criticize this article

Conflict theory and racial profiling: An empirical analysis

of police traffic stop data

Matthew Petrocellia, Alex R. Piquerob, Michael R. Smithc,*

aDepartment of Sociology and Criminal Justice, Southern Illinois University, Edwardsville, IL 62026, USA bCenter for Studies in Criminology and Law, University of Florida, 201 Walker Hall, P.O. Box 115950,

Gainesville, FL 32611-5950, USA cCriminal Justice Program, Washington State University, Spokane, 668 North Riverpoint Boulevard, Box B,

Spokane, WA 99202-1662, USA

Abstract

Using data collected by the Richmond, Virginia Police Department, this article applies conflict theory to police

traffic stop practices. In particular, it explores whether police traffic stop, search, and arrest practices differ

according to racial or socioeconomic factors among neighborhoods. Three principal findings emanate from this

research. First, the total number of stops by Richmond police was determined solely by the crime rate of the

neighborhood. Second, the percentage of stops that resulted in a search was determined by the percentage of

Black population. Third, when examining the percentage of stops that ended in an arrest/summons, the analyses

suggest that both the percentage of Black population and the area crime rate served to decrease the percentage of

police stops that ended in an arrest/summons. Implications for conflict theory and police decision-making are

addressed.

D 2002 Elsevier Science Ltd. All rights reserved.

Introduction

Conflict theory holds that law and the mecha-

nisms of its enforcement are used by dominant

groups in society to minimize threats to their interests

posed by those whom they label as dangerous,

especially minorities and the poor. Over the past

several years, racial profiling by police has become

an issue of national significance. In his first speech to

Congress on February 27, 2001, President Bush

addressed racial profiling and directed Attorney Gen-

eral John Ashcroft to develop a set of recommenda-

tions to end racial profiling by America’s police

forces. Although empirical data on racial profiling

is scarce (Government Accounting Office, 2000),

conflict theory suggests that police may indeed target

minorities when conducting traffic stops or field

interrogations.

Using data collected by the Richmond, Virginia

Police Department, this article tests the application of

conflict theory to police traffic stop practices. In

particular, it explores whether police traffic stop,

search, and arrest practices differ according to racial

or socioeconomic factors among neighborhoods (e.g.,

Smith, 1986). Previous research on conflict theory

and the police used data from multiple cities or states

to examine differences in minority treatment by the

police at the macro level. This article extends the

current research by presenting a micro-level analysis

of police practices using census tract data from a

single city. It begins with a discussion of prior

research on conflict theory and its relationship to

racial profiling.

0047-2352/02/$ – see front matter D 2002 Elsevier Science Ltd. All rights reserved.

PII: S0047 -2352 (02 )00195 -2

* Corresponding author. Tel.: +1-509-358-7711.

E-mail address: [email protected] (M.R. Smith).

Journal of Criminal Justice 31 (2003) 1–11

Conflict theory and racial profiling

Conflict theory

According to Simmel (1950), conflict is a fun-

damental social process. As such, society is largely

molded and shaped by the competing interests of

social groups who vie for dominance in order to enact

or maintain a social structure most beneficial to them.

Conflict theory asserts that the relative power of a

given social group dictates social order in that power-

ful groups not only control the lawmakers, but also

the law enforcement apparatus of the state. In

essence, laws are made which serve the interests of

the privileged and the police are used to suppress and

control any segment of society that poses a threat to

the status quo (Black, 1976; Dahrendorf, 1959;

Quinney, 1970; Turk, 1969; Vold, 1958).

The notion of ‘‘threat’’ underlies the conflict

perspective. In a capitalist society where economic

resources equate to power, it is in the interest of the

ascendant class to maintain economic stratification in

order to dictate the legal order (Chambliss, 1976;

Chambliss & Seidman, 1971; Quinney, 1974, 1975;

Taylor, Walton, & Young, 1973). According to Quin-

ney (1974, p. 24), ‘‘the dominant economic class,

through its use of the legal system, is able to pressure

a domestic order that allows its interests to be

promoted and maintained.’’ Indeed, economic strati-

fication is so important to the vitality of the advan-

taged that they will pressure legislators to enact

repressive measures intended to control groups con-

sidered volatile and threatening (Tagaki, 1974). The

larger the gap in economic disparity, the more pro-

nounced the dynamic, and as ‘‘the more economically

stratified a society becomes, the more it becomes

necessary for dominant groups to enforce through

coercion the norms of conduct that guarantee suprem-

acy’’ (Chambliss & Seidman, 1980, p. 33).

Turk (1969) also claims that culturally dissimilar

groups are viewed as threats to the existing social

order. More specifically, racial minorities are consid-

ered a threat to the dominant class in the United

States (Blalock, 1967; Quinney, 1970; Turk, 1969).

Swigert and Farrell (1976) report that Whites main-

tain criminal stereotypes about non-Whites. This is

compounded by the fact that Whites perceive the

proportion of non-Whites in their communities as an

indicator of a crime problem and that interracial

victimization is considered particularly threatening

to Whites as compared to non-Whites (Lizotte &

Bordua, 1980). Using data from the National Crime

Survey, Liska, Lawrence, and Sanchirico (1982)

demonstrated that fear of crime was related to the

percentage of African-Americans in cities, while

Chiricos, Hogan, and Gertz (1997) linked fear of

crime among Whites to the perception that they were

the racial minority in their neighborhood.

Hence, economic and racial minorities are seen as

a threat to the ruling class. Conflict theory maintains

that the privileged, acting on the perception of threat,

will use the crime control apparatus of the state to

restrain and limit those who threaten their interests.

Practically, this means that one should expect more

aggressive law enforcement practices in areas with

greater percentages of poor and non-White citizens.

This ‘‘threat hypothesis’’ has been tested in several

arenas of American policing.

Conflict theory and the police

Chambliss and Seidman (1971, p. 269) summarize

the process of law enforcement, from a conflict/

Marxian point of view, with six propositions:

1. The agencies of law enforcement are bureaucratic

organizations.

2. An organization and its members tend to substi-

tute for the official goals and norms of the

organization ongoing policies and activities that

will maximize rewards and minimize the strains

on the organization.

3. This goal substitution is made possible by:

(a) The absence of motivation on the part of the

role-occupants to resist pressures toward goal-

substitution;

(b) The pervasiveness of discretionary choices

permitted by the substantive criminal law

and the norms defining the roles of the mem-

bers of law enforcement agencies; and

(c) The absence of effective sanctions for the

norms defining the roles in those agencies.

4. Law enforcement agencies depend on political

organizations for resource allocation.

5. Organizations will minimize strains on themselves

by processing those who are politically weak and

powerless, while refraining from processing those

who are politically powerful.

6. Therefore, it may be expected that law enforce-

ment agencies will process a disproportionately

high number of the politically weak and power-

less, while ignoring the violations of those with

power.

Researchers have been particularly interested in

testing conflict theory against police practices

because of the unique status police hold in society.

If law can be seen as the nails that hold society

together, then police can certainly be viewed as the

hammer of the state. In their analysis of the U.S.

police in The Iron Fist and the Velvet Glove, the

Center for Research on Criminal Justice (1975) goes

M. Petrocelli et al. / Journal of Criminal Justice 31 (2003) 1–112

so far as to suggest that the police often-times

function with an ‘‘iron fist.’’ The unique position of

the police is troubling from the conflict perspective

because they are seen as ‘‘agents of the ascendant

class’’ (Sorenson, Marquart, & Brock, 1993, p. 418)

who are, by oath, required to enforce laws that only

serve the privileged at the expense of the underclass.

In effect, they are seen as armies of occupation doing

the ‘‘dirty work’’ of a rather insidious system

designed to protect the wealthy and the White (Blau-

ner, 1972; Chamlin, 1989). These claims have been

empirically tested as a function of police resources,

arrests, homicides by police, the killing of police

officers by civilians, and civil rights complaints

against police.

Police resources

Jacobs (1979) tested the conflict proposition that

law enforcement personnel should be most numerous

in metropolitan areas where differences in economic

resources are greatest. Using census tract data in large

metropolitan areas, he found that economic inequality

was correlated with police strength (measured as the

total number of police and other law enforcement

personnel). Similarly, Liska, Lawrence, and Benson

(1981) explored the possibility that the size of a

police agency was driven by the perceived threat of

the dominant class by analyzing 109 U.S. municipal

police departments from 1950 to 1972. Testing the

assertion that relatively small, culturally dissimilar

groups may not be perceived as posing much of a

threat while a relatively large culturally dissimilar

group (composing 20–30 percent of the population),

which is also racially dissimilar, may be perceived as

a substantial threat to the social order, they found that

the percentage of non-White increases in the popu-

lation substantially influenced police size. This effect

was particularly apparent in the South, just after the

advent of the civil disorders of the 1960s. These

findings are consistent with conflict theory, as both

the actual number of non-Whites and the increased

perception of threat (meaning civil rights activism)

impacted the size of the crime control apparatus.

Interestingly, the effects of segregation were also

tested vis-a-vis the conflict perspective. Liska et al.

found that more racially segregated jurisdictions

experienced a decrease in police size. This finding

is explained through the work of Blauner (1972) and

Spitzer (1975), who argued that segregation acted as

a means of social control. When races are segregated,

the visibility of ‘‘threatening classes’’ to the White

majority is diminished, as is the actual incidence of

interracial crime. Thus, White perception of the

criminal threat is reduced and demands for greater

police protection wane.

Jackson and Carroll (1981) looked at the alloca-

tion of police resources through police expendi-

tures. Hypothesizing that more money would be

spent on law enforcement in cities with a high

concentration of non-Whites and minority political

activity, they found that the racial composition of a

city and the level of Black mobilization activity

were significant predictors of police expenditures.

More specifically, they found the size of the Black

population was a significant predictor of police

salaries and operational budgets, while Black polit-

ical mobilization was a significant predictor of

capital expenditures. Unlike Liska et al. (1981),

the frequency of riots in the 1960s did not predict

allocation of police resources.

Arrests

The threat hypothesis predicts that as the percent-

age of racial and economic minorities increases, the

perceived threat to the privileged class will intensify

leading to increased pressure to enact more stringent

crime control measures. As a result, the total number

of arrests should rise independent of the actual crime

rate. Liska and Chamlin (1984) tested the threat

hypothesis against arrest rates and found that income

inequality did indeed predict total arrests for both

property and personal crimes. Liska, Chamlin, and

Reed (1985) again tested this hypothesis using UCR

data, and found that consistent with conflict theory,

percent non-White, income inequality (as measured

by the Gini index), and a low level of segregation

increased the certainty of arrests for Whites and non-

Whites.

Homicides by the police

Proponents of the conflict model argue that the

use of lethal force by the police is influenced by the

racial and class standing of the perpetrator/victim

(Knoohuizen, Fahey, & Palmer, 1972; Tagaki,

1974). Tagaki (1974, p. 30) goes so far as to state

that police have a lower threshold of suspicion for

Blacks as opposed to Whites, which manifests itself

in the police having ‘‘one trigger finger for Whites,

another for Blacks.’’ Several empirical studies do

report that Blacks are disproportionately fired upon

and killed by police (for a full discussion of this body

of literature, see Binder & Scharf, 1982), although the

disparity has been decreasing steadily since the mid-

1980s (Walker, 1992).

Because the threat or use of violence is crucial

to maintaining class order, conflict theorists hypo-

thesize that state coercion, including lethal force,

will be greatest in jurisdictions with the greatest

inequalities. Jacobs and Britt (1979) first tested this

M. Petrocelli et al. / Journal of Criminal Justice 31 (2003) 1–11 3

assertion and reported that inequalities in the dis-

tribution of economic resources predicted the use of

lethal force by the police. Sorenson et al. (1993)

used UCR Supplemental Homicide Reports of the

largest U.S. cities from 1980 to 1984 to test the

threat hypothesis and also found economic inequal-

ities to be the most accurate predictor of felon

killings by police. They also found absolute pov-

erty (i.e., the percentage of citizens living below

the poverty line) and percent Black to be signific-

ant predictors of police caused homicides. Lastly,

Jacobs and O’Brien (1998) studied 170 cities and

concluded that economic stratification along racial

lines best predicts the use of deadly force by the

police, while cities with more Blacks and a recent

growth in the number of Blacks experience a

higher rate of police killings of Blacks.

Police killings by civilians

Because the literature does support the threat

hypothesis in the realm of police killings of civilians,

Chamlin (1989) hypothesized that states with higher

degrees of racial and economic inequality should also

exhibit a higher rate of police killed by civilians.

Specifically, he tested the proposition that increases

in the relative population size of threatening groups

(e.g., Blacks, Hispanics, and the poor) will increase

the level of antagonism between crime control agents

and civilians. Increased levels of antagonism are

predicted to make police–citizen encounters more

volatile and thereby increase the rate of police kill-

ings. Using multivariate analyses, he found that while

economic inequality had little effect on police kill-

ings, increases in the population of racial minorities

and the percent poor did positively impact the rate of

police killings.

Civil rights complaints

In the latest test of the threat hypothesis, Holmes

(2000) regressed the average number of civil rights

violations criminal complaints (alleging police bru-

tality) reported to the Department of Justice on

relevant independent variables to include city po-

pulation, index crime rate, percent Black, percent

Hispanic, region, and majority/minority income

inequality. Consistent with the conflict perspective,

he found that ‘‘threatening people’’ (i.e., percent

Black, percent Hispanic, and majority/minority

income inequality) were positively related to the

average number of civil rights criminal complaints,

while ‘‘threatening acts’’ (index crime rate) were not.

Using data from the Houston Police Department,

Kessler (1999) found that, although officers working

in areas where community policing had been imple-

mented received significantly fewer complaints than

officers working in other areas, the percentage Black

in the area was related to sustained violent, criminal,

conduct, and total complaints. Similarly, Lawton,

Hickman, Piquero, and Greene (2001) found that

complaints against Philadelphia police tended to be

higher in areas with high unemployed males as well

as areas with high female-headed households with

children.

Racial profiling and conflict theory

Research on the policing of certain classes of

people has generated a number of important insights.

Chambliss (1994, p. 177), for example, engaged in a

series of ride-a-longs in Washington, D.C. and

observed that the Rapid Deployment Unit, a unit

designed to target drugs and potential riots, seemed

to focus their efforts on the ‘‘urban ghetto,’’ an area

of Washington where 40 percent of the Black popu-

lation lives below the poverty level. The end result of

this selective targeting, at least according to Cham-

bliss, was a focus on young Black males, which

negatively affected families and education, created

moral panic, and swelled prison populations that were

comprised predominantly of minorities—especially

young Black males.

Despite the large body of literature on conflict

theory, racial profiling by police, per se, is a relatively

new topic for empirical inquiry; consequently,

research on racial profiling is limited. In one of the

largest and most sophisticated studies of racial pro-

filing to date, the New York Attorney General’s

Office (1999) analyzed more than 181,000 field

interrogation cards completed by NYPD officers from

1998 to 1999 and found that although Blacks com-

prised only 25.6 percent of New York City’s popu-

lation, they accounted for 50.6 percent of all persons

stopped by the NYPD. Hispanics were also over-

represented among persons stopped, while Whites

were significantly underrepresented.

Thus, even after controlling for the differential

rates at which minorities commit criminal offenses

within precincts (as measured by arrests), Blacks

(23 percent more) and Hispanics (39 percent more)

were still stopped more frequently than Whites

across all crime categories. These findings support

the work of Liska et al. (1985) who found that

cities with higher percentages of non-Whites pro-

duced higher arrest rates independent of crime

rates. In the case of New York, minority citizen

involvement in crime did not explain the rates at

which minorities were stopped by police relative to

Whites.

As the result of litigation over the allegedly

discriminatory traffic stop practices of New Jersey

M. Petrocelli et al. / Journal of Criminal Justice 31 (2003) 1–114

state troopers, the State of New Jersey undertook a

study of the stop and search activities of troopers

in two State Police districts. Examining the stops

that occurred from April 1997 through February

1999, and including most of 1996 and a few

months from 1994, a New Jersey Attorney Gen-

eral’s team found that 627 of the 87,489 traffic

stops involved a vehicle search. Of those searches,

77.2 percent involved Black or Hispanic motorists.

During a similar time period, only 33.9 percent of

the total traffic stops made in the two districts were

of Blacks and Hispanics (Office of the Attorney

General, 1999).

Under existing research findings, the disparities in

searches among minorities and Whites cannot be

explained by a difference in the probability that

minorities will be found in possession of contraband

or illegal weapons. For example, in his study of the

stop and search practices of the Maryland State

Police, Lamberth (1997) found that although only

17.5 percent of speeders along the I-95 corridor

through Maryland were Black (January 1995 through

September 1996), 72.9 percent of persons searched

were Black. Once searched, however, Blacks were no

more likely than Whites to be in possession of

contraband.

Likewise, Zingraff et al. (2000) analyzed 1998

traffic stop data from the North Carolina Highway

Patrol and found that African-Americans were

slightly more likely to be ticketed than Whites

when compared to their percentage among licensed

drivers in North Carolina. Moreover, they found

that Blacks were significantly more likely than

Whites to be searched even though they were

slightly less likely than Whites to be in possession

of contraband.

In sum, conflict theory asserts that representatives

of the dominant social class, such as police who

maintain social control, view minority citizens as

posing an increased risk of criminality (Blalock,

1967; Center for Research on Criminal Justice,

1975; Lizotte & Bordua, 1980; Quinney, 1970; Swi-

gert & Farrell, 1976; Turk, 1969). Research findings

that show lower minority ‘‘hit rates’’ (number of

successful searches/number of stops) compared to

Whites are indicative of this perspective. Consistent

with conflict theory, police perhaps are more likely to

search minority drivers because they erroneously

expect to find contraband more frequently among

the disadvantaged and minority class (e.g., Cham-

bliss, 1994).

The results from most reported racial profiling

studies indicate that minorities are stopped, searched,

and sometimes ticketed at rates that exceed those for

Whites when compared to some benchmark popu-

lation (GAO, 2000; Harris, 1999; Lamberth, 1997;

New York Attorney General’s Office, 1999; Office of

the Attorney General, 1999; San Diego Police

Department, 2000; San Jose Police Department,

1999; Zingraff et al., 2000). Using individual stops

as the unit of analysis, however, Smith and Petrocelli

(2001) found that although African-Americans in

Richmond, Virginia were stopped at rates that

exceeded their proportion in the driving-eligible

population, they were no more likely to be searched

than Whites and were actually less likely than Whites

to be ticketed or arrested. Moreover, race of the

officer did not predict the race of the motorist

stopped, nor did it predict whether a search or an

arrest took place.

Current focus

The analysis presented below builds on prior

research by focusing on how neighborhood context

may influence police behavior. This is particularly

important since much of the previous research on

police decision-making (i.e., traffic stops, arrests,

etc.) tended to focus on city- and state-level differ-

ences. One exception to this research focused on

identifying the manner in which neighborhood con-

text shaped police behavior.

Using data from the Police Services Study,

Smith (1986) studied five measures of police

behavior and eleven neighborhood characteristics

to test the neighborhood context hypothesis. He

found that suspects confronted in lower-status

(SES) neighborhoods incurred a higher risk of

being arrested, while those encountered in non-

White or racially mixed communities were more

apt to be handled coercively by police. In sub-

sequent analyses, Smith observed a significant

interaction between the suspect’s race and the racial

composition of the neighborhood in which coercive

confrontations occurred such that police were more

likely to exercise coercive authority toward Black

offenders in primarily Black neighborhoods. In fact,

Black suspects in White neighborhoods were

handled less coercively by police compared to

Black suspects in Black neighborhoods. In sum,

Smith concluded that police responded differently

depending on the type of neighborhood in which

encounters occurred such that police respond to

both ‘‘places and people.’’

Given recent evidence on the nature of police

behaviors across different social (state, city, neigh-

borhood) contexts, the current analysis examines the

extent to which police stop, search, and arrest deci-

sions are a function of neighborhood demographic

and socioeconomic characteristics in the City of

Richmond.

M. Petrocelli et al. / Journal of Criminal Justice 31 (2003) 1–11 5

Study site and methods

Richmond is the capital of Virginia and a city

of approximately -200,000 people. The Richmond

Police Department is comprised of 690 sworn

officers. Thirty percent of the officers are Black,

67 percent are White, and 3 percent are persons of

other races. Women make up 13 percent of the

department’s sworn personnel. The police depart-

ment serves a city whose population is 57.2 percent

Black, 38.3 percent White, and 2.6 percent His-

panic. In 1990, median household income in Rich-

mond was US$23,551, and 17 percent of families

lived below poverty level.1 Like many cities in the

south, Richmond has many racially segregated

neighborhoods where persons of one racial group

(usually Blacks or Whites) make up almost the

entire neighborhood population.

Traffic stop data were collected by the Richmond

Police Department from January 17, 2000 through

March 31, 2000. During traffic stops, officers

equipped with mobile data computers (MDCs) in

their cars recorded preselected information on the

driver and the stop itself directly into their com-

puters. Eventually, these data were matched with

demographic data from the officers. A full descrip-

tion of the data fields can be found in Appendix A.

During the ten-week data collection period, 6,699

traffic stops were recorded by the city’s computer-

aided dispatch system.2 Of those stops, the city

estimated that 179 (2.7 percent) were conducted

by officers not equipped with MDCs in their auto-

mobiles. Officers themselves entered data on 4,782

traffic stops, yielding a compliance rate with the

data collection protocol of 73 percent. These data,

along with census-tract level crime and demographic

information, serve as the basis for the findings

below (see Table 1).

Hypotheses

Consistent with conflict theory, the following

hypotheses are tested:

H1: Racial and socioeconomic variables will influ-

ence overall traffic stop rates within census tracts;

relatively speaking, Black and poor census tracts will

experience higher stop rates than predominantly

White and wealthy census tracts.

H2: Racial and socioeconomic variables will influ-

ence the percentage of traffic stops that result in a

search; relatively speaking, Black and poor neighbor-

hoods will experience a higher percentage of stops

resulting in searches than predominantly White and

wealthy neighborhoods.

H3: Racial and socioeconomic variables will influ-

ence the percentage of traffic stops that result in a

summons or arrest; relatively speaking, Black and

poor neighborhoods will experience a higher percent-

age of stops resulting in an arrest than predominantly

White and wealthy neighborhoods.

The primary analytic device used to test these

hypotheses was ordinary least squares regression.3

This technique is used to examine whether and to

Table 1

Description of variables

Measurement Mean S.D. Minimum Maximum Description

Dependent variables

Stop rate Interval 67.31 257.03 0.93 2,130.43 Stop rate per 1,000 population

Search Interval 8.65 7.50 0 36.92 Percent of stops resulting

in a search

Arrest Interval 61.74 14.44 38.30 100 Percent of stops resulting in

at least one arrest/summons

Independent variables

Black Interval 55.05 36.37 0.6 99.4 Percent Black population

Other Interval 1.45 2.03 0 14.86 Percent population of

non-Black minorities

Poverty Interval 18.74 17.84 1.00 80.80 Percent of families below

poverty

Unemployment Interval 8.78 10.62 0.90 76.90 Percent unemployed

Income Interval 38,879 28,223 7,718 209,274 Mean family income

Part I rate Interval 212.89 748.53 14.07 6,043.48 Part I crime rate per

1,000 population

M. Petrocelli et al. / Journal of Criminal Justice 31 (2003) 1–116

what degree racial and socioeconomic variables influ-

ence stop, search, and arrest practices within Rich-

mond’s seventy census tracts.

Results

Table 2 presents the results of a multiple regres-

sion predicting the overall stop rate by the police per

1,000 citizens. Only one of the six coefficients

exerted a significant effect on the total number of

stops: Part I crime rate. That is, police stops were

significantly higher in areas with higher crime rates.

It is also worth pointing out that neither percent Black

nor any of the socioeconomic characteristics were

significantly related to the total number of stops by

police.4

Next, the same set of independent variables was

employed to predict the percentage of total stops that

resulted in a search. As can be seen from Table 3,

only one of the six coefficients was significant.

Namely, the percent Black population was positively

and significantly related to the percentage of total

stops that resulted in a search. This result suggests

that searches were more prevalent in neighborhoods

that were characterized by a high percentage of

Blacks.5

The final model regresses the percentage of stops

involving at least one arrest/summons on the same six

independent variables. The results may be found in

Table 4. Two of the six coefficients attained signific-

ance in this model: percent Black population and Part

I crime rate per 1,000 citizens. The negative sign of

both coefficients requires comment. Notice that in

areas where there is a higher percentage of Black

population, the chance of stops involving at least one

arrest/summons is lower. Similarly, in high crime rate

areas, the chance of stop involving at least one arrest/

summons is lower.

Discussion

Under the backdrop of conflict theory, this article

set out to examine how police decisions to stop,

search, and arrest were a function of the demo-

graphic and socioeconomic characteristics of Rich-

mond neighborhoods. The previous analyses lead to

three main conclusions. First, the total number of

stops by Richmond police was determined solely by

the crime rate of the neighborhood. Neighborhoods

with higher crime rates were likely to evidence a

higher number of stops by the police. In fact, none

of the demographic and socioeconomic character-

istics exerted direct effects on the number of police

stops. At first glance, these results appeared troub-

ling for conflict theorists. Second, the percentage of

stops that resulted in a search was determined by

only one characteristic: the percentage of Black

population. This result suggested that in areas pre-

dominantly inhabited by Blacks, police stops were

likely to result in a search. Third, when examining

the percentage of stops that ended in an arrest/

summons, the analysis revealed a slightly different

pattern of substantive results; namely, both the

percent Black population as well as the crime rate

served to decrease the percentage of police stops

that ended in an arrest/summons.

Table 3

Search regressed against predictor variables

Independent

variables

b S.E. b

Black 0.121 * * 0.039 0.570

Other � 0.643 0.905 � 0.104 Poverty � 0.0093 0.077 � 0.217 Unemployment � 0.0032 0.215 0.026 Income 1.058E� 05 0.000 0.038 Part I rate � 0.0009 0.014 � 0.073 Constant = 4.628

R2=.305

** P< .05.

Table 4

Arrest regressed against predictor variables

Independent variables b S.E. b

Black � 0.169 * 0.069 � 0.437 Other 1.411 1.591 0.125

Poverty � 0.0014E� 02 0.136 � 0.018 Unemployment � 0.198 0.378 � 0.090 Income � 4.617E� 05 0.000 � 0.092 Part I rate � 0.0058 * 0.025 � 0.267 Constant = 79.402

R2=.355

* P< .01.

Table 2

Stop rate regressed against predictor variables

Independent variables b S.E. b

Black � 0.0064 0.093 � 0.087 Other � 2.629 2.137 � 0.123 Poverty 0.0094 0.183 0.061

Unemployment 0.406 0.508 0.097

Income 1.265E� 05 0.000 0.013 Part I rate 0.337* * * 0.034 0.819

Constant =� 0.340 R2=.674

*** P < .001.

M. Petrocelli et al. / Journal of Criminal Justice 31 (2003) 1–11 7

These results seem to point to a ‘hurdle-effect’

when it comes to police officers differentially apply-

ing their power across communities in Richmond. For

example, the first hurdle, being stopped, seems more

of a function of the crime rate of the area rather than

any sort of demographic and/or socioeconomic char-

acteristic. Moreover, since it is difficult to determine

suspect race in many stops, it is not surprising to find a

null effect for the percent Black in the population.

Once the hurdle of being stopped has been crossed,

however, searches appear more prevalent in areas

comprised of a high percentage of Black residents.

This particular result could indicate that officers are

more likely to search Black residents based on general

perceptions (Piliavin & Briar, 1964). Even further,

searches could be a function of the areas in which

officers patrol. For example, it may be the case that

officers possess some sort of ‘ecological attribution

bias’. As Smith (1986) suggests, police may respond

differently depending on the type of neighborhood in

which the encounters occur such that police assess-

ments of individuals may reflect the kinds of people

who live in a particular neighborhood. Thus, Black

residents may be searched more frequently because of

the location in which they are stopped. Once this

search hurdle is crossed, however, it seems that arrests

may not be likely to be characterized by macro-level

factors; instead, the decision to arrest may be more

likely a function of other unmeasured macro-level

and/or individual characteristics. In sum, Richmond

police search Black suspects at a higher rate; however,

this trend is reversed when it comes to arrest since the

percent Black coefficient is negatively related to

arrest. Thus, since Richmond police are more likely

to search stopped motorists in areas of high Black

concentration, the nonsignificant effect for percent

Black on arrests may indicate that police make too

many searches that are either unsubstantiated or do not

yield evidence that could meet/exceed the bar for

arrest. A correction, then, is made. As Chambliss

(1994, p. 179) found in Washington, D.C., vehicle

stops (and subsequent searches) generally yielded hit

rates of 10 percent (for finding illegal drugs, weapons,

or someone who was wanted by authorities), although

officers believed that they found serious violations in

about a third of vehicle stops.

Before discussing the relevance of the above

findings for conflict theory in general, and police

practice in particular, several limitations must be

mentioned that preclude any sort of definitive state-

ment regarding racial profiling (or the lack thereof) in

Richmond. First, the data came from one city and

during a particular time period. Thus, the extent to

which the results would hold in other cities in other

time periods remains an open question. In addition,

because of a contractual agreement with the Rich-

mond Police Department, data were only collected for

a ten-week period. Although the data provided suf-

ficient statistical power for the analyses conducted, it

remains possible that more data or data from a

different time period could yield different results.

Clearly, future research should attempt to address

such issues. Second, the data set provided by the

police department did not contain information on

individual characteristics, specifically suspect-level

data on demeanor (Worden & Shepard, 1996). Given

the importance of suspect antagonism in determining

police decision-making (Klinger, 1994; Smith, 1986),

and how controlling for such micro-level factors

typically overshadows the effect of macro-level var-

iables, future efforts may wish to concentrate on an

examination of both macro- and micro-level charac-

teristics. Third, the data did not examine the influence

of organizational factors. Prior research has shown

that organizational characteristics influence police

decision-making at the street level (Smith, 1984);

although one would not expect much variation within

a single police force, the influence of organizational

characteristics on decision-making remains an empir-

ical question. Finally, although fairly traditional stat-

istical techniques were employed in this analysis,

future efforts may wish to investigate models that

take into consideration the potential overdispersion of

stops, searches, and arrests throughout the City of

Richmond, as well as potential spatial autocorrela-

tion.

With these limitations in mind, the results bear

import for conflict theory explanations of police

decision-making as well as police practice. Recall

that conflict theory presents a theory of the behavior

of the criminal law. According to conflict theorists,

relatively powerless people are more likely to be

officially defined as criminal, processed by the crim-

inal justice apparatus, and possess little ability to

infiltrate the criminal justice and legislative

decision-making systems (Center for Research on

Criminal Justice, 1975). These assumptions led to

the hypotheses that demographic (Blacks) and socio-

economic (poor) factors would be related to police

decision-making.

For the most part, the findings from the current

analysis appear mixed on this front. One of the most

important factors in starting the wheels of the crim-

inal justice process is the police officer’s decision to

stop a motorist. Conflict theorists would hold that

officers would be more likely to profile poor, minor-

ity (typically Black) areas, and to make a higher

percentage of stops in such areas. The analysis did

not uncover such a finding, thereby suggesting that it

may not be the area per se that determines the police

officer’s decision to stop; as the results showed, the

crime rate appears to be a more important factor in

M. Petrocelli et al. / Journal of Criminal Justice 31 (2003) 1–118

the decision to stop.6 Still, areas characterized by a

higher percentage of Blacks tend to have drivers

searched more often than areas characterized by a

lower percentage of Blacks. This result seems to

provide some support for conflict theorists (see

Chambliss, 1994). Still, police decisions to stop,

search, and arrest were not determined from the

socioeconomic characteristics of Richmond neighbor-

hoods.7

In the end, this study is neither the definitive nor

the last statement on racial profiling, as empirical

research in this area is just beginning. As researchers

chart the future study of racial profiling, they must

bear in mind that the collection of multiple levels of

data are necessary for a more complete understanding

of how police officers make decisions on the street

with regard to stopping, searching, and arresting

motorists. To deny that individual- and neighbor-

hood-level characteristics do not influence such deci-

sions would be to deny the fact that police ‘‘patrol

people and places’’ (Smith, 1986). Only when

researchers, citizens, and policymakers adopt this

perspective will people be better equipped to deal

with issues related to racial profiling.

Appendix A. Variables captured

Notes

1. Income and poverty data are not yet available from

the 2000 census.

2. To be sure, the ten-week data collection period is

somewhat short. The authors return to this point in the

Discussion section.

3. Supplemental analysis indicated that the assump-

tions underlying the OLS model were met.

4. In this and subsequent models, percent minority

population was excluded due to problems with multi-

collinearity. In Richmond, minorities other than African-

Americans make up less than 3 percent of the population.

Thus, percent Black appears to be the best variable to

measure the influence of race.

Scale

Driver variables

Age (year of birth) interval

Gender nominal

Race nominal

Asian

Black

Hispanic

Native American

Middle Eastern descent

White (Caucasian)

Officer variables

Race nominal

Asian

Black

Hispanic

Native American

Middle Eastern descent

White (Caucasian)

Gender nominal

Age ordinal

1 = 21–25

2 = 26–30

3 = 31–35

4 = 36–40

5 = 41–45

6 = 46–50

7 = 51 +

Length of service ordinal

1 = 0–5 years

2 = 6–10 years

3 = 11–15 years

4 = 16–20 years

5 = 21–25 years

6 = 26+ years

Shift (day, evening, power) nominal

Event variables

Census tract of stop nominal

Reason for stop nominal

Defects (no city decal, equipment,

expired registration or inspection)

Investigation

Moving violation

Disposition of stop (up to three possible) nominal

Advised (warning)

On-view felony arrest

On-view misdemeanor arrest

Summons issued

Warrant served

Mental detention order

Parking/city decal citation

DUI arrest

Offense report

Miscellaneous report

Juvenile violation report

Other report

Suspension notification issued

Vehicle towed

Vehicle search

Information received

Stolen vehicle recovered

Property found or seized

Guns/weapons found or seized

Search conducted (yes or no) nominal

Type of search conducted nominal

Consent

Incident to arrest

Inventory

Pat-down

Appendix A (continued)

M. Petrocelli et al. / Journal of Criminal Justice 31 (2003) 1–11 9

5. Percent minority population was again ex-

cluded from the model due to problems with multi-

collinearity.

6. To be sure, this does not fully negate the possibility

that police officers do not target Black motorists.

7. Although not technically a conflict theory, some of

the core propositions from Donald Black’s (1976) theory of

the behavior of law also imply that more powerful social

actors have the ability to use law against less powerful

actors. Basically, Black describes law (i.e., government

social control) as a quantitative variable such that there can

be more law at certain times and places and less law at other

times and places. Although not the focus of this presenta-

tion, future efforts aimed at studying racial profiling may

find Black’s theory useful.

References

Binder, A., & Scharf, P. (1982). Deadly force in law

enforcement. Crime and Delinquency, 28 (1), 1–23.

Black, D. (1976). The behavior of law. New York: Academic

Press.

Blalock, H. (1967). Towards a theory of minority group

relations. New York: Capricorn Books.

Blauner, R. (1972). Racial oppression in America. New

York: Harper and Row.

Center for Research on Criminal Justice (1975). The iron fist

and the velvet glove: An analysis of the U.S. police.

Berkeley, CA: Author.

Chambliss, W. (1976). The state and criminal law. In W.

Chambliss & M. Mankoff (Eds.), Whose law? What or-

der? A conflict approach to criminology. ( pp. 66–106).

New York: Wiley.

Chambliss, W. J. (1994). Policing the ghetto underclass: the

politics of law and law enforcement. Social Problems,

41, 177–194.

Chambliss, W. J., & Seidman, R. (1971). Law, order, and

power. Reading, MA: Addison-Wesley.

Chambliss, W. J., & Seidman, R. (1980). Law, order and

power. Reading, MA: Addison-Wesley.

Chamlin, M. (1989). Conflict theory and police killings.

Deviant Behavior, 10, 353–368.

Chiricos, T., Hogan, M., & Gertz, M. (1997). Racial com-

position of neighborhoods and fear of crime. Criminol-

ogy, 35, 107–131.

Dahrendorf, R. (1959). Class and class conflict in industrial

societies. Stanford, CA: Stanford University Press.

Government Accounting Office (2000). Racial profiling:

Limited data available on motorist stops. Washington,

DC: U.S. Government Accounting Office.

Harris, D. (1999, December). The stories, the statistics, and

the law: Why ‘‘driving while Black’’ matters. Minnesota

Law Review, 84, 265–326.

Holmes, M. (2000). Minority threat and police brutality:

Determinants of civil rights complaints in U.S. munici-

palities. Criminology, 38(2), 343–365.

Jackson, P., & Carroll, L. (1981). Race and the war on

crime: The sociopolitical determinants of municipal po-

lice expenditures in 90 non-southern cities. American

Sociological Review, 46, 390–405.

Jacobs, D. (1979). Inequality and the police use of deadly

force. Social Problems, 26, 403–412.

Jacobs, D., & Britt, D. (1979). Inequality and police

use of deadly force: An empirical assessment of

a conflict hypothesis. Social Problems, 26(4),

403–412.

Jacobs, D., & O’Brien, R. (1998). The determinants of

deadly force: A structural analysis of police violence.

American Journal of Sociology, 103, 837–862.

Kessler, D. A. (1999). The effects of community policing on

complaints against officers. Journal of Quantitative

Criminology, 15, 333–372.

Klinger, D. A. (1994). Demeanor or crime? Why ‘‘hostile’’

citizens are more likely to be arrested. Criminology, 32

(3), 475–493.

Knoohuizen, R., Fahey, R. P., & Palmer, D. J. (1972). The

police and their use of fatal force in Chicago. Chicago:

Chicago Law Enforcement Study Group.

Lamberth, J. D. (1997). Report of John Lamberth, Ph.D.

American Civil Liberties Union Web site: http://www.

aclu.org/court/lamberth.html.

Lawton, B. A., Hickman, M. J., Piquero, A. R., & Greene,

J. R. (2001). Using GIS to analyze complaints against

police: A research note. Justice Research and Policy, 3,

95–108.

Liska, A., & Chamlin, M. (1984). Social structure and crime

control among macrosocial units. American Journal of

Sociology, 90, 383–395.

Liska, A., Chamlin, M., & Reed, M. (1985). Testing the

economic production and conflict models of crime con-

trol. Social Forces, 64, 119–138.

Liska, A., Lawrence, J., & Benson, M. (1981). Perspectives

on the legal order: The capacity for social control. Amer-

ican Journal of Sociology, 87, 413–426.

Liska, A., Lawrence, J., & Sanchirico, A. (1982). Fear of

crime as a social fact. Social Forces, 60(3), 760–770.

Lizotte, A., & Bordua, D. (1980). Firearms ownership for

sport and protection. American Sociological Review, 45,

229–243.

New York Attorney General’s Office. (1999). The New York

City Police Department’s ‘‘stop & frisk’’ practices. New

York: Author.

Office of the Attorney General. (1999). Interim report of the

state police review team regarding allegations of racial

profiling. New Jersey: Author.

Piliavin, I., & Briar, S. (1964). Police encounters with juve-

niles. American Journal of Sociology, 17, 206–214.

Quinney, R. (1970). The social reality of crime. Boston:

Little, Brown and Company.

Quinney, R. (1974). A critical theory of law. In R. Quinney

(Ed.), Criminal justice in America (pp. 1–25). Boston:

Little, Brown and Company.

Quinney, R. (1975). Crime control in capitalist society: A

critical philosophy. In I. Taylor, P. Walton, & J. Young

(Eds.), Critical criminology (pp. 181–212). London:

Routledge and Kegan Paul.

San Diego, California Police Department. (2000). Vehicle

stop study: Mid-year report. San Diego, CA: Author.

San Jose, California Police Department. (1999). Vehicle stop

demographic study. San Jose, CA: Author.

Simmel, G. (1950). The sociology of Georg Simmel. (Trans-

M. Petrocelli et al. / Journal of Criminal Justice 31 (2003) 1–1110

lation and Introduction by K.H. Wolff). Glencoe, IL:

Free Press.

Smith, D. A. (1984). Organizational context of legal control.

Criminology, 22(1), 19–38.

Smith, D. A. (1986). The neighborhood context of police

behavior. In A. J. Reiss Jr., & M. Tonry (Eds.), Com-

munities and crime, crime and justice: A review of re-

search (Vol. 8). Chicago: University of Chicago Press.

Smith, M., & Petrocelli, M. (2001). Racial profiling?: A

multivariate analysis of police traffic stop data. Police

Quarterly, 4(1), 4–27.

Sorenson, J., Marquart, J., & Brock, D. (1993). Factors

related to killings of felons by police officers: A test

of the community violence and conflict hypotheses. Jus-

tice Quarterly, 10, 417–440.

Spitzer, S. (1975). Toward a Marxian theory of deviance.

Social Problems, 22, 638–651.

Swigert, V., & Farrell, R. (1976). Murder, inequity and the

law: Differential treatment in the legal process. Lexing-

ton, MA: Heath.

Tagaki, P. (1974). A garrison state in a ‘democratic’ society.

Crime and Social Justice, 1, 27–33.

Taylor, I., Walton, P., & Young, J. (1973). The new crimi-

nology. New York: Harper Torchbooks.

Turk, A. T. (1969). Criminality and legal order. Chicago:

Rand McNally.

Vold, G. (1958). Theoretical criminology. New York:

Oxford.

Walker, S. (1992). Taming the system. New York: Oxford

University Press.

Worden, R., & Shepard, R. (1996). Demeanor, crime, and

police behavior: A reexamination of the police services

study data. Criminology, 34, 83–105.

Zingraff, M., Mason, M. H., Smith, W. R., Tomaskovic-

Devey, D., Warren, P., McMurray, H. L., & Fenlon,

C. R. (2000). Evaluating North Carolina State Highway

Patrol data: Citations, warnings, and searches in 1998.

North Carolina State University Web site: http://

www.nccrimecontrol.org/shp/ncshpreport.htm.

M. Petrocelli et al. / Journal of Criminal Justice 31 (2003) 1–11 11