Community by Design

25
Community by Design* Karen A. Franck, New Jersey Institute of Technology Findings and speculations about the effects of physical design on social behavior in planned residential settings have accumulated without the benefit of any explicit theoretical models (Newman, 1972, 1973; Cooper, Day, and Levine, 1972; Cooper, 1975). Recent research about urban neighborhoods has included the formulation and estimation of models of community attachment, but these models do not incorporate any physical design features (Kasarda and Janowitz, 1974; Hunter, 1975). I attempt here to connect these two areas of research by developing and estimating a model of community attachment that includes the effects of physical design. The sites for the research are federally assisted, moderate income housing developments. Both physical and social features of the sites prove to have significant effects on the level of community attachment in these settings. Effects of physical design on social behavior in residential settings were first observed in studies of spatial proximity and friendship formation (Merton, 1948; Festinger, Schacter, and Back, 1950; Caplow and Forman, 1950; Kuper, 1953). Following this research, sociologists expressed serious doubt that design could have any influence on social behavior except in special circumstances (Gans, 1961, 1968; Keller, 1966). Nonetheless, interest in the physical design of communities was rekindled by the observations of the Pruitt-Igoe housing project in St. Louis (Rainwater, 1970; Yancey, 1971; Newman, 1972). Despite the continuing accumulation of findings and speculations since then (Cooper, 1975; Zeisel and Griffin, 1975; Becker, 1977; Choldin, 1978; Newman, 1980), no explicit theoretical models have been developed to explain the expected effects of physical design. During this same period an alternative was proposed to the “loss of community” argument propounded by Wirth (1938), Nisbet (1953), and Stein (1960). This alternative defines local community as multidimensional, comprising a variety of functions, activities, and sentiments (Sutties, 1968, 1972, 1975; Hunter, 1974, 1975, 1978; Kasarda and Janowitz, 1974; Fischer et al., 1977). Rather than asking whether a few presumably essential ingredients of local community are present, the authors explore what pattern of elements is present and how this pattern may differ across locations, time, and people. Theoretical models developed to examine these patterns have focused on community attachment in urban neighborhoods but have not included physical design features (Kasarda and Janowitz, 1974; Hunter, 1975). Indeed, these two fields, focusing primarily on physical design or local community, have remained largely separate. The objective of the present research is to connect them: to incorporate some of the ideas of Newman (1972, 1973) and Cooper (1970, 1975; Cooper, Day, and Levine, 1972) about the effects of physical design in planned residential settings into a

Transcript of Community by Design

Community by Design*

Karen A. Franck, New Jersey Institute of Technology

Findings and speculations about the effects of physical design on social behavior in planned residential settings have accumulated without the benefit of any explicit theoretical models (Newman, 1972, 1973; Cooper, Day, and Levine, 1972; Cooper, 1975). Recent research about urban neighborhoods has included the formulation and estimation of models of community attachment, but these models do not incorporate any physical design features (Kasarda and Janowitz, 1974; Hunter, 1975). I attempt here to connect these two areas of research by developing and estimating a model of community attachment that includes the effects of physical design. The sites for the research are federally assisted, moderate income housing developments. Both physical and social features of the sites prove to have significant effects on the level of community attachment in these settings.

Effects of physical design on social behavior in residential settings were first observed in studies of spatial proximity and friendship formation (Merton, 1948; Festinger, Schacter, and Back, 1950; Caplow and Forman, 1950; Kuper, 1953). Following this research, sociologists expressed serious doubt that design could have any influence on social behavior except in special circumstances (Gans, 1961, 1968; Keller, 1966). Nonetheless, interest in the physical design of communities was rekindled by the observations of the Pruitt-Igoe housing project in St. Louis (Rainwater, 1970; Yancey, 1971; Newman, 1972). Despite the continuing accumulation of findings and speculations since then (Cooper, 1975; Zeisel and Griffin, 1975; Becker, 1977; Choldin, 1978; Newman, 1980), no explicit theoretical models have been developed to explain the expected effects of physical design.

During this same period an alternative was proposed to the “loss of community” argument propounded by Wirth (1938), Nisbet (1953), and Stein (1960). This alternative defines local community as multidimensional, comprising a variety of functions, activities, and sentiments (Sutties, 1968, 1972, 1975; Hunter, 1974, 1975, 1978; Kasarda and Janowitz, 1974; Fischer et al., 1977). Rather than asking whether a few presumably essential ingredients of local community are present, the authors explore what pattern of elements is present and how this pattern may differ across locations, time, and people. Theoretical models developed to examine these patterns have focused on community attachment in urban neighborhoods but have not included physical design features (Kasarda and Janowitz, 1974; Hunter, 1975).

Indeed, these two fields, focusing primarily on physical design or local community, have remained largely separate. The objective of the present research is to connect them: to incorporate some of the ideas of Newman (1972, 1973) and Cooper (1970, 1975; Cooper, Day, and Levine, 1972) about the effects of physical design in planned residential settings into a

290 KAREN A. FRANCK

causal model of community attachment that is adapted from the research of Hunter (1975) and Kasarda and Janowitz (1974) on urban neighbor- hoods.

T?w Causal Model

“Community by design” refers to two assumptions: (1) planned residential settings are, to some extent, communities by definition and (2) physical design, in combination with other site characteristics, affects community conditions, activities, and sentiments. The sites for this research are housing developments built and managed with federal subsidies. In line with the first assumption, a housing development possesses some of the attributes of a local community before it is even occupied. Its architectural uniformity and clear boundaries fulfill one of the traditional criteria for the existence of local community: a territorial unit delimited by observable boundaries (Hillery, 1955; Keller, 1973). It has a predeter- mined name, which fulfills another requirement for community (Ross, 1974; Hunter, 1974, 1975). A housing development also possesses an image based in part on the sponsor’s objective in building it and, in subsidized housing, on the federal programs under which it is financed. For these reasons housing developments have been called “planned com- munities” (Merton, 1948), “artificial neighborhoods” (Suttles, 1972), and “minimal named areas” (Suttles, 1975).

Even though these features are common to all housing developments, the intensity of community activities and sentiments in these settings varies considerably. The underlying assumption of the causal model developed to explain this variation is that certain physical, social, and organizational characteristics of planned residential environments facilitate or discourage various community conditions and activities that, in turn, either facilitate or discourage certain perceptions and sentiments, including community attachment (see Figure l ) . ’

The research of Kasarda and Janowitz (1974) and Hunter (1975) suggested the basic causal sequence of variables in the model. Relatively enduring characteristics of the developments are the independent variables, community activities are intervening variables, and community attachment is the final outcome variable. The independent variables are six attributes of housing developments that, except for age of site, are largely determined by the federal housing program under which a development is built and financed, by the objectives of the sponsor, and by management policies. In this sense they are subject to planned change: they can be controlled and sometimes changed by policy decisions.

The existence of a tenants’ association is not determined by policy but is dependent instead on residents’ initiative. Since a tenants’ association is usually formed soon after a development is occupied, it is placed

FIGU

RE 1

O

utlin

e of

Mod

el o

f C

omm

unit

y A

ttac

hmen

t

asso

ciat

ion

I Tenan

ts’

I C

omm

univ

Act

iviti

es

Com

mun

ity C

ondi

tions

Vic

timiz

atio

n ra

te

Tur

nove

r ra

te

292 KAREN A. FRANCK

immediately after the site characteristics and before community conditions in the causal model. The placement of community conditions before activities and sentiments is determined by one of the objectives of this research, to examine how such conditions affect community activities and sentiments. Perceived problems are placed between activities and sentiments because activities, in the context of this model, are viewed as causally prior to attitudes or feelings. This view is consistent with Hunter’s (1975) model and makes logical sense‘ in that patterns of activities are likely to be more stable over time than perceptions or sentiments, which may change abruptly in reaction to a particular event.

The choice of the physical design feature (number of apartments in a group) is based on the work of Newman (1972, 1973) and Cooper (1970; Cooper, Day, and Levine, 1972) who suggest that the number of households that share the space immediately adjacent to their apartment doors is likely to affect a wide range of community conditions, activities, and perceptions. The greater the number of households sharing the space, the worse the consequences are expected to be. For this study, the number of apartments in a group in walk-up and high-rise buildings is the number of apartments that share a corridor or stairway landing. In row houses the area adjacent to the individual apartment entry is considered to be the private domain of that family. Since this area is shared with no other families, the value for the variable number of apartments in a group takes the value of zero in row house sites.

According to both Newman (1972, 1973) and Cooper (1970, 1975; Cooper, Day, and Levine, 1972), the greater the number of apartments in a group, the less responsibility residents will take for the care, use, and safety of the spaces outside their apartments and the less socially involved they will be with one another. Apartment group size is therefore likely to have positive effects on turnover and victimization rates and negative effects on acquaintance, friendship-kinship, sense of safety, and perceived quality of maintenance. Through these negative effects on the intervening variables in the model, apartment group size will exert a negative indirect effect on community attachment. In addition, this design feature will exert a direct negative effect on attachment because the larger the group, the less commitment residents will feel (Newman, 1972; Cooper, 1975).

The number of individual residential units sharing a common space adjacent to the units has also been featured as an important variable in recent research on the effects of crowding in residential environments (Baum, Harpin, and Valins, 1975; Baum and Valins, 1977; McCarthy, 1978). The expected negative effects of apartment group size on acquain- tance, friendship-kinship bonds, and attachment are consistent with this body of work.

The proportion of one-parent welfare families is included in the model as the leading social characteristic of sites because Newman’s findings

COMMUNITY BY DESIGN 293

(1972, 1973) indicate that this population characteristic is an important predictor of crime rate and of other problems such as high turnover and fear of crime. The explanation for this is that the higher the proportion of families of this type, the less responsibility residents are able to take for the care and safety of their environment.

One would also expect percent one-parent welfare families to have a negative effect on attachment given that residents in lower social classes are less content with their neighborhoods than residents in higher social classes (Hunter, 1974; Fischer et al., 1977). At the same time, however, the proportion of one-parent welfare families may have a positive effect on attachment through friendship-kinship bonds. Members of lower social classes tend to have a greater proportion of their friends and relatives living in their neighborhoods (Kasarda and Janowitz, 1974) and people with friends and relatives living nearby show stronger feelings of attachment (Kasarda and Janowitz, 1974; Fischer et al., 1977).

The proportion of families with minors and the proportion of heads of household aged 20 to 35 are included in the model to explore how these features affect community attachment. The findings on such effects for the individual family in urban neighborhoods are mixed (Hunter, 1974, 1975; Kasarda and Janowitz, 1974; Fischer et al., 1977). As formal social organizations, cooperative ownership and tenants’ association are likely to increase the level of acquaintance, friendship-kinship, and attachment (Kasarda and Janowitz, 1974; Hunter, 1975). Meetings of a cooperative or a tenants’ association provide opportunities for residents to meet each other, to try to solve local problems, and to build up feelings of commitment.

Primary ties of friendship and kinship were traditionally treated as essential ingredients of local community. The presumed weakening of these ties in urban settings was used as evidence for the disappearance of local community (Bell, 1968). More recently, both primary ties and more informal social relationships between residents have been considered (Keller, 1973; Kasarda and Janowitz, 1974; Hunter, 1974, 1975; Fischer et al., 1977). Newman (1972, 1973), in fact, places greater importance on casual interaction and mutual recognition among residents in housing develop- ments than on more intimate relationships. Research findings indicate that both forms of social interaction enhance feelings of community attachment, although the effect of primary ties does appear stronger than the effect of casual acquaintance at least in urban neighborhoods (Kasarda and Janowitz, 1974; Fischer et al., 1977). Friendship-kinship and acquaintance are included in the present model as separate types of community activity to compare their respective effects on attachment in housing developments. Each is likely to have a positive effect on attachment and sense of safety.

Friendship-kinship and acquaintance are both expressive forms of community activity. Since local community fulfills expressive and instru- mental functions (Suttles, 1972), it seemed important to include some form

294 KAREN A. FRANCK

of instrumental community activity. To that end, an additional intervening variable-previous experience in trying to solve a common problem-was added to the model. This variable is expected to link tenants’ association to community attachment because the existence of an association will facilitate attempts to resolve problems and such efforts are likely to enhance feelings of commitment.

Data and Method

The sites for this research are 35 moderate income, federally assisted housing developments in Newark, St. Louis, and San Francisco. They include row houses, walk-ups, and high-rise buildings. Those developments with two building types were treated as two sites, so the total number of sites in the study is 43: 7 high-rise, 25 walk-up, and 11 row house sites. The high-rise buildings have from 9 to 20 apartments on a floor and the walk-up buildings from 1 to 6. The sites are small, ranging from 40 to 500 apartment units, with an average of 141 units. The developments were between 2 and 12 years old at the time of the survey; many of them were attractive and well maintained and none was seriously deteriorated. The developments were built under federal housing programs that provide a share of the equity to sponsors and guarantee low-interest mortgage loans. Additional subsidies are provided to residents through various rent supplement programs. The developments house both low- and moderate- income households: half of the households in these sites reported a yearly income of $5,500, adjusted for family size. The majority of households are families with children and most of the residents are black.

The unit of analysis in this study is the housing site, rather than the individual resident or household, because the effects of the site characteristics specified by the model are hypothesized to operate at the group level. The model specifies, for example, that the number of apartments in a group will have a negative effect on the mean level of attachment: that sites where many apartments are grouped together will have lower levels of attachment than sites where few apartments are grouped together. These expected group-level effects were drawn largely from the work of Newman and Cooper. I have taken other expectations about particular relationships from studies that correlated individual characteristics (Kasarda and Jan- owitz, 1974; Hunter, 1975; Fischer et al., 1977) and placed them at the group level.

The primary source of data is a household survey of a stratified probability sample of 1,615 male and female adult residents. I obtained supplementary information from housing records, site plans, site visits, and interviews with managers. After constructing indices of activities, perceptions, and sentiments, I computed a mean score for each of these variables for each site (Table 1).

The attachment index included the following four items:

COMMUNITY BY DESIGN 295

“Suppose it wasn’t possible for you to live here any more and you had to move out of [name of development], how sad or happy would you be to leave?” (five-point scale from “very happy” to “very sad”).

“On the whole, how good or bad is [name of development] as a place to live?” (five-point scale from “very bad” to “very good”).

“What do most people from around [name of city] who have heard of [name of development] think of it? How good or bad a place to live do they think it is?” (five- point scale from “very bad” to “very good”).

“Right now, if you could have your way about it, how likely is it that you would move out of this development?” (five-point scale from “very likely” to “very unlikely”).

The friendship-kinship index included two items:

“How many close adult friends and close adult relatives do you have who live here at [name of development]?” (actual number coded into 10 categories from “no friends or relatives” to “more than 30”).

“How often do you get together with close adult friends and close adult relatives who live here at [name of development], for instance to visit or go out together?” (seven-point scale from “about once a year” to “more than once a week”).

And the three items in the acquaintance index are the following:

“In general, how often do you have casual conversations with other residents here at [name of development]?” (seven-point scale from “less frequently than once a month” to “several times a day”).

“How many families do you feel there are at [name of development] whom you can count on in an emergency?” (five-point scale from “none” to “very many”).

“How many families are there at [name of development] where you know at least one adult resident by name?” (actual number coded into 10 categories from “none” to “more than 30”).

Because the model is recursive and the errors are assumed to be uncorrelated, I used ordinary least squares regression to estimate the path coefficients (Hanushek and Jackson, 1977). I computed coefficients for the following four types of effect: total effects, direct effects, total indirect effects, and individual indirect effects (Alwin and Hauser, 1975). Coeffi- cients for total and direct effects were tested for significance using an alpha level of .05. Not to interpret a coefficient only because it is not statistically significant may, however, be a mistake when it is comparable in magnitude to other coefficients that are being interpreted (Heise, 1975). Therefore, a criterion of relative magnitude was also used: any direct or total effect equal to or larger than .30 is considered to be of substantive importance.

No techniques are available within path analysis for determining whether an indirect effect is statistically significant, that is, for simultaneously assessing the significance of all the links in an indirect path. Although there are suggestions for how best to interpret indirect effects (e.g., Alwin and Hauser, 1975), there is as yet no consensus on the criteria to use for judging their importance. It seemed reasonable to use a standard of relative magnitude: to be interpreted at all the coefficient for an individual indirect effect in this study must be equal to or larger than .05. Using this standard may of course result in the interpretation of indirect paths that are composed

TABLE 1

Mea

sure

men

t of

Var

iabl

es i

n M

odel

of

Co

mm

un

ity

Att

ach

men

t (N

= 4

3 si

tes)

Stan

dard

V

aria

bles

D

escr

iptio

n M

ean

Dev

iatio

n

No.

of

apts

./gro

up

% o

ne-p

aren

t w

elfa

re f

amili

es

I w

ith m

inor

s %

age

d 20 t

o 35

Age

of

site

C

oope

rativ

e ow

ners

hip

Ten

ants

’ as

soci

atio

n

Tur

nove

r ra

te

Vic

timiz

atio

n ra

te

Frie

ndsh

ip-k

insh

ip

bond

s

In h

igh-

rise

s an

d w

alk-

ups

this

is

the

num

ber

of a

part

men

ts

Prop

orti

on o

f ho

useh

olds

tha

t ar

e si

ngle

-par

ent

fam

ilies

wit

h

Prop

ortio

n of

hou

seho

lds

with

mem

bers

age

d 20

or,

youn

ger.

Pr

opor

tion

of h

ouse

hold

s w

ith h

ead

aged

20

to 35.

Num

ber

of y

ears

dev

elop

men

t ha

s be

en o

ccup

ied.

D

umm

y va

riab

le:

whe

ther

or

not

deve

lopm

ent

is

Dum

my

vari

able

: w

heth

er or

not

ther

e is

a t

enan

ts’

Num

ber

of f

amili

es w

ho m

oved

out

in 1976,

divi

ded

by

Tot

al n

umbe

r of

vic

timiz

atio

n ex

peri

ence

s in

site

dur

ing

on a

floor;

in r

ow h

ouse

s th

e va

lue

is z

ero.

wel

fare

as

maj

or s

ourc

e of

inc

ome.

coop

erat

ivel

y ow

ned

by r

esid

ents

.

asso

ciat

ion.

num

ber

of o

ccup

ied

unit

s at

end

of

that

yea

r.

prev

ious

yea

r re

port

ed b

y re

spon

dent

s,

divi

ded

by

num

ber

of r

espo

nden

ts,

mul

tiplie

d by

1,000.

and

freq

uenc

y of

vis

jts w

ith t

hem

. In

dex

of t

wo

item

s: n

umbe

r of

rel

ativ

es a

nd f

rien

ds i

n si

te

3.81

5.28

12.28

2.09

61.77

3.71

45.72

17.15

6.60

3.20

(5 c

oope

rativ

e, 38

nonc

oope

rativ

e si

tes)

(28

site

s w

ith

tena

nts’

ass

ocia

tion,

15 s

ites

wit

hout

) 20.76

16.95

.78

.42

61.48

11.76

Cas

ual

acqu

aint

ance

Exp

erie

nce

tryi

ng t

o pr

oble

m

solv

e a

Sens

e of

saf

ety

Perc

eive

d m

aint

enan

ce q

uali

ty

Com

mun

ity

atta

chm

ent

Inde

x of

thr

ee i

tem

s: f

requ

ency

of

casu

al c

onve

rsat

ions

with

17

.07

1.91

re

side

nts,

nu

mbe

r kn

own

by n

ame,

num

ber

can

turn

to

in e

mer

genc

y.

Inde

x of

tw

o ite

ms:

whe

ther

res

iden

ts h

ave

ever

got

ten

toge

ther

to

solv

e a

prob

lem

and

whe

ther

res

pond

ent

part

icip

ated

.

2.18

.4

2

Sing

le i

tem

: ho

w u

nsaf

e/sa

fe d

evel

opm

ent

is.

2.33

.5

1 Si

ngle

ite

m:

how

bad

lgoo

d m

aint

enan

ce i

s.

3.26

.6

4 In

dex

of f

our

item

s: h

ow s

ad t

o le

ave,

how

goo

d as

pla

ce

13.5

8 92

.24

to l

ive,

how

lik

ely

to m

ove,

how

goo

d im

age

is.

N~

E:

T

he s

ourc

e of

dat

a fo

r te

nant

s’ a

ssoc

iatio

n, c

oope

rativ

e, a

nd a

ge o

f si

te w

as a

sur

vey

of t

he m

anag

ers

of t

he d

evel

opm

ents

. T

he s

ourc

e of

dat

a fo

r tu

rnov

er r

ate

was

hou

sing

man

agem

ent

file

s. T

he s

ourc

e of

dat

a fo

r nu

mbe

r of

ap

ts./

grou

p w

as s

ite v

isits

an

d in

spec

tion

of

site

pl

ans.

T

he

sour

ce o

f al

l ot

her

data

was

a h

ouse

hold

sur

vey

of r

esid

ents

. R

espo

nses

to

the

item

s in

the

ind

ices

for

att

achm

ent,

fri

ends

6ip-

kins

hip.

and

acq

uain

tanc

e w

ere

wei

ghte

d on

the

bas

is o

f a

fact

or a

naly

sis

of t

hose

ite

ms.

298 KAREN A. FRANCK

of statistically nonsignificant direct paths. The argument being made here is that the indirect path is nonetheless substantively important.

The small sample size in this study (N = 43) causes the confidence intervals around the coefficients to be very large. Other than creating large sampling errors for the coefficients, the small sample size presents no problems for the use of path analysis as a technique. The small sample size is of course a constraint on the interpretation of Coefficients because sample size is taken into account in testing the significance of coefficients. Coefficients of the same magnitude, which are not significant in the present study, might prove to be significant with a larger sample.

Findings

Table 2 lists the different types of effects that each of the independent and intervening variables has on community attachment. The total effect that a given variable, X, exerts on another variable, Y, is the standardized multiple regression coefficient associated with X when none of the inter- vening variables in the relationship between X and Y have been included in the regression equation (Alwin and Hauser, 1975). The direct effect of X on Y is the coefficient associated with X when all variables that are expected to affect Y, independent and intervening, have been included in the regression equation. The total indirect effect of X on Y is the difference between the total effect and the direct effect (Alwin and Hauser, 1975) and is the sum of all the individual indirect effects. The coefficient for any individual indirect effect is the product of the Coefficients for the component direct effects. The total association is the zero-order correlation. The noncausal component of the relationship is the unanalyzed portion of the total association and is computed by subtracting the total effect from the zero-order correlation.

The number of apartments grouped together and the percent of one- parent welfare families are the only two site characteristics that have significant total effects on attachment. Both of these effects are negative: the greater the number of apartments grouped together and the higher the percent of one-parent welfare families, the lower the level of attachment. Of the intervening variables, three have significant total effects on attach- ment: the presence of a tenants’ association, the degree of acquaintance among residents, and the sense of safety. Each of these effects is positive. Although the level of friendship-kinship bonds among residents has only a relatively small and nonsignificant total effect on attachment, it has a significant direct effect that is negative.

Number of Apartments in a Croup

The effects of apartment group size on attachment confirm many of the expectations derived from Newman (1972, 1973) and Cooper (1970; Cooper, Day, and Levine, 1972). As shown in Figure 2, this design feature

COMMUNITY BY DESIGN 299

TABLE 2

Effects of Site Characteristics and Intervening Variables on Community Attachment

Site Characteristics Total and Intervening Total Direct Indirect Noncausal Total Variables Effect Effect Effect Component Association

1. No. apts./group 2. % one-parent welfare 3. % aged 20 to 35 4. 76 with minors 5. Cooperative 6. Age of site 7. Tenants’ association 8. Turnover 9. Victimization

10. Friendship-kinship 11. Acquaintance 12. Solving problem 13. Safety 14. Maintenance

-.53* -.46* -.lo - . lo

.19

.01

.24*

.09 -.19 -. 13

.36*

.02

.31*

.21

-.44* -.11 .03 -.49

-.05 -.05 -.15 .05

.02 .17 -.08 .09

.05 .19

.05 .04 - -.19

-.31* .18 .46* -.lo .09 -.07 .31* - .21 -

-.06 -.11 -.14 -.22

.18 -.08 -. 10 -.47 -.06

. l l

.07

.06

.37

.47

-.59 -.57 -.24 -.32

.37 -.07

.14 -.38 -.25 -.02

.43

.08

.68

.68

‘Coefficient significant at the .05 level

shows several of the expected negative indirect effects on attachment: through acquaintance (coefficient = - .06), sense of safety (coefficient = - .16), and perceived quality of maintenance (coefficient = - .06).* The strong negative direct effect supports the idea that large groupings of* dwelling units inhibit residents’ feelings of commitment to their residential environment (Newman, 1972, 1973; Cooper, 1971); Cooper, Day, and Levine, 1972).

Number of apartments in a group does not, however, have any significant negative effects on acquaintance or friendship-kinship bonds (Table 4). Large groupings of apartments do not seem to lead to the anonymity and social isolation that have sometimes been associated with high-rise living (Cooper, Day, and Levine, 1972; Newman, 1972, 1973). Similarly, these findings do not bear out the expectation of social withdrawal derived from the social overload theory of crowding in residential settings

TA

BL

E 3

Zer

o-or

der

Cor

rela

tions

bet

wee

n V

aria

bles

in

Mod

el o

f C

omm

unity

Atta

chm

ent

(N =

43

site

s)

Var

iabl

es

1 2

3 4

5 6

7 8

9 10

11

12

13

14

1. No. a

ptd

gro

up

2.

W

one-

pare

nt w

elfa

re

3. W

aged

20

to 3

5 4.

%

with

min

ors

5. C

oope

rativ

e 6.

Age

of

site

7.

Ten

ants

’ as

soci

atio

n

8. T

urno

ver

9. V

ictim

izat

ion

10. F

rien

dshi

p-ki

nshi

p

11. A

cqua

inta

nce

12. S

olvi

ng p

robl

em

13. S

afet

y

14. M

aint

enan

ce

1 .oo

.07

1.00

.0

6 .0

4 1.

00

-.03

.4

2 .3

1 1.

00

-.15

-.1

4 -.2

9 -.

03

1.00

.01

.10

-.03

-.

21

1.00

.3

1 -.

lo

-.27

-.

04

.27

.27

.12

.41

.16

-.29

-.26

.44

.19

.34

-.05

-

.34

-.06

.1

9 .3

4 -.

lo

-.I0

-.

23

.04

.42

.21

-.01

-.

23

.02

.30

-.36

-.71

.0

1 -.

29

.12

-.25

-.

61

-.08

-.3

2 .4

6

-.08

.03

.07

.07

.24

-.18

-.

I1

-,16

1 .oo

-.43

1.

00

-.21

.2

8 1.

00

.19

-.23

.1

5 1.

00

.42

-.40

-.12

.6

3 1.

00

.77

-.40

.04

.41

.40

1.00

.0

2 -.

15

-.37

-.

16

.06

-.12

1.

00

.10

-.24

-.

43

.08

.32

.04

.61

1.00

NU

TE: C

orre

latio

ns f

or a

ttac

hmen

t ar

e sh

own

in T

able

2.

FIG

UR

E 2

Maj

or C

ausa

l A

ntec

eden

ts o

f C

omm

unit

y A

ttac

hmen

t T

enan

ts’

asso

ciat

ion

Fri

ends

hip-

kins

hip

Acq

uain

tanc

e

Solv

ing

prob

lem

\

*Coe

ffic

ient

sig

nifi

cant

at

.05

leve

l

t

‘/

*.

Ant

enan

ce

302 KAREN A. FRANCK

(Baum, Harpin, and Valins, 1975; Baum and Valins, 1977; McCarthy, 1978).

In fact, number of apartments in a group has a beneficial effect on one form of social interaction. It has a positive effect on tenants’ association, suggesting that large groupings of apartments encourage the formation of such organizations. One explanation for this finding is suggested by the negative effects of apartment group size on sense of safety and perceived quality of maintenance. Residents in buildings with large groupings of apartments may form tenants’ associations to try to resolve these problems. Despite whatever attempts residents may make to improve conditions in buildings with many apartments grouped together, their feelings of com- mitment remain low. This is indicated by the direct negative effect of apartment group size on attachment. Thus, the presence of tenants’ associations in buildings with large groupings of apartments cannot over- come the negative effect that group size exerts on attachment.

As expected, experience in trying to solve a problem does link tenants’ association to attachment, but this indirect effect of tenants’ association is small (coefficient = .06) because experience in trying to solve a problem has only a small direct effect on attachment. The expressive forms of community activity, friendship-kinship and acquaintance, have far stronger effects on 3ttachment than this instrumental activity.

To discover whether the level of attachment differs between all pairs of building types, attachment was residualized, controlling for all inde- pendent variables except number of apartments in a group. The univariate F for this residualized variable by the three categories of building type was significant (p < .05) so the differences between each pair of building types were tested using Fisher’s procedure of least significant differences. As shown in Table 5, feelings of attachment are significantly lower in high-rise buildings than in either row houses or walk-up buildings, but there is no difference between the latter two types. This is consistent with the work of both Newman and Cooper who contrast the high-rise prototype with both types of low-rise buildings.

Percent of One-Parent Welfare Families

As expected, the percent of one-parent welfare families has a strong total negative effect on attachment that is completely accounted for by its indirect effects (Table 2). The indirect effects are mediated by sense of safety (coefficient = - .23) and perceived quality of maintenance (coefficient = -.09). These findings confirm the expectation drawn from Newman (1972, 1973) that a high percentage of one-parent welfare families has an adverse effect on safety and maintenance. The strong positive effect of percent one-parent welfare families on victimization rate (Table 4) also confirms Newman’s findings although victimization rate does not affect community attachment.

TA

BL

E 4

Stan

dard

ized

Coe

ffic

ient

s fo

r D

irec

t E

ffec

ts i

n M

odel

of

Com

mun

ity

Att

achm

ent

(N =

43

site

s)

Var

iabl

es

1 2

3 4

5 6

7 8

9 10

11

12

7. T

enan

ts'

asso

ciat

ion

8. T

urno

ver

9.

Vic

timiz

atio

n 10

. Fr

iend

ship

-kin

ship

11

. A

cqua

inta

nce

12.

Solv

ing

prob

lem

13

. Sa

fety

14

. M

aint

enan

ce

,378

.40'

-.27

.l

l

-.12

.I7

-.49

* -.

30

-.I3

-.

01

.39'

.35

-.06

-.

04

-.73

' -.

45

*

-.26

.I

1 .2

1 .0

9 .1

5 .I

2 .1

3 .0

4

.16

-.0

1

-.01

.04

.01

.I1

-.I2

-

.23

-.04

.03

,398

,3

49

.07

-.04

.29*

-.03

-

-.48

'

.05

-.04

.I3

.04

-.26

.0

9 .3

7'

.32

-.I3

-.

09

-.I2

.6

5'

-.23

.3

1'

.01

.29

.I2

-.26

.4

0*

- -.

07

.08

.I1

-.32

' .2

7 -.

32

-.20

-.

09

-

'Coe

ffci

ent

sign

ifica

nt a

t th

e .0

5 le

vel.

NO

IT:

Var

iabl

es 1

thr

ough

6 a

re n

amed

in

Tab

le 3

. D

irec

t ef

fect

s fo

r at

tach

men

t ar

e lis

ted

in T

able

2.

304 KAREN A. FRANCK

TABLE 5

Paired Multiple Comparisons of Residualized Means for Attachment by Building Type

I. Residualized Means

Row house Walk-up High-rise

5.26 1.75 -14.54

(N = 11) (N = 25) (N = 7)

11. Differences between Means

Row/Walk-up Row/High-rise Walk-up/High-rise

3.51 19.80* 16.29*

*CoeRicient significant at the .05 level. NWE: Attachment was residualized controlling for % one-parent welfare, % with minors, 56 aged 20 to 35, cooperative, and age of site.

The relationship of percent one-parent welfare families to attachment is consistent with earlier research showing that residents in lower social classes are less content with their neighborhoods (Hunter, 1974; Fischer et al., 1977). That there is no direct effect of percent one-parent welfare families on attachment bears out Fischer’s finding that once the quality of respondents’ housing and neighborhoods is taken into account, there is no difference between social classes with respect to contentment.

The strong positive effect of percent one-parent welfare families on the level of friendship-kinship bonds is consistent with the urban neigh- borhoods study of Kasarda and Janowitz (1974) showing that individuals of lower economic status have a larger portion of friends and relatives

COMMUNITY BY DESIGN 305

living in their own communities. It is also consistent with case studies of low-income, working-class urban neighborhoods, which show that residents have close ties with each other (Fried and Gleicher, 1961; Gans, 1962; Suttles, 1968). The positive relationship is inconsistent with case studies of public housing projects that imply that a concentration of low-income families in planned residential environments breeds anonymity among residents (Rainwater, 1970; Yancey, 1971; Moore, 1969). Even though in this study a high proportion of low-income families has an enhancing effect on the degree of mutual knowledge and intimacy among residents, this intimacy has an adverse effect on community attachment. Consequently, percent one-parent welfare families has another negative indirect effect on attachment through friendship-kinship bonds (coefficient = - . 1 1).

Friendship-Kinship and Acquaintance

The primary ties of friendship-kinship were separated from the more casual ties of acquaintance to explore the possibility that the two types of ties would have different effects on attachment, and indeed they do. The positive effect of acquaintance is consistent with previous findings showing that sense of community is positively affected by informal neighboring or informal social activities (Kasarda and Janowitz, 1974; Hunter, 1975; Fischer et al., 1977). The direct negative effect of friendship-kinship bonds on attachment stands in sharp contrast to the positive effect of acquaintance and to previous research that shows that both informal neighboring and primary ties have positive effects on attachment (Kasarda and Janowitz, 1974; Fischer et al., 1977).3

One explanation for the adverse direct effect of primary ties on attachment, and for the discrepancy with earlier findings, lies with the size and homogeneity of the area where the friends and relatives are located. In the studies of Kasarda and Janowitz (1975) and Fischer and others (1977) the areas being considered were urban neighborhoods whereas in the present study the areas are subsidized housing developments that are both smaller and more physically and socially homogeneous than many urban neighborhoods. The concentration of primary ties and the probable overlap of relationships within such relatively small, planned communities may lead to demands and to a lack of privacy that residents find unsatisfactory. It may even be oppressive. The size and character of these developments may cause high levels of friendship-kinship bonds to lead to excessive demands and lack of privacy that, in turn, result in low attachment.

The plausibility of this explanation receives some support from the reasons respondents gave for liking their development as a place to live. Although these answers were not systematically analyzed, a few excerpts illustrate some residents’ desire to maintain their distance from other residents.

306 KAREN A. FRANCK

I’ve always been comfortable here. I like my apartment and my neighbors seem to be nice but not nosey. I think it’s nice because nobody bothers anybody. I don’t get too familiar with anybody. Nobody bothers me and I don’t bother anybody. I’m not around with other people. They don’t know what’s going on with me. I like it here. I know my neighbors but I don’t get into problems. I don’t bother anyone and no one bothers me.

That close relationships with neighbors can create problems for residents, leading them to wish to forgo such ties, has been a finding in case studies of low-income public housing projects (Rainwater, 1970; Cooper, 1975). At Pruitt-Igoe residents expressed concern that neighbors should not learn too much about them for fear they would use this knowledge against them (Rainwater, 1970). Residents may have similar concerns in housing devel- opments occupied by a mixture of low- and moderate-income families.

Gove, Hughes, and Galle (1979) have treated excessive demands and lack of privacy within the home as intervening variables linking objective crowding to different forms of pathology. The authors found that the number of persons per room strongly affected felt demands and lack of privacy, both of which led to desire for and actual withdrawal. I am suggesting that a similar process is operating with respect to friendship- kinship bonds outside the home. The greater the number of such bonds within a small residential area and the greater the overlap among these bonds, the more demands and the less privacy residents are likely to experience in that setting and, hence, the greater their wish to withdraw either from those relationships in particular or from that community in general.

The direct negative effect of friendship-kinship bonds on attazhment is offset primarily by a positive indirect effect transmitted by sense of safety (coefficient = .12). As the level of primary ties in a site increases, the sense of safety increases and this, in turn, increases the level of attachment. Close relationships may make residents feel safe by giving them a strong sense of social support and mutual aid. Friendship-kinship bonds thus have two contrasting effects on attachment. First, close ties between residents may provide social support, thereby making residents feel safe and more attached to the community. Second, they may create excessive demands and a lack of privacy, thereby making residents feel less attached to the community.

Casual acquaintance also has contrasting effects on attachment. It has a direct positive effect, which is reasonable in the present context because acquaintance is unlikely to create excessive demands or a lack of privacy. It has an indirect negative effect that is transmitted by sense of safety (coefficient = - .lo). While friendship-kinship bonds increase the sense of safety, acquaintance decreases it. Close relationships may provide residents

COMMUNITY BY DESIGN 307

with support and mutual aid, thereby reducing fear, whereas more casual relationships may provide a means for communicating about crimes that have occurred without providing residents with support, thereby increasing fear. Unlike the friendship-kinship, the beneficial data effect of acquaintance on attachment outweighs its adverse effect so that weaker ties among residents have a large positive net effect on attachment, whereas stronger ties have only a small and nonsignificant net effect.

Implications

Local Community

Housing developments are different from urban neighborhoods in that they are named, centrally managed, and possess a high degree of physical homogeneity. Nevertheless, like neighborhoods, they can be viewed as one type or one level or local community and the findings from this study indicate that they perform some of the same functions as urban neighbor- hoods. They are places of informal social interaction, settings of primary ties of kinship and friendship, and objects of feelings of attachment and commitment. These findings join many earlier studies in demonstrating the continued survival of local community activities and sentiments in urban areas (Hunter, 1978).

The findings also raise a question about the presumed beneficial consequences of close relationships among residents. The tentative expla- nation for the direct negative effect of friendship-kinship bonds on attach- ment is that the presence and overlap of such bonds within small planned communities lead to excessive demands and reduced privacy that, in turn, cause residents to wish to retreat from their involvement in the community.

The idea that high levels of social stimuli cause excessive demands and lead to social withdrawal is the mainstay of the classic essays on urban life (Simmel, 1950; Wirth, 1938; Milgram, 1970). One difference between this idea and its application in the study by Gove, Hughes, and Galle (1979) or in the present study is the source of the social stimuli and the accompanying demands. In the earlier essays the demands are assumed to arise from contacts with strangers, whereas in the study by Gove, Hughes, and Galle (1979) the source is contacts with members of one’s own household. In the present study the source is assumed to be contacts with friends and relatives living outside the home but in the same development. In Gove, Hughes, and Galle’s research high levels of contact with members of one’s own household lead to desire for and actual withdrawal. In the present study high levels of contact with friends and relatives living nearby appear to lead to a desire to withdraw, as indicated by low levels of attachment.

The number of apartments in a group can be used as an index of the amount of contact residents have with other people, both friends and

308 KAREN A. FRANCK

strangers, near their homes. In the present study high levels of this type of contact do not lead to actual withdrawal because apartment group size has no effect on friendship-kinship or acquaintance. It does, however, have a negative effect on attachment, suggesting that high levels of contact with other people near one’s home may lead to a desire to withdraw. Apartment group size and friendship-kinship may have negative effects on attachment for the same reason: they both may cause excessive demands and lack of privacy. At the very least, the findings demonstrate that friendship-kinship bonds within small planned communities have both good and bad conse- quences and that the assumption of exclusively good consequences ought to be questioned.

Physical Design

After the studies of spatial proximity and friendship formation were conducted in the early fifties, interest among sociologists in the physical design of residential settings subsided until the early seventies. This hiatus may have been caused in part by a strong reaction against anything that appeared to be a form of architectural or physical determinism (Gans, 1968; Broady, 1972). There was also considerable skepticism about how much influence physical design features could have on social behavior (Keller, 1966; Gans, 1961, 1968), particularly compared to social and cultural factors.

Yet, whether the built environment is at all important and, if it is, how much so are empirical, not rhetorical, questions. As shown by the results of this study, when the questions are posed with reference to specific settings, the answers can be both complicated and enlightening. In this study physical design turns out to be as important as the major social characteristic of the sites in determining the level of community attachment among residents. While the influence of design is largely direct, the influence of social composition is mediated, primarily by sense of safety. Thus, even when design features and social characteristics have effects of comparable magnitude, it is unlikely that the causal dynamics of the effects will be the same.

While design has a powerful effect on attachment, it by no means affects all the attitudes and behaviors included in the causal model and with respect to those it does influence, it does not have the same type of effect on each one. Much as Gans (1961) would have predicted, the number of apartments in a group does not influence the nature or intimacy of relationships among residents. These relationships are primarily affected by social and organizational characteristics. A large number of apartments in a group does, however, have detrimental effects on turnover rate, sense of safety, and perceived quality of maintenance. In contrast, it has a beneficial effect on tenants’ association, suggesting that it may be the perceived problems in buildings with large groupings of apartments that

COMMUNITY BY DESIGN 309

lead to the more frequent formation of tenants’ organizations in such buildings. And finally, the findings on physical design and attachment suggest a kind of threshold effect: community attachment is significantly lower in high-rise buildings than in either row houses or walk-up buildings, but there is no difference in attachment between the latter two types. Physical design features may have an observable effect on activities and sentiments only when these features reach a particular level or threshold.* Where this threshold is will depend on the particular design features and behaviors being studied.

Unlike earlier research on physical design and local community, the present study involved the formulation and estimation of a causal model of the expected effects of physical design. With such a model one can trace out both the good and the bad effects of a particular design feature. More important, such a model requires the positing of intervening variables, thus forcing one to answer the question: “Why do you expect this feature of the built environment to have this effect?” Empirical research that measures the indirect effects mediated by these intervening variables allows one to support, modify, or reject one’s original expectations.

One of the frequent criticisms of physical determinism has been that it assumes only direct effects of environment on behavior (Gans, 1968; Broady, 1972). The explicit formulation and subsequent measurement of indirect relationships between the built environment and various behaviors and attitudes not only resolve that problem but are essential if any theoretical or empirical progress in research about the built environment is to be made. Another weakness of physical determinism is that the built environment is always treated as the independent variable. This is not necessary. One might well consider the factors that determine the design of particular environments; in this case features of the built environment would be treated as dependent variables (Franck, 1980).

Overcoming the weaknesses of the physical determinist perspective will make it easier to incorporate physical design variables into research about local community. There are, however, other obstacles to such a step. There is no precedent for sustained consideration of the built or natural environment in mainstream sociology (Catton and Dunlap, 1980). In research on local community in particular the belief is reiterated that other factors are more important than physical design and therefore should be given priority (Suttles, 1975).

Given current trends and incentives, these obstacles are not insur- mountable. Sociologists are giving more consideration to built and natural environments than ever before, and the view of local community as a multidimensional phenomenon is more compatible with a sustained con- sideration of the built environment than many other areas in sociology. Recently proponents of that view have begun to introduce concepts of the built environment into their discussions (Hunter, 1981). One incentive for

310 KAREN A. FRANCK

more frequent consideration of physical design is the current need for social scientists to assess the social impacts of proposed projects such as highways, urban development, nuclear power plants, and new housing. Such assessments, which require consideration of the potential effects of the physical design features of the proposed projects, will be more valuable if they can be based on previous research findings.

ENDNOTES

The data for this study were collected as part of a larger research project funded by the Law Enforcement Assistance Administration, National Institute for Law Enforcement and Criminal Justice, Department of Justice, grant 76-NI-990036-S-1. I would like to express my gratitude to Oscar Newman, the principal investigator of the project, for making these data available to me and for his guidance. This paper is based in part on my Ph.D. dissertation completed at the City University of New York. I would like to thank the members of my thesis committee (Maxine Wolfe, Harold Proshansky, and Oscar Newman) for their help. I am also grateful to Arnold Simmel for his advice. An earlier version of this paper was presented at the annual meeting of the American Sociological Association, Boston, Massa- chusetts, August 1979. I appreciate comments made by Illana Samets, Noel Dunivant, and anonymous reviewers on earlier drafts of this paper.

’ Figure 1 is a simplified representation of the model. The various sets of variables are enclosed in boxes to reduce the number of arrows and make the diagram legible. In the analysis each variable in each box is treated as a separate variable and is expected to affect all subsequent variables both directly and indirectly. No causal relationships are posited between the variables in the same box.

For ease of presentation the variables included in Figure 2 are limited to community attachment, the six independent and intervening variables that have significant direct or significant total effects on attachment, and other intervening variables that account for substantive individual indirect effects from these other variables to attachment. The paths in Figure 2 depict the significant direct effects of these variables on attachment, the individual indirect effects of these variables on attachment that are larger than .05, and the residual effect on attachment.

Some readers may attribute the unexpected negative effect of friendship-kinship on attachment to the relatively high correlation between friendship-kinship and acquaintance (r = .63). This correlation does reduce the precision of the estimated effects of both friendship kinship and acquaintance on attachment by increasing the sampling variance of these estimates. The correlation does not bias the estimated effect of either variable on attachment (Hanushek and Jackson, 1977). Thus, the correlation between friendship-kinship and acquaintance may reduce the amount of confidence one places in the estimated effects of each of these variables, but it does not constitute an explanation for the sign of those effects.

* Cillis (1979) has made a similar argument with respect to the effects of household density.

REFERENCES

Alwin, Duane F. and Robert M . Hauser 1975 “The decomposition of effects in path analysis.” American Sociological Review

40:37-47. Baum, Andrew, R. Edward Harpin, and Stuart Valins

COMMUNITY BY DESIGN 31 1

1975 “The role of group phenomena in the experience of crowding.” Environment and Behavior 7: 185-198.

Architecture and Social Behavior. Hillsdale, N.J.: Lawrence Erlbaum.

1977 Housing Messages. Stroudsburg, Pa.: Dowden, Hutchinson, and Ross.

1968 “The city, the suburb, and a theory of social choice.” Pp. 181-196 in Scott Greer, Dennis McElrath, David Minar, and Peter Orleans (eds.), The New Urbanization. New York: St. Martin’s.

“Social theory and architectural design.” Pp. 170-185 in Robert Gutman (ed.), People and Buildings. New York: Basic Books.

“Neighborhood interaction in a homogeneous community.” American Sociological Review 15:357-366.

Catton, William R. and Riley E. Dunlap “A new ecological paradigm for post-exuberant sociology.’’ American Behavioral Scientist 24: 15-47.

Baum, Andrew and Stuart Valins

Becker, Franklin D .

Bell, Wendell

1977

Broady, Maurice 1972

Caplow, Theodore and Robert Forman 1950

1980

Choldin, Harvey M. 1978 “Social life and physical environment.” Pp. 352-386 in David Street (ed.),

Handbook of Contemporary Urban Life. San Francisco: Jossey-Bass.

“Residents’ attitudes toward the environment at St. Francis Square, San Fran- cisco.” Working Paper No. 126. Berkeley: Institute of Urban and Regional Development. Easter Hill Village. New York: The Free Press.

“Analysis of the design process at two moderate-income housing developments.” Working Paper No. 80. Berkeley: Institute of Urban and Regional Development.

1972 “Resident dissatisfaction in multi-family housing.” Working Paper No. 160.

Cooper, Clare 1970

1975

1968 Cooper, Clare and Phyllis Hackett

Cooper, Clare, Noel Day, and Beatrice Levine

Berkeley: Institute of Urban and Regional Development.

“Path analysis: sociological examples.” American Journal of Sociology 72: 1-16.

Social pressure in informal groups. New York: Harper.

Duncan, Otis Dudley

Festinger, Leon, Stanley Schacter, and Kurt Back

Fischer, Claude, Robert M. Jackson, C. Anne Stueve, Kathleen Gerson, Lynne M. Jones, and Mark Baldassare

Networks and places: social relations in the Urban Setting. New York: Free Press. Franck, Karen A.

“Exorcising the ghost of physical determinism.” Paper presented at the annual meeting of the American Sociological Association, New York.

“Sources of residential satisfaction in an urban slum.” Journal of the American Institute of Planners 27:305-315.

“Planning and social life.” Journal of the American Institute of Planners 27:134- 140. The Urban Villagers. New York: The Free Press.

(ed.), People and Plans. New York: Basic Books.

1966

1950

1977

1980

Fried, Mark and I? Gleicher 1961

Gans, Herbert J . 1961

1962 1968 “The effect of a community on its residents.” Pp. 12-24 in Herbert J. Gans

312 KAREN A FRANCK

Gillis, A. R. 1979 “Household density and human crowding: unraveling a non-linear relationship. ”

Journal of Population 2:104-117. Gove, Walter R., Michael Hughes, and Omer R. Galle

“Overcrowding in the home: an empirical investigation of its possible pathological consequences.” American Sociological Review 44:59-80.

“Site planning and social life.” Journal of Social Issues 22:103-115. People and Buildings. New York: Basic Bcoks.

1979

Gutman, Robert 1966 1972

1977 Statistical Methods for Social Scientists. New York: Academic Press.

1975 Causal Analysis. New York: Wiley.

1955 “Definitions of community: areas of agreement.” Rural Sociology 20: 1 1 1-123.

1974 Symbolic Communities. Chicago: University of Chicago Press. 1975 “The loss of community: an empirical test through replication.” American

Sociological Review 40:537-552. 1978 “Persistence of local sentiments in mass society.” Pp. 133-162 in David Street

(ed.), Handbook of Contemporary Urban Life. San Francisco: Jossey-Bass. 1981 “Material neighborhoods and social neighborhoods. ” Paper presented at the

annual meeting of the Eastern Sociological Society, New York.

“Community attachment in mass society.” American Sociological Review 39:328- 339.

Keller, Suzanne “Social class in physical planning.” International Social Science Journal 18:494- 512. “Neighborhood concepts in sociological perspective.” Pp. 190-212 in John Gabree (ed.), Surviving the City. New York: Ballantine Books.

“Blue print for living together.” Pp. 1-202 in Leo Kuper (ed.), Living in Towns. London: Cresset Press.

“Principles of path analysis.” Pp. 3-37 in Edgar F. Borgatta (ed.), Sociological Methodology 1969. San Francisco: Jossey-Bass.

Effects of Tenant Population Size on Low-Income Public Housing Residents. Ph.D. dissertation, City University of New York.

“Toward an understanding of community satisfaction.” Pp. 299-352 in Amos H. Hawley and Vincent P. Rock (eds.), Metropolitan America in Contemporary Perspective. New York: Halsted Press.

Hanushek, Eric A. and John E. Jackson

Heise, David R.

Hillery, George A.

Hunter, Albert

Kasarda, John D. and Morris Janowitz 1974

1966

1973

Kuper, Leo 1953

Land, Kenneth C . 1969

McCarthy, Dennis 1978

Marans, Robert W. and Willard Rogers 1975

Merton, Robert K. 1948 “The social psychology of housing.” Pp. 163-217 in William Dennis (ed.),

Current Trends in Social Psychology. Pittsburgh: University of Pittsburgh Press.

Man and His Urban Environment, Reading, Mass.: Addison-Wesley. “The reconciliation of ‘subjective’ and ‘objective’ data on physical environment in the community.” Sociological Inquiry 43:147-173.

Michelson, William H. 1970 1973

COMMUNITY BY DESIGN 313

Milgram, Stanley

Moore, William

Newman, Oscar

1970

1969 The Vertical Ghetto. New York: Random House.

1972 Defensible Space. New York: Macmillan. 1973 Architectural Design for Crime Prevention, Washington, D.C.: National Institute

of Law Enforcement and Criminal Justice. Community of Interest. New York: Doubleday.

The Quest for Community. New York: Oxford University Press.

Behind Ghetto Walls. Chicago: Aldine.

“The local community: a survey approach.” Pp. 46-60 in Scott Greer and Ann Lennarson Greer (eds.), Neighborhood and Ghetto. New York: Basic Books.

“High density environments: their personal and social consequences.” Pp. 259- 281 in Andrew Baum and Yakov M. Epstein (eds.), Human Response to Crowding. Hillsdale, N.J. : Lawrence Erlbaum.

“The metropolis and mental life.” Pp. 409-424 in Kurt H . Wolf (ed.), The Sociology of Georg Simmel. Glencoe, Ill.: Free Press.

“The experience of living in cities.” Science 67:1461-1468.

1980 Nisbet, Robert

Rainwater, Lee

Ross, H . Laurence

1953

1970

1974

Saegert, Susan 1978

Simmel, Georg 1950

[1903]

Stein, Maurice

Suttles, Gerald D. 1960

1968 1972 1975

The Eclipse of Community. Princeton: Princeton University Press.

The Social Order of the Slum. Chicago: University of Chicago Press. The Social Construction of Communities, Chicago: University of Chicago Press. “Community design: the search for participation in a metropolitan community.” Pp. 235-297 in Amos H . Hawley and Vincent P. Rock (eds.), Metropolitan America in Contemporary Perspective. New York: Halsted Press.

“Urbanism as a way of life.” The American Journal of Sociology 44:l-24. Wirth, Lewis

Yancey, William L. 1938

1971 “Architecture, interaction, and social control: the case of a large-scale public housing project.” Environment and Behavior 3:3-21.

Charlesview Housing. Cambridge, Mass. : Architectural Research Oflice, Harvard University Graduate School of Design.

Zeisel, John and Mary Griffin 1975