-
The University of Chicago Press is collaborating with JSTOR to
digitize, preserve and extend access to American Journal
ofSociology.
http://www.jstor.org
Social Structure and Crime Control Among Macrosocial Units
Author(s): Allen E. Liska and Mitchell B. Chamlin Source: American
Journal of Sociology, Vol. 90, No. 2 (Sep., 1984), pp.
383-395Published by: The University of Chicago PressStable URL:
http://www.jstor.org/stable/2779220Accessed: 26-03-2015 21:31
UTC
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 contentin 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 [email protected].
This content downloaded from 200.130.19.139 on Thu, 26 Mar 2015
21:31:45 UTCAll use subject to JSTOR Terms and Conditions
-
Social Structure and Crime Control among Macrosocial Unitsl
Allen E. Liska and Mitchell B. Chamlin State University of New York
at Albany
Recent research, drawing on the conflict perspective, has
examined the effect of the racial/economic composition of
macrosocial units on the capacity for crime control (police size).
This paper extends this work to actual crime control (arrest
rates). The results show that there is considerable variation in
arrest rates between cities and that racial/economic composition
substantially affects them, indepen- dently of reported crime
rates. The effects are specified by type of arrest (property and
personal) and race of offender (white and non- white).
This paper is concerned with crime control among macrosocial
units. Most research, based on deterrence theory, examines the
effect of crime control on crime rates. Our research, drawing on
the conflict perspective, examines the structural causes of crime
control.
The conflict perspective conceptualizes crime control as an
instrument used by dominant and powerful groups to control those
actions and groups which threaten their interests. Turk (1969), for
example, argues that culturally dissimilar groups are perceived by
authorities as threats to the social and political order and that
crime control can be understood as an instrument for controlling
them. With the exception of various histor- ical and case studies
(Chambliss and Seidman 1982), there have been few explicit
empirical tests of the conflict perspective of crime control-
specifically, of the threat hypothesis-at the macro level. Recently
Jacobs (1979), Jackson and Carroll (1981), Liska, Lawrence, and
Benson (1981), and Loftin, Greenberg, and Kessler (1981) have
examined the effect of the racial/economic composition of cities
and SMSAs, as an indicator of the threat perceived by authorities,
on the capacity for crime control as mea-
' We would like to thank Joseph Lawrence for assistance with
data collection; Larry Sherman and Paul A. Zolbe for assistance in
obtaining the arrest data; and Judith Blau, Terry Blum, Ronald A.
Farrell, Richard B. Felson, Reid Golden, John Logan, Kenneth
Mazlen, Mark Reed, and Glenna Spitze for their comments on various
drafts of the paper. Requests for reprints should be sent to Allen
E. Liska, Department of Sociology, State University of New York at
Albany, Albany, New York 12222.
? 1984 by The University of Chicago. All rights reserved.
0002-9602/85/9002-0006$0 1.50
AJS Volume 90 Number 2 383
This content downloaded from 200.130.19.139 on Thu, 26 Mar 2015
21:31:45 UTCAll use subject to JSTOR Terms and Conditions
-
American Journal of Sociology
sured by police size and expenditures. Our research extends this
work in two directions. One, it examines the effects of three
dimensions of racial/ economic composition central to the studies
above (percentage nonwhite, segregation, and economic inequality)
on the actual volume of crime control as measured by arrests. Two,
it explicitly examines three social processes (threat, power, and
benign neglect) which, according to the conflict perspective
hypothesis, underlie the relationship between racial/ economic
composition and crime control.
PERCENTAGE OF NONWHITES The researchers above argue that
nonwhites, as a culturally and racially dissimilar subordinate
group, are perceived by authorities as threats to the social and
political order and that police are used to control them. Although
a relatively small culturally dissimilar subordinate group may not
be perceived as posing much of a threat, a relatively large
culturally dissimilar group constituting 20%-30% of the population
may be per- ceived as posing a substantial threat and as a problem
of social control. In particular, nonwhites are viewed as criminal
threats (Swigert and Farrell 1976; Lizotte and Bordua 1980; and
Liska et al. 1981). For example, Liska et al. (1981) report that,
when one controls for crime rates, the percentage of nonwhites in a
city substantially affects the fear of crime. Hence, we might
expect a strong relationship between the percentage of nonwhites in
a social unit and the level of crime control. Jackson and Carroll
(1981), Jacobs (1979), Liska et al. (1981), and Loftin et al.
(1981) consistently show that, when one controls for reported crime
rates, the percentage of nonwhites in cities and SMSAs relates
substantially to police size.
The implications of the percentage of nonwhites for the actual
volume of crime control, as measured by arrest rates, are much more
complex. The conflict perspective implies three distinct causal
processes.
To clarify these processes, it may be useful to think of the
total arrest rate as composed of the rates for whites and nonwhites
weighted by their proportions of the city population. Conflict
theory assumes that non- whites have a substantially higher arrest
rate than whites because, rela- tive to whites, they are less able
to resist arrest and because authorities share common stereotypes
linking them with crime. Therefore, as the percentage of nonwhites
increases, the total arrest rate of a city should increase (power
hypothesis). Statistically, this is a compositional or an aggregate
effect.
The threat hypothesis underlying the recent work on police size
sug- gests that a high percentage of nonwhites produces an emergent
property, "perceived threat of crime," which increases arrest rates
through increas-
384
This content downloaded from 200.130.19.139 on Thu, 26 Mar 2015
21:31:45 UTCAll use subject to JSTOR Terms and Conditions
-
Crime Control
ing pressure on police to control crime. For, if increasing the
size of the crime control apparatus is one response to a perceived
threat of crime, then pressuring the existing police apparatus to
control crime would seem to be an equally logical response. Hence,
the threat hypothesis suggests that the percentage of nonwhites
affects the total arrest rate because it affects the actual arrest
rate of nonwhites and perhaps of whites as well.
The conflict perspective also suggests that the percentage of
nonwhites may affect arrest rates by constraining the racial
composition of crime. For nonwhites, as for all social categories,
as their percentage in the population increases, the proportion of
their in-group to out-group in- teraction in everything from
marriage to crime increases (Blau 1977); so, as the percentage of
nonwhites increases, the ratio of intra- to interracial crimes
increases for nonwhite offenders and decreases for white offend-
ers. For nonwhite offenders intraracial crime may have a lower
arrest rate than interracial crime. It may be viewed by both
victims and police as more a personal and family problem than a
matter requiring official intervention. Nonwhite victims may be
less prone to report it to the police, and when they do report it,
they frequently may be unable to legitimate their complaint as a
crime and to pressure police to allocate resources to resolve it.
Thus, the percentage of nonwhites may negatively affect their
arrest rate (the benign-neglect hypothesis). For white offend- ers,
because intraracial crime involves higher- as well as equal-status
victims, the implications of the ratio of intra- to interracial
crime on arrest rates is not so clear.
In sum, the conflict perspective suggests three related but
distinct causal processes underlying the effect of the percentage
of nonwhites on arrest rates. The net effect of the percentage of
nonwhites on the arrest rate depends on the relative strengths of
these causal processes.
SEGREGATION Spitzer (1975) argues that the segregation of
problematic groups into urban ghettos functions as a vehicle of
social control, thereby reducing the need for a large crime control
apparatus; and Liska et al.'s (1981) research shows a substantial
inverse relationship between racial segrega- tion and police size
per capita in major U.S. cities. Extending this thesis to the
actual volume of crime control is quite straightforward because the
threat and benign-neglect hypotheses make the same prediction: an
in- crease in the segregation of problematic populations
(nonwhites) de- creases the arrest rate. However, the hypotheses
assume that different causal processes underlie the effect. The
threat hypothesis suggests that, by reducing the threat of crime
perceived by authorities, the segregation of nonwhites reduces the
pressure on police to control crime, thereby
385
This content downloaded from 200.130.19.139 on Thu, 26 Mar 2015
21:31:45 UTCAll use subject to JSTOR Terms and Conditions
-
American Journal of Sociology
decreasing the arrest rate, especially that of nonwhites. The
benign- neglect hypothesis suggests that, by increasing the ratio
of intra- to inter- racial crime for nonwhite offenders, the
segregation of nonwhites de- creases the pressure on police to
control crime, thereby decreasing the arrest rate, especially that
of nonwhites.
ECONOMIC INEQUALITY Conflict theorists argue that economic
stratification or inequality accen- tuates economic conflict, which
in turn increases the threat perceived by the dominant class, and
that the greater the conflict, economic stratification, and
perceived threat, the greater the disposition of the dominant class
to use coercion to maintain a social order favorable to its
interests (Chambliss and Seidman 1982). Therefore, the greater the
in- come inequality, the greater the level of crime control.
We have been able to locate only a few studies bearing directly
on this proposition. Jacobs (1979) reports a substantial
relationship between in- come inequality and police size for major
SMSAs; however, Loftin et al. (1981), in a study of cities, are
unable to replicate these findings. As to the actual volume of
crime control, for states Jacobs (1978) reports a relation- ship
between income inequality and the certainty of punishment, and for
SMSAs Williams and Drake (1980) find a relationship between income
inequality and arrest rates. These findings must be viewed with
caution, because other important variables, such as reported crime
rates and racial composition, are not controlled. Our research
extends this research by examining the effect of income inequality
on arrest rates, with racial composition and reported crime rates
controlled.
In sum, in an extension of past work on police size, our
research examines the effect of racial/economic composition on
actual crime con- trol (arrest rates) as it operates directly and
indirectly through its effect on the capacity for crime control
(police size).
PROCEDURES The sample is part of a sample of 109 cities,
originally selected because their residential segregation levels
have been calculated since 1940. Arrest data are available for 76
of these cities, approximately one-half of all cities over 100,000
in population.
Crime Control The volume of crime control is measured by arrest
rates (ratio of arrests to population) for both property (robbery,
larceny, burglary, and auto theft)
386
This content downloaded from 200.130.19.139 on Thu, 26 Mar 2015
21:31:45 UTCAll use subject to JSTOR Terms and Conditions
-
Crime Control
and personal (homicide, aggravated assault, and rape) index
crimes. Put- ting these arrests into personal and property
categories minimizes incon- sistencies in classification among
police departments. For example, a rape could be misclassified as
an assault but is unlikely to be misclassified as a larceny.
Capacity for Crime Control Capacity is measured by the number of
police employees per capita (Fed- eral Bureau of Investigation
1972). Civilian clerical employees are in- cluded in the measure
because they free uniformed personnel for the immediate task of
crime control.
Racial/Economic Composition Racial residential segregation is
measured by the Dissimilarity Index, which describes the extent to
which the racial composition of city blocks reflects the racial
composition of the city as a whole (Sorenson, Taeuber, and
Hollingsworth 1975). Income inequality is measured by the Gini
index, which expresses the average difference in income between all
pairs of individuals in a city relative to the average income of
that city (Blau 1977).
Control Variables Population size, the percentage of poor people
(measured as the percent- age of families below the "poverty
line"), and reported crime rates are included as "control"
variables. Urbanism theory suggests that a large population is
associated with increased reliance on formal means of social
control, a traditional Marxian perspective suggests that the poor
are least able to resist arrest, and deterrence theory implies that
reported crime rates affect crime control. Since theory and
research also suggest that these variables are related to the
dimensions of racial/economic composi- tion discussed here, their
omission from the analysis might bias estimates of the effects of
these dimensions on arrest rates.
Analysis Regression analysis is used to estimate the direct
effect of the causal variables on arrest rates and their indirect
effect on them through their influence on police size. Because of
the time lag involved in budget deci- sions, police size is
measured two years later (1972) than the other causal variables
(1970). The 1972 budget reflects decisions made sometime in
387
This content downloaded from 200.130.19.139 on Thu, 26 Mar 2015
21:31:45 UTCAll use subject to JSTOR Terms and Conditions
-
American Journal of Sociology
late 1971, which can only be based on information (reported
crime rates) for 1970 or before. To maintain the logical temporal
order between police size and arrest rates, the latter are also
measured in 1972.
The equations were first estimated using cross-sectional OLS.
Yet, even with the two-year time lag between arrest rates and
reported crime rates, this estimation method does not disentangle
their effects. If arrest rates decrease crime rates, a lag measure
of crime rates (1970) may be affected by past levels of arrest
rates, and cross-sectional OLS may yield biased estimates of the
effect of reported crime rates (1970) on arrest rates (1972).
Therefore, we also estimated the equations using dynamic (cross-
lag) OLS and simultaneous equation (2SLS) methods. The dynamic
(cross-lag) analysis estimates the effect of the causal variables
(1970) on arrest rates (1972) with a lag measure of arrest rates
(1967) in the equa- tion, thereby controlling for the effects of
arrest rates on reported crime rates. The 2 SLS analysis uses
instrumental variables to yield a measure of reported crime rates
(1970) purged of the effect of arrest rates. Exploiting the panel
design of this research, the analysis uses a lag measure of
reported crime rate (1965) as an instrument, thereby circumventing
the locating of suitable instruments, a major problem in this
research area (Kessler and Greenberg 1981). Because all three
estimation methods yield extremely similar estimates, only the
cross-sectional OLS estimates are reported in the tables.
RESULTS The analysis proceeds as follows. First, because the
conflict perspective suggests that the income inequality effect is
contingent on the type of crime (personal or property), the
equations are estimated separately for property and personal arrest
rates. Second, to isolate the contextual from the aggregate effect
of percentage of nonwhites, the property and per- sonal arrest rate
equations are estimated separately for whites and non- whites.
Third, in order to examine explicitly the extent to which the
benign-neglect process mediates the effects of segregation and the
per- centage of nonwhites, the robbery equation is estimated for a
subsample of cities for which inter- and intraracial robbery rates
can be measured.
Property and Personal Arrest Rates Property and personal arrest
rates are normally distributed with means of 11.0 and 1.8, and
standard deviations of 5.5 and 1.1, respectively, per 1,000
population. Clearly, the chance of being arrested for an index
crime is significantly higher in some cities than in others.
Table 1 presents the cross-sectional OLS estimates of the
effects of the
388
This content downloaded from 200.130.19.139 on Thu, 26 Mar 2015
21:31:45 UTCAll use subject to JSTOR Terms and Conditions
-
Crime Control
TABLE 1
CROSS-SECTIONAL ESTIMATES OF TOTAL ARREST RATES
PROPERTY PERSONAL
v3 B v B
Populationa ........... .06 .064 - .05 - .010 Police sizea
............. .18 - .986 .15 .163 Poor (%) ............... -.16
-.259 -.03 -.010 Crime ratea ............ .29 .112** .34 .218**
Income inequality ....... .25 43.59** .17 6.09* Segregation
............ - .30 - . 180** - .28 - .034** Nonwhite (%) ..........
.34 .136** .35 .028**
R2 . . . ..37 .54
a B's are expressed in population units per 100,000, in police
per 1,000 population, and in crimes per 1,000 population.
* 1.5 times SE. ** 2.0 times SE.
causal variables on arrest rates. The first and second columns
present the standardized (,B's) and unstandardized (B's)
coefficients for property ar- rest rates, and the third and fourth
columns present those for personal arrest rates. The correlation
matrix was examined for evidence of mul- ticollinearity. Only a few
correlations exceed .50, the correlation of .61 between the
percentage of nonwhites and of poor people being the strongest.
When we reestimated the equations deleting percentage of poor
people, the findings did not change significantly from those
presented in table 1. The effect of each variable on arrest rates
was also estimated by the decrease that occurs in the R2 when it is
deleted from the equation; the findings rank the causal variables
in the same relative order as do the ,B's in table 1.
The results are clear. The cross-sectional, dynamic, and
simultaneous equation estimates all show that only income
inequality, segregation, percentage nonwhite, and reported crime
rates substantially affect arrest rates. Consistent with the
conflict perspective, the effect of income in- equality is somewhat
stronger for property arrest rates than personal ones, whereas the
effects of the other causal variables are about equal for both
arrest rates, and the effects of income inequality, segregation,
and the percentage of nonwhites are clearly independent of the
effect of re- ported crime rates. Interestingly, police size does
not positively affect arrest rates; there are therefore no indirect
effects of racial/economic com- position on arrest rates through
police size, although our research, like past research, shows that
racial/economic composition does affect police size.
389
This content downloaded from 200.130.19.139 on Thu, 26 Mar 2015
21:31:45 UTCAll use subject to JSTOR Terms and Conditions
-
American Journal of Sociology
Racial Subsamples The observed positive effect of the percentage
of nonwhites is difficult to interpret. It may be because nonwhites
have a higher arrest rate than whites and therefore, as the
percentage of nonwhites increases, the total arrest rate increases
(power hypothesis); or it may be because the percent- age of
nonwhites positively influences not only the arrest rate of non-
whites but even that of whites (threat hypothesis). This is a
special case of the classic problem of isolating compositional and
contextual effects. The contextual effect is examined here by
estimating the effect of the percent- age of nonwhites on arrest
rates of whites and nonwhites separately, thereby controlling for
compositional effects.
The equations were again estimated using cross-sectional,
dynamic, and simultaneous equations methods. As before, because the
estimates are so similar, we report only the cross-sectional
estimates here (table 2). As we expected theoretically, the effects
of population size, reported crime rates, income inequality, and
the percentage of poor people do not differ significantly among the
racial subsamples, although the effect of police size becomes
inconsistent, dependent on the type of crime and race. The focus of
the racial subsample analysis, however, is on the effect of the
percentage of nonwhites. The threat hypothesis suggests that it
should be positive, but the benign-neglect hypothesis suggests that
it should be negative; both hypotheses suggest that the effect,
positive or negative, should be stronger for nonwhites than whites.
The estimates support the benign-neglect hypothesis, showing that
it is negative and substantially and statistically significant only
for nonwhites. In fact, the percentage of nonwhites is the most
important variable in the nonwhite equations and about the least
important variable in the white equations.
The benign-neglect hypothesis further suggests that the effect
of the percentage of nonwhites should be stronger for personal
crimes, where victims can frequently identify the race of the
offender, than for property crimes, where they cannot. However, the
P's (table 2) provide little sup- port for this hypothesis. This
may occur because robbery, where victims can identify the race of
the offender, is classified as a property crime because it is
directed toward property rather than people, which is a major
distinction in the income inequality hypothesis. To address this
issue further, the equations for nonwhites were estimated for each
specific arrest rate. The findings are consistent with the
benign-neglect hy- pothesis.2 For auto theft, burglary, and
larceny, where victims cannot
2 For rape, where the victim can also identify the race of the
offender, the ,( is only - .11. We are suspicious of this finding,
because the total model explains only 3% of the variance in the
arrest rate, whereas for the other crimes it explains from 18% to
47%.
390
This content downloaded from 200.130.19.139 on Thu, 26 Mar 2015
21:31:45 UTCAll use subject to JSTOR Terms and Conditions
-
* ****
r- OC -0 0o-00
o o
N00 0
0)
st 00 0000 t -
z z ******
00 000 cn J in 0o t~- 0E 00 C. (c. C.
CN CY in C14 r-e X 0000 )0 c00
04 ~ ~ -
0) 0 0)~~~~~~~~~~~~C
Cd ~~~~~~~0
m t - N so N oo o - C
X~~~~~~~0 0 o o - t O 0 000 in t- O - O
ot 00 CDt CD 0 CD C
o 00'4'O 0
-Q 0
Ps~~~~~~. .n 0) 0~~~~~~~~~~~C
. ~f. -\. , C . 0=
z - - fi -s 49 : e E E 2 0~~~~~~~~~~~~aU 0)) *~~~C *d * o * * *
*~~~~~~~~c
400c.)0t0 \00000.-
U ~ 00~0t~-00
This content downloaded from 200.130.19.139 on Thu, 26 Mar 2015
21:31:45 UTCAll use subject to JSTOR Terms and Conditions
-
American Journal of Sociology
generally identify the race of the offender, the P3's for the
percentage of nonwhites are - .19, - .21, and - .35, respectively;
for assault and rob- bery, where victims generally can identify the
race of the offender, the P3's are -.41 and -.47, respectively; and
for homicide, where victims (be- fore they die) and witnesses can
sometimes identify the race of the of- fender, the ,3is -.41.
National Crime Survey Subsample The benign-neglect hypothesis
assumes (1) that the percentage of non- whites and segregation
decrease the arrest rate of nonwhites because for nonwhites these
variables increase the ratio of intra- to interracial crimes and
(2) that nonwhites are arrested less often for intraracial than for
interracial crimes. Unfortunately, this process cannot be directly
tested with official crime and arrest data because the race of
offenders and victim is not recorded. This has been reported for
the 26 cities of the National Crime Survey (NCS), 25 of which are
included in our sample. The NCS interviewed about 20,000 households
in each city from 1972 to 1974 (approximately the years during
which our data were collected) and asked the crime victims to
identify the race of the offender. This is possi- ble for robbery,
rape, and assault. From this data intraracial and interra- cial
crime rates and the proportion of intra- to interracial crime can
be estimated. We examined only robbery for nonwhites because the
robbery data are thought to be much more reliable than the rape and
assault data (Hindelang 1978) and because the proportion of robbery
which is intrara- cial for white offenders is too small to affect
white arrest rates significantly.
Because this subsample consists of only 25 cities, the limited
degrees of freedom require that the analysis be limited to the
following crucial vari- ables: intraracial and interracial robbery
rates, the percentage of non- whites, and segregation. Using OLS,
we first estimated the extent to which the percentage of nonwhites
and segregation influence the racial composition of robbery. In
support of the benign-neglect hypothesis, the ,3 estimates show
that both moderately increase the intraracial robbery rate (.27 and
.26, respectively) and strongly decrease the interracial rob- bery
rate (- .63 and - .39, respectively), thereby substantially
increasing 3 It can be argued that there is a definitional negative
correlation between the percent- age of nonwhites and the nonwhite
arrest rate. Contemporary opinion seems to be that this is not the
problem it was once thought to be, especially when, as is the case
here, the ratios are theoretical variables, not just constructions
for the purpose of standardi- zation (Long 1979). In addition, a
definitional correlation cannot account for the fact that the
effect of the percentage of nonwhites is strongest for those crimes
where the victim can identify the race of the offender-exactly
where the benign-neglect hy- pothesis suggests that it should be
strongest.
392
This content downloaded from 200.130.19.139 on Thu, 26 Mar 2015
21:31:45 UTCAll use subject to JSTOR Terms and Conditions
-
Crime Control
the ratio of intra- to interracial crime for nonwhite offenders.
Next, we estimated the effects of intraracial robbery, interracial
robbery, segrega- tion, and the percentage of nonwhites on the
robbery arrest rate. Again, the ,3 estimates support the
benign-neglect hypothesis. The effect of inter- racial robbery
(.47) is four times stronger than the effect of intraracial robbery
(.12), and a substantial portion of the total negative effect of
both segregation and percentage of nonwhites on the robbery arrest
rate is mediated through the interracial robbery rate, reflecting
the benign- neglect process. Generally, the findings suggest that
the percentage of nonwhites is important mostly because it
constrains the racial composi- tion of robbery, as its indirect
effect through interracial robbery (-.30) is about double its
positive direct effect (. 16); and that segregation is impor- tant
mostly because it reduces the perceived threat of crime, as its
direct effect (- .32) is about double its indirect effect (-.18)
through interracial robbery.
Of course, these findings must be viewed with caution. Because
the subsample is small, the analysis is limited to cross-sectional
OLS esti- mates of equations including only a few variables. Yet
the findings pro- vide consistent support for the benign-neglect
interpretation of the findings for the full sample.
DISCUSSION This paper, drawing on the conflict perspective and
extending studies of police size, examines the effect of
racial/economic composition on arrest rates directly and indirectly
through its effect on police size.
The analysis shows that the police size effect is inconsistent,
varying by type of crime and race, and that it is frequently
insignificant substantively and statistically. An explanation for
the findings may lie in the allocation of police resources.
Organizational theory suggests that, as police depart- ments grow,
a greater proportion of their personnel is absorbed by admin-
istrative duties, making a lesser proportion directly available for
crime control. In fact, Wilson and Boland (1978) report that police
size explains only 23% of the variance in the number of police
units on the street. Whatever the explanation, in this paper it is
important to emphasize that the observed effect of racial/economic
composition on arrest rates is not mediated by police size.
Generally, the analysis provides support for the conflict
perspective on crime control, showing that the racial/economic
composition of cities sub- stantially affects arrest rates.
Economic inequality moderately affects arrest rates for both
whites and nonwhites for both personal and property crimes. The
findings seem to support the conflict perspective thesis that
income inequality accentuates
393
This content downloaded from 200.130.19.139 on Thu, 26 Mar 2015
21:31:45 UTCAll use subject to JSTOR Terms and Conditions
-
American Journal of Sociology
economic conflict and the greater the economic inequality and
conflict, the greater the disposition of dominant groups to use
coercion to maintain a social order favorable to their interests.
Yet, more research is required to document that this process
underlies the income inequality effect, be- cause, although the
formulation and enforcement of numerous laws may be explained in
terms of ruling class interests (Chambliss and Seidman 1982), the
control of urban street crime seems to be in the interests of all
economic classes. Indeed, it may be even more in the interests of
the blue- collar and lower-middle than of the upper class, which
can afford to re- side in relatively safe neighborhoods, locate
businesses in these neighbor- hoods, and insure its property
against theft.
A segregation effect is clearly evident for both races for
property and personal arrests, supporting both the threat and
benign-neglect hypoth- eses of the conflict perspective. The
analysis of the NCS subsample pro- vides further support for the
benign-neglect hypothesis, showing that a substantial portion of
the segregation effect on nonwhite arrest rates is mediated through
its effect on the racial composition of crime. Yet, a substantial
direct effect remains, suggesting that other processes, such as
perceived threat, may also be operating.
The effect of the percentage of nonwhites is a function of both
composi- tional and contextual effects. Although the total effect
cannot be precisely partitioned into these effects without data on
both individual and social units, the analysis of the arrest rate
disaggregated by race suggests a negative contextual effect for
nonwhites; as the percentage of nonwhites increases, the arrest
rate of nonwhites decreases. Note that this theoreti- cally
significant contextual effect is obscured when the total arrest
rate is not disaggregated by race. Because nonwhite arrest rates
are so much higher than white arrest rates (five times higher for
property and six for personal), an increase in the percentage of
nonwhites increases the total arrest rate, even though it actually
substantially decreases the nonwhite arrest rate. The contextual
effect of the percentage of nonwhites supports the benign-neglect
hypothesis. Additional support for this process is given in the
analysis of NCS cities, which shows that the negative effect of the
percentage of nonwhites on the nonwhite robbery arrest rate is
completely mediated through its effect on the racial composition of
robbery.
In sum, arrest rates are a significant social fact, varying
considerably among U.S. cities. Our analysis shows that this
variation reflects to a large extent the economic/racial
composition of cities, independently of police size and reported
crime rates, thus providing support for the con- flict perspective.
By disaggregating the sample by race and type of crime and by
examining a subsample of cities where the racial composition of
robbery can be measured, the analysis further examines power,
threat,
394
This content downloaded from 200.130.19.139 on Thu, 26 Mar 2015
21:31:45 UTCAll use subject to JSTOR Terms and Conditions
-
Crime Control
and benign-neglect processes assumed to underlie the effect of
economic/ racial composition on arrest rates.
REFERENCES
Blau, Peter M. 1977. Inequality and Heterogeneity. New York:
Free Press. Chambliss, William J., and Robert B. Seidman. 1982.
Law, Order and Power. Read-
ing, Mass.: Addison-Wesley. Federal Bureau of Investigation.
1972. Uniform Crime Reportsfor the United States.
Washington, D.C.: Government Printing Office. Hindelang, Michael
J. 1978. "Race and Involvement in Crimes." American Sociolog-
ical Review 43 (February): 93-109. Jackson, Pamela Irving, and
Leo Carroll. 1981. "Race and the War on Crime."
American Sociological Review 46 June): 290-305. Jacobs, David.
1979. "Inequality and Police Strength." American Sociological
Review
44:913-24. . 1978. "Inequality and the Legal Order: An
Ecological Test of the Conflict
Model." Social Problems 25 (June): 5 15-25. Kessler, Ronald C.,
and David F. Greenberg. 1981. Linear Panel Analysis: Models of
Quantitative Change. New York: Academic Press. Liska, Allen E.,
Joseph J. Lawrence, and Michael Benson. 1981. "Perspectives on
the
Legal Order: The Capacity for Social Control." American Journal
of Sociology 87 (September): 413-26.
Lizotte, Alan J., and David J. Bordua. 1980. "Firearms Ownership
for Sport and Protection." American Sociological Review 45 (April):
229-43.
Loftin, Colin, David F. Greenberg, and Ronald C. Kessler. 1981.
"Income Inequality, Race, Crime, and Crime Control." Paper
presented at the annual meeting of the American Society of
Criminology.
Long, Susan B. 1980. "The Continuing Debate over the Use of
Ratio Variable: Fact and Fiction." Pp. 37-67 in Sociological
Methodology, edited by Karl F. Schuessler. San Francisco:
Jossey-Bass.
S0renson, Annemette, Karl E. Taeuber, and Leslie J.
Hollingsworth, Jr. 1975. "In- dexes of Racial Residential
Segregation for 109 Cities in the United States, 1940- 1970."
Sociological Focus 8:125-42.
Spitzer, Steven. 1975. "Toward a Marxian Theory of Deviance."
Social Problems 22 (June): 638-51.
Swigert, Victoria L., and Ronald A. Farrell. 1976. Murder,
Inequality and the Law: Differential Treatment in the Legal
Process. Lexington, Mass.: Heath.
Turk, Austin T. 1969. Criminality and Legal Order. Chicago: Rand
McNally. Williams, Kirk and Susan Drake. 1980. "Social Structure,
Crime and Criminalization:
An Empirical Examination of the Conflict Perspective."
Sociological Quarterly 21 (Autumn): 563-76.
Wilson, James Q., and Barbara Boland. 1978. "The Effect of the
Police on Crime." Law and Society Review 12 (Spring): 366-90.
395
This content downloaded from 200.130.19.139 on Thu, 26 Mar 2015
21:31:45 UTCAll use subject to JSTOR Terms and Conditions
Article
Contentsp.383p.384p.385p.386p.387p.388p.389p.390p.[391]p.392p.393p.394p.395
Issue Table of ContentsAmerican Journal of Sociology, Vol. 90,
No. 2, Sep., 1984Front MatterKarl Marx and the Satanic Mills:
Factory Politics Under Early Capitalism in England, the United
States, and Russia [pp.247-282]The Cumulative Texture of Local
Urban Culture [pp.283-304]Black Family Formation and Tenancy in the
Farm South, 1900 [pp.305-325]The Transition from Youth to Adult:
Understanding the Age Pattern of Employment [pp.326-358]Trends in
Parental Socialization Values: Detroit, 1958-1983
[pp.359-382]Research NotesSocial Structure and Crime Control Among
Macrosocial Units [pp.383-395]The Dynamics of Self-Esteem and
Delinquency [pp.396-410]Military Keynesianism in the United States,
1949-1976: Disaggregating Military Expenditures and Their
Determination [pp.411-417]
Commentary and DebateComment on Denzin's "Note on Emotionality,
Self, and Interaction" [pp.418-422]Reply to Baldwin
[pp.422-427]Comment on Barrie S. Morgan's "An Alternate Approach to
the Development of a Distance-Based Measure of Racial Segregation"
[pp.427-428]The Utility of Distance-Based Segregation Indexes:
Reply to Mitra [pp.428-429]A Defense of Rational Sociological
Discourse [pp.429-431]What Is Rational, What Is Not: Reply to Mnch
[pp.432-434]
Review EssayCenturies of Death and Dying [pp.435-439]
Book Reviewsuntitled [pp.440-442]untitled [pp.443-444]untitled
[pp.444-445]untitled [pp.445-447]untitled [pp.448-449]untitled
[pp.450-453]untitled [pp.453-455]untitled [pp.455-457]untitled
[pp.457-458]untitled [pp.459-461]untitled [pp.461-463]untitled
[pp.463-464]untitled [pp.465-466]untitled [pp.466-469]untitled
[pp.469-471]untitled [pp.472-474]untitled [pp.474-475]untitled
[pp.475-477]untitled [pp.477-480]untitled [pp.480-482]
Back Matter