, life-
long socialization is not continuous but rather rhythmical, being boosted by
life course transitions and slowing down during the phases in-between, be-
cause in transitions familiar contexts of reference loose relevance and are
replaced by new, less well-known ones. One may extend this argument to
integrate the apparently contradicting ideas about when major phases of
socialization take place in a lifetime: birth and the very first years of life
represent the first biographical transition of a human being, the person’s
first entry into a social field,4 and this field’s active and passive exploration
while being in a particularly fragile – and therefore subjectively highly
significant, affectively ‘‘mobilizing’’ – situation. Consequently, the social-
ization taking place in these circumstances is bound to be particularly im-
pressive and rich in consequences. In comparison, later spells of socializat-
ion, e.g., in adolescence, concern an already (at least partly) structured, i.e.,
socialized person and need considerable social and psychological weight to
supersede or basically relativize the elements already acquired in prior so-
cialization processes; moreover, it concerns an individual that is already
‘‘biographically constructed’’ and becomes, at least potentially, an active
and critical participant in her/his own socialization. We may see these el-
ements as a background to some of Furstenberg’s arguments, based on the
idea that earlier socialization may be part of selection processes (e.g., prob-
lematic socialization outcomes increase peoples’ chances of ending up in
out-of-schedule situations) that create some of the unobserved variation in
results about negative consequences of non-normative transitions, leading
to simplistic or at least hasty conclusions about the negative consequences of
such non-normative transitions in women’s or children’s life courses.
Incitations for Interdisciplinarity in Life Course Research 367
A fourth corollary of transitions is related to the management of the
change they imply. Transitions will be envisaged by the concerned persons
with more apprehension than the stable periods between them, possibly also
with more ambivalence (Luscher) because they represent – to varying de-
grees, of course – potential risks in a person’s biography. The ways actors
themselves and their social environment handle these risks and the subjective
insecurity they entail is an obviously important theme for life course re-
search; this theme can be conceptualized in terms of individual and collective
adaptation or coping. There may be a vast array of forms and resources for
coping related to diverse transitions, some of them specialized, some of them
of a general nature: rituals, other forms of dramatization (including private
ones), emergence of professionalized transition specialists (gate keepers as
well as transition helpers), problem-solving literature for both professionals
and lay persons, repair institutions (such as hospitals, rehabilitation clinics,
training institutions for vocational reorientation, etc.), but also general re-
sources and styles of coping that are not situation-specific.
All these arguments point to the fact that life course transitions are not so
much instantaneous moments of switching from one situation to another –
Bird and Kruger warn us against a ‘‘guillotine-like perception of transi-
tions’’ – but rather processes that may be of considerable duration and
complexity. Transitions, if they are not triggered by unexpected events, are
mostly prepared for and subject to anticipation. They may be composed of
several differentiated changes, and can imprint their reality on the person’s
everyday life progressively. All of this does not happen in a tick, but takes
biographical time. This reason is enough in itself to justify Elder’s (1998)
insistence on the necessity to consider each transition as a series of mini-
transitions or decisional moments; Levinson’s (1990) concept of transitional
periods may often be more appropriate than the simple term of transition;
the same holds a fortiori for the term of event.
Another important aspect is the way in which transitions are embedded in
the whole trajectory of a person: what are the factors that trigger and mod-
ulate a transition to begin with, and what consequences do transitions and
their outcomes have for the subsequent trajectories? How can we theorize the
oft-cited cumulativity of life course developments? Let us mention just one
example. According to research in several – especially European – countries,
the transition from the situation of a cohabiting couple without children
(pre-child phase) to the one of a family with a pre-school child is quite
systematically accompanied by a switch of the couple to a more traditionally
gendered task organization, relatively egalitarian convictions of the partners
notwithstanding (Li & Currie, 1992; Born, Kruger, & Lorenz-Meyer, 1996;
RENE LEVY ET AL.368
Kalicki, Fthenakis, Peitz, & Engfer, 1998; Kalicki, Fthenakis, & Peitz, 1999;
Widmer, Kellerhals, & Levy, 2003a; Widmer et al., 2003b). This
‘‘retraditionalization’’ of the families’ internal structure seems to be quite
resilient in later family phases and is not strongly influenced by the female
partner’s degree of resuming paid work. It may well be that it is precisely in
analyzing closely how transitions are initiated and produced, and how they
produce in turn their consequences, that we will be able to better understand
how the agency-in-structure vision advocated by Settersten and Gannon can
be more finely modeled theoretically.
Several contributions rightly underline the variability and complexity of life
courses and specific transitions (especially Furstenberg and Mortimer et al.)
and of diverse factors to be uncovered behind this variability (Mortimer et al.,
Marshall). Bird and Kruger remind us with a powerful argumentation that
matters are more complex than a merely sequential conceptualization might
suggest, because the complexity of transitions is not restricted to this ‘‘linear’’
aspect. This is what they call ‘‘inline’’ transitions, which have to be completed
by adding to the overall picture also ‘‘competing’’ and ‘‘coupled’’ transitions.
They warn us usefully against several kinds of substantively inadequate
complexity reductions that lie in wait for researchers. We have not only to
take into account the fact that most transitions are not just events and that all
three types of transitions may occur together, but we should also resist the
simplifying ‘‘offers’’ suggested by some easygoing technical terms, such as
formal status definitions that may not fully coincide with the practical reality
of persons’ lives (see their examples of being married while living alone), or
technical terms like event history analysis.
An aspect that has long been at the core of life course analysis is what has
been generally called ‘‘the timing of events’’, including the precise timing of
important events and transitions as well as the duration of the phases be-
tween them, and often also the normative schedules concerning them (age
norms). Furstenberg shows convincingly that we should be much more
circumspect about the social and psychological meaning of transitions being
out-of-time or out-of-order. The fact that a life course does not replicate the
‘‘normative’’ pattern, when such a pattern exists, is ostensibly not enough in
itself to diagnose a problem for the future of the non-respectful life course
passengers. Furstenberg considers a series of additional conditions that are
likely to play a crucial role for such a situation to be socially and subjectively
problematic – or on the contrary enhancing, by way of mobilizing latent
potentials and resources. This is certainly a research area in need of further
conceptual refinement as well as interdisciplinary treatment, and touching
directly on agency, for that matter.
Incitations for Interdisciplinarity in Life Course Research 369
What about transitions in a psychological or social–psychological per-
spective? Stage models, which would seem to be prima vista candidates for
psychological equivalents of social transitions, however prominent they have
been in the developmental psychologies of authors like Piaget, Kohlberg or
Erikson, are not supported by research on cognitive functioning. The topic
may be more promising with respect to identity changes, in line with our
above hypotheses. The weak or non-existent empirical basis of psychological
stage models is also one of the main reason for the absence of stages in
Baltes’ SOC model of life-span development (see next section). This situation
on the side of life-span psychology contrasts strangely with many life course
sociologists’ stressing the timing of life events. The idea of more or less
abrupt endogenous changes is put into question in a significant part of life-
span psychology in favor of a gradual view of age-related changes. In some
respects similarly, life course sociology shows a strong commitment to con-
sidering age as the basic dimension for the study of life-long development,
even if it seems to have some difficulties to conceptualize its meaning
(Settersten & Mayer, 1997); however, more than psychology, sociological
approaches insist strongly on events or transitions. Can these views be
brought together? Maybe a renewed attention to transitions and their var-
ying (and not necessarily close) relationship with age is in order from both
the psychological and the sociological side, developing a stand that is less
‘‘naturally’’ and immediately oriented toward age (without neglecting it) and
theoretically more outspoken about what can provoke change. We shall
push forward on this tread in the section about interdisciplinarity.
PERSONALITY, BIOGRAPHICAL
RECONSTRUCTIONS AND THE LIFE COURSE:
TOWARD A SYSTEMIC AND DYNAMIC APPROACH
Personality and Identity Across the Life Course
As already mentioned, the relationship between agency and structure is a
central theme of the sociological approach to the life course. However, as
often underlined by sociologists, psychologists tend to neglect the impor-
tance of the structural components of the life course. The inverse is also true;
psychologists often consider the theories of sociologists, which usually insist
on the motivation and goal orientation of the individual across the life
course, as oversimplifications with respect to their disciplinary knowledge.
RENE LEVY ET AL.370
Recent advances in the psychological and psychosocial study of the indi-
vidual across the life span are illustrated by several chapters of this volume
and have important implications for the development of a truly interdis-
ciplinary approach to the life course. Putting into perspective different as-
pects of such developments in personality and identity theory will enable us
to better articulate the different contributions of this volume.
Personality as a Framework for Studying Individuals across Time
In the framework of personality theory, there is some consensus about the
important dimensions of personality that must be distinguished and related
with each other. However, there are also fundamental debates, especially
around two questions that have to be tackled seriously by life course
theorists. The first concerns the interplay of biology and personality and the
second the question of the degree to which personality interacts with the
social environment and with trajectories in different life domains.
Since the seminal work of Allport (1937), personality psychology has been
broadly defined as the scientific study of the individual person, and per-
sonality development across the life span has been a central theme. Theories
of personality development across the life span have been proposed on the
basis of the psychodynamic tradition (Sigmund Freud & Carl Gustav Jung,
for example) and developed in an offspring known as life-cycle psychology
(Erikson, 1963, 1968; Havighurst, 1972; Levinson, 1978, 1996; Neugarten,
1977). However, these theoretical frameworks have been repeatedly criti-
cized for lacking empirical support to their theoretical claims.
More recently, alternative systemic models (Hooker & McAdams, 2003;
McAdams, 1996; McCrae & Costa, 2003; Mischel & Shoda, 1998) have
emerged and offer a synthesis of empirical results and theoretical develop-
ments. The model of McCrae and Costa (2003) is known as the Five-Factor
Theory (FFT). This model makes the assumption that our personality is
founded on basic tendencies, which are composed mainly by the Big Five
dimensions of personality (traits of neuroticism, extraversion, openness,
agreeableness and conscientiousness), but also by sexual orientations, some
cognitive abilities and artistic talents. These basic traits appear to be fairly
stable across the life course, influenced mainly by biological factors (some
changes appear but are described as predictably linked to age; see Srivatava,
John, Gosling, & Potter, 2003), and to influence the other two components
of personality across the life course, namely, the characteristic adaptations –
personal strivings, attitudes, worldviews, strategies or processes of coping
Incitations for Interdisciplinarity in Life Course Research 371
and adaptation (Baltes & Baltes, 1990; Brandtstater, Krampen, & Heil,
1993; Heckhausen, 1999) – and the self-concept system (self-schemas, iden-
tity, personal myths, life narratives). Contrary to the basic tendencies, these
two elements are based on learning and are subject to external influences like
cultural norms and life events. Finally, we find the objective biography,
which represents the main observable outcome of basic personality tenden-
cies and characteristic adaptations.
McAdams (1996 and this volume) proposes a comprehensive model of the
person distinguishing three levels (traits, personal action constructs or char-
acteristic adaptations, and the life story) that has been further elaborated in
a more recent version by Hooker (2002) and Hooker and McAdams (2003).
These authors’ new model, named the Six-Foci Model of personality (SFM),
is based on three structural components and three process-related compo-
nents (processes producing corresponding structure components). The three
structural components correspond to the three levels of personality already
described by McAdams (1996). The corresponding processes are respectively
states (emotions, moods, hunger, fatigue, anxiousness), self-regulatory
processes (especially primary and secondary control; see Schulz &
Heckhausen, 1996), and processes of self-narrating (remembering, reminis-
cence and storytelling). This new systemic model of personality, in comple-
ment with the FFT model, allows for a heuristic description of the aging
person and focuses researchers’ attention on the self and its relations
to contents (mainly present-oriented), goals (mainly future-oriented) and
reconstructions of biography (mainly past-oriented).
Personality, Identity and the Social Environment
Some of the main issues tackled in this volume concern, on the one hand,
relationships between different areas of personality and identity, and on the
other, how dynamics in personality and identity are related to dynamics in
the social environment. The relations between the different components of
the personality and identity systems are very complex; FFT and SFM mod-
els have different views in this respect. FFT postulates a causal chain of
influence starting from biology and traits, mediated by the experiences and
learning of the individual across the life span, which finally results in its
biographical trajectory. Clearly, here, differences in traits between individ-
uals are thought to influence the other dimensions of personality (identity
and characteristic adaptations). Perrig-Chiello and Perrig exemplify this
kind of model; they argue that the relationships between well-being and
RENE LEVY ET AL.372
autobiographical recollection or episodic memory are indeed influenced by
personality traits like extraversion, conscientiousness and neuroticism, as
predicted by FFT. However, as in other studies, the direction and strength of
this influence of traits, as well as its variability across domains of the per-
sonality structure, remains an open question (Mischel, 2004). Moreover, in
order to evaluate more precisely the relationships between life trajectories
and personality or identity structures and process, two additional issues
need to be developed in future research. The first is the necessity to combine
different methodological strategies for life course research (nomothetic–
idiographic; qualitative–quantitative, individual versus aggregated hierar-
chical data, etc.), with special attention to longitudinal designs. The second
is the need to articulate different levels of analysis in order to better
understand the relationships between the ‘‘objective’’ and ‘‘subjective’’
biography and between the individual, relational and collective levels. In
Perrig-Chiello and Perrig, for example, transitions and life events are re-
corded on the basis of memory, and the social context is simplified to a small
number of general variables (mostly gender and education). Sociologists and
social demographers would wish to include more detailed information on
the social position of individuals and on the institutionally monitored life
transitions. Life trajectories cannot be assessed only with individuals’ mem-
ory-based recollection as autobiographical memory, even in such areas as
job history or parents’ occupations (see Scott & Alwin, 1998), because they
are subject to memory biases and to well-known processes of reconstruction
that Perrig-Chiello and Perrig interestingly also develop in their chapter.
The chapter by Emler shows that the relationships between identity
changes and life transitions are indeed complex and should be modeled
more explicitly. On the basis of social-psychological research, he shows that
identity development does not correspond to a general pattern of qualita-
tively different hierarchical stages across the life span as postulated in the
Piagetian and in particular in the life-cycle tradition. His results indicate
that self-categorization processes are related to the social environment in
which they take place. Consequently, the content and processes related to
the self are primarily associated with time and location, two central dimen-
sions of the life course approach (Settersten, 1999). Emler shows also that
social identity is related to social relationships and locations, and that
changes in social identity (related to life transitions or turning points) are
often followed or anticipated by geographical moves leading to changes in
the social participation of individuals.
This systemic relationship between social identity (defined and construct-
ed through interaction with relevant others in different settings and types of
Incitations for Interdisciplinarity in Life Course Research 373
relationships) and personality on the one hand and the social and cultural
context on the other is also at the heart of McAdams and colleagues’ work,
as illustrated in their research about generativity in midlife. Generativity is
a central theme of identity, which is in turn related to a wide range of
engagements within society. It is in direct interaction with the social envi-
ronment, in which individuals continuously reformulate their identity sto-
ries. The life-stories orientation reminds us of an essential element in the
way individuals and life stories are related: meanings must not be ignored
when considering how the individuals interact with their social environment.
It is through meanings that individuals are related to their social networks,
to their engagements in different life spheres and to their life trajectories
(past, present and future). It is also through shared meanings or social
representations that individuals are related to ideologies and their socio-
cultural context (see, for example, the observations of Marshall on how
different institutions may create different stories and ideologies about death
and dying; see also Chryssochoou, 2003). As such, the two socialized com-
ponents of personality (characteristic adaptations and identity) are in con-
stant interaction with the social environment and biography. If individuals
are indeed active in the construction of the meaning they attribute to their
life, one conclusion of the chapters discussed here is that the social context
participates directly in the definition of the personality development of in-
dividuals by way of communication processes (socialization, social influ-
ence, conformity) and through the effects of non-normative events that
induce specific social–psychological processes of adaptation and coping
(e.g., in the sphere of health, Taylor & Brown, 1988; Taylor, Kemeny, Reed,
Bower, & Gruenewald, 2000). In this regard, social psychologists in some
way depict a more complex individual life course than the motivated actor
often seemingly referred to by sociologists.
To sum up, personality and social psychologists have developed models
like FFT and SFM that share the idea that three components should be
considered in any global comprehension of personal development. These
include, at the structural level, personality traits (relatively stable and struc-
tured early in the life course), psychosocial regulations (goals, secondary
control, coping processes, etc.) and identity (life story, self-categorization
processes, etc.). The relationships among these three components (and their
associated processes), and especially their relationships with the social
structure and the structure of the life course, are still in question. However,
personality psychology and social psychology have now developed a com-
prehensive model that should be useful for the interdisciplinary life course
perspective.
RENE LEVY ET AL.374
METHODOLOGICAL AND DATA-ANALYTICAL
APPROACHES
Collection of Life Course Data
Following Scott and Alwin (1998), we may distinguish two important issues
regarding the question of life-history data. The first is about what should be
measured, i.e., what kind of data to collect, while the second is about how to
do these measurements; it refers to data design.
Concerning the first, Scott and Alwin distinguish between three kinds of
measures in life course research: events, experiences and meanings. The term
of event refers to the collection of event data in different life domains (e.g.,
family, professional career) with the aim to analyze chronologies, sequences
or interactions between life events. This corresponds to the definition of a
life history given by Elder (1992): ‘‘a lifetime chronology of events and
activities that typically and variably combine data records on education,
work-life, family and residence’’ (p. 1122). Events are typically investigated
by demographers (Billari; Oris & Ritschard), to some extent also by soci-
ologists. The second approach measures cumulated experiences during an
individual’s past in order to analyze his/her present situation or his/her
expectations about the future. For example, the work experience of an ad-
olescent is considered as a predictor of the transition to the labor market or
to high school, as shown by Mortimer et al. The third kind of measure
focuses on the meaning or the evaluation a person attributes to her/his past
trajectory. This past can be contrasted with the present situation or future
expectations. The approach to narrative identity proposed by McAdams
exemplifies research collecting meanings of life courses.
Some contributions to this volume, and more generally most examples in
the literature, suggest that a fourth type of measurement in the life course
approach has to be considered, referring to the context or the institution-
alization of life courses. Data on context can be succinct information used as
a complement to life-event data and helps to depict the structure of con-
straints and opportunities in persons’ social environment. This kind of in-
formation is illustrated by Bird and Kruger by their taking into account the
legislation on motherhood leaves in Germany that influences the duration of
career interruptions. Context data include also interviews of life course
agents, i.e., persons who are present or give support at a specific stage or
transition in the life course, as developed by Marshall’s analysis of the
transition to death. One methodological difficulty is the integration of these
contextual informations and life course histories.
Incitations for Interdisciplinarity in Life Course Research 375
Concerning the second issue, two types of data collection design are cur-
rently used in life course research. The first one is the cross-sectional col-
lection of retrospective data. The reconstruction of individual life courses
with archive or administrative data forms a first subtype of retrospective
data. For historians interested in somewhat remote periods, this is practi-
cally the only possible strategy; its main disadvantage is that these data were
not originally collected for research purposes (Oris & Ritschard); research-
ers using them have no other choice than to accept them – and their lim-
itations – as ‘‘given’’ (Billari).
The second subtype of retrospective data is generated by surveys in which
persons are interviewed about events or experiences in their past. The quality
of data collected in retrospective surveys depends strongly on the memory of
respondents. Tools like life event history calendars minimize the potential
biases (Freedman, Thornton, Camburn, Alwin, & Young-DeMarco, 1988;
Belli, 1998). Retrospective data are often interesting to collect in order to
analyze the influence of a historical event or of a specific historical period
(economic crisis, war, etc.) on life courses. However, various studies on the
impact of a historical event on trajectories show generally that this impact
varies according to the situation or the stage in the life course where persons
are interviewed. A range of samples allowing to compare the impact of such
a historical event on trajectories in different generations, i.e., of people born
at different dates can be used in order to analyze this impact in terms of
cohort effects (Ryder, 1965). This strategy of combining the choice of design
and of sample characteristics is also appealing to analyze the effects of con-
textual changes (Bird & Kruger). However, retrospective data present several
limitations, the main one being that only factual data can be collected. This
implies that only information on events and experiences can be analyzed.
The other longitudinal design is the repeated collection of prospective
data. In psychology, this type of methodology is usually referred to as
longitudinal design, in other disciplines of the social sciences as panel design.
In this case, the same measures are applied to the same sample of persons at
various points in time. This kind of data collection allows taking into ac-
count current experiences as well as intentions about the future. This enables
life course researchers to confront intentions with their realization in sub-
sequent waves of the survey, or to investigate how expectations evolve
across the life course (Nurmi & Salmela-Aro, 2000). Panel surveys are es-
pecially interesting for research centered on agency. Several defaults of the
panel design have also been underlined, especially the attrition of cases
across succesive vawes of the panel or the risk to break time series if ques-
tions or experimental modalities are changed underway (MacArdle).
RENE LEVY ET AL.376
What might then be the best strategy for data collection in interdiscipli-
nary research? As mentioned above, events are the research topic favored by
demographers, historians, and, to a lesser extent, sociologists. Experience is
a topic common to all social science disciplines, but it is probably most
prominent in sociology and psychology. Meanings are a research topic of
particular interest to social psychology, but also to sociology. The necessity
to collect context data is more often invoked by demographers or socio-
logists than by psychologists. A preliminary condition of interdisciplinarity
is then to combine measures of events, experiences, meanings and context.
Moreover, a panel design allows to take into account each topic, especially
when it is completed by a qualitative survey. This type of panel design is also
especially interesting to analyze micromechanisms during a transition or a
change in the life course. It should be noted that retrospective surveys allow
also doing interdisciplinary work.
ANALYSIS OF LIFE COURSE DATA
Recently, certain data-analytical tools have come to play a critical role in
advancing knowledge about the life course. If integrated and combined,
these tools promise further interesting applications. We can identify four
general families of such tools that we consider especially helpful to study the
life course in an interdisciplinary perspective. While some of these tech-
niques have developed in parallel in different disciplines, others did so
mainly within a given discipline and may hence be less known by scholars of
other research fields. Life course scientists have a vital interest to get famil-
iarized with these methods even though some of them may seem rather
exotic at first sight. Some of them are directly treated by the contributions in
this volume, others are only hinted at, so it may be helpful to give a struc-
tured overview of this rapidly evolving array.
The first set of data-analytical tools we would like to mention falls under
the label of structural equation modeling (SEM) and has been adopted es-
pecially in psychological research. This set of techniques aims at explaining
the interrelationships observed among a set of chosen variables, usually by
the means of a correlation matrix. One of the major advantages of SEM
is that the researcher has complete freedom as to how to represent the
structure of the data. The researcher adopting SEM must define what the
underlying structure accounting for the interrelationships within the data
might be and translate that hypothesis (or series of hypotheses if several
alternative specifications are formulated) into a testable and rejectable
Incitations for Interdisciplinarity in Life Course Research 377
model, to be tested against the data at hand. This set of techniques further
has the desirable property of partitioning the variance of single variables or
indicators into a portion said to be common among the chosen variables,
hence representing latent constructs that cannot be observed nor measured
directly, but are thought to influence the measurable properties of the var-
iables, and a portion said to be unique to each variable. Hence, measure-
ment problems may often be circumvented with the application of SEM.
Some promising advanced applications of SEM to the study of the life
course are illustrated by McArdle’s contribution. There, changes between
adjacent repeated measures of a set of variables assessed on the same in-
dividuals are defined as latent differences, so that occasion-specific meas-
urement threats may be isolated and eliminated from the important
information gathered on the individuals. One may then test the existence
of variance in latent difference scores, which translates into interindividual
differences in change. Once differential change is established, correlates and
antecedents of change may be tested, one of the major goals of longitudinal
research (Baltes & Nesselroade, 1979).
A second set of techniques is that of event history or survival analysis
(EHA), adopted most heavily in demography (Billari and Ritschard & Oris).
The basic question motivating the application of EHA is what affects the
probability of the occurrence at a given time of a specific event (e.g., mar-
riage, birth of the first child, onset of a disease, death). Unlike SEM, a so-
called survival model does not have to be specified by the researcher. The
most popular example of EHA indeed is Cox’ proportional hazards model,
in which a semi-parametric hazard function provides very reasonable esti-
mates of the influences exerted by chosen covariates on the occurrence of the
event scrutinized. Particularly appealing is the test of time constancy of each
predictor (in practice unfortunately often overlooked). By introducing in the
model not only chosen predictors, but also their interactions with the un-
derlying time dimension, it is possible to test whether the potential effect of a
predictor holds across time or is manifested only at certain time periods.
Hence, what is believed to be an important predictor of the occurrence of an
event can be tested for its temporal relevance, much in the vein of what Bird
and Kruger remind us, namely that life course scholars ought to pay close
attention to the time dimension. Moreover, the regularity of time-ordered
events may also be assessed with this set of techniques.
A third set of techniques is that of multilevel models (MLM), also known
as random effects, mixed effects, and hierarchical linear models. This set
of techniques more than any other recognizes the possibility that the data
under inspection are structured according to either pre-defined (and usually
RENE LEVY ET AL.378
hierarchical) organizations (such as households within neighborhoods and
family members within households) or to configurations that were not
planned, but nevertheless resulted because of empirical research contingen-
cies (e.g., multistage sampling). Such situations do not meet the basic as-
sumption validating results from ordinary linear regression that the units of
observation are independent of each other, and ignoring this structure of the
data usually results in biased standard errors of the parameters (Goldstein,
1987). MLM can hence be conceived as regression equations with not one
source of variance (usually referred to as the errors, or the residuals), but
several sources of variance, the sources themselves being organized according
to the structure of the data (e.g., a first error may be associated with members
within a household and a second with households within a neighborhood).
Hence, variables at the individual and at the contextual levels can be modeled
together, making sure their distinctions are properly respected and not ar-
tificially removed.5 This approach is particularly promising for enriching the
‘‘agency and structure’’ discussion previously presented. Indeed, variables
measured at the individual (i.e., agentic) level can be analyzed in concom-
itance with variables assessed at the contextual (i.e., structural) level. More-
over, the so called cross-level interactions may be defined that allow for the
estimation not only of main effects at the individual and contextual levels,
but also of their interaction. It is this interaction between agentic and struc-
tural variables that most often motivates life course scholars.
A fourth, promising analytical approach is represented by exploratory
analyses of repeated measures data. These techniques are newer but are
quickly gaining popularity in life course research, thanks to their capacity to
address not only quantitatively but also qualitatively motivated questions.
Ritschard and Oris discuss Markov transition models and longitudinal data
mining. Both techniques are concerned with sequences of events that do not
need to follow specific assumptions. Mathematical rules can be established to
explain the probability of switching from one event to another and the effect
exerted by chosen covariates on these switches. While Markov models are
parametric, data mining is non-parametric and aims at deriving
association rules among the most typical sequences and their frequencies.
A typical practical application is that of online bookstores, where custo-
mers purchasing item A are notified of similar items bought by preceding
customers who also purchased item A. Another promising longitudinal ex-
ploratory technique is that of optimal matching (Abbott, 1995), of which Bird
and Kruger remind us in their contribution. This technique, unlike EHA,
is not focused on a specific event, but aims instead at producing typical
sequences, i.e., in the case of life courses, longitudinal constellations of states,
Incitations for Interdisciplinarity in Life Course Research 379
allowing to study whole trajectories. The assumptions of the states are min-
imal, so that their complexity may drive the synthesis (Widmer et al., 2003b).
Through pairwise comparisons this inductive method computes the minimal
distance separating each pair of individual trajectories available in the sample
according to pre-defined ‘‘costs’’ needed to transform one trajectory into
another. The resulting distance matrix may then be fed into a clustering
procedure to obtain groups as a function of their shared types of trajectories.
These four families of techniques6 have developed separately, with little or
no links between them. In recent years, however, methodological advances
have allowed to combine appealing features of these methods to benefit
further from their applications (Billari, Ritschard & Oris, and McArdle).
Examples include the combination of EHA and MLM (Ritschard & Oris)
that allow for conditioning the probability of the occurrence of an event on
information at different levels of the data organization. Similarly, the recent
addition of SEM latent variables in EHA refines the measurement properties
of predictors, so that their effects are less attenuated by unrelated variance
such as error. SEM and MLM have also recently been combined to provide
for another methodological synergy (Ghisletta & Lindenberger, 2004;
Rovine & Molenaar, 2000). Here, the advantages of SEM with respect to
measurement properties are joined with the power of MLM to disentangle
effects stemming from different, hierarchically organized sources of variance.
Methodological refinements have contributed much to the advancement
of not only empirical but also theoretical knowledge about the life course. At
the same time, the methodological and data-analytical techniques have ad-
vanced themselves, geared as they have become to address further theoretical
questions raised by life course scholars. The combination of now well-
established techniques and the quickly evolving field of linear and non-linear
dynamical systems will further contribute to paving the road of life course
research. We believe that a fundamental ingredient for successful life course
research is the presence of continued synergetic communication between the
different disciplines concerned. Good data-analytical tools have emerged in
each discipline, and we are confident that yet better tools will be developed
by combining discipline-specific existing techniques. These ameliorations are
instrumental to acquire deeper knowledge of life course phenomena.
INTERDISCIPLINARITY
After having sifted through the four sections of this volume with a view to
common or interacting themes between its contributions, let us take up the
RENE LEVY ET AL.380
three axes we propose for developing interdisciplinarity in life course
research, common concepts, theoretical bridges, and transversal themes –
what elements can this volume contribute to each of these?
We have to realize that common concepts and transversal themes are
closer to each other than both are to theoretical bridges, since we almost
inevitably use concepts to refer to themes. This allows us to treat both
aspects together. Common concepts have to develop from interdisciplinary
work on transversal themes – they need not be identical from the outset,
otherwise interdisciplinarity would be restricted to the rare instances where
the same concept is used in more than one discipline (examples could be
socialization or coping), and even if the words were not the same, inter-
disciplinary connections would boil down in such cases to simple termino-
logical translation (as in the case of ‘‘lifespan’’ and ‘‘life course,’’ or of the
different names used for some data-analytical methods). Conceptual com-
parison and elaboration become interesting if concepts are not the same
between two or more disciplines, but substantively close enough to favor
interdisciplinary exchange and elaboration, which is most likely to be fruit-
ful when starting from common themes.
Among the candidates for becoming common or transversal themes that
have emerged in this volume, let us single out socialization, age, identity
change and life course transitions, agency, coping, and gender. In varying
configurations, each of these substantive areas promises for interdisciplinary
collaboration to bring about mutual enrichment, greater strength
of analytical grip, and more complete understanding of life course phe-
nomena, especially if we start from the principle to look first at inter-
dependencies rather than causal chains because the latter would fix a priori
hierarchies of causation between the disciplines. Socialization is perhaps the
most traditionally common area, at least between psychology, social psy-
chology and sociology, because the three disciplines consider it to be one of
the most basic processes of personal development; the particular perspec-
tives developed by each of them is a crucial source of complementarity.
The situation is different for age, the analytical status of which is clearly
less obvious and more controversial. But then, interdisciplinary discussion
of this situation should lead to spell out more explicitly the discipline-
specific perspectives and assumptions, facilitating to arrive at a more en-
compassing and articulate conceptualization of the phenomena concerning
age and the related processes. As an illustration, let us elaborate somewhat
further on this theme. In psychology, the influential article by Wohlwill
(1970) pinpointed this problem. This author characterized the status of age
as a convenient, descriptive and data-organizing tool that lacked however
Incitations for Interdisciplinarity in Life Course Research 381
theoretical meaning. He urged scholars to investigate the aspects of behavior
that might be lawfully related to age. Among the few examples embracing
this deeper analysis of the variable of age we can cite the search for markers
of biological age (or biomarkers; e.g., Anstey, Lord, & Smith, 1996), the
opposition of chronological age to other time definitions that are theoret-
ically better justified in the light of the phenomenon under analysis (e.g.,
time left to the onset of pre-clinical dementia when studying memory per-
formance in older adults; Sliwinski, Hofer, Hall, & Buschke, 2003), and,
from a more analytical point of view, transformations of age in relation to
the variable considered in order to better understand the age-related mech-
anisms (e.g., McArdle, 1986). In much the same vein, sociologists Settersten
and Mayer (1997) have advised, ‘‘chronological age itself is an ‘empty’ var-
iable...it is whatever age presumably indexes that is thought to be impor-
tant.’’ The major difference between psychological and sociological uses of
age in the life course/life-span field is probably that in developmental psy-
chology, it indicates changes implied in forms of physical and cognitive
maturation and physical or physiological aging, i.e., biological age, whereas
sociologists think in terms of social age, again with several specific meanings
that are not always clearly distinguished, especially in the sense of age norms
versus the more structural sense of specific roles or role sets that define the
way individuals are embedded in the social world. In sum, life course schol-
ars more than others are urged to move the age variable from the right side
to the left side of the equation: age should not be used to explain behavior,
but should itself become subject to analysis. A further hint at a socio-
logical contribution to the substantive interpretation of age can be seen
in Kohli’s (1985) discovery of biographical chronologization as a rather
recent historical process, related to what Weber already analyzed as
the growing bureaucratization and rationalization of modern societies. Ac-
cording to Kohli, the sequencing and temporal ordering of modern life
courses that has taken place in the last 2–3 centuries is explained mainly by
the structural changes brought about by modernization, reliance on age for
the legal and procedural attribution of a series of rights and duties corre-
sponding to what Weber called bureaucratical rationality (among such in-
stitutional innovations, let us mention the implementation of compulsory
scolarization of all children along with the prohibition of children’s
employment, the fixation of a series of legal age thresholds, and various
welfare-state regulations with life course incidences, be they age related or
not). Seen from this vantage point, the more or less regular timing of some
crucial events in modern life courses, especially ‘‘normative transitions’’, no
longer appears as a definitional element of life courses as such, but as one of
RENE LEVY ET AL.382
the various ways in which they may be socially institutionalized and stand-
ardized. Facing this empirical situation, we turn out to be rather poor in
theoretical tools permitting to conceptualize such findings – take the pure
timing dimension away and see what remains in terms of life course an-
alytical tools! Not much, at least in the conceptual traditions we mentioned
up to now; we shall come back to this question with respect to conceptual
bridges.
For identity change and life course transitions, we have proposed some
hypotheses in an interdisciplinary perspective; this may be somewhat of a
test area for our postulate of a priori symmetry between the disciplines –
sociologists as well as demographers will certainly have a tendency to as-
sume that social regulations trigger identity changes rather than the other
way round, but a more agentic and also a (social-) psychological perspective
will want to consider with equal interest the possibilities of active individual
construction of transitions. Again, the various disciplines can only gain at
working together in this area because they have developed different and
potentially complementary conceptions of an issue they share, but that goes
beyond their specific horizon.
Let us pass more summarily on the remaining examples of transversal
themes. Agency as a theme is almost per definition a meeting place for
psychological, social–psychological and sociological aspects, including cog-
nitive, affective, cultural, behavioral and interactional dimensions; coping
may be less common a topic for demography and sociology than for the
other two disciplines considered here, but its importance has long been
acknowledged there, too. Finally, gender may need less insistence on the
fruitfulness of an interdisciplinary approach than any of the other themes,
but it should be underscored that probably few other perspectives than the
study of life courses are equally apt at highlighting processes of gendering,
and notably not only on the level of interindividual doing gender, but also
on an institutional level – provided, of course, research is done in a gender-
sensitive way (Eichler, 1988).
What about theoretical bridges? We may distinguish between two levels of
theorizing, a Merton-like level of middle-range theorizing that corresponds
to the concepts used in subject areas like those we have just mentioned, and
a more metatheoretical or abstract level of general conceptual thinking. One
track of interdisciplinary development on this second level is indicated by
the recent FFT and SFM models in social psychology mentioned above.
They provide explicit entry points for interactions with the person’s social
environment as conceptualized, e.g., by sociological approaches – an aspect
they share with Bronfenbrenner’s principle of ecological psychology or
Incitations for Interdisciplinarity in Life Course Research 383
Erikson’s opening for social aspects entering into the stage-defining dilem-
mas of epigenesis he postulates, but their advantage can be seen in their
firmer grounding in empirical research.
This theoretical ‘‘offer’’ from personality psychology may be seen as a
major anchor point for building conceptual bridges between psychology and
sociology – let us try to develop, as a counterpart, a sociological anchor
point, taking up the question of how and in which ways the link between
specific role sets and their corresponding ages is socially constructed, espe-
cially on the level of macro- and mesosocial institutions. There is a prom-
ising track in sociological thinking, indicated, for example, by Kerckhoff
(1993, p. 13) in a passage where he thinks about concepts for analyzing
professional careers: ‘‘...we need to chart the movementyof individuals over
time as they pass through a number of stages in the life course and occupy
positions within hierarchically structured social organizations. At each
stage, we need to be able to identify a set of locations that are hierarchically
ordered, and we need to measure the personal characteristics of the indi-
viduals occupying those locations. yCharting the flow of individuals be-
tween structural locations across stages in the life course constitutes
describing the ‘careers’ of those individuals, (i.e.,)ythe pathway (they fol-
low)ybetween positions in the social structure occupied at different points
in the life course.’’7
In a more general stance, we can define this structural part of the life
course as a movement through social space. Taking into account the basic
understanding that social space is generally organized in relatively well-
defined and well-delimited social fields (Lewin, 1935; Bourdieu, 1980) and
that – due to multiple participation – our social participation is generally
definable by status/role sets or profiles rather than by single statuses and
roles, we can reformulate this heuristic definition of the life course as a
sequence of participation profiles (Levy, 1977, 1991).8 The linking to age of
specific features of such sequences, especially of specific transitions between
subsequent participation profiles, for instance, on the basis of cultural age
norms (Neugarten, Moore, & Lowe, 1965; Settersten, 1997) some of which
may be more officially institutionalized (Kohli’s chronologization), is but
one of the multiple ways of life course standardization. The sequential or-
dering of a series of institutional participations (Kruger, 2001, Bird &
Kruger) is another one. It is, however, important to stress that life course
standardization is not a necessary ingredient of this analytical vision, it is
one of its variable dimensions. Analyzing life courses as sequences of par-
ticipation profiles, e.g., in a social structural perspective, enables us to more
fully characterize the mechanisms that relate biographically changing social
RENE LEVY ET AL.384
participation to ongoing time without allowing the social structural aspects
to go unobserved or to remain poorly defined (Settersten, 2005).
Turning back to articulating the sociological and psychological perspec-
tives, we may then ask to what extent psychological development and aging
(cognitive, emotional, moral, identity), besides its ontogenetic ‘‘push fac-
tors’’, can also be shown to develop according to a threshold rather than
gradual model that would, however, not express endogenous stages, but
materialize in relation with transitions in participation profiles, and more
specifically to what extent some changes of participation profiles, rarely
studied in sociology, could have a psychological impact, e.g., profile exten-
sions or restrictions in the sense that the number of simultaneous partici-
pations before and after a transition grows larger or smaller.9
Another substantive theoretical axis that may help articulate or integrate
conceptual contributions of different disciplines can be seen in the distinc-
tion of various system levels, once different usages of some terms are spelled
out and possibly agreed upon (e.g., what constitutes micro- and macrolevel
phenomena differs dramatically between psychologists and sociologists). It
would certainly be simplistic to assign to each discipline its proper system
level, thus restraining it to the level in question (as Devereux, 1967, once
proposed with his ‘‘complementaristic’’ conception of ethnopsychoanalysis).
One discipline can legitimately study several such levels, some of which are
also focused by other disciplines. This is especially evident for the subject
matters shared by social psychology and psychology on the one hand, and
sociology and social psychology on the other, but also and probably even
more so of sociology and social demography. In this perspective, Settersten
and Gannon’s formulation of agency within structure is not only a formula
to overcome the artificial opposition of these complementary aspects of
social reality, but to integrate more generally individual- and social-level
explanations. Even if the classical distinction between structure and agency
may have a singular ideological and political interest, it can be seen as a
special case of this more general dimension of system-level differentiation.
Spelling out and studying the mechanisms that relate different system levels
with each other remains one of the lesser studied and poorly conceptualized
areas in all concerned disciplines, and calls intrinsically for inter-
disciplinarity (Diewald, 2001). Saying this, we should also note that this
volume (and probably most similar ones) does not cover all the possible
system types and levels that are relevant for life course research. The social
sciences certainly have to develop more explicit and integrative theoretical
and methodological interfaces toward the biological system and its processes
of ontogenesis on one hand (Shanahan, Hofer, & Shanahan, 2003) and, on
Incitations for Interdisciplinarity in Life Course Research 385
the other, toward historical and cultural processes as well as actual processes
of institutional framing that intervene heavily on the life course (see Elder,
1974; Heinz, 1992; Kruger & Levy, 2001).
Let us end this volume with a very general statement about inter-
disciplinarity. Once we are convinced of interdisciplinary work as a superior
way to study life courses, we have to define the ways we want to ‘‘mix’’ the
conceptual tools our various disciplines can bring to bear on the topic. One
basic principle has already been stated, it is what we could call the principle
of a priori disciplinary symmetry. A second basic principle should inform
our attempts, we could call it an articulation principle: the various disci-
plines cannot and should not be simply melted into some unifying theo-
retical mold, given that they focus on complementary aspects of the same
phenomena as well as on different and complementary levels of the systemic
organization of reality. For that reason, to develop its potential, their sci-
entific synergy needs explicit articulations, not fusion, and even less con-
ceptual vagueness. The way leading to this goal is certainly difficult and
laborious; we hope that this volume is more dynamic than the proverbial
signpost that points to a direction without going there itself, i.e., that we
have not only been able to point out some promising tracks, but also to
undertake some real steps in that direction.
NOTES
1. Throughout this final chapter, all references to contributing authors concerntheir contributions to this volume.2. Let us recall Marshall’s distinction between the principle of agency as an in-
gredient of human nature and the practical competence and possibility to act as avariable characteristic. Freire’s (1972) concept of conscientization can be cited asaiming at empowering oppressed individuals in order to give them the practicalcapacity to actualize this latent part of their being humans.3. ‘‘Critical’’ does not necessarily mean threatening and even less unexpected – the
following considerations are not limited to ‘‘non-normative’’ transitions or events.What is implied here is simply the idea that compared to the more or less stableperiods between transitions, the latter tend to entail to a larger extent processes ofchange and adaptation on various levels. It goes without saying that this postulate isnot meant to exclude the possibility of important processes and consequences at-tributable to what goes on during the phases between transitions.4. We use this term in a rather general sense, much as it was already used by
Lewin (1951) in his field theory approach where he defined a field as the totality ofsimultaneous and interdependent features that form a situation; however, we wouldinsist more than he did on the systemic character of social fields (see also Bourdieu,1984).
RENE LEVY ET AL.386
5. Two methodological problems that may be avoided by properly applyingMLMs are the atomistic and the ecological fallacies, which incorrectly assign groupeffects at the individual level or, vice versa, individual effects at the group level,respectively.6. A fifth, particularly promising set of techniques that life course scholars have an
interest to adopt is that of linear and non-linear dynamical systems (e.g., Kaplan &Glass, 1995; in this volume, it has only been alluded to by McArdle). In this large setof techniques, the outcome of interest is not the value of a variable at a given timepoint, but the change in that variable over time. These methods presuppose that thehuman organism can be conceived as an open living system, continuously fluctu-ating, achieving a dynamic equilibrium through self-organizing structures tendingtowards homeostasis (not limited to its physiological meaning). These techniquesprovide a way to characterize many of the basic phenomena of development, in-cluding change, stability, variability, stages, continuity, and the combination ofquantitative and qualitative change as well as the emergence of new forms of struc-ture and function.7. Similar ideas can be found in the thinking of Rosow (1976).8. Without neglecting the fact that participation in a social field implies also
holding a position in the field’s internal structure and assuming the correspond role.All these aspects have multiple implications that may be of importance when anal-yzing life course sequences, but cannot be spelled out here.9. Among the rare studies of this aspect, let us cite Thoits (1986) and Moen,
Dempster-McCain, & Williams (1989, 1992) who show beneficial effects of multiplerole occupancy on women’s health. In this perspective, retirement corresponds to amajor profile-narrowing.
REFERENCES
Abbott, A. (1992). From causes to events: Notes on narrative positivism. Sociological Methods
& Research, 20, 428–455.
Abbott, A. (1995). Sequence analysis: New methods for old ideas. Annual Review of Sociology,
21, 93–113.
Allport, G. W. (1937). Personality: A psychological interpretation. New York: Holt, Rinehart &
Winston.
Anstey, K. J., Lord, S. R., & Smith, G. A. (1996). Measuring human functional age: A review of
empirical findings. Experimental Aging Research, 22, 245–266.
Baltes, P. B., & Baltes, M. M. (Eds) (1990). Successful aging: Perspectives from the behavorial
sciences. New York: Cambridge University Press.
Baltes, P. B., Lindenberger, U., & Staudinger, U. M. (in press). Lifespan theory in develop-
mental psychology. In: R. M. Lerner (Ed.), Theoretical models of human development
(Vol. 1 of the Handbook of child psychology). New York, NY: Wiley.
Baltes, P. B., & Nesselroade, J. R. (1979). History and rationale of longitudinal research. In:
J. R. Nesselroade & P. B. Baltes (Eds), Longitudinal research in the study of behavior and
development. New York: Academic Press.
Baltes, P. B., Reese, H. W., & Nesselroade, J. R. (Eds) (1977). Life-span development meth-
odology: Introduction to research methods. Monterey, CA: Brooks-Cole.
Incitations for Interdisciplinarity in Life Course Research 387
Belli, R. (1998). The structure of autobiographical memory and the event history calendar:
Potential improvements in the quality of retrospective reports in surveys. Memory, 6(4),
383–406.
Berger, P., & Luckmann, T. (1966). The social construction of reality: A treatise in the sociology
of knowledge. Garden City, New York: Doubleday.
Born, C., Kruger, H., & Lorenz-Meyer, D. (1996). Der unentdeckte Wandel. Annaherung an das
Verhaltnis von Struktur und Norm im weiblichen Lebenslauf. Berlin: Sigma.
Bourdieu, P. (1980). Questions de sociologie. Paris: Minuit.
Bourdieu, P. (1984). Distinction, a social critique of the judgement of taste. Cambridge: Harvard
University Press.
Brandtstater, J., Krampen, G., & Heil, F. E. (1993). Adaptative resources of the aging self:
Outlines of an emergent perspective. International Journal of Behavioral Development,
16(2), 323–349.
Campbell, A. A. (1968). The role of family planning in the reduction of poverty. Journal of
Marriage and the Family, 30(2), 236–245.
Chryssochoou, X. (2003). Studying identity in social psychology: Some thoughts on the
definition of identity and its relation to action. Language and Politics, 2(2), 225–242.
Devereux, G. (1967). From anxiety to method in the behavioral sciences. Den Haag-Paris: Mouton.
Diewald, M. (2001). Unitary social science for causal understanding: Experiences and prospects
of life course research. Canadian studies in population. Special Issue on Longitudinal
Methodology, 28(2), 219–248.
Eichler, M. (1988). Nonsexist research methods: A practical guide. London: Allen & Unwin.
Eisenstadt, S. (1968). Social institutions. In: D. L. Sills (Ed.), Encyclopedia of the social sciences
(pp. 409–429). New York: Macmillan and Free Press.
Elder, G. H. (1974, 1999). Children of the great depression. Chicago: University of Chicago
Press.
Elder, G. H. (1992). Life course. In: E. F. Bogatta & M. L. Bogatta (Eds), Encyclopedia of
sociology, (Vol. 3, pp. 1120–1130). New York: Macmillan.
Elder, G. H. (1996). The life course paradigm: Social change and individual development. In:
K. Luscher (Ed.), Examining lives in context: Perspectives on the ecology of human de-
velopment (pp. 101–139). Washington, DC: APA Press.
Elder, G. H. J. (1998). The life course and human development. In: R. M. Lerner (Ed.),
Handbook of child psychology (5th ed., Vol. 1, pp. 939–991). New York, NY: Wiley.
Erikson, E. H. (1963). Childhood and society (2nd ed.). New York: Norton.
Erikson, E. H. (1968). Identity: Youth and crisis. New York: W. W. Norton & Company, Inc.
Freedman, D., Thornton, A., Camburn, D., Alwin, D., & Young-DeMarco, L. (1988). The life
history calendar: A technique for collecting retrospective data. Sociological Methodol-
ogy, 18, 37–68.
Freire, P. (1972). Pedagogy of the oppressed. New York: Herder & Herder.
Ghisletta, P., & Lindenberger, U. (2004). Static and dynamic longitudinal structural analyses of
cognitive changes in old age. Gerontology, 50, 12–16.
Goldstein, H. I. (1987). Multilevel models in educational and social research. London, UK:
Oxford University Press.
Havighurst, R. H. (1972). Developmental tasks and education (3rd ed.). New York: David
McKay.
Heckhausen, J. (1999).Developmental regulation in adulthood: Age-normative and sociostructural
constraints as adaptative challenges. New York: Cambridge University Press.
RENE LEVY ET AL.388
Heinz, W. (Ed.) (1992). Institutions and gatekeeping in the life course. Weinheim: Deutscher
Studienverlag.
Hooker, K. (2002). New directions for research in personality and aging: A comprehensive
model for linking levels, structures, and processes. Journal of Research in Personality, 36,
318–334.
Hooker, K., & McAdams, D. P. (2003). Personality reconsidered: A new agenda for aging
research. Journal of Gerontology: Psychological Sciences, 58B(6), 296–304.
Kalicki, B., Fthenakis, W. E., & Peitz, G. (1999). The emergence of traditional gender-roles at the
transition to parenthood. Poster, SRCD Biennial Meeting, 15–18 April 1999,
Albuquerque, NM.
Kalicki, B., Fthenakis, W. E., Peitz, G., & Engfer, A. (1998). Gender-roles at the transition to
parenthood. Poster, XVth Biennial ISSBD Meeting, July 1st–4th, Berne.
Kaplan, D., & Glass, L. (1995). Understanding nonlinear dynamics. New York, NY: Springer.
Kerckhoff, A. C. (1993). Diverging pathways. Social structure and career deflections. Cambridge:
Cambridge University Press.
Kernis, M. H., & Goldman, B. M. (2003). Stability and variability inself-concept and
self-esteem. In: M. R. Leary & J. P. Tangney (Eds), Handbook of self and identity
(pp. 106–127). New York: The Guildford Press.
Kohli, M. (1985). Die institutionalisierung des lebenslaufs. Kolner Zeitschrift fur Soziologie und
Sozialpsychologie, 37, 1–29.
Kohli, M. (1986). The world we forgot: A historical review of the life course. In: V. W. Marshall
(Ed.), Later life: The social psychology of aging (pp. 271–303). Beverly Hills, CA: Sage.
Kruger, H. (2001). Geschlecht, territorien, institutionen. Beitrag zu einer soziologie der le-
benslauf-relationalitat. In: C. Born & H. Kruger (Eds), Individualisierung und verflech-
tung. Geschlecht und generation im deutschen lebenslaufregime (pp. 257–299). Munchen:
Juventa.
Kruger, H., & Levy, R. (2001). Linking life courses, work, and the family: Theorising a not so
visible ne xus between women and men. Canadian Journal of Sociology, 26(2), 145–166.
Levinson, D. J. (1978). The seasons of man’s life. New York: Ballantine.
Levinson, D. J. (1990). A theory of life structure development in adulthood. In:
C. N. Alexander & E. J. Langer (Eds), Higher stages of human development. Perspec-
tives on adult growth (pp. 35–53). Oxford: Oxford University Press.
Levinson, D. J. (1996). The seasons of a woman’s life. New York: Ballantine.
Levy, R. (1977). Der Lebenslauf als Statusbiographie. Die weibliche Normalbiographie in
makrosoziologischer Perspektive. Stuttgart: Enke.
Levy, R. (1991). Status passages as critical life course transitions. A theoretical sketch. In:
W. R. Heinz (Ed.), Theoretical advances in life course research (pp. 87–114). Weinheim:
Deutscher Studienverlag.
Lewin, K. (1935/1951). Field theory in social science. New York: Harper & Bros.
Li, P. S., & Currie, D. (1992). Gender differences in work interruptions as unequal effects of
marriage and childrearing: Findings from a Canadian national survey. Journal of Com-
parative Family Studies, 23(2), 217–229.
Mayer, K. U., & Muller, W. (1986). The state and the structure of the life course. In:
A. B. Sørensen, et al. (Eds), Human development and the life course (pp. 217–245).
Hillsdale: Lawrence Erlbaum Associates.
Mayer, K. U., & Schoepflin, U. (1989). The state and the life course. Annual Review of
Sociology, 15, 187–209.
Incitations for Interdisciplinarity in Life Course Research 389
McAdams, D. P. (1996). Personality, modernity, and the storied self: A contemporary frame-
work for studying persons. Psychological Inquiry, 7(4), 295–321.
McArdle, J. J. (1986). Latent growth within behavior genetic models. Behavior Genetics, 16,
163–200.
McCrae, R. R., & Costa, P. T. (2003). Personality in adulthood: A five-factor theory perspective
(2nd ed.). New York: The Guildford Press.
Mischel, W. (2004). Toward an integrative science of the person. Annual Review of Psychology,
55, 1–22.
Mischel, W., & Shoda, Y. (1998). Reconciling processing dynamics and personality disposi-
tions. Annual Review of Psychology, 49, 229–258.
Moen, P., Dempster-McClain, D., & Williams, R. M. (1989). Social integration and longevity:
An event history analysis of women’s roles and resilience. American Sociological Review,
54(4), 635–647.
Moen, P., Dempster-McCain, D., & Williams, R. M. (1992). Successful aging: A life-course
perspective on women’s multiples roles and health. American Journal of Sociology, 97(6),
1612–1638.
Moen, P., & Han, S.-K. (2001). Reframing careers: Work, family, and gender. In:
V. W. Marshall, W. R. Heinz, H. Kruger & A. Verma (Eds), Restructuring work and
the life course (pp. 424–445). Toronto: University of Toronto Press.
Nesselroade, J. R. (1991). The warp and woof of the developmental fabric. In: R. M. Downs,
L. S. Liben & D. S. Palermo (Eds), Visions of aesthetics, the environment, & development:
The legacy of Joachim F. Wohlwill (pp. 213–240). Hillsdale, NJ: Lawrence Erlbaum
Associates, Inc.
Neugarten, B. (1977). Personality and aging. In: J. E. Birren & K. W. Schaie (Eds),Handbook of
the psychology of aging (pp. 626–649). New York: Van Nostrand Reinhold.
Neugarten, B., Moore, J. W., & Lowe, J. C. (1965). Age norms, age constraints, and adult
socialization. American Journal of Sociology, 70(6), 710–717.
Nurmi, J.-E., & Salmela-Aro, K. (2000). Goal construction, reconstruction and depressive
symptomatology in a life span context: The transition from school to work. Journal of
Personality, 70(3), 385–420.
Rosow, I. (1976). Status and role change through the life span. In: R. H. Binstock & E. Shanas
(Eds), Handbook of aging and the social sciences. New York: Van Nostrand Reinhold.
Rovine, M. J., & Molenaar, P. C. M. (2000). A structural modeling approach to a multilevel
random coefficients model. Multivariate Behavioral Research, 35, 51–88.
Ryder, N. B. (1965). The cohort as a concept in the study of social change. American Soci-
ological Review, 30, 29–46.
Schulz, R., & Heckhausen, J. (1996). A life span model of successful aging. American Psy-
chologist, 51, 702–714.
Scott, J., & Alwin, D. F. (1998). Retrospective versus prospective measurement of life histories
in longitudinal research. In: J. Z. Giele & G. H. Elder, Jr. (Eds), Methods of life course
research: Qualitative and quantitative approaches. Thousand Oaks, CA: Sage.
Settersten, R. A. (1997). The salience of age in the life course.Human Development, 40, 257–281.
Settersten, R. A. (2005). Toward a stronger partnership between life-course sociology and life-
span psychology. Research in Human Development, 2(1+2), 25–41.
Settersten, R. A., Jr. (1999). Lives in time and place. Amityville, NY: Baywood Publishing
Company, Inc.
RENE LEVY ET AL.390
Settersten, R. A., & Mayer, K. U. (1997). The measurement of age, age structuring, and the life
course. Annual Review of Sociology, 23, 233–261.
Shanahan, M. J., Hofer, S. M., & Shanahan, L. (2003). Biological models of behavior and the
life course. In: J. T. Mortimer & M. J. Shanahan (Eds), Handbook of the life course.
New York: Kluwer Academic/Plenum Publishers.
Sliwinski, M. J., Hofer, S. M., Hall, C., & Buschke, H. (2003). Modeling memory decline in
older adults: The importance of preclinical dementia, attrition, and chronological age.
Psychology and Aging, 18, 658–671.
Srivatava, S., John, O. P., Gosling, S. D., & Potter, J. (2003). Development of personality in
early and middle adulthood: St like plaster or persistent change? Journal of Personality
and Social Psychology, 84(5), 1041–1053.
Taylor, S. E., & Brown, J. D. (1988). Illusion and well-being: A social psychological perspective
on mental health. Psychological Bulletin, 103(2), 193–210.
Taylor, S. E., Kemeny, M. E., Reed, G. M., Bower, J. E., & Gruenewald, T. L. (2000).
Psychological resources, positive illusions, and health. American Psychologist, 55
(99–109).
Thoits, P. A. (1986). Multiple identities: Examining gender and marital status differences in
distress. American Sociological Review, 1(2), 259–272.
Widmer, E., Kellerhals, J., & Levy, R. (2005). What pluralization of the life course? An analysis
of personal trajectories and conjugal interactions in contemporary Switzerland. In:
H. Kriesi, P. Farago, M. Kohli & M. Zarin-Nejadan (Eds), Contemporary Switzerland:
revisiting the special case. Houndmills: Palgrave, Macmillan.
Widmer, E., Kellerhals, J., & Levy, R., with the collaboration of M. Ernst Stahli & R. Hammer
(2003a). Couples contemporains – cohesion, regulation et conflits. Une enquete soc-
iologique. Zurich: Seismo.
Widmer, E., Levy, R., Pollien, A., Hammer, R., & Gauthier, J.-A. (2003b). Entre standard-
isation, individualisation et sexuation: Une analyse des trajectoires personnelles en Suisse
[Between standardization, individualization, and sexuation: An analysis of personal
trajectories in Switzerland]. Swiss Journal of Sociology, 29, 35–67.
Wohlwill, J. F. (1970). The age variable in psychological research. Psychological Review, 77,
49–64.
Incitations for Interdisciplinarity in Life Course Research 391
AUTHOR INDEX
Abbott, A. 13, 188, 274, 309, 364, 379
Abelson, R. 238
Abrams, D. 202
Ackerman, S. 62
Adler, J. 247
Agrawal, R. 301
Ahrons, C. 160
Aisenbery, S. 188
Alea, N. 219–220
Alexander, K.L. 132
Allison, P.D. 266, 289
Allport, G.W. 239, 371
Alpert, A. 333
Alter, G. 287, 291–293
Alwin, D.F. 37, 69, 85, 373, 375–376
Ambady, G. 241
Anderson, E. 317, 326, 351
Andres, D. 226
Anstey, K.J. 382
Anyadike-Danes, M. 277
Anyidoho, N.A. 245
Appice, A. 303
Aranza-Ordaz, F.J. 289
Arbuckle, J.L. 351
Arbuckle, T.Y. 226
Archer, M.S. 70, 85–86
Arnett, J.J. 243–244
Attanucci, J. 14
Attias-Donfut, C. 95
Axinn, W.G. 266, 293
Ayers, M.S. 220
Bakan, D. 62, 246
Baltes, P.B. 11–12, 14, 28, 43, 48, 132,
221, 231, 241, 315–317, 361, 378
Bandura, A. 132
Barber, B.L. 168
Barber, J.S. 266, 293
Bartlett, F.C. 219–220, 229
Barutchu, A. 202
Bates, D.M. 351
Bauer, J.J. 245
Bauman, Z. 99
Baumeister, R.F. 239
Bawin-Legros, B. 95
Buchel, F. 190
Beck, U. 42, 174, 179
Becker, H.S. 156
Becker, S.O. 271–272
Becker-Schmidt, R. 176–177
Beck-Gernsheim, E. 179
Beckmann, P. 180, 190
Beekink, E. 291–292
Begin, J. 107
Bell, R.Q. 317, 333, 351
Belli, R. 376
Belsky, J. 123
Bengtson, V.L. 95
Berchtold, A. 295–296, 299
Berger, P.L. 69–70, 73, 363
Bernard, H.R. 206
Berntsen, D. 219
Berquo, E. 286
Bickel, J.-F. 96
Billari, F.C. 261–262, 273–274, 287,
299, 301–303
Billig, M.G. 202
Bird, K. 180–182
Birkett, H. 342
Blakeslee, S. 159
393
Blechler, M. 46
Bleuler, E. 97
Block, J. 342
Blockeel, H. 299, 301–302
Blossfeld, H.-P. 174, 177, 187,
189, 265–268, 278, 287–288,
290, 296
Bluck, S. 219–221, 243, 250
Bock, R.D. 350
Bocquier, P. 266, 287
Bogen, S. 122
Boker, S.M. 348, 351
Born, C. 179, 182, 189, 368
Bouchard, T.J. 240
Bourdieu, P. 300, 384, 386
Bower, J.E. 374
Bowman, P.J. 246–247
Bradway, K.P. 318, 320, 348
Brainerd, C.J. 220
Brand, C. 109
Brandtstadter, J. 40, 43, 132, 372
Brannen, J. 102, 108
Breiman, L. 26, 263, 300,
303, 306
Brewer, W.F. 220
Bronfenbrenner, U. 28, 38
Brooks-Gunn, J. 164
Brostrom, G. 291–293
Brown, C. 245
Brown, J.D. 374
Brown, R. 200
Browne, M. 351
Browne, W. 293
Bruner, J.S. 238, 242, 252
Bryk, A.S. 293, 328, 335, 351
Bryson, A. 270
Buckholdt, D.R. 85
Bundy, R.F. 202
Burkhardt, A. 109
Buschke, H. 382
Busch-Rossnagel, N.A. 60, 84
Butler, R.N. 82, 230
Bynner, J. 210
Cain, L.D. 72
Call, K.T. 135, 144
Camburn, D. 376
Campbell, A.A. 157, 364
Campbell, C. 68–69
Campbell, D.T. 345
Campbell, R.T. 85
Cantor, N. 241
Carstensen, L.L. 47
Caselli, G. 285
Casey, K. 238
Cattell, R.B. 316, 338, 342
Cavanaugh, J.C. 226
Ceci, M. 303
Chaikelson, J. 226
Charme, S. 239, 243
Chartrand, E. 107
Cherlin, A.J. 110, 159
Christensen, K. 271
Chryssochoou, X. 374
Ciampi, A. 303, 309
Cinnirella, M. 17
Cinotta, S. 202
Clarke, P. 69
Clausen, J.A. 40, 132, 135, 156, 168
Clausen, J.S. 68, 80, 218
Clogg, C. 144
Cnaan, A. 324
Coenen-Huther, J. 106, 111
Cohler, B.J. 99, 109, 114, 238, 244
Colby, A. 199, 248
Coleman, J.S. 138
Collins, R. 85
Colvin, C.R. 240
Congdon, R. 293
Conger, R.D. 123
Connidis, I.A. 106, 121
Converse, P. 209
Conway, M.A. 219–220, 238
Cook, T.D. 44, 345
Costa, P.T. 224, 237, 239–240, 252, 371
Counts, D. 73
Courage, M.L. 242
AUTHOR INDEX394
Courgeau, D. 12, 19, 266, 287, 289
Cox, D.R. 264, 289
Crawford, J.R. 324
Cremer, C. 190
Criqui, M.H. 240
Crompton, R. 189
Crosnoe, R. 173, 187
Crowley, J. 309
Cudeck, R. 329, 333, 351
Cumming, E. 73
Curran, P. 351
Currie, D. 368
Czikszentmihalyi, M. 204
Dahrendorf, R. 69
Damon, W. 248
Dannefer, D. 41, 47, 59, 64, 85, 237
Day, R. 246
De Bruijn, B.J. 14
de Coninck, F. 19
De Rose, A. 303
de St. Aubin, E. 241, 245, 247–249
Deary, I.J. 239–240, 324
Dehaspe, L. 301
Dehejia, R.H. 269
Del Vecchio, W. 240
Demeny, P. 285
Dempster-McClain, D. 387
Denzin, N. 250
Devereux, G. 385
Diamond, A. 247, 249
Diehl, M. 60
Diewald, M. 40, 42, 385
Diggle, P.J. 341
Dilthey, W. 81
Dixon, R.A. 227
Doise, W. 17, 201
Donaldson, G. 316–317
Donati, P. 95
Dorsett, R. 270
Doyle, G.C. 218
Draper, D. 293
Drobnic, S. 177, 189
du Toit, S.H.C. 329, 333, 351
Duncan, L.E. 241
Duncan, S.C. 333
Duncan, T.E. 333
Dupaquier, J. 284
Dupaquier, M. 284
Durante, M. 202
Durkheim, E. 42
Durkin, K. 16
Dykstra, P.A. 262, 287
Eccles, J.S. 44, 168
Edel, L. 239
Edelstein, W. 14
Edmunds, J. 100
Eerola, M. 266
Ehmer, J. 118
Ehrsam, R. 222
Eich, E. 227
Eichler, M. 383
Eisenstadt, S. 362
Elder, G.H.Jr. 6, 12, 17, 44, 60, 63,
65–66, 68, 72, 106, 123, 127–128,
134, 155–156, 173–174, 187, 189, 218,
232, 237, 244, 252, 273, 361, 363–364,
368, 375, 386
Eliason, S.R. 144, 151
Ely, M. 238
Emler, N. 14, 197, 203, 205–206,
209–210, 212
Emmons, R.A. 241
Engelbrech, G. 190
Engfer, A. 369
Engstler, H. 111, 124
Entwisle, D.R. 132
Epstein, D.B. 317, 326, 329, 335,
338, 351
Epston, M. 251
Erikson, E.H. 14, 16, 82, 198,
217, 230, 237, 239, 241,
243, 371
Erzberger, C. 23, 183, 188
Esping-Andersen, G. 49
Author Index 395
Esser, H. 190
Essex, M.J. 16
Fahrenberg, J. 223, 226
Farkas, J.I. 174
Ferrer, E. 351
Ferrer-Caja, E. 331
Filipp, S.-H. 106–107
Finch, J. 95
Finch, M.D. 151
Fingerman, K.L. 108
Fitzgerald, J.M. 225, 229
Flaherty, B.P. 139
Flament, C. 202
Flanagan, C.A. 46
Flavell, J.H. 198
Foley, J.M. 247
Foner, A. 155
Forrest, J.D. 29, 309
Foster, E.M. 46
Frank, E. 303
Frankel, M.R. 293
Freedman, D. 376
Freire, P. 386
Freud, S. 98, 217
Freund, A.M. 40, 132, 241
Friedman, H.S. 240
Friedman, J.H. 303, 306
Friis, H. 95
Furnkranz, J. 274, 299, 301–303
Fthenakis, W.E. 369
Funder, D.C. 240
Furstenberg, F.F. 44, 46, 155, 157–159,
164
Gabriel, Y. 238
Gallagher, M. 159
Gannon, L. 35
Gardiner, J.M. 219
Garrett, M. 202
Gauthier, A. 95
Gauthier, J.-A. 366, 369, 380
Gecas, V. 47, 132
George, L.K. 11, 62, 69, 86
Gergen, K.J. 238, 244
Ghisletta, P. 361, 380
Giarusso, R. 95
Gibbons, R. 351
Giddens, A. 61, 65, 70, 85,
174, 241, 244
Giele, J. 40
Gifford, E.J. 46
Gigerenzer, G. 41
Giles, H. 201
Gilligan, C. 14
Gjerde, P.F. 251
Glass, L. 387
Godard, F. 19
Goffman, I. 75, 81, 112
Gold, D.P. 226
Goldberg, L.R. 240
Goldman, B.M. 366
Goldstein, H. 293, 351, 379
Goode, W. 213
Goody, J. 86
Gosling, S.D. 371
Gottschall, K. 182
Grabowski, L.J. 135, 144
Grambsch, P.M. 290, 293
Green, E.E. 226
Greenhalgh, S. 285
Gregg, G. 239, 250
Griffin, D.W. 105
Gruenewald, T.L. 374
Grunebaum, H.U. 99, 114
Gubrium, J.F. 85, 238
Gutmann, D. 61, 244
Gutschner, P. 118
Habermas, T. 243, 250
Haely, M. 293
Hagestad, G.O. 16, 47, 156, 174, 287
Hakim, C. 177
Hall, C. 382
Hall, M. 303
Hallahan, M. 241
AUTHOR INDEX396
Hamagami, F. 318, 320–321, 329,
331, 333–335, 337–338, 345,
347–349, 351
Hamerle, A. 266, 288
Hammer, R. 366, 369, 380
Hampel, R. 223, 226
Han, J. 300
Han, S.-K. 186, 189, 366
Hand, D.J. 300
Hankiss, A. 243
Harlow, S.D. 315, 352
Harootyan, R.A. 95
Harris, F. 189
Hashtroudi, S. 219
Havighurst, R.H. 16, 371
Hay, E. 108
Hayes, C. D. 158
Heckhausen, J. 43, 132, 372
Heckman, J. 270
Hedecker, D. 351
Heidrich, S.M. 218
Heil, F.E. 372
Heilbrun, C. 250
Heinz, W.R. 64, 85, 173–174, 185, 187
Hennessy, J. 202
Henry, W.E. 73
Herman, M.R. 44
Hermans, H.J.M. 239, 246
Hershberg, T. 157
Hertzog, C. 227
Hetherington, E.M. 168
Hewstone, M. 201
Hohn, C. 286
Hobcraft, J. 262, 266
Hoch, H. 110, 125
Hofer, D. 226
Hofer, S.M. 382, 385
Hoffman, B.J. 246
Hogan, D.P. 16, 174
Hogdson, D. 285
Hogg, S.A. 303, 309
Holford, T.R. 289
Holland, P.W. 270
Holman, T.B. 107
Holmes, J. 238
Holstein, J.A. 85, 238
Hooimeijer, P. 286
Hooker, K. 239, 241, 244,
371–372
Horn, J.L. 316–317
Hotz, V.J. 269–271
House, J.S. 133
Howe, M.L. 242
Hopflinger, F. 222, 224
Hradil, S. 179
Huang, Y.T. 245
Hughes, E.C. 72
Huinink, J. 183
Hultsch, D.F. 227
Ichino, A. 271–272
James, A. 40, 45
Java, R.I. 219
Jekeli, I. 106
Jenks, C. 40, 45
Jepperson, R.L. 59, 65–66
John, O.P. 240–241, 371
Johnson, K.M. 60, 63, 173, 187
Johnson, M.J. 107, 155
Johnson, M.K. 219
Jolesz, F. 321
Jones, G. 211
Jones, K. 321
Jones, R.L. 86
Josselson, R. 225
Joreskog, K.G. 338
Julien, D. 107
Junn, J. 209
Kahana, B. 61
Kahana, E. 61
Kahn, R. 48, 202
Kaiser, A. 222, 224
Kalicki, B. 369
Kalish, R.A. 75–76
Author Index 397
Kamber, M. 300
Kangas, J. 320
Kaplan, B. 245
Kaplan, D. 387
Kass, G.V. 304
Kastenbaum, R. 85
Katz, D. 202
Keane, M.R. 273
Kellerhals, J. 106, 366, 369
Kelly, J. 168
Kemeny, M.E. 374
Kennedy, S. 46
Kenny, D.A. 205
Kenrick, D.T. 205, 240
Kenward, M.G. 341
Kerckhoff, A.C. 384
Kernis, M.H. 366
Keyes, C.L.M. 219, 230
Kikinis, R. 321
Killworth, P.D. 206
King, L. 245
Kiser, E. 65
Kish, L. 293
Kissel, E. 247
Klein, D. 104
Klein, V. 180
Klineberg, J. 204
Kling, K.C. 16
Kling, V. 226
Klohnen, E.C. 241
Kluckhohn, C. 239
Kluge, S. 188
Knapp, G.-A. 177
Knellessen, O. 122
Knox, W.E. 135
Kohlberg, L. 198–199
Kohler, H.-P. 271
Kohli, M. 12, 187, 243, 363, 367, 382
Krampen, G. 372
Krebs, E. 222
Kruger, H. 50, 134, 150, 173–174,
176, 179, 182, 189, 363, 368,
384, 386
Kruglanski, A.W. 14
Kottig, M. 29
Kuh, D. 218
Kuijsten, A. 286
Kurtz, B. 180
Labouvie-Vief, G. 14
Laird, N.M. 324, 334, 351
Lalive d’Epinay, C. 96, 125
Landwehr, N. 303
Lang, F.R. 47, 102, 115
Langbaum, J. 239
Larson, R. 204
Lash, S. 174
Lautmann, R. 110
Lawton, M.P. 61
Layder, D. 69
Le Goff, J.-M. 111, 268, 361
Leary, T. 101
Leblanc, M. 309
Lee, J.C. 131
Lee, S. 97, 125
Lelievre, E. 12, 19, 266, 287, 289
Lemmon, H. 324
Lerner, R.M. 60, 84
Lettke, F. 102, 104, 106–107, 122
Leveille, S. 107
Levine, D.I. 270–271
Levinson, D.J. 237, 244, 252, 368, 371
Levy, R. 3, 11, 50, 175, 361, 363, 366,
369, 380, 384, 386
Lewin, K. 384, 386
Lewis, M. 246–247
Li, A. 50
Li, F. 333
Li, P.S. 368
Li, S.-C. 48
Lianos, G. 202
Lieberson, S. 160
Liefbroer, A.C. 291–292
Lillard, L.A. 267, 278, 290
Lim, K.O. 345
Lin, D.Y. 293
AUTHOR INDEX398
Lindenberger, U. 14, 28, 43, 221, 315,
361, 380
Lindsay, D.S. 219
Lindsay, P. 135
Littell, R.C. 351
Little, B.R. 241
Little, R.J.A. 324, 341
Little, R.T.A. 351
Littwak, E. 95, 125
Loevinger, J. 198
Loftus, E.F. 220
Logan, R.L. 121, 241
Lord, S.R. 382
Lorence, J. 135
Lorenz-Mayer, D. 179, 182, 189
Lorenz-Meyer, D. 107, 116, 368
Lowe, J.C. 17, 155–156, 384
Lowenstein, A. 117
Luscher, K. 93, 95–96, 102, 106–107,
110–111, 115, 118, 122
Luckmann, T. 69–70, 73, 363
Ludewig-Kedmi, R. 113
Lykken, D.T. 240
Lynch, K.A. 295
Mabry, B. 95, 123
Macaulay, D. 227
MacDonald, I.L. 295
Machado, M.A. 245
MacIntyre, A. 238
Maclean, M. 218
Macmillan, R. 144, 151
Maes, H.H. 351
Magnusson, D. 342
Malerba, D. 303
Malo, M.A. 309
Mandler, J.M. 242
Mannila, H. 300–301
Mansfield, E.D. 246–247, 249
Manting, D. 18
Manton, K.G. 291
Maples, J. 266, 293
Marcia, J. 198
Marini, M. 156
Markman, H.J. 107
Markus, H.R. 80, 132
Marshall, V.W. 57, 66, 69–70,
72–74, 78, 81–83, 85–86, 95,
173, 185
Maruna, S. 238
Marx-Ferree, M. 189
Mason, J. 95
Matthews, G. 239–240
Matthews, S.H. 95
Mayer, A.-K. 106–107
Mayer, K.U. 7, 12, 16, 42, 49, 174, 177,
181, 183, 187, 189, 266–267, 273, 288,
363, 370, 382
Mazzuco, S. 273
McAdams, D.P. 14, 17, 121, 126, 217,
230, 233, 237–239, 241, 243–250, 252,
371–372
McArdle, J.J. 315, 317–318, 320–321,
324, 326, 329, 331, 333–335, 337–338,
341, 344–345, 347–349, 351–352,
382
McCrae, R.R. 224, 237, 239–240,
252, 371
McCullagh, P. 310
McGue, M. 240
McKinney, S. 303, 309
McLanahan, S. 159
McLoyd, V. 46
McMullin, J.A. 69–70, 121, 124
McNamara, S. 206
McNicoll, G. 285
McVicar, D. 277
Mead, G.H. 100, 119
Meier, B. 226
Mercer, R.T. 218
Meredith, W. 318, 320, 328, 348,
350–351
Merton, R.K. 156
Metha, P.D. 351
Meyer, J.W. 59, 65–66
Meyerhoff, B.G. 86
Author Index 399
Miech, R.A. 68, 134
Milardo, R. 107
Milgram, S. 204
Milhoj, P. 95
Miliken, G.A. 334, 351
Mills, C.W. 49
Mills, M. 268
Milofsky, E. 241
Mischel, W. 240–241, 371, 373
Mitchell, J.C. 202
Mitteness, L.S. 114
Muller, W. 187, 363
Modell, J. 157
Moen, P. 174, 176, 186, 189,
366, 387
Mojardin, A.H. 220
Molenaar, P.C.M. 16, 380
Monopoli, M. 303
Moore, J.W. 17, 155–156, 384
Morgan, S.P. 164
Mortimer, J.T. 131, 133–136, 138–140,
144, 150–151, 174–175
Moskowitz, D.S. 62
Mounoud, P. 16
Mouw, T. 274, 276–277
Mueller, M.M. 66, 83, 173
Mugny, G. 17
Mullen, B. 230
Mullin, C.H. 269–271
Munoz, F. 309
Murphy, M. 262, 266
Murphy, S.A. 266, 293
Murray, H.A. 239
Murray, S.L. 238
Muthen, B.O. 342, 351
Muthen, L.K. 342, 351
Myrdal, A. 180
Nagin, D. 342
Neale, M.C. 351
Nelder, J.A. 310
Nesselroade, J.R. 316–317, 321, 333,
344–345, 352, 366, 378
Neugarten, B.L. 16–17, 155–156, 174,
371, 384
Nichols, E.G. 218, 233
Nie, N.H. 209
Noam, G. 14
Nurius, P. 80, 132
Nurmi, J.-E. 376
Nydegger, C.N. 114
Oakes, J.M. 39
Oesterle, S. 138, 140, 174
Ogg, J. 117
Olshen, R.A. 303, 306
Olson, L.S. 132
Ondrich, J. 190
O’Rand, A. 63, 65–66, 85
O’Rand, A.M. 174
Oris, M. 283, 287, 291–293, 299
Otscheret, E. 122
Painter, G. 270–271
Pajung-Bilger, B. 96, 102
Paliwal, K.K. 295
Pallara, A. 303
Panis, C.W.A. 267, 278
Parker, R. 111
Parsons, T. 95, 156
Pascual-Leone, J. 14
Passel, J.S. 39
Patten, A. 246–247
Peitz, G. 369
Perren, S. 225
Perrig, W.J. 217, 222, 226–227
Perrig-Chiello, P. 217, 222–227,
233
Petersen, T. 266
Peterson, C. 247
Pfefferman, A. 345
Phillips, M. 44
Philo, C. 45
Piaget, J. 198
Piccarreta, R. 274
Pillemer, K. 95–96, 106, 112, 118
AUTHOR INDEX400
Pinheiro, J.C. 351
Plakans, A. 118
Platow, M. 202
Plewis, I. 293
Pleydell-Pearce, C.W. 238
Polkinghorne, D. 238
Pollien, A. 366, 369, 380
Pomerantz, E. 237, 240
Pothoff, R.F. 350
Potter, J. 371
Poulain, M. 287
Powesland, P.F. 201
Prein, G. 23
Prescott, C.A. 337, 351
Pressat, R. 17
Pross, H. 180
Prout, A. 40, 45, 49–50
Prskawetz, A. 274, 299, 301–303
Potter, U. 278, 290
Purdon, S. 270
Pyke, K.D. 123
Quinlan, J.R. 303, 306
Rabiner, L.R. 295
Raftery, A.E. 295
Rakotomalala, R. 310
Ramon, J. 301
Ramsey, C.M. 245
Rao, C.R. 328
Rasbash, J. 293
Raudenbush, S.W. 293, 328,
335, 351
Reder, L.M. 220, 231
Reed, G.M. 374
Reicher, S.D. 202, 205
Rein, M. 95
Reinharz, S. 96
Reyna, V.F. 220
Reynolds, J. 246–247
Richards, L.N. 95
Ricoeur, P. 239, 242
Riegel, K.R. 14, 84
Riley, J.W. 37
Riley, M.W. 37, 72, 155
Rindfuss, R.R. 16, 87, 155, 174, 262
Ritschard, G. 283, 299, 303, 307
Robb, R. 270
Roberts, B.W. 237, 240
Roberts, K. 134
Roberts, R.E.L. 95, 127
Robertson, E.B. 123, 127
Rodgers, B. 218
Rogosa, D. 317, 350
Rohwer, G. 265–267, 278, 287,
290, 296
Romney, D. 210
Rosenbaum, P.R. 271–272
Rosenbloom, M.J. 345
Rosenfeld, R.A. 16, 87, 155, 174
Rosenthal, C.J. 95
Rosenthal, G. 29
Rosenthal, R. 241
Rosenwald, G. 250
Rosow, I. 387
Ross, K. 46
Rossi, A.S. 62, 95
Rossi, P.H. 39, 95
Rossier, C. 111
Rost, H. 190
Rothermund, K. 43
Rotter, J.B. 132
Rovine, M.J. 380
Rowe, J. 48
Roy, S.N. 350
Rubin, D.B. 271–272, 351
Rubin, D.C. 219–220
Rubin, M. 201
Ruch, M. 226
Rudorf, S. 109
Ruetzel, K. 247
Rumbaut, R. 46
Rusca, E. 231
Rutter, M. 217–218
Rybash, J.M. 225
Ryczkowska, G. 303
Author Index 401
Ryder, N.B. 156, 376
Ryff, C.D. 16, 69, 80, 218–219, 230
Ryu, S. 151
Sackmann, R. 175
Salmela-Aro, K. 376
Sameroff, A. 44
Sandefur, G. 159
Sanders, S.G. 269–271
Sarbin, T. 238
Sauvain-Dugerdil, C. 111
Sayer, A.G. 345, 351
Schacter, D.L. 219, 230
Schaie, K.W. 12
Schank, R. 238
Schmidt, B. 177
Schneider, N.F. 190
Schoeni, R. 46
Schoepflin, U. 12, 363
Schutze, Y. 181
Schulz, R. 372
Schutz, A. 70
Schwartz, J.E. 240
Schwartz, S.H. 241
Schwartzman, A. 226
Schwarz, N. 219
Scollon, C.K. 245
Scott, J. 373, 375
Seber, G.A.F. 329
Segal, M.R. 309
Segal, N.L. 240
Selg, H. 223, 226
Seligman, M.E.P. 247
Serafin, D. 226
Settersten, R.A. 4–5, 16, 28, 35, 39–40,
42–44, 46, 48, 52, 60, 69, 156, 174, 267,
273, 370, 373, 382, 384–385
Sewell, W.H. 37, 70, 85
Shanahan, L. 385
Shanahan, M.J. 50, 68, 133–134, 139,
151, 385
Shanas, E. 95
Sheldon, K.M. 239
Shifley-Grove, S.S. 225, 229
Shoda, Y. 241, 371
Shuey, K.M. 106
Sibeon, R. 43
Silverstein, M. 95
Singer, J.A. 241, 245–246
Singer, J.D. 351
Skevington, S. 202
Skinner, M.L. 123
Skytthe, A. 271
Slasor, P. 324
Slater, P.E. 208
Sliwinski, M.J. 382
Small, B.J. 227
Smelser, N.J. 70, 99, 113, 122
Smith, G.A. 382
Smith, J. 40
Smyth, P. 300
Spangler, D. 106
SpieX, K. 190
Spini, D. 361
Spiro, A. 321
Sorbom, D. 338
Sørensen, A. 174
Srikant, R. 301
Srivastava, S. 240, 371
Staff, J. 131, 138, 140
Stallard, E. 291
Starr, J.M. 324
Staudinger, U. M. 14, 28,
43, 100, 120,
221, 315, 361
Stehlik-Barry, K. 209
Stehouwer, J. 95
Stewart, A.J. 241
Stahelin, H.B. 222, 226
Stone, C.J. 303, 306
Stoup, W.W. 334, 351
Strack, F. 219, 234
Strassen, J.-F. 95
Strauss, A.L. 156
Stringfield, D.O. 205
Struyf, J. 301
AUTHOR INDEX402
Strycker, L.A. 333
Sturzenegger, M. 222, 224
Stuss, D.T. 219
Sugarman, L. 218
Suitor, J.J. 106, 112
Sullivan, E.V. 345
Suls, J. 230, 234
Sussman, M.B. 95
Swicegood, C.G. 16, 87,
155, 174
Szreter, S. 285
Szydlik, M. 95
Tajfel, H. 200, 202–203
Tashakkori, A. 23
Tavare, S. 295
Taylor, S.E. 374
Teddlie, C. 23
Tellegen, A. 240
Therneau, T.M. 290, 293
Thiffault, J. 303, 309
Thoits, P.A. 387
Thom, R. 16
Thomas, W.I. 63
Thompson, C.W. 320
Thompson, M.M. 105
Thorne, A. 244–245
Thornton, A. 376
Thurlemann, F. 122
Tisak, J. 328, 350–351
Tolke, A. 183
Toivonen, H. 301
Tomasello, M. 242
Tomlinson-Keasy, C. 240
Tostain, M. 14
Townsend, P. 95
Tsay, A. 274
Tucker, J.S. 240
Tucker, L.R. 328
Tulving, E. 219
Turiel, E. 14
Turner, B.S. 100
Turner, J.C. 201, 203
Uhlenberg, P. 59
Vaillant, G.E. 218, 241
Vallin, J. 284–285
Van de Kaa, D.J. 285
Van de Water, D. 247
van der Heijden, P.G.M. 274
van der Maas, H.L. 16
Van Imhoff, E. 286
van Poppel, F. 291–292
van Wissen, L.J.G. 262, 286–287
Vaupel, J.W. 291
Verkamo, A.I. 301
Viaud, J. 11, 16
von Allmen, M. 106
von Matt, P. 97
Veron, J. 285
Wadsworth, M.E.J. 218
Wahba, S. 269
Waite, L. 159
Wallace, C. 211
Wallerstein, J.S. 159
Walshe, J. 202
Ware, J.H. 334, 351
Weber, M. 64, 69
Weddeburn, D. 95
Wei, L.J. 293
Wellman, H.M. 242
Wertlieb, D. 218
West, M.A. 202
West, S.G. 351
Whalley, L.J. 324
Wheeler, M.A. 219
White, M. 251
Whiteman, M.C. 239–240
Whittlesea, B.W.A. 220
Widmer, E. 361, 366, 369, 380
Wiese, B.S. 132
Wild, C.J. 329
Wilensky, H. 157
Willekens, F.J. 284
Willett, J.B. 317, 333, 345, 350–351
Author Index 403
Williams, N. 202
Williams, R.M. 387
Williams, T. 245
Willms-Herget, A. 179
Willson, A.E. 106
Wingard, D.L. 240
Wingens, M. 175
Winsborough, H. 156
Winter, D.G. 241
Wishart, J. 328, 331
Wohlwill, J.F. 5, 381
Woike, B. 239, 245–246
Wolfinger, R.D. 334, 351
Wolpin, K.I. 273
Woodcock, J.R. 326, 331, 351
Woodcock, R.W. 331
Woodhouse, G. 293
Wothke, W. 351
Wright, R. 220
Wu, L. 50
Wu, L.L. 188, 266–267, 274
Wunsch, G. 285
Wurzbacher, G. 180
Xenos, P. 286
Xie, G. 351
Yamaguchi, K. 266, 289
Yang, M. 293
Yang, Q. 190
Young-DeMarco, L. 376
Zanna, M.P. 105
Zeger, S.L. 352
Zighed, D.A. 303, 307, 310
Zima, P.V. 105
Znaniecki, F. 63
Zucchini, W. 295
Zuroff, D.C. 62
AUTHOR INDEX404
SUBJECT INDEX
accelerated failure time model 289
ACT effects 316–317
action-structure problem 69–70
adaptation 61
adjacent institutionalization 51
adolescence 45, 144, 146, 248
affection 107
agency 39–41, 44, 49, 58–69, 83,
144–149, 246, 362–365
agency and structure 21–22, 35, 71–83,
131, 364–365, 370, 379
agency within structure 41–43, 51–52,
385
agentic decision-making and behavior
133
agentic strategies 133
aging and dying 58, 71–72
AIC criteria 298, 307, 308
algorithmic culture 274–278
ambivalence 93, 95, 97–105,
110–119
ambivalence diversification 108–110
ambivalence understanding elements 98
analysis of covariance 334–335
assessment and differentiation of
ambivalence 105–108
assessment of relationships 105–108, 112
attitude 220, 230
autobiographical coherence 243
autobiographical memory 24, 219–220
autobiographical reflection selectivity
81
autobiographical self 242
average effect of treatment 270–271
awareness and timeframes 186
baby break 178–182
Bartlett’s schema 220
basic tendencies 371–372
BIC criteria 298, 307–308
Big Five traits 240, 371
biographical experiences 217–218
biographical memories 225–227
biographical recollection 227–230
biographical reconstruction 23–25, 195,
370–374
biographical transition 217–218, 222,
224
Bradway–McArdle collection 335
Bradway–McArdle Longitudinal Study
318, 320–321
Captivation 104
Caring 113–117
causal approach 265–269
causal coherence 243
causal impacts of events 269–273
causal inference 270
‘causality’ culture 265–273
challenges and contradictions of life
course 35
changes in identity 81–83
changing social identity 208–209
characteristic adaptations 241, 371–372
child-rearing leave 178–180
childhood and adolescence 44–45
childhood friends and playmates 208
class-theoretical models 38
classification tree 303, 309
cognitive functioning 198–199, 226
405
cognitive growth for males and females
340–342
cognitive skills 243
communion 62, 246
conceptualizing ambivalence 97–104
constructivist views 40
contamination sequences 246–247
contemporary discourse reframing
186–187
continuous-time model 288–289
conventional approach of structure 37
coping 368
core social identities 203–208
covariates 265–267, 271–272,
288–290
Cox’s model 289–290, 293–295
cultural variations 76, 212–213
culture 76, 250–253
data-analytical approach 375–377
data-analytical tool 377, 380
data collection 376
data-mining approaches 283, 309
data-mining method 300
data modeling culture 26, 264, 278, 300
decisional ambivalence 116–117
demographic analysis 283–288
demographic density 262
demographic events significance 155
demographic evolution 285
demography 261–263, 283
developmental data analysis steps 317
developmental psychology 315, 382
developmental regulation model 43
developmental scores 333–337
developmental shapes 324–333, 338–344
dialectical processes 69–70
discrete-time model 289
dispositional traits 239–242
diversification context of research status
108–110
divorce 110, 158–161, 168–169
double chain Markov model (DCMM)
295–296
dynamic determinants 344–348
dynamic structure 37–38
early adulthood through midlife 45–47
early childbearing 157–159
early marriage 166
early work investment 136–140
educational attainment 131, 134, 142,
144
educational influence 335–337
educational orientations 136–140
ego strength 61–62
emancipation 102
emerging adults 244
emotional valence 222, 225, 228–229
end of the life course 57, 60, 83–84
Note: See also last chapters of life
environmental proactivity 61
episodic memory recollection 227
episodic recollection 225–227
Erikson’s theory 121
event-based approach 263–264
event-based culture 265–273
event history analysis (EHA) 263,
265–269, 378
event-history model 300
event history regression models 288–295
event-sequential association rules mining
300–302
events to echo effects 181–182
evidence in life transitions 64–65
expectations 331–333
extended transition 211
external causality 18
extravert person 240
factor-analytic study 240
family relationship 45, 364
fertility project types 111
first-order homogenous transition
matrices 296–298
SUBJECT INDEX406
fitting latent difference score models
347–352
Five-Factor Theory (FFT) 371–373
formal aspects of life course 11–20
friends and acquaintances 210
gender 176
gender difference 5–6, 138
general developmental model 316
generalized linear model 293
generativity 61–62, 107, 121, 241, 248,
374
gerontology 20, 47–48
Gini heterogeneity index 275
good life course 72–77
group differences 160, 318, 338–344
growth curve 317, 320, 340
growth-curve analyses 321, 324–325
growth mixture model 342–343
hazard model 267–268, 309
hidden Markov model (HMM) 295,
297–298
hierarchical linear model 335, 378
highly generative American adults
247–250
historical force types 178
holistic approaches 26, 263
holistic culture 273–275
homogenous-Markov model 295
human agency 35, 131, 251
human individuality 238–242, 247
human infants 242
identity 24, 81–83, 100, 119, 197,
208–209, 211, 366, 370–374
identity change 24, 83, 197, 211, 383
identity formation process 209–210,
243
incomplete information 323–324
individual agency 362–365
individual differences 61, 221, 240
individual differences predictors
modelling 333–337
individual life course 273–274, 362
individual orientations and actions 133
individuals across time 371–372
individuals and life stories 374
induction tree 303–308
inheritance 118–119
inline transitions 174–175, 369
innovative methods of analysis 25–27
institutional framing 50–51
integrative life narrative 241
interdependent lives see linked lives
interdisciplinarity 361, 380–386
interdisciplinarity pathways 9–10
interdisciplinary perspectives 3
intergenerational ambivalence 102
intergenerational relations 93, 95–97
intergenerational relations research
module 100–104
intergenerational-social transition 303
interlacing transition 176, 178, 182, 185
Interlacing transitions and
contemporary discourse 186
internal causality 18
International Union for the Scientific
Study of Population 279, 286
introversion/extraversion 239–240
Keane and Wolpin model 273
knowledge provided by induction tree
306–307
Knowledge-Verbal score 348
Kohlberg’s theory 198–199
Konstanz Inheritance Survey 118
labor force 136–137
labor market 36, 111, 134, 150, 167,
176–178, 182
last chapters of life 71, 81
latent basis curves 328–329
latent curve model (LCM) 315, 317, 351
Subject Index 407
latent difference scores modeling
344–345
latent mixture model 318, 343–344
latent scores 325, 335–337
layered life course patterns 176–178
learning theory 198
leave taking 178–180
Liberal Market States 49–50
life course analysis 173, 261, 303, 369
life course context 375
life course data 265, 283, 375, 377
life course events 11, 17–20
life course pattern 176–178
life course perspective 83, 119, 173, 308
life course research 3–5, 361
life course shaping elements 261
life course sociology 36, 38
life course stages 11, 13–15
life course terminology 58
life course theories 63, 244
life course trajectories 76, 133, 267, 278
life course transition 155, 185, 197, 227,
383
life event inventory 223, 234
life event research 19
life events moderation 162–170
life narrative 239, 246–247
life review 120, 230
life-span developmental psychology 223,
315
life-span psychology 42
life stories 245–250
Life Story Interview 245
life table 285
life trajectory 373
lifetime period 219
linear growth models 324–327
linear spline models 324–331
linked lives 6, 12, 22
living context 217, 222
lone motherhood 182–187
long-suppressed tendencies 244
longitudinal-analytical tools 308
longitudinal data 26, 199, 299–300, 317,
320, 324, 338, 379
longitudinal life-span data 315
longitudinal-regression model 288
loose coupling 63–64, 66, 168
macro–micro links 5
Note: See also micro–macro links
macrostructural perspective of life
course 131–132
making sense of agency 67–69
management of change 368
marital disruption 158–159
marital status 182–185
Markov transition model 288, 295–296,
379
Marriage 18–19
masculine trait 61–63
maternity and parental leave 178
memorized life transition 219
memory and personality relationship
226–227
memory research 219
methodological approach 374
methodological preliminaries of research
104–105
methodological strategy 160
methods of analysis 25–27
micro–macro links 133
Note: See also macro–micro links
microfication 47
middle age 224–225
mixed-effects model 27, 310, 334
mixture model 342–343
mixture transition distribution (MTD)
295
MLE parameter 337
mobility 285, 303
model-based expectations about growth
331
model-restricted fuzzy-set cluster
analysis 342
modeling approaches 267
SUBJECT INDEX408
monothetic divisive algorithm (MDA)
274
moral reasoning 199
mortality tables 283–284
mother–adult daughter relationship 114
motherhood 179–184
movement through social space 384
MTD model 295
multi-level model 26, 378–379
multi-level modeling 290–293
multilevel and multiprocess modeling
265–269
multilevel growth model 334–335
multiple ambivalence 117
multiple group perspective 338–340
multiple-regression analyses 223
multiprocess model results 268
narrative 250–253
narrative approaches 13, 237–239,
252
narrative identities 241–244
national identity 202, 249–250
negative transition 228
neuroticism 223, 225–226
new father 187
new–old view of intergenerational
relations 93
nonlinearity 328–331
non-normative event 157–159, 364
non-normative life course 155
nonparametric stage 271
Non-Verbal abilities 348
normative transition 382–383
objective facts versus subjective
reconstructions 218–221
observed data 321
occupational career structures 178
off-time transitions 112
old age 47–49, 222–229
ontogenesis 13
ontogenetic push factors 385
operationalizing the life course 77–81
optimal matching 26, 274, 277–278, 309,
379
overcoming resistance 64
parent–child relationship 93, 96,
106–107, 109
part-time work 142, 179, 363
partnership translations 228
past life transition 219, 222
Person’s Education 337
personal ambivalence 116–117
personal attribute models 38
personal constructs 132
personal relationship 96, 116
personality 101, 133, 198,
205, 217, 221–222, 238,
245, 363, 370
personality and identity 370–371
personality–cognition relation 226
personality dimensions 371–372
personality psychology 239, 245, 371,
384
physical displacement 212–213
planful competence 21, 40, 68, 132
political engagement 209
political identity 209–210
polynomial model 328, 333
polynomial nonlinearity in growth 328
Population Index database 296
post-modernity 98
potential predictor 303–304
predictors of early work 136
pre-life course perspective 84
principle of life course 63–64
production of life 60–61
program evaluation literature 269–272
propensity scores matching 271
proportional-hazard model 289
proportional hazard-odd model 289
prototypical transitions 198–200
psychological analyzer 365–370
psychological well-being 223–225
Subject Index 409
random-coefficients 334–335
random effect model 292, 310
rational choice theory 65
reconstructive recollection 229
redemption sequences 246–247
redemptive self themes 248–249
regression analyses 112, 225, 277
regression tree 303
regulation of well-being 217–218
residuals 331
response trajectories 169–170
responsibility 65–67
retraditionalization 369
retrospective data 376
rituals of transition 208, 212
role of the state 187
SCT and SIT 202
second demographic transition 285
second-order homogenous transition
296
self-categorisation 200–201, 373
Self-Categorisation Theory (SCT) 24,
197, 201, 203
self-concept 372
self-defining memories 246
sequence analysis 263, 274
sequence-mining approach 300
sequential institutionalization 51
shaping life courses 261
shared conception 72
shared heterogeneity 290–293
simple linear regression 291
simultaneous institutionalization 51
Six-Foci Model (SFM) 372, 374, 383
slopes as outcomes 335
small-world phenomenon 204
social address model 38
social analyzer 365–370
social behavior 363
social categorisation 200, 203–204
social demographers 9
social environment 372
social identity 24, 197, 200, 208, 211, 373
social identity change 197, 211–213
Social Identity Theory (SIT) 200
social-mobility analysis 299
social niche model 38
social psychological perspective 131,
370, 383
social psychology 8, 14, 16, 18, 197, 200,
385
social sciences 157, 237, 283
social structure 35, 37–39, 41, 64,
69, 362
social theory 59
social transition analysis with induction
trees 303
socialization 9, 70, 367, 381
socioeconomic achievement trajectories
140–144
sociology 67
socio-personal framing of transition 174
solidarity 102
solidarity perspective 95
specific life periods 44–49
stage theories 237–238
stage transition 199
statistical modeling of life events 288
statistical models 351
status attainment 135
status changes 81–83
status definitions to ambiguity 181
stopped working 179–180
stories 250–251
story development 242–245
storytelling 239, 242, 247, 252
strands of life course 185–189
stratification algorithmic modeling
culture 264
structural agency 363–365
structural ambivalence 117
structural equation modeling (SEM)
techniques 317
structure and agency see agency and
structure
SUBJECT INDEX410
structure 36–52
studying lives in time 237
subjective meanings of lives 238
survival analysis 378
teenage childbearing 157–158, 168–169,
269–270
temporal coherence 243
the good death 72–73
thematic coherence 243–244
theoretical bridges 383–384
theoretical linkage of interdisciplinarity
10–11
theory-making in social sciences 252
timeframes 186
time-stamped event 288
timing of events 156, 174, 369
traditional demographic analysis 286
trait conception 238
transition 11–13, 21–24, 188–189, 197,
365–370, 381–385
transition analysis instruments
187–188
transition characteristics 15
transition process 174, 211–213
transition reduction 175–176
transition to adulthood 131, 144–149,
156, 225, 262, 269
transitions and life events 218–221
Transitions and Life Perspectives in
Middle Age 224
transversal substantive themes 20–27
transversal themes 381–383
tree-growing principle 304
true novel 243
turning points 110–113, 178
two-staged latent class analysis 144
U.N. Convention on the Rights of the
Child 45, 48
understanding lives 252
unexplained variance 63–64, 83–84
United Nations Principles for Older
Persons 48
unplanned pregnancy 163–164
Verbal-Knowledge score 320–321, 331
Verbal–Non-Verbal bivariate coupling
model 347
vignettes use 105
Warmr process 301–302
welfare-state regime 49–50
well-being 222
work investment 136–144, 149–150
work pattern 137–138, 140–142, 146
World Fertility Survey (WFS) 262
Youth Development Study (YDS), the
135–136
youth 46, 94, 100, 134, 138, 139
Subject Index 411