You are on page 1of 19

The Perception and Fear of Crime: Implications for Neighborhood Cohesion, Social Activity,

and Community Affect


Author(s): Timothy F. Hartnagel
Source: Social Forces, Vol. 58, No. 1 (Sep., 1979), pp. 176-193
Published by: Oxford University Press
Stable URL: http://www.jstor.org/stable/2577792 .
Accessed: 14/06/2014 09:51

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Oxford University Press is collaborating with JSTOR to digitize, preserve and extend access to Social Forces.

http://www.jstor.org

This content downloaded from 195.78.108.60 on Sat, 14 Jun 2014 09:51:47 AM


All use subject to JSTOR Terms and Conditions
The Perception and Fear of Crime:
Implications for Neighborhood
Cohesion, Social Activity,
and Community Affect*

TIMOTHY F. HARTNAGEL, Universityof Alberta

ABSTRACT
This researchexaminesthe relationshipbetweenthe perceptionand
fearof crimeon the one handand neighborhood cohesion,socialactivityand
affectfor the communityon the other,usingsurveydatacollectedfrominter-
views with a sampleof residentsof a westernCanadiancity. Thehypotheses
that theperceptionof increasedcrimeand thefearof crimewouldbeinversely
relatedto neighborhood cohesionand socialactivitywerenot supported.But
as hypothesized,the fear of crimewas negativelyrelatedto affectfor the
community.And the predictionthat the experienceof actual victimization
wouldnotaffectthesehypothesized relationshipswassupported.Whenvarious
socialand residentialvariableswereincludedwith fearof crimein a multiple
regressionto predictcommunityaffect,lowfearand olderage werefound to
result in greateraffectbothfor the neighborhood and the city. In addition,
femalesand the less well-educatedhadmoreaffectfor the city.An exploration
ofpossibleinteractioneffectsbetweenfearof crimeandthesocialandresidential
variablesdid notyieldany significantresults.

The Problem and Related Theory

In a recent collection of essays, James Q. Wilson remarked that "predatory


crime does not merely victimize individuals, it impedes and, in the extreme
case, prevents the formation and maintenance of community. By disrupting
the delicate nexus of ties, formal and informal, by which we are linked
with our neighbors, crime atomizes society and makes of its members mere
individual calculators estimating their own advantage. . . . " (21). Similar
concerns had earlier been voiced by McIntyre in her discussion of the
*Revisedversion of a paper presented at the annual meeting of the PacificSociologicalAsso-
ciation, 1978. Data were obtained from the 1977 Edmonton Area Study administeredby the
PopulationalResearch Laboratoryand the Departmentof Sociology, University of Alberta.
L. W.Kennedy provided useful comments on a previous draft.
C 1979, University of North Carolina Press. 0037-7732/79/010176-93$01.80
176

This content downloaded from 195.78.108.60 on Sat, 14 Jun 2014 09:51:47 AM


All use subject to JSTOR Terms and Conditions
Fearof Crime / 177

significance of the fear of crime as revealed by surveys of public attitudes.


Social interaction is reduced and people restrict their activities. "They forego
opportunities for pleasure or cultural enrichment, and they become less
sociable, more suspicious. The level of interaction and mutual trust in the
society is reduced; public places become less safe than they otherwise
might be" (46). And at a more abstract level, Cohen discussed how de-
viance may erode social organization through its impact on the trust that
members of a group have that other members will, by and large, play
by the rules or norms of the group. "Distrust, even if it is unfounded,
weakens organization by undermining motivation; to distrust others is to
see one's own effort as pointless, wasted, and foolish, and the future as
hazardous and uncertain. One is then inclined to 'pull out of the game' if
he can, and to invest his resources with those whom he can trust, because
deviance . . . destroys faith in future performance" (5).
But along with others, Cohen also recognized that under certain
circumstances deviance may contribute to social organization. Thus a com-
mon enemy-external or internal to the group-may unite the group
against the deviant, thereby contributing to group integration or solidarity.
So deviant behavior may lead to increased frequency and intensity of inter-
action among members of the group or community. This, of course, recalls
Durkheim's argument that crime increases social solidarity by binding
people together in opposition to the law violator: "Crime brings to-
gether upright consciences and concentrates them. We have only to notice
what happens, particularly in a small town, when some moral scandal has
just been committed. They stop each other on the street, they visit each
other, they seek to come together to talk of the event and to wax indignant
in common" (102). Durkheim went on to point out that if the traversed
state "is strong, if the offense is serious, the whole group attacked masses
itself in the face of the danger and unites, so to speak, in itself" (103).
Similarly, Mead argued that ". .. . the attitude of hostility toward the
lawbreaker has the unique advantage of uniting all members of the com-
munity in the emotional solidarity of aggression.... The criminal does
not seriously endanger the structure of society by his destructive activities,
and on the other hand he is responsible for a sense of solidarity, aroused
among those whose attention would otherwise be centered upon interests
quite divergent from those of each other."
Research attention to these issues has been minimal. For example,
some support for the notion that crime unites members of the community
appears in Poveda's observations of a town's reactions to adolescent drug
use. Community discussion groups were formed and a mass meeting held
to discuss drug use and the most appropriate community response to it.
Similarly, Seibel found that deviance and reactions to it enhanced the level
of social integration among groups in small African societies, probably
because the units for integration are the families of the offender and the

This content downloaded from 195.78.108.60 on Sat, 14 Jun 2014 09:51:47 AM


All use subject to JSTOR Terms and Conditions
178 / Social Forces / vol. 58:1, september 1979

victim-that is, groups rather than individuals. Conklin (a, b) interviewed


respondents in an urban and a suburban community in an eastern metro-
politan area in 1968 to investigate the possibility that the fear of crime
undermines the social fabric of community life. Using a perception of crime
scale, with items on the relative size of local crime rates as the independent
variable, Conklin reported inverse relationships between this measure and
reported concern for personal safety, interpersonal trust of others and gen-
eral affect for the community (how satisfactory as a place to live) among the
urban but not suburban sample. Admitting the absence of a good measure
of the extent of social interaction in the two communities, Conklin reported
the perception of crime scale to be unrelated in either locale to participation
in local organizations, clubs and social groups or to a question on the
number of best friends living in town.
It should be noted that Conklin's independent variable was the
perception of local crime-what others have sometimes called concern
about crime. This should be distinguished from the personal fear of victim-
ization or crime since, as Conklin and others (Erskine, a; Furstenberg) have
indicated, perception of the amount of crime and fear of crime involve
distinct components of personal reaction to the crime problem. In fact,
Furstenberg found that those most concerned about the problem of crime
were no more or less afraid of personal victimization and Erskine (a)
reported that although perception of the amount of crime changed several
times in the years 1963 to 1973, fear for safety had risen steadily since 1965.
Conklin (b) found for the urban sample only that those feeling safer were
also more trusting of others (r = .19) and were more satisfied with the
community, although the latter relationship was barely significant. He con-
cluded that crime produces insecurity, distrust and a negative view of
the community and reduces trust and attachment to neighbors and to
the community as a whole. Crime thus reduces social solidarity and per-
haps contributes to increased crime rates through weakened informal social
controls in the community resulting from reduced solidarity or integration.
Yancey, reporting on the work of Wolfe et al., concluded that the
level of personal integration into networks is strongly related to the per-
ception of human dangers in the environment. Persons who were not
integrated into such networks were more likely to express concern over
allowing their children out of the house, felt that they were vulnerable
to strangers, felt unsafe on the street at night and that children in the
neighborhood were out of control. Rainwater discussed the interpersonal
consequences for the poor of living in a world characterized by danger. A
generalized watchfulness and touchiness in interpersonal relations and an
avoidance of strangers and neighbors results, he suggests, from the percep-
tion of danger. And more recently, Savitz et al. reported that among black
adults residing in Philadelphia the greater the fear of criminal victimization
the greater the number of reported avoidances of previous actions (e.g.,

This content downloaded from 195.78.108.60 on Sat, 14 Jun 2014 09:51:47 AM


All use subject to JSTOR Terms and Conditions
Fearof Crime / 179

stay home at night, visit friends less, don't talk to strangers, etc.). How-
ever, there was little relationship between actual victimization and altered
behavior.

Methods

HYPOTHESES

The present study was designed to examine further the relationship be-
tween the perception and fear of crime on the one hand and neighborhood
cohesion, social activity and affect for the community on the other, using
data collected in an amalgam survey conducted by the Population Research
Laboratory, University of Alberta in the spring of 1977. Several hypotheses
were formulated to guide the present research. First, the perception of
increased crime was hypothesized to be inversely related to neighborhood
cohesion, social activity and affect for the community. Second, the fear of
crime was also hypothesized to be inversely related to neighborhood co-
hesion, social activity and community affect. We expect (Hypotheses One
and Two) that those respondents who perceive more of an increase in
crime and/or who are more fearful of crime will be less integrated into their
neighborhoods, will engage in less social activity, and will be less satisfied
with their communities. Third, since there is little evidence that the actual
experience of criminal victimization is related to the perception and fear of
crime (Ennis; Thomas and Hyman), it was hypothesized that the relation-
ships expected in Hypotheses One and Two would be unaffected by the
experience of actual victimization.
In addition to these hypotheses, selected demographic and residen-
tial factors will be introduced as control variables in a multivariate analysis
in order to explore possible spurious, intervening and interactive relation-
ships. These variables were selected from the available data after reviewing
the literature that deals with the correlates of the perception and fear of
crime, neighborhood cohesion, social activity, and community affect. This
literature can be briefly summarized as follows.

DEMOGRAPHIC CORRELATES

Sex, age, family income and educational level have been shown to be re-
lated to our independent and/or dependent variables. Reviews of a number
of different opinion polls dealing with crime and justice topics (Erskine, a;
Hindelang) have shown that females, nonwhites, the poor, the least edu-
cated and the elderly are the most fearful of crime. Similar results were
reported by Creechan et al. from the same data set as used in the present
research. These authors also found that females, the least educated and the
elderly were more likely to perceive more violent crime today than five

This content downloaded from 195.78.108.60 on Sat, 14 Jun 2014 09:51:47 AM


All use subject to JSTOR Terms and Conditions
180 / Social Forces / vol. 58:1, september 1979

years ago and Erskine (a) reported finding no consistent relationship be-
tween the perception of crime and various measures of socioeconomic
status (SES).
The long history of empirical interest in social class differences in
primary relationships and neighboring is relevant for the present work
since social cohesion in the community is enhanced by close, intimate,
informal interaction. A number of authors have concluded that lower class
persons exhibit less informal social activity and fewer social relationships
with their neighbors (Axelrod; Bell and Boat; Cohen and Hodges; Knupfer;
Smith et al.).1 Gans (a) has argued that many of the assertions about ways
of life in cities and suburbs can be analyzed in terms of class and life cycle
characteristics while more recently, Athanasious and Yoshioka reported
that age was related to friendship choices regardless of distance to friends'
homes but family income was a significant factor only at small distances.
Finally, research by Foote et al., Fried, and Yancey has indicated
that working class groups are as satisfied with their neighborhoods as the
middle class although Yancey reported that neighborhood satisfaction was
related to proximity of friends in the lower, but not the higher, SES cate-
gories. Michelson has concluded that no significant differences in the pre-
ferred form of homes, neighborhoods and cities have been shown to be
related to social class differences.

RESIDENTIAL CORRELATES

Home ownership, dwelling type, residence location, number of children at


home, spouse at home and length of residence represent the residential
variables used as controls. Boggs concluded that significantly fewer central
city compared to rural, small town and suburban residents considered
their own neighborhoods as safe. Central city residents were also more
likely to expect some serious crime to happen and less likely to expect help
from their neighbors. There is also a substantial body of literature based on
an ecological orientation toward informal neighborhood social relations
(Greer, a, b; Martin; Shevky and Bell; Young and Willmott). Tallman and
Morgner, summarizing this position, point out that a high ratio of single
family to multiple dwellings in a community may promote more home and
family centered orientations as well as similarity of life routines among
neighbors and that these factors, along with more shared open space,
may increase informal interaction among neighbors. Both Tomeh (a) and
Tallman and Morgner found greater neighborhood intimacy in suburban
communities after controlling for social class and other factors. Athanasious
and Yoshioka reported that both number of children and marital status
were related to friendship choices regardless of distance to friends' homes
while Tomeh (b) concluded that there was no relationship between length
of residence and frequency of contacts with neighbors. Finally, Kennedy,

This content downloaded from 195.78.108.60 on Sat, 14 Jun 2014 09:51:47 AM


All use subject to JSTOR Terms and Conditions
Fear of Crime / 181

using the same data set as the present study, found that respondents (Rs)
in multiple family dwellings are less likely to know their neighbors but
are more likely to spend time with friends and go out for entertainment
than those in single family dwellings. Similarly, suburban residents know
their neighbors but are less likely than urban residents to spend time
with friends and go out for entertainment. Increased length of residence
was related to an increased number of neighbors known while time spent
with friends and entertainment frequency declined. Finally, Rs in multiple
family dwellings expressed less satisfaction with their neighborhood when
social interaction with neighbors and friends was low.

SAMPLE

The data were collected by interviews completed in the Spring of 1977 with
341 Rs, which represented a response rate of 85 percent. A sample of
households was selected from the July 1976 Edmonton City Telephones
Street Address Directory2 using a multi-stage area cluster design with
stratification according to income, probability proportional to size and
equal probability of selection of each household. Rs were selected from
household members over eighteen years of age and were chosen on a sex
quota basis so as to represent both sexes.3

MEASUREMENT

The perception of crime was measured by asking Rs how much of a change


they thought there had been in violent crime in the city during the past five
years. Two measures of fear of crime were selected, the degree of personal
safety Rs felt in their own neighborhoods; and the safety of the city as a
whole as they rated it.
The dependent variable, neighborhood cohesion, was measured by
asking Rs how often they got together with any of their neighbors just for a
chat and how many of the adults in their neighborhood they would know
by name if they met them on the street. Two indicators were also chosen as
measures for social activity. Rs were asked how often they went out for
entertainment and how often they spent a social evening with friends who
live outside the neighborhood. Affect for the community was measured by
asking Rs to indicate their degree of satisfaction with the neighborhood
and the city as places to live.
Given the amalgam nature of the original study, only two questions
could be asked about the experience of criminal victimization. Rs were
asked whether, during the last year, anyone entered illegally into their
house or apartment and, second, whether during the last year anyone took
something directly from them by using force. Responses to the two items
were combined and dichotomized into victim and non-victim categories.

This content downloaded from 195.78.108.60 on Sat, 14 Jun 2014 09:51:47 AM


All use subject to JSTOR Terms and Conditions
182 / Social Forces / vol. 58:1, september 1979

Finally, standard demographic data on the Rs' sex, age, family in-
come and level of education were collected. In addition, information was
gathered on whether the R lived in single or multiple family housing,
rented or owned the property, length of residence, location of the residence
(inner or outer city neighborhood),4 and whether a spouse and/or children
lived in the home.
The data are analyzed using correlation and multiple regression
techniques. Dichotomized variables are treated as dummy variables and
ordinal measures are treated as interval given the robustness of regression
analysis (Bohrnstedt and Carter; Boyle).

DATA ANALYSIS

The data for an initial evaluation of Hypotheses One and Two can be found
in Table 1 where the zero-order correlation coefficients are presented. The
coefficients in the first row of Table 1 offer scant support for the first
hypothesis that the perception of increased crime would be inversely re-
lated to neighborhood cohesion, social activity and community affect. The
two coefficients relating perception of crime to neighborhood cohesion
(number of neighbors known and frequency of talking with neighbors) are
not significantly different from zero. Similarly, the relation between per-
ception of crime and frequency of spending a social evening with friends
was not statistically significant. However, the second indicator of social ac-
tivity-frequency of going out for entertainment-shows a significant but
only slight negative relation to the perception of increased crime ( -.135),
indicating that those Rs perceiving more of an increase in crime are only
slightly less likely to go out frequently for entertainment. But when con-
trols5 for relevant social characteristics and residential factors were intro-
duced even this slight relationship disappeared. Therefore we conclude
that the zero-order relation between perception of increased crime and
entertainment frequency is the spurious product of their relations with
other variables, particularly age, home ownership and length of residence.
Finally, both of the indicators of community affect exhibit small but statisti-
cally significant relationships with perception of crime. But both of these
relations are in the opposite direction from that predicted: those who per-
ceive more crime are more satisfied with the community. In the case of
satisfaction with the city, the partial correlations indicate that its relation
with crime perception is a spurious function of age and length of residence.
But the relation between neighborhood satisfaction and crime perception
persists with the introduction of control variables. In an attempt to further
interpret this unexpected result the first-order partial correlations between
perception of crime and neighborhood satisfaction, controlling in turn for
fear of crime in the city and in the neighborhood, the indicators of neigh-
borhood cohesion, and satisfaction with the city as a place to live, were

This content downloaded from 195.78.108.60 on Sat, 14 Jun 2014 09:51:47 AM


All use subject to JSTOR Terms and Conditions
Fear of Crime / 183

CD
c:
en
I
r-_
I
-
I
(U

U-~ ~ _
c~~~~~~~r
r_ _n
~~z.

C- 0

4LL 4- ru -c-1

>-r
D Q)
4-
C)
4-
<:1: C_

C-) 4-J ~0 0
o
> 0 .-41
z 4J 4- 4-
< E LU m O
-. -. L\

>-
E
0
0
-04-
co C4J 00
e ~ C4
C4
0
r-.
em

H) L) .Ci

Z/ Cr
L0 LtJ Cs: C
E (D U N N\ 0
0 4-
o 4. C I I

>-
0
ot
cn _UJ * .
_

O > N

oD
o 4-
(U m)
Cr u IJ )o o
Uo)O
O) <n >
_:
0
u UI 0- 0 r-. o t
0
or o, --

z r 4

LLi

o
- 04
0 c0 J
~~~C
cr-
0 0
~ 0
I
0
I
04
1
04
Lf\
(L -
< (UG)
WL 0 HZ

wL -o
0
C4
- 0
-c
CO
0o
._) N N co
0
o o 0~ 0 '0
\OU\

z u I. . * L -4I
-D o o co -
U.. ~~~0 -OC 04 ) 4 r- a) e or- rL

oD
z) Ix) o Q) u _ o

-4 U *- L.- LU m
- tr u
L m *- *-
(U.-
I-Ut; - m. U._

This content downloaded from 195.78.108.60 on Sat, 14 Jun 2014 09:51:47 AM


All use subject to JSTOR Terms and Conditions
184 / Social Forces / vol. 58:1, september 1979

calculated but the zero-order results remained unchanged. So we are left


with a small but puzzling positive relationship between perception of crime
and neighborhood satisfaction which probably is the result of the action of
some as yet unknown variable.
The second hypothesis shifts the focus from the perception to the
fear of crime and it was predicted that such fear, both in the neighborhood
and the city at large, would be negatively related to neighborhood cohe-
sion, social activity and affect for the community. The data for evaluating
this hypothesis are also found in Table 1. As can be readily seen, none
of the coefficients dealing with neighborhood cohesion or social activity
reaches statistical significance. However, the coefficients measuring the
relation between fear of crime in the neighborhood and in the city and
affect for the community are statistically significant, moderate in magni-
tude, and in the predicted direction. Those Rs who are more fearful of
crime evaluate their neighborhood and the city less favorably as places to
live. Furthermore, these relationships persist in the presence of controls
for social characteristics and residential factors. We therefore conclude that
these findings are not the spurious product of at least these test factors
and that the evidence supports that part of the second hypothesis that
predicted fear of crime to be negatively related to community affect. We
will return to these data shortly when we examine the direct and indirect
effects of these variables on affect for the community and explore the
possibility of interaction among variables.
The third hypothesis predicted that any relationships found be-
tween the perception and fear of crime on the one hand and neighborhood
cohesion, social activity, and community affect on the other would be
unaffected by the experience of criminal victimization. The first-order par-
tial correlations between the original variables, controlling for criminal vic-
timization, fail to differ appreciably from the zero-order correlations in
Table 1 suggesting, in support of the hypothesis, that the experience of
victimization does not affect the relationship (or lack thereof) between the
original variables. The additional possibility of interaction between victimi-
zation and the perception and fear of crime was tested through the inclu-
sion of the relevant multiplicative terms in the regression equations for
each of the dependent variables. The hierarchical F-test (Nie et al., 337)
was employed to evaluate the contribution of each of the three interactions
(victim-perception of crime; victim-fear of crime in city; victim-fear of crime
in neighborhood) to the explained variation. None proved to be statistically
significant. This finding adds further support for the hypothesis that the
experience of actual victimization does not have a significant impact on the
relationships predicted in hypotheses one and two.
The rest of this analysis is more exploratory, as no specific hypothe-
ses of interaction between the perception and fear of crime on the one
hand and the various demographic and residential variables on the other

This content downloaded from 195.78.108.60 on Sat, 14 Jun 2014 09:51:47 AM


All use subject to JSTOR Terms and Conditions
Fear of Crime / 185

were formulated. Furthermore, given the findings reported above, we will


restrict this analysis to the fear of crime and affect for the community
relationship. We will examine this relationship at both the neighborhood
and city levels. As an aid to the analysis to follow, Figure 1 depicts a path
model of the assumed causal order and direction of effects among the
variables, with affect for the community as the dependent variable. Certain
of the variables have been grouped into blocks and within these blocks
curved arrows represent intercorrelation. Straight arrows indicate causal
direction and the error terms have been omitted to sharpen the dia-
grammatic representation. Path coefficients were derived from successive
reduced-form regression equations: coefficients were derived initially for
the equation containing only exogenous variables and subsequently for
equations which add intervening variables in sequence from cause to effect.
The decomposition of these coefficients into direct and indirect effects
(Alwin and Hauser) to clarify the linkages among variables is presented in
Tables 2 and 3.
Considering affect for the neighborhood first, fear of neighborhood

3
x~~~~~~~~~l

Xl= SEX X8 TYPE


= DWELLING

X2= AGE Xg= LOCATION


RESIDENCE

X3= INCOME X OF RESIDENCE


= LENGTH

X4
X= EDUCATION = VICTIM

AT HOME
= CHILDREN Xl2 = FEAROF CRIME

X = SPOUSEAT HOMEX = COMMUNITY


AFFECT

X = HOME
OWNERSHIP

FIgu 1. PATHMODELOF EFFECTSON AFFECTFOR THE COMMUNITY

This content downloaded from 195.78.108.60 on Sat, 14 Jun 2014 09:51:47 AM


All use subject to JSTOR Terms and Conditions
186I Social Forces I vol. 58:1, september 1979

4-J

4- 0O1 0O1 C14 C14 0O%00 1- O"% r-_ LCr% LC%


LU
L O - LC% LC\ C14 _-r C14 \-O LC% C14 LCr r-
4~~~J 0 cN 00-I-N~ -00 0 0 0\0

c'~~J
'~O '~O ' r,- 0D ao 0D \. r-_ LC\
a)L. C'J LC~~~~~
~~~ LC~~~~
0
O~~'O -~~~O~~ Lcn
C~~~~~~~J
00 LCD 0:
00 00 0~~~~CD
C CD 0 0 0

0
S. C-

Z V)
o 4-J
u
E
~ 0
'JO
c
c14
r\D
'-
a r 00 c
n o-,1 c14cJ
C0 c
U)
C/) a1) 4 0 0 0 0 0 0
LUj 4- 0 0 0 0 0

O LU I II

uLJ

0~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~01
D-
- ~~ 4-J U Li..

Z) a)- 0a
a) \.O r-. Lr :
H- 0 0 a O,1
o4- ... - a)
LL L--
UJ

Z r-. L(\

o 4-
'-0
\-O -'O
0,1
\.O
0.
LCn
-
0a)
r-_
'-
LC\
\ Lr\ Lt\
-_ LCr\
Lr\
-
1

cu a) LC\ LC\ r-_ ' 0Oa -:r %D Lr\r_


o 4-i 4- 0 C'J c0 0 0 0
uJJ
o ~~~~~~~~~~~~~~~~~~~~
\

-CE a) -
o a) ~~~~~
~~~~~~~~~~~~~~~~0
-C 0. 0
0
<
~S-~~~~~~~~~
4-J -C
4-J
.~~~~~4-J
U,
a)
U)
-
E
0~~~ a)
~~~2~~~~~~~~~ . ~~~~~~~~
a) C
u 0
4-0 U
0 c
a)
~~~~~~~
4- -
0-
0
n (
< 4-J 4-J L- a) 0 - ) 4-J -C- E 0 -na U,
U)
o ~~E
0
(
u
0 CU
a) U-
4-.
0) Un
-. -
4-J L.--
0)E
>
a
o a)> X a) 0 E a)(l a 0 m) 0 U,
o L. ) a-- )n 0 a).
au a) ()

0
m~~~~~
a)J -J 0-C -uCJ

* a).- ~~~~~~~~O4-J II 01
)4 a) 4 a
I- 0> ~~~~~~CY
4-QJ(/

This content downloaded from 195.78.108.60 on Sat, 14 Jun 2014 09:51:47 AM


All use subject to JSTOR Terms and Conditions
Fear of Crime I187

4-J

4- (n C4 Lr\ -: Lr\ r- 00 _-r _: '


LUI -: 0 0 - LC\ 0 00 \D 0O% c c .
cn Lr\ o \'O O oac c cOJ 0 0o
CD
0 C-0 CD 0.- 0 00 _-

OL \-'D LC\ O'~co _:- r-_t_ 0 C*'J -:t


L) O" ( -?r %D - r-. Lr\ .- -l_
a- 00000

L.J

4-J E C*'J ,1
O
cn LC\ -tr O,1% \ C*'J _- C*'
u 0 LC\ 0 'c Cn O- 0 %\D
a1) 4-J 0 0 .- .- - 0 00 0 0 0
4- U 0 0 00000 C C 0 0 0
4- ....

C,)
LUJ - 0Qa) %
cr-~
(5
4-J U
c 0Z
Li.
wL
cr-
a)a) 0~
LU ~ ~0 LC~ (~~fl
\
. . . .

4-OJ

4- 0 00 0 0.
D 0 ... .

a--
zCY
C/) -c a)

u -j 00~~~~~~~~~~~~~~
%D

LUJ 4-J 00
0 _-r LC\ Lt\ \O c0 - '- -
cn -0 %D O"
- .,
o -LU
cua) an00 0
-:r
'%o 0n
.
r-.- --I
0
O
c'J
_-
r-._
r-_.
Z 0 4- ..

C/)a) 0C*'
0 a
o C 4-J
~~~~~~~~~~~~~~~~0
-c
> fl
-C
LU C 4-J a1) a) C 4- U .-0
C L 0 C ~C CY 00 OC L -
(1) cu 3a) C C - a1) 4- U (f (
4-J 4) - a 0*- a) 4J -C E 0 t
o a)L E -0 -) -0(U 4- -- >-)c
F- cu0 U - = a1) - --U CY 4-J L4-J >
O a)> X a1) U :3 .- 0 E a1) (n 0 Ca) U (cJ.- 0
L
CY) f
< L- a) Cy C 0 -c 0 3 a1) - 1) L . .) u C a) a.)
LL 0- U, < U
(w . J > U-LL < Y
C,)

U)
~~~0
a
V)C )u
a)- 4-J

3 0L ~~~~4-J
(U
CD a)~ c
c
C4 a1)
1I- ::> cY: V

This content downloaded from 195.78.108.60 on Sat, 14 Jun 2014 09:51:47 AM


All use subject to JSTOR Terms and Conditions
188 / Social Forces / vol. 58:1, september 1979

crime clearly has the largest effect ( -.378). The only other variable to have
a sizeable effect on neighborhood satisfaction is age (.251), with satisfaction
increasing with age. Furthermore, most (.205/.251 = 82%) of the effect
of age is direct and unmediated by other variables in the model. Family
income and type of dwelling have smaller total effects with 94 percent of
the latter being direct, while 53 percent (.085/.159) of the effect of family
income is mediated by the residential variables and 28 percent (.045/.159)
being a direct effect. It is also interesting to note that no indirect effects are
mediated by the variable of victimization and its total and direct effects
are also negligible. This provides additional confirmation for our earlier
conclusion with respect to Hypothesis Three. Similarly, there are no sub-
stantial effects mediated by the fear of crime except in the case of sex.
While the total effects of sex appear to be negligible, there seems to be
some evidence that fear of crime is operating to suppress the effect of sex
on neighborhood affect. The direct effect of sex is -.077 (females are more
satisfied with their neighborhood) but this appears to be negated by its
indirect effect through the fear of crime (.072). A similar, but smaller,
suppressor effect of fear of crime appears to occur in the case of the spouse-
at-home variable. In the main, however, less fear of crime and older age
result in greater affect for the neighborhood, with residence in a single
family dwelling showing a similar but smaller effect.
When we turn to satisfaction with the city as a place to live as
the indicator of the dependent variable we can immediately observe from
Table 3 that fear of crime in the city has by far the greatest effect, accounting
for 23 percent of the explained variation. Age also has a significant effect
on satisfaction with the city, 89 percent (.250/.281) of which is direct. And
here again increasing age results in greater satisfaction. The inclusion of
the variable of fear of crime in the city reduces the unexplained variation in
satisfaction with the city sufficiently for two variables-sex and education
to emerge as statistically significant predictors although neither was
reported as statistically significant in previous regression equations. Hence
control for fear of crime in the city permits the previously suppressed
effects of sex and education to emerge. And there is some evidence of a
similar effect of fear on the effect of the home ownership variable on
satisfaction with the city. Females, the less well-educated and, to some
extent, those who rent are more satisfied with the city. So in addition to its
substantial direct effect on satisfaction with the city, fear of crime in the city
also operates to suppress the effects of certain other factors. Finally, it is
also apparent from Table 3 that the experience of criminal victimization
again fails to operate as an intervening variable. So it appears that the
results at both the city and neighborhood levels are fairly consistent.
Tests for the presence of interaction effects in the prediction of both
neighborhood and city satisfaction were also undertaken. Not all pos-
sible first-order interactions among independent variables were examined

This content downloaded from 195.78.108.60 on Sat, 14 Jun 2014 09:51:47 AM


All use subject to JSTOR Terms and Conditions
Fear of Crime / 189

given the absence of a compelling theoretical rationale. Rather, the possible


interaction between fear and each of the other independent variables was
explored by including the products of these variables (e.g., fear x sex;
fear x age; etc.) as additional terms in a multiple regression equation.
Blalock (464) has pointed out that when the original variables belong in
blocks (as they do here) and/ or when multicollinearity exists between origi-
nal variables and the cross-product terms (the correlations ranged from
.969 to .175 with the average being .662), it is reasonable to measure the
extent to which the entire set of cross-product terms adds significantly to
the explained variance since the assessment of the effects of particular
cross-product terms may be too risky given the large sampling errors in-
volved. Neither the entire set of ten interaction terms nor the block of
four fear-social characteristics interaction terms or the block of six fear-
residential interactions could be added to the initial results.6 Therefore, we
can conclude that this exploration of possible interaction effects between
fear of crime on the one hand and social characteristics and residential
factors on the other has not yielded any significant results.

Discussion and Conclusions

Several of these findings warrant further consideration. We should first


recall the absence of any relationship between the perception of fear of
crime on the one hand, and the indicators of neighborhood cohesion and
social activity on the other. No doubt this result is partly statistical; that is,
the product of insufficient variation in the perception and fear of crime.
Perhaps not surprisingly, 85 percent of the Rs perceived a substantial in-
crease in violent crime over the past five years. But more interestingly, only
7 percent indicated a high fear of crime in their neighborhood while only
17 percent expressed a high degree of fear of crime in the city as a whole.
So Canadians, at least in this city of approximately one-half million resi-
dents, did not give evidence of a high degree of fear for their personal
safety from criminal victimization, particularly in their own neighborhood.
In addition, the Rs' fear of crime-and perhaps the level of crime
itself-may not be sufficiently high to alter established behavior patterns.
The frequency of various social activities may become entrenched and may
therefore require a high level of fear to be substantially altered. In fact, it is
possible that some threshold of crime or fear must be reached before cer-
tain behaviors will be altered. Furthermore, for the perception and fear of
crime to have the hypothesized negative relationship with neighborhood
cohesion such perception and fear may not only have to be above some
threshold level, they may also have to be widely shared in the particular
neighborhood communities. Relatively isolated individuals experiencing
fear of crime may not be sufficient.

This content downloaded from 195.78.108.60 on Sat, 14 Jun 2014 09:51:47 AM


All use subject to JSTOR Terms and Conditions
190 / Social Forces / vol. 58:1, september 1979

Of course, it could also be argued that the level of crime and fear
of it may not be great enough to generate an increase in neighborhood
cohesion or integration. Furthermore, the communal spirit in urban neigh-
borhoods of North American society may not be great enough for the fear
of crime to pose either a significant threat to neighborhood cohesion or
to produce any substantial intensification of solidarity or integration. For
example, only 14 percent of our Rs knew most of the adults in their neigh-
borhood and only 22 percent knew more than half. The neighborhood,
except perhaps in certain ethnic enclaves (see Gans, b), may simply not be
the basis of social integration in our society. This latter point would appear
to fit with Seibel's findings from Africa concerning crime and the units of
integration.
Finally, the absence of a relationship between the perception and
fear of crime and neighborhood cohesion, may result from the perception
and fear being too unfocused and not specific enough to threaten whatever
common values may be shared in the neighborhood. Thus the failure to
stimulate an increase in neighborhood cohesion. In other words, for an
increase in solidarity to occur, perception and fear of crime may have to
be focused on a specific external threat-a criminal rather than crime in
general. Furthermore, in our segmented, urbanized and bureaucratized
world, crime and the response to the criminal have increasingly become
impersonal phenomena to be responded to and controlled through formal
mechanisms and they may therefore fail to elicit much in the way of in-
formal response. Formal, bureaucratic crime control has taken over the
action, if not the function, of informal community response and control.
A second finding which calls for some comment was the observed
negative relation between the fear-but not the perception-of crime and
affect for the community. As predicted, those who were more fearful of
crime were less satisfied with their neighborhood and the city as places to
live. This finding reemphasizes the need to distinguish the perception of
crime from the personal fear of criminal victimization. Further, the fear of
crime appears to be related to an attitude, affect for the community but not
the behavioral measures of neighborhood interaction and social activity. As
suggested previously, behavior patterns may be too well established to be
significantly affected by the fear of crime-at least at the levels of fear
experienced here-while the more general and perhaps vague sense of
satisfaction with the community is influenced to some extent by the ex-
pressed fear of criminal victimization. Furthermore, fear of crime had the
strongest effect, net of other predictors, on community affect compared to
the demographic and residential factors included in the regression. So the
fear of crime may be an important variable to consider in studies of the
consequences of lowered community affect. For example, the decision to
move from one city or neighborhood to another may, among other things,
reflect satisfaction with or affect for the community which, as we have

This content downloaded from 195.78.108.60 on Sat, 14 Jun 2014 09:51:47 AM


All use subject to JSTOR Terms and Conditions
Fear of Crime I 191

seen, is significantly affected by the fear of crime. So it is possible that such


fear may have behavioral consequences but perhaps only through such an
attitudinal phenomenon as affect for the community.
Finally, we should reiterate the perhaps surprising absence of signi-
ficant interaction effects. This may in part reflect the fairly rigorous testing
procedure employed here-the inclusion of cross-product terms in a mul-
tiple regression with a hierarchical significance test-compared to such
procedures as statistical control through cross-classification. However, we
should not dismiss the possibility of more complex interactions than tested
here. Second- or even higher-order interactions are possible even in the
absence of first-order interactions. But, in the absence of a well-developed
theoretical rationale this possibility is more remote and, given the statistical
problems involved, unlikely to produce interpretable results. So we can
conclude, with Cohen that "much remains to be done by way of specifying
the conditions under which deviance of various sorts has this consequence
or that, and at what point the spread of deviance becomes destructive of
organization" (11).

Notes
1. However, Yancey found no relationship between social status and casual neighboring. But
lower status Rs were more likely to have friends living nearby than were those of higher
status, suggesting that once casual acquaintances are distinguished from friends, lower class
Rs are more closely tied to their neighbors. Athanasious and Yoshioka also found that social
class failed to differentiate friendship choices for next door neighbors but became a factor at
greater distances.
2. Several other sources were explored for a list of dwellings, including the City of Edmonton
Elections Office and Henderson's Directory. However, the City did not have its Civic Census
listing in a usable form and the most recent Henderson's Directory available was the 1975
edition which was considered too out of date given the rapid population growth in Edmonton.
The Street Address Directory has the advantage of being relatively up to date-listings in this
directory are approximately two months old at issue date-and is very convenient to use.
Two or more households in a single dwelling can usually be easily identified since in most
cases they have separate telephones. Also, apartment buildings have every separate unit with
a phone listed separately and multiple tenancy complexes are segregated and arranged in an
easily identifiable form. The main disadvantages of using this directory were the absence of
subscribers with silent numbers or households which were not subscribers. The City of
Edmonton Utilities Department advised us that 6.6 percent of the total subscribers had silent
numbers. No figures were available for those without telephones but a random cross-check of
the Street Address Directory with Henderson's Directory indicated that considerably less
than 10 percent of the addresses were affected. It is possible that the exclusion of households
without telephones may have slightly biased the sample since transients and those of lower
income are less likely to have telephones. It is also conceivable that those with silent numbers
might be more fearful of crime.
3. In order to provide adequate population estimates it was determined that 320 completed
questionnaires would be necessary. Given an anticipated 80 percent completion rate, an
original sample of 400 was necessary. This provides a sampling frame of 3 in 1,000 with
households as the sampling unit. In a family household either a husband or wife was eligible
for interviewing while in non-family households any member over the age of eighteen was
eligible. Because of a concern that women were more likely to be home and thus be over-
sampled, interviewers were instructed to try to obtain an equal number of male and female Rs
and were to request to talk to the male head of the household in the first third of the
interviews; thereafter either the male or the female head of the household was interviewed.

This content downloaded from 195.78.108.60 on Sat, 14 Jun 2014 09:51:47 AM


All use subject to JSTOR Terms and Conditions
192 I Social Forces I vol. 58:1, september 1979

4. Inner and outer city neighborhoods were defined by grouping census tracts accordingto
natural boundaries (a river, ravines, railroadtracks) and radius from the downtown core.
Innercity neighborhoods,therefore,were those fallingwithin the census tractslocatedwithin
a radius of less than two to three miles, depending on naturalboundaries, of the downtown
core. Outercity neighborhoodswere those fallingwithin all other census tractslocatedwithin
the city limits.
5. A table of partialcorrelationcoefficientsis availablefrom the author.
6. Forwardstepwise regression, with the significancelevel set at .05 and the toleranceat .01,
also failed to include any of the interactionterms.

References
Alwin, D., and R. Hauser. 1975. "The Decomposition of Effects in Path Analysis."
American SociologicalReview 40(1):37-47.
Athanasiou, R., and G. A. Yoshioka. 1973. "The Spatial Character of Friendship
Formation." Environmentand Behavior5(1):43-65.
Axelrod, M. 1956. "Urban Structure and Social Participation." American Sociological
Review 21(1):13-18.
Bell, W, and M. Boat. 1957. "Urban Neighborhoods and Informal Social Relations."
AmericanJournalof Sociology 62(4):391-98.
Blalock, H. 1972. Social Statistics. New York: McGraw-Hill.
Boggs, S. L. 1971. "Formal and Informal Crime Control." SociologicalQuarterly 12
(Summer):319-27.
Bohrnstedt, G. W., and T. M. Carter. 1971. "Robustness in Regression Analysis." In
H. L. Costner (ed.), SociologicalMethodology.San Francisco: Jossey-Bass.
Boyle, R. P 1970. "Path Analysis and Ordinal Data." American Journal of Sociology
75(4):461-80.
Cohen, A. K. 1966. Deviance and Control. Englewood Cliffs: Prentice-Hall.
Cohen, A. K., and H. Hodges. 1963. "Characteristics of the Lower-Blue Collar
Class." Social Problems10(4):303-34.
Conklin, J. E. a:1971. "Dimensions of Community Response to the Crime Problem."
Social Problems18(3):373-85.
. b:1975. The Impactof Crime. New York: Macmillan.
Creechan, J. H., T. F. Hartnagel, and R. A. Silverman. 1978. "Attitudes Toward
Crime and Law Enforcement." Unpublished manuscript.
Durkheim, E. 1893. The Division of Laborin Society. New York: Free Press, 1933.
Ennis, P H. 1967. CriminalVictimizationin the United States. Washington: Government
Printing Office.
Erskine, H. a:1974. "The Polls: Fear of Violence and Crime." Public Opinion Quarterly
38(1):131-48.
. b:1974. "The Polls: Causes of Crime." Public Opinion Quarterly38(2):288-98.
. c:1974. "The Polls: Control of Violence and Crime." Public Opinion Quarterly
38(3):490-502.
Foote, N. N., J. Abu-Lughod, M. Foley, and L. Winni. 1960. Housing Choices and
Housing Constraints. New York: McGraw-Hill.
Fried, M. 1963. "Grieving for a Lost Home." In L. Duhl (ed.), The Urban Condition.
New York: Basic Books.
Furstenberg, F. F., Jr. 1971. "Public Reaction to Crime in the Streets." The American
Scholar 40(4):601-10.
Gans, H. J. a:1962. "Urbanism and Suburbanism as Ways of Life." In A. M. Rose
(ed.), Human Behaviorand Social Processes. Boston: Houghton Mifflin.
. b:1962. The Urban Villagers.New York: Free Press.
c:1967. The Levittowners.New York: Pantheon Books.

This content downloaded from 195.78.108.60 on Sat, 14 Jun 2014 09:51:47 AM


All use subject to JSTOR Terms and Conditions
Fear of Crime / 193

Greer, S. a:1956. "Urbanism Reconsidered." American SociologicalReview 21(Febru-


ary): 19-25.
. b:1960. "The Social Structure and Political Process of Suburbia." American
SociologicalReview 25(August):514-26.
Hindelang, M. J. 1974. "Public Opinion Regarding Crime, Criminal Justice, and
Related Topics." Journalof Researchin Crimeand Delinquency 11(2):101-16.
Kennedy, L. W. 1977. EnvironmentalOpportunityand Social Contact.Edmonton: Popu-
lation Research Laboratory, University of Alberta.
Knupfer, G. 1953. "Portrait of the Underdog." In R. Bendix and S. M. Lipset (eds.),
Class, Status, and Power.New York: Free Press.
McIntyre, J. J. 1967. "Public Attitudes Toward Crime and Law Enforcement." An-
nals 374 (November):34-46.
Martin, W. 1956. "The Structuring of Social Relationships Engendered by Suburban
Residence." American SociologicalReview 21(August):446-52.
Mead, G. H. 1918. "The Psychology of Punitive Justice." American Journal of Soci-
ology 23:585-92.
Michelson. W. 1970. Man and His Urban Environment.Reading: Addison-Wesley.
Nie, N., C. Hull, J. Jenkins, K. Steinbrenner, and D. Bert. 1975. StatisticalPackagefor
the Social Sciences. New York: McGraw-Hill.
Poveda, T. G. 1972. "The Fear of Crime in a Small Town." Crime and Delinquency
18(2):147-53.
Rainwater, L. 1966. "Fear and the House-as-Haven in the Lower Class." Journalof
the AmericanInstitute of Planners 32(1):23-31.
Savitz, L., M. Lalli, and L. Rosen. 1977. City Life and Delinquency. Washington:
Government Printing Office.
Seibel, H. D. 1972. "Social Deviance in Comparative Perspective." In R. A. Scott
and J. D. Douglas (eds.), TheoreticalPerspectiveson Deviance. New York: Basic
Books.
Shevky, E., and W. Bell. 1955. Social Area Analysis. Stanford: Stanford University
Press.
Smith, J., W. H. Form, and G. P Stone. 1954. "Local Intimacy in a Middle-Sized
City." AmericanJournalof Sociology 60(November):276-84.
Tallman, I., and R. Morgner. 1970. "Life-Style Differences Among Urban and Sub-
urban Blue-Collar Families." Social Forces 48(March):334-48.
Thomas, C. W., and J. M. Hyman. 1977. "Perceptions of Crime, Fear of Victim-
ization, and Public Perceptions of Police Performance." Journalof Police Science
and Administration 5(3):305-17.
Tomeh, A. K. a:1964. "Informal Group Participation and Residential Patterns."
AmericanJournalof Sociology 70(july):28-35.
. b:1969. "Empirical Considerations on the Problem of Social Integration."
SociologicalInquiry 39(Winter):65-76).
Wilson. J. Q. 1975. Thinking about Crime.New York: Basic Books.
Wolfe, A., B. Lex, and W. Yancey. 1968. The Soulard Area. St. Louis: Social Science
Institute, Washington University.
Yancey, W. L. 1971. "Architecture, Interaction, and Social Control." Environment
and Behavior3(1):3-21.
Young, M., and P Willmott. 1957. Family and Kinship in East London. Glencoe: Free
Press.

This content downloaded from 195.78.108.60 on Sat, 14 Jun 2014 09:51:47 AM


All use subject to JSTOR Terms and Conditions

You might also like