You are on page 1of 14

Fear of Crime in the United States: A Multivariate Analysis

Author(s): Frank Clemente and Michael B. Kleiman


Source: Social Forces, Vol. 56, No. 2, Special Issue (Dec., 1977), pp. 519-531
Published by: Oxford University Press
Stable URL: http://www.jstor.org/stable/2577738 .
Accessed: 14/06/2014 15:26

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 62.122.72.154 on Sat, 14 Jun 2014 15:26:57 PM


All use subject to JSTOR Terms and Conditions
Fear of Crime in the United States:
A Multivariate Analysis
F RANK CL E M E NT E, Pennsylvania State University
MICHAEL B. KLEIMAN, Pennsylvania State University

ABSTRACT
Fear of crime in the United States has becomea problemas serious as
crime itself. Many commentatorshave pointed out that fear is greatly out of
proportion to the objectiveprobabilityof being victimized. But to date, few
multivariate analyses of fear of crime have been undertaken. The present
research moves in this direction by combining and analyzing two national
samplesfrom 1973 and 1974 in regard to fear (n = 2,700). Weemployedfive
variablescentral to the victimization literature-sex, race, age, socioeconomic
status, and community size. Multivariate Nominal Scale Analysis (MNA)
was employedto assess the independentability of each variable to predict re-
spondentswho indicateda fear of crime (42 %) and those who did not (58 %/6).
Findings indicated that sex and city size are strong predictorsof fear.
Age and race were somewhat less important than has generally been supposed
and the socioeconomicvariables, income and education, had small effects.
Merely on the basis of this system of explanatoryvariables, however, it was
possibleto classify correctlyalmost 72 percent of the entire sample in regardto
fear. Implicationsof this high explanatorypower as well as limitations of the
analysis are presentedand discussed.

As Weis and Milakovich recently pointed out, fear of crime in the United
States has become a problem as serious as crime itself. Findings ranging
from the President's Commission on Law Enforcement to the Gallup Poll
and the national victimization surveys of the Department of Justice (LEAA)
have all documented the pervasive fear of crime in this country. Such
widespread anxiety over possible victimization takes a serious toll in the
daily lives of many Americans. People are forced to change their usual
behavior. They stay off the streets at night, avoid strangers, curtail social
activities, keep firearms, buy watchdogs and may even move to other
neighborhoods. The situation was best summed up by McIntyre:

*Research supported in part by the College of Liberal Arts and the Environmental Policy
Center, Pennsylvania State University. Frank Andrews and Robert Messenger generously
provided advice in regard to the MNA program. Claudia Clemente served as research
assistant.

519

This content downloaded from 62.122.72.154 on Sat, 14 Jun 2014 15:26:57 PM


All use subject to JSTOR Terms and Conditions
520 / Social Forces / vol. 56:2, december 1977

... fear [of crime] is impoverishing the lives of Americans. People stay behind the
locked doors of their homes.... The general level of sociability is diminished....
Society is suffering from what economists label opportunity costs ... [people] are
not enjoying the opportunities of their communities . . . they ignore jobs in some
neighborhoods . . . meetings are poorly attended . .. library use is decreasing
and recreationfacilitiesremainunused (41).
In short, the cost of crime goes far beyond the economic and physi-
cal losses imposed by criminals. It extends to the forced alteration of daily
living habits as well as to the negative psychological effects of living in a
state of constant anxiety. Further, fear of crime has a deleterious effect on
the general social order. As fear becomes manifest in the avoidance of
strangers, sociability, mutual trust and the willingness to help others dis-
appear. Such phenomena, labeled indirect victimizations by Conklin (374),
signify a serious erosion in the quality of life in the United States (see also
Brooks; Ennis; Holden; Reiss, a,b).
Yet, despite widespread agreement that fear of crime is a major
social problem there have been few systematic efforts to ascertain the
determinants of that fear. The bulk of empirical investigation in criminol-
ogy has centered around either the causes of crime (e.g., Nettler), the
correlates of victimization (e.g., Ennis) or the treatment of offenders (e.g.,
Robinson and Smith). Even the new sub-area, victimology, is largely con-
cerned with the compensation of direct victims rather than with the issue
of indirect victimization (cf. McGrath).
In 1967 the President's Commission on Law Enforcement (52) stressed
the lack of adequate research on the determinants of anxiety over possible
victimization. And, as Brooks recently noted, the situation has not greatly
improved during the interim: "So little attention has been paid to fear of
crime that we do not know what tactics might be most effective in combat-
ing [this] fear . . ." (244). The goal of the present research is to tackle this
problem through a multivariate approach to the fear of crime. Even a
cursory review of the literature reveals that previous research has suffered
from inadequate statistical controls (cf. Ennis). This paucity of multivariate
work has seriously inhibited the development of causal explanations of
differential fear of crime.

Background

Most commentators have noted an important element of irrationality in the


public's fear of crime: fear of crime is far out of the proportion to the
objective probability of being victimized (cf. Boggs; Brooks; Furstenberg;
Morris and Hawkins). Further, the crimes people fear most (personal
violence) are those which occur least (President's Commission on Law

This content downloaded from 62.122.72.154 on Sat, 14 Jun 2014 15:26:57 PM


All use subject to JSTOR Terms and Conditions
Fear of Crime / 521

Enforcement). As Jane Jacobs stressed in Death and Life of Great American


Cities it does not take many incidents of violence to make people fear the
streets. Mass media, sensationalistic journalism, and the tendency to ste-
reotype crime as invariably violent all contribute to what may be an irra-
tional fear. (For further discussion of this issue see President's Commission
on Law Enforcement; Sutherland and Cressey).
But regardless of the extent to which the widespread fear of crime is
unwarranted or irrational, the fact remains that it exists and demands
alleviation. As W. I. Thomas commented years ago, what people define as
real is real in its consequences. The rate of females being victimized on the
street may be only twelve per thousand population (Department of Jus-
tice). But if 60 percent of women are afraid to walk alone in their neighbor-
hood at night (see Erskine), the definition of the situation has clearly
become an important social problem.
As to which segments of the population are the most fearful, the
evidence is very scanty. On the one hand, some survey data indicate that
those least in danger are the most afraid (cf. Harris Poll). On the other
hand, similar data suggest that residents of high crime areas also display
substantial fear of victimization (cf. Biderman et al.). Part of the reason for
these seemingly contradictory findings lies in a conceptual confusion. As
Furstenberg has pointed out, most research has tended to use fear of crime
and concern for crime interchangeably. This has resulted in ambiguous
findings since a person may be troubled by the problem of crime but not
afraid of being personally victimized.
Further reason for the ambiguity as to which segments of the popu-
lation most fear crime is the lack of multivariate research in the area. For
example, when simple bivariate survey findings indicate that the elderly
are highly afraid of crime we do not know if this is because they are old,
poor, disproportionately female, relatively undereducated, concentrated
in inner cities as opposed to suburbs or, as is most likely, a combination of
all of the above. Without multivariate analysis it is impossible to assess the
independent effects of relevant variables. In short, the entire question of
spuriousness remains unanswered.

Potential Explanatory Factors

The strategy employed here will be to fix on those variables most often
included in studies of victimization and fear of victimization. In general,
five variables predominate in the literature: sex, race, age, socioeconomic
status, and size of community. Keying on these five variables will maintain
continuity with previous work, and advance our research toward a causal
perspective.

This content downloaded from 62.122.72.154 on Sat, 14 Jun 2014 15:26:57 PM


All use subject to JSTOR Terms and Conditions
522 / Social Forces / vol. 56:2, december 1977

SEX

An extensive array of survey data indicates that females are less likely to be
victims of personal crime than males (cf. Department of Justice; Reiss, a).
This pattern is documented in Table 1 where the victimization rate for
males is almost double that of females. (The data in Table 1 are drawn from
the Justice Department's Criminal Victimization survey. These data are su-
perior to Uniform Crime Reports because, unlike the latter, they include
unreported crimes. They suffer this limitation, however: they cannot be
disaggregated by the nature of the relationship between criminal and
victim.)

Table 1. VICTIMIZATION
RATES IN THE UNITED
STATES BY SELECTEDCHARACTERISTICS,1973*

Crimes against Persons


Group
GrouP (rate per 1,000 population)
Population 17.2

Sex:
Males 23.2
Females 11.9

Race:
White 16.2
Black 25.5

Age
20-34 24.7
35-49 10.9
50-64 5.6
65 plus 4.4

Income($)
7,499 18.6
7,500-9,999 16.1
10,000-14,999 15.5
15,000 plus 13.4
*Data drawn from Department of Justice, 16-26.

Although males are much more likely than females to be victims of


crimes against the person, survey data indicate that about three times as
many women are afraid of crime than men (cf. Erskine). Unfortunately, it is
not possible to control for the obvious reluctance of men to admit fear.
Given the extremely wide disparity between the sexes, however, we would
hesitate to dismiss such differences entirely as a function of masculinity.

This content downloaded from 62.122.72.154 on Sat, 14 Jun 2014 15:26:57 PM


All use subject to JSTOR Terms and Conditions
Fear of Crime / 523

RACE

Blacks are more likely to be victims of crimes against the person than
whites. As Reiss (a) noted: "For all major crimes against the person . . the
probability of being a victim is greater for any Negro man or woman than
for any white man or woman . . ." (45).
There is evidence that blacks are also more afraid of victimization
than whites. Feagin, for example, found blacks were more likely to feel it
necessary to arm themselves to defend their homes against criminals.
Further, Biderman et al. found blacks in Washington, D. C. scored higher
on a crime anxiety scale than whites.

AGE

For the population over twelve there is an inverse relation between age
and victimization. The data reported in Table 1 clearly document the
decrease in victimization in each ascending age cohort.
As Gubrium has emphasized, however, this extremely low victim-
ization rate among the elderly is not widely recognized. Consequently, fear
of crime among the aged is very high. Consider, for example, recent state-
ments drawn from the social gerontology literature:
Fear of criminal victimization causes self-imposed "house arrest" among older
people . .. many refuse to venture out of doors (Goldsmith and Tomas,10).
. . . crime has increasingly become a source of public fear and anxiety . . . I don't
believe that such effects are anywhere more apparent than among aged persons in
the United States (Cunningham,3).

SOCIOECONOMIC STATUS (SES)

This is inversely related to victimization for crimes against the person. (The
reverse is true for property crimes.) As the data in Table 1 indicate, indi-
viduals in the higher income level have the lowest rate of victimization.
When fear of crime is considered (as opposed to concern over the
crime rate) survey data generally indicate a negative relationship with SES.
A Gallup poll of 1974, for example, found both income and education to be
inversely associated with fear. Similarly, Biderman et al. found scores on
their anxiety index to be lowest among Washington, D. C. residents with
the highest incomes. They concluded that people with greater financial
resources are better able to protect themselves from harm and, therefore,
have less fear of being victimized. And, of course, individuals in the higher
income brackets can afford to live in safe neighborhoods.

This content downloaded from 62.122.72.154 on Sat, 14 Jun 2014 15:26:57 PM


All use subject to JSTOR Terms and Conditions
524 / Social Forces / vol. 56:2, december 1977

SIZE OF COMMUNITY

This has generally been found to be directly related to the crime rate (e.g.,
Wolfgang). In discussing the results of a national victimization survey by
the Department of Justice Ennis concluded: ". . . it is clear that as one
moves from the central city to the suburbs out into smaller towns and rural
areas, the crime rates decline, but much more drastically for crimes against
the person than for property crimes . . ." (29).
Data on fear of crime seem to parallel actual crime rates in relation to
city size. Boggs found that residents of large cities in Missouri express
significantly greater fear than residents of suburbs or rural areas. Similarly,
Conklin compared an urban sample to suburban residents and fotlnd the
latter to be considerably less afraid of crime. Finally, a number of national
surveys indicate a positive association between fear of being victimized
and size of community (cf. Erskine).

Methods

DATA

The data were drawn from the 1973 and 1974 General Social Surveys
conducted by NORC at the University of Chicago. This social indicators
program is directed by James A. Davis.
Both the 1973 and 1974 surveys were multistage area probability
samples designed to represent the non-institutionalized adult population
of the continental United States (see Davis, a, b, for details). Combining
these groups yields a total sample size of 2,700. We use the total sample in
order to insure enough minority individuals, e.g., aged and blacks, for our
analysis.

OPERATIONAL DEFINITIONS

The variables are straightforward. Sex and race are natural dichotomies,
coded as (0) male and (1) female and (0) white and (1) black, respectively.
Age was broken down into four groups: (1) 18 to 34, (2) 35 to 49, (3) 50 to
64, and (4) 65 and over.
SES is measured along two traditional dimensions. First, income-
annual family income coded as (1) less that $7,000, (2) $7,000 to $9,999,
(3) $10,000 to $14,999, and (4) $15,000 and over. Second, education-coded
as (1) less than high school, (2) high school, and (3) greater than high
school.
Community size was measured in the NORC studies by the follow-
ing ordered categories:

This content downloaded from 62.122.72.154 on Sat, 14 Jun 2014 15:26:57 PM


All use subject to JSTOR Terms and Conditions
Fear of Crime / 525

(5) Large city (250,000 plus)


(4) Medium size city (50,000-250,000)
(3) Suburb of large city
(2) Small town (2,500-50,000)
(1) Rural (under 2,500)
Fear of crime was ascertained in the NORC surveys through the
following question: "Is there any area right around here-that is, within
a mile-where you would be afraid to walk at night?" Responses were
dichotomized as "yes" or "no."
While this question is limited in scope, it is especially appropriate
for our analysis for several reasons: (1) it was asked in identical form in
both the 1973 and 1974 surveys, (2) it has consistently been used in
previous research on the fear of crime (cf. Ennis), (3) it has face validity in
that it assesses fear of crime rather than concern over the crime rate, (4) it is
pertinent to our conceptual scheme since, as was mentioned earlier, crimes
against the person generate the most fear. Finally, and perhaps most
important, this particular question brings fear of crime brusquely down to
the personal level. If members of major segments of the population are
afraid to walk within a mile of their homes it is not being overly dramatic to
conclude that such people have a significant fear of crime. Perhaps this line
of reasoning was best stated by Reiss: "The quality of life in a community is
measured in part by the degree to which residents are free of fear as they
go about in public . .. one should develop a measure of the extent to
which the individual feels personally safe in public . . ." (b, 434). Clearly,
the question used in the NORC surveys is compatible with Reiss' sugges-
tion. On the other hand, however, it should be recognized that the depen-
dent variable does exclude the broader range of crime. As one reader has
noted, for example, corporate, political, and war crimes may also generate
fear. Future analyses, we hope, will study these dimensions of fear of
crime.

STATISTICAL PROCEDURES

Simple bivariate distributions were calculated to yield a gross idea of the


relationship between each independent variable and fear. Given the di-
chotomous nature of the dependent variable and the nominal or ordinal
nature of the predictor variables, Multivariate Nominal Scale Analysis
(MNA) was employed to obtain a multivariate perspective on the determi-
nants of fear. MNA provides an assessment of the controlled effects of a set
of nominal or ordinal explanatory variables on a categorical dependent
variable-in this instance fear of crime.
MNA derives from the work of Andrews and Messenger, who pro-
vide a useful summary of the technique:

This content downloaded from 62.122.72.154 on Sat, 14 Jun 2014 15:26:57 PM


All use subject to JSTOR Terms and Conditions
526 / Social Forces / vol. 56:2, december 1977

The relationshipof MNA to (MultipleClassificationAnalysis) and dummy variable


regression is extremely close.... MNA is designed to handle problems where:
(a) the dependent variableis a set of mutually exclusive categories-i.e., a nominal
scale, and (b) where the independent variables may be measured at any level of
measurement,including nominal measurement....
In brief, MNA is a series of Multiple Classification Analysis (MCA) runs in
parallel. While MCA and dummy variable regression analysis are applied
to one intervally scaled dependent variable, MNA converts a nominally
scaled dependent measure to several dummy variables, one dummy vari-
able for each category of the original nominal scale. These dependent
dummy variables then indicate whether a given case is in a given category
of the original dependent variable. Of course, since each of these newly
created dummy dependent variables has only one interval-that between
(1) presence in and (0) absence from the given category of the original
dependent variable-each meets the equal-intervals requirement of an in-
terval scale. It is thus appropriate to use such dummy dependent measures
in MCA. The MNA procedure simply provides a means of generating sum-
mary statistics analogous to those of MCA and dummy variable regression
analysis by performing parallel MCA's using each of these dummy vari-
ables in turn as a dependent variable. (See Andrews and Messenger for
further details.)
The MNA technique is especially useful in the present analysis be-
cause it provides information on explained variance. As Andrews and
Messenger note, MNA yields insight into two basic questions relevant to
our analysis: (1) What is the relationship of a particular independent
variable to the dependent variable after statistically holding constant all
other independent variables? and (2) Taken all together, how well do the
independent variables explain the variability in the dependent variable?
Clearly, these questions are crucial to our efforts to take a causal multivari-
ate approach to the fear of crime.
As an aside at this point it should be mentioned that prior to using
MNA we analyzed our data for possible problems generated by statistical
interaction. This preliminary analysis did not reveal any problems of statis-
tical interaction in our data. Accordingly, we proceeded with the MNA
technique. (Cf. Andrews and Messenger for further discussion of the
assessment of interaction effects for MNA.)

Findings

Before turning to the results of the MNA analysis it is useful, as well as


interesting, to examine some distributional aspects of the data. Table 2
disaggregates the sample on fear by each independent variable.
The data presented in Table 2 closely parallel previous survey find-

This content downloaded from 62.122.72.154 on Sat, 14 Jun 2014 15:26:57 PM


All use subject to JSTOR Terms and Conditions
Fear of Crime / 527

ings on fear of crime (cf. Erskine). Females, blacks, and residents of large
cities all display significantly more fear than their respective counterparts.
Differences by age, income, and education are not quite as substantial but
are nevertheless apparent. In general, the data provide strong support for
expectations drawn from earlier research.

Table 2. PERCENTAGEDISTRIBUTIONON FEAROF WALKINGALONE AT NIGHTWITHIN


ONE MILEOF HOME FOR SAMPLEAND RELEVANTSUBSAMPLES, 1973 AND 1974

Category N % Afraid* Category N % Afraid*

Sample 2,700 42 Income ($):


6,999 821 48
Sex: 7,000-9,999 429 41
Males 1,291 22 10,000-14,999 665 43
Females 1,409 61 15,000 plus 785 35

Race: Education:
White 2,386 40 <HS 938 44
Black 314 57 HS 898 44
>HS 864 38
Age:
18-34 972 41 Community Size:
35-49 721 40 Large city 621 60
50-64 611 43 Medium city 320 51
65 plus 396 50 Suburb 706 41
Small town 520 36
Rural 533 24
*Percent not afraid is, in each case, 100 minus the percent afraid.

Our discussion of the MNA analysis corresponds to the two ques-


tions posed earlier by Andrews and Messenger. First, what is the con-
trolled effect of each independent variable? The MNA technique yields
beta and beta-squared values which are directly analogous to the standard-
ized partial regression coefficients from conventional regression analysis
(Andrews and Messenger 81). These values are presented in Table 3.
Here we find a marked disparity in the predictive power of the in-
dependent variables. Sex with a beta-squared of .153 has relatively high
explanatory utility. City size, while not as influential as sex, also has
explanatory power. The other variables, however, are relatively less im-
portant.
The second piece of information yielded by the MNA technique is
the explanatory strength of all the independent variables acting in concert.
One index of this composite strength was presented in Table 3, showing
that the independent variables accounted for over one-fifth (.23) of the
variance in the dependent variable.

This content downloaded from 62.122.72.154 on Sat, 14 Jun 2014 15:26:57 PM


All use subject to JSTOR Terms and Conditions
528 / Social Forces / vol. 56:2, december 1977

Table 3. BETA AND BETA-SQUAREDVALUESFOR


SIX EXPLANATORYVARIABLES

Variable Beta Beta-Squared


Sex .39 .153
Race .05 .002
Age .09 .008
Income .06 .003
Education .02 .000
Size .24 .058
R = .48 R2 = .230
Note: MNA does not provide for assessment of statistical signifi-
cance because the technique is designed for use with large
samples such as ours where statistical significance is likely to be
an artifact of sample size (Andrews and Messenger, 37).

In addition to an R2, however, the MNA analysis also indicates


number and percent of respondents who were correctly classified as afraid
and not afraid by the system of independent variables. The results of the
classification portion of the MNA analysis are presented in Table 4.

Table 4. PERCENTAGEDISTRIBUTIONOF SAMPLEOBSERVATIONS


CLASSIFIEDINTOBOTH GROUPS BY THE MULTIVARIATE
NOMINALSCALE
ANALYSIS

Percentage of Observations in Group


at Left Classified into Group Below
Group Afraid Not Afraid

Afraid (1,139) 66% (757) 34% (382)


Not Afraid (1,561) 25% (383) 75% (1178)

Correctly classified = 757 + 1178 = 1935 = 71.7% of sample.


Incorrectly classified = 382 + 383 = 765 = 28.3% of sample.

The percentages reported in Table 4 indicate that 66 percent of the


1,139 respondents who said they were afraid were classified correctly by
the five independent variables. Similarly, 75 percent of the 1,561 persons
who said they were not afraid were classified correctly. Looking at these
results from the opposite perspective, only one-third of the afraid group
and a fourth of the nonafraid group were misclassified. These data suggest
that while the complex of variables does quite well in classifying individ-
uals into both groups it has more power for the non-afraid group. In other
words, it is somewhat easier to predict who is not afraid of crime than who
is afraid. This pattern derives from the general tendency of men to report
themselves as not afraid.
The classification results can be examined from an overall perspec-
tive as well. Of the 2,700 respondents in the sample 1,935 were correctly

This content downloaded from 62.122.72.154 on Sat, 14 Jun 2014 15:26:57 PM


All use subject to JSTOR Terms and Conditions
Fear of Crime / 529

classified into one of the two fear categories. This means that merely on the
basis of these six variables 71.7 percent of the cases were correctly pre-
dicted. In addition, as was noted earlier, the R2 for the analysis indicates
that our system of variables explains 23 percent of the variance in fear of
crime (see Table 3). This lends strong credence to the argument that these
variables are important determinants of fear of crime. (Some readers may
question the interpretation of 71.7% as significant since 58% of the sample
could be correctly classified by chance alone-the percentage who were
not afraid. This line of reasoning is flawed, however, because it completely
removes prediction from explanation. Both theory construction and pro-
gram development require a base more substantial than that provided by
mere chance. That our system of variables correctly predicted 71.7% of the
sample provides a very useful point of departure for future analysis.)

Implications

Several other important implications for research can be indicated. First,


the findings demonstrate the need for a multivariate approach in assessing
the determinants of fear of crime. The MNA analysis provided information
as to the impact of each independent variable while the effects of the others
were removed. This procedure enabled us to rank the variables by their
relative predictive power. Further, it revealed sizeable disparities in impor-
tance of the variables. Finally, and perhaps most important, the multi-
variate approach showed that the effects of race, age and SES are less than
would be expected on the basis of the simple percentage distributions
which dominate previous research on the problem. Race provides a clear
example of this situation. The data reported in Table 2 indicated a 17 per-
cent gap between blacks and whites on fear. In the MNA analysis, how-
ever, race ranked only fifth of the six explanatory variables. Similarly, the
controlled effect of age upon fear was certainly less than one would
suppose upon reviewing the social gerontology literature (cf. Goldsmith
and Tomas).
Second, implications for future research can be drawn from the limi-
tations of our analysis. For example, while the NORC question on fear of
crime was adequate for our purposes it could easily be improved by
eliminating the dichotomous format and including a broader range of re-
sponses. Further, other dimensions of fear of crime, e.g., afraid in one's
own home, also deserve careful attention.
Another limitation of our study revolved around the reluctance of
many males to say they were afraid. As we stated earlier, it is impossible to
disaggregate the sample on people who simply refuse to admit fear and
those who are genuinely not afraid. While our data are no different from
any other survey data in regard to this problem it is important to recognize

This content downloaded from 62.122.72.154 on Sat, 14 Jun 2014 15:26:57 PM


All use subject to JSTOR Terms and Conditions
530 / Social Forces / vol. 56:2, december 1977

the issue of respondent bias. Sensitive issues such as fear are especially
susceptible to contamination. Future research might pose the question
somewhat less bluntly than was the case in the NORC study, giving
respondents an opportunity to admit fear without feeling embarrassed.
Finally, our study demonstrates the need to examine the disjuncture
between objective conditions and the subjective interpretation of those
conditions. Comparison of the data reported in Tables 1 and 2 reveals wide
discrepancies between actual victimization rates and fear of crime. Not all
of these differences can be accounted for by the argument that fear of crime
makes some people stay home and thus lowers their victimization rate.
Rather, the fear of crime is only one of a variety of areas where the defini-
tion of the situation is out of phase with the empirical world. As Campbell
and Converse (8) point out, this facet of social research is underdeveloped:
We generally know very little about the detailed interplaybetween objective situa-
tions and the way people assess them subjectively.... Wehave become deeply im-
pressed at the degree to which subjectivestates can 'pull apart'from what might be
deduced on the basis of our currentways of understandingobjectivesituations (8).
In short, the relationship (or lack of relationship) between hard rates and
the social-psychological state of the population requires careful examina-
tion if we are to achieve a deeper understanding of the human meaning of
social change.

References
Andrews, Frank M., and R. C. Messenger. 1973. Multivariate Nominal Scale Analysis.
Ann Arbor: Institute for Social Research, University of Michigan.
Biderman, Albert D., L. Johnson, J. McIntyre, and A. Weir. 1967. Report on a Pilot
Study in the District of Columbiaon Victimizationand Attitudes Tozward
LazwEnforce-
ment. Washington: Government Printing Office.
Boggs, S. L. 1971. "Formal and Informal Crime Control: An Exploratory Study of
Urban, Suburban, and Rural Orientations." SociologicalQuarterly 12(Summer):
319-27.
Brooks, J. 1974. "The Fear of Crime in the United States." Crime and Delinquency
20(July):241-44.
Campbell, A., and P. E. Converse. 1973. "Social Change and Human Change." In
Angus Campbell and P. E. Converse (eds.), The Human Meaning of Social
Change. New York: Russell Sage.
Conklin, J. E. 1971. "Dimensions of Community Response to the Crime Problem."
Social Problems 18(Winter):373-85.
Cunningham, C. L. 1974. "The Sceriario of Crimes Against the Aged." In Noel
Tomas (ed.), Reducing Crimes Against Aging Persons. Philadelphia: Task Force
Report of Department of Health, Education, and Welfare.
Davis, James A. a:1973. General Social Survey. Chicago: National Opinion Research
Center.
. b:1974. General Social Survey. Chicago: National Opinion Research Center.
Department of Justice. 1974. Criminal Victimizationin the United States. Washington:

This content downloaded from 62.122.72.154 on Sat, 14 Jun 2014 15:26:57 PM


All use subject to JSTOR Terms and Conditions
Fear of Crime / 531

Law Enforcement Assistant Administration.


Ennis, Philip H. 1967. Criminal Victimizationin the United States. Washington: Gov-
ernment Printing Office.
Erskine, H. 1974. "The Polls: Fear of Crime and Violence." Public Opinion Quarterly
38(Spring):131-45.
Feagin, J. R. 1970. "Home-Defense and the Police: Black and White Perspectives."
AmericanBehavioralScientist 13(May-August): 797-814.
Furstenberg, F. A. 1971. "Public Reaction to Crime in the Streets." The American
Scholar 40(August):601-10.
Gallup, G. 1972. The Gallup Opinion Index-Political, Social and Economic Trends.
Report No. 82 (April).
Goldsmith, J., and N. Tomas. 1974. "Crime Against the Elderly: A Continuing
National Crisis." Aging (June-July):10-14.
Gubrium. J. F. 1974. "Victimization in Old Age." Crime and Delinquency 20(July):
245-50.
Harris, Louis. 1969. Special survey of public reaction to crime in Baltimore. Commis-
sioned by LifeMagazine (January).
Holden, M. 1969. "The Quality of Urban Order." In H. Schmandt and W. Bloom-
berg (eds.), The Quality of Urban Life. Beverly Hills: Sage.
Jacobs, Jane. 1961. Death and Life of Great American Cities. New York: Random
14ouse.
McGrath, W T. 1968. "Compensation to Victims of Crime." Canadian Journal of
Corrections10:591-99.
McIntyre, J. 1967. "Public Attitudes Toward Crime and Law Enforcement." The
Annals 374(November):34-46.
Morris, Norval, and G. Hawkins. 1971. The Honest Politicians Guide to Crime Control.
Chicago: University of Chicago Press. E
Morgan, James N., and Robert C. Messenger. 1973. THAID. Ann Arbor: Institute
for Social Research, University of Michigan.
Nettler, Gwynn. 1974. Explaining Crime. New York: McGraw Hill.
President's Commission on Law Enforcement. 1967. TaskForceReport. Washington:
Government Printing Office.
Reiss, Albert J., Jr. a:1967. Studies in Crime and Law Enforcementin Major Metropolitan
Areas. Washington: Government Printing Office.
. b:1973 "Monitoring the Quality of Criminal Justice Systems." In Angus
Campbell and P. Converse (eds.), The Human Meaning of Social Change. New
York: Russell Sage.
Robinson, J., and G. Smith. 1971. "The Effectiveness of Correctional Programs."
Crimeand Delinquency 17(January):67-80.
Sutherland, E. H., and Donald R. Cressey. 1970. Criminology. Philadelphia: Lippin-
cott.
University of Michigan. 1970. OSIRIS140 User's Manual. Ann Arbor: Institute for
Social Research.
Weis, K., and M. Milakovich. 1975. "Politics and Measure of Success in the War on
Crime." Crimeand Delinquency 21(anuary):1-10.
Wolfgang, M. E. 1967. The Subcultureof Violence. London: Tavistock.

This content downloaded from 62.122.72.154 on Sat, 14 Jun 2014 15:26:57 PM


All use subject to JSTOR Terms and Conditions

You might also like