Parent adolescent

24
Child Development, June 1997, Volume 68, Number 3, Pages 507-529 Parenting Styles, Adolescents' Attributions, and Educational Outcomes in Nine Heterogeneous High Schools Kristan L. Glasgow, Sanford M. Dornbusch, Lisa Troyer, Laurence Steinberg, and Philip L. Ritter This article examined the contemporaneous and predictive relations between parenting styles, adolescents' attributions, and 4 educational outcomes. Data were collected from adolescents attending 6 high schools in California and 3 high schools in Wisconsin during the 1987-1988 and 1988-1989 school years. The results of path analyses partially confirmed the central hypotheses. Adolescents who perceived their parents as being nonauthoritative were more likely than their peers to attribute achievement outcomes to external causes or to low ability. Furthermore, the higher the proportion of dysfunctional attributions made for academic suc- cesses and failures, the lower the levels of classroom engagement and homework 1 year later. Although adoles- cents' attributional style provided a bridge between parenting style and 2 educational outcomes, it did not fully explain the impact of parenting on those outcomes. Additional analyses within gender and ethnic subgroups reinforced the overall pattern of findings observed within the entire sample. INTRODUCTION Within the last decade, researchers in psychology, so- ciology, and education have shown that parental in- fluence does not decline as children mature into ado- lescents (Astone & McLanahan, 1991; Baker & Stevenson, 1986; Baumrind, 1991; Steinberg, Lam- born, Darling, Mounts, & Dornbusch, 1994; Steven- son & Baker, 1987). General parenting styles and spe- cific parenting practices continue to shape the development of adolescent competence, especially in the area of educational achievement. Yet, our under- standing of the processes through which parenting styles influence adolescent development remains limited. This research focuses on the role of attribution, a social psychological process, as a potential link be- tween parenting style and adolescents' educational outcomes. We present path models that relate parent- ing styles to adolescents' attributional style, which in turn is related to several achievement-related behav- iors and outcomes. Attributional style is a central concept in the literature on social cognition and plays a pivotal role in contemporary theories of achieve- ment motivation. It refers to a "tendency to make particular kinds of causal inferences, rather than oth- ers, across different situations and across time" (Met- alsky & Abramson, 1981, p. 38). Adolescents' attribu- tional style is conceptualized here as both a consequence of family interaction patterns and as a determinant of their achievement behaviors, perfor- mances, and expectations. This study builds on previous research in two ways. First, it portrays parenting style as an interper- sonal determinant of adolescents' attributions. This view differs from traditional psychological perspec- tives that portray causal attributions as products of personality or internal states, such as need for achievement, self-esteem, anxiety, or time perspec- tive (Arkin, Detchon, & Maruyama, 1981; Bar-Tal, 1978; Wolf & Savickas, 1985). The analyses in this ar- ticle test whether parenting styles affect the likeli- hood of expressing a particular style of causal in- ference. Although a number of studies have investigated parents' attributions for their children's school performance (e.g., Eccles-Parsons, Adler, & Kaczala, 1982; Yee & Eccles-Parsons, 1988), seldom is parenting behavior examined as a predictor of chil- dren's attributions (Hess & HoUoway, 1984). Second, the study uses a heterogeneous sample of high school students to evaluate the over-time rela- tions between attributional style and educational out- comes. Theorists view attributions as having affective and cognitive consequences that act as motivational determinants of achievement behavior on subse- quent tasks (Bar-Tal, 1978; Weiner, 1985). Although we do not directly test all aspects of this process, we do examine the effect of adolescents' attributional style on classroom engagement, homework, aca- demic achievement, and educational expectations. Parenting Style Parenting styles are constellations of parental atti- tudes, practices, and nonverbal expressions tbat char- es) 1997 by the Society for Research in Child Development, Inc. All rights reserved. 0009-3920/97/6803-0001$01.00

Transcript of Parent adolescent

Child Development, June 1997, Volume 68, Number 3, Pages 507-529

Parenting Styles, Adolescents' Attributions, and Educational Outcomesin Nine Heterogeneous High Schools

Kristan L. Glasgow, Sanford M. Dornbusch, Lisa Troyer, Laurence Steinberg,and Philip L. Ritter

This article examined the contemporaneous and predictive relations between parenting styles, adolescents'attributions, and 4 educational outcomes. Data were collected from adolescents attending 6 high schools inCalifornia and 3 high schools in Wisconsin during the 1987-1988 and 1988-1989 school years. The results ofpath analyses partially confirmed the central hypotheses. Adolescents who perceived their parents as beingnonauthoritative were more likely than their peers to attribute achievement outcomes to external causes orto low ability. Furthermore, the higher the proportion of dysfunctional attributions made for academic suc-cesses and failures, the lower the levels of classroom engagement and homework 1 year later. Although adoles-cents' attributional style provided a bridge between parenting style and 2 educational outcomes, it did not fullyexplain the impact of parenting on those outcomes. Additional analyses within gender and ethnic subgroupsreinforced the overall pattern of findings observed within the entire sample.

INTRODUCTION

Within the last decade, researchers in psychology, so-ciology, and education have shown that parental in-fluence does not decline as children mature into ado-lescents (Astone & McLanahan, 1991; Baker &Stevenson, 1986; Baumrind, 1991; Steinberg, Lam-born, Darling, Mounts, & Dornbusch, 1994; Steven-son & Baker, 1987). General parenting styles and spe-cific parenting practices continue to shape thedevelopment of adolescent competence, especially inthe area of educational achievement. Yet, our under-standing of the processes through which parentingstyles influence adolescent development remainslimited.

This research focuses on the role of attribution, asocial psychological process, as a potential link be-tween parenting style and adolescents' educationaloutcomes. We present path models that relate parent-ing styles to adolescents' attributional style, which inturn is related to several achievement-related behav-iors and outcomes. Attributional style is a centralconcept in the literature on social cognition and playsa pivotal role in contemporary theories of achieve-ment motivation. It refers to a "tendency to makeparticular kinds of causal inferences, rather than oth-ers, across different situations and across time" (Met-alsky & Abramson, 1981, p. 38). Adolescents' attribu-tional style is conceptualized here as both aconsequence of family interaction patterns and as adeterminant of their achievement behaviors, perfor-mances, and expectations.

This study builds on previous research in twoways. First, it portrays parenting style as an interper-

sonal determinant of adolescents' attributions. Thisview differs from traditional psychological perspec-tives that portray causal attributions as products ofpersonality or internal states, such as need forachievement, self-esteem, anxiety, or time perspec-tive (Arkin, Detchon, & Maruyama, 1981; Bar-Tal,1978; Wolf & Savickas, 1985). The analyses in this ar-ticle test whether parenting styles affect the likeli-hood of expressing a particular style of causal in-ference. Although a number of studies haveinvestigated parents' attributions for their children'sschool performance (e.g., Eccles-Parsons, Adler, &Kaczala, 1982; Yee & Eccles-Parsons, 1988), seldom isparenting behavior examined as a predictor of chil-dren's attributions (Hess & HoUoway, 1984).

Second, the study uses a heterogeneous sample ofhigh school students to evaluate the over-time rela-tions between attributional style and educational out-comes. Theorists view attributions as having affectiveand cognitive consequences that act as motivationaldeterminants of achievement behavior on subse-quent tasks (Bar-Tal, 1978; Weiner, 1985). Althoughwe do not directly test all aspects of this process, wedo examine the effect of adolescents' attributionalstyle on classroom engagement, homework, aca-demic achievement, and educational expectations.

Parenting Style

Parenting styles are constellations of parental atti-tudes, practices, and nonverbal expressions tbat char-es) 1997 by the Society for Research in Child Development, Inc.All rights reserved. 0009-3920/97/6803-0001$01.00

508 Child Development

acterize the nature of parent-child interactions acrossdiverse situations (Darling & Steinberg, 1993). Con-temporary research on parenting styles derives fromBaumrind's (1971, 1978) well-known studies of chil-dren and their families. Baumrind's conceptualiza-tion of parenting style is based on a typological ap-proach to the study of family socialization practices.This approach focuses on the configuration of differ-ent parenting practices and assumes that the impactof any one practice depends, in part, on the arrange-ment of all others. Variations in the configuration ofmajor parenting elements (such as warmth, involve-ment, maturity demands, and supervision) producevariations in how a child responds to parental influ-ence. From this perspective, parenting style is viewedas a characteristic of the parent that alters the effec-tiveness of family socialization practices and thechild's receptiveness to such practices (Darling &Steinberg, 1993).

Baumrind's (1971) parenting style typology identi-fied three qualitatively different patterns of parentalauthority—authoritarian, authoritative, and permis-sive. Maccoby and Martin (1983) subsequently trans-formed this typology by categorizing families ac-cording to their levels of parental demandingness(control, supervision, maturity demands) and re-sponsiveness (warmth, acceptance, involvement).Redefining parenting styles in terms of the interac-tion between these two underlying dimensions pro-duced a fourfold typology. A primary difference be-tween Baumrind's earlier model and Maccoby andMartin's refinement is that the latter differentiates be-tween two types of permissive parenting.

The extended parenting style typology distin-guishes between nondemanding families that vary intheir level of responsiveness. Parents characterizedby low demandingness and high responsiveness en-gage in an indulgent style of parenting. These parentsare tolerant, warm, and accepting. Yet, they exerciselittle authority, make few demands for mature behav-ior, and allow considerable self-regulation by thechild or adolescent. By contrast, parents who are nei-ther demanding nor responsive display a neglectfulor uninvolved pattern of parenting. These parents donot monitor their children's behavior or support theirinterests. Whereas indulgent parents are committedto their children, neglectful parents, often preoccu-pied with their own problems, are disengaged fromparental responsibilities (Baumrind, 1991; Lamborn,Mounts, Steinberg, & Dornbusch, 1991; Maccoby &Martin, 1983).

The parenting typology also distinguishes be-tween demanding families that vary in their level of

responsiveness. Authoritarian parents are highly de-manding and unresponsive. These parents attempt tomold and control the behavior and attitudes of theirchildren according to a set of standards. They tendto emphasize obedience, respect for authority, andorder. Authoritarian parents also discourage verbalgive-and-take with their children, expecting rules tobe followed without further explanation (Baumrind,1991).

Authoritative parents maintain an effective bal-ance between high levels of demandingness and re-sponsiveness. These parents establish and firmly en-force rules and standards for their children'sbehavior. They consistently monitor conduct and usenonptmitive methods of discipline when rules are vi-olated. Socially responsible and mature behavioris expected and reinforced. Authoritative parentsare also warm and supportive. They encourage bi-directional communication, validate the child's indi-vidual point of view, and recognize the rights of bothparents and children (Baumrind, 1991; Dornbusch,Ritter, Leiderman, Roberts, & Fraleigh, 1987).

Although the parenting style typology was origi-nally developed for research on family socializationpractices during childhood, it also has been used tostudy the links between family interaction patternsand areas of adolescent functioning. This work showsthat the beneficial influence of authoritative parent-ing does not diminish during adolescence. Authorita-tively reared adolescents consistently score higher onmeasures of psychosocial competence and schoolachievement, and lower on measures of internal dis-tress and problem behavior, than do adolescentsfrom nonauthoritative families (Baumrind, 1989,1991; Hein & Lewko, 1994; Lamborn et al., 1991; Paul-son, 1994; Steinberg, Lamborn, Dornbusch, & Dar-ling, 1992). Although there are ethnic and culturalvariations in the impact of parenting styles (Chao,1994; Darling & Steinberg, 1993; Dornbusch et al.,1987), this empirical pattern appears to transcendgender, family structure, age, and social class divi-sions.

By contrast, adolescents with neglectful parentsshow the lowest level of adjustment among the fourtypes of parenting styles. These adolescents are themost disadvantaged with respect to measures of so-cial competence, academic achievement, and psycho-logical adjustment (Baumrind, 1991; Lamborn et al.,1991). Moreover, the deleterious effects of neglectfulparenting accumulate over time (Steinberg et al.,1994).

Despite years of research on parenting styles, thereare surprisingly few studies of the mechanisms that

Glasgow et al. 509

intervene between parenting and adolescents'achievement outcomes. Attempts to explain the in-fluence of parenting style have generally focused onspecific parental behaviors and on internal character-istics of youth. A recent study, for example, foundthat parental involvement in schooling partly medi-ated the relation between authoritative parenting andadolescent school performance (Steinberg et al.,1992). It has also been shown that authoritative par-enting promotes academic success through a positiveeffect on adolescents' psychological orientation to-ward work (Steinberg, Elmen, & Mounts, 1989). Re-search on authoritarian parenting found that highparental control was associated with heightened self-consciousness among young adolescents. Increasedself-awareness, in turn, predicted lowered self-assessments of ability in math, social, and sports do-mains (Yee & Flanagan, 1985).

In this article, we attempt to show that attribu-tional processes provide an additional mechanismlinking parenting styles to adolescents' educationaloutcomes. Reviews of the research literature con-clude that there have been too few attempts to iden-tify the social determinants of attributions (Critten-den, 1983; Hess & Holloway, 1984; Stryker &Gottlieb, 1981). One exception is the work of Dweckand her colleagues, who found that the attributionsof children are shaped by teacher feedback within theschool (Dweck, Davidson, Nelson, & Enna, 1978).This study investigates whether interaction patternswithin the family have a similar influence on adoles-cents' style of causal inference.

Causal Attributions, Achievement Behavior,and Outcomes

Attributions are inferences about the perceivedcauses of observed behaviors and events. Attributinga cause or set of causes to a particular outcome affectsthe ways social actors understand and manage eventsin their environment (Heider, 1958). In effect, causalinferences create a guide for future behavior (Kelly,1967). Although the majority of attributional researchhas taken place in experimental settings (Weiner,1985), a number of studies have examined studentattributions for outcomes on actual academic tasks(Arkin & Maruyama, 1979; Arkin et al., 1981; For-syth & McMillan, 1981; Griffin, Combs, Land, &Combs, 1983). In both artificial and natural contexts,the perceived causes of task success and failure aremost often attributed to ability, effort, luck, or taskdifficulty.

These attributions figure prominently in Weiner's

(1979,1980,1985) attributional model of achievementmotivation, one of the influential perspectives onachievement-related behavior. His perspective ex-plains how causal attributions link achieved out-comes to future performances on identical or relatedtasks. According to the model, causal attributionsproduce affective reactions and expectations for fu-ture success that, in turn, guide achievement-relatedbehaviors on subsequent tasks (e.g., task choice, per-sistence, intensity, and diligence).

Attributions are classified along three imderlyingcausal dimensions in Weiner's model. The locationof an attribution on these dimensions determines itsimpact on affect, level of expectancy, and achieve-ment behavior (Weiner, 1985). The locus of controldimension identifies whether the observed outcomewas caused by factors within the individual (internal)or the situation (external). This dimension influencesaffective reactions to task outcomes. The stability di-mension classifies perceived causes as either fixed orvariable in time and determines the subjective proba-bility of success on future tasks. The controllabilitydimension distinguishes whether causes are subjectto personal control.

There is general agreement in the literature thatsome patterns of attribution facilitate studentachievement more than others do. Attributing aca-demic success to high ability or effort and academicfailure to insufficient effort is considered the optimalresponse style (Bar-Tal, 1978; Platt, 1988). Ability isa stable-internal cause. Attributing success to abilityproduces pride and an enhanced expectation for con-tinued success in the future (Weiner, 1985). Effort isan unstable-internal cause that is subject to personalcontrol. Attributing either success or failure to effortgives rise to an expectation that future success can beachieved through continuing or greater expendituresof effort (Bar-Tal, 1978; Dweck, 1975; Dweck & Bush,1976). These attributions, through enhanced expec-tancy, positive affect, and controllability, are pre-dicted to increase the likelihood of engaging inachievement-maximizing behaviors on subsequenttasks.

By contrast, the most debilitating pattern of causalinference occurs when success is attributed to luckand failure to lack of ability (Maehr & Nicholls, 1980;Willig, Harnisch, Hill, & Maehr, 1983). Attributingfailure to lack of ability produces feelings of shameand an expectation that similar failures will occur onlater tasks, regardless of the level of personal effort(Weiner, 1985). Failure attributions to stable causes,like ability, decrease the expectancy for future suc-cess. Luck is an external-unstable cause that cannot

510 Child Development

be personally controlled. Accordingly, attributingsuccess or failure to luck results in an expectation thatfailure may nevertheless occur in the future (Bar-Tal,1978). These attributions are expected to indirectlylower subsequent levels of achievement behavior.

There has been some debate regarding the appro-priateness of using measures of actual achievementoutcomes (e.g., test scores, grade point average) totest Weiner's attributional model. Weiner (1983) ar-gued against this strategy, claiming that these out-comes are determined by too many factors outsidethe scope of his theory. Others have disagreed withhis position, noting instances where Weiner previ-ously used achievement outcomes in his formula-tions (Covington & Omelich, 1984a, 1984b).

Although a few studies have disconfirmed predic-tions derived from Weiner's model (Bernstein, Ste-phan, & Davis, 1979; Covington & Omelich, 1979,1984b), others support the claim that attributionshave consequences for academic outcomes (Wagner,Spratt, Gal, & Paris, 1989; Willig et al., 1983). Platt(1988), for instance, used the attributional model toexplain variations in first-term college performance.He found that ability and effort attributions for highschool success had indirect positive effects on collegegrades. These attributions influenced measures of ex-pectancy and achievement behavior, which in turnaffected college grades. Related studies on learnedhelplessness also found that children who had a mal-adaptive attributional style performed less well inschool than did their peers (Fincham, Hokoda, &Sanders, 1989; Nolen-Hoeksema, Seligman, & Gir-gus, 1986).

We use Weiner's framework in this study as a ba-sis for evaluating the relations between attributionsand educational outcomes in a heterogeneous popu-lation of high school youth. Adolescents' attributionsfor success and failure in four academic subjects areexamined. The pattern of attributions across the foursubjects is characterized as a functional or dysfunc-tional attributional style, which represents an adoles-cent's tendency to make particular kinds of causal in-ferences across different academic domains.

A functional attributional style refers to a patternof causal inference that reflects a sense of personalconfidence in one's performance capacities, as well asa willingness to accept responsibility for one's actionsand their consequences. Adolescents who express afunctional attributional style are likely to view them-selves as efficacious agents, capable of successfullyperforming the behaviors that are necessary to bringabout a desired outcome (cf. Bandura, 1977). In con-crete terms, a functional attributional style denotesthe tendency to make ability or effort ascriptions for

task successes and to make effort ascriptions follow-ing task failures. As noted, these attributions pro-mote achievement behavior and outcomes throughpositive affect, expectancy, and a perception of con-trollability.

A dysfunctional attributional style, by contrast,implies lack of faith in one's performance capacitiesand a reluctance to assume responsibility for one'sbehavior and the outcomes it may generate. Adoles-cents who use a dysfunctional attributional style arelikely to view themselves as less able and may there-fore place responsibility onto external factors whenmaking causal inferences. This pattern of causal in-ference is manifest by the tendency to make externalattributions for either task success or failure, or attri-butions to low ability following failure. Dysfunc-tional attributions are theoretically expected toreduce the likelihood of engaging in achievement-maximizing behaviors and, therefore, of succeedingacademically.

Linking Parenting Style, Attributional Style, andEducational Outcomes

This study is an initial attempt to integrate and ex-tend existing research on parenting styles, attribu-tion, and areas of adolescent achievement. We be-lieve that the integration will produce a greatercontribution to each theoretical framework and toour understanding of adolescence if central conceptsfrom each tradition are retained. The literature onfamily processes provides the conceptual basis for re-lating the four distinct types of parenting to adoles-cents' attributional style. Weiner's theoretical modelprovides the basis for linking dysfunctional attribu-tions to achievement-related outcomes. Dysfunc-tional attributions are viewed here as a bridge be-tween parenting styles and adolescents' educationaloutcomes.

General styles of parenting may be enactedthrough concrete practices that either foster or inhibitthe development of self-reliance, responsibility, anda sense of performance adequacy in adolescents, thequalities underpinning attributional tendencies. Ifcertain modes of parenting create family environ-ments that inhibit these developing qualities, thenthey may have the concurrent effect of increasing anadolescent's propensity for expressing a dysfunc-tional style of causal inference. Previous research in-dicates that nonauthoritative styles of parenting areless effective than authoritative parenting in promot-ing a sense of instrumental competence among ado-lescents (Baumrind, 1978, 1991).

Instrumentally competent adolescents are able to

Glasgow et al. 511

"balance other-oriented, rule-following tendencieswith individualistic, autonomous, active thinking"(Darling & Steinberg, 1993, p. 492). The demanding-ness and responsiveness characteristic of authorita-tive parenting contribute to the development of in-strumental competence in adolescents. For example,the bi-directional communication and joint decisionmaking associated with this mode of parenting helpsadolescents to understand the links betw^een their be-havior and its consequences (Dornbusch, Ritter, Leid-erman, Mont-Reynaud, & Chen, 1990). Adolescentswho describe their parents as authoritative reporthigher self-perceptions of academic ability and aremore self-reliant than are youth from indulgent, au-thoritarian, or neglectful families (Baumrind, 1991;Cooper, Grotevant, & Condon, 1983; Lamborn et al.,1991; Steinberg et al, 1994).

Indulgent parenting fosters instrumental compe-tence to sortie extent in adolescents, but to a lesserdegree than authoritative parenting (Baumrind,1991). Adolescents from indulgent families score rel-atively high on measures of social competence andself-reliance, but slightly lower than authoritativelyreared adolescents in work orientation and self-perception of academic ability (Lamborn et al., 1991).

The high demandingness and lack of respon-siveness characteristic of authoritarian parentsthwarts the development of instrumental compe-tence in adolescents (Baumrind, 1991). The emphasison conformity to external rules and standards, cou-pled with a lack of parental responsiveness, commu-nicates to youth that self-regulation, individuality,and autonomous thinking are not valued or ex-pected. Continual use of extrinsic reinforcements andthe overemphasis on obedience to authority under-mines adolescents' perceptions of competence, self-reliance, and internal motivations to achieve (Lam-born et al., 1991; Lepper & Greene, 1978; Steinberget al., 1994).

Neglectful parents' lack of both responsivenessand demandingness compromises diverse areas ofadolescent functioning. Adolescents who describetheir parents as neglectful exhibit the lowest levels ofself-regulation and cognitive competence (Baumrind,1991; Lamborn et al, 1991; Steinberg et al., 1994). Be-cause neglectful parents seldom portray behavioralexpectations and standards as meaningful or reason-able, these youth are not aided in understanding theconnections between self-initiated behaviors andtheir potential consequences.

The pattern of associations between parentingstyles and instrumental competence provides thefoundation for positing additional links to adoles-cents' attributional tendencies, which reflect per-

ceived levels of self-reliance, responsibility, and per-formance capacity. The study tests two hypotheses.First, adolescents from authoritarian, indulgent, andneglectful families will be more likely to evidence adysfunctional attributional style than will authorita-tively reared adolescents. Second, the greater the ad-olescents' tendency to make dysfunctional attribu-tions for academic successes and failures, the lowerthe subsequent levels of classroom engagement,homework, academic achievement, and educationalexpectations. These achievement-related behaviorsand outcomes are important because of their impacton a student's likelihood of entering an institution ofhigher education. Postsecondary educational creden-tials have become prerequisites for entry into manyhigh-status occupations, whereas high school gradu-ation no longer brings substantial economic returns(Hout, 1988).

METHOD

Sample

This study analyzes a subset of data from ques-tionnaires administered to approximately 11,000 highschool students, attending six high schools in Califor-nia and three high schools in Wisconsin. The nineschools were selected to provide a socially and eco-nomically diverse sample of adolescents. Question-naires were given to all students in attendance duringthe 1987-1988 and 1988-1989 school years. Becauseof its length, the survey was divided into two partsand was administered on two separate days in eachschool year. For each survey administration, approxi-mately 80% of the students filled out questionnaires,approximately 5% chose not to participate, and ap-proximately 15% were absent on the day of question-naire administration. Informed consent was obtainedfrom all participating students (Steinberg et al., 1994).

The total sample across the two academic years in-cluded 6,902 students. Responses from 2,353 stu-dents, a subset of the total over-time sample, are ana-lyzed in this article. The smaller size of the studysample is predominantly due to the method of classi-fying families into one of the four parenting styles.All families that were in the middle third on one ormore of the parenting dimensions were eliminatedfrom the analysis (described below).

Measures

Measurements of parenting styles, attributionaltendencies, and adolescents' educational outcomeswere created from items in the student question-

512 Child Development

naires. There has been some debate in the literatureregarding the relative merits of using adolescent- ver-sus parent-reported information to study family so-cialization practices. Adolescents' and parents' per-ceptions of family environments do not alwayscoincide (Collins, 1990; Paulson, 1994; Smetana,1988), and neither source of data is completely freeof bias. There are important theoretical and method-ological reasons for identifying the bases of these per-ceptual differences. Although there may be points ofdivergence, there is a clear role for the study of be-haviors reported by adolescents. The ways in whichadolescents socially construct and interpret eventswithin the home are crucial for understanding howthey react to those events. The subjective image ofparenting styles reported by adolescents providesone window through which to view the internal dy-namics of the family environment.^

It is possible, however, that adolescents who re-port dysfunctional attributions will be biased towardperceiving their parents as nonauthoritative. A re-verse causal process may be occurring, in whichadolescents influence their parents' modes of social-ization. Unfortunately, we are prevented from evalu-ating the effects of adolescents on their parents be-cause the 1988-1989 survey instrument did notinclude comparable measures of parenting styles(Steinberg et al., 1994). Because the sample containsonly two waves of data, this temporal sequence can-not be ruled out by the present study. We leave openthe prospect that the link between parenting stylesand adolescents' cognitions may be more complexthan is represented by existing theory and research.

Demographic Characteristics of Respondents

The demographic characteristics of students in thetotal and study samples are described in Table 1.Analyses comparing the demographic profiles of thetotal and study samples revealed no significant dif-ferences in the percentages of youth representingeach social category. The study sample is as ethni-cally and socially diverse as the total over-time sam-ple and is nearly evenly divided between males andfemales.

Students were asked to select their primary ethnicidentification from 16 possible categories. Problem-

1. An earlier study compared students' self-reports to highschool transcripts and to parent responses (Dornbusch et al.,1990). Student responses were highly correlated to objectiveschool data. Although parents' and students' reports of familyinteractions often differed markedly, identical patterns of rela-tions between family interactions and adolescent outcomes werefound when the two sources of information were compared.

Table 1 Demographic Characteristics of Respondents in theTotal (N = 6,902) and Study (N = 2,353) Samples

Characteristics

Ethnicity:African AmericanAsian AmericanHispanic AmericanNon-Hispanic White

Sex:FemaleMale

Parental education:Some college or 2-year degree4-year college degree or beyond

Family structure:IntactNonintact

Age group:15 years or younger16 years or older

TotalSample (%)

8.715.115.360.9

50.149.9

51.848.2

66.433.6

60.339.7

StudySample (%)

9.114.813.962.2

51.848.2

51.948.1

66.633.4

58.941.1

atic identifications were resolved by comparison withschool records. There were sufficient cases for analy-sis of four major ethnic groups: African American,Asian American, Hispanic American, and non-Hispanic White. Three ethnic dummy variables werecreated to represent African Americans, AsianAmericans, and Hispanic Americans. Non-HispanicWhites were the comparison group.

The indicator of family socioeconomic status wasparental education. Students were asked to indicatethe highest level of education completed by each bio-logical parent. The options ranged from some gradeschool to professional or graduate degree. Responsecategories were subsequently recoded to approxi-mate the number of years of schooling. The numberof years completed by each respondent's mother andfather were averaged to yield a single, continuousmeasure of parental education. For descriptive pur-poses, the level of parent education was dichoto-mized into two categories at the approximate me-dian: up to "some college or 2-year degree" (low) and"4-year college graduate" or beyond (high).

Family structure was determined from the re-sponses students gave to the question, "Which par-ents or guardians do you live with?" A dummy vari-able was created for intact families (coded 1) versusall other major family forms (i.e., single parent andstep-parent families, coded 0).

The age of respondents was calculated in terms oftheir age in months in November 1987, the date thefirst survey was administered. Descriptive informa-tion was computed by classifying students into two

Glasgow et al. 513

age groups; 15 years and your\ger and 16 years andolder. In all other analyses, age was treated as a con-tinuous variable.

Parenting Style

The parenting style index used in this study wasdeveloped by Steinberg and his colleagues (Lambornet al., 1991; Steinberg et al., 1994). It was originallycreated to approximate the categorical typology pro-posed by Baumrind (1971) and Maccoby and Martin(1983). Qualitatively different types of parenting en-vironments are identified by characterizing parentsin terms of their responsiveness and demandingness.We used two scales as indicators of these dimen-sions—acceptance / involvement and strictness / Su-pervision, respectively. The acceptance/involvementscale assesses the degree to which adolescents viewtheir parents as being responsive, loving, and in-volved. The scale was computed by taking the meanof 10 items from the 1987-1988 survey (e.g., "When[my mother or father] wants me to do something, [heor she] explains why"; "1 can count on [him or her]to help me out if I have some kind of problem"; a =.72, M = .81, SD = .11, range = .25-1.0; Steinberg etal., 1994). The strictness/supervision scale measuresadolescents' perceptions of parental monitoring andbehavioral supervision. Nine items from the first sur-vey were included in this scale (e.g., "My parentsknow exactly where I am most afternoons afterschool"; "In a typical week, what is the latest you canstay out on school nights?"; a = .76, M = .74, SD =.13, range = .30-1.0; Steinberg et al., 1994). Several ofparenting items were scaled according to a three-point Likert format, whereas others were true-falsedistinctions. Composite indices were computed byweighting items to adjust for these scaling differ-ences. A listing of all items contained in each scalecan be found in Lamborn et al. (1991).

In keeping with the typological approach to familysocialization, the scales were used to assign familiesto one of the four parenting style categories (Lam-born et al., 1991; Steinberg et al, 1994). The samplewas trichotomized on the acceptance/involvementscale and on the strictness/supervision scale. To becertain that these four groups defined distinct catego-ries, families scoring in the middle third on either di-mension were removed from the analyses. Familiesscoring in the upper third on both scales were de-fined as authoritative (N = 817), whereas neglectfulfamilies (N = 834) ranked in the lowest third on eachscale. Authoritarian families (N = 451) scored in thehighest third on strictness and in the lowest third onacceptance. Indulgent families (N = 251) were de-

fined by the reverse pattern, highest in acceptance/involvement and lowest in strictness/supervision.

A total of 2,353 families were classified into one ofthe four parenting styles suggested by Maccoby andMartin (1983). The means and standard deviations onacceptance/involvement and on strictness/supervi-sion for each of the four groups are reported inSteinberg et al. (1994). Three dummy variables werecreated here to represent adolescents' perceptions ofauthoritarian, indulgent, and neglectful parenting.Authoritative parenting formed the comparisongroup.

As Steinberg and his colleagues (Lamborn et al.,1991; Steinberg et al., 1994) have previously argued,the tertile split procedure allows us to be cor\fidentthat the four parenting categories represent qualita-tively different types of family environments. Theclassification of families into these categories is madefor heuristic, not diagnostic, reasons (Lamborn et al.,1991). We recognize that this operationalization lim-its the external validity of the study. The split andassignment procedure is sample specific and elimi-nates a large number of cases from the analysis. De-spite these caveats, the decision to use tertile splitprocedure is appropriate given the goals of thisstudy, and it is consistent with prior research. Thisdecision should not minimize the importance of for-mulating and validating alternative measures of par-enting style.

We also performed analyses using three continu-ous measures of parenting styles. The measuresavailable to us were of authoritarian, authoritative,"and permissive parenting. These indices were origi-nally created by Dornbusch and his colleagues(Dornbusch et al., 1987). This earlier study did notdifferentiate between indulgent and neglectful par-enting. The use of continuous parenting scores didnot alter the overall pattern of results. Adolescents'perceptions of permissive and authoritarian parent-ing were positively associated with dysfunctional at-tributions, and authoritative parenting was nega-tively associated.

Dysfunctional Attributional Style

Attributions were elicited for grades received infour different academic subjects (English, math, so-cial studies, science). In 1987-1988, students wereasked: "Everyone gets a good [poor] grade some-times. When you get a good [poor] grade in each ofthese classes, which reason usually causes the good[poor] grade?" Students were instructed to select themost important one or two reasons for each subject.The perceived causes were luck, effort, teacher bias.

514 Child Development

task difficulty, and ability.^ Functional responses in-cluded ability and effort attributions for good gradesand effort attributions for poor grades. Dysfunctionalresponses included attributions to lack of ability inresponse to failure and to luck, teacher bias, or taskdifficulty for either success or failure.

The dysfunctional attributional style index wascomputed by summing attributional responsesacross the four academic subjects for each respon-dent, then calculating the proportion of the totalnumber of responses that were dysfunctional {range= 0-1). The proportion of dysfunctional responsesreported by adolescents indicates their tendency toattribute academic outcomes to either external causesor to low ability. When each of the four subjects wasconsidered as an item making up the total attribu-tional score, Cronbach's alpha was .61. Analysescomparing the proportions of dysfunctional attribu-tions reported in the total and study samples showedno significant difference between the two samples.

Classifying the success attribution "it was an easytask" as dysfunctional differs slightly from Weiner'stheoretical model. Task difficulty is sometimesviewed as a stable cause (Bar-Tal, 1978). Given a suc-cessful outcome, this attribution is purported to in-creases the expectation for success on later tasks. Inpractical terms, however, "task difficulty" is not afunctional attribution in an educational setting (cf.Platt, 1988). Students who make attributions to taskdifficulty are not claiming personal responsibility fortheir academic outcomes. Because courses changeover time, these students may be less inclined to com-mit themselves to future tasks that are seeminglymore difficult.

In another respect, it could be argued that perceiv-ing a task as easy implies high ability. This functionalinference is justified when a student also believes thatthe task was not easy for other students. We expectthat a student in this situation will select "ability"rather than "task difflculty" as the most importantcause of a good grade. If the student perceives thetask as easy for all other students, then "task diffi-culty" does not necessarily imply exceptional ability.Accordingly, the student should attribute success tothe ease of the task. This normative response is ap-propriately classified as a dysfunctional attribution.A student who attributes success to ability and to taskdifficulty is probably unsure of whether or not the

2. For good grades, the attributions were: "I was lucky," "Iworked hard," "The teacher liked me," "The class was easy,"and "I'm good at this subject." The attributions for poor gradeswere: "I had bad luck," "I didn't work hard," "The teacher wasunfair," "The class was hard," and "I'm not good at this sub-ject." For each of these questions, students could also choose"Never taken this subject," if appropriate.

task was easy for other students. Selecting both ofthese attributions effectively neutralizes their contri-bution to attributional style.

Educational Outcomes

Classroom engagement. The index of classroom en-gagement, one indicator of achievement behavior,measures the extent to which students are attentiveto and involved in academic classwork (Steinberg etal., 1992). In each year, students were asked to indi-cate the frequency with which they pay attention toclasswork, concentrate, try hard, and let their mindswander (reverse coded) in four academic subjects(math, English, social studies, and science). All re-sponses were coded on a 5 point scale, rangingfrom "never" to "always" or to "almost every day."Responses to each item were averaged across the foursubjects, then the mean of the four averaged itemswas computed to create a single index (a = .70, range= 1-5).

Homework. Spending time on homework, a secondindicator of achievement behavior, usually has a ben-eficial impact on academic performance (Natriello &McDill, 1986), particularly for low-ability students(Keith, 1982). In each survey, students reported theamount of time spent on homework each week (in-cluding reading assignments) in the four main aca-demic subjects: None, about 15 min, about 30 min,about an hour, about 2 or 3 hr, about 4 hr or more.Responses were recoded to approximate the numberof hours spent on homework each week and wereaveraged across the four classes into a single index(range = 0-5).

Academic achievement. Students' self-reportedgrades provided the indicator of academic schoolperformance. Self-reported grades are highly corre-lated with the distribution of grades on official schooltranscripts (Dornbusch et al., 1987). In each survey,students were asked, "Which of the following bestdescribes your grades so far in high school? MostlyA's, about half A's and half B's, mostly B's, about halfB's and half C's, . . . or mostly below D." Responseswere converted into the standard four-point gradingscale [range = 0-4).

Educational expectations. A key social psychologicalvariable affecting educational attainment and occu-pational mobility is a student's educational expecta-tions (Sewell & Hauser, 1980). After students wereasked how far they would like to go in school, theywere asked to give a more realistic response: "Con-sidering your situation, what is the highest level thatyou really expect to go in school?" There were six pos-sible responses, ranging from "Leave school as soonas possible" to "Finish college and take further train-

Glasgow et al. 515

ing (medical, law, graduate school, etc.)" {range =1-6).

Table 2 presents the intercorrelations, means, andstandard deviations of variables from the entire sam-ple (i.e., including adolescents who were not as-signed to one of the four parenting groups). Analysescomparing the means on the educational outcomesfrom the total and study samples revealed no signifi-cant differences between the two samples (althoughthere was a tendency, p < .10, for students in thestudy sample to do slightly more homework in 1988-1989). Thus, although smaller in size, the study sam-ple's demographic characteristics and outcomes didnot differ appreciably from those in the total over-time sample.

Plan of Analysis

Path analytic techniques were used to test thestudy's central hypotheses. Path coefficients were de-rived from a series of OLS regression equations (us-ing listwise deletion of missing values). These coeffi-cients are equivalent to standardized regressioncoefficients. Because the path coefficients are "scalefree," effects within the same equation may be com-pared to assess the relative impact of each explana-tory variable on the outcome of interest.

Four separate path models were analyzed, one foreach educational outcome. These models show boththe contemporaneous and predictive relations be-tween parenting styles, dysfunctional attributions,and the four outcomes. Each model depicted the de-mographic characteristics and the parenting indicesas determinants of attributional style and a 1987-1988 outcome. All of these causally prior variables,in turn, were posited as predictors of the relevantoutcome in 1988-1989.

Using the Time 1 outcome scores as covariates inthe analysis generates constrained estimates of the re-lations between attributional style and Time 2 out-comes. The high correlations between the outcomescores over time limit the amount of variance the re-maining predictors can explain.^ Nevertheless, failingto include these paths increases the risk of attributingmore explanatory significance to attributional stylethan may be warranted. We anticipated that the effectof attributional style on the 1988-1989 outcomeswould be relatively small once the earlier outcome

3. The intercorrelations of the outcome variables over the 1year period were as follows: classroom engagement, r = .57;homework, r = .50; academic achievement, r = .73; educationalexpectations, r = .64. These correlation coefficients differ fromthe values reported in Table 2 because they were generatedwithin the regression analyses. All associations were significantat the p < .01 level.

levels were controlled. To evaluate this possibility,the Time 2 outcome score in each model was esti-mated with and without the 1987-1988 covariate.Comparisons across these equations confirmed thisexpectation. These results are presented in the Ap-pendix.

Although the equation predicting attributionalstyle is the same in each of the four models, theseanalyses do not represent four independent tests ofthe first hypothesis. We analyzed information frommany of the same respondents in each path model.Yet the samples are not identical across the four mod-els. The number of cases analyzed in each model var-ied, depending on the educational outcome. The totalnumber of cases ranged from 1,275 to 1,576 students.Rather than conducting all analyses on a single sam-ple of youth, therefore excluding roughly 300 stu-dents and further biasing the sample, four separatemodels were estimated. To the extent that the fournonindependent tests produce similar results, onecan conclude that the findings are sufficiently robustso as to be unaffected by small changes in the sampleor by changes in the specific educational outcome be-ing examined.

Path analyses were also performed separatelywithin each gender and ethnic group. Recent studiesshow that the impact of parenting styles on areas ofadolescent adjustment differ depending on the socialcontext that gives them meaning (e.g., Chao, 1994).Cultural systems are interpretative frameworks thatgive meaning to daily interactions. Parenting stylesmay have very different implications for adolescentfunctioning when considered in terms of the culturalsystem operating within the family and the localcommunity. In Asian cultures, for example, some as-pects of high parental demandingness symbolize par-ents' concern for and involvement with their chil-dren. According to Chao (1994), parental control inAsian families often represents a type of organiza-tional control that is used to promote and maintainfamily harmony. The path analyses by gender andethnicity explore the extent to which family processesin diverse cultural settings create similar outcomesfor adolescents.

RESULTS

The results of the path analyses are shown in Figures1 through 4. The path diagrams include only statisti-cally significant path coefficients (p < .05 or better)from the first, second, and fourth regression equa-tions in each model. The standard errors, unstan-dardized regression coefficients, and explained vari-ance from each analysis appear in the Appendix asTables A1-A4.

(M

* * * is o<N lO (N 00 QQt-^ ^ - ^ (M

* * * * t ^\O CO CO CN • ^CO CO CN CO ^

Ig

rfCNJi-HtN^C^CO'

IIoI I

i2

.so

D

I I I I I I

'S.I I I I

2f2

I I I I

I I

n

oe

IQ

I I

* * * *

I I I I I I

oII I I I I I I I

to

T3

I•a

r l O c O C O ^ ( N C ^O C N J i — l O O O OI I I I I

I I

cgu3-a

01 jH

c: <u

1

let

CC

i H

c -2

I

o

ii V

Glasgow et al. 517

DysfunctionalAttributional

Style

ClassroomEngagement

(Time 2)

Figure 1 Standardized regression coefficients for the classroom engagement path model. All coefficients shown are statisticallysignificant at the p < .05 level.

Path Model for Classroom Engagement

The results of the path analysis linking parentingstyles, dysfunctional attributions, and classroom en-gagement are diagrammed in Figure 1. The first twosets of paths in the figure show the contemporaneousrelations among these variables. As predicted by thefirst hypothesis, adolescents who described their par-ents as authoritarian, indulgent, or neglectful weresignificantly more likely to report a higher propor-tion of dysfunctional attributions than were adoles-cents in the authoritative parenting group. Accordingto Figure 1, the path coefficient between neglectfulparenting and attributional style was the largest inthe equation (P = .24). Although these results aresupportive, this equation explained only a small pro-portion of the total variance in attributional tenden-cies (6%; see Appendix, Table Al),

The results of regressing the 1987-1988 index ofclassroom engagement on the respondents' demo-graphic characteristics and the parenting style indi-ces were consistent with the trends reported in theliterature. Relative to the comparison group, adoles-cents from nonauthoritative families were signifi-cantly less engaged in classroom activities during the1987-1988 academic year.

The third and fourth equations in the model esti-mate the predictive relations between parentingtypes, attributions, and 1988-1989 classroom engage-ment. Table Al (col, 3) shows the unconstrained rela-tion between attributional style and 1988-1989 en-

gagement, b = -.50, p < .001, As expected,controlling for the 1987-1988 outcome score in thefourth equation dramatically decreased the magni-tude of this relation, b = - .18, p < .05, This coefficientshows the effect of attributional style on changes inclassroom engagement over the 1 year period. As hy-pothesized, there was a significant negative relationbetween the proportion of dysfimctional attributionsreported in 1987-1988 and the level of classroom en-gagement in 1988-1989 (P = -.06),

Figure 1 shows that attributional style bridges par-enting styles and one form of achievement behavioramong adolescents. Each parenting style was indi-rectly related to reductions in classroom engagementthrough its positive association with dysfunctionalattributions. Yet, attributions did not fully explain theimpact of parenting style. Significant direct effects ofparenting remained once attributional style and theTime 1 outcome score were controlled. The combinedinfluences of these variables explained 34% of thevariance in 1988-1989 classroom engagement (seeTable Al),

Path Model for Homework

The results of the path analysis relating parentingstyles, dysfiinctional attributions, and the amount oftime spent on homework are presented in Figure 2.Despite slight variations in sample size, the patternof relations between the three parenting styles and

518 Child Development

Homework(Time 2)

Figure 2 Standardized regression coefficients for the homework path model. All coefficients shown are statistically significant atthe p < .05 level.

adolescents' attributions is identical to the previousmodel. Adolescents who perceived their parents asnonauthoritative were more likely than their peers toreport dysfunctional attributions for academic suc-cesses and failures. The neglectful mode of parentingcontinued to show the strongest relation to attribu-tional style (P = .22).

By contrast, parenting styles were only weakly as-sociated with homework in 1987-1988. According toFigure 2, the only significant path among this set ofpredictors linked neglectful parenting to homework(P = -.19). The absence of paths between the re-maining two parenting groups and homework indi-cates that adolescents from these groups did not dif-fer from their authoritatively reared peers on thisoutcome.

Consistent with the second hypothesis, a dysfunc-tional attributional style predicted a significant de-cline in the amottnt of time adolescents spent onhomework in 1988-1989 (P = -.07), controlling forhomework in 1987-1988. Once again, attributionalstyle indirectly linked the three parenting styles tochanges in homework, but it did not account for theserelations. Controlling for dysfunctional attributionsin the third equation did not eliminate the direct ef-fects of parenting (see Table A2, col. 3). The adverseeffects of nonauthoritative parenting on this form ofachievement behavior operated both directly and in-directly through attributional style. The final regres-

sion equation explained 30% of the total variation inhomework.

Path Model for Academic Achievement

Figure 3 presents the results of the path analysisjoining parenting styles, attributions, and high schoolgrades. The results of estimating the first regressionequation are comparable to those already described,with one exception. In this sample, adolescents whoviewed their parents as indulgent were no morelikely than adolescents from the authoritative parent-ing group to express dysfunctional attributions. Per-ceptions of authoritarian and neglectful parentingshowed the predicted relations to attributional style(P = .11 and p = .22, respectively). The combined in-fluence of social background and parenting ex-plained 6% of the variation in adolescents' attribu-tions.

The contemporaneous relations between parent-ing styles and 1987-1988 grade point average wereestimated by the second equation in the model. Fig-ure 3 shows that adolescents who characterized theirhome environments as indulgent or neglectfulearned significantly lower grades in 1987-1988 thanadolescents from authoritative homes. The path coef-ficient for the authoritarian parenting group was alsonegative, but only at a tendency, P = -.05, p < .10.

Glasgow et al. 519

.05

DysfunctionalAttributional

Style

Grade PointAverage(Time 2)

Figure 3 Standardized regression coefficients for the academic achievement path model. All coefficients shown are statisticallysignificant at the p < .05 level.

Table A3 (col. 3) shows that the unconstrained re-lation between attributional style and 1988-1989grade point average was significant, b = -.43, p <.001. However, this relation disappeared once the di-rect effect of earlier grades was taken into account,b = -.08, ns. The failure of dysfunctional attributionsto predict year-to-year differences in grades is notsurprising given the large correlation between theoutcome scores over time. Comparing the unstan-dardized regression coefficients from the third andfourth regression equations in Table A3, it is clearthat the 1987-1988 covariate substantially reducedmany of the effects in the model. With the Time 1outcome score controlled, attributional style was nota linking mechanism between parenting and laterschool performance in this path model. Even so, thefinal equation explained 55% of the total variation inhigh school grades.

Path Model for Educational Expectations

The final path analysis examined the hypothesizedrelations between parenting styles, dysfunctional at-tributions, and educational expectations. The resultsdiagrammed in Figure 4 are equivalent to the resultsfrom the last analysis. Adolescent descriptions of au-thoritarian and neglectful parenting were positivelyrelated to the tendency to report dysfunctional attri-

butions. Perceptions of indulgent parenting, how-ever, showed no relation to attributional style.

The paths in Figure 4 also indicate that the levelof schooling adolescents expected to attain during1987-1988 was associated with how they perceivedtheir parents. Adolescents in each of the nonauthori-tative parenting groups reported significantly lowereducational expectations than their peers. The great-est difference was observed between the neglectfuland authoritative parenting types (p = -.17).

This path analysis did not support the second hy-pothesis. Expressing a high proportion of dysfunc-tional attribution in 1987-1988 was not associatedwith a significant decline in educational expectationsover the 1 year period. This result is similar to thepattern observed in the academic achievementmodel. Although the unconstrained effect of attribu-tions was significant, b = -.56, p < .001, controllingfor the Time 1 outcome score overwhelmed the con-tribution of attributional style, b = -.06, ns. In thismodel too, attributions failed to bridge parentingstyles with adolescents' educational expectations.

In summary, the results of all four path analysespartially support the two central hypotheses. Adoles-cents who perceived their parents in nonauthorita-tive terms were more likely to report a dysfunctionalstyle of causal inference (except for the indulgent par-enting group in the models for academic achieve-ment and educational expectations). Dysfunctional

520 Child Development

DysfunctionalAttributional

Style

EducationalExpectations

(Time 2)

Figure 4 Standardized regression coefficients for the educational expectations path model. All coefficients shown are statisticallysignificant at the p < .05 level.

attributions were negatively related to each outcomein 1988-1989, but only when the equation did not in-clude the 1987-1988 covariate. Including the Time 1outcome score as a predictor in each model dramati-cally reduced, and in some cases eliminated, the im-pact of attributions. In the final analyses, there werenegative direct effects of dysfunctional attributionson subsequent levels of classroom engagement andhomework, but no impact on changes in grade pointaverage or educational expectations.

As the last step in the analyses, we estimated thepath models within each gender and ethnic group.The purpose of these analyses was to assess whetherfamily and attributional processes produced similaroutcomes in diverse social contexts. Only the rela-tions bearing upon the two central hypotheses are re-viewed. All of the results reported below were statis-tically significant at the .05 level or better. Amongfemales, all three parenting indices were significantlyassociated with dysfunctional attributions across thefour path models. Dysfunctional attributions, in turn,predicted significant reductions in classroom engage-ment, homework, and grades during 1988-1989, con-trolling for the 1987-1988 outcome scores.

The pattern of results for males was less support-ive than the pattern for females. Only neglectful par-enting was significantly related to dysfunctional at-tributions in each of the four path models. Although

the unconstrained effect of dysfunctional attributionson each 1988-1989 outcome was significant, control-ling for the 1987-1988 outcome scores reduced theserelations to nonsignificance. Dysfunctional attribu-tional style did not predict changes in any of the fouroutcomes over the 1 year period.

When path analyses were performed within eachethnic group, the number of cases dropped substan-tially. Because there were too few cases in each par-enting category to perform separate analyses onAfrican American and Hispanic American students,these two ethnic groups were pooled for the analysis.The number of students classified in the indulgentfamily type was as low as 15 in the combined ethnicgroup, as well as in the Asian ethnic group. The totalnumber of cases in each subgroup ranged from 216to 238 for Asian American students, from 832 to 1,042for non-Hispanic White students, and from 227 to 296for African and Hispanic American students com-bined.

Among non-Hispanic White students, perceptionsof neglectful and authoritarian parenting predictedsignificantly more dysfunctional attributions in threeof the four models. The path coefficient for authori-tarian parenting was not significant in the homeworkpath model. Perceptions of indulgent parenting, bycontrast, did not show a significant relation to attri-butional style in any model. Dysfunctional attribu-

Glasgow et al. 521

tions predicted significant reductions in subsequentlevels of classroom engagement and homework, con-trolling for the earlier levels of these outcomes.

Similarly, for Asian American students, descrip-tions of authoritarian and neglectful parenting weresignificantly associated with a higher proportion ofdysfunctional attributions in each path model. Theattributional style index failed to predict changes inany of the outcomes over the 1 year period.

Among African and Hispanic American students,only neglectful parenting showed the expected rela-tion to attributional style in all four path models. Theeffect of authoritarian parenting on attributions wassignificant only in the GFA path model. The index ofdysfunctional attributions, in turn, predicted signifi-cant declines in educational expectations, holdingconstant the Time 1 outcome score.

The results of the subgroup analyses reinforce theoverall pattern of findings observed within the entiresample. There were no significant relations oppositeto the directions predicted for any of the subgroups.We elaborate on these results in the discussion sec-tion of this article.

An Alternative Hypothesis

One alternative explanation of the main resultsfrom this study is that attributional style and the1988-1989 levels of classroom engagement andhomework are causally dependent on prior levels ofacademic performance. If this alternative is correct,then the observed relations between dysfimctionalattributions and these outcomes could be spurious.To examine this possibility, two additional path mod-els were estimated in which 1987-1988 grade pointaverage was included as an additional predictor of1987-1988 attributional style and of 1988-1989 class-room engagement and homework. Because earliergrades and attributional style were measured con-temporaneously, this is not an ideal test of the alter-native hypothesis. Nevertheless, it is sufficient for as-sessing whether the observed relations betweenattributions and these outcomes are the result of pos-sessing a common association with prior schoolachievement.

The results of these final path analyses did notsupport the alternative hypothesis. Earlier gradepoint average was negatively associated with attribu-tional style and positively associated with the two ed-ucational outcomes in 1988-1989 {p < .01 or better).Yet, independent of these relations, dysfunctional at-tributions predicted significant reductions in eachoutcome {p < .05), controlling for the relevant index

in 1987-1988. Therefore, although the endogenousvariables in each model were associated with aca-demic performance in 1987-1988, these associationsdid not eliminate the effect of attributional style.

DISCUSSION

A central premise of this study is that the nature of .parent-child interactions in the home affects the waysin which adolescents understand and manage eventsin their lives. This premise, combined with Weiner'sresearch on attribution, provided the basis for pos-iting links between parenting styles, adolescents' at-tributions, and educational outcomes.

Different constellations of parental behaviors andaffective expressions produce variations in adoles-cents' perceptions of their own performance capaci-ties, as well as in their self-reliance and willingnessto act responsibly. Among the four distinct parentingstyles, authoritative parenting is the most successfulin fostering personal and social responsibility in ado-lescents, without limiting their emerging autonomyand individuality. Given this empirical generaliza-tion, we hypothesized that adolescents' subjective ex-periences of nonauthoritative parenting styles alsowould be associated with a greater tendency to ex-press dysfunctional attributions in high school. Aspredicted, adolescents from these family environ-ments were less inclined to view their academicachievements as the products of their own capacitiesand persistence. Compared to their peers, these stu-dents more often reported a pattern of attributionthat emphasized either external causes for highschool grades or low ability as the cause of poorgrades.

The research literature on attributional processesprovided the basis for further hypothesizing that adysfimctional attributional style would lower subse-quent levels of achievement behavior and outcomes.The data used in this study lacked information ontwo key mediators in Weiner's theoretical model, af-fect and expectancy. Nevertheless, we were able toexamine those aspects of the model that linked attri-butions to achievement behaviors (classroom en-gagement and homework) and to outcomes (gradepoint average and educational expectations). The re-sults were partially consistent with the secondhypothesis. Dysfunctional attributions loweredstudents' willingness to engage in achievement-maximizing behaviors, but it did not affect their aca-demic achievement or educational expectations overtime. In view of these findings, we can speculate thatthe cumulative impact of dysfunctional attributions

522 Child Development

on achievement behavior might be much greater overa longer time span.

Attributional style was not a primary mediator ofthe relations between parenting styles and adoles-cents' educational outcomes. Controlling for dys-functional attributions in the path models did noteliminate the direct effects of parenting style. Bothparenting and attributions made independent contri-butions to the prediction of adolescents' achievementbehaviors. This general lack of mediation does notsuggest, however, that attributions played no role inthese relations. At a minimum, attributional style isone of many potential processes by which parents in-directly influence their children.

To assess whether the process operated similarlyacross diverse social and cultural contexts, we per-formed analyses separately within gender and ethnicsubgroups. The expected relations between nonau-thoritative parenting and dysfunctional attributionswere often found. The clearest pattern to emergefrom these analyses was the consistent relation be-tween perceptions of neglectful parenting and ado-lescents' attributional style. Adolescents from eachsubgroup were doubly disadvantaged by the absenceof both parental warmth and behavioral standards.Perhaps just as noteworthy was the lack of associa-tion between indulgent parenting and attributionalstyle. This parenting style failed to consistently pre-dict higher levels of dysfunctional attributionsamong males and among three ethnic groups. Thesefindings highlight the importance of distinguishingamong nondemanding parents who vary in theirlevel of responsiveness. They also complement theresults of previous studies, which have found thatadolescents from neglectful families show the lowestlevels of cognitive and psychosocial functioning(Baumrind, 1991; Lamborn et al,, 1991; Steinberg etal,, 1994),

Only a few significant links were observed be-tween dysfunctional attributions and the educationaloutcomes within the subgroups. The absence of theserelations among males, non-Hispanic White stu-dents, and among the combined group of Africanand Hispanic American students was the result ofcontrolling for the Time 1 outcome scores. All of theunconstrained effects of attributional style were sig-nificant within these subgroups. The failure of attri-butions to predict outcomes among the Asian-American students, with or without the 1987-1988covariate, requires further investigation. It appearsthat unmeasured aspects of the social context forAsian American students prevented dysfunctionalattributions from producing the expected declines inachievement behavior and outcomes. Most of our

knowledge of adolescent and family processes comesfrom studies of middle-class Whites, The failure ofgeneral hypotheses to apply within a particular sub-group provides an impetus for conducting further re-search that is specifically designed to explicate groupdifferences (see Chao, 1994).

The low explained variance in the four main pathmodels, as well as in the subgroup analyses, suggeststhat the combined effects of background characteris-tics, parenting styles, and attributions explain only asmall portion of a more complex story. Other socialstructural factors may also shape adolescents' ten-dencies to make particular kinds of attributions acrosssituations. For example, adolescents who are placedin different curriculum tracks may evidence differentstyles of causal inference. Curriculum tracking pro-duces inequalities in educational experiences and op-portunities, which could affect the ways in which stu-dents interpret the reasons for their academicsuccesses and failures.

The limitations of this study point to directions forfurther research. Path analyses that include measure-ments at more time points are needed to help un-scramble the causal relations among variables. As isoften the case, the availability of longitudinal datawould permit more complete models of explicit andimplicit causal mechanisms. Parenting styles, for ex-ample, were theorized to be interpersonal determi-nants of adolescents' attributional tendencies. Be-cause of the data's panel structure, these relationscould be examined only cross-sectionally. Further-more, a plausible alternative model, based moreclosely on Weiner's work, could not be examinedwith these data because of the temporal restrictions.The alternative formulation predicts that dysfunc-tional attributions, through affect and expectancy,would lower subsequent levels of classroom engage-ment and homework, and these achievement behav-iors, in turn, would lower later high school grades,

A natural extension of the present research wouldbe to examine attributional tendencies within specificacademic subjects, assessing the extent to which adysfunctional attributional style operates similarlyacross different subjects. For example, the conse-quences of making dysfunctional attributions for tasksuccesses and failures in math or science may bemore severe than the consequences of making similarattributions in English or social studies. In addition,members of particular social groups may be morelikely to use a dysfunctional attributional style in onesubject area rather than in another. Work on genderdifferences in academic achievement, for example,suggests that females express more dysfunctional at-tributions for their performance in math and science

Glasgow et al. 523

than do males (Hansen & O'Leary 1985; Oakes, 1990).Studies of such issues would further our understand-ing of the links between adolescents' attributionalstyle and their achievement-related outcomes.

Studies that focus on content-specific parentingpractices may further clarify how parent-child inter-actions shape attributional tendencies. The influenceof parenting style on adolescent development is pri-marily indirect, through parenting practices andthrough the adolescent's openness to socialization(Darling & Steinberg, 1993). Parenting behaviors thatare domain-specific (e.g., educationally oriented)may shape adolescents' attributional styles morepowerfully than a global parenting style. In addition,general styles of parenting can moderate the impactof specific parenting practices (Darling & Steinberg,1993). The effects of parenting practices on adoles-cents' attributional tendencies may be altered whenconsidered in relation to the more general parentingenvironment in which the practices are embedded.

A goal of this research was to derive additionalimplications from the conceptual framework devel-oped by Baumrind (1971) and Maccoby and Martin(1983). In keeping with this perspective, we bor-rowed a technique from related studies (Lamborn etal, 1991; Steinberg et al., 1994) to classify adolescentsinto qualitatively different family environments. Thefour parenting styles created by this technique wereideal types based on sample characteristics. A largepercentage of the adolescents in this study could not

be classified into one of the parenting groups, indicat-ing that a substantial number of adolescents were notsubject to these very distinct types of parenting. Tomaximize the external validity of this research, futurestudies should examine how mixed or inconsistentstyles of parenting affect adolescents' cognitive pro-cesses and educational outcomes.

Finally, it is impressive that parenting styles wereassociated with adolescents' attributional tendenciesat this later period of development, despite an estab-lished history of school performance. These findingssuggest that the influence of parenting styles on attri-butional processes may be greater during earlierstages of child development.

ACKNOWLEDGMENTS

The research reported here was supported by theSpencer Foundation, the Drown Foundation, and Jilland Boyd Smith.

ADDRESSES AND AFFILIATIONS

Corresponding author: Kristan L. Glasgow, Depart-ment of Sociology, Building 120, Stanford University,Stanford, CA 94305-2047. Sanford M. Dornbusch is atStanford University; Lisa Troyer is at the Universityof Iowa; Laurence Steinberg is at Temple University;and Philip L. Ritter is at Stanford University.

524 Child Development

APPENDIX

Table Al Unstandardized Regression Coefficients for Equations in the ClassroomEngagement Path Model (AT = 1,576)

Independent Variables

Female

Parental education

Intact family

Years of age

African American

Asian American

Hispanic American

Authoritarian

Indulgent

Neglectful

1987-1988 DAS

1987-1988 engagement

ConstantAdjusted R̂

1987-1988DAS

(1)

,02+(,01)

-,01»»*(,00)

- ,01(,01)

- ,01(,01)

-.07***(,02)

-.04*(.02)

-.03+(,02).05**'

(.02).05*

(.02),11*'*

(.01)

.60***

.06

1987-1988Engagement

(2)

.04(.03).01

(,01)- .00(.03).01

(.02).05

(.06).21***

(,04).16**

(.05)-.29»**(.04)

-.30***(.05)

-.58***(.04)

3.76*»*.16

1988-1989Engagement

(3)

.11**(.04)

- .00(.01)

- .00(,04)

-.04»(.02).12+

(.07).10+

(.05)- .01(,06)

-.26***(.05)

- .05(.06)

-.42***(,04)

-.50***(.08)

4.62***.12

1988-1989Engagement

(4)

.09**(,03)

- ,00(.01)

- .00(.03)

-.04**(,02)

. 1 1 +

(.06)- .02(.04)

• - . 1 0 *

(.05)-.10*(,04).12*

(.05)-.10*(,04)

-.18*(.07).61***

(.03)2.12***

.34

Note: Dysfunctional Attributional Style (DAS). Standard errors are in parentheses. Sig-nificance tests are two-tailed.+ p < .10; *p < .05; **p < .01; ***p < .001.

Glasgow et al. 525

Table A2 Unstandardized Regression Coefficients for Equations in the HomeworkPath Model (N = 1,275)

Independent Variables

Female

Parental education

Intact family

Years of age

African American

Asian American

Hispanic American

Authoritarian

Indulgent

Neglectful

1987-1988 DAS

1987-1988 homework

ConstantAdjusted R^

1987-1988DAS

(1)

.01(.01)

- . 01*(.00)

- .01(.01)

- .01(.01)

-.07**(.02)

-.04**(.02)

- .03(.02).05**

(.02).05*

(.02).11***

(•02)

.60***

.05

1987-1988Homework

(2)

.23**(•08).09***

(.01).09

(.08).17***

(.04)-.54***(.14).38***

(.10)- .16(.13)

- .18^(.10)

- .20(.13)

-.56***(.09)

-2.06**.12

1988-1989Homework

(3)

.40***(.07).08***

(.01).16*

(.08).11**

(.04)-.57***(.13).26**

(.10)-.23+(.13)

-.33**(.10)

-.26*(.13)

-.52***(.09)

-.76***(.17)

- .96.15

1988-1989Homework

(4)

.30***(.07).04***

(.01).13+

(.07).04

(•04)-.32**(.12).12

(.09)- .16(.12)

-.27**(.09)

-.20+(.12)

-.33***(.08)

-.42**(.15).41***

(.02)- .32

.30

Note: Dysfunctional Attributional Style (DAS). Standard errors are in parentheses. Sig-nificance tests are two-tailed.+ p < .10; * p < .05; ** p < .01; *** p < .001.

526 Child Development

Table A3 Unstandardized Regression Coefficients for Equations in the AcademicAchievement Path Model (N = 1,371)

Independent Variables

Female

Parental education

Intact family

Years of age

African American

Asian American

Hispanic American

Authoritarian

Indulgent

Neglectful

1987-1988 DAS

1987-1988 GPA

ConstantAdjusted R^

1987-1988DAS

(1)

.01(.01)

- . 0 1 *. (-00)- .01(.01)

- .01(.01)

-.08***(.02)

-.04**(.02)

- .02(.02).06***

(.02).03

(.02).10***

(.01)

.57***

.06

1987-1988GPA

(2)

.12**(.04).07***

(.01).15***

(.04)- .03(.02)

-.30***(.07).34***

(.05)-.20**(.07)

-.10+(.06)

-.24***(.07)

-.33***(.05)

2.41***.19

1988-1989GPA

(3)

.08*(.04).05***

(.01).20***

(•04)-.04+(.02)

-.33***(.07).29***

(.05)-.20**(.07) ,

-.10+(.05)

-.19**(.07) .

-.27***(.05)

-.43***(.09)

2.88***.20

1988-1989GPA

(4)

.00(.03).01**

(.01).11***

, (.03)-.02(.02)

- . 11*(.05).09*

(.04)- .07(.05)

- .05(.04)

- .05(.05)

-.10**(.04)

- .08(.07).64***

(.02)1.12***

.55

Note: Dysfunctional Attributional Style (DAS). Standard errors are in parentheses. Sig-nificance tests are two-tailed.+ p < .10; *p < .05; ** p < .01; ***p < .001.

Glasgow el al. 527

Table A4 Unstandardized Regression Coefficients for Equations in the EducationalExpectations Path Model (N = 1,444)

Independent Variables

Female

Parental education

Intact family

Years of age

African American

Asian American

Hispanic American

Authoritarian

Indulgent

Neglectful

1987-1988 DAS

1987-1988 expectations

ConstantAdjusted R^

1987-1988DAS

(1)

.02(.01)

-.01**(.00)

- .02(.01)

- .01(.01)

-.08***(.02)

-.04*(.02)

- .02(.02).06***

(•02).03

(.02).11***

(.01)

.59***

.06

1987-1988Expectations

(2)

.05(.06).12***

(.01).14*

(.06)-.06*(.03).07

(.10).30***

(08)-.30**(.10)

-.33***(.08)

-.24*(.10)

-.43***(.07)

4.13***.18

1988-1989Expectations

(3)

.25***(.06).10***

(.01).09

(.06)-.07*(.03).02

(.10).27***

(.08)-.28**(.10)

-.24**(.08)

- .10(.10)

-.35***(.07)

-.56***(.13)

4.72***.18

1988-1989Expectations

(4)

.21***(.05).04***

(.01).03

(.05)- .03(.02).02

(.09).12+

(.06)- .10(.08)

- .09(.06).02

(.08)-.16**(.06)

- .06(.11).55***

(.02)2.15***

.43

Note: Dysfunctional Attributional Style (DAS). Standard errors are in parentheses. Sig-nificance tests are two-tailed.* p < .10; * p < .05; ** p < .01; *** p < .001.

REFERENCES

Arkin, R. M., Detchon, C. S., & Maruyama, G. M. (1981).Causal attributions of high and low achievement moti-vation college students for performance on examina-tions. Motivation and Emotion, 5, 139-152.

Arkin, R. M., & Maruyama, G. M. (1979). Attributions, af-fect, and college exam performance. Journal of Educa-tional Psychology, 71, 85-93.

Astone, N. M., & McLanahan, S. S. (1991). Family structure,parental practices, and high school completion. Ameri-can Sociological Review, 56, 309-320.

Baker, D., & Stevenson, D. L. (1986). Mother's strategies ofchildren's school achievement: Managing the transitionto high school. Sociology of Education, 59, 156-166.

Bandura, A. (1977). Self-efficacy: Toward a unifying theoryof behavioral change. Psychological Review, 84, 191-215.

Bar-Tal, D. (1978). Attributional analysis of achievement-related behavior. Review of Educational Research, 48,259-271.

Baumrind, D. (1971). Current patterns of parental author-ity. Developmental Psychology Monograph, 4, 1-103.

Baumrind, D. (1978). Parental disciplinary patterns and so-

cial competence in children. Youth and Society, 9, 239-276.

Baumrind, D. (1989). Rearing competent children. InW. Damon (Ed.), Child development today and tomorrow(pp. 349-378). San Francisco: Jossey-Bass.

Baumrind, D. (1991). The influence of parenting style onadolescent competence and substance use. Journal ofEarly Adolescence, 11, 56-95.

Bernstein, W. M., Stephan, W. G., & Davis, M. H. (1979).Explaining attributions for achievement: A path analyticapproach. Journal of Personality and Social Psychology, 37,1810-1821.

Chao, R. K. (1994). Beyond parental control and authoritar-ian parenting style: Understanding Chinese parentingthrough the cultural notion of training. Child Develop-ment, 65, 1111-1119.

Collins, W. A. (1990). Parent-child relationships in the tran-sition to adolescence: Continuity and change in interac-tion, affect, and cognition. In R. Montemayor, G. R. Ad-ams, & T. P. Gullotta (Eds.), Advances in adolescentdevelopment: Vol. 2. Prom childhood to adolescence: A transi-tional period? (pp. 85-106). Newbury Park, CA: Sage.

Cooper, C. R., Grotevant, H. D., & Condon, S. M. (1983).

528 Child Development

Individuality and connectedness in the family as a con-text for adolescent identity formation and role-takingskill. In H. D. Grotevant & C. R. Cooper (Eds.), New di-rections for child development: Vol. 22. Adolescent develop-ment in the family (pp. 43-59).San Francisco: Jossey-Bass.

Covington, M. V., & Omelich, C. L. (1979). Are causal attri-butions causal? A path analysis of the cognitive modelof achievement motivation. Journal of Personality and So-cial Psychology, 37, 1487-1504.

Covington, M. V., & Omelich, C. L. (1984a). The troublewith pitfalls: A reply to Weiner's critique of attributionresearch. Journal of^Educational Psychology, 76,1199-1213.

Covington, M. V., & Omelich, C. L. (1984b). An empiricalexamination of Weiner's critique of attribution research.Journal of Educational Psychology, 76, 1214-1225.

Crittenden, K. S. (1983). Sociological aspects of attribution.Annual Review of Sociology, 9, 425-446.

Darling, N., & Steinberg, L. (1993). Parenting style as con-text: An integrative model. Child Development, 113, 487-496.

Dombusch, S. M., Ritter, P., Leiderman, P., Mont-Reynaud,R., & Chen, Z. (1990). Family decision making and aca-demic performance in a diverse high school population.Journal of Adolescent Research, 5, 143-160.

Dombusch, S. M., Ritter, P., Leiderman, P., Roberts, D., &Fraleigh, M. (1987). The relation of parenting style to ad-olescent school performance. Child Development, 58,1244-1257.

Dweck, C. S. (1975). The role of expectations and attribu-tions in the alleviation of learned helplessness. Journalof Personality and Social Psychology, 31, 674-685.

Dweck, C. S., & Bush, E. S. (1976). Sex differences in learnedhelplessness: I. Differential debilitation with peer andadult evaluators. Developmental Psychology, 12,147-156.

Dweck, C. S., Davidson, W., Nelson, S., & Enna, B. (1978).Sex differences in learned helplessness: II. The contin-gencies of evaluative feedback in the classroom. III. Anexperimental analysis. Developmental Psychology, 14,268-276.

Eccles-Parsons, J., Adler, T. F., & Kaczala, C. M. (1982). So-cialization of achievement attitudes and beliefs: Parentalinfluences. Child Development, 53, 310-321.

Fincham, F. D., Hokoda, A., & Sanders, S., Jr. (1989).Learned helplessness, test anxiety, and academicachievement: A longitudinal analysis. Child Develop-ment, 60, 138-145.

Forsyth, D. R., & McMillan, J. H. (1981). Attributions, affect,and expectations: A test of Weiner's three dimensionalmodel. Journal of Educational Psychology, 73, 393-403.

Griffin, B. Q., Combs, A. L., Land, M. L., & Combs, N. N.(1983). Attributions of success and failure in college per-formance. Journal of Psychology, 114, 259-266.

Hansen, R. D., & O'Leary, V. L. (1985). Sex-determined at-tributions. In V. E. O'Leary, R. K. Unger, & B. S. Wall-ston (Eds.), Women, gender, and social psychology (Vol. 31,pp. 67-99). Hillsdale, NJ: Erlbaum.

Heider, F. (1958). The psychology of interpersonal relations.New York: Wiley.

Hein, C, & Lewko, J. H. (1994). Gender differences in fac-

tors related to parenting style: A study of high per-forming science students. Journal of Adolescent Research,9, 262-281.

Hess, R. D., & HoUoway, 5. D. (1984). Family and schoolas educational institutions. In R. D. Parke (Ed.), Review ofchild development research (Vol. 7, pp. 179-222). Chicago:University of Chicago Press.

Hout, M. (1988). The American occupational structure inthe 1980s. American Journal of Sociology, 93, 1358-1400.

Keith, T. Z. (1982). Time spent on homework and highschool grades: A large-sample path analysis. Journal ofEducational Psychology, 74, 248-253.

Kelly, H. H. (1967). Attribution theory in social psychology.In D. Levine (Ed.), Nebraska Symposium on Motivation(Vol. 15, pp. 192-238). Lincoln: University of NebraskaPress.

Lamborn, S., Mounts, N., Steinberg, L., & Dombusch, S. M.(1991). Patterns of competence and adjustment from au-thoritative, authoritarian, indulgent, and neglectful fam-ilies. Child Development, 62, 1049-1065.

Lepper, M. R., & Greene, D. (1978). The hidden costs of re-wards: New perspectives on the psychology of motivation.Hillsdale, NJ: Erlbaum.

Maccoby, E., & Martin, J. (1983). Socialization in the contextof the family: Parent-child interaction. In E. M. Hether-ington (Ed.), P. H. Mussen (Series Ed.), Handbook of childpsychology: Vol. 4. Socialization, personality, and social de-velopment (pp. 1-101). New York: Wiley.

Maehr, M. L., & Nicholls, J. G. (1980). Culture and achieve-ment motivation: A second look. In N. Warren (Ed.),Studies in cross-cultural psychology (Vol. 3, pp. 221-267).New York: Academic Press.

Metalsky, G. 1., & Abramson, L. Y. (1981). Attributionalstyles: Toward a framework for conceptualization andassessment. In P. C. Kendall & S. D. HoUon (Eds.), As-sessment strategies for cognitive-behavioral interventions(pp. 13-58). New York: Academic Press.

Natriello, G., & McDill, E. D. (1986). Performance stan-dards, student effort on homework and academicachievement. Sociology of Education, 59, 18-31.

Nolen-Hoeksema, S., Seligman, M. E. P., & Girgus, J. S.(1986). Learned helplessness in children: A longitudinalstudy of depression, achievement, and explanatorystyle. Journal of Personality and Social Psychology, 51, 435-442.

Oakes, J. (1990). Opportunities, achievement, and choice:Women and minority students in science and mathe-matics. Review of Research in Education, 16, 153-222.

Paulson, S. E. (1994). Relations of parenting style and pa-rental involvement with ninth-grade students' achieve-ment. Journal of Early Adolescence, 14, 250-267.

Platt, C. (1988). Effects of attributions for success on first-term college performance: A covariance structuremodel. Journal of Educational Psychology, 80, 569-578.

Sewell, W. H., & Hauser, R. M. (1980). The Wisconsin longi-tudinal study of social and psychological factors in aspi-rations and achievement. In A. Kerckhoff (Ed.), Researchin the sociology of education and socialization (Vol. 1, pp.59-100). Greenwich, CT: JAI Press.

Glasgow et al. 529

Smetana, J. G. (1988). Concepts of self and social conven-tions: Adolescents' and parents' reasoning about hypo-thetical and actual family conflicts. In M. R. Gunnar &W. A. Collins (Eds.), Minnesota symposia on child psychol-ogy: Vol. 21. Developing during the transition to adolescence(pp. 79-122). Hillsdale, NJ: Eribaum.

Steinberg, L., Elmen, J., & Mounts, N. (1989). Authoritativeparenting, psychosocial maturity, and academic successamong adolescents. Child Development, 60, 1424-1436.

Steinberg, L., Lambom, S. D., Darling, N., Mounts, N. S., &Dornbusch, S. M. (1994). Over-time changes in ad-justment and competence among adolescents from au-thoritative, authoritarian, indulgent, and neglectful fam-ilies. Child Development, 65, 754-770.

Steinberg, L., Lambom, S. D., Dornbusch, S. M., & Darling,N. (1992). Impact of parenting practices on adolescentachievement: Authoritative parenting, school involve-ment, and encouragement to succeed. Child Develop-ment, 63, 1266-1281.

Stevenson, D. L., & D. P. Baker. (1987). The family-schoolrelation and the child's school performance. Child Devel-opment, 58, 1348-1357.

Stryker, S., & Gottlieb, A. (1981). Attribution theory andsymbolic interactionism: A comparison. In J. H. Ickes,W. J. Kidd, & R. E. Kidd (Eds.), New directions in attribu-tion research (Vol. 2, pp. 425-453). Hillsdale, NJ: Eri-baum.

Wagner, D. A., Spratt, J. E., Gal, I., & Paris, S. G. (1989).Reading and believing: Beliefs, attributions, and reading

achievement in Moroccan schoolchildren. Journal of Edu-cational Psychology, 81, 283-293.

Weiner, B. (1979). A theory of motivation for some class-room experiences. Journal of Educational Psychology, 71,3-25.

Weiner, B. (1980). Human motivation. New York: Holt, Rein-hart & Winston.

Weiner, B. (1983). Some methodological pitfalls of attribu-tion research. Journal of Educational Psychology, 74, 530-543.

Weiner, B. (1985). An attributional theory of achievementmotivation and emotion. Psychological Review, 92, 548-573.

Willig, A. C, Harnisch, D. L., Hill, K. T., & Maehr, M. L.(1983). Sociocultural and educational correlates ofsuccess-failure attributions and evaluation anxiety inthe school setting for Black, Hispanic, and Anglo Chil-dren. American Educational Research Journal, 20, 385-410.

Wolf, E. M., & Savickas, M. L. (1985). Time perspective andcausal attributions for achievement. Journal of Educa-tional Psychology, 77, 471-480.

Yee, D. K., & Eccles-Parsons, J. S. (1988). Parent perceptionsand attributions for children's math achievement. SexRoles: A Journal of Research, 19, 317-333.

Yee, D. K., & Flanagan, C. (1985). Eamily environments andself-consciousness in early adolescence. Journal of EarlyAdolescence, 5, 59-68.