|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
URBAN DISADVANTAGE, SOCIAL DISORGANIZATION AND RACIAL
PROFILING: AN ANALYSIS OF ECOLOGY AND POLICE OFFICERS' RACE-
SPECIFIC SEARCH BEHAVIORS
ERIN C. LANE
A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF ARTS
UNIVERSITY OF FLORIDA
Erin C. Lane
This is dedicated to Aunt Cheryl and Grandpa Lane
In my studies at the University of Florida I have come in contact with a number of
individuals who have made positive contributions to my learning and development as a
graduate student. Dr. Beeghley was so influential; he guided me through my senior year
and directed me to the right resources when I was trying to put together a senior thesis.
He has had a profound impact on my life, and I am truly grateful.
I thank Alex Piquero for his patience and guidance, as he was instrumental in
helping me write and publish my senior thesis. Jodi Lane provided me with the tools and
guidance I needed to get things done and keep it all organized. Matt Nobles kept me
focused and somehow always knew what was going on. He was always ready to help. I
am forever grateful to Karen Parker, who helped me develop my research focus and
guide me through my graduate career. The care and expertise she provided are
Finally, I would like to thank my family for their support. They never gave up and
their persistence helped me get where I am today. My stepmother, Cheryl, my
grandmother, Aunt Jan, Uncle Perry, Uncle Paul, and had so much confidence in me. My
sisters, Kristin and Michelle, kept me motivated by always acting so impressed when I
informed them of my research. I thank my father for everything I have, his consistency,
love, concern, support, determination, and his wealth of knowledge.
TABLE OF CONTENTS
A C K N O W L E D G M E N T S ................................................................................................. iv
LIST OF TA BLE S .............................. ....... ...... .. .............. .. vii
A B S T R A C T .......................................... .................................................. v iii
1 IN TRODU CTION ................................................. ...... .................
2 REVIEW OF THE LITERATURE ........................................ ......................... 6
Public Perception L literature ............................................................................. 7
A actual Police Behavior ............................... ... ............ ....... ............. 9
Neighborhood Context and Police Behavior............................................................13
3 THEORETICAL PERSPECTIVES....................................... ......................... 17
Origins of Urban Disadvantage Perspective............................................................17
Development of Urban Disadvantage Perspective ...................................................20
Urban Disadvantage and Racial Profiling Linkages .......................................23
Origins and New Perspectives in Social Disorganization Theory.............................24
Social Disorganization and Racial Profiling Linkages............................26
A additional H ypotheses ............................................................. .....................27
4 D A T A A N D M E TH O D S ........................................ ............................................29
Unit of Analysis ............................. .............................. ........ 29
M ia m i D ata ................................. ......................................................................... 3 0
Citizen Contact Data.......................... .....................30
Dependent Variable................... ................ .... ....................... ........ 31
C en su s D ata .............. ...... .. ...................................... ............... .... .3 2
U rban D disadvantage Indicators ........................................ ....................... 33
Social D isorganization Indicators.................................... ........................ 33
O officer D ata ............................................................................ 34
C rim e D ata ............... ................................... ......... ............... 35
5 R E S U L T S ............................................................................................................. 3 8
Principle Com ponent Analysis ............................................................................38
A nalytic M ethod ....................................................... ..... .............. 39
T otal Sam ple A analysis .. .. .... .......................................................... .. .... .. .... .. 40
A analysis by A rea Type ..................... .. ...... ................... ...... .. ............... 44
6 DISCUSSION AND CONCLUSIONS ............................................................... 50
A PRINCIPLE COMPONENTS ANALYSIS AFTER VARIMAX ROTATION ........57
B DESCRIPTIVES: MEANS AND (STANDARD DEVIATIONS) FOR
VARIABLES IN SEARCH MODELS FOR NON-DISADVANTAGED AND
D ISA D V A N T A G E D ....................................................................... .....................58
LIST OF REFEREN CES ............................................................ ................... 59
B IO G R A PH IC A L SK E TCH ..................................................................... ..................64
LIST OF TABLES
4-1 Descriptive statistics: Means (and standard deviations) for variables in race-
specific search m odels............ .................................................... .. .... ........ 37
5-1 Seemingly unrelated regression (SUR) coefficients (and standard errors)
for the total sam ple ......... .. .................................... .... .... .... ........ ........ .... 41
5-2 Seemingly unrelated regression (SUR) coefficients (and standard errors)
for non-disadvantaged and disadvantaged areas ............................................. 46
Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Arts
URBAN DISADVANTAGE, SOCIAL DISORGANIZATION, AND RACIAL
PROFILING: AN ANALYSIS OF ECOLOGY AND POLICE OFFICERS' RACE-
SPECIFIC SEARCH BEHAVIORS
Erin C. Lane
Chair: Karen F. Parker
Major Department: Criminology, Law and Society
Racial profiling by police officers has become a popular subject in the field of
criminology in the recent past, gaining extensive coverage in the literature. The study of
racial profiling varies greatly in scope and research method, resulting in three primary
forms in the literature, including measurement of citizen perceptions of police practices,
citizen self-report data on contacts with police, and actual police stop data. However,
findings from research in these veins vary greatly, often failing to take into account
ecological context and neglecting to incorporate theory as a factor guiding research. This
research seeks to compensate for these deficiencies and add to the literature by enhancing
our understanding of racial profiling through theory-based research. Using data collected
from more than 61,000 police stops in Miami Dade County, I will examine a number of
factors that may influence race-specific search behaviors independently for
neighborhoods of high and low disadvantage in an attempt to explore and compare the
contributions of urban economy/disadvantage and disorganization theories to the study of
Minorities in general, and specifically black males, are drastically overrepresented
in the criminal justice system (Walker, 2001). Scholars agree that a primary cause of this
disproportionality may stem from law enforcement officers' use of race as the key
determinant for stopping, interrogating, searching, and/or arresting individuals of
minority group status. Support for the salience of this phenomenon has been presented
countless times through various forms of empirical research (Hagan and Albonetti, 1982;
Harris, 1997; Lamberth, 1997; Walker et al., 2000; Brown, 2001; Engel et al.' 2002;
Weitzer and Tuch, 2002; Zingraff et al., 2002; and Lundman and Kaufman, 2003).
The act of racial profiling by police officers, defined as the process by which law
enforcement agents use race as the key factor in determining whether to stop, search, cite,
and/or arrest minority group members, has recently become a prominent focus within the
field of criminology (Engel et al., 2002). We know that the practice of racial profiling
exists, but why? What factors cause police officers to use race as a primary determinant
in making the decision to conduct stops and searches?
Leading scholars have advanced many reasons for officers' more punitive
behavior toward minority individuals. One of the more popular explanations concerns
crime statistics, which show that African Americans are more likely to commit crime,
particularly violent crime, than individuals of other races. However, these rates may be
misleading, and should be analyzed beyond face value (Walker, 2001). For example,
conviction rates of African Americans are proportionately higher than that of other racial
groups. The simple fact that racial profiling elicits such public reaction forces individuals
to consider alternatives to these rather basic measures of arrest, conviction, sentencing,
and incarceration. For example, African Americans living in high crime areas typically
have more frequent contact with police due to the increased presence of police officers,
and are, therefore, more likely to have their criminal activity discovered (Krivo and
Peterson, 1996). Some research has indicated that African Americans and other
minorities are more likely to drive vehicles with broken taillights, cracked windshields,
and other equipment violations, giving officers more reason to make a traffic stop, and
again increasing the likelihood that contraband or criminal activity is detected (Mac
Donald, 2003). Still, other scholars cite the underlying laizze faire brand of racism that
exists today in the United States as a primary factor causing high rates of racial profiling
(Brown, 2001; Massey et al., 1994; Wilson et al., 2004).
What seems to be less salient in this research is an explanation of police officers'
decisions to conduct race-specific searches subsequent to traffic stops, relative to
neighborhood characteristics. While a small number of studies do take into account
officers' behavior relative to ecological conditions, findings have been inconsistent.
Therefore, it is still unclear how and to what degree contextual factors may contribute to
higher rates of African American searches as compared to that of whites. Findings from
this small body of literature have, however, consistently demonstrated that police officers
decision-making processes are affected by neighborhood conditions, being more likely to
handle suspects more punitively in neighborhoods with high poverty and crime levels,
and equate neighborhood characteristics with that population residing in these
neighborhoods (Robison, 1936; Werthman and Piliavin, 1967; Bayley and Mendelsohn,
1969; Lundman, 2004; Meehan and Ponder, 2004; and Engel and Calnon, 2004).
Further, researchers have found that officers often judge seriousness of a situation based
on the neighborhood context (Terrill and Reisig, 2003).
Two perspectives have been consistently successful in explaining the relationship
between ecology and crime urban disadvantage and social disorganization (Bursik,
1988; Massey, 1990; Wilson, 1987; and Sampson and Wilson, 1995). These perspectives
are highly interrelated, as urban disadvantage perspective finds its origins in the social
disorganization tradition. The primary difference is that the urban disadvantage
perspective posits that racial discrimination in housing markets is the primary
contributing factor to the concentrated disadvantage of poor African Americans, resulting
in pronounced racial residential segregation across the United States. This type of
discrimination caused a clustering of minorities living in extreme social isolation and
poverty (Massey, 1990). Conversely, the social disorganization perspective holds that
high crime rates and other negative neighborhood conditions are the result of the inability
of residents to realize common goals and, therefore, are unable to regulate the activity of
youth or other at-risk groups within their communities. This cause of this social distance
can be found in neighborhood instability, residential mobility and communication
barriers, and the like (Sampson, 1988; and Sampson and Wilson, 1995).
Neighborhoods with concentrated disadvantage have higher crime rates because the
social isolation they are subjected to, in combination with lack of access to legitimate
employment, creates a sort of sub-cultural tolerance of criminal behavior (Cullen and
Agnew, 2003). In effect, this sub-cultural tolerance leads to higher crime rates and
therefore to higher levels of police presence, resulting in increased rates of stops,
searches, arrests and incarceration in these areas. These conditions perpetuate a cycle of
crime and poverty geographically concentrated in urban neighborhoods (Rose and Clear,
1998). As a result of racial residential segregation, there exists a greater social distance
between officers and residents of highly disadvantaged areas. By this reasoning, due to
the prejudice that is perpetuated through discrimination in housing markets, police
officers will be more likely to treat African Americans more punitively.
Social disorganization theory explains concentrated disadvantage as a result of a
lack of informal social control, stemming from high residential mobility, population
heterogeneity, concentrated poverty, and family dissolution (Bursik, 1988; and Sampson
and Wilson, 1995). These characteristics developed in response to significant changes in
the urban economy beginning in the 1970s. After the relocation of many manufacturing
jobs into suburban areas and overseas, inner city residents were left without access to
employment. Coupled with the exodus of many working- and middle-class African
Americans from inner cities, so began the existence of the underclass (Wilson, 1987).
Essentially, residents of these neighborhoods lack communication and are unable to
realize their common goals and interests, and thus are unable to maintain informal social
control over local youth and other residents (Sampson, 1988; Sampson and Wilson, 1995;
and Cullen and Agnew, 2003). As a result of residents being incapable of community-
level regulation, more formal social control is necessary to maintain social order, and
therefore, more police officers are dispatched to these areas due to higher prevalence of
criminal activity (Bursik, 1988).
This paper uses these perspectives to guide the research design in examining the
relationship between police officers' race-specific search practices and ecological
context, uncovering the dynamics of race, context, and police behavior. Using the urban
disadvantage and social disorganization perspectives, this research seeks to extend
explanations of racial profiling at the census tract level, using citizen contact and officer
data from more than 61,000 traffic stops in Miami-Dade County, collected by the Miami-
Dade Police Department during a six-month period in 2001. Further, both crime data
relevant to geographic location and census tract data from the 2000 Census will be used
to provide a clearer picture of the reality of racial profiling and race-specific stops and
searches by police in Miami-Dade County.
In the next section, I review supporting literature on racial profiling, including
research from each type of methodology. Following that, I review the central tenants of
urban disadvantage and social disorganization perspectives, finishing with a brief
discussion of revitalizations of the latter. In the next section, I report details on the data
set used, as well as research methodology. Finally, I present a discussion the findings of
the research and suggest direction for future research in the study of racial profiling.
REVIEW OF THE LITERATURE
Racial profiling is the process by which law enforcement agents use race as the key
factor in determining whether to stop, search, cite, and/or arrest individuals of minority
status (Engel et al., 2002). The phrase 'driving while black' was coined in the early
1990s as the US began to see a rise in public awareness of and concern about
discriminatory practices by police officers (Harris, 1997). This increase can be partially
attributed to highly publicized cases of racial discrimination by members of our justice
system, particularly the heinous beating of Rodney King and the events that took place in
Los Angeles following the acquittal of the officers responsible for such abuses of power
(Brown, 2001). However, the recent growth of this body of literature is not indicative of
the true beginnings of this phenomenon; evidence of the existence of racial profiling has
long been documented.
Research dating back as early as 1936 cites evidence that police used race as a key
factor in their decision-making processes (Robison, 1936). Further, official data on
citizen approval ratings of police practices documented since 1967, show that the
relationship between minority citizens and police is at best, strained (Werthman and
Piliavin, 1967; and Walker et al., 2000). Racial profiling has become one of the most
common explanations for this differential treatment of minorities by police officers, thus
illustrating the importance of research in this area (Harris, 1997; Lamberth, 1997; Walker
et al., 2000; Brown, 2001; Weitzer and Tuch, 2002; Lundman and Kaufman, 2003; and
Three distinct forms of research on the subject matter have consistently appeared in
the literature. Citizen public perceptions have been used to measure prevalence of racial
profiling and perceptions of the overall legitimacy of police behavior (Hagan and
Albonetti, 1982; Weitzer and Tuch, 2002; and Lundman and Kaufman, 2003). The
second form of research focuses on actual police behavior by counting and analyzing data
from actual traffic stops and searches, but findings from this research have been relatively
inconsistent (Lamberth, 1997; Spitzer, 1999; and Harris, 1997). The final perspective is
the most recently developed and consequently provides the smallest body of literature; it
focuses on racial profiling behavior relative to neighborhood context. Specifically, this
vein analyzes neighborhood characteristics and compares them with rates of racial
profiling behavior by officers (Robison, 1936; Werthman and Piliavin, 1967; Smith,
1986; Fagan and Davies, 2000; and Terrill and Reisig, 2003). It is this last group of
literature we seek to expand with this research. A brief review of each form of literature
Public Perception Literature
Hagan and Albonetti (1982) pioneered the study of citizen perceptions of police
behavior and racial profiling. In their analysis of the United States' Department of
Justice survey on public perceptions of police practices, they found evidence that being a
member of a minority group significantly increases the likelihood of perceptions of
injustice. Of particular interest is their finding that African Americans scored an average
of 13 points higher than whites on the "perceptions of injustice" scale. This indicates that
relationships between police and minority citizens suffer from some external aggravating
factors, most likely racial profiling and differential treatment (Hagan and Albonetti,
In 2002, Weitzer and Tuch conducted a study measuring citizen perceptions of
racial profiling, including citizen approval/disapproval of police use of the practice,
prevalence rates, citizen perceptions of the utility of the practice, and personal experience
with police officers among others. The researchers used characteristics of respondents
including race, gender, age, type of neighborhood (urban vs. rural), past experiences with
police, and so forth, to check for patterns in citizen perceptions of police. The results
indicated first that 94.3% of African Americans and 84.4% of whites disapproved of the
use of racial profiling as a method of crime prevention by police officers. Results also
showed that African Americans were 3 times more likely than whites to hold negative
opinions of local law enforcement agents, and 4 times more likely to have negative
opinions of state law enforcement. Race and personal experience with police officers had
the strongest influence on attitudes toward the police, with minority members being more
likely to hold negative opinions. Finally, African Americans who had some form of
contact with police officers within the past 12 months from the time of the interview were
significantly less likely than their white counterparts to believe they had been treated
fairly by police officers; 16.6% of the sample was eligible for this comparison. Findings
revealed that 26.1% of African Americans felt they had been treated unfairly, compared
with 7% of whites. The authors concluded that perceptions of racial profiling vary by
race, class, and gender; the most significant indicators of negative perceptions were race
and personal experience (Weitzer and Tuch, 2002).
Lundman and Kaufman (2003), using data from a nationally representative survey,
searched for a relationship between race and ethnicity of drivers and police decisions to
conduct traffic stops. The authors used the National Crime Victimization Survey, a
nationally representative survey that among other things includes information on citizen
encounters with police, to perform their analysis. The results indicated that African
Americans report being stopped more often than both Whites and Hispanics. The results
further indicated that African Americans stopped by police were significantly more likely
to believe that the officer's reason for the traffic stop was illegitimate or based on extra-
legal factors (Lundman and Kaufman, 2003). This type of research lends credence to the
study of racial profiling, but public perceptions are sometimes less credible than analyses
of actual police behaviors.
Actual Police Behavior
The New Jersey Turnpike study (Lamberth, 1997) is one such study that accounts
for actual police stop behavior. The researchers first determined the racial composition
of the driving population on a set section of the Turnpike to establish a base line, and then
determined the racial composition of speeding drivers. The researchers found that
although African Americans composed 13.5% of all motorists and only 15% of all
speeding motorists, they accounted for 35% of persons stopped and/or searched by law
enforcement officers. African Americans were 4.85 times more likely to be stopped or
searched by police. Further inquiry based on police records indicated that African
Americans composed 73.2% of all arrests during a 3.5-year period, making them 16.5
times more likely to be arrested as compared to drivers of other races (Lamberth, 1998).
In 1999, Spitzer, in cooperation with the New York Attorney General's Office
(1999), released the results of a year long study of stop and frisk practices of New York
Police Department officers. They found that although African Americans represented
only 25.6% of the population, they composed 50.6% of all persons stopped by officers.
Hispanics were also over-represented compared to their proportion of the population;
they composed 23.7% of the population and 33% of persons stopped. This pattern
persisted even after researchers controlled for the differential rates at which minorities
commit crimes. The researchers concluded that there was a clear over-representation of
minorities being stopped by police as compared to their proportion of the population
Chambliss (1994) conducted a study in which assistants observed behaviors of
police officers assigned to the Rapid Deployment Unit of Washington D.C.'s police
department. Although this research was not originally intended to be used to explain
racial profiling, the observations collected were disturbing enough to merit further
analysis. Data on police behaviors observed by researchers during ride-alongs were
collected and analyzed, illustrating the level of disparity in treatment of minorities by
police officers. In general, police officers were found to treat African Americans more
harshly during traffic stops. Officers were also quoted admitting to the use of racial
profiling and reliance on pre-textual factors used to stop minority drivers in certain areas
of Washington D.C. Although the paper is without statistical analysis, the observations
made were disconcerting and illustrate that, at least in South Central Washington D.C.,
racial profiling was openly practiced; officers were relatively unconcerned with
concealing their methods of policing (Chambliss, 1994).
Harris (1997) conducted a study examining racial profiling activity of police
officers in three of the largest metropolitan areas of Ohio based on court records. The
researchers compared court record violator rates of African American and white
motorists with their respective percentages of the population to gage disproportionality of
police stop practices. The results indicated that African Americans were at least 2 times
more likely to be stopped as compared to whites, a clear indication of racial profiling
With the intent of gauging the effectiveness of a drug interdiction initiative along a
stretch of highway of Interstate 95 in Florida, the Orlando Sentinel conducted a study of
police stop practices (Cole, 1999). Each police car was equipped with video camera so
that researchers could collect detailed information of each stop. Data from 1,100 stops
were captured on video and reviewed by researchers. The videos provided clear evidence
that approximately 70% of drivers stopped were African American, while they composed
less than 5% of the driving population along that specific stretch of highway. The
disparity in this study is significant and shows officers propensity to stop minorities more
often than whites while attempting to stop drug trafficking (Cole, 1999).
Some research on actual police stop behavior has found variations in support for
different types of police action. For example, Smith and Petrocelli (2001) analyzed data
collected by the Richmond, Virginia, Police Department on traffic stops, checking for
patterns of racial profiling in police traffic stop behaviors. The data consisted of both
officer and citizen static characteristics and officers' work histories. Using multivariate
analysis, the researchers found that the proportion of traffic stops of minority members,
and particularly African Americans, was significantly higher than their proportion of the
population. African Americans accounted for 51% of the population of licensed drivers,
but accounted for more than 64% of all police traffic stops. Minorities accounted for
nearly 78% of all arrests subsequent to traffic stops, and 63% of those ticketed or arrested
in the sample. Interestingly, however, they found no significant disparity between search
and seizure rates and driver's race (Smith and Petrocelli, 2001).
One problem with the measurement of actual police stop behavior is that research
studies have yielded inconsistent results, at times finding no evidence of racial profiling
by officers. After compiling information on all police traffic stops conducted during a 4-
month period, researchers in cooperation with the Florida Highway Patrol (2000) found
that Florida Highway Patrol officers were stopping minorities at a rate almost equivalent
to their proportion of the population. While the largest amount of disparity occurred with
Hispanics, composing 12% of the population and 17.9% of all persons stopped, African
American motorists were stopped at a rate almost equal to their percentage in the
population. The findings of this study, however, may be explained by subject reactivity,
the concept that individuals change or modify their behavior when they have knowledge
that it is being monitored. The bottom line is, though, that there is little evidence of racial
profiling among officers employed by the Florida Highway Patrol (Florida Highway
Cordner et al., in collaboration with the San Diego Police Department, conducted
an examination of police practices based on police-reported data on traffic stops, searches
of vehicle, and vehicle search hit rates. Results showed that vehicles of only Hispanic
drivers were more likely than vehicles of whites to be searched. Researchers found no
evidence of disparity in search rates between whites and African Americans. It should be
noted, however, that vehicle searches of whites were 2 times more likely to yield search
hits as compared to vehicle searches of Hispanics. While the white/African American
disparity was not significant, higher search rates of Hispanics remains an indicator of
racial profiling (Cordner et al., 2000).
Preliminary findings of a study conducted by Zingraff et al. and the North Carolina
State Highway Patrol were released in 1998. To assess the prevalence of racial profiling
among its officers, the North Carolina State Highway Patrol collected police-reported
data on search and hit rates of motorists on North Carolina's highways. While the
occurrence of searches by officers was rare, data indicated that African Americans were
more likely to be searched than whites, at rates of 13 searches per 10,000 written
warnings or tickets vs. 8 searches per 10,000 written warnings or tickets, respectively.
Discovery of contraband occurred in 33% of searches of whites, while only 26.3% of
searches of African Americans resulted in a hit. There is clear evidence of disparity;
however, it may be the result of a small sample size (n=826 searches out of 906,758
written warnings or citations) (Zingraff et al., 2000).
Neighborhood Context and Police Behavior
A small body of research exists which has documented factors influencing police
behavior. A consistent finding from this body of literature is that police behavior is
affected by the characteristics of neighborhoods in which officers work. It is important to
examine the ecological characteristics of urban disadvantage research shows that there is
extreme variation in the structural characteristics between predominantly African
American and predominantly white neighborhoods.
Robison (1936) examined the relationship between delinquency and the accuracy of
its measurement through official police data. Results suggested that the decision to arrest
youth by police officers was made after considering two factors, severity of rule-breaking
behavior and the juvenile's moral character. The latter factor, she argued, was
determined primarily by place of residence, with arrests more likely to result when a
juvenile suspect claimed residence in a poor area (Robison, 1936).
Werthman and Piliavin (1967) conducted a study in which they observed and
interviewed police officers in the San Francisco Police Department to analyze factors
influencing police behavior. According to the researchers, police actions are influenced
by the ecological composition of their patrol area due to officers' tendencies to define
geographic locations as suspicious in general. Further, the researchers proposed and
supported the assertion that neighborhood residence was a primary indicator in police
decisions to select potential law violators (Werthman and Piliavin, 1967). Bayley and
Mendelsohn (1969) found similar evidence in support of the ecological influence on
police behavior. Specifically, they found that there existed a greater social distance
between the police and the poor, which often resulted in more punitive and/or aggressive
behavior toward citizens in lower-class areas. They further noted that police attributed an
individual's presence in a poor or low class neighborhood to the individual's level of
threat, and therefore treated them more punitively (Bayley and Mendelsohn, 1969).
Using data from the Police Services Survey, Smith (1986) examined the effects of
various community characteristics on police behaviors. The neighborhood characteristics
included crime rates, racial composition, poverty levels, and other socioeconomic factors.
The author found that officers were significantly more likely to use force against African
Americans. That effect, however, was mediated by neighborhood context, with
individuals stopped in minority neighborhoods being more likely to be treated coercively.
These individuals were 3 times more likely to be arrested than individuals encountered
outside minority neighborhoods (Smith, 1986).
Fagan and Davies (2000) analyzed patterns of police "stop and frisk" practices of
the New York City police department. Using measures of social disorganization (i.e.,
broken windows, graffiti, etc.), or what the author referred to as "Broken Windows"
theory, as well as structural characteristics, including racial composition and poverty
levels, Fagan attempted to find racial bias in police decisions to "stop and frisk"
individuals. The author's hypothesis, that measures of social disorganization and
structural characteristics are the most significant indicators race-specific "stop and frisk"
practices by police, was confirmed. The results indicated that African Americans and
Hispanics were significantly more likely to be "stopped and frisked" by police officers.
Further, the standards required by law to justify stopping these minority groups more
often were not met or were borderline unconstitutional. "Stop and frisk" practices were
concentrated in poor neighborhoods with large minority populations. Researchers
concluded that police focus on regulating the poor in poor areas. Due to racial
composition of poverty in the United States, African Americans and Hispanics are
overrepresented in police "stop and frisk" practices (Fagan and Davies, 2000).
Terrill and Reisig (2003) also examined neighborhood context relative to police use
of force using data collected from 2 US cities. Their findings were consistent with that of
other research in this vein. The authors found that officers were more likely to treat
young, male, lower class suspects with more force. More significantly, findings
demonstrated that police officers were significantly more likely to use force when in
contact with suspects in neighborhoods with high crime rates and with high levels of
concentrated disadvantage. The significance of this finding remained even after
controlling for behavior of the suspect (Terrill and Reisig, 2003).
The consistency of the findings in this form of research indicates that police
officers may very well be influenced by contextual factors of the neighborhoods where
they encounter suspects. While the cited research tends to focus on use of force and
coercion by officers, it demonstrates that officer behaviors vary across neighborhoods,
based on context, with officers acting more punitively toward suspects in low class, high
crime, minority neighborhoods. Accordingly, in the next chapter I provide a discussion
of theories linking officers' racial profiling behaviors to neighborhood conditions.
Origins of Urban Disadvantage Perspective
Urban disadvantage perspective finds its origins in the work of William Julius
Wilson. In his book, The Truly Disadvantaged (1987) he describes how the changing
structure of the urban economy of the 1970s combined subsequently with the
development of a cultural value of weak labor force attachment contributed to the rise in
poverty, and specifically, its concentration in inner city communities. Wilson argues that
the extreme rise in the concentration of poverty was caused by three inter-related factors;
(1) the changing economic structure from goods producing to service producing, (2) the
relocation of manufacturing jobs into suburban areas and overseas, and (3) the out-
migration of working- and middle-class African Americans from the inner-city to the
suburbs (Wilson, 1987; and1992).
During the 1970s the United States, particularly the Midwest and Northeast regions
began experiencing drastic shifts in the local manufacturing-focused economies. The
advancements made in technological industries contributed to these changes, as did the
motivation to secure cheaper labor. Both technological and service industries
experienced significant growth at the same time that the industrial sector was declining.
This radically decreased the number of manufacturing jobs available to the poor,
unskilled, undereducated minorities living in inner cities.
Another contributing factor to the creation of the underclass was the relocation of
manufacturing jobs from the inner city to suburban areas and overseas. The outcome was
a spatial mismatch between the unskilled, undereducated industrial workers of the inner
city and the physical locations of potential employment. Such great distance between
place of residence and place of employment resulted in the inability of poor inner city
manufacturing workers to maintain employment in that field and significantly reduced
the number of potential job opportunities available (Wilson, 1987; and 1992).
Inner city unemployment rates soared as its residents did not have access to the
education, training and skills required to gain employment in the technological industry.
High unemployment rates coupled with intermittent economic recessions created a loose
labor market, with too few jobs for too many people. Consequently, wages for unskilled
service workers were significantly reduced, making it difficult for inner city African
Americans to support a family. As Erbe (1975) explained, the decline of proper
functioning of local institutions directly affects the quality of the local education system.
Loss of employment caused this decline, leaving residents of inner city neighborhoods
educationally unprepared to handle the shift; they simply did not have the same access to
training and quality education that others outside the inner city had. This further
contributed to the already weak labor force attachment of African American males
(Wilson, 1987; and 1992).
The decentralization of the labor market clearly had the effect of decreasing
opportunity for employment of African Americans residing in inner cities in general, as
measured by the ratio of available employment to people and the average commute time
to and from places of employment. Moreover, the newly created loose labor market and
the subsequent decrease in wages for what was left of available inner city jobs
contributed to the ever-increasing social isolation of the underclass and intra-class
segregation of African Americans. This was evidenced through a comparison of less-
educated African Americans living in inner cities and less-educated African Americans
living in the suburbs; less-educated African Americans living in the suburbs tend to
receive higher wages as compared to those living in the inner city. Inner city residents
were forced to take employment for wages barely large enough to provide for families
Finally, the social and economic changes brought about by the Civil Rights
movement for a short time provided more access to housing markets outside urban areas
for working- and middle-class African Americans. Access to housing markets outside
the inner city led to the out-migration of working- and middle-class African Americans,
thus leaving a concentration of poor African Americans socially isolated from
conventional role models. Social isolation of the poorest minorities caused substandard
socialization; inner city residents were no longer exposed to positive, hard-working
African American role models. This hindered their ability to see that there was
opportunity beyond the inner city, that there was a possibility for upward mobility for
those who are determined to improve their situation. Therefore, they adapted to what
they believed was the only way of life.
The effects of this concentration of poor, unskilled and undereducated African
Americans resulted in the continued deterioration of access to employment opportunities
and networks, decreased the already substandard quality of education and diminished the
number of marriageable partners, thus contributing to higher levels of family dissolution,
and further aggravating the weak labor attachments of inner city residents. These factors
together work to produce what Wilson termed the underclass, poverty-stricken groups of
minorities living in inner cities lacking access to employment, education, social networks
and conventional role models (Wilson, 1987).
Development of Urban Disadvantage Perspective
Massey (1990), in cooperation with Denton and Eggers, finds support for Wilson's
theory of the inner city underclass and factors contributing to the concentration of
poverty, but argues that the primary cause of the rise in concentrated poverty among
African Americans was due to high levels of racial residential segregation (Massey,
Eggers and Denton, 1994). While excluding the influence of African American out-
migration due to lack of significance in statistical analyses, they theorized that additional
factors including the African American poverty rate, intra-racial residential segregation
by class, and most significantly, levels of African American/white residential segregation
acted to create the underclass. The residential preferences of whites worked to eliminate
any chance of racial integration.
The goal of the researchers was to identify potential inter-relationships between
these factors and their influence on the negative social conditions representative of
African American, poverty-ridden, inner city neighborhoods as outcomes. Examples of
negative social conditions include high crime rates, decreased neighborhood social
networks, low social control, increased rates of out-of-wedlock births, lack of availability
of marriageable partners, increased unemployment rates, higher percentages of female-
headed households, concentrated poverty, and so forth (Massey, 1990; Massey & Eggers,
1990; Massey, Gross and Eggers, 1991; Massey, Eggers and Denton; 1994).
The fact that the concentration of poverty increased during the 1970s has
consistently been confirmed empirically. According to Massey and colleagues, the
answer to explaining the concentration of poverty disproportionately experienced by
African Americans lies in racial residential segregation stemming from discrimination in
housing markets. The interaction between high levels of racial residential segregation
and inequality in income distribution created a concentration of poverty within the
poorest African American neighborhoods. Massey claimed that Wilson's concept of the
out-migration of working- and middle-class African Americans did not explain why the
concentration effects disproportionately affected African Americans. He posits that
while more African Americans became upwardly mobile in the 1970s and 1980s, there
was not in fact a mass exodus of these classes out of urban areas. Although the degree of
African American intra-class segregation did increase, it was still much lower than the
intra-class segregation experienced by other comparable minority groups. Furthermore,
results from multivariate analysis show that the tendency for African Americans of
opposing classes to live in areas segregated from each other is not related to the rates of
African American poverty concentration in inner cities (Massey, 1990).
In general, African Americans experience higher levels of poverty than whites, a
condition that is consistent throughout the majority of cities in the United States. When
racial residential segregation is imposed, shifts in the economy are disproportionately
concentrated in poor minority neighborhoods. The absorption of the negative effects
resulting from a downward economic shift by already poor minority neighborhoods acts
to exacerbate the already meager fiscal situations of residents in small geographic
locations, thus concentrating levels of poverty and creating an urban underclass (Massey,
To support his claims, Massey performed a series of simulated experiments to
examine the effects of an imposed downward economic shift on both racially segregated
and non-segregated hypothetical cities generally representative in racial composition and
economic situation of the majority of U.S. cities. He confirmed the hypothesis that the
concentration of poverty would be more equally distributed in areas with low levels of
racial segregation, as compared to areas with high levels of racial segregation. Findings
included the discovery that when income deprivation was applied to racially segregated
areas, some whites were actually better off economically, whereas all African Americans
were worse off, regardless of class (Massey, 1990). The results indicated that as racial
residential segregation increases, the absorption of downward economic shifts becomes
more concentrated within poor minority neighborhoods.
In their 1994 study of the causes of concentrated poverty in urban centers, Massey,
Eggers and Denton found evidence that the strongest effect on the level of African
American poverty came from the earnings that African American workers could expect to
receive in manufacturing and services. Higher income levels in these industries translate
into lower levels of poverty; unfortunately, the reciprocal was reality for the vast majority
of urban residents. Further, they found that the concentration of poor African Americans
in urban areas was significantly and negatively impacted by any shifts in the economy.
Racial residential segregation played a compelling role in the concentration of poverty
and its endogenous, interacting components including income inequality, majority
prejudice toward African Americans and racial discrimination in housing markets
(Massey, Eggers and Denton, 1994).
Urban disadvantage perspective has been applied to help understand the
relationships between property and drug crime, homicide, and racial profiling. One study
used urban disadvantage perspective in an analysis of property crime levels, finding that
as levels of segregation within given areas increases, so too do levels of property crime
(Akins, 2003). Using census data from 1990, Krivo and Peterson (1996) found that racial
differences in structural disadvantage was linked to higher rates of violent crime (Krivo
and Peterson, 1996). Parker and Pruitt (2000) analyzed race-specific homicide rates
relative to poverty and poverty concentration. Their finding that these factors
differentially affected rates of black and white homicide provides support for urban
disadvantage perspective (Parker and Pruitt, 2000). Finally, Meehan and Ponder found
that as measures of urban disadvantage decreased, African Americans were subject to
higher rates of stops by police (Meehan and Ponder, 2004).
Urban Disadvantage and Racial Profiling Linkages
The key component of urban disadvantage perspective is that concentrated
disadvantage was created out of racially motivated discrimination in the housing markets,
due to white preferences to not live in integrated neighborhoods. African Americans are
significantly more likely to live in areas of concentrated disadvantage as compared to
whites; even the poorest whites are significantly less likely to live in areas of
concentrated disadvantage. The absence of whites in poor, high crime areas may lead
officers to equate African Americans with higher levels of criminal activity; the areas
African Americans inhabit often have significantly higher crime rates, so officers may be
more likely to believe that they are involved in crime. Consequently, as measures of
urban disadvantage increase, police officers will be more likely to use race as a primary
discriminatory factor in their decisions to conduct searches. Research has shown that
increases in the levels of urban disadvantage increase levels of political dislocation,
which can have the effect of creating strained relationships between African Americans
and police officers. High rates of unemployment may increase the likelihood of contact
with police; this may have the effect of increasing levels of racial profiling. It then
follows that areas with higher levels of disadvantage will experience higher search rates
of African Americans as compared to that of whites in the same areas. Based on these
assumptions, we derive the following hypotheses on the relationship between urban
disadvantage and racial profiling:
* Hypothesis One: As indicators of concentrated disadvantage increase, search rates
of African Americans will also increase.
* Hypothesis Two: As indicators of concentrated disadvantage increase, white search
rates will not increase.
Should evidence be found in support of these hypotheses, the evidence will also lend
support to the urban disadvantage perspective in explaining higher rates of African
American searches by police officers in neighborhoods of high disadvantage in
comparison to that of low disadvantage neighborhoods.
Origins and New Perspectives in Social Disorganization Theory
Social disorganization theory finds its origins in the work of Shaw and McKay,
pioneers of the study of social processes through observation. The central tenets were
extracted from Burgess's concept of the division of cities. Essentially, a central city can
be dissected into 5 distinct zones- Zone 1 is the business or industrial zone, Zone 2 is the
transition zone inhabited by poor immigrants, ready to leave the area at the first
opportunity. Zone 3 is composed of working class domiciles, while Zone 4 is
characterized by middle class neighborhoods and Zone 5 inhabited by commuters (Cullen
and Agnew, 2003). The researchers believed that if this dissection of a central city was in
fact indicative of the maturation process of a real urban center, then rates of delinquency
would be highest in zone 2, the zone in transition (Shaw and McKay, 1942).
To provide evidence for their theory, Shaw and McKay mapped the location of
every juvenile delinquent in Chicago's urban center, according to police records, and then
computed delinquency rates over time. The hypothesis that delinquency rates would be
highest in zone 2 was confirmed. The authors concluded that delinquency rates were
generally stable over time, regardless of race or ethnicity of the dominant population in a
given time period. This evidence supported the assertion that it is not characteristics of
the individuals inhabiting the area, but characteristics of the area itself that increased
delinquency rates. According to Shaw and McKay, social disorganization refers to the
breakdown of social institutions within a community. Factors contributing to this
breakdown included high rates of family disruption, ethnic heterogeneity, and residential
instability (Cullen and Agnew, 2003).
It was not until the mid-1980s that social disorganization theory began to regain
popularity, taking the focus off of micro-level theories of crime prevalent after the 1970s.
Robert J. Sampson was one of the pioneers of the new social disorganization perspective.
Drawing on the work of Shaw and McKay, Sampson proposed that the cause of high
crime rates in areas of concentrated disadvantage was due to the neighborhood's lack of
ability to exercise informal social control, e.g., dispersing a group of loitering teenagers.
While Sampson agreed that the breakdown of community controls was rooted in
structural conditions, he further argued that instead of a natural process of large urban
city maturation, variation in levels of disorganization are closely linked to racial
This revised form of social disorganization has been used throughout
criminological literature. Specifically, Sampson et al., (2001) found that concentrated
disadvantage in combination with low levels of informal social control and kinship
networks increased rates of homicide. Snell (2001) found that poor urban neighborhoods
suffered from higher crime rates, as explained by higher levels of disorder and lower
levels of friendship and kinship networks (Snell, 2001). Lowenkamp et al., (2003)
replicated a prior experiment by Sampson and Groves, analyzing crime rates in the UK;
their results provided further support for the validity of this perspective (Lowenkamp et
Gibson, Zhao, and Lovrich found that as perceptions of high levels of collective
efficacy increased, fear of crime and victimization were significantly decreased (Gibson
et al., 2002). Lane and Meeker (2003) used the perspective to investigate fear of crime
levels in urban neighborhoods (Lane and Meeeker, 2003). Another study conducted by
Browning (2002) extended the perspective to intimate partner violence, showing that
higher levels of collective efficacy and informal social control resulted in lower levels of
violence among partners, and further increased the likelihood that female victims would
report violence to police (Browning, 2002). Finally, Sun et al., (2004) applied social
disorganization to their analysis of perceptions of police and local government, finding
that as concentrated disadvantage increased, perceptions of police became more became
more negative (Sun et al., 2004).
Social Disorganization and Racial Profiling Linkages
According to social disorganization perspective, residents of highly disorganized
areas lack the ability to unite and realize common goals and initiatives; lack of these
social networks leads to a breakdown of informal social control, and thus higher crime
rates. As a result, neighborhood residents must rely on formal social control, specifically,
police officers. Higher rates of police presence increase the likelihood of traffic stops,
and consequently, the rates of searches. Officers' perceptions of neighborhood
characteristics may also influence behavior independent of crime rates. For example, a
concentration of poverty as manifested by vacant housing units, and unsupervised youth
"hanging out" on streets are each indicators of social disorganization that may cause an
officer to become suspicious of local residents. The presence of these indicators may
lead officers to believe residents, regardless of race, are more likely to be involved in
crime, and therefore are less likely to practice racial profiling. Therefore, we theorize that
as measures of social disorganization increase rates of both African American and white
searches will increase. There should be no racial disparity in search rates in highly
* Hypothesis Three: As indicators of social disorganization increase, we expect to
find an increase in search rates regardless of race.
* Hypothesis Four: The impact of social disorganization indicators on search rates
will not differ by racial group.
The validation of these hypotheses would point to the efficiency of social disorganization
perspective in explaining higher rates of searches in highly disorganized neighborhoods
for both white and African American residents.
Research on police behavior has shown that officers have a tendency to act more
punitively toward minority group members, particularly African Americans. To be sure,
police during the 1960s and 1970s shot and killed 7 African American suspects for every
1 fatal shooting of a white suspect (Alpert and MacDonald, 2001). Research findings
vary, but have shown that officers are more likely to use force or brutality on suspects
being arrested or showing disrespect (Terrill and Reisig, 2003). Further research notes
that African Americans are more likely to have a disrespectful demeanor while in contact
with police (Terrill and Mastrofski, 2002, and Mastrofski, Reisig, and McCluskey, 2002).
Therefore, it follows that officers may exhibit higher levels of use of force with African
Americans as compared to whites.
* Hypothesis Five: As police use of force reports increase, so too will African
American search rates, but will have no impact of white searches.
DATA AND METHODS
It is the goal of this research to determine whether there is a clear link between
characteristics of a geographic location and officers' race-specific search behaviors. In
an attempt to compare the contributions of both urban disadvantage and social
disorganization perspectives to an explanation of racial profiling in context, this research
will use indicators of urban disadvantage and social disorganization to test whether they
independently explain higher rates of either race-specific or class-specific searches based
on characteristics of the census tracts. To allow for an appropriate analysis of these
factors, data include actual police-citizen contact data, census data including measures of
various census tract characteristics, and crime data specific to the census tracts under
study in Miami-Dade County.
Unit of Analysis
Given this paper's focus on urban disadvantage, social disorganization and macro-
level characteristics influencing race-specific search behaviors of police officers, all data
included are aggregated to the census tract level. This allows for a clearer picture of
neighborhood characteristics. Miami-Dade County is composed of 348 census tracts.
While the census tract is not explicitly considered a neighborhood, census tracts have
been used in extant research as the unit of analysis because they are more likely to
correspond to actual community boundaries (Krivo and Peterson, 1996, and Land et al.,
Miami-Dade County offers a particularly ideal domain in which to study racial
profiling because of the area's divergent racial and ethnic make-up. The Miami-Dade
population is composed of approximately 68% white residents and 22% African
American residents; further, approximately 54% of all residents are of Hispanic origin.
The ethnic diversity of the area under study has also increased by 8.3% from 1990 to
2000 and continues to grow (Parker et al., 2004). Table one lists descriptive statistics of
the Miami-Dade area as well as that of all variables used in this analysis.
This data set was compiled by the Miami-Dade Police Department in cooperation
with academic researchers in an effort to detect and identify patterns of racial profiling in
police behaviors. The data were collected from April 1, 2001 to October 31, 2001.
Composed of 66,109 traffic stops, the data set includes geo-coding of stop locations,
citizen characteristics, reason for stop, and the events that transpired after the initial stop,
aggregated to the census tract. However, due to the inability to geo-code some incidents,
the original sample is reduced by 7.3%, n = 61,255.
Citizen Contact Data
It was necessary to collect information on all stops that occurred in Miami census
tracts during the data collection period. The Miami-Dade Police Department does not
require information to be collected on each stop. Instead, they require police officers to
call in a specific code to dispatch when making a traffic stop. Unfortunately, this does
not allow for the collection of information as detailed as is necessary to study racial
profiling. To get a clear picture of the events that occurred before, during and after the
police stop, it would be necessary for officers to record detailed information on each stop
To alleviate this issue, the local City Commission appointed an Advisory Board to
help determine proper information to be used for inclusion on the Citizen Contact Forms,
forms that each officer would be required to complete after making a traffic stop. With
suggestions from the academic research group, the Advisory Board determined the
information to be included on the Citizen Contact Forms based on the two primary forms
of data considered necessary; the research group needed to know who was being stopped
and the outcomes of those stops. Examples of information collected on Citizen Contact
Forms included reason for making the stop, driver race, driver age, driver residence, and
numerous other details pertinent to the detection of racial profiling.
The officers of Miami-Dade County collected a total of 86,232 Citizen Contact
Forms. Due to consistency and training issues, the forms collected during February and
March were not analyzed. The forms analyzed were those collected between April 1,
2001 and October 31, 2001. They were sent to Miami-Dade Police Department
Headquarters where county employees entered the data by hand. Forms that contained
incorrect or missing data were returned to the officers as needed and re-collected. A
series of recoding and recurrent validity testing ensures the quality and consistency of the
cards with actual police records. As a result, officials could not include only 118 forms.
This analysis includes data from 66,109 Citizen Contact Forms, and is reduced to 61,255
traffic stops after accounting for missing data.
The dependent variable is limited to the natural logs of two racially disaggregated
counts of search rates per 10,000 log of black searches per 10,000 and log of white
searches per 10,000. The former is computed as the log of the total number of black
searches divided by the black population per 10,000. The latter is computed as the log of
the total number of white searches divided by the white population per 10,000. To get a
clear indication of search behaviors, all forms of searches were included in the analysis;
each search-related variable is coded as 1 = Yes and 2 = No. The survey included a
number of search-related variables, including search of person, search by consent,
inventory search, search of vehicle, search of driver, and search of passenger. It was
necessary to use the log of these rates because of the high variation about the means of
each indicator. In using the log of the search rates, the distributions become more
normal, and thus more appropriate for linear regression analysis. Further, the use of the
natural log reduces heteroscedasticity among variables (Land et al, 1990, and Oussey,
1999). To avoid zero counts as being counted as missing data, one unit was added to
each incident with a zero count, and then transformed to the natural log, again equaling
To calculate race-specific indicators, variables selected for analysis include the
following: race, coded as 1 = American Indian or Alaska Native, 2 = Oriental/Asian, 3 =
Black, 4 = Native Hawaiian or Pacific Islander, and 5 = White. The data were arranged
to select out only searches with a code of 3 for black searches, and 5 for white searches.
Appendix A defines the operational properties of each variable used in the analysis.
The measures included in this data set were retrieved from the 2000 Census,
downloaded from the Census Bureau webpage (www.factfinder.census.gov), and Miami-
Dade County and modified to accommodate this research. The data are used to estimate
social, economic, and demographic measures at the census tract level, which have been
used as indicators in the measurement of the urban disadvantage perspective and social
Urban Disadvantage Indicators
Variables commonly used to reflect urban disadvantage are employed in this
research and include race-specific measures of percent persons living below poverty
level, defined as percentage of white/black persons living below the poverty level in the
total population, and race-specific measures of percent not employed, age 16 and over,
defined as percent white/black not employed age 16 and over in the total population.
Each of these measures have been included in analysis of urban disadvantage in extant
research (Land et al., 1990; Ousey, 1999; Parker and McCall, 1999; and Stretsky et al.,
2004). Percent of black and white female-headed households with children under the age
of 18 is defined as the number of racially disaggregated households that are female-
headed with children under the age of 18. This final measure was included as a
disadvantage indicator because it is a condition that is nearly unique to African American
families; African Americans are more than twice as likely to experience the condition of
female headship, as compared to whites. It is appropriate to include female-headed
households with children under 18 in this index, as opposed to its inclusion as a measure
of social disorganization (Land et al., 1990; and Ousey, 1999).
Social Disorganization Indicators
Indicators of social disorganization include percent of Spanish-speaking population
speaking English not well or not at all, percent Hispanics in the total population,
residential mobility, percent of vacant housing units, and percent persons who journey to
work 25 minutes or more, and are defined below. Percent Spanish-speaking population
speaking English not well or not at all, age 5 and above is defined as the percentage of
Spanish-speaking persons speaking English not at all or not well in the total population,
ages 5 and above. Percent Hispanics in the total population is defined as the percentage
of individuals of Hispanic descent in the total population. These measures are included
to provide a better understanding of potential communication issues among residents and
therefore, a lack of potential for collective efficacy and informal social control.
Residential mobility is defined as having a different residence within the past 5
years. This variable is used as an indicator of neighborhood residential stability.
According to Sampson, neighborhood instability weakens the ability and desire of
residents to understand their collective goals (Sampson, 1987). As residential mobility
increases, it becomes more likely that the neighborhood experiences higher levels of
disadvantage. Percent of vacant housing units is defined as the percentage of vacant
housing units as compared to the total number of housing units.
Percent persons who journey to work 25 minutes or more is defined as percent
persons of a given census tract who travel to work 25 minutes or more. This measure was
included to account for local social disorganization through a lack of available local
employment. According to Wilson, the relocation of manufacturing jobs to the suburbs
caused unemployment rates to increase substantially, due to inner city workers'
geographic proximity to available employment (Wilson, 1987). Therefore, this measure
is computed as an indicator of social disorganization.
The Miami data set also includes information on officers employed by the Miami-
Dade Police Department during the time of the study. Information on each of the 1,659
officers in the citizen contact database was included; examples consist of race, age, rank,
sex, and the like. Data are also provided by Professional Compliance Bureau, which
includes information on disciplinary actions, complaints and use of force reports, limited
to the previous 5 years (1997-2000), and aggregated to the census tract. All identifiers
were removed from the data set to ensure anonymity. We include use of force reports,
defined as the rate of use of force reports an officer has received within the past 5 years.
As was the case with the dependent variable, it was necessary to use the natural log of the
rates of use of force reports due to significant variation about the mean.
Crime data was provided by the Miami-Dade Police Department, spanning 5 years
from 1997 to 2001. This data set includes both Part I and II crime arrests for the time
span. The file contains data on 121,763 cases, and includes the following information:
incident date, Uniform Crime Report coding, UCR descriptions, details of the incident,
incident location, police district and grid information, and census tract identifiers. Using
the crime data, I calculated the total and race-specific rates per 10,000 for each type of
crime based on the arrest records spanning that time period, all of which are aggregated
to the census tract level; the types of crimes include homicide, violent, property, and drug
crimes, as well as a total crime rate.
Homicide rate per 10,000 is a racially disaggregated measure of violent crime, and
is defined as the black/white homicide rate per 10,000. Research has consistently shown
a link between racial composition of a given neighborhood and violent crime rates
(Robison, 1936; Werthman and Piliavin, 1967; Smith, 1986; and Parker and McCall,
1999). If an officer knows that he or she is making a stop in an area with higher
homicide rates, the officer may be more likely to act in a more punitive fashion, as
manifested by a search.
Racially disaggregated rates of arrest were also included as controls, and are
defined as rates at which an arrest was made of a black or white per 10,000 stops. The
arrest rates were calculated from the variable, Arrest Made, and coded as 1 = Yes, and
2 = No. All types of police arrests are included, with the exception of custody arrest, in
which an individual is not truly arrested, but taken into police custody.
Percent of males in the total population is used as a control measure due to their
increased likelihood of participation in criminal activity and an increased likelihood of
arrest by police officers as compared to females. It is defined as the percentage of males
in the total population.
Finally, we control for the rate investigative stops, defined as stop resulting from
officer suspicions, to detect tendencies for police to be more suspicious in certain
contexts as compared to others. We created a variable to compute the percentage of
investigative stops; all investigative stops were computed from the variable Reason for
stop, coded as 1 = investigative, 2 = traffic violation, 3 = equipment violation, 4 = BOLO
(be on look-out), 5 = other, 6 = hazardous moving violation, and 7 = non-hazardous
Table 4-1 Descriptive statistics: Means (and standard deviations) for variables in race-
specific search models
Log Search Rates per 10,000 (1.82) (1.52)
Urban Disadvantage Indicators
Urban Disadvantage Index (30.63) (19.7
Social Di i :,'l:, t. J Indicators
S T70.79 71.23
Hispanic Immigration Index 7.7 .
Residential Mobility (2.57
% Vacant Housing Units (8.04)
% Who Travel to Work > 25 Minutes 51.46
or More (10.66)
Officer Conduct Variables
Log Officer Use of Force Report 4.27
Homicide Rate Per 10,000 (3.6) (3.
% Males in Total Population (4.85)
Investigative Stop Rate (73.44)
Population Composition ofMiami-
% White Population (30.121)
% African American Population (29.871)
S n 54.364
% Hispanic Population (29.112)
Principle Component Analysis
Structural indicators tend to be highly correlated. Therefore, to avoid problems
with multicollinearity, we performed a principle components analysis on indicators of
urban disadvantage and social disorganization (Land, et al, 1990, Parker and McCall,
1999, and Stretsky, Schuck and Hogan, 2004). The factor loadings produced separate,
theoretically distinct indices of each perspective. Further, the factor loadings were the
same for both white and black models. The principle components analysis matrix after
varimax rotation is provided in Appendix B.
The urban disadvantage index includes three measures of disadvantage that are
consistent with extant literature as indicators of urban disadvantage, disaggregated by
race (Krivo and Peterson, 1996; Wilson, 1987; Massey and Eggers, 1990; Massey, 1990;
and Ousey, 1999). The race-specific measures of percent living below poverty level,
percent not employed, age 16 and over, and percent of female-headed households with
children under 18. All were highly correlated and thus loaded together, creating the black
and white disadvantage indices.
The Hispanic immigration index was created from the loadings of the following
measures related to the Hispanic population: percent Spanish-speaking population
speaking English not well or not at all, age 5 and above, and percent Hispanics in the
total population. Therefore, two conceptually distinct indices emerged from the analysis
and are treated as independent variables in the multivariate models: The Urban
Disadvantage Index, disaggregated by race, and Hispanic Immigration Index.
This analysis contains two separate models estimated theoretically in white and
black indicators on race-specific search rates in the total sample. As was stated
previously, it was necessary to perform log transformations for some variables so that
their distributions were more normal, and to reduce heteroscedasticity among
independent variables, thus making the data more appropriate for linear regression
analysis. Each of the model regression equations is estimated from the same unit, which
may result in cross-equation error correlations. Therefore, seemingly unrelated
regression (SUR) was chosen as the most appropriate test for the analysis. This method
uses generalized least squares to estimate each individual model regression equation. It is
designed to account for correlations in error terms across equations so that comparisons
between models can be made without encountering issues with collinearity while using
measures taken from the same sampling unit (Greene, 1996, Land et al., and Ousey,
1999). Finally, I employed a Breusch-Pagan test of independence to ensure that
heteroscedasticity among variables would not be significant enough to affect analysis.
The test revealed no evidence of heteroscedasticity (chi2 = 1.619, p > 0.05).
Because the theoretical arguments and hypotheses propose differences across racial
groups, it is necessary to conduct a test for race-specific differences in the magnitude of
the effects of each indicator on black and white models. Essentially, we needed to
conduct a test measuring the variance of the effects of indicators across the models. To
obtain this information we computed the F statistic through the SUR analysis, which
measures the significance of the differences of coefficients across models. It is necessary
to employ such a test because other options of statistical analysis tend to be negatively
biased; this effect appears because certain types of analysis commonly used simply
compare regression coefficients (Paternoster, et al., 1998). The results from the
seemingly unrelated regression analysis and F test are reported below in Table 2. A brief
explanation of findings follows.
Total Sample Analysis
Turning first to indicators related to urban disadvantage, we see that the
disadvantage index in the black model was found to have a negative but non-statistical
effect on the rate of black searches. This finding is contrary to the predictions of
Hypothesis 1 that higher scores on the index would result in higher rates of searches of
African Americans. Hypothesis 2, that increased levels of urban disadvantage would
have no effect on rates of white searches, was also not supported by the results of the
analysis. Again, we find that the reciprocal of Hypothesis 2 was the outcome. The
results show a positive and significant relationship between the rates of white searches by
police and the white disadvantage index. That is, as levels of white urban disadvantage
increase, the rate of police searches of white drivers also increases. Further, the F
statistic reveals that the difference in these coefficients across models is statistically
We now consider the hypotheses related to the social disorganization perspective.
While we offer multiple measures of social disorganization, no measures reach statistical
significance in the black models. Only one measure was found to have a significant
impact on rates of white searches percent vacant housing units. In disorganized areas,
whites are not likely to be searched by police, meaning that as the percentage of vacant
housing units increase, the rate of white searches by police actually decrease. Hypothesis
Table-5-1 Seemingly unrelated regression
the total sample
(SUR) coefficients (and standard errors) for
Black White Coeff. Comparison F
(and P) Values
Urban Disadvantage Indicators
Urban Disadvantage Index
Hispanic Immigration Index
% Vacant Housing Units
% Who Journey to Work 25
Officer Conduct Variables
Log of Officer Use of Force
Homicide Rate Per 10,000
% Males in Total Population
Investigative Stop Rate
Model Chi Square
*P < 0.05 ** P < 0.01
3, that these factors will increase both black and white search rates, is therefore not
supported by the analysis. However, we do find some evidence supporting Hypothesis 4
that these factors will impact the black and white models similarly. The F statistics show
that social disorganization indicators do not differentially affect the white and black
models. That is, measures of disorganization have a similar effect on rates of black and
The log of officer use of force reports has a significant and positive effect on rates
of both black and white searches. Its effect on white search rates is stronger than its
effect on black search rates, however, giving us 99% confidence and 95% confidence
respectively that as reports of use of force increase, so too do rates of white and black
searches by police. Therefore, Hypothesis 5 is partially supported; while officer use of
force reports do have a positive and significant effect on the black search rate, it also was
found to increase white search rates. However, there is clear evidence of racial variance
in the impact of these coefficients concerning the relationship between force and race-
specific search rates. The F statistic shows that the impact of police use of force rates
differs significantly by race.
Concerning the controls, we find that the homicide rate per 10,000 does not impact
rates of race-specific searches, and there is no evidence of racial variance in the way
homicide rates affect each model. The arrest rate was highly significant in both models,
giving 99% confidence that as arrest rates increase, so too will rates of race-specific
searches. However, this effect may have an indirect relationship the impact of inventory
searches, a procedure always completed upon all arrests. That is, when an officer
conducts a search during a traffic stop, the decision to search was based on the officer's
discretion. An inventory search is part of police arrest protocol, which is not considered
a search that can be motivated by racial bias, and thus may inflate the actual rates of race-
specific searches by officers. Interestingly, the arrest rate differentially affects rates of
race-specific searches, a finding that should be investigated in future research. We find
that the percentage of males in the total population had no effect on race-specific searches
by police, and this measures was without evidence of racial variance. Finally, the
investigative stop rate had a positive and significant effect on searches by police in both
models. That is, when an officer makes a stop for investigative purposes, it increases the
likelihood of a search for whites and blacks alike. However, the impact of this control
does not differ by race.
In summary, the findings from the preliminary analysis provide modest support for
both urban disadvantage and social disorganization hypotheses. The urban disadvantage
index was found to have a statistically significant, positive effect on white searches by
police, while having a negative and non-significant effect on black searches. None of the
social disorganization measures used in the current analysis resulted in statistical
significance in the black model. A single measure of social disorganization, percent
vacant housing units, was found to have a negative impact on rates of white searches.
However, consistent with the social disorganization hypotheses, we found evidence of
racial invariance. That is, social disorganization measures affect both black and white
search rates similarly. Finally, turning to police use of force reports, we found that the
hypothesis was partially confirmed. This finding suggests that increases in use of force
reports have a positive and significant impact on not only rates of black searches, but also
rates of white searches by police.
Analysis by Area Type
To check for differences across types of areas, disadvantaged and non-
disadvantaged, a supplemental analysis was conducted, using the same set of hypotheses
from the primary analysis. The disadvantaged model selected out census tracts having at
least 30% of African American residents living below the poverty level. The non-
disadvantaged areas were composed of those tracts with less than 30% of African
Americans living below the poverty level. Poverty level cut-off points vary in the
literature, but researchers have typically used a percentage between 20% defined as low,
and 40%, defined as extreme. Therefore, a high African American poverty level of 30%
was used as the cut-of point to distinguish disadvantaged areas from non-disadvantaged
areas (Wilson, 1987, Massey, 1990, Massey and Eggers, 1990, Krivo and Peterson, 1996,
and Stretesky, Schuck, and Hogan, 2004). Disaggregating the data by area type resulted
in n = 95 disadvantaged census tracts, and n = 170 non-disadvantaged census tracts.
Descriptive statistics for the supplemental analysis are listed in Appendix C. The results
from the SUR analysis and F tests are given in Table Three.
The focus on African American poverty is consistent with urban disadvantage and
social disorganization perspectives in that structural factors causing concentrated poverty
are said to disproportionately affect this minority group as compared to structural effects
on whites. Therefore, we find it appropriate to disaggregate census tracts based on the
proportion of African American residents living in poverty in a given census tract for
both the black and white models. It allows for the isolation of these areas in which
African Americans are affected disproportionately by factors that create conditions
conducive to high poverty concentrations.
Concerning the urban disadvantage indicators in the non-disadvantaged model, we
find that neither is significantly affected by the urban disadvantage index. Again, we see
that these findings do not support Hypothesis 1, that urban disadvantage indicators will
have a significant, positive impact on search rates of African Americans. However,
results do lend some support to Hypothesis 2, that white searches will not be affected by
increases in urban disadvantage. The F statistic further reveals no evidence of racial
variance. That is, differences between coefficients illustrate that the impact of the urban
disadvantage index does not differ by racial group.
For measures of social disorganization, we find that the impact of these indicators
on race-specific searches is not statistically significant, with the exception of percent
vacant housing units. The coefficients are identical for percent of vacant housing units in
the white model, while no other significant relationships were identified. The black
model again, shows no significant relationships among social disorganization indicators
and rates of black searches. These findings are contrary to the predictions of Hypothesis
3, which states that as social disorganization indicators increase, so too will rates of race-
specific searches by police. However, the F statistic reveals that no indicators
differentially affect the white and black models, which lends some support to Hypothesis
4 that the impact of social disorganization indicators will not differ by racial group.
The log of officer use of force reports has a different relationship to rates of black
and white searches by police in non-disadvantaged areas. The black model indicates no
significant relationship between police use of force and rates of black searches.
However, the white search rate is significantly affected by the indicator, giving 99%
confidence that as police use of force reports increase, so too do rates of white searches
Table-5-2 Seemingly unrelated regression (SUR) coefficients (and standard errors) for
non-disadvantaged and disadvantaged areas
Non-Disadvantaged Areas Disadvantaged Areas
N = 170 N = 95
Black White Comparison F Black White Comparison F
(and P) Values (and P) Values
Urban Disadvantage Indicators
Urban Disadvantage Index
Social L', i,:,,, ,o ,i.
Hispanic Immigration Index
% Vacant Housing Units
% Who Journey to Work 25
Log of Officer of Use of Force
Homicide Rate Per 10,000
% Males in Total Population
Investigative Stop Rate
Model Chi Square
*P < 0.05 ** P < 0.01
by police. Finally, we see that the F coefficient is significant; officer use of force reports
do, in fact, differentially affect the black and white models. Again, we do find partial
support for Hypothesis 5 in the differential effects of this indicator on rates of black and
The homicide rate per 10,000 was found to have no effect on rates of race-specific
searches. The arrest rate was found to have a highly significant effect on both rates of
race-specific searches, while the F statistic reveals significant differences in the impact of
arrest rates by racial group. Percentage of males in the total population had no effect on
race-specific searches, and no racial differences in its influence were identified. The
investigative stop rate was found to positively influence the rates of black and white
searches. The investigative stop rate has a slightly stronger, but statistically non-
significant, positive impact on rates of white searches by police, as compared to its
influence on the black search rate. However, the F statistic reveals no evidence of
variation by race.
Similar to findings from the primary analysis, we find that the urban disadvantage
index has a positive and significant relationship to rates of white searches; however, we
do see a slight reduction in significance from the primary model to the current model. No
significant relationship between the disadvantage index and rates of black searches was
identified. Further, based on the statistic from the F test, we see no evidence that the
disadvantage index differentially affects the black and white models. Again, these
findings lend no support to Hypothesis 1 and Hypothesis 2.
Indicators of social disorganization have no significant effect on either race-specific
search rates. These results are in opposition to Hypothesis 3 that both black and white
search rates will increase. However, we again find support for Hypothesis 4, as there is
no evidence of a differential effect of the indicators on either model.
Finally, we find significant the influence of the log of officer use of force report
rates for each model. The relationship between the indicator and black searches is
positive and significant, with 95% confidence. As in previous models, we see that the
influence of officer use of force reports is stronger for white search rates, in this case
yielding 99% confidence that as use of force report rates increase, white searches by
police will also increase. However, we do not see a difference in the way this indicator
affects each model. The F statistic reveals that the impact of officer use of force reports
similarly affects both black and white search rates, a finding we do not see in the non-
The impact of two controls in the disadvantaged areas is slightly different from the
non-disadvantaged areas. Consistent with findings from the non-disadvantaged areas, we
see that homicide rate per 10,000 has no significant influence on race-specific searches,
without evidence of racial variance. The arrest rate is continuous across models as well,
having a highly significant impact on rates of black and white searches, while showing
considerable evidence of racial variance. However, the percentage of males in the total
population, while having no impact on searches, was found to have a different influence
on each model. An additional difference was found in the impact of the investigative
stop rate on black searches by police, as the coefficient increased in significance for the
black model, while the impact on white searches was the same as in the non-
disadvantaged model. Racial variance, however, was again not identified.
Results from the supplemental analysis were similar to findings from the
preliminary analysis. The urban disadvantage index had a significant and positive effect
on rates of white searches, but not black searches, in the disadvantaged model only, while
the index had no statistically significant impact in the non-disadvantaged model. We
again see no significant impact of social disorganization indicators on black search rates
in both disadvantaged and non-disadvantaged areas or on the white search rates in
disadvantaged areas. However, we see positive statistical significance in the impact of
percent vacant housing units on rates of white searches by police in the non-
Finally, we see a decrease in the impact of officer use of force reports for black
search rates in the non-disadvantaged model. We also find statistical significance in the
magnitude of the F statistic, meaning that the impact of use of force reports is different
based on racial group. The disadvantaged model also yields results similar to those from
the preliminary analysis; use of force provides a 95% confidence interval for its impact
on black search rates, and a 99% confidence interval for its impact on the white searches,
as well as a change in the F statistic indicating no evidence of racial variance. That is, the
impact of use of force reports on black and white search rates by police becomes more
similar in the disadvantaged model.
DISCUSSION AND CONCLUSIONS
The goal of this research was to determine the effects of urban disadvantage and
social disorganization on racial profiling as manifested through a police search
subsequent to a traffic stop. For the analysis, we compared indicators of social
disorganization and an urban disadvantage index to police rates of black and white
searches. Overall, the results indicate that these perspectives do not significantly
contribute to our knowledge of the causes of racial profiling.
The urban disadvantage index significantly influenced white search rates in the
total sample, while it had no impact on search rates of African Americans. This is the
case across each black model; the findings held when relationships were estimated by
non-disadvantaged and disadvantaged area types, thus exhibiting the robustness of the
findings from the analysis. The F statistic generated shows that the impact of the urban
disadvantage index differs by race. This is also consistent with findings from previous
studies, which included similar measures. Krivo and Peterson (2000) found in their
analysis of black and white structural conditions and homicide rates, that concentrated
disadvantage was 1 of only 2 variables that had significant racial differences. Perhaps
this is a result of a significantly lower likelihood for whites to live in concentrated
disadvantage (Krivo and Peterson, 2000). It then follows that when whites do live in
areas of concentrated disadvantage, their economic and social situations are at their
worst. Therefore, a possible explanation of this finding may be that white searches by
police increase with increases in concentrated disadvantage because when whites
experience conditions such as these, the impact of disadvantage is exacerbated, resulting
in whites living in the poorest and most dangerous communities.
The measures of social disorganization showed no statistically significant impact
on the black model, while percent vacant housing units had a negative and significant
impact on rates of white searches by police. Krivo and Peterson (1996) found that higher
vacancy rates as well as higher rates of rental occupancy have positive effects on rates of
property crime. Moreover, this effect was found to have a more significant impact on
white neighborhoods than in predominantly black neighborhoods with extremely low
numbers of professional residents. This finding again points to the extremely different
structural living conditions of poor whites and poor blacks. As I explained above, the
likelihood of poor whites to live in the poorest areas is so low that when this condition
does occur, whites are living in the worst of areas. It may be more common to find a
higher number of professionals, who tend to be more stable in their living situations,
residing in more populated, poor white neighborhoods because of a higher likelihood of
better education, and thus, betterjobs. However, in the worst conditions, whites are
residing in areas composed of residents of the lowest socioeconomic status, therefore,
having a lower likelihood of professional residents, and thus higher levels of vacancy
(Krivo and Peterson, 1996).
Looking at the descriptive statistics from the supplementary analysis (see Appendix
C), we see that African Americans are more likely to be searched by officers in non-
disadvantaged areas, while whites are more likely to be searched in disadvantaged areas.
These statistics are in opposition to the assumptions of the hypotheses, that indicators
consistent with each neighborhood type would have a strong influence on black searches,
but not white searches in the case of urban disadvantage, or that they would have a
similar effect on both black and white search rates in the case of social disorganization.
Police officers have a great deal of discretion in the field; the decisions they make
concerning law enforcement are shaped by, among other things, past experiences. Police
officers are trained to look for things that are out of the ordinary (Smith, Makarios, and
Alpert, Forthcoming). African Americans are more likely to reside in areas having
characteristics that typify urban disadvantage, while only 2% of the poorest whites live in
neighborhoods with an extreme concentration of poverty (Wilson, 1987, and 1992). It
then follows that a white person in a particularly poor or predominantly African
American neighborhood may prompt an officer to suspect criminal activity, and
subsequently stop and search that individual because they don't fit (Smith, Makarios, and
Alpert, forthcoming). Therefore, we may expect an increase in white searches as we
move into more disadvantaged areas, as is illustrated in the supplementary models.
As I stated before, it is not likely that whites live in areas of concentrated poverty
and high crime. When whites do reside in these areas, it is possible that they are put in a
different social class, separate from poor whites living in more affluent neighborhoods.
Perhaps the social distance between these two groups created by the difference in
neighborhood residence is enough to influence police suspicions and thus, increase white
The racial threat hypothesis offers another potential explanation for the results. It
claims that the percentage of minorities in a given area is positively related to the level of
formal social control. If segregation is imposed and minority groups are concentrated in
certain areas, they become less of a threat, resulting in less police activity (Liska and
Chamlin, 1984, and Terrill and Mastrofski, 2002). Perhaps police focus more on whites
in these areas because they come to expect certain types of suspicious or criminal
behaviors from blacks in disadvantaged areas. Black crime becomes commonplace,
making officers less sensitive to African American presence, and thus more attentive to
suspicious behaviors of whites.
The benign-neglect hypothesis argues that larger African American populations
have a higher likelihood of intra-racial crimes, which are less likely to be handled via
legal response. This is because intra-racial crimes do not pose a threat to the majority
(Liska and Chamlin, 1984, and Smith and Holmes, 2003). Again we see potential for
more focus on white crime in areas of high disadvantage. According to Black's theory of
law, whites are higher in status than African Americans because they compose the
majority. We may therefore see higher levels of legal response to suspicious behaviors of
whites because higher status elicits more law (Doyle and Luckenbill, 1991).
The officer use of force report rate was found to be significant in both the white and
black models. However, we see that it has a stronger impact on white search rates, and as
demonstrated by the F statistic generated, there is significant variation in the way this
variable affects whites and African Americans. The racial difference in the impact of this
variable is contradictory to existing literature on the determinants of use of force and
neighborhood context. Consistent with extant literature, we hypothesized that higher
rates of officer use of force reports would increase the likelihood of African American
search rates, but not white search rates. Numerous studies show that officers are more
likely to use force against minority suspects than whites (Smith, 1986, Smith, 1987, Liska
and Chamlin, 1984, Doyle and Luckenbill, 1991, Terrill and Mastrofski, 2002, Terrill and
Reisig, 2003, and Smith and Holmes, 2003).
Turning to the supplementary analysis we see that in the non-disadvantaged
models, use of force rates show no impact on African American searches, while it has a
significant impact on white searches. The disadvantaged models show that both search
rates are influenced by use of force rates, but there is a stronger impact in the white
model. Perhaps these findings reflect levels of aggression in police officers. As
previously stated, officers look for things that are out of the ordinary. It may make sense
that a majority of officers are searching African Americans in more affluent
neighborhoods because they are less likely to take residence there. However, it appears
that for a white to be searched in an affluent area, it takes a particularly aggressive officer
to make that decision one that has a higher occurrence of use of force reports.
The findings of the non-disadvantaged models may be imparted to the total sample
models and the disadvantaged models. Essentially, all officers are doing theirjobs by
stopping and searching African Americans, but only some officers are searching whites.
This may be evidence of racial profiling because it assumes African Americans are
committing crimes, as evidenced by higher rates of use of force when initiating the search
of a white suspect, while it is less likely for the search of an African American to be
conducted by officers who have high rates of use of force.
The potential contributions of this research are many. This is one of only a small
number of studies on racial profiling conducted using such a large data set, containing
such detailed information on police officers, citizens, and stop information. Further, the
data were gathered using a combination of sources outside of the individual level data,
including the 2000 census, and the Uniform Crime Report, allowing for a more thorough
investigation of the problem (Lundman, 2004). This research is theory-based, which is
unique to this field. Extant research has generally neglected to incorporate theory prior to
statistical analysis (Engel, Calnon and Bernard, 2002). As a theory-based research paper,
the units of analysis were ultimately determined by tenets of social disorganization and
urban disadvantage in an attempt to explain the differential behaviors of police relative to
neighborhood context. This research also provides a comparison between the effects of
each context-related perspective, and may help to support one over the other in
explaining racial differences in search rates.
In designing this analysis, it was often difficult to discern which indicators were
more consistent with each perspective. Perhaps future research should consider using
measures of social disorganization that are more distinct from those of urban
disadvantage. Community characteristics such as collective efficacy may be one avenue
that can provide a clearer dichotomy between the two perspectives. Further, taking into
account characteristics of community-level institutions may provide an improved
conceptualization of a community's organization.
This study used racially-disaggregated measures of search rates by police as an
indicator of racial profiling. Due to the actual number of searches conducted during the
period of data collection, the counts for searches were low. Indeed, the counts were so
low and varied so significantly across census tracts that the natural log had to be used to
make the data appropriate for the regression analysis. The act of conducting a search is
highly discretionary for police officers because it is the officer alone who takes into
account various circumstances, conditions or cues that may warrant a search. Therefore,
looking at arrest rates as indicators of racial profiling as well as stop rates could provide a
clearer picture of racial profiling by police officers.
Perhaps racial profiling is a phenomenon that needs to be studied through more
micro-level theories. Future research should attempt to include more data on the static
characteristics of officers, and that of citizens stopped. The inclusion of an interview
component may help to inform the perspectives officers and citizens. Further, the data
were taken from a single geographical area, Miami-Dade County. Not only is this area
unique because of its ethnic make-up, but it is also a southern city, which may present
quite different characteristics and conditions from those in other regions of the US.
Further, research should focus on the comparison of a number of cities from different
regions to prevent regional bias. Future research should also seek to include in its
analysis a Hispanic component, especially when studying an area so ethnically diverse.
PRINCIPLE COMPONENTS ANALYSIS AFTER VARIMAX ROTATION
Black Model White Model
Variables Component Component
1 2 1 2
% Spanish-speaking population
who speak English not well or 0.97 0.96
not at all, age 5 and above
% Hispanics in Total Population 0.96 0.97
% White non-Hispanics living 0.9
below poverty level
% White non-Hispanic not
employed, age 16 and over
% White female-headed 0.53
household with children < 18
% Black persons living below
% Blacks not employed, age 16 0.90
% Black female-headed 0
household with children < 18
Eigen Values 2.15 1.79 1.9 1.73
% Variance Explained 43.01 35.69 38.09 34.49
DESCRIPTIVES: MEANS AND (STANDARD DEVIATIONS) FOR VARIABLES IN
SEARCH MODELS FOR NON-DISADVANTAGED AND DISADVANTAGED
N = 170 N = 95
Black White Black White
Log of Search Rates per 10,000
Urban Disadvantage Indicators
Urban Disadvantage Index
Social De i, :i.,t, ,, Indicators
Hispanic Immigration Index
% Vacant Housing Units
% Who Travel to Work > 25
Officer Conduct Variables
Log of Officer Use of Force
Homicide Rate Per 10,000
% Males in Total Population
Investigative Stop Rates
% White Population
% African American Population
% Hispanic Population
LIST OF REFERENCES
Akins, S. 2003. Racial segregation and property crime: Examining the mediating effect of
police strength. Justice quarterly, 20, 675-695.
Alpert, G., and Dunham, R. 1988. Policing multi-ethnic neighborhoods: The Miami study
and findings for law enforcement in the United States. New York: Greenwood Press.
Alpert, G. P., and MacDonald, J. M. 2001. Police use of force: An analysis of
organizational characteristics. Justice quarterly, 18, 393-409.
Bayley, D.H., and Mendelsohn, H. 1969. Minorities and the police: Confrontation in
America. New York: Free Press.
Borg, M. J., and Parker, K. F. 2001. Mobilizing law in urban areas: The social structure
of homicide clearance rates. Law & society review 35, 435-466.
Browning, C. R. 2002. The span of collective efficacy: Extending social disorganization
theory to partner violence. Journal of marriage &family, 64, 833-850.
Bursik, R. J. 1988. Social disorganization and theories of crime and delinquency:
Problems and prospects. Criminology, 26, 519-551.
Chambliss, W. 1994. Policing the ghetto underclass: The politics of law and law
enforcement. Social problems 41, 177-194.
Cole, D. 1999. Is public concern about feral police using racial profiling justified? Insight
on the news 15, 24-27.
Cordner, G., Williams, B., and Zuniga, M. 2000. Vehicle stop study: Mid-year report.
San Diego, CA: San Diego Police Department.
Cullen, F. T., and Agnew, R. 2003. Criminological theory: Past topresent. 2nd ed. Los
Angeles: Roxbury Publishing Company.
Doyle, D. P. and Luckenbill, D. F. 1991. Mobilizing law in response to collective
problems: A test of Black's theory of law. Law & society review, 25, 103-116.
Engel, R.S., Calnon, J.M., and Bernard, T. 2002. Theory and racial profiling:
Shortcomings and future directions in research. Justice quarterly, 19, 249-274.
Fagan, J., and Davies, G. 2000. Street stops and broken windows: Terry, race, and
disorder in New York City. Fordham urban law journal, 28, 457-504.
Florida Highway Patrol. 2000. Race study of the Florida highway patrol [Online].
Gibson, C. L., Zhao, J., and Lovrich, N. P. 2002. Social integration, individual
perceptions of collective efficacy, and fear of crime in three cities. Justice
quarterly, 19, 537-564.
Greene, W. H. 1996. Economic analysis. Upper Saddle River, NJ: Prentice-Hall, Inc.
Guerra, S. 1992. Domestic drug interdiction operations: Finding the balance. Journal of
criminal law & criminology, 82, 1109-1992.
Hagan, J., and Albonetti, C. 1982. Race, class, and the perception of criminal injustice in
America. American journal of sociology, 88, 329-355.
Harris, D. A. 1997. "Driving while black" and all other traffic offenses: The Supreme
Court and pretextual traffic stops. Journal of criminal law & criminology, 87, 544-
Jacobs, D., and Kleban, R. 2003. Political institutions, minorities, and punishment: A
pooled cross-national analysis of imprisonment rates. Social forces, 82, 725-755.
Kane, R. J. 2002. The social ecology of police misconduct. Criminology, 40, 867-897.
Krivo, L. J. and Peterson, R. D. 1996. Extremely disadvantaged neighborhoods and urban
crime. Social forces, 75, 619-650.
Krivo, L. J. and Peterson, R. D. 2000. The structural context of homicide: Accounting for
racial differences in process. American sociological review, 65, 547-559.
Lamberth, J. (1998). Driving while black: A statistician proves that prejudice still rules
the road. Hartford. [Online] Available:
Land, K. C., McCall, P. L., and Cohen, L. E. 1990. Structural covariates of homicide
rates: Are there any invariances across time and social space? American journal of
sociology 95, 922-963.
Lane, J., and Meeker, J. W. 2003. Fear of gang crime: A look at three theoretical models.
Law & society review, 37, 425-456.
Liska, A. E., and Chamlin, M. B. 1984. Social structure and crime control among
macrosocial units. American journal of sociology, 90, 383-395.
Lowenkamp, C. T., Cullen, F. T., and Pratt, T. C. 2003. Replicating Sampson and
Groves's test of social disorganization theory. Journal of research in crime &
delinquency, 40, 351-373.
Lundman, R. 2004. Driver race, ethnicity, and gender and citizen reports of vehicle
searches by police and vehicle search hits: Toward a triangulated scholarly
understanding. Journal of criminal law & criminology, 94, 309-349.
Lundmand, R., and Kaufman, R. 2003. Driving while black: Effects of race, ethnicity,
and gender on citizen self-reports pf traffic stops and police actions. Criminology,
MacDonald, H. 2003. Are cops racist? Chicago: Ivan R. Dee.
Massey, D. S. 1990. American apartheid: Segregation and the making of the underclass.
American journal of sociology, 96, 329-357.
Massey, D. S., and Denton, N. A. 1988. Suburbanization and segregation in US
metropolitan areas. American journal of sociology, 94, 592-626.
Massey, D. S., and Denton, N. A. 1993. American Apartheid: Segregation and the
making of the underclass. Cambridge: Harvard University Press.
Massey, D. S. and Eggers, M. L. 1990. The ecology of inequality: Minorities and the
concentration of poverty, 1970-1980. American journal of sociology, 96, 1153-
Massey, D.S., Eggers, M. L. and Denton, N. A. 1994. Disentangling the causes of
concentrated urban poverty. International journal of group tensions, 24, 267-316.
Mastrofski, S. D., Reisig, M. D., and McCluskey, J. D. 2002. Police disrespect toward the
public: An encounter-based analysis. Criminology, 40, 519-552.
Meehan, A. J. and Ponder, M. C. 2002. Race and place: The ecology of racial profiling
African American motorists. Justice quarterly, 19, 399-430.
Morenoff, J. D., Sampson, R. J., and Raudenbush, S. W., 2001. Neighborhood inequality,
collective efficacy, and the spatial dynamics of urban violence. Criminology, 39,
Ousey, G. 1999. Homicide, structural factors, and the racial invariance assumption.
Criminology, 37, 405-426.
Parker, K. F., and McCall, P. L. 1999. Structural conditions and racial homicide patterns:
A look at the multiple disadvantages in urban areas. Criminology, 37, 447-477.
Parker, K. F., Stults, B., and Rice, S. K. 2005. Racial threat, concentrated disadvantage
and social control: Considering the macro-level sources of variation in arrests.
Manuscript in preparation.
Paternoster, R., Brame, R., Mazerolle, P., and Piquero, A. R. 1998. Using the correct
statistical test for the equality of regression coefficients. Criminology, 36, 859-866.
Robison, Sophia M. 1936. Can delinquency be measured? New York, Pub. for the
Welfare Council of New York City by Columbia University Press.
Rose, D. R. and Clear, T. R. 1998. Incarceration, social capital, and crime: Implications
for social disorganization theory. Criminology, 36, 441-479.
Russell, K. 2001. Racial profiling: A status report of the legal, legislative, and empirical
literature. Rutgers race & the law review, 61, 1-16.
Sampson, R. J. 1987. Urban black violence: The effect of male joblessness and family
disruption. American journal of sociology, 93, 348-382.
Sampson, R. J., and Raudenbush, S. W. 1999. Systematic social observation of public
spaces: A new look at disorder in urban neighborhoods. American journal of
sociology, 105, 603-651.
Sampson, R. J. and Wilson, W. J. 1995. Toward a theory of race, crime, and urban
inequality. In J. Hagan and R.D. Peterson (Eds), Crime and inequality. Stanford:
Stanford University Press.
Smith, B. W., and Holmes, M. D. 2003. Community accountability, minority threat, and
police brutality: An examination of civil rights criminal complaints. Criminology
Smith, D. A. 1986. The neighborhood context of police behavior. In A.J. Reiss and M.
Tonry (Eds), Communities and crime. Chicago: University of Chicago Press.
Smith, D. A. 1987. Police response to interpersonal violence: Defining the parameters of
legal control. Social forces, 65, 767-782.
Smith, M., and Petrocelli, M. 2001. Racial profiling?: A multivariate analysis of police
traffic stop data. Police quarterly, 4, 4-27.
Smith, M. R., Makarios, M., and Alpert, G. P. 2005. Differential suspicion: Theory and
gender effects in the traffic stop context. Manuscript in preparation.
Snell, C. 2001. Neighborhood structure, crime, and fear of crime: Testing Bursik and
Grasmick's neighborhood control theory. New York: LFB Scholarly Publishing
Spitzer, E. 1999. The New York City Police Department's "stop and frisk" practices: A
Report to the people of the state of New York from the Office of the Attorney
General. Albany: New York Attorney General's Office.
Stretesky, P. B., Schuck, A. M., and Hogan, M. J. 2004. Space matters: An analysis of
poverty, poverty clustering, and violent crime. Justice quarterly, 21, 817-841.
Sun, I. Y., Triplett, R. A., and Gainey, R. R. 2004. Social disorganization, legitimacy of
local institutions, and neighborhood crime: An exploratory study of perceptions of
the police and local government. Journal of crime &justice, 2, 33-60.
Terrill, W., and Mastrofski, S. D. 2002. Situational and officer-based determinants of
police coercion. Justice quarterly, 19, 215-248.
Terrill, W., and Reisig, M. D. 2003. Neighborhood context and police use of force.
Journal of research in crime & delinquency, 40, 291-321.
Walker, S. 2001. Searching for the denominator: Problems with police traffic stop data
and early warning system solution. Justice research & policy, 3, 63-95.
Walker, S., Spohn, C., and DeLone, M. 2001. The color ofjustice: Race, ethnicity, and
crime in America. 2nd ed. United States: Wadsworth.
Ward, J. 2002. Race, ethnicity, and law enforcement profiling: Implications for public
policy. Public administration review, 62, 726-735.
Weitzer, R., and Tuch, S. 2002. Perceptions of racial profiling: race, class, and personal
experience. Criminology, 40, 435-457.
Werthman, C. and Piliavin, I. 1967. Gang members and the police. In D. Brodura (Ed)
The police: Six sociological essays. New York: John Wiley.
Wilson, G., Alpert, G., and Dunham, R. 2004. Prejudice in Police Profiling. American
behavioral scientist, 47, 896-909.
Wilson, W. J. 1987. The truly disadvantaged: The inner city, the underclass, and public
policy. Chicago: University of Chicago Press.
Wilson, W.J. 1992. Another look at the truly disadvantaged. Political science quarterly
Zingraff, M., Mason, H. M., Smith, W., Tomaskovic-Devey, D., Warren, P., McMurray,
H., and Fenlon, C. R. 2000. Evaluating North Carolina state highway patrol data:
Citations, warnings, and searches in 1998. North Carolina State University.
In May of 2003, I received my Bachelor of Arts in sociology from the University of
Florida. My senior thesis, Correlates of female juvenile delinquency, was published in
December of 2003 in The International Journal of Sociology and Social Policy. The
completion of my senior thesis allowed me to graduate with highest honors. I plan to
pursue my PhD in criminology at UF, continuing my research on racial discrimination,
social structure and crime, and race and poverty issues.