BIAS CRIME INCIDENCE IN UNITED STATES COUNTIES, 2000-2009: AN
APPLICATION OF SOCIAL DISORGANIZATION THEORY
By
Ryan Brevin Martz

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Criminal Justice – Doctor of Philosophy
2014

ABSTRACT
BIAS CRIME INCIDENCE IN UNITED STATES COUNTIES, 2000-2009: AN
APPLICATION OF SOCIAL DISORGANIZATION THEORY
By
Ryan Brevin Martz
This goal of this dissertation is to identify predictors of bias criminality in the United
States at the county level from 2000 - 2009. There is relatively little known about bias crime
occurrence in the United States. In addition, increased public attention to bias criminality
requires additional social science research examining the predictors of bias crime in American
communities. By examining traditional indicators of social disorganization theory, this
dissertation seeks to explore the likelihood of bias crime occurrence at the macro-level. As such,
the unit of analysis is United States counties. The N is 3,141. The data upon which this
dissertation is based come from the Federal Bureau of Investigation (FBI), the United States
Census Bureau (USCB), the Association of Religious Data Archives (ARDA), and
Congressional Quarterly’s Voting and Elections Collection.
From the data, measures of economic deprivation, social heterogeneity (diversity), social
cohesion, and residential mobility were created. These measures represent traditional indicators
of social disorganization theory. Four models are introduced in this dissertation in order to
answer several research questions that explore the differences between how these predictors
affect various types of bias crime. Negative binomial regression and OLS regression are used to
analyze the data and address the research questions. Specifically, anti-race motivated bias crime,
anti-sexual orientation motivated bias crime, and anti-religion motivated bias crime types are
considered.

Although the findings were not resoundingly supportive of the application of social
disorganization theory to the understanding of bias criminality, there are remarkable conclusions
nonetheless. Measures of social heterogeneity – or diversity – seem to yield the most conclusive
evidence toward predicting the risk of bias crime occurrence. Specifically, a county’s (higher)
percentage of Muslim residents was the strongest predictor of bias criminality across the United
States from 2000 – 2009. Similarly, the percentage of Jewish residents, the percentage of nonWhite residents, and the percentage of foreign-born persons were positively related. Individual
measures of residential mobility and social cohesion were also helpful to predicting bias crimes.
In addition, county population was useful in predicting bias criminality. Urban areas and
urbanized clusters were more likely to experience bias crime occurrence than were rural areas.
In addition, results are inconclusive on whether hate crime legislation decreases the risk of bias
crime occurrence. The findings indicate that more research is needed. Specifically,
understanding a community’s level of diversity seems to be important to the prediction of bias
criminality in American counties.

Copyright by
RYAN BREVIN MARTZ
2014

This dissertation is dedicated to my parents, Eric and Brenda Martz, who have always
encouraged and supported my education and inquisitive nature, and to my grandmother, Betty
Carraway, who continues to inspire me daily.

v

ACKNOWLEDGMENTS

I would like to acknowledge the efforts of those who have contributed toward the
completion of this dissertation. From Bloomington to East Lansing, Dr. Steve Chermak has
supported my educational goals the entire way. He inspired me to pursue a career in academia
and has always provided support to that end. He has reviewed countless drafts and has always
been available to guide my academic growth. I am extremely grateful for his efforts, his
mentorship, and the education he has provided me. I would also like to thank Dr. Tim Bynum,
Dr. Ed McGarrell, and Dr. Steve Gold for serving on my committee. Their expertise and
guidance were instrumental in my completion of this project.
I would like to acknowledge and recognize my friends and colleagues from Michigan
State University and Saginaw Valley State University. Without them, I could not have begun or
completed this journey. Dr. Carol Zimmermann provided encouragement, support, and her
theoretical and methodological expertise throughout the dissertation process. I am infinitely
grateful for her guidance, kindness, and friendship. Dr. Joni Boye-Beaman and Dr. Bob Lane
provided institutional support and patiently guided me through numerous challenges.
Dr. Eric Grommon provided guidance with research design and was my go-to data
consultant. His friendship and academic support are much appreciated. Dr. Mary Braz
contributed her methodological advice when I was designing this dissertation. Dr. Tim
Rowlands read drafts and provided feedback throughout the writing process. Dr. Jennifer
McCullough offered much methodological and statistical support. Her assistance is much
appreciated. Dr. Sher Ratnabalasuriar was always there for a cup of coffee and to discuss the
many angles from which this research could be considered. In addition, Dr. James Bowers

vi

shared his SPSS and statistical expertise on a regular basis. I am grateful for all of the generous
efforts so many friends and colleagues have provided.
I would also like to thank my family for their unrelenting support of my education.
Brenda Martz, Eric Martz, and Lauren Martz have each encouraged me during frustrating points
in the writing process and have been patient with my reactions to the many ups and downs of the
dissertation process. In addition, they have listened to hours of discussion about the many
unrealized research ideas I have had. Lynn Michelsen, George Michelsen, Jeff Michelsen, and
Jeanine Michelsen have been supportive in helping me identify my contributions to the field of
criminal justice. They have also encouraged me to move forward during times of doubt. I am
lucky to have such a supportive family; this dissertation would not have been possible without
their efforts.
Dr. Melissa Garmo has been by my side during my entire doctoral program at MSU. I
cannot thank her enough for the daily encouragement and friendship. She has been there for me
in so many ways throughout many challenging times during the Ph.D. program. For that, I
cannot express my gratitude enough.
Christopher Michelsen has sacrificed greatly so that I could earn my degree. From East
Lansing, to Ann Arbor, to Saginaw, to Washington, and back to Saginaw again – he has been
present throughout my entire doctoral program. He has not wavered in his confidence in me,
even when I could not muster the same. Chris has listened to hundreds of hours of discussion
about my academic life, including the process of completing this dissertation. In addition, his
expertise with data organization and analyses in Microsoft Excel was helpful and much
appreciated during the data collection process. I cannot express enough appreciation as Chris is

vii

the person who has sacrificed the most so I could achieve my goal of completing a Ph.D. and
beginning a career in academia. I could not have completed this degree without him.

viii

TABLE OF CONTENTS

LIST OF TABLES ....................................................................................................................... xiii
LIST OF FIGURES ...................................................................................................................... xv
CHAPTER 1: INTRODUCTION ................................................................................................... 1
Understanding the Risk of Bias Crime in the United States ....................................................... 1
The Scope of Bias Crime in the United States ........................................................................ 3
Understanding Bigotry and Hate ............................................................................................ 4
The Impact of Bias Crime on Victims ........................................................................................ 6
Race......................................................................................................................................... 8
Religion ................................................................................................................................... 8
Sexual Orientation .................................................................................................................. 9
The Impact of Bias Crime on Communities ............................................................................. 11
The Importance of Research on Bias Crime ............................................................................. 13
Celebrated Cases and Public Awareness of Bias Crime ........................................................... 14
Lack of Theory on Bias Crime Occurrence .............................................................................. 15
Research Questions, Data, and Methodology ........................................................................... 16
Lack of Research on Bias Crime with Counties as Units of Analysis ...................................... 18
Direction and Contribution of Research ................................................................................... 22
Counties as Units of Analysis ............................................................................................... 22
Organization of Dissertation ..................................................................................................... 23
Summary ................................................................................................................................... 23
CHAPTER 2: REVIEW OF BIAS CRIME RESEARCH ............................................................ 25
Bias Crime Victimization ......................................................................................................... 25
Anti-Race Bias Crime ........................................................................................................... 28
Anti-Religion Bias Crime ..................................................................................................... 30
Anti-Sexual Orientation Bias Crime ..................................................................................... 33
Superordinate Group Victimizations .................................................................................... 35
Bias Crime Offending ............................................................................................................... 36
Individuals and Intolerance ................................................................................................... 36
Hate Groups .......................................................................................................................... 37
The Influence of Religion ..................................................................................................... 41
Theoretical Perspectives on Bias Crime Incidence ................................................................... 42
Psychological Perspectives ................................................................................................... 42
Prejudice ............................................................................................................................ 42
Authoritarianism ................................................................................................................ 44
Sociological Perspectives...................................................................................................... 46
Prejudice and Group Position ............................................................................................ 46
Social-Psychological Perspectives ........................................................................................ 47
Social Learning Theory ..................................................................................................... 48
Economic Perspectives ......................................................................................................... 49
ix

Realistic Group Conflict Theory........................................................................................ 49
Defended Neighborhoods Theory ...................................................................................... 50
Scapegoat Theory .............................................................................................................. 52
Political Perspectives ............................................................................................................ 52
Cultural Ideology Influence ............................................................................................... 53
Social Movement Theory................................................................................................... 53
Policy Perspectives on Bias Crime ........................................................................................... 55
Bias Crime Legislation and Controversy .............................................................................. 55
Bias Crime and Law Enforcement ........................................................................................ 56
Spatial Characteristics of Bias Crime ................................................................................... 58
Strengths and Weaknesses of Prior Studies .............................................................................. 60
Summary ................................................................................................................................... 61
CHAPTER 3: SOCIAL DISORGANIZATION THEORY .......................................................... 63
Social Disorganization Theory ................................................................................................. 64
A Resurgence of Social Disorganization Theory.................................................................. 66
Blau and Blau (1982) ......................................................................................................... 66
Sampson (1986) ................................................................................................................. 67
Skogan (1986) .................................................................................................................... 68
Bursik (1988) ..................................................................................................................... 69
Sampson and Groves (1989) .............................................................................................. 71
Sampson and Wilson (1990) .............................................................................................. 72
Bursik and Grasmick (1993) .............................................................................................. 73
Sampson, Raudenbush, and Earls (1997) .......................................................................... 74
Morenoff, Sampson, and Raudenbush (2001) ................................................................... 75
Warner (2003) .................................................................................................................... 76
Sampson (2012) ................................................................................................................. 77
Testing and Applying Social Disorganization Theory.......................................................... 78
Chicago Area Project (CAP).............................................................................................. 78
Project on Human Development in Chicago Neighborhoods ............................................ 79
Urban Renewal .................................................................................................................. 80
Community Policing .......................................................................................................... 81
Weed and Seed ................................................................................................................... 84
Community Disorganization and Racially Motivated Crime ............................................ 85
Disorganization and Far Right Violence............................................................................ 86
Social Disorganization Theory and Bias Crime in the United States ....................................... 88
Hypotheses ................................................................................................................................ 89
Low Economic Status ........................................................................................................... 90
Social Heterogeneity ............................................................................................................. 90
Social Cohesion .................................................................................................................... 91
Residential Mobility.............................................................................................................. 92
CHAPTER 4: DATA AND METHODOLOGY .......................................................................... 93
Classifying Bias Crime Events in Unites States Counties ........................................................ 93
Using Official Data ................................................................................................................... 94
Advantages of Official Data ................................................................................................. 95
x

Drawbacks to Using Official Data ........................................................................................ 96
Data Collection Strategy ........................................................................................................... 98
Counties as the Unit of Analysis............................................................................................. 100
Operationalization of Dependent and Independent Variables ................................................ 103
Operationalization of Dependent Variables ........................................................................ 103
Creation of Dependent Variables ..................................................................................... 105
Operationalization of Independent Variables ..................................................................... 111
Low Economic Status ...................................................................................................... 112
Social Heterogeneity ........................................................................................................ 113
Social Cohesion ............................................................................................................... 114
Residential Mobility ........................................................................................................ 116
Control Variables ................................................................................................................ 116
Codebooks........................................................................................................................... 118
Data Analyses ......................................................................................................................... 120
Primary Models ................................................................................................................... 121
Negative Binomial Regression ........................................................................................ 121
OLS Regression ............................................................................................................... 123
Summary ................................................................................................................................. 125
CHAPTER 5: THE CONTEXT OF BIAS CRIME IN UNITED STATES COUNTIES .......... 126
The Nature of Counties in the United States: Descriptive Findings ....................................... 126
Description of Independent Variables .................................................................................... 133
Economic Deprivation ........................................................................................................ 133
Social Heterogeneity ........................................................................................................... 135
Social Cohesion .................................................................................................................. 138
Residential Mobility............................................................................................................ 139
Control Variables ................................................................................................................ 144
Summary ................................................................................................................................. 148
CHAPTER 6: CONSIDERING BIAS CRIME OCCURRENCE IN UNITED STATES
COUNTIES ................................................................................................................................. 149
Bivariate Analyses .................................................................................................................. 149
Model One .......................................................................................................................... 149
Models Two, Three and Four .............................................................................................. 156
Model Two ....................................................................................................................... 157
Model Three ..................................................................................................................... 158
Model Four ...................................................................................................................... 158
Bivariate Analyses for Continuous Independent Variables ................................................ 159
Key Findings ........................................................................................................................... 167
Conclusion .............................................................................................................................. 169
CHAPTER 7: USING SOCIAL DISORGANIZATION THEORY TO PREDICT BIAS CRIME
AT THE COUNTY LEVEL ....................................................................................................... 170
Question 1: .............................................................................................................................. 170
Question 2: .............................................................................................................................. 175
Question 3: .............................................................................................................................. 179
xi

Question 4: .............................................................................................................................. 184
Question 5: .............................................................................................................................. 188
Social Heterogeneity ........................................................................................................... 190
Social Cohesion .................................................................................................................. 191
Residential Mobility............................................................................................................ 191
Population ........................................................................................................................... 191
Hate Crime Legislation ....................................................................................................... 192
Additional Control Variables .............................................................................................. 193
Summary ................................................................................................................................. 194
CHAPTER 8: DISCUSSION AND CONCLUSION ................................................................. 195
A Review of the Dissertation and Notable Findings .............................................................. 195
Notable Findings ................................................................................................................. 196
Measures of Social Disorganization Theory .................................................................... 196
Additional Variables ........................................................................................................ 198
Limitations .............................................................................................................................. 200
Implications for Policy............................................................................................................ 202
Implications for Future Research ............................................................................................ 203
Conclusion .............................................................................................................................. 205
APPENDICES ............................................................................................................................ 206
APPENDIX A ......................................................................................................................... 207
APPENDIX B ......................................................................................................................... 209
APPENDIX C ......................................................................................................................... 211
APPENDIX D ......................................................................................................................... 213
APPENDIX E ......................................................................................................................... 215
APPENDIX F ......................................................................................................................... 217
REFERENCES ........................................................................................................................... 219

xii

LIST OF TABLES

Table 1.1: Bias Crime Incidents by Year (1996-2010)……………………………………………4
Table 4.1: Characteristics of United States Counties by State………………………………….101
Table 4.2: Reported Bias Crime Totals by State………………………………………………..107
Table 4.3: Bias Crime Motivations (2000-2009)……………………………………………….110
Table 4.4: Codebooks for County, Social Disorganization, and Control Variables...………….119
Table 5.1: Bias Crime Reporting by State……………………………………………………...126
Table 5.2: Survey of County Populations across the U.S………………………………………129
Table 5.3: Descriptive Statistics for U.S. County Population (2000)…………………………..130
Table 5.4: Characteristics of United States Counties (2000-2009)……………………………..131
Table 5.5: Descriptive Statistics for Measures of Economic Deprivation (2000)……………...134
Table 5.6: Descriptive Statistics for Measures of Social Heterogeneity (2000)………………..137
Table 5.7: Descriptive Statistics for Measures of Social Cohesion (2000)…………………….139
Table 5.8: Descriptive Statistics for Measures of Residential Mobility (2000)………………..140
Table 5.9: Context of Bias Crime Laws in the United States (2000)…………………………..141
Table 5.10: Additional Descriptive Statistics for Control Variables (2000)…………………...145
Table 5.11: Descriptive Statistics for Measures of Anti-White Bias Occurrence (2000)………147
Table 6.1: Model One Chi-Square Tests for Independence (DV1)…………………………… 150
Table 6.2: Logistic Regression on Measures of Low Economic Status………………………..151
Table 6.3: Logistic Regression on Measures of Social Heterogeneity………………………....152
Table 6.4: Logistic Regression on Measures of Social Cohesion………………………………153
Table 6.5: Logistic Regression on Residential Mobility……………………………………….154

xiii

Table 6.6: Logistic Regression on Control Variables…………………………………………..155
Table 6.7: Kruskal-Wallis Test for Models Two, Three, and Four…………………………….157
Table 6.8: Spearman’s Rho Correlations for Model Two………………………………………159
Table 6.9: Spearman’s Rho Correlations for Model Three……………………………………..162
Table 6.10: Spearman’s Rho Correlations for Model Four…………………………………….165
Table 7.1: Model 1 – Negative Binomial Regression…………………………………………..171
Table 7.2: Model 2 – OLS Regression Analysis………………………………………………..176
Table 7.3: Model 3 – OLS Regression Analysis………………………………………………..181
Table 7.4: Model 4 – OLS Regression Analysis………………………………………………..185
Table 7.5: Summary of Multivariate Findings/Regression Coefficients (Standardized)……….189
Table A.1: Characteristics of Economic Deprivation (2000)…………………………………..207
Table B.1: Characteristics of Social Heterogeneity (2000)…………………………………….206
Table C.1: Characteristics of Social Cohesion in States (2000)………………………………..211
Table D.1: Characteristics of Residential Mobility in States (2000)…………………………...213
Table E.1: Descriptive Statistics for Control Variables………………………………………...215
Table F.1: Controlling for Anti-White Bias Crime (2000)……………………………………..217

xiv

LIST OF FIGURES

Figure 1: Percentage of States with Hate Crime Laws................................................................143

xv

CHAPTER 1: INTRODUCTION
This dissertation investigates various factors influencing the occurrence of bias crimes in
United States counties. By examining traditional indicators of social disorganization theory, this
dissertation seeks to explore the likelihood of bias crime occurrence in United States counties.
The data upon which this dissertation is based come from the Federal Bureau of Investigation
(FBI), the United States Census Bureau (USCB), the Association of Religious Data Archives
(ARDA), and Congressional Quarterly’s Voting and Elections Collection.
However, before engaging in further research on bias crime in the United States, it is
important to review the breadth of bias crime knowledge available in the literature and to
understand the risks such crimes create for communities. The remainder of Chapter One is
devoted to the descriptive analysis of bias-motivated crime as well as an introduction to this
dissertation.
Understanding the Risk of Bias Crime in the United States
The risk associated with bias crime is usually focused on the individual victims of said
crimes. In order to have a complete picture of the nature of bias crime risk, however, it is
imperative to investigate the nature of the crime itself. This information will inform the current
study as it relates to the risk and subsequent harm imposed on United States counties. The
following discussion uncovers this understanding by introducing discussion on the scope of bias
crime in the United States, the nature of bigotry and hate, and the social construction of
difference and the impact of power.
Although the United States government has a relatively broad definition of what
constitutes a bias crime, not all states have adopted policies protecting all oppressed social
groups. Michigan, for instance, protects individuals based on race, religion, color, gender, or

1

national origin (Michigan Legislature, 2011), but not on sexual orientation, gender identity,
disability, or ethnicity. These omissions are not uncommon. Traditionally, with regard to sexual
orientation and gender identity, fewer than half of all states with bias crime laws include sexual
orientation and gender identity in the definition of bias crimes (Wang, 1994). However,
according to Gerstenfeld (2011), as of 2009, the National Gay and Lesbian Task Force (2009)
reports that only fifteen states with bias crime legislation omit sexual orientation from inclusion
in the definition. Although there is much debate as to which groups should and should not be
protected by bias crime legislation, sexual orientation is often omitted as legislators are reluctant
to lend political legitimacy the circumstances of the lesbian, gay, and bisexual community
(Jenness & Broad, 1997). Women and persons with a disability are also groups that face an
uphill battle in gaining protection via bias crime legislation. Many individuals do not view
gender (i.e. females) to be a minority1, and thereby overlook that women retain an oppressed
status in society. Additionally, there are those who would argue that persons who have
physically or mentally disabilities are not often victims of bias crime (Jenness & Broad, 1997).
Nevertheless, according to Gerstenfeld (2011), as of 2009, there were still five states with no bias
crime laws on the books at all. Given the political difficulty of bias crime legislation, even
limited progress in providing protection to marginalized groups is notable.
As stated, bias crime has traditionally been defined as a criminal act motivated by animus
toward the victim’s affiliation with an oppressed social group (Gerstenfeld, 2004). Definitions
of bias crime vary by time and location – even the concept itself has numerous frequently used
terms such as hate crime, ethnic intimidation, ethnoviolence, and civil rights crimes. Hate crime
is really about prejudice, or bias, and less about hate; therefore, bias crime appears to be a more

1

Although ‘minority’ is a commonly used word to describe subordinate social groups, ‘oppressed social group’ is
used to more accurately reflect the situation of a marginalized group of people.

2

accurate designation of the phenomenon (Jacobs & Potter, 1998). Publicly, the two terms are
frequently used interchangeably; however, bias crime is largely considered by scholars to be a
more accurate term.
Regardless of the debate as to which groups should be protected by bias crime legislation,
it is clear that any crime based on bigotry is particularly hurtful to the victim, both physically and
psychologically, and consequently is important as a topic of investigation (Iganski, 2001).
Moreover, Iganski (2001) suggests that bias crime not only affects the individual victim, but also
distresses the oppressed - and thereby subordinate - social group to which the victim belongs.
For instance, in the case of James Byrd, an African American individual who was murdered in
Texas in 1998, Byrd was not randomly selected as a victim; he was chosen as a victim because
of his racial identification. The message of hatred and intolerance was meant not only for Byrd,
but for the entire African American community in Jasper, Texas and the United States alike. In
this way, messages of hate and intolerance terrorize the entire group with which the victim
identifies (Tsesis, 2003; Boekmann & Turpin-Petrosino, 2002; Green, McFalls, & Smith, 2001;
Iganski, 2001). Consequently, the African American community, particularly in Jasper, Texas,
was left with feelings of insecurity, fear, and terror following the murder. For this reason, many
scholars argue that crimes of this type warrant legislative, social, and scholarly attention.
The Scope of Bias Crime in the United States
Since the Hate Crimes Statistics Act of 1990, whereby the Federal Bureau of
Investigation (FBI) was required by Congress to collect bias crime data from local law
enforcement agencies, scholars and government officials have gained a better understanding of
the frequency of reported bias crimes. As reported by the FBI and is evident in Table 1, total
reported bias crimes for the United States were 10,706 in 1996 and by 2008 amounted to 7,783.

3

Overall, the majority of reported bias crimes were motivated by bias toward a victim’s race,
religion, or sexual orientation. This fact informs the current research in that it forms the basis for
understanding the context of bias crime occurrence.
Table 1.1: Bias Crime Incidents by Year (1996-2010)
1996
2000
2004
2008
2010
Total Incidents
10,706
8,063
7,649
7,783
6,628
Race
6,767
4,337
4,042
3,992
3,135
Ethnicity/National
1,163
911
470
894
847
Origin
Religion
1,500
1,472
1,374
1,519
1,322
Sexual Orientation
1,256
1,299
1,197
1,297
1,277
Disability
36
57
78
43
Multiple Bias
20
8
7
3
4
*Hate Crime Statistics (Federal Bureau of Investigation, (1996); (2000); (2004); (2008); (2010)
Understanding Bigotry and Hate
Bigotry is difficult to comprehend for many; however, it is often described by scholars as
being a thinking disorder (Beck, 1999). As a thinking disorder, or even as a strong emotion,
bigotry can be very destructive to those who encounter it. Moreover, the causes of bigotry can
also be difficult to discern. Similar to other social phenomena, there is great variability in a
person’s reasons for falling into a pattern of bigoted thinking or behavior.
Many scholars argue that one of the more influential components to bigoted thinking is
ignorance. Similarly, the lack of experience with difference can be attributed to bigotry as well.
For instance, Jacobs and Potter (1997) explain that the fear of social diversity propels some
individuals to feel bigoted toward another person based on that person’s identification with a
subordinate social group. Specifically, the socially “undesirable” groups tend to be those of
oppressed statuses such as race, ethnicity, sexual orientation, and religion (Perry, 2003b).
Moreover, the fear of diversity can lead to prejudice, whereby a person comes to have a negative
opinion about someone based on a preconceived idea of a group with which the person may
identify (Jacobs & Potter, 1997).
4

Although bigotry is generally not considered to be a positive personal characteristic, it is
not as detrimental as the criminal act it sometimes supports - that of a bias crime. Moreover, it is
important to note that bigotry is itself not criminal in the United States; it is protected by the First
Amendment of the United States Constitution (Iganski, 2001). However, as Steinberg, Brooks,
and Remtulla (2003) suggest, bias crime is defined as learned aggressive behavior completely
motivated by prejudice, bigotry, or racism.
Other scholars differentiate between bigotry and hate in how the concepts inform bias
crime offending. For instance, Gordon Allport’s (1954) book The Nature of Prejudice
introduces the concept of prejudice as relevant to the understanding of the construction of hate.
In this seminal work, Allport (1954) suggests that though it is possible to hate individuals and
groups, it is often easier to hate social groups. After all, it is easier to the see humanity in an
individual.
Allport (1954) continues to discuss prejudice as being influential in the commission of
five certain actions: 1) antilocution, 2) avoidance, 3) discrimination, 4) physical attack, and 5)
extermination. These five actions can be viewed as scalar, with antilocution being the least
serious. In this case, individuals are able to identify an out-group and participate in negative
discourse related to that group. More seriously, persons may choose to create social distance
between those whose group identification is viewed as undesirable and themselves. There is no
direct harm inflicted here. Discrimination is yet more serious, with social groups experiencing
the deprivation of certain opportunities or rights. Allport (1954) continues to explain that
physical attack is generally more serious than discrimination, as individuals are often gravely
harmed. At this point, it is evident that prejudice has contributed toward criminal victimization
of someone identified with an out-group. The most severe level is extermination. At this stage,

5

Allport (1954) explains that the out-group’s complete annihilation is the goal. Examples of this
include the Holocaust in Europe (Allport, 1954) or the Rwandan genocide in Africa in the 1990s.
With regard to hatred, it is perhaps a common misunderstanding that love and hate are
opposite in nature. To combat this notion, Sternberg and Sternberg (2008) introduce a
multifaceted structural theory of hate. The authors use a triangular explanation of love to explain
the closely related central tenets of hate. Sternberg and Sternberg (2008) suggest love can be
explained by the convergence of three factors: commitment, passion, and intimacy. Conversely,
hate is constructed by the melding of commitment, passion, and a negation of intimacy. In
effect, there is relatively little difference between love and hate. Given the concepts are similar
in nature, Sternberg and Sternberg (2008) suggest that the hatred held toward individuals and
groups can, in part, be attributed to the convergence of the three pillars of the structural theory of
hate.
The Impact of Bias Crime on Victims
Across the United States, when a bias crime occurs, organized hate groups such as the Ku
Klux Klan, skinheads, or neo-Nazis are often suspected of perpetrating the crime and are thereby
blamed. However, in general, organized hate groups do not commit the majority of bias crimes.
Although bias crime offenders generally commit their crimes in groups, they are not usually
affiliated with an organized hate group (Craig, 2002). Bias crime offenders tend to be young,
Caucasian males with no prior criminal record from backgrounds that are generally not
impoverished (Craig, 2002). Moreover, in many cases, the offender may even be a neighbor or
live in close proximity to the victim (McDevitt, Levin, & Bennett, 2002). Rather than blatant
extremists dressed in regalia, the bias crime offender may be living next door and be seemingly
otherwise harmless.

6

Bias crime offenders’ motives can vary. As such, this dissertation focuses on exploring
the individual motives related to bias crime occurrence. Currently, there is little macro-level
research capable of providing context for such incidents. Therefore, it is important to remember
that the underlying factor found in all bias crimes is bigotry (Levin & McDevitt, 1993).
McDevitt, Levin, and Bennett (2002) explain the various motivating factors involved with bias
crime offending. For instance, in certain bias crimes the offender is in search of a sense of power
and excitement. These crimes are considered thrill crimes and are the most common. Other bias
crime offenders feel the need to protect their territory or resources and are called defensive
offenders. Yet others that commit bias crimes in a reactive manner avenging a perceived wrong
are regarded as retaliatory criminals. Those that victimize based on a desire to “cleanse the
world of evil” are known as mission offenders.
During the 1980s and 1990s, a series of celebrated crimes involving victims of
subordinate status invigorated much interest and disagreement by American legislators and
scholars alike. Since that time, many resources – scholarly, financial, and legislative - have been
allocated toward the understanding and remedy of bias crime. While some consensus exists on
the detriment of bias crime offending, disagreement still surrounds the causes of bias crime
victimization and the policy actions that should be taken to combat such crimes. Although bias
crime occurs across a wide range of groups – from political affiliation to race – this study is
limited to the three most frequently occurring types of bias crimes: those motivated by bias
against the victim’s race, religion, and sexual orientation.

7

Race
African Americans are the most common victims of bias crime in the United States
(Gerstenfeld, 2004). This should not be surprising as African Americans have been subjected to
intolerance, violence, racism, and inequality for centuries. American slavery is perhaps the most
massive bias crime in history (Gerstenfeld, 2004), and therefore it set the stage for the
occurrence of more recent bias criminality. For decades after slavery, African Americans
endured lynchings and other forms of violence nationwide, though arguably much of it occurred
in the South. Historically these crimes were not necessarily recorded or viewed as bias crimes.
As an example, Nolan, Akiyama, and Berhanu (2002) point out that though the South has the
second highest number of index crimes and the second highest crime rate in the four regions of
the United States, it somehow reports the lowest number of bias crimes. Prejudicial views of
African Americans are present today and are made painfully visible every time a bias crime
against an African American occurs. Incidents occur more frequently than one would presume,
though they are not always as brutal as the James Byrd murder. For example, in the South
hundreds of black churches have been bombed or burned in recent years (Gerstenfeld, 2004).
Numerous cross burnings have occurred in the front lawns of African Americans as well.
According to Green (1998) although considered free speech, these cross burnings, usually carried
out by whites, often predict more direct and violent attacks. These incidents are equally as
troubling given that they are motivated by the same bigotry and intolerance that took James
Byrd’s life.
Religion
Although religiously-motivated bias crime can happen to a member of any subordinate
religious group, the Jewish community has long endured bias-based criminal victimization.

8

Anti-Semitism is perhaps among one of the oldest and deepest forms of prejudice and intolerance
in existence. Historically, Jewish persons were persecuted by the Egyptians, Greeks, and
Romans and have been confined to ghettoes without the ability to own land (Gerstenfeld, 2004).
The most well-known account of anti-Semitism occurred in Nazi Germany where six million
Jewish persons perished throughout Europe. On the American front, anti-Semitism often
involves the extremist beliefs that blame Jewish persons for economic troubles, communism, and
disloyalty to the United States (Gerstenfeld, 2004). Additionally, many fringe Americans
believe that a Jewish conspiracy runs the country and that Jewish persons have led various social
movements, such as the feminist and Civil Rights movements (Gerstenfeld, 2004). When these
extremist beliefs are acted upon, anti-Jewish incidents ranging from swastika spray paintings on
synagogues to harassment and more violent assaults occur across the United States (Gerstenfeld,
2004). For instance, as Altschiller (1999) describes, in 1994 a Lebanese immigrant shot at a van
carrying fifteen Hasidic Jewish students in New York. One student died and three were injured.
Sexual Orientation
Individuals in the lesbian, gay, and bisexual community experience acts of intolerance
and violence, both criminal and non-criminal, at a rate similar to members of other oppressed
social groups (Gerstenfeld, 2004), yet attention to these injustices is often overlooked and
ignored. Although there are numerous reasons for this, the sentiment that members of the LGB
community are deviant, abnormal, and unworthy of legitimacy and concern is widely accepted in
American society (Gerstenfeld, 2004). Taking this assumption into consideration, it should
come as no surprise that there is a limited availability of information about, and therefore a
significant gap exists in the literature, the effects of bias crime on the LGB community
(Comstock, 1991).

9

Specifically, anecdotal evidence exists that suggests bias crime may affect LGB
individuals in a manner that differs from other bias crime or non-bias crime. It is frequently
noted that a theme of anti-LGB violence is the presence of an elevated level of severity and
brutality of attack. Many scholars argue that anti-LGB bias crimes tend to be the most brutal and
physically severe of all bias crimes (Comstock, 1991; Levin & McDevitt, 1993; Miller &
Humphries, 1980; Rayburn & Davison, 2002). Certainly, there are numerous examples of bias
crimes committed against other oppressed subordinate groups that result in equally brutal and
severe injuries. The difference seems to be that bias crime against the LGB community tends to
be particularly brutal and are often considered to be overkill (Bufkin, 1999), sending a
particularly strong message of intolerance to victims and to the entire LGB community.
As research and anecdotal accounts suggest, whether an assault or homicide, the effects
of attacks against those persons who identify as part of the LGB community tend to be severe in
nature. Police officers are often witnesses to the aftermath of an anti-LGB bias crime. One
Miami, Florida police officer noted that anti-LGB attacks have been among the worst beatings
that he has seen (Comstock, 1991). In more severe cases, Comstock (1991) references a
psychiatrist as stating, “Multiple and extensive wounds are not uncommon in the fury of antihomosexual murder.” (p. 47). In their 1980 study of anti-LGB assaults and homicides, Miller
and Humphreys remarked that, “An intense rage is present in nearly all homicide cases involving
gay male victims. A striking feature... is their gruesome, often vicious nature. Seldom is the
homosexual victim simply shot. He is more apt to be stabbed a dozen or more times, mutilated,
and strangled” (p. 179). According to Berrill (1992), yet another example of anti-LGB bias
crime brutality comes from, Melissa Mertz, a hospital employee in New York City, “Attacks
against gay men were the most heinous and brutal I encountered. They frequently involved

10

torture, cutting, mutilation, and beating, and showed the absolute intent to rub out the human
being because of his [sexual] preference” (p. 25).
The Impact of Bias Crime on Communities
Although there are various perspectives on the immediate outcomes of bias crime
offending, many scholars have focused on concept of interchangeability. McDevitt, Balboni,
Garcia, and Gu (2001) suggest that bias crime offending is conceptually different from non-bias
based crimes given the victim potentially could have been any member of the targeted group.
For example, in the event that an African American individual is the victim of a bias crime, the
larger African American community realizes this act of criminality could have been targeted
toward them as well. In effect, people are left thinking, “It could have been me.” In this way,
bias crime offenders are inherently sending a negative message to oppressed subordinate groups.
This message is largely based on the sentiment that these individuals are viewed as different and
thereby not welcome to be part of the community. Subsequently, other members of the affected
subordinate groups may experience secondary victimization (McDevitt J. , Balboni, Garcia, &
Gu, 2001).
Likewise, Iganski (2001) suggests the harm of bias crime victimization can be considered
to have a wave-like effect. Others have discussed this as being a “ripple effect” (Noelle, 2001).
When one person is victimized, the result is a wave of victimization throughout the community –
affecting not only the victim, but others in the subordinate group. Moreover, it is certain that a
bias crime victim is generally subjected to physical harm; however, bias crime victims also
experience psychic harm (Iganski, 2001). Although Iganski (2001) notes that the findings are
not unanimous, emotional and psychological impacts usually follow bias-based criminal events.
Moreover, victims generally do experience what Iganski (2001) considers an “in terrorem”

11

effect. Essentially, “in terrorem” refers to the threatened state a victim and community
experience after the occurrence of a bias crime. In addition, this terror, or psychological trauma,
for victims can last over five years past the criminal event – a time period that is double that of
non-bias motivated crimes (Craig, 2002). In their study of effects of ethnoviolence, Ehrlich,
Larcom, and Purvis (1994) compare the psychophysiological symptoms resulting from posttraumatic stress of ethnoviolence victims, personal violence victims, and non-victims. The
results of this nationwide survey suggest that 18 of the 19 stress items occurred for a higher
percentage of ethnoviolence victims than others in the study. Among some of the most common
reported symptoms were sleep difficulties, anxiety, hardships at work and with concentration,
anger, fear, and feelings of unexplained exhaustion (Ehrlich, Larcom, & Purvis, 1994). As is
evident, the violent message of intolerance that meets the victims of bias-based victimization is
quite accurately considered to be a form of domestic terrorism.
As suggested, bias crime is harmful to individuals and to society alike – both physically
and emotionally to individuals and entire subordinate social communities; however, this
dissertation seeks to focus attention on the effects of bias crime on Unites States counties. All
types of criminal activity are detrimental for communities; there is no doubt. However, bias
crime presents a unique challenge. Given the complexities of bias criminality, public policy
options are much debated. It is certain, that the motive behind one policy option - bias crime
laws – is to publically acknowledge the ill effects bias crime has on the community. Jacobs
(1998) calls this symbolic denunciation. In effect, society has recognized that a certain type of
crime – in this case, bias crime – is particularly harmful to the community and takes action to
state its collective recognition of and condemnation for the problem.

12

The Importance of Research on Bias Crime
Although bias crime is a relatively new topic of study in the discipline of sociology and
field of criminal justice, there has been significant research conducted on this type of crime from
the policing, legal, behavioral, and political perspectives. Moreover, various contemporary
theories have been explored to help explain bias crime offending.
The effects of intolerance, bigotry, and bias are numerous; however, bias-motivated
criminal acts invoke fear and confusion that disturb entire communities (Iganski, 2001), leading
many to understand that bias-related crimes warrant increased attention. Bias-motivated crime is
not limited to a small number of communities. Its effects, however, are most often experienced
by oppressed social groups – typically racial, ethnic, religious, and sexual minorities, as well as
persons with disabilities and persons targeted because of their gender identity.
With the understanding that bias crime is a social problem that creates a cause for
concern, importance lies in looking to theory to explain why bias crimes occur. Although it
might be easy to blame hatred alone for the occurrence of bias crimes, hatred does not
adequately explain why bias crimes are committed. Hatred can manifest itself in many forms,
some of which actually include criminal offending; however, other forms of hatred such as hate
speech are constitutionally protected (Gerstenfeld, 2004). Bigotry and intolerance are not
strangers to the United States, and it is likely that most Americans hold biases to which they may
or may not admit or of which they may not be aware. However, not all Americans choose to
criminally victimize others based on bigotry, prejudice, or hatred. Predicting such criminal
activity can be particularly difficult; thus, looking to theory for guidance can provide insight into
complex issues such as bias crimes. Before discussing theoretical perspectives, however, this
dissertation seeks to describe the current relative lack of empirical evidence on bias crime.

13

Celebrated Cases and Public Awareness of Bias Crime
Bias-motivated crimes have come to the American public’s attention only in recent
decades; however, the underlying bigotry and intolerance of such crimes are not novel
phenomena in the United States. The recent attention to bias crime does not imply that bias
crime did not exist previous to its identification and naming. In actuality, examples of bias crime
are not only evident throughout American history, but world history as well. American slavery
and the Holocaust are prime examples, yet these merely scratch the surface of the bias-based
brutalities humans have inflicted upon one another in the past. However, not until the 1980s did
the terms hate crime and bias crime2 advance into public consciousness and discourse (Green,
McFalls, & Smith, 2001). Since that time, there has been flurry of attention given by the media,
the police, and various legislators to crimes motivated by the bigotry toward an individual based
on their identification or their perceived identification with a certain social group. According to
Chakraborti and Garland (2009), many victims of bias crime are chosen because of their group
association related to race, religion, disability, or sexual orientation. In two well-known cases of
bias crime offending, the brutal murders of James Byrd, an African American man, and Matthew
Shepard, a gay college student, have made Americans painfully aware of the consequences that
such bigotry and intolerance can have on society. Although bias crime can surface in many
forms and include people of various subordinate social identities, the underlying factor found in
all bias crime is always bigotry (Levin & McDevitt, 1993). A bigot is defined by MiriamWebster (1991) as, “one obstinately or intolerantly devoted to his opinions and prejudices” (p.

2

The term bias crime is used in this dissertation as opposed to the term hate crime. Although the term hate crime is
often used in the literature, scholars have moved toward the use of bias crime as a more precise wording of the crime
(McDevitt & Williamson, 2003). The FBI defines hate crimes as crimes motivated by the offender’s bias based on
the victim’s race, sexual orientation, religion, ethnicity, national origin, and disability (Federal Bureau of
Investigation, 2010).

14

149). Certainly, understanding bigotry can help to inform the discussion of bias crime
occurrence.
Lack of Theory on Bias Crime Occurrence
Theoretical explanations of bias crime offending and victimization are largely
underdeveloped in the social science literature. As suggested, bias-based crime is not new to the
United States; rather, importance lies in understanding and appreciating such crimes through the
conceptual lenses of the various social sciences. Social scientists have posited theoretical
explanations for bias crime from the disciplines of criminology, psychology, sociology, and
political science. Most importantly, the discipline of sociology is largely lacking a cogent
explanation of the phenomenon.
Moreover, bias crime does vary contextually; each incident is differentially based on
time, space, victim, and offender. It is important to consider who the victims are does indeed
matter. After all, the basis for labeling a bias crime as such is due to its unique characteristic of
having a victim targeted based on her or his larger subordinate social group affiliation – in short,
who the victims are. Thus, it comes not as a surprise that bias crimes targeted toward a variety
of subordinate social groups may have differential outcomes. Specifically, a goal of this
dissertation is to contribute to the discussion of bias crimes occurrence in the United States at the
county level. In accepting the realization that social science does not possess much in way of
explaining bias crime, it is important to focus on various factors that may contribute to an
enhanced understanding of the particular criminal event at the county level.
With the motivations of bias crime offenders in mind, it is difficult to identify with any
certainty the underlying reasons why individuals choose to commit bias crimes. Although there
are theories that attempt to explain bias crime offending, the scholarly literature is largely

15

underdeveloped in this area. Therefore, after a review of existing theories presented in Chapter
Two, this dissertation will examine the perspective of social disorganization theory with regard
to the occurrence of bias crime at the county level.
Social disorganization theory is increasingly important to consider when conducting bias
crime research as it provides a macro-level ecological perspective to a concept traditionally
examined through the lens of micro-level theories. There is much value in understanding the
numerous effects of bias crime; however, the way community groups affect bias crime
occurrence – is largely unknown in the social science literature. With its ability to speak to
macro-level effects of bias crime occurrence, social disorganization theory is optimal for
examining county-level variation in bias crime occurrence.
Research Questions, Data, and Methodology
This dissertation seeks to address several research questions regarding bias crime
occurrence at the county level of analysis. The following are research questions informed by
traditional indicators of social disorganization theory. The research questions are as follows:
Question 1: What are the relationships between traditional indicators (low economic status,
social heterogeneity, family disruption, residential mobility, and violent crime rate) of social
disorganization concepts (as measured by US Census Data, ARDA, and Congressional
Quarterly) and bias crime at the county level?
Question 2: What are the relationships between traditional indicators (low economic status,
social heterogeneity, family disruption, residential mobility, and violent crime rate) of social
disorganization concepts (as measured by US Census Data, ARDA, and Congressional
Quarterly) and anti-race motivated bias crime at the county level?

16

Question 3: What are the relationships between traditional indicators (low economic status,
social heterogeneity, family disruption, residential mobility, and violent crime rate) of social
disorganization concepts (as measured by US Census Data, ARDA, and Congressional
Quarterly) and anti-sexual orientation motivated bias crime at the county level?
Question 4: What are the relationships between traditional indicators (low economic status,
social heterogeneity, family disruption, residential mobility, and violent crime rate) of social
disorganization concepts (as measured by US Census Data, ARDA, and Congressional
Quarterly) and anti-religion motivated bias crime at the county level?
Question 5: What are the differences between anti-race, anti-religion, and anti-sexual orientation
bias crime occurrences at the county level?
The purpose of this dissertation is to examine the relationships between indicators of
social disorganization theory and the incidence of bias crime occurrence at the county level of
analysis. Four primary models will be introduced – one examining bias crime occurrence in the
aggregate and three examining different types of bias crime motivation: anti-race, anti-religion,
and anti-sexual orientation. Additionally, each motivation will be compared against the others in
order to identify meaningful differences.
Given the lack of empirical evidence informing bias crime occurrence, especially at the
county level, this dissertation will use quantitative methods to add to the discussion about bias
criminality in the United States. The data used for this dissertation will be aggregated from
existing data available from United States government agencies – namely, the FBI and the
USCB. The primary data used to capture the dependent variables and some independent
variables will be provided by the FBI’s Uniform Crime Report. After the introduction of the
Hate Crime Statistics Act of 1990, the FBI has been collecting reported bias crime data

17

nationwide. In addition, relevant data from the USCB will be used to construct appropriate
independent variables. These variables will be assembled using indicators of social
disorganization theory.
Quantitative analyses will be performed to address the specific research questions asked
in this dissertation. In addition to descriptive statistics, this dissertation will provide the
appropriate bivariate analyses to determine existing relationships. Moreover, the research
questions will be answered through multivariate analyses. Further details about the data and
methodology will be provided in Chapter Three.
Lack of Research on Bias Crime with Counties as Units of Analysis
Given bias crime research is in its infancy, there is relatively little scholarship available
for review. A significant contributor to this problem is the reality that it was not until the early
1990s that the United States began to collect data on bias-motivated crimes. In 1990, the United
States Congress passed the Hate Crime Statistics Act. It was at this point, and for the first time,
that the FBI began to collect data from local law enforcement agencies about the occurrence and
frequency of bias crimes (McDevitt J. , Balboni, Bennett, Weiss, Orchowsky, & Walbolt, 2000).
Similar to other types of crime, there are numerous entities that measure bias crime. The
most relevant is the United States Attorney General’s Office which was required to collect data
on the information surrounding hate crimes due to the passage of the Hate Crime Statistics Act of
1990 (Perry, 2001). This legislation was the first attempt of the United States government to
collect accurate information about hate crimes. The accuracy of these statistics is often debated,
however. Regardless, every year, law enforcement agencies around the country identify and
report the occurrence of bias crime to the FBI (Nolan, Akiyama, & Berhanu, 2002). In response
to the new support for bias crime data collection participation, the FBI realized a need to develop

18

an additional data collection program supplemental to the Uniform Crime Report (UCR) and
designed wholly for bias crime. Although the definition of bias crimes is rather narrow, the FBI
reports that bias crime reports have increased since the beginning of data collection (Perry,
2001). Unfortunately, there is a large discrepancy between the bias crime incidents that the UCR
captures as a result of the Hate Crime Statistics Act of 1990 and the incidents that other antiviolence agencies record. This suggests there are far more bias crime incidents that occur than
are reported.
Even with the FBI’s annual attempts to collect bias crime data, difficulty remains in
obtaining an accurate picture of the scope of bias crime in the United States as local law
enforcement agencies are not required to report (Green, McFalls, & Smith, 2001). McDevitt,
Balboni, Bennett, Weiss, Orchowsky, and Walbolt (2000) report that though the number of
reporting agencies grew substantially throughout the 1990s - from about 2,000 to approximately
12,000 – incident frequencies remained markedly constant. In 1992, the FBI reported 7,755 bias
crimes as having occurred in the United States; by 1996, that number had risen only to 8,759.
This is curious given the disproportionate increase in reporting agencies compared to reported
bias crimes.
In addition to legislative and law enforcement efforts, the Anti-Defamation League
(ADL) collects data on various anti-Semitic hate crimes. The data that are collected are
generally self-reports from victims and can therefore provide much more qualitative data. This
is an area in which the UCR is lacking as it traditionally collects quantitative data (Nolan,
Akiyama, & Berhanu, 2002). The qualitative data that the ADL collect tend to be rich in detail
and inclusive of a wide range of events including non-criminal incidents. For instance, incidents
that range from anti-Semitic slurs to campus crimes are collected annually (Perry, 2001). An

19

advantage of the ADL’s data collection efforts is that it also provides a place for victims to report
incidents where they are able to feel comfortable. Although anti-Semitic events are still likely
underreported, perhaps the ADL provides a less bureaucratic environment that is more appealing
to victims than a law enforcement agency.
The Southern Poverty Law Center (SPLC) has also been active in measuring bias crime
for numerous years. Although there is an emphasis on tracking and recording hate motivated
incidents by hate groups, the SPLC gathers significant information every year in its Intelligence
Report. Unlike the ADL’s specialization on anti-Semitic incidents, there is not a group that
collects specifically race-related bias crime information (Perry, 2001). However, most of the
bias-motivated incidents that the SPLC collects involve anti-racial and anti-LGB incidents
(Perry, 2001). Like the ADL, the SPLC offers qualitative supplements to the data available in
the UCR.
The National Gay and Lesbian Task Force (NGLTF) is another advocacy group that
collects hate crime data. Each year, the NGLTF selects six or nine cities around the country in
which to measure the occurrence of anti-LGB bias-based offending (Perry, 2001). The LGB
population in these cities is then surveyed and reports of victimization are aggregated.
Unfortunately, one drawback is that the surveys are conducted mainly through gay and lesbian
publications – from newsletters to websites. Often, closeted gay, lesbian, or bisexual persons
avoid these publications and consequently, the survey suffers from underreporting. However,
every year the data from just a few cities outpace the reports to the UCR for anti-LGB bias crime
(Perry, 2001). This illuminates the inconclusiveness of the UCR with regard to bias crime.
More recently created, the United States Extremist Crime Database (ECDB) identifies
and follows the occurrence of ideologically motivated homicides (Gruenewald, 2011). The focus

20

of this database is primarily on tracking far-right extremists; therefore, there is notable overlap
between the homicidal actions of hate crime offenders and, for example, white supremacists and
other far-right groups. This database is helpful in understanding the entire scope of homicide –
especially as it related to extremist activity. Since 85% of bias crime consists of violent offenses
(Freilich & Chermak, 2013), examining ideologically motivated homicides is particularly
relevant to understanding bias criminality.
Additionally, there are various other advocacy groups and anti-violence proponents that
anecdotally collect bias crime data. However, there seems to be a lack of cohesion between
victims and the groups collecting the information. As such, whether it is data from the UCR, the
ADL, the SPLC, the ECDB, or the NGLTF, there are likely many bias crimes that go undetected.
Given the limited sources for accurate bias crime data, there should be no surprise that
quality scholarship is lacking. There are numerous scholarly accounts of the frequency of bias
crime occurrence; however, the extant literature remains largely descriptive in nature (Green,
McFalls, & Smith, 2001). As Green and colleagues (2001) suggest, bias crime data are largely
available only from government agencies (such as the FBI) and watchdog groups that monitor
anti-government, militia, and bias-motivated activities.
Although the descriptive analyses that result are valuable, the limitations are significant.
Without detailed accounts of bias crime victims, offenders, and the context in which the crimes
occur, it is difficult to conduct research able to answer directed questions. Moreover, collecting
original bias crime data presents a problem for researchers. Bias crime incidents themselves are
difficult to identify and observe, thereby leaving researchers to rely on unreliable records from
the government and watchdog groups. Access to would-be offenders is not often available especially given the easily identified hate groups are not responsible for the majority of bias

21

crime. Moreover, discerning motive is difficult at best, providing little room for scholars to
retrieve accurate data on bias offending. In addition, Green, McFalls, and Smith (2001) suggest
more robust bias crime scholarship would result from access to reliable data sources such as
surveys of victims, ethnographies, and official records.
Direction and Contribution of Research
Although there are likely to be several limitations to this study, a number of contributions
to the bias crime discussion are expected to emerge. Since empirical studies about bias crime are
relatively limited in the social science literature, this study will add to the general discussion of
macro-level effects of bias crimes – more specifically, at the county level. Moreover, this
dissertation will add to the literature with regard to social disorganization theory and its
contribution to understanding the occurrence of bias crime at the county level. In addition, this
dissertation will examine the unique characteristics of anti-LGB victimization as compared to
those of other subordinate groups such as race and religion.
Counties as Units of Analysis
This dissertation seeks to uncover findings regarding traditional indicators of social
disorganization theory and its connection with the occurrence of bias crimes at the county level
of analysis. Although there are likely numerous factors that contribute to a certain county’s
prevalence of bias crime occurrence, this dissertation will provide insight into whether the
differential opportunities available in a community to remain socially organized affect bias crime
occurrence. This dissertation is unique due its contribution based on counties as the units of
analysis. Although many social scientists have provided much insight into the nature of bias
crime offending and victimization, there is a significant gap in the literature with regard to
offending at the county level. Given that criminal occurrence at all levels requires government

22

intervention, an understanding of county level factors that influence bias crime in the United
States will be a beneficial contribution to the bias crime literature as it can ultimately aid in
informing local governments – and the federal government - in bias crime prevention. Thus, this
dissertation seeks to contribute toward a greater body of knowledge regarding the various social
factors creating a climate ripe for the occurrence of bias crime in the United States.
Organization of Dissertation
This dissertation will be segmented into eight chapters. After the introductory chapter,
Chapter Two will discuss the existing research on bias crime and Chapter Three will introduce
theoretical contributions to the literature. Chapter Four will be composed of research questions,
hypotheses, and a discussion about the proposed data and methodology for the dissertation. The
results of descriptive analyses will be presented in Chapter Five. Chapter Six will present the
findings from appropriate bivariate findings. The multivariate findings addressing the research
questions will be presented in Chapter Seven. Chapter Eight will conclude the dissertation with
relevant discussion about the findings of the analyses along with the policy implications and
future research recommendations.
Summary
No single explanation is able to provide a framework for understanding the extent and
nature of bias crime. However, scholars have posited various social science theories in an
attempt to contribute to the explanation of the occurrence of bias crimes. Based on the existing
literature, this dissertation will attempt to contribute to the knowledge of bias crime occurrence
at the county level through the lens of social disorganization theory. The aim of this dissertation
is to explore the extent to which the traditional indicators of social disorganization theory inform

23

not only inform the occurrence of bias crime at the county level, but also aid in understanding the
meaningful differences in frequency between oppressed social group victimizations.

24

CHAPTER 2: REVIEW OF BIAS CRIME RESEARCH
The extant bias crime literature is largely descriptive in nature. Given the scholarly
exploration of bias crime offending and victimization largely dates back only to the 1980s,
scholars are just beginning the journey toward an empirical understanding of the phenomenon.
However, the existing literature is important nonetheless as it informs the current study and its
attempt to uncover the factors that affect the prevalence of bias crime incidence. Not only will
this literature review explain what is known about bias criminality in the United States, but is
will also provide a foundation for this dissertation. The following literature review will include a
discussion of what is known about bias crime including: bias crime victimization, bias crime
offending, perspectives on bias crime incidence, policy perspectives, and strengths and
weaknesses of the bias crime literature. Since there is not significant empirical evidence
informing the literature regarding macro-level bias criminality, this literature review will provide
insight into the phenomenon.
Bias Crime Victimization
Not surprisingly, victims of bias crimes find the experience to be quite traumatic.
Although some controversy exists in the literature, it seems there is a possibility that bias crimes
could be more traumatic than other crimes (Gerstenfeld, 2011). At the victim level, it is not
uncommon for individuals to experience lowered self-esteem, marked fear, anger, and sadness
(Barnes & Ephross, 1994). Furthermore, Barnes and Ephross (1994) suggest these emotional
reactions to bias-motivated crime are not unlike the reactions to other criminal victimizations.
Moreover, Jacobs and Potter (1998) note that all crime victims experience negative
psychological and emotional effects, not just bias crime victims. Certainly, there is merit to the
argument that criminal victimization is universally traumatic; nevertheless, this remains to be

25

proven in the scholarship (Gerstenfeld, 2011). Given this, and as previously mentioned, bias
crimes are considered by many scholars to be more hurtful than non-bias offending and thereby
warrant legislative and policy attention (Iganski, 2001).
There are several studies finding bias crime victimization to be more harmful than other
types of criminal victimization. Perhaps most convincingly, and echoing the cited effects of bias
crime by of Barnes and Ephross (1994), in a study of lesbian, gay, and bisexual individuals,
responses indicated that those who had been victims of bias crimes reported greater levels of
anger, depression, and anxiety than those individuals reporting non-bias crime victimization
(Herek, Cogan, & Gillis, 2002). Moreover, in a study of bias crime victimization in Boston,
Massachusetts, bias crime victims reported more fear and a lower level of safety as opposed to
non-bias crime victims (McDevitt J. , Balboni, Garcia, & Gu, 2001). The time it takes for a
victim to recover is considered as well. Craig (2002) notes that victims of bias crimes often take
five or more years to recover from the crimes; this is two times greater than the recovery period
necessary for non-bias crime victims to recover from their incidents. Although the reality of the
psychological and emotional effects bias crime victims face is still in need of further research,
importance lies in looking at the significance of anecdotal instances of more severe physical
injury resulting from bias crime victimization.
Injuries caused by bias crime offending often tend to be quite severe. According to
Messner, McHugh, and Felson (2004), bias-based offenses tend to be more likely to result in
injury than all other types of non-bias crime. Although this finding is not universal across victim
groups, evidence seems to suggest such attacks do result in extreme brutality. This evidence of
excessive physical trauma is reported by Levin and McDevitt (1993); however, the supporting
data were from official police data in Boston, Massachusetts. Certainly, underreporting of bias

26

crime victimization mitigates the relevance of this finding. However, anecdotally, it has long
been observed that bias crime victims seem to experience extreme physical brutality in their
attacks. It is not insignificant that anti-LGB attacks seem to be particularly extreme. In such
cases, the minority of incidents involved a gun - a potential indication of the offender’s intent to
brutalize the victim (Lieberman, Arndt, Personius, & Cook, 2001). After all, the intense animus
toward LGB persons could result in a particularly personal – and therefore, brutal - attack on the
victim.
As previously discussed, bias crime is viewed as affecting more than simply the victim,
but the larger social group of which the victim is a part as well. In effect, the bias-based offense
is used to send a message of intolerance and terror to an entire community of persons (Tsesis,
2003; Boekmann & Turpin-Petrosino, 2002; Green, McFalls, & Smith, 2001; Iganski, 2001).
These directed messages of hate can contribute to the occurrence of secondary victimization
(Noelle, 2001). Often, secondary victimization comes in the form of psychological effects –
effects that are reported to be more negative than those arising from non-bias attacks (Rose &
Mechanic, 2002). Although reactions to bias crime victimization in a community can vary, there
is certainly consistency in the effects of the violence on the general community.
In addition to community-level effects of bias-motivated crime, scholars suggest
individuals affected by bias-motivated criminality are less likely to report the incidents due to the
fear of potential retaliation by the offender (Cogan, 2001). This reality has led bias crime experts
to have a skeptical view on the accuracy of bias crime statistics. After all, the “dark figure of
crime” is applicable to bias-motivated crimes as well. Moreover, if the victim identifies as LGB,
she or she may be less likely to report the crime given the particularly low level of lack of
acceptance and tolerance for difference that exists in American society.

27

Anti-Race Bias Crime
Given that anti-race bias crime is the most frequently occurring bias-based type of crime
in the United States, it is rather surprising to most that there is a relatively underdeveloped – and
largely lacking – body of literature on anti-race victimization (Perry, 2003b). Although bias
crime victimization is most certainly problematic for all racial groups not identified with the
white community, bias crime has been especially apparent in its persistence and frequency in
occurrence against the African American community. Although the data relied upon are official
data – for which there are known inadequacies – African Americans are the most frequently
targeted social group for bias crime victimization (Gerstenfeld, 2011). However, not all scholars
agree, as Messner, McHugh, and Felson (2004) found that African Americans’ risk for
victimization is similar to other socially marginalized racial groups. Certainly, FBI data do not
confirm these findings, as 61% of racially-motivated bias crimes were committed against African
Americans in 1995 (Torres, 1999).
As previously mentioned, many scholars consider American enslavement of African
Americans to be one of the earliest – and most egregious – bias crimes ever to occur in the
United States, with the institution’s roots beginning in the 1600s (Gerstenfeld, 2011). After
Reconstruction, bias-based attacks against African Americans continued. For instance, lynching
was used frequently in the South as a terroristic tool to prevent African American citizens from
exercising their civil rights – including voting (Torres, 1999). According to Torres (1999), this
method of terror was used 1,481 times between 1889 and 1918. The overwhelming majority of
the victims were African American.
Moreover, the African American community has long been subject to intimidation in
many forms across the United States. From cross burnings to church arson, bias-based offending

28

against the African American community has been used to send messages of intolerance and
bigotry. Torres (1999) explains that churches have long been important to the African American
community as these are historically one of the first and only independent institutions.
Although the majority of anti-race motivated bias crimes are committed against African
Americans in the United States, other racial groups are negatively affected by bias crime as well.
For instance, Asian Americans have long been exposed to intolerance in the United States.
Internment camps for Asian Americans during World War II are often considered to be one of
the most prominent examples of institutionalized oppression and intolerance against persons of
Asian descent. After the Pearl Harbor bombing in 1941, significant anti-Japanese sentiment
existed in the United States (Chen, 2000). This resistance was extreme enough that the federal
government created internment camps to isolate those Americans with Japanese ancestry during
World War II. Although many defended this bigoted act as being necessary for national security,
it is curious that German Americans were not interned – nor were Italian Americans (Chen,
2000).
More recently, and perhaps one of the most publicized accounts of anti-Asian American
violence, was the murder of Vincent Chin in 1982. Chin was brutally murdered with a baseball
bat by two white men in Detroit, Michigan who espoused anti-Asian sentiments – reportedly
because of recent American automobile manufacturing layoffs (Chen, 2000). At the time,
increased sensitivity to the decline of the American automobile industry – and the perceived
“threat” of Japanese automobile manufacturers’ rising market share – seemed to fuel much of the
animus toward those who identified as Asian American.
Certainly, a long history of bias-based racial oppression and victimization exists in the
United States. Given the nature of this dissertation and the tendencies of African Americans to

29

be singled out for bias-based offending, concentration is placed on the African American
community as opposed to extensive attention to various other racial and ethnics groups. For
instance, in 2007, 36% of bias crimes were committed against African Americans versus 9.5%
against whites, 3.2% against multiracial groups, 8.7% against Hispanics, 2.5% against Asians
and Pacific Islanders, and 5.4% against other ethnicities (Federal Bureau of Investigation, 2007).
Anti-Religion Bias Crime
Second to anti-race based bias crime occurrence is anti-religion based bias crime. There
are innumerable religious faiths in the United States; however, this project does not seek to
identify and survey the experiences of each. Instead, focus will be placed on those of Jewish,
Muslim, and Amish faiths.
With nearly 78.4% of Americans identifying with the Christian faith – whether Catholic
or Protestant – Christians hold a clear majority (Pew Forum on Religion and Public Life, 2008).
Given this, Christians enjoy the ability to control their surroundings to their interests; namely,
this religious group possesses power. This renders others from differing faiths to be part of a
subordinate status. Historically, in the United States, persons of the Jewish faith have been
excluded and persecuted. As previously discussed, this should come as no surprise given Jewish
individuals’ struggle for acceptance and tolerance has been challenging across the world as well.
Anti-Semitism has been in existence for more than a millennium – making it one of the
oldest examples of hatred and intolerance in human history (Gerstenfeld, 2011). Still, antiSemitic bias crime research is limited; as mentioned, the majority of anti-Semitic bias crime data
collection and research is conducted by the Anti-Defamation League (ADL) – a group that has
been monitoring anti-Semitic incidents since 1979 (Anti-Defamation League, 2001). The ADL’s
efforts have largely captured incidents of anti-Jewish crime, bigotry, or intolerance. It should be

30

noted, however, that the ADL’s data do go further than government data in that they include noncriminal events as well as criminal events (Gerstenfeld, 2011).
Although the great majority of anti-religion bias is focused on anti-Semitism, since the
September 11, 2001 terrorist attacks in New York, Pennsylvania, and Virginia, there has been
increasing attention paid to anti-Muslim and anti-Arab bias crime (Gerstenfeld, 2011). As
previously mentioned, there is unsubstantiated perception by some Americans that those of the
Muslim faith – or those who appear to be of Arab descent, though not one in the same – pose a
threat to the security of the United States.
Adding to this discussion is a study conducted by Disha, Cavendish, and King in 2011.
This study examined the reported upsurge in anti-Muslim and anti-Arab violence in the United
States since the terrorist attacks of September 11, 2001. Furthermore, Disha, Cavendish, and
King (2011) were interested in exploring the “spaces of hate” and the “times of hate.” Included
in the discussion is the power differential hypothesis. Similar to its presentation earlier in this
dissertation, this hypothesis suggests that members of a majority group are empowered, and
thereby implicitly encouraged, to act out when minorities are present in low numbers (Levine &
Campbell, 1972). Taking this into consideration, the authors found that temporal and spatial
aspects of anti-Arab and anti-Muslim bias crime led to three salient findings. First, historical
events such as September 11, 2001 can act as a catalyst that allows individuals – in this case,
Americans – to identify a “new enemy” upon which to place acts of hatred (Disha, Cavendish, &
King, 2011). Regardless of nationality, many Americans would be likely to have a perception
that the new enemy is Arab or Muslim. Second, this study accentuates the importance of
understanding that bias crime incidents might occur more frequently in areas where there is a
larger population of a targeted group. For instance, in communities with high populations – or

31

relatively high populations – of Muslim or Arab people, there is a greater likelihood of bias
crime victimization. However, Disha, Cavendish, and King (2011) caution that in reality, the
communities in which Arabs and Muslims were most at risk following September 11th were in
places in which they were few in number. Essentially, the authors contend that there is greater
risk because small oppressed social groups have high visibility and little protection. As they
suggest, there is strength in numbers. This finding falls in line with Levine and Campbell’s
(1972) power-differential hypothesis.
Bias crime is also committed against smaller groups of religious minorities. Often, when
one considers such crimes, anti-Semitic or anti-Muslim bias crime comes to mind. However,
other religious groups – such as the Amish – are affected by bias crime as well. Byers and
Cryder (2002) studied anti-Amish bias offending in the rural Midwest. In an act called
“claping,” the authors qualitatively examined the contributing factors toward the occurrence of
such activity. The study exposed several types of “claping” – including acts such as forcing
Amish buggies off streets with automobiles, turning over Amish outhouses facilities, and
throwing flour at Amish buggies (Byers & Cryder, 2002). Given the authors’ examination of
routine activities theory, it comes as no surprise that offenders view the Amish as a vulnerable,
and thereby suitable, target – especially given their relatively few numbers and differential dress
and lifestyle characteristics. Moreover, in accordance with religious tenets, the Amish are
reluctant to call the police when such crime happens; this is likely to mitigate the effects of
capable guardianship. Motivation was found to be based not only on prejudice, but also on the
notion that the Amish deserved the harm occurring from their victimization (Byers & Cryder,
2002). Although understanding aspects of anti-Amish bias crime does not explain all offending

32

against religious minority groups, it does provide a qualitative perspective of offending through
the theoretical lens of routine activities theory.
Anti-Sexual Orientation Bias Crime
According to the FBI, bias crime committed against individuals due to their actual or
perceived sexual orientation is increasingly problematic. For instance, in 2008 the FBI reported
1,297 bias crimes committed against LGB individuals (Federal Bureau of Investigation, 2008).
Historically, the figures reflecting anti-LGB bias crime place third in frequency – behind race
and religion (Gerstenfeld, 2011). In addition, the figures released from the FBI are only reported
crimes; certainly many anti-LGB bias crimes remain unreported to authorities. McDevitt and
Williamson (2003) note that victims of LGB bias crimes are least likely of all bias crime victims
to report the incidents. Although the exact reasons for this are not known, it is reasonable to
acknowledge that bias against LGB persons is pervasive throughout American culture. Yang
(1997) suggests that anti-LGB behaviors – such as bias crimes - are essentially actions taken
with the support of values present in a community’s culture. This differs from other types of
bigotry as anti-LGB prejudice – and often violence - is frequently overlooked by many social,
religious, and governmental institutions (Herek, 1989).
In perhaps one of the earliest and most salient studies of anti-LGB violence, Comstock
(1991) provides an overview of both anti-LGB bias based victimization and offending. With
regard to the gender of the anti-LGB victim, males do tend to be victimized more frequently –
but not by much. This finding is similar for rape. In addition, persons of color are more
frequently victimized than are those identified as white (Comstock, 1991). However, these
conclusions only reflect physical effects of anti-LGB victimization – not the well-noted
emotional damage that often occurs as well. Moreover, Comstock (1991) notes that LGB

33

persons are far more likely – regardless of racial identity – to be victims of violent crimes than
are citizens of the general population.
In addition, Comstock (1991) notes that the context of anti-LGB bias crime is important
to consider. Essentially, Comstock (1991) suggests that – all things equal – the majority of antiLGB bias crimes occurred near “LGB” spaces. For instance, 59% of victims reported that the
bias crimes occurred near a public gay/lesbian area such as a bar or “bath house.” It follows that
LGB individuals more likely to be present in such areas, and therefore are vulnerable to attacks.
In these settings, LGB people are both present and in many cases, identifiable (Comstock, 1991).
However, caution should be taken when assuming all individuals in a “LGB space” identify as
such, as the perception of having an LGB identity is enough to incite victimization.
In his 1991 bias crime survey, Comstock provided a foundation of knowledge upon
which much understanding of anti-LGB bias crime offending is based. In comparing his survey
data to national crime statistics, Comstock (1991) concluded that contextual perpetrator
information if similar to other bias crime offending incidents. For instance, 94% of perpetrators
of anti-LGB violence are male; this is slightly greater than the figure for violence against the
general population. It is commonly known that males are engaged in the most criminal
offending; in this case, males are the perpetrators in 78% of violent victimizations (Comstock,
1991). Additionally, Comstock (1991) notes that the majority of the anti-LGB perpetrators are
white (67%), which is in contrast to 69% for all violent crimes. Moreover, Comstock (1991)
reports that anti-LGB offenders are younger than offenders of general crimes. This differential is
comparatively noticeable with 46% of anti-LGB offenders being under the age of 22. This figure
is 29% for offenders in the general population (Comstock, 1991). Also noteworthy is the fact
that the many anti-LGB violent crimes are committed in groups. These are not necessarily

34

organized hate groups; rather, they typically are comprised of groups of young males. Comstock
(1991) found that in the general population, only 27% of violent crimes are committed in groups;
for anti-LGB violent crimes, that figure climbs to 48%.
Furthermore, in a sample of 290 LGB identified individuals, Rose and Mechanic (2002)
report that 73% of the respondents had been victim in at least one homophobic event. The
seriousness of the victimization varied considerably, from sexual assault to threats of violence or
other acts. Given the purpose of the study was to determine the psychological effects of antiLGB bias crime victimization, it is not surprising that homophobic sexual assaults had the
highest incidence of psychological distress (Rose & Mechanic, 2002). In addition, Rose and
Mechanic (2002) found that those victimized by anti-LGB sexual assault were more likely to
report these offenses to the police than were other anti-LGB bias crime victims.
Superordinate Group Victimizations
In the discussion regarding bias crime victimization, many members of superordinate
groups often inquire about their inclusion in bias crime victimization. According to some
scholars (Jacobs & Potter, 1997) attention to bias crime is relegated to mere identity politics and
ignores the victimization of majority groups.
Although it is certainly true that any individual can be targeted because of her or his
identification with a social group, the problem of bias crime victimizations is most often a result
of power differentials. For instance, Gerstenfeld (2004) suggests it is not usually the straightidentified individuals concerned about being targeted by LGB-identified individuals.
Nonetheless, there have been instances in which super-ordinate group members have been
victimized because of their larger social group identification. To accommodate these instances,
most bias crime language is framed to be inclusive of all social groups of people. This includes

35

majority groups such as Whites, Christians, and heterosexuals. Violence against any individual
is certainly worthy of attention; however, this dissertation seeks to focus efforts on the
victimization of marginalized social communities – specifically communities of color, religious
minorities, and the LGB community.
Bias Crime Offending
In order to fully understand the scope of the bias crime problem in the United States, it is
important to consider the context surrounding bias crime offenders as well as victims. Therefore,
this dissertation offers a discussion of not only typical offenders, but the social context that
surrounds these offenders. Hate and intolerance and their dissemination throughout American
culture are certainly imperative to the understanding of bias crime occurrence. Individuals
promulgating hateful and intolerant speech are considered as well. In addition, discussion of a
major hate organization – the Ku Klux Klan – provides context as well.
Individuals and Intolerance
Given that bias crime research is a relatively new scholarly endeavor, understanding of
bias crime offenders is limited in the scholarly literature. However, as mentioned in his
ethnographic study of hate groups, Ezekiel (1995) notes that the majority of bias crimes are
actually committed by individuals – not organized hate groups. Of course, as noted by
(Comstock, 1991) many of these individuals act together as a group during the commission of
bias crimes. This fact stands in opposition to what most common view of hate groups as the face
of hatred in the United States. In actuality, the typical bias crime offender tends to be young and
male – sometimes offending in small unaffiliated groups (Franklin, 2002). Moreover, from their
study of bias-motivated assaults, Messner, McHugh, and Felson (2004) found that a large
number of bias crime offenders were intoxicated at the time of the incident.

36

Although there has been some progress with the bias crime offender research,
significance lies in examining the bigoted attitudes and speech that fuels the individual acts of
bias-motivated crimes. Certainly, hate speech is protected by the First Amendment to the United
States Constitution; however, the vitriol espoused by some may be linked to the encouragement
of bias-motivated criminality (Tsesis, 2000).
Hate Groups
Bias, prejudice, and hate are not relegated to individuals. There are – and long have been
– many hate groups in the United States. Perhaps the best known voice of hate, the Ku Klux
Klan has had various periods of popularity nationwide. With origins dating back to
Reconstruction Era Tennessee, the Ku Klux Klan has always stood for hate and intolerance
disguised as an organization seeking “100% Americanism.”
Although the Klan originated in Tennessee during Reconstruction, its ugly head has
surfaced many times in American history – from 1965 to the present day. The second wave of
Ku Klux Klan activity started in the early 1920s as a response to a changing racial and religious
climate in the United States. Given this dissertation seeks to focus on counties, it is important
acknowledge in depth the influence The Ku Klux Klan had over one particular community. As
such, the Ku Klux Klan reared its ugly head in Detroit, infiltrating the political and social stage
of the nation’s fourth largest city at the time (Widick, 1972).
The Ku Klux Klan was founded on fundamentalist principles of Christianity and thereby
its main purpose was to promote hate and intolerance toward those who were not White,
Protestant, and heterosexual. In their own publication dating back to the early 1900s, the Ku
Klux Klan defined many of their ideals, one of which was white supremacy. Americanism - a
form of nationalism, and Protestantism were also promoted (Widick, 1972). Therefore, in

37

Detroit, a city with a growing number of European Catholics and Southern African Americans,
the Ku Klux Klan found much inspiration for flourishing throughout much of the early 1900s
through the 1950s (Widick, 1972).
According to Jackson (1967) the first Kleagle, or chapter of the Detroit Ku Klux Klan,
was formed in 1921. Although the organization had a slow start, by 1925 there were over
100,000 Ku Klux Klan members in Detroit (Boyle, 2005). The membership consisted of mainly
White, Protestant, and working class individuals from the Detroit area (Widick, 1972). There
were other Kleagles in Michigan, but Detroit was the stronghold for the entire state (Jackson,
1967). Although the Ku Klux Klan had a significant membership in the early 1920s, it was not
quite as socially unacceptable to be a member as would be today. Without a doubt, the Ku Klux
Klan was considered a hateful group in the 1920s; however, the following was much stronger,
perhaps empowering the unfortunate reality for which the organization stood. The Ku Klux Klan
became so socially acceptable that even President Warren G. Harding (OH) was a member of the
certified hate group (Boyle, 2005). Certainly, the President’s association with the hate group
indicates the level of national acceptance of the exclusion and oppression of certain social
groups.
The Ku Klux Klan was fairly prevalent in Detroit’s culture as well. It was not
uncommon during the early 1920s to see Ku Klux Klan advertisements in The Detroit Free Press
seeking more members or announcing rallies (Jackson, 1967). In addition, Jackson (1967)
explains that there were smaller subgroups within the Detroit Ku Klux Klan for boys and for
women called the Junior Klan and the Women of the Ku Klux Klan, respectively. Therefore,
there was no getting around the fact that the Ku Klux Klan had a strong grip on Detroit. As the

38

city kept changing demographically and socially, the Ku Klux Klan continued to grow in
strength.
Although not always visible to Detroit residents, the Ku Klux Klan was alive and
prosperous in the early 1920s. This led to the common occurrences of cross burnings and rallies
in the Detroit area. The first major rally by Ku Klux Klan members was held in Royal Oak in
April of 1923 (Jackson, 1967). Approximately eight thousand Ku Klux Klan members gathered
to listen to the grand goblin discuss the superiority of the white race, both physically and
mentally, in the Great Lakes region (Jackson, 1967). According to Jackson (1967), in an
apparent attempt to prevent the Ku Klux Klan from holding future rallies, the State of Michigan
passed the 1923 Burns Act which outlawed public meetings of masked men. However, this did
not stop the Ku Klux Klan from rallying. Soon after the passing of the law, an unmasked group
of 25,000 to 50,000 Klan members rallied in Dearborn, Michigan (Jackson, 1967; Boyle, 2005).
Boyle (2005) tells the story of one of the most infamous of Detroit’s race-related events
that occurred on a summer evening in 1925. Dr. Ossian Sweet, an African American doctor who
had moved into a prominently white neighborhood, found himself subject to a crowd of jeering
Whites surrounding his house in protest. Although there were 17 police officers posted for
protection at Dr. Ossian Sweet’s new house, the crowd of angry Whites overwhelmed Sweet and
his family. Armed with rifles for protection, Dr. Sweet and his family spent the first night in the
home without incident. The second night, violence broke out and two White men in the angry
mob were shot from inside Sweet’s house. One man was killed and another was only injured.
The Sweet family was arrested and detained, but later acquitted of all charges. While the Ku
Klux Klan was never officially involved in the incident, most believe that the organization had
significant influence on the planned mob scene at Dr. Sweet’s home. With the proliferation of

39

Ku Klux Klan members in Detroit at the time, it is highly likely that there was involvement of
the Klan.
The Ku Klux Klan was influential to many aspects of Detroit in the 1920s, one of which
was local politics. In one of the Klan’s first attempts at politics, the organization proposed to
outlaw parochial schools in Michigan (Widick, 1972). The proposal was defeated; however the
political move by the Ku Klux Klan reflected the still deep hatred for Catholics in the Detroit
area by Klan members.
In yet another political campaign in 1924, the Ku Klux Klan sponsored write-in mayoral
candidate Charles Bowles who nearly won the election (Jackson, 1967). Bowles ended up losing
the election by a narrow margin, although many in Detroit contended that he actually won.
Claims were then made that the votes were tallied wrong in an attempt to prevent Bowles from
taking the office of mayor of Detroit (Jackson, 1967). The winner of the election, Mayor John
Smith, though not a Ku Klux Klan supporter, ended up appeasing their desires for hatred and
intolerance. In a published letter, Smith told African Americans to stay in “their own
neighborhoods” and not to move to White spaces of Detroit (Jackson, 1967). While this action
had nothing to do with the Ku Klux Klan, Smith essentially pleased many of the hateful Whites
in the Detroit area by writing the letter.
To many, the Ku Klux Klan was one of the most influential organizations that could be
found in Detroit in the 1920s. Although it was not always visible, the Ku Klux Klan made every
effort possible to recruit new members, integrate itself into the local press, and legitimize itself
with White citizens based on its desire for the attainment of “100% Americanism.” However,
through its many cross burnings, rallies, sour political endeavors, and hateful protests, Detroit
citizens became wise to the error of the Klan’s ways. By 1934, the Ku Klux Klan became

40

disorganized and died out in Detroit. The hatred that was the driving force of the Klan was not
necessarily dead; however, an organized outlet no longer existed in the Detroit area (Jackson,
1967).
The Influence of Religion
Many scholars have associated religion to the promotion of bigotry in a variety of ways.
For instance, Craig (2002) suggests that many religions have anti-LGB views that guide their
interpretations of sacred doctrine. Moreover, it is no secret that organized religion has been
exclusive of women throughout history. Although neither of these examples is universal – given
there are numerous religious institutions supportive of social justice and acceptance of diversity
– organized religion in the United States has provided a vehicle to mobilize those who are
inclined to retain bias against certain communities.
Primarily, this is true of Christianity in the United States. For instance, interpretations of
the Christian bible have been used to justify slavery, anti-Semitism, segregation, the oppression
of women, and the persecution of the LGB community, among others. Unfortunately, this is not
limited to history books; rather, the intolerance fueled by the religious community is still seen in
the 21st century. To be equitable, there are many Christian organizations that pride themselves
on their contribution to the advancement of human rights and social justice. Specifically, the
United Church of Christ has been supportive of the LGB community for decades; more recently,
the Episcopal Church has pronounced its support as well. Moreover, at times the Roman
Catholic Church and varieties of Lutheran, Presbyterian, and United Methodist churches, have
vowed to support a mission of social justice – albeit with a discriminatory notion of inclusion.
As time progresses, it is likely that the number of parishes supporting social justice and
acceptance of LGB populations will grow. Such has been the reality over the recent decades.

41

Theoretical Perspectives on Bias Crime Incidence
Given that consideration of bias crime is in its infancy, the volume of research
surrounding bias crime offending remains largely underdeveloped (McDevitt, Levin, & Bennett,
2002). Nevertheless, various scholars have introduced limited theoretical perspectives useful for
understanding bias crimes and the related policy implications. Green, McFalls, & Smith (2001)
provide a framework of typologies in which bias crime theories can be organized, while
considering that theoretical perspectives generally either consider an individual or societal unit of
analysis. In order to provide a comprehensive overview and critique of current theoretical
perspectives on bias crime, the following typologies are used: psychological; socialpsychological; sociological; economic; and political (Green, McFalls, & Smith, 2001).
Psychological Perspectives
Bias crimes can be examined from a societal perspective or an individual perspective
(Green, McFalls, & Smith, 2001). Psychological theories used to contribute to an understanding
of bias crimes suggest an individual – or micro - unit of analysis. Given the difficulty of
studying individuals involved in bias crimes, most theoretical accounts of bias crimes tend to
consist of a societal unit of analysis (Green, McFalls, & Smith, 2001). However, there are
psychological perspectives that can be useful in attempting to explain various components of
bias crimes. It is common for bias crime scholars to acknowledge the influence of prejudice and
the authoritarian personality in bias crime offending.
Prejudice
Although prejudice does not directly cause a person to commit a bias crime, it can
undoubtedly become an integral part of an individual’s personality and therefore an informant to
their outward actions. It is commonly understood that prejudice is a learned behavior and not

42

innate (Allport, 1954). Allport (1954) also suggests that the formation of prejudice begins early
and occurs in several stages. First, children learn to take part in social categorization in various
ways. For instance, a child might begin categorizing others by race, gender, and age (Allport,
1954). However, there are numerous ways that children and adults can categorize other
individuals. Second, children begin to learn about in-groups and out-groups and therefore learn
the concepts of and differences between “us” and “them” (Allport, 1954). Third, children learn
about the various stereotypes affiliated with different social groups (Allport, 1954). Allport
(1954) describes these stereotypes as exaggerated beliefs that are linked with certain social
groups. These judgments about individuals that identify with certain groups can be positive or
negative, but are generally always exaggerated and incorrect (Gerstenfeld, 2004). Moreover,
Allport (1954) suggests that these stereotypes aid in the creation of prejudice which is defined as
“an avertive or hostile attitude toward a person who belongs to a group, simply because he
belongs to that group, and is therefore presumed to have the objectionable qualities ascribed to
that group” (p. 7).
Prejudice can have various effects on individuals. No one is a neutral observer; however,
there is a difference between those who perceive others with prejudice and those who do not
(Gerstenfeld, 2004). Optimally, individuals find themselves without the need to prejudge others
based on various characteristics. However, when an individual is influenced by prejudice, the
outcome can be quite severe. In his ethnographic study of the American radical right, Ezekiel
(1995) presents numerous examples of extreme prejudice. One such example involves Arthur, a
member of the Ku Klux Klan, who prejudicially suggested that “Blacks don’t make as much
‘cause they are all out raping and robbing White people” (p. 21). Ezekiel (1995) also presents
Arthur’s anti-Semitic views that Jewish people engineered and invented abortion, and, “They

43

crucified Christ! They crucified Christ. In my opinion, they did it” (p. 22). It is evident that not
only was Arthur particularly misinformed, but he held very deep prejudices against at least two
categorical groups. Although not all prejudices are as hateful and intolerant as those presented
here, it is plausible that prejudicial attitudes, when coupled with affective disorders, can
contribute toward acts of avoidance, discrimination, assaults, and even social extermination
(Allport, 1954).
The effects of prejudicial attitudes can be quite disagreeable; however, research
consistently shows that education is negatively related to prejudice (Wilson & Ruback, 2003).
For instance, in their study of racial attitudes, Schuman, Steeh, Bobo, and Krysan (1997) found
that the higher the education level of a Caucasian community, the greater acceptance of African
American individuals in schools and homes. That education is negatively related to prejudice
leads Wilson and Ruback (2003) to suggest that it is possible for more highly educated
communities to have lower bias crime statistics – and less severe bias crimes – than lesseducated communities. Undoubtedly, that education negatively affects prejudice is particularly
helpful in efforts to understand bias crimes. However, how education affects prejudice can only
be used to provide information about psychological traits that might influence bias crime
offending.
Authoritarianism
The authoritarian personality was first introduced by Adorno, Frenkel-Brunswik,
Levinson, and Sanford (1950) in their study examining right-wing extremism. The authors
suggest that that authoritarian personality could be characterized by the following factors: a need
for rigid and stereotyped views; a low tolerance for ambiguity; and an attraction to
uncomplicated, authoritative, and conspiratorial ideologies (Adorno, Frenkel-Brunswik,

44

Levinson, & Sanford, 1950). Given that prejudice can be learned, Adorno, Frenkel-Brunswik,
Levinson, and Sanford (1950) introduce the idea that parents with an authoritarian personality
are likely to raise children who view the world as hierarchical instead of equal and who view
power and authority as the most important components in any relationship. Moreover, Adorno,
Frenkel-Brunswik, Levinson, and Sanford (1950) suggest that these children have been taught to
view internal impulses, such as disregard for rules, as evil and therefore have been encouraged to
fight them. Children may project these impulses on others; consequently, this suggests they
cannot be trusted (Adorno, Frenkel-Brunswik, Levinson, & Sanford, 1950). Additionally, the
children of authoritarian parents fear showing aggressive behaviors to their parents and
consequently transfer their aggression to other children, who in their view are vulnerable
(Adorno, Frenkel-Brunswik, Levinson, & Sanford, 1950).
Authoritarianism can certainly provide some insight into the workings of the mind of
some bias crime offenders. There seems to be some anecdotal truth to Adorno, FrenkelBrunswik, Levinson, and Sanford’s (1950) perspective; however, the perspective should be
considered with caution. In Green, Adelson, and Garnett’s (1999) study, attitudinal surveys were
used to determine if there was a particular personality profile for bias crime offenders. From the
authors’ findings, it is apparent that there is no uniform psychological profile for bias crime
offenders. There may be many people with authoritarian personalities; however, there are very
few individuals who espouse the authoritarian personality and commit bias crimes (Green,
Adelson, & Garnett, 1999).
From the perspectives presented, prejudice and authoritarianism are likely only to provide
insight into an individual’s personality. It is evident that neither psychological trait is likely to
be causally related to bias crime offending. The two perspectives discussed do seem to

45

contribute toward an understanding of bias crime offending; however, they also provide insight
into less serious instances of intolerance. Therefore, the inconclusive psychological perspectives
might lead scholars to examine broader, macro-level perspectives.
Sociological Perspectives
Unlike psychological and social-psychological perspectives which focus more on the
individual involved in bias crime offending, sociological accounts provide a broader, more
abstract explanation (Green, McFalls, & Smith, 2001). To begin, Blumer’s group position
perspective provides a sociological interpretation of prejudice.
Prejudice and Group Position
Bigotry or prejudice toward another group is always present in bias crime offending
(Levin & McDevitt, 1993). Blumer (1958) suggests that racial prejudice exists in a form of
group position as opposed to particular sentiments that one group has toward another. In other
words, a group with power over another simply acts in a manner that opposes the subordinate
group’s desire for power and privilege (Sidanius & Pratto, 1999). Essentially, the dominant
group functions to preserve its position, which is that of the dominant group (Blumer, 1958).
Blumer (1958) describes race prejudice as the result of four feelings that are always
present in the dominant group. Although prejudice is not limited to dominant racial groups, it is
no secret that dominant groups that generally express the majority of prejudicial- and therefore,
oppressive - sentiments. In the United States, Caucasians tend to find themselves in dominant
positions more frequently than other racial groups. For example, some whites often embrace the
following four feelings toward African Americans and other oppressed racial groups: that they
are superior, that the subordinate race is intrinsically alien or different, that they have a right to
claim certain privileges and advantages, and that they often possess a fear or suspicion that the

46

subordinate race has plans to overtake the interests of the dominant race (Blumer, 1958). Racial
prejudices, then, can be seen as stemming from a perceived challenge to these feelings by
subordinate groups.
Prejudiced members of dominant racial groups also maintain that they have a claim to
certain privileges and advantages (Blumer, 1958). For instance, from informal observation, the
“average” American is often described as a Caucasian man and not a member of any other racial,
gender, or ethnic group. This implies, of course, that America is a “White man’s country” and
with that ownership comes privilege. With white privilege, a Caucasian person is “entitled” to
certain advantages such as claims of property; the right to certain jobs; occupations and
professions; access to the law; positions of control in the government; and the right to certain
schools, clubs, and churches (Blumer, 1958). When these privileges are threatened, prejudice
can appear as a result (Blumer, 1958); and as suggested, prejudice forms the perfect foundation
for bias crime offending.
Additionally, Blumer (1958) explains racial prejudice through group function and adds a
sociological bent to what is usually considered a psychological problem. While the group
position perspective offered by Blumer (1958) can be quite helpful in understanding the
underpinnings of bias crimes, it is not able to explain the phenomena in their entirety.
Social-Psychological Perspectives
Social-psychological explanations of bias crimes focus on the relationship between
individual psychological characteristics and larger societal influence (Green, McFalls, & Smith,
2001). Given most theoretical development pertaining to bias crimes has occurred at the societal
level, social-psychological perspectives are well suited to aid in the explanation of how social

47

conditions affect individuals. Consequently, there is a perspective scholars have attempted to
use to understand and explain bias criminality - social learning theory.
Social Learning Theory
Recently, social learning theory has been considered helpful in explaining bias crimes.
Social learning theory is defined as a cognitive process in which an individual’s personality and
environment continuously interact (Beirne & Messerschmidt, 2000). Through a social learning
perspective, there is some evidence that bias crime perpetrators are influenced by the various
attitudes around them (Gerstenfeld, 2002). Whether affected by the attitudes of parents, friends,
or the media, bias crime offenders have shown that they espouse many of the same bigoted views
and fear of diversity as those who influence them (Gerstenfeld, 2002).
Two studies have supported social learning theory as an explanation for bias crime
offending. One study conducted by Byers, Crider, and Biggers (1999) examined incidents of
claping in the Midwest. Claping can be defined as anti-Amish bias crime. Findings suggest that
participants in bias crimes against the Amish gained their belief and attitudes from their peers
and families. Another study examined bias crime offenders who had committed anti-LGB bias
crimes. Similar the Byers, Crider, and Biggers (1999) study, Franklin (2000) found that those
who had committed an anti-LGB bias crime had friends who shared their anti-LGB ideologies.
Although it is clear that bias crime offenders must be learning their prejudices from those
who surround them, social learning theory is relatively weak in explaining bias crimes. Social
learning theory does help explain how certain individuals gain their prejudiced or bigoted beliefs,
but the theory does little to help us understand why these particular individuals take part in bias
crime offending.

48

Economic Perspectives
Examining bias crimes from a broader, societal perspective, economic theories of bias
crime present closely aligned arguments. Like other typologies of bias crime theories, economic
perspectives are still unable to fully explain bias crime offending. However, of the bias crime
theories, those with an economic basis have historically been the most popular explanations of
bias crime offending (Hamm, 1994). Consequently, there are numerous anecdotal examples of
how economic conditions might affect bias crime offending.
Realistic Group Conflict Theory
Realistic group conflict theory suggests that inter-group hostility emerges from power
differentials and conflict over scarce economic resources (Gerstenfeld, 2002; Campbell, 1965;
Green, McFalls, & Smith, 2001). Inter-group hostility can arise from a number of situations.
Early on, researchers suggested that group conflict theory could explain the negative relationship
between the economic conditions of the South and the frequency of lynching of African
Americans (Hovland & Sears, 1940). Essentially, Hovland and Sears (1940) found that as cotton
prices lowered and the economy turned downward, the rate of lynching increased, implying
Whites attacked African Americans when they believed that their privilege and access to scarce
resources were threatened. Moreover, in a more recent study, Jacobs and Wood (1999) found
cities that had greater economic competition between African Americans and Caucasians had
greater white killings of African Americans.
Perhaps a reason for realistic group conflict theory’s popularity is the apparent
availability of related anecdotal stories. For instance, in 1982, Vincent Chin - a Chinese
American man was murdered in Detroit, Michigan by laid off American automobile workers
(Perry, 2001). Apparently Chin was mistaken for a Japanese American, a group that the

49

automobile workers blamed for losing their jobs and viewed as taking over their privileged
access to the American automobile market. Yet another example tells a story of the Ku Klux
Klan terrorizing Vietnamese shrimpers in Texas in 1981. White shrimpers in Galveston Bay felt
that newly immigrated Vietnamese shrimpers were competing unfairly and therefore called upon
the Ku Klux Klan to intimidate immigrants into leaving the area (Levin & McDevitt, 1993).
Months of terror ensued, resulting in property destruction, violence, and intimidation (Stanton,
1992). In both cases, Vincent Chin and the Vietnamese shrimpers, economics were at least part
of the drive toward committing bias crimes. American Whites placed blame on another racial
group for their economic losses, regardless of any merit to their claims that those groups were
responsible for those losses. After all, a social construct such as race – at its sociological base is unlikely to contribute to a prediction of economic downturn.
In a slightly different manner, some scholars suggest that strain theory contributes to bias
crime offending. Rather than a discussion of group conflict, in this context, strain theory
explains how individuals become economically and socially strained when there is a shortage of
jobs and opportunity to engage in the socially accepted aspects of society. Potentially, would-be
bias crime offenders reason that their situation would be brighter if it were not for members of
other racial or socially oppressed groups taking “their” jobs. Mindsets such as this could
therefore lead to a rise in bias crimes against people of different ethnicities and races (Anderson,
Dyson, & Brooks, 2002).
Defended Neighborhoods Theory
Given the popularity of economic perspectives for explaining bias crimes, it would be
rather easy to assume that scholars have arrived at a definitive understanding of the cause of bias
crimes. Rather, recent research on economic conditions and bias crime offending has provided

50

surprising results. Green, Strolovitch, and Wong (1998) provide evidence that economics do not
in fact affect bias-motivated behavior. In their study, the researchers examined unemployment
data in New York City between 1987 and 1995. They found no link between the city’s monthly
unemployment rate and bias crimes (Green, Strolovitch, & Wong, 1998). Instead, the findings of
the study suggest that it was an influx of ethnically diverse people moving into formerly
homogeneous neighborhoods that is linked to bias crime offending. Essentially, Green,
Strolovitch, and Wong (1998) propose the perspective that residents of homogenous
neighborhoods use bias crimes as a method of “defending” their neighborhood from new
residents who may be different in some way. The defended neighborhoods thesis differs from
the realistic group conflict theory in that the new in-migrants of a neighborhood are not creating
new power differentials; rather, they are changing the collective identity of the neighborhood
(Green, McFalls, & Smith, 2001). However, as residents of the neighborhood familiarize
themselves with the newcomers, the hostility is likely to transform into indifference or even
acceptance (Green, McFalls, & Smith, 2001).
Green and colleagues’ (1998) findings were perhaps among the most important to the
development of bias crime theory. In one study, the authors contested the long-held beliefs that
downturns in the economy are likely to encourage an increase in bias crimes in a community and
introduce a new perspective - defended neighborhoods. The defended neighborhoods
perspective can certainly aid in the understanding on bias crimes in neighborhoods, and even at
larger levels such as cities; however, the perspective should be considered with caution as it
ignores various influences such as those of politics (Green, McFalls, & Smith, 2001).

51

Scapegoat Theory
Closely related to realistic group conflict theory and strain theory, scapegoat theory
(Allport, 1954) suggests that when a downturn in the economy occurs, people lash out at
convenient groups. It is important to emphasize that the out-groups are those that are convenient
to blame even if there is little evidence of their involvement. However, the groups that are
selected are generally the out-groups for which the offender maintains some level of frustration
and anger (Gerstenfeld, 2004).
Perhaps the prime example of scapegoat theory is the representation of Middle Easterners
as violent zealots and terrorists after events such as September 11, 2001 or the Oklahoma City
bombing (Gerstenfeld, 2002). Although much of the scapegoating can be implied by the media,
these sentiments can be initiated by politicians and other interest groups (Green, McFalls, &
Smith, 2001). For instance, in an attempt to find groups to blame, or scapegoat, following the
attacks of September 11, 2001 and the economic downturn that was a result, the Christian pastor
Jerry Falwell blamed the attacks on so-called “abortionists, pagans, feminists, the LGBT
community, and the American Civil Liberties Union” (Gerstenfeld, 2004; Green, McFalls, &
Smith, 2001). Regardless of the facts, the usual lineup of out-groups is often blamed for events
such as a September 11, 2001 and downturns of the economy.
Political Perspectives
As a societal-level approach to explaining bias crimes, political typologies have been
used to understand, at least in part, why bias crime offending occurs. Many would even argue
that bias crimes exist as a construction of use for political gain (Jacobs & Henry, 1996).
However, most agree that bias crimes do indeed exist as a problem and require attention political

52

attention. With that taken into consideration, the influence of cultural ideologies and social
movement theory has been discussed in the literature.
Cultural Ideology Influence
The influence of cultural ideology speaks to the values, shared beliefs, and customs of a
society (Herek & Berrill, 1992). Although bias crimes occur within an extremely broad cultural
context, cultural ideologies do affect bias crime offending (Craig, 2002). One such ideology is
that of violence. Craig (2002) suggests that violence is increasingly considered as an option
when facing conflict; and at times, ethnic and racial oppressed social groups are used as
scapegoats for these bias crimes.
In addition, if a political culture is sympathetic to intolerance, the result can be prejudice
and bigotry – potential indicators of bias crime (Craig, 2002). For instance, during the Reagan
era of the 1980s, progressive legislation aimed at civil rights for all people was attacked (Craig,
2002). Specifically, the Reagan administration opposed the previously passed Voting Rights Act
of 1965 and the Equal Rights Amendment, which was defeated in 1983. This political culture
created a climate of intolerance, leading some to believe that bigotry, prejudice, and bias crimes
were acceptable (Pinkney, 1994).
Social Movement Theory
Social movement theory has been used to help explain bias crime in terms of a political
perspective. Green and colleagues (2001) suggest that bias crime offenders are motivated to take
action based on “political opportunity” structures. In addition to any bias these offenders may
already have toward their victims, social movement theory suggests that various political factors
may contribute to acting on one’s bias. For instance, any political legitimization of biases may
be likely to embolden a would-be offender and is considered a political opportunity (Karapin,

53

1996; Koopmans, 1996). In addition, the authors suggest the current likelihood of punishment
based on bias offending is yet another political opportunity.
There are different ways to interpret the salience of the cultural ideologies and social
movement perspectives. As Craig (2002) implies, political conservatism may very well provide
the foundation for “culturally approved” intolerance. Moreover, lack of political action may
leave the door open for bias-related crimes to be tolerated. For instance, in a radio interview
following the September 11, 2001 attacks, U.S. Representative and Republican John Cooksey
from Louisiana made the statement, “Someone who comes in that’s got a diaper on his head and
a fan belt wrapped around that diaper on his head, that guy needs to be pulled over” (Gerstenfeld,
2004). It is clear that Representative Cooksey was interested in transferring his bigoted views to
the public. This politician’s influence certainly affects the general cultural ideology in America;
however, it is difficult to determine whether such accounts directly cause bias crimes to occur.
In contrast, given that bias crimes first attracted wide attention in the 1980s (Green,
McFalls, & Smith, 2001), it is evident that attention was paid to the brutal outcomes of bigotry
and prejudice – even during the ultra-conservative Reagan administration. The lack of
credibility these perspectives are given in the criminological literature is likely due to the
assumption that the cultural ideology and social movement claims are grounded in ideological
bickering. After all, it was the socially conservative President George H.W. Bush who signed
the Hate Crime Statistics Act of 1990 into law (Gerstenfeld, 2004) and several of the most
publicized bias crimes occurred during President William J. Clinton’s administration. As
suggested, the influence of cultural ideology speaks to the values, shared beliefs, and customs of
a society (Herek & Berrill, 1992) and it likely affects political social movements as well.

54

Perhaps more research is necessary in determining the affect politics can have on cultural
ideologies and social movements as they pertain to bias crime.
There are very few empirical studies explaining the nature of bias criminality. The
reason is likely due to the relatively short amount of time that bias crimes have come to the
attention of the public and the controversy that seems to surround the topic. However, a few of
the existing theories have made an impact on our understanding of bias crimes. Many scholars
still find some truth to economic theories since studies have provided telling results. At the same
time, Green and colleagues’ (1998) defended neighborhoods thesis seems to move us away from
economic theories and toward perspectives focused on demographic movement. Therefore, this
dissertation seeks to explore the macro-level effects bias criminality has on American counties.
As such, social disorganization theory is most appropriate for the current study examining the
relationship between counties and bias crime occurrence.
Policy Perspectives on Bias Crime
Inevitably, when a social problem emerges as salient, it becomes part of a larger dialogue
within the realm of public policy. Namely, once bias crime became an issue for which attention
was paid, individuals and interest groups alike began to put pressure on the government to solve
the problem through public policy. Although there have been a variety of ideas presented, the
most frequent proposed solution to a social problem such as bias crime is legislation. As such,
an overview of the legislative attention to bias crime incidence is warranted.
Bias Crime Legislation and Controversy
In the aggregate, there is relatively little controversy over the negative judgment assigned
to the occurrence of bias crime. Most fair-minded individuals realize that no person should be
victimized for any reason; however, the debate begins to expand when taking a victim’s identity

55

into account. Most social scientists proffer that in the name of equality, all victims should be
treated the same under the law. Yet others take into account the disproportionate levels of social
power afforded groups in society and the resultant direct harm placed on marginalized
populations through various types of victimization. Regardless, this presents a situation where
public figures must decipher a solution to the problem of bias crime victimization. Generally,
this comes in the form of legislation. As expected – and with most topics handled by way of
politics – there is great controversy over the nature of bias crime laws.
The advent of bias crime legislation took place in 1990 with the Hate Crime Statistics Act
approved by the United States Congress. This legislation required the FBI to collect data about
the frequency of bias crime offending. Essentially, this meant that for the first time in American
history, the federal government would have an official overview of the frequency of bias crime
offending as reported by each state. Nevertheless, even this law allowing the FBI to collect data
on bias crime was met with controversy. In addition to resistance from conservative Republican
politicians, many law enforcement officials were skeptical of bias crime data collection as well.
Bias Crime and Law Enforcement
Regardless of potential good will, law enforcement – both the police and prosecutors –
have had difficulty navigating their reactions to bias crime offending and the related legislation.
Perhaps the most salient reason for this difficulty is the ambiguity of bias crime incidence. As
previously stated, bias crimes are often difficult to identify. Unlike all other crimes, for bias
crime prosecution, the offender’s motive must be proven (Franklin, 2002). Consequently, law
enforcement officials often have difficulty determining this given an offender’s motive is often
unknown (Franklin, 2002). Law enforcement officials are tasked with not only protecting the
public, but also identifying social problems for further public attention. As such, Steinberg,

56

Brooks, and Remtulla (2003) have suggested a “but for” rule is of aid to the law enforcement
community. In essence, the “but for” rule helps the police and prosecutors determine whether a
certain criminal act was indeed a bias crime. For instance, using this rule, a law enforcement
official would be able to state, “but for the hate motivation, the crime would never have been
committed” (Steinberg, Brooks, & Remtulla, 2003). In addition, enforcement of bias crime is
largely discretionary (Haider-Markel, 2002). Discretion is a significant aspect of law
enforcement activity – whether when determining the type of crime that has occurred at the
scene or when determining the proper charge for the defendant. After all, police officers have
long been considered “street level bureaucrats” who are directly responsible for numerous public
services – including the identification of bias crime occurrence (Lipsky, 1980). Although there
had been an upsurge in training for law enforcement in the identification of bias crime, this
activity has mainly been relegated to large cities (Haider-Markel, 2002).
Additionally, ideological influences have had an effect on attitudes regarding bias crime
in the law enforcement community. Notably, the majority of prosecuting attorneys tend to be
politically and socially conservative (McPhail & DiNitto, 2005). Given this, it is no surprise that
prosecutors have been reluctant to pursue bias crime charges. Moreover, scholars and
prosecutors alike acknowledge the evidentiary burden that bias crimes add to a case (McPhail &
DiNitto, 2005). Since motivation is difficult to determine, and prosecuting attorneys tend to
pursue cases they know they can win, bias crime charges tend to remain unpopular. After all,
intent is generally the only requirement for prosecution for all other crimes – both violent and
property-related. McPhail and DiNitto (2005) also note prosecutors face pressure from the
public to prosecute crimes in accordance to community values. As such, the authors note that

57

prosecutors tend to shy away from prosecuting anti-LGB bias crime in cities where there is a
high proportion of its population adhering to religiosity and social conservatism.
Spatial Characteristics of Bias Crime
In a study of bias crime at the county level, Wilson and Ruback (2003) found that bias
crime was most likely to occur in rural areas – as opposed to urban areas. Given that social
disorganization theory is widely respected among criminologists, these findings come as a
surprise. As such, urban areas often are characterized by ethnic and racial heterogeneity, dense
populations, and at times – poverty (Wilson & Ruback, 2003). The authors suggest that bias
crime may be more prevalent in rural areas as opposed to urban areas because urban areas have a
higher level of interaction between people from different social groups than do rural areas.
Alternatively, in an area with great homogeneity, the frequency of interaction with diverse
populations will be relatively low. Perhaps, it is the case that rural areas provide an atmosphere
where individuals have low levels of interaction with difference and perhaps may exhibit greater
prejudice. Thus, rural residents are more accepting of bias-motivated behaviors (Wilson &
Ruback, 2003).
It seems that areas with greater racial interactions – such as is the case in urban areas –
there may have higher bias crime rates by virtue of their demographic make-up (Wilson &
Ruback, 2003). Perhaps the more frequently one interacts with individuals with whom she or he
differs the greater opportunity there is for bias-motivated crime to occur. However, Wilson and
Ruback (2003) suggest this occurs due to a “threshold effect.” In essence, citizens in urban areas
may be less likely to report bias crime – or perceive one as such – and dismiss the incident based
on a sentiment that nothing can or will be done to resolve the problem.

58

As mentioned, bias crime research largely has been lacking. It is important, however, to
examine bias crime at the city level. After all, bias crime does more than affect subordinate
groups; it affects entire communities. Importance lies in the pursuit of the understanding of how
communities are affected by bias crime at the community level.
In perhaps one of the most salient articles regarding location and bias crime, McVeigh,
Bjarnason, and Welch (2003) examined bias crime reporting at the county level. In this study,
the authors found that there is relevance in examining the local contexts present in a county when
observing the incidence of bias crime. For instance, the existence of various instruments of
social movements at the local level – such as political and civil rights organizations - can be
associated with higher rates of bias crime reporting. Certainly, relevance lies in understanding
that even when bias crime social movements occur – official reporting or policy creation – it
does not necessarily mean that policies will be enforced evenly across locations (McVeigh,
Bjarnason, & Welch, 2003).
McVeigh, Bjarnason, and Welch (2003) mention that there is importance in using official
data and provide argument against Kitsuse and Cicoural’s (1963) contrasting claim. In effect,
official data are known to lack reliability and should therefore be avoided (Kitsuse & Cicourel,
1963). Alternatively, McVeigh, Bjarnason, and Welch (2003) suggest it is due to this lack of
reliability that there is value in analyzing official data. After all, it is by virtue of the
unpredictability of results obtained from official data – differing in space and time – that allow
social scientists to understand the ever-changing nature of crime and, in this case, bias crime.
Moreover, by examining bias crime occurrence at the county level, McVeigh, Bjarnason, and
Welch (2003) add rural areas to the discussion. When concentrating on standard Metropolitan
Statistical Areas (MSAs), many of the bias crimes are overlooked (McVeigh, Bjarnason, &

59

Welch, 2003). Examining bias crime reporting at the county level allows for the inclusion of
urban, suburban, and rural areas.
Strengths and Weaknesses of Prior Studies
Bias crime research has grown tremendously in volume in the past few decades. There is
no doubt this is the result of increased attention paid to bias crime events. As late as the 1980s,
bias crime research was still in its infancy (Gerstenfeld, 2011). Although it has expanded as a
field of study, there is still much to be uncovered with regard to bias crime offending, victim
experience, legislation, and public policy approaches. Certainly, this dissertation seeks to
contribute toward that end – providing additional information about bias crime and the
community. It is upon the numerous strengths of bias crime research that this dissertation builds.
Much of the early research focused on descriptive analyses of bias crime events.
Essentially, this endeavor led to only a marginal understanding of the basic structural elements of
a bias crime. Bias crime offenders were examined as were the victims. However, little was
developed with regard to social science theory and the way such theories could explain the
phenomenon of bias crime occurrence.
Moreover, from a victimology standpoint, the research has expanded to survey issues
related to traditionally marginalized groups. Understanding that bias crime, unlike most
interpersonal crime, is related to intergroup struggles for power, researchers moved into
exploring systems of oppression. It is evident that the preservation of power is particularly
related to the incidence of hate criminality. This dissertation takes that and applies it to social
disorganization theory.
One of the strengths of the bias crime literature is its coverage of the spatial
characteristics of bias crime events. Specifically, understanding the nature of and location of

60

bias crime events helps to elucidate the factors associated with the risk of bias crime occurrence.
Although scholars have posited various theoretical contributions to this end, there is somewhat of
a consensus in the literature that those areas that are more rural in nature are more likely to
experience bias crime offending. Perhaps this is curious given the previous discussion of the
defended neighborhoods perspective – one which assumes a certain level of diversity in a larger
community, such as a county. Conversely, rural areas – often encapsulated by the concept of a
county - tend to be rather homogenous compared to larger cities. This fact can inform our
understanding of the risk of bias criminality and its occurrence in United States counties.
Along with the numerous strengths, there are many weaknesses of bias crime research.
Primarily, is it evident that bias crime research is greatly underdeveloped as there are relatively
few academic studies that have been conducted informing social science about the nature of bias
crime incidence. Not only is there a lack of empirical work to try to explain the nature of bias
criminality, but also there are very few studies that explain the effect of bias crimes on American
communities – whether at the state, city, or county level. Moreover, there is a lack of work
comparing and contrasting the various types of bias crimes that affect the American public.
Essentially, there is relatively little known about the differences between anti-religion, anti-race,
and anti-sexual orientation bias crime events. This dissertation seeks to explore each of these
significant gaps in the literature.
Summary
Bias crime is a phenomenon about which the social science literature is relatively silent.
Over approximately the past 30 years, scholars have been able to provide many descriptors of
bias crime victimization and offending. Through this endeavor, it has become clear that greater
interdisciplinary exploratory and explanatory research is necessary. Based on the current and

61

presented scholarly information, this dissertation seeks to add to the discussion surrounding bias
crime. Through an exploratory approach, this research will add to the understanding of bias
crime victimization at the county level. With a gained comprehension of this knowledge,
perhaps various municipalities across the United States will be able to implement public policies
to mitigate the occurrence – and the effects of – bias crime offending.

62

CHAPTER 3: SOCIAL DISORGANIZATION THEORY
This chapter examines the macro-level social disorganization theory and its application to
bias criminality from the perspective of numerous social sciences. The majority of
criminologists have attempted to explain crime through the lens of micro-level theories, thus
focusing on individual characteristics or situations. Conversely, communities and crime scholars
lean toward a macro-level approach, examining the various ways that communities both
influence crime and are influenced by it. Specifically, social disorganization theory has been
used to explain crime in American communities. Social disorganization theory suggests that a
community which is unable to identify its common goals and effect appropriate social controls is
more likely to suffer from crime (Sampson & Groves, 1989). Social disorganization theory
posits that communities facing high residential mobility, a lack of social cohesion (collective
efficacy), urbanism, low socioeconomic status of residents, unsupervised peer groups, and high
rates of family disruption are more likely to experience criminal events. This theory has been
tested quite thoroughly and the results of those studies are presented in this chapter.
Studies of bias crime have largely been descriptive in nature. Moreover, such studies
focus on the perpetrators and, at times, the offenders. This dissertation seeks to examine
American communities – in this case, counties. Given the unit of analysis for this dissertation is
the county, a theoretical framework from the communities and crime perspective is most
important. The purpose of this dissertation is to add to the knowledge of bias criminality and its
effects on communities. Therefore, social disorganization theory – with its emphasis on
communities and crime – presents a solid theoretical framework with which to examine bias
criminality.

63

Social disorganization theory also informs the research design of this dissertation.
Common indicators of social disorganization theory are considered and included as the basis for
creation of independent variables for this dissertation. Additionally, in order to provide an
appropriate explanation of bias crime occurrence at the county level, the historical development
of social disorganization theory is presented and discussed. An historical presentation of the
social disorganization literature provides insight into the development of the theory, from its
genesis to the most recent contributions, thereby providing theoretical and methodological
guidance for this dissertation.
Social Disorganization Theory
Social disorganization is one of the most popular theories in the study of communities
and crime. The roots of social disorganization theory extend back to Robert Park (1952) and his
theory of human ecology. The main concept was that each city had a certain organic unity and
natural areas where various groups of people lived. In addition, Park theorized that these
natural areas could change based on the ecological process of “invasion, dominance, and
succession.” Robert Park and Ernest Burgess (1925) contributed the notion of concentric zones
in cities. Burgess suggested that at the city center was Zone I - or the business district. Zone II
was typically the oldest section of the city and included frequent “invasion, dominance, and
succession.” Farther out in the city is Zone III, which was comprised of modest homes and
apartments, occupied by residents who had escaped Zone II. Zone IV was occupied by residents
who lived in single-family homes and expensive apartments. Zone V was the commuter zone.
Park and Burgess (1925) suggested that the changes that occurred throughout the zones
exemplified the “invasion, dominance, and succession” perspective in which residents moved in
a radial fashion outward into a different zone.

64

Informed by the research of Park and Burgess (1925), Clifford R. Shaw and Henry D.
McKay (1942) examined juvenile delinquency and its uneven dispersion in space and time
across Chicago. The Chicago School researchers found that juvenile delinquency tended to be
concentrated in areas of the city with low economic status. These particular neighborhoods
where delinquency and crime were concentrated were also near areas of heavy industry or
commerce, had declining populations, and had numerous abandoned buildings. In addition,
Shaw and McKay found that these disorganized neighborhoods remained in such a manner
regardless of the race, ethnicity, or national origin of its residents. Such findings were still new
to criminology and positioned Shaw and McKay in the direction of examining communities for
contributing factors to crime, rather than assuming individual culpability.
Ultimately, Shaw and McKay (1942) suggested that neighborhoods that were in transition
tended to be the most socially disorganized. These neighborhoods often had particularly high
rates of residential mobility compared to other areas of the city, had a high level of racial
heterogeneity, and had low socioeconomic statuses. Given these unstable structural elements,
socially disorganized neighborhoods were characterized as having low social control. Therefore,
the institutions that might provide social control in more stable and organized neighborhoods
were comparatively weak in socially disorganized neighborhoods. Essentially, the researchers
implied that in the inner city families could be disrupted, institutions such as schools and
churches would be poorly attended and in a state of disorder, and political groups would be
ineffective (1942). Shaw and McKay (1942) then theorized that this community breakdown, as
opposed to a delinquent’s individual shortcomings, led to higher rates of juvenile delinquency
and crime.

65

A criticism of social disorganization theory was that it emphasizes middle class values
and is biased against differentially organized neighborhoods (Pratt & Cullen, 2005). There are
numerous cultural and ethnic differences that are evident between any given neighborhoods in a
city. To place judgment on “disorganized” neighborhoods does not necessarily make them
disorganized; rather, it allows them to be seen as differentially organized. Consequently,
numerous scholars viewed social disorganization theory as irrelevant and excluded it from
modern criminology. Although researchers regarded social disorganization theory as relevant
throughout the 1950s and 1960s, by the 1970s Shaw and McKay’s theory had been marginalized
and nearly abandoned by criminologists (Pratt & Cullen, 2005).
A Resurgence of Social Disorganization Theory
During the 1980s, criminologists found a renewed interest in Shaw and McKay’s social
disorganization theory. As crime and disorder seemed to creep back into the public’s attention
during the 1980s, there was a need to further examine macro-level issues related to crime and
delinquency. Some scholars suggest the second wave of interest in social disorganization theory
came as a result of the conservatism and social inequality present in Reagan Administration
policies in the early 1980s (Cullen & Agnew, 2003). Whether the cause of the renewed interest
was due to unequal policy or increasing crime rates, social disorganization theory was reignited
to help explain the relationship between communities and the crime they experience on a day to
day basis. In order to present the progression of social disorganization theory since the 1980s,
the following sections provide discussion of significant contributions to the perspective.
Blau and Blau (1982)
In their 1982 study, Blau and Blau examined urban violence in 125 of the largest cities in
the United States. The authors found that socioeconomic inequality between African American

66

and Caucasian residents, in addition to socioeconomic conditions generally, increased violence.
However, Blau and Blau (1982) reported that once economic inequalities were controlled, the
rates of violent crimes were no longer influenced by poverty. Moreover, Southern location and
the proportion of African Americans in a specific location had a minimal influence on violence.
This suggests that if there is indeed a culture of violence, that particular culture is rooted in
economic inequality and not in race.
From the results of the Blau and Blau (1982) study, there is further evidence that macrolevel factors could influence crime rates. In this case, economic inequality, a major tenet of
Shaw and McKay’s social disorganization theory, positively affects the violent crime rate.
Although Blau and Blau (1982) examine several theoretical perspectives, including labeling
theory and Marxist theory, their study exemplifies the relevance of community-level factors and
their effects on crime. Although the authors exposed a similarity with Shaw and McKay’s theory
in that poverty influences crime rates, Blau and Blau (1982) suggest that race and Southern
location are not as influential in predicting criminality. Therefore, it is worth noting that even
with slightly different variables, the Blau and Blau (1982) provide some indication that economic
inequality and poverty seem to provide a better explanation of criminality.
Sampson (1986)
Sampson is one of the leading scholars that have advanced social disorganization theory.
In 1986, Sampson proposed adding family disruption to the structural factors of poverty,
urbanism, residential mobility, and racial/ethnic heterogeneity. The relevance of this addition to
social disorganization theory is that if a neighborhood has a number of households with
inadequate parents or single-parent households, and it has the original structural factors of the
theory present, social disorganization is likely to result. It is important to note, however, that

67

Sampson (1986) explained that a neighborhood would need to have numerous households with
family disruptions - not just an isolated case. The idea is that disrupted families would lead to a
lack of social control. This, in turn, would contribute toward the inability of a neighborhood to
control delinquent or criminal behavior.
Sampson’s contribution was particularly useful due to his extension of social
disorganization theory. Since the renewed interest began in the early 1980s, no scholars have
posited extensions to Shaw and McKay’s social disorganization theory. Although Sampson
retained much of Shaw and McKay’s original structural components of the theory, he included
family disruption as well. Blau and Blau (1982) may have reintroduced discussion about social
disorganization theory, but unlike Sampson, they did not add any structural components to the
theory. Therefore, it is clear that Sampson began the trend of important extensions to social
disorganization theory that would come for the next ten years.
Skogan (1986)
Yet another study that has informed social disorganization theory is Skogan’s 1986
examination of the effect of fear of crime on neighborhood change. Skogan (1986) suggested
that fear can cause dramatic impacts on communities such as the following: withdrawal from
community life; a weakening of the informal social control process that is known to inhibit
crime, delinquency, and disorder; a weakening of the organization in the community and its
ability to mobilize to reach its goals; poor business conditions; the presence of crime and
delinquency from outside the community and the growth of the same within the community; and
changes in the population. According to Skogan (1986), even though fear of crime does not
necessarily reflect the actual crime rate, it can affect the overall organization of the community.

68

Those communities that are gripped with the fear of crime can be excluded from social and
economic benefits of the mainstream society.
Skogan’s contribution to the development of social disorganization primarily concerns
the effect of fear of crime in communities. This adds a new variable that is based on the overall
feelings of the residents of the community and mediates the theorized outcome of crime. Being
aware that only the fear of crime is enough to cause social disorganization is helpful in two ways.
First, awareness of this theory could help some communities share facts about crime rates. After
all, in some communities, there is always the real possibility that residents are overly concerned
about the crime rate. Attenuating their fears might help to rebuild social cohesion and prevent
crime and disorder caused by social disorganization. Second, even if a community is particularly
troubled with a high crime rate, combating residents’ fears could have an impact on their
willingness to get involved in their communities, and ultimately lower social disorganization and
crime.
Bursik (1988)
Perhaps some of the largest contributions to the development of social disorganization
theory were Bursik’s criticisms that he posited in 1988. The first criticism is that many dismiss
social disorganization because it cannot address individual-level behaviors. Bursik (1988)
argued that since the objective of the theory is to examine community-level factors that lead to
disorganization, it is inappropriate to expect social disorganization theory to contribute to the
social psychological literature. Bursik’s (1988) second criticism lies in the fact that social
disorganization theory assumes stable ecological structures. He suggested more emphasis should
be placed on the dynamics of urban change in order to better assess social disorganization in any
community. A third criticism is the ambiguity of the measurement of social disorganization.

69

Bursik (1988) suggests that Shaw and McKay (1942) did not properly differentiate between
social disorganization and the phenomena it supposedly causes. For instance, is delinquency
considered social disorganization or rather the outcome of social disorganization? Fourth, there
are discrepancies in how crime and delinquency is measured. This may the result of differential
handling by the criminal justice system on many levels. However, there is likely some variation
in the records, whether they are self-report or official records. Last, there are differences in the
definition of social disorganization that vary by location. If the definition is defined as the
inability of a community to attain goals that are agreed upon by the community, then there is an
assumption that the community agrees upon one definition of community goals. Bursik (1988)
suggested that since communities differ, then so do definitions of social disorganization.
Also important to the development of the social disorganization perspective are Bursik’s
extensions of the theory. First, Bursik (1988) suggested the use of contextual analyses with the
use of official data and self-report data. Although difficult to attain, the use of these two data
types would significantly improve criminological knowledge of the contextual effects of social
disorganization. Second, victimization data could be used to better understand issues
surrounding social disorganization. Additionally, Bursik (1988) suggested that other theoretical
perspectives, such as routine activities theory, can be used concurrently to social disorganization
to aid in the understanding of neighborhood factors.
Bursik (1988) contributed to the development of social disorganization in numerous
ways. Not only did he criticize the theory and analyze prior studies, but he provided suggestions
for future research. This approach differed from much of the other research available on the
topic of social disorganization in that it did not test the theory or any part of the theory.

70

However, the perspectives and theoretical development that Bursik proposed are useful in tracing
the development of social disorganization over the years.
Sampson and Groves (1989)
Sampson and Groves’s 1989 test of social disorganization theory was a first of its kind.
To some, it is considered a classic in criminology (Lowencamp, Cullen, & Pratt, 2003). Up until
that point, Shaw and McKay’s social disorganization theory had never been tested by social
scientists. The authors hypothesized that ethnic heterogeneity, low economic status, residential
mobility, and family disruption would lead to social disorganization. Sampson and Groves
(1989) also measured social disorganization with three factors: sparse local friendship networks,
unsupervised teenage peer groups, and low organizational participation. As a consequence of
social disorganization, crime rates were hypothesized to increase. The theory was first tested by
administering a survey to 10,905 residents of Great Britain. The results from the survey
supported social disorganization theory. In the same study, the researchers administered a
replication and surveyed another 11,030 residents of Great Britain with the same results.
Although Sampson and Groves suggested that there are a few limitations, most importantly that
social disorganization was only measured with three variables, the overall results were
theoretically consistent with Shaw and McKay’s social disorganization theory.
To give more credit to Sampson and Groves’s test of social disorganization theory, it is
important to mention that Lowencamp, Cullen, and Pratt conducted a replication in 2003. These
authors used data from the 1982 British Crime Survey and used similar models and measures to
those that Sampson and Groves used. The authors’ results suggested Sampson and Groves’s
results were not artifactual and are therefore consistent with social disorganization theory.

71

Sampson and Groves’s (1989) testing of social disorganization theory was particularly
influential in terms of criminological theory development. Not only was this the first test of the
theory, but the results of both studies supported the theory. Another notable contribution of
Sampson and Groves’s study was the addition of “family disruption” to the theory. Although
Sampson included this factor from his 1986 study, for the 1989 study, the authors extended the
theory and tested it. Not only did Sampson and Groves discover there may be more merit to
social disorganization theory than has previously been thought, but the addition of “family
disruption” allowed a more thorough examination of the theory and it exemplified the
importance of macro-level theories. To date, no other researchers had made such an impact on
social disorganization theory since Shaw and McKay. Sampson and Groves’s theory testing
paved the way for replications and further development of social disorganization theory.
Sampson and Wilson (1990)
After numerous of extensions of social disorganization theory, Sampson and Wilson
(1990) acknowledged that the meaning of social disorganization had become the inability of a
community structure to realize the common values of its residents and maintain effective social
controls. Therefore, the authors suggested that there are various macro-social forces (e.g.
housing discrimination and segregation) that interact with community-level factors to hinder the
formation of social organization. Moreover, Sampson and Wilson (1990) suggested cultural
disorganization provided another way to examine crime and race. The authors discussed
cognitive landscapes, or ecologically structured norms that help people know how to navigate the
expectations of conduct. Also pertinent is the idea of social isolation. Essentially, this is the lack
of contact with institutions or individuals from the mainstream society. The authors suggest that

72

these factors are not necessarily present in the neighborhood context. Rather, they result from
structural inequalities that many poor, urban communities routinely experience.
Sampson and Wilson (1990) contributed significantly to the discussion on social
disorganization and communities and crime. Although their discussion of social disorganization
is similar in many ways to previous studies, they proposed the idea of cultural disorganization.
This concept provided a fresh view on issues such as race and crime and poverty and crime. By
acknowledging that there are external factors that influence cognitive landscapes and social
isolation of community residents, Sampson and Wilson set the stage for the later examination of
cultural issues such as Elijah Anderson’s (1999) study, Code of the Street and Warner’s (2003)
study of attenuated culture and social disorganization theory.
Bursik and Grasmick (1993)
Neighborhoods and Crime: The Dimensions of Effective Community Control was another
influential contribution to social disorganization theory. Although the authors cover a lot of
theoretical ground in their book, the systemic theory of neighborhood crime control emerged as
the take away contribution. Bursik and Grasmick (1993) suggested that in the systemic model,
there is a link between local crime control and the numerous public control systems present in
any community. For instance, every neighborhood or community has a variety of economic and
political institutions that control it. Using this systemic model, the authors examined various
aspects of urban community life, such as fear of crime and gang involvement.
In past studies, scholars have added to social disorganization theory, tested it, and
examination various factors that mediate a community’s ability to prevent social disorganization.
Bursik and Grasmick (1993) take social disorganization theory further by introducing the
systemic model, therefore expanding the understanding of the theory. Although the authors

73

broaden the understanding of social disorganization theory, they do not take into account the
influence that culture has on social institutions and crime rates. They claim that crime can be
explained without culture taken into account, therefore proposing that the structural perspective
of social disorganization theory is sufficient. Although Bursik and Grasmick (1993) largely
ignored culture, they made this apparent, and therefore prompted further studies based on
cultural aspects of urban community life.
Sampson, Raudenbush, and Earls (1997)
One of the largest contributions to social disorganization theory came in the way of a
concept that is nearly oppositional to disorganization. Sampson, Raudenbush, and Earls (1997)
applied the term “collective efficacy” and used its presence to explain the lack of crime and
disorder in public spaces. Collective efficacy is similar to the notion of social capital. Coleman
(1988) explains social capital as a network between various individuals that helps people achieve
common goals. The authors argued that when there is much social capital in a community, there
is a lower amount of observable crime. Utilizing the theory of social capital, collective efficacy
theory suggests that when there is “cohesion and mutual trust,” residents are likely to take action
to achieve community goals of public order (Sampson, Raudenbush, & Earls, 1997). This
amounts to residents of a community getting together to organize a neighborhood watch group or
making complaints about disorder to the police.
In a test of their theory of collective efficacy, Sampson, Raudenbush, and Earls (1997)
videotaped both sides of the street in a large portion of Chicago, Illinois in order to observe
instances of physical disorder (such as abandoned properties or automobiles). Additionally, the
authors interviewed residents of the various communities about the apparent social disorder. The
results of the study indicated that physical and social disorders were closely related to differential

74

land use and poverty. Therefore, the authors were able to conclude that lessened collective
efficacy and structural factors were instrumental in the explanation of crime.
Sampson, Raudenbush, and Earls’s (1997) addition to social disorganization theory is
perhaps among the most salient. Since the early 1980s, when social disorganization theory
began to emerge back into favor, numerous scholars have attempted to examine factors
associated with neighborhood disorder. However, these authors concentrated on the presence of
organization (informal social control) as opposed to the lack of social cohesion, or social
disorganization. The authors’ efforts to examine this new perspective allow the examination of
factors that can influence social organization. For instance, with collective efficacy theory,
researchers can examine the reasons why individuals and communities do not have the ability to
achieve common goals. They also can study why certain communities do have the ability to
achieve varying levels of collective efficacy. However, the authors stated that because collective
efficacy matters, that does not mean that various forms structural inequality should be ignored.
Sampson, Raudenbush, and Earls contributed one of the most significant pieces of the social
disorganization puzzle and provided insight on how collective efficacy fit together with
residential mobility, socioeconomic status, urbanism, ethnic heterogeneity, family disruption,
and unsupervised peer groups (Pratt & Cullen, 2005).
Morenoff, Sampson, and Raudenbush (2001)
More recently, efforts were made by Morenoff, Sampson, and Raudenbush (2001) to
examine the linkage of concentrated disadvantage and collective efficacy and its effect on
homicide. The results indicated that both factors independently predict increased homicide. The
authors also found that spatial proximity to homicide is related to increased homicide rates.

75

Essentially, the authors used homicide to represent urban violence. The overall results suggested
that spatial dynamics and neighborhood inequalities are important in explaining urban violence.
The findings from the Morenoff, Sampson, and Raudenbush (2001) study represent a
different attempt to examine social disorganization theory. Rather than studying crime or
delinquency in general, the authors chose to use homicide as an indicator of urban violence. This
is an important contribution in that is exemplifies how social disorganization theory can be used
to explain a specific type of crime as opposed to general references of crime, delinquency, or
disorder. Certainly, this perspective allows a testing of social disorganization theory as it affects
bias crime occurrence. Similar to other studies, however, the authors continued to use the
construct of collective efficacy in order to examine social cohesion.
Warner (2003)
Most of the studies in the social disorganization literature discuss structural
disorganization. Although there seems to be merit to this perspective, Warner (2003) examined
the role of attenuated culture and its effect on communities. Using survey data, the author
examined residents from 66 neighborhoods from a Southern community. Prior research has
suggested that when cultural elements are reduced, it cannot provide for effective social control
in the community (Kornhauser, 1978). Warner’s results support that conclusion as she found
that the level of social ties and concentrated disadvantage influence cultural strength. This, in
turn, has a significant impact on social control.
Warner’s recent study is an important contribution because it suggests the inclusion of
culture into the social disorganization perspective. Sampson and Wilson (1990) suggested that
cultural disorganization become included in social disorganization theory along with the
structural perspective. Although there has previously been discussion about this in past studies,

76

some scholars such as Bursik and Grasmick (1993) have dismissed culture in favor of a
structural approach. Warner brought culture back into the discussion, and for some, has
provided hope that further integration between structural and cultural factors will occur with
social disorganization theory.
Sampson (2012)
More recently, Robert Sampson (2012) has reexamined communities and crime in
Chicago. Perhaps the most salient contribution of this work is Sampson’s offering of a general
theoretical approach to understanding neighborhoods and crime in American cities. Given this is
a significant contribution to the literature on communities and crime, there a several aspects of
Sampson’s work worth noting. For instance, Sampson (2012) suggests that with regard to racial
segregation and concentrated disadvantage, neighborhoods replicate themselves. Given racial
segregation and concentrated disadvantage are not new phenomena in American communities;
they tend to continue from decade to decade with little change. Sampson (2012) refers to this as
“stability of change.” In addition, Sampson posits that racial segregation and concentrated
disadvantage affect neighborhoods in that they influence collective efficacy, cynicism about the
legal system, residents’ reception of community disorder, and neighborhood stigmatization.
Perhaps another important contribution of Sampson’s (2012) work includes his assertion that
communities are influenced by the context of larger community dynamics. It is therefore an
ineffective approach to treat each community as if it were independent from its surroundings.
What is most important to recognize from Sampson’s work on Chicago neighborhoods is
that much of what affects a community’s level of disorganization is continuous – even with the
relative increase in placelessness. This is highly important to the current literature on

77

communities and crime as it contributes the notion that macro level effects are often overlooked
and should be included in future studies.
Testing and Applying Social Disorganization Theory
According to Sampson (2003), learning from macro-level social disorganization theory, it
is apparent that there exists more of a need to change communities rather than to change people –
and by virtue of that, their behavior. Therefore, implementing policy and directing community
action toward affecting a community as a whole is particularly appropriate. Although social
disorganization theory has not been used to examine bias crime offending at the county level, it
has been approached in a variety of other settings. For instance, social disorganization theory
can be and has been applied to communities in a variety of ways including Shaw and McKay’s
Chicago Area Projects (CAP) in Chicago, Project on Human Development in Chicago
Neighborhoods (HDCN), urban renewal, community policing, disorganization and far-right
extremist violence, and Weed and Seed. At times, these tests of the theory overlap; however,
they are common implementations of social disorganization theory, regardless of their
effectiveness. These applications of social disorganization theory are helpful in that they can
inform public policies created to combat bias crime occurrence in American communities.
Chicago Area Project (CAP)
Perhaps most salient is a discussion of communities and crime policy with a summary of
Shaw and McKay’s application of their own social disorganization theory. CAP was a program
put in place in Chicago, Illinois in 1932 to lower delinquency across the city. Several projects
lasted through the 1960s and had a variety of goals. Shaw and McKay selected high crime
neighborhoods in which to implement the CAP program. Of particular importance to the
researchers and the Chicago Area Project was the mobilization of community groups to aid in the

78

development of social organization and social control. Existing groups, such as labor unions and
churches were targeted first, but the aim was to include as many law-abiding citizens as possible
in hopes that their involvement would help create community committees (Vito, Maahs, &
Holmes, 2006). In addition, Shaw and McKay set out to develop informal relationships with the
delinquent youth, especially gangs, in crime ridden neighborhoods (Vito, Maahs, & Holmes,
2006). This allowed troubled youth to have access to recreational programming all the while
providing exposure to “conventional” residents. Other work was initiated as well, such as efforts
to clean up neighborhoods and rid them of garbage, graffiti, and other forms of physical decay
(Vito, Maahs, & Holmes, 2006). All of these efforts exemplify social disorganization theory’s
proposition that disorganization is a problem of the community rather than the individual
residents.
Unfortunately, there is no rigorous evaluation of the CAP program in Chicago. Given the
length of time that CAP was in effect, there were various reports written throughout years. At
best, CAP provided mixed results (Vito, Maahs, & Holmes, 2006). In some communities with
strong community committees, there were reductions in juvenile delinquency. However, many
neighborhoods reported no results or even an increased delinquency rate. Although the Chicago
Area Project was unable to provide clear results related to social disorganization theory, the
efforts likely set the stage for later applications of the theory.
Project on Human Development in Chicago Neighborhoods
In 1995, Sampson (2012) notes that a research program named Project on Human
Development in Chicago Neighborhoods (PHDCN) was begun in order to study criminality in
Chicago while accounting for the context of community. Essentially, the methodological basis
for this study was a rather extensive survey of Chicago residents – amounting to approximately

79

8,782. In designing the survey, Sampson (2012) notes that two specific components were
constructed to measure “informal social control” and “social cohesion” – certainly, two
important indicators of social disorganization theory. The result of this was the creation of a
collective efficacy scale used to measure the same in Chicago’s neighborhoods. Additionally,
this study aimed to inform the growing communities and crime literature.
Although PHDCN was rather extensive, several findings emerged from the endeavor.
First, researchers found that collective efficacy in Chicago’s neighborhoods varied widely,
certainly by time and space. However, that collective efficacy that did exist at greater levels was
associated with lower crime rates. Second, Morenoff et al. (2001) report that the density of
neighborhood-based organizations and the higher participation by residents in volunteer
programs predicts high levels of collective efficacy; in turn, creating controls for poverty and
crime.
Urban Renewal
On the surface, it seems relatively easy to build a case for urban renewal – or
gentrification. In continuing with social disorganization theory, urban renewal would add
higher-income households to poverty-stricken neighborhoods. Some believe urban renewal
functions as a balancing act and equalizes the neighborhood with social, political, and economic
benefits (Vito, Maahs, & Holmes, 2006). However, certainly the gentrification of a
neighborhood exacerbates the tensions - or strain - between various social classes. Gentrification
has been associated with creating evidence of social dislocation. However, some scholars hope
that the renewal of socially disorganized neighborhoods would improve social cohesion and
lower crimes rates. In reality, and as suggested, this is a failed policy as urban renewal often
forces poorer residents to leave the area (Vito, Maahs, & Holmes, 2006). As wealthier residents

80

move into the area, property taxes increase and cause the neighborhood to be out of reach for its
original residents, thereby worsening social cohesion. In many cases, this is an unintended and
undesired effect of urban renewal. However, historically urban renewal has been a tool for
furthering racial inequality.
A particularly troubling example of gentrification occurred in Detroit, Michigan during
the 1950s. Detroit has a long and bitter history with racial inequality, though the urban renewal
that occurred in the neighborhoods of the Lower East Side and in Paradise Valley was extremely
difficult for African American residents of the city (Sugrue, 2005). This urban renewal, as it was
called, was really a relocation of African Americans and an attempt by the city to “clean up.” In
the midst of a housing crisis, the city removed “slum” areas of the city in which many African
American residents lived. Not only did this leave these disadvantaged residents without a home,
but it overcrowded the unaffected African American neighborhoods of Detroit. In addition, the
city of Detroit also was undergoing a massive highway project which further displaced numerous
African American families. Detroit’s urban renewal plan was nothing more than another
example of structural inequality. There may be other examples of urban renewal programs that
are well-intentioned. However, the problem with the application of social disorganization theory
to urban renewal programs is that it may be implemented without regard to those who do not
have the resources to continue living in the community.
Community Policing
Community policing is not itself a policy; rather it is a philosophy of policing. In
addition, it can be considered a direct application of social disorganization theory. Skogan
(2006) suggests community policing is a process of changing police culture and decision making
processes. Essentially, it leaves many of the decisions of how to police neighborhoods up to the

81

residents and the officers who patrol those neighborhoods (Skogan, 2006). There are three
processes that are often considered to be part of community policing. Skogan (2006)suggests
that decentralization, problem solving, and citizen involvement are instrumental to the practice of
community policing. It is capable of tackling a wide range of community problems, not just
crime. Therefore, community policing takes on many community problems and works with
community member to solve them (Vito, Maahs, & Holmes, 2006). Therefore, it can be viewed
more as a type of community action and it can involve various criminological theories such as
social disorganization and broken windows.
The various practices of community policing can be seen as community action related to
social disorganization theory. The policies and community actions that are often achieved under
a community policing framework are thought to build collective efficacy and restore social
control to neighborhoods. It should be noted, however, that Sampson, Raudenbush, and Earls
(1997) discovered that community policing can build collective efficacy for neighborhoods that
already have it. These neighborhoods are generally white and have numerous resources
available to them. Conversely, the authors note that the philosophy of community policing has
yet to provide evidence that it builds collective efficacy in neighborhoods that did not previously
engender collective efficacy (Skogan, 2006).
It would seem inappropriate to discuss community policing and social disorganization
theory without mention of broken windows theory. Although broken windows theory is an
entirely different theory than social disorganization theory, community policing is so expansive
that broken windows theory is often used to inform community policing policies and community
actions. Broken windows theory refers to the theory that Wilson and Kelling developed in 1982
to explain that social disorder and physical decay undermined a community’s ability to defend

82

itself against crime. Abandoned cars, graffiti, panhandling, street prostitution are all forms of
disorder which Wilson and Kelling suggested a community has given up (Wilson & Kelling,
1982). Therefore, crime and disorder spiral out of control and unless there is significant
intervention, crime pushes the community downhill. Community policing is often employed to
develop policies and community actions that can reduce social disorder and physical decay. In
many respects, the move toward community policing has allowed broken windows theory an
opportunity to become a popular theory for the explanation of neighborhood crime and disorder.
Using this perspective to inform social disorganization theory is rather salient in exposing the
ways a community can continue to build collective efficacy.
Keeping in mind that community policing can appease both the social disorganization
theorist and the broken windows proponent, there are numerous functions that a community
policing department can have. For instance, many police departments take part in engaging with
the community in various ways such as holding citizen police training academies, conducting
surveys to gather information about community satisfaction, opening storefront police offices,
walking door-to-door to identify problems, mailing out newsletters, conducting education classes
on issues such as drugs, and working with municipalities to enforce health and safety rules
(Skogan, 2006).
Perhaps the best example of community police in practice is the Chicago Alternative
Policing Strategy (CAPS). Although CAPS was based on broken windows theory, from its
community policing framework, numerous community building policies and community actions
resulted. For instance, according to Skogan (Skogan, 2006), there were four central elements
that were integral to community policing in Chicago: turf orientation, community involvement,
problem solving, and interagency partnerships. First, turf orientation can be best described as the

83

presence of community service officers or problems solvers whose job is to respond to the
“problem of the week.” Second, community involvement was a large component of CAPS.
There were various forms of community involvement such as identifying local problems, helping
police set priorities, taking individual actions and actions in conjunction with the police, and the
usage of beat meetings to hold police officers accountable for their efforts. Third, problem
solving involved numerous topics, not all of which were criminal matters. The stipulation was
that something was defined as a problem only if it was an ongoing situation or a group of related
events. Last, the police worked with other agencies to create partnerships. This allows the
police to have abandoned automobiles towed quickly and to have vacant lots mowed. In other
cases, police worked with agencies who would immediately paint over graffiti.
In conjunction with the community residents in Chicago, Illinois, the police were able to
make some progress in the city (Skogan, 2006). Crime was down since the program was
implemented, but nationwide trends indicated similar results. Skogan (2006) found that African
Americans and Caucasians saw improvement due to CAPS, but the Latino community did not.
All four of the CAPS community policing strategies discussed address pertinent theoretical
issues relating to urban decay, crime, and disorganization. Given community policing is a broad
concept in policing, it allows for various community activities and policies to be developed.
Regardless of how and where community policing is practiced, the core idea is to leave many of
the decisions of how to police neighborhoods up to the residents and the officers who patrol
those neighborhoods.
Weed and Seed
Terence Dunworth and Gregory Mills (1999) describe Weed and Seed as an incubator for
social change. Social change is brought about by strategies that mobilize and coordinate

84

resources in targeted communities. Essentially, Weed and Seed is a policy directed at stabilizing
conditions in high crime neighborhoods and promoting community restoration (Dunworth &
Mills, 1999). As mentioned, policies do overlap in scope, and Weed and Seed is no exception.
Although Weed and Seed brings many new ideas to the table to rebuild communities, there is
some familiarity with the policy.
Weed and Seed policies generally have four essential components (Dunworth & Mills,
1999). First, communities engage in analysis and planning of the problems that need to be
addressed in their communities. In addition, strategies are implemented to address these
problems. Second, neighborhood weeding takes place. Weeding involves concentrated efforts
of law enforcement in order to identify, apprehend, and prosecute various criminals. Third,
community policing is implemented in order to bridge the gap between weeding and seeding.
Fourth, the community begins to seed. This involves numerous human services including youth
activities, adult literacy classes, and parental counseling.
In a nationwide evaluation of Weed and Seed, Dunworth and Mills (1999) found that the
policy is a rather strong stimulant of community building. It aids in creating public and private
partnerships to develop programs that benefit communities. In addition, Weed and Seed policies
concentrated on building trust and on encouraging community participation in efforts to
revitalize communities. It is evident that Weed and Seed displays the factors of social
disorganization theory in that the goal is to involve the community and various agencies in
attaining a social cohesion in order to lower crimes rates together.
Community Disorganization and Racially Motivated Crime
In his 2007 study, Lyons explores the relationship between community context and the
various crimes motivated both by animus for black persons and white persons. Lyons used

85

official data from the City of Chicago Police Department, U.S. Census data, and community
survey data from Chicago. In doing so, he was able to discern the contextual factors that
affected the creation of social organization – or collective efficacy. Lyons (2007) found that
perhaps in opposition to general assumptions about bias or hate criminality, anti-black crime is
actually most likely to occur in relatively organized communities with high levels of social
control. On the other hand, anti-white crime tends to take place in rather disorganized
neighborhoods – spaces known for residential instability. With these findings taken into
account, Lyons (2007) certainly establishes a new perspective through which to view bias-based
criminality. As mentioned, applications of social disorganization theory to bias crime are almost
non-existent; therefore, any scholarly attention to community organization and residential
instability is quite salient.
Disorganization and Far Right Violence
More recently, Freilich, Adamzcyk, Chermak, Boyd, and Parkin (2013) introduce a
county-level test of social disorganization theory as applied to far-right extremist violence.
Primarily, Freilich et al. (2013) approached the study with the intention of examining variations
in the counties in which far-right extremists lived during the time of the criminal event. Social
disorganization theory was used in order to help uncover some of the variations in the spatial
aspect of the study – the county. Moreover, the authors examined the differences that American
counties provided in understanding far-right extremist attacks. However, Freilich et al.’s (2013)
study adds to the literature in a significant way in that it uses social disorganization theory to
explain these variations. The results of the study indicate that perhaps social disorganization
theory is helpful in this explanation as it provides details into the counties in which perpetrators
resided at the time of the attack. This is then juxtaposed against the findings with regard to the

86

location of the attack. In this way, the authors suggest that certain variables from social
disorganization theory do work to explain “regular crime” and far-right extremist criminality
alike. However, there are some differences in that not all attributes can explain both types of
crime (Freilich, Adamczyk, Chermak, Boyd, & Parkin, 2013). It is certainly worth noting that
this study is quite new in its usage of social disorganization theory to speak to far-right violence.
Given this dissertation’s focus is on bias crime occurrence at the county level, Freilich et al.’s
(2013) is particularly relevant.
In summary, various policies and community actions designed to lower crime rates can
be informed by testing social disorganization theory. Shaw and McKay’s CAP program used
their social disorganization theory and found mixed results. Urban renewal projects, such as
those in Detroit, suggest that injecting social, economic, and political stability can build social
cohesion, however they are usually met with failure. Community policing perspectives often use
social disorganization theory and broken windows theory in order to create a different culture of
policing, which in turn provides community actions that support community involvement and
social cohesion. Chicago’s CAPS project found limited success in its white and AfricanAmerican communities, but failed in the Latino community. The PHDCN primarily identified
ways of understanding informal social control and social cohesion – all in an effort to examine
collective efficacy. Researchers found Weed and Seed to be a strong stimulant of community
building in a national evaluation. Freilich et al. (2013) found that social disorganization theory
does help to explain some attributes of far-right extremist violence, but not other. The
involvement of the community and the promotion of social cohesion and change reflect the basic
implications of social disorganization theory.

87

Social Disorganization Theory and Bias Crime in the United States
From Shaw and McKay’s (1942) original social disorganization theory to the 2010s,
there has been significant theoretical development of this important communities and crime
perspective. Social disorganization theory began as a macro-level theory to examine the
differences in crime between neighborhoods. Shaw and McKay theorized that these
neighborhoods often had particularly high rates of residential mobility compared to other areas
of the city, had a high level of racial heterogeneity, and had low socioeconomic statuses.
Although this theory went through a period of several years where it was considered irrelevant, it
returned to life during the early 1980s. Since that time, there have been numerous contributions
by various scholars. There have been tests of different variations of the theory over the years and
nearly all studies have reported results supporting social disorganization theory’s relevance. To
date, the variables that are most often considered to be central to social disorganization are
socioeconomic status, urbanism, racial heterogeneity, residential mobility, unsupervised peer
groups, family disruption, and collective efficacy. These variables are used to inform this
dissertation as it relates to bias crime. Moreover, there have been suggestions that social
disorganization theory include structural and cultural variables. Although there is some debate
about that issue, there is overall support for social disorganization theory as is currently exists
(Pratt & Cullen, 2005).
The bias crime literature is largely underdeveloped with regard to theory – both at the
micro and macro levels. The aim of this dissertation is to investigate whether traditional
indicators of social disorganization theory inform the knowledge base on bias crime occurrence
in United States counties. Therefore, the following discussion presents the necessary
components of social disorganization upon which this dissertation is based. It is with regard to

88

the indicators of social disorganization theory explored by other scholars in their studies of crime
and violence, that this dissertation expands knowledge of bias criminality in the United States.
Although the aforementioned discussion of social disorganization theory is pertinent to
understanding the perspective through which this dissertation is considered, of most relevance
are the traditional indicators of social disorganization theory that will be used to measure its
presence. Freilich, Adamczyk, Chermak, Boyd, and Parkin (2013) considered social
disorganization theory and its relevance in explaining far-right terrorism occurrence in various
contexts. Freilich et al. (2013) explore whether social disorganization theory can be used to
account for county-variations in where far-right terrorists lived during the time of incidence.
Additionally, the authors created a unique measure of social cohesion by which social
disorganization theory could be tested. This was achieved by aggregating opinion data from the
General Social Survey. The work of Freilich et al. (2013) informs this dissertation in that it
identifies eight conceptual indicators of social disorganization theory that can be used to
understand terrorism at the macro-level. Given the similarities between terrorism and bias crime,
several of these indicators are useful in informing the construction of appropriate measures to
test social disorganization theory and its ability to inform bias crime occurrence at the county
level. The following indicators identified by Freilich et al. (2013) will be used to inform the
measures social disorganization theory in this dissertation: 1) deprivation; 2) diversity; 3)
disorganization; and 4) social cohesion.
Hypotheses
Given this dissertation seeks to explore the bias crime occurrence in American counties,
several of Freilich, et al.’s (2013) indicators of social disorganization theory are used. Based on
the authors’ study, this dissertation will similarly measure social disorganization with the

89

following indicators: low economic status, social heterogeneity, social cohesion, and residential
mobility.
Low Economic Status
Deprivation can be measured in a variety of ways; however, this dissertation will
consider deprivation as an economic indicator. In terms of economics, there is importance in
identifying one’s ability to engage in the larger economy. For instance, attaining and
maintaining employment, the ability to purchase goods are two examples of ways economic
participation can be measured. As such, the lack of presence of these indicators could indicate a
specific economic deprivation. Lacking meaningful employment and the inability to have
purchasing power are certainly worthy of consideration as they represent the inability of a person
to take part in the larger economy. Therefore, this dissertation examines whether higher
percentages of unemployment in United States counties are more likely to experience bias crime
occurrence (H1).
Social Heterogeneity
Although social heterogeneity – or diversity - refers to a broad range of concepts, it is
certainly relevant with regard to bias crime occurrence. Perhaps the very concept of bias crime
is related to the notion of diversity – the presence of “difference” creating a sense of
disorganization in a community. This can be measured in a variety of ways; however, the most
feasible way is by examining the demographics of a given community. In the case of this
dissertation, that means examining counties for various levels of diversity. This will be
measured by accounting for the percentage of racial minorities living in the given counties.
Moreover, diversity will be measured by accounting for the number of foreign born persons
living in each county as well.

90

This dissertation examines whether higher levels of social heterogeneity in United States
counties are related to greater bias crime occurrence. More specifically, diversity is measured by
identifying the presence of specific racial, ethnic, and religious populations in United States
counties. Therefore, this dissertation explores whether higher percentages of non-white
residents in United States counties is related to greater bias crime occurrence (H2). As
mentioned, diversity is also accounted for in this dissertation by recognizing the presence of
religious minorities. As such, this dissertation examines whether higher percentages of foreign
born persons living in United States counties is related to bias crime occurrence (H3).
Social Cohesion
In order to capture more traditional indicators of social disorganization theory, another
aspect of the theory is measured. With relation to the notion of social cohesion, it is possible to
examine the maintenance of marital family bonds in American counties. This will be
accomplished by measuring the percentage of divorced persons residing in each county in the
United States. Therefore, this dissertation examines whether high percentages of divorce in
United States counties affect the level of bias crime occurrence in the respective counties (H4).
Moreover, scholars note that a community’s violent crime rate can be used to measure the level
of social disorganization present in that community. After all, when citizens are organized in a
manner that contributes to collective efficacy, it follows that crime is less likely to occur in that
community. Additionally, the presence of violent crime in a county is used to measure the social
cohesiveness of such communities. As such, this dissertation seeks to determine whether a
higher crime rate in a United States county affects the level of bias crime occurrence (H5).

91

Residential Mobility
This dissertation also examines the impact of residential mobility on the likelihood of
bias crime occurrence in United States counties. If residents are moving to and from specific
neighborhoods – or counties – it is possible that these communities do not have the opportunity
to organize and thereby prevent criminality. In addition, perhaps the more people in a
community that rent their homes, the less likely a community is to be socially organized. With
that lack of social organization, it is then possible that these neighborhoods may experience bias
crime occurrence at a differential level from other neighborhoods. Therefore, this dissertation
explores whether a higher percentage of residential mobility impacts the level of bias crime
occurrence in United States counties (H6).

92

CHAPTER 4: DATA AND METHODOLOGY
Chapter Four presents the data and the methodological approach used for this
dissertation. Researchers have studied the numerous reasons why bias crime offending might
take place; however, the extant literature is relatively silent in its discussion of factors that
potentially create a community climate that is rich for bias crime offending. By examining
various characteristics of American counties, their populations, and the frequency of bias crime
reporting in them, this dissertation strives to explore the potential predictors of bias crime using
United States counties - or their equivalent - as the units of analysis. The aim is to be able to test
social disorganization theory’s ability to explain bias crime occurrence at the county level.
Essentially, this dissertation seeks to examine macro-level communities as they affect the
occurrence of bias crimes. As such, the essence of bias crime awareness and prevention may lie
in understanding the factors of social disorganization theory and the links to bias crime
victimization. Given this, Chapter Four includes a discussion of classifying bias crime events,
using official data, a data collection strategy, the creation of variables for the study, and a
discussion of relevant methodological techniques.
Classifying Bias Crime Events in Unites States Counties
One of the challenges scholars have faced with regard to studying bias crime is simply
the defining of such crimes. As noted, bias crime is a relatively newly identified type of crime in
the United States and the world, alike. Although bias crime has historically existed for many
years, it is only recently that it has been explored in the academic literature. Whether in the
psychological, sociological, or criminological literature, the scholarly attention is relatively
recent. With this taken into consideration, it is important to acknowledge that the definition of
bias crime differs by space and time as well. However, this dissertation uses the official

93

definition of hate crime as introduced by the FBI. The FBI (2008) defines hate crimes as crimes
motivated by the offender’s bias based on the victim’s race, sexual orientation, religion,
ethnicity, national origin, and disability. Although there are numerous additional social groups
that could be included – such as age, residential status, and political affiliation – this dissertation
follows the direction of the FBI with regard to definition. Since the FBI is the only government
entity that collects annual hate crime data that can be analyzed at the county level, those data are
used to measure bias crime incidence in United States counties.
It is also important to consider how bias crime occurrences exist within the context of
Unites States counties. Although there is little known about bias crime in general, there is even
less known about the relationship between communities and bias crime events. Scholars have
examined bias crime as it exists at the national, state, and even city levels. However, bias crime
occurrence in Unites States counties has not been examined in the prior research. Certainly, it is
plausible that a variety of factors may influence the rate of bias crime occurrence. For instance,
by examining counties, this study may be able to uncover their differences between different
sizes of communities and their effects of bias crime occurrence. It is likely that there is a
meaningful difference between suburban, urban, and rural communities and their influence on
bias crime occurrence. By utilizing counties as the unit of analysis, this dissertation is able to
account for those differences by way of including all appropriate data.
Using Official Data
This dissertation seeks to add to the scholarly literature by way of analyzing official data
collected at the national, or macro-level. By testing social disorganization theory and its efficacy
in explaining bias crime occurrence, this study speaks to the effects that are known from official
data. Although official data are used in this study, it is important to highlight the strengths and

94

weaknesses this type of data possesses. The following paragraphs explain the importance of
official crime data in criminological studies.
Advantages of Official Data
Official data are frequently used in criminological studies due to their availability and
accessibility. In fact, most of the records used in criminological research tend to be arrest reports
– official records collected by local police departments (Maxfield & Babbie, 2011). Considering
this definition, it is essential to note that much of the official crime data in existence in the
United States are collected and maintained by local, state, and federal government entities.
Although this can be at the local, state, or federal level, governments have a vested interest in
recording crimes – and working to combat those crimes based on crime trends extracted from the
data.
In a seminal study of self-reporting and official data as they relate to delinquency,
Hindelang, Hirschi, and Weiss (1979) present several influential findings. Primarily, the authors
suggest that official crime data do in fact provide valid representations of the demographics of
criminality in the United States. This is extremely salient as it substantiates the use of official
crime data – especially when compared to the utilization of other data types, such as self-reports.
In addition, most official crime data record serious crimes – as opposed to minor crimes (Kleck,
1982). This is important to consider in that the reported bias crimes in the United States tend to
be more severe in nature than they are trivial. Moreover, when the findings of self-report studies
were compared to those from official records, the findings were remarkably consistent – both in
terms of reliability and validity (Hindelang, Hirschi, & Weis, 1979); (Maxfield, Weiler, &
Widom, 2000).

95

Additionally, official crime data are a matter of public record, and are therefore
accessible by researchers and citizens alike. Official data are certainly available in paper format
in libraries and government agencies; however, the Internet allows for a more accessible
approach. For instance, by navigating to the FBI’s website – and the USCB’s – researchers can
have access to annual crime statistics and various demographic figures. Certainly, there are
limitations to official data – as is discussed later in this chapter – however, there is a need to
consider bias criminality at the county level. Given the breadth of this approach, using official
data is not only one of the few feasible approaches to data, but also perhaps the most inclusive as
well. After all, using official data allow this dissertation to explore bias criminality across the
United States as a whole – as opposed to a mere selection of communities. The result could be
perhaps a more thorough – if not generalizable – set of findings about bias crime occurrence in
the United States.
Drawbacks to Using Official Data
Although there are several advantages of using official crime data, there are also several
negatives to the data type as well. Primarily, official crime data have a reputation for being
rather conservative by nature. Essentially, because of the voluntary nature of official crime data
collection, researchers often note missing data. Generally, when considering the “dark figure of
crime,” we know that more crime occurs than is reported to the police in the United States. This
is certainly the case for bias crime as well; in fact, underreporting may even be more pronounced
here than with other crime.
In addition, Hindelang, Hirschi, and Weiss (1979) suggest that official crime records can
be too stringent to uncover the intricacies of criminality. Moreover, the limited nature of official
records might include the fact that variation is difficult to ascertain. It may be that official

96

records will have more minor crimes reported than major crimes. For instance, it is likely that
official crime data include most reported traffic offenses, but do not account for the same
proportion of robberies, rapes, and arsons. It follows that bias crime may be reported in less
serious situations more frequently than serious examples. With the exception of celebrated
cases, which are most likely to be reported, it may be that the bias homicide does not get reported
as a being related to bias. Whether this is due to the aforementioned lack of gravity with which
some police departments treat bias crime or due to oversight, bias crime is often unrecorded.
Certainly, this includes the phenomenon of underreporting as well.
Official crime records also reflect the biases that may be present in criminal justice
actors. Due to the disproportionate representation of lower socio-economic status individuals,
males, and African Americans in the criminal justice system (Piquero, MacIntosh, & Hickman,
2002), it is probable that a similar disproportionate representation exists in the official crime
records. This would certainly skew the findings of any study with dependent variables
constructed from official crime data. Certainly, this limitation is acknowledged as problematic;
however, when considering bias crime occurrence – given the power structure and relations – it
is also necessary to consider whether the entire reporting system for bias crime reflects reality.
Given the victims of bias crime tend to be from social minority groups, such as African
Americans, Jewish persons, and lesbian and gay citizens - and since the police are
overwhelmingly white males, it is important to consider whose story is being captured in official
crime records. If official crime records reflect the power structure present in the United States, it
would presumably reflect the stories of those in power – in essence, those who are part of social
majority groups. After all, those in power often dictate the nature of the local culture – whether
it be related to sex, race, sexual orientation, or ability.

97

Data Collection Strategy
The data used for this dissertation are collected from two United States government
agencies and two private entities. The primary data used to capture the dependent variables and
certain independent variables were provided by the Federal Bureau of Investigation’s Uniform
Crime Report. After the introduction of the Hate Crime Statistics Act of 1990, the FBI has been
collecting reported bias crime data nationwide. These data are reflected annually in the Uniform
Crime Report.
In order to properly collect the relevant data to construct dependent variables representing
bias crime occurrence in United States counties, the website of the Inter-University Consortium
for Social and Political Research (ICPSR) at the University of Michigan was accessed. The
ICPSR has archived a variety of hate crime data from the FBI – including data for the relevant
years of 2000-2009 that are necessary for this study. Each file is archived by year; therefore,
each of the ten appropriate years was downloaded for the purposes of this study. The files are
quite large and do include appropriate information about the frequency of hate crime occurrence,
motivation of offender, and location of criminal offense. Each data file was downloaded from
ICPSR in Microsoft Excel format, reorganized, and exported to a master SPSS file.
The FBI’s UCR data are organized by incident, thereby inhibiting analyses at the countylevel unit of analysis. Given this dissertation is specifically aimed at understanding bias crime at
the county level, the incident-level data from the FBI were aggregated in order to create a
county-level measure. The data were rearranged in Microsoft Excel so that the unit of analysis
was transformed from incident-level to county-level. The created measure allowed for the
number of bias crimes that occurred in each county between 2000 and 2009 to be counted. The
resultant data allow for the answering of the research questions posed in Chapter Three.

98

In addition, data from the United States Census Bureau are used to construct the
independent variables. As mentioned, the 2000 census was used as a marker for the subsequent
decade. These data were gathered by accessing the United States Census Bureau’s website and
using the American FactFinder tool. From this, data representing the appropriate measures of
social disorganization as well as control variables were extracted from the database. Data were
downloaded in Microsoft Excel format and were later reorganized and added to the master SPSS
file.
Other county-level data were provided by various sources. Religious data are provided
by the Association of Religious Data Archives (ARDA). Since the USCB is prevented by law
from collecting information of Americans’ religious affiliations and practices, accessing the
ARDA’s website proved fruitful in obtaining data related to religious congregants present in the
United States. In order to remain consistent with United States Census Bureau data used to
construct various independent variables, data for the year 2000 were accessed and downloaded to
represent the following decade. Data were available for download in SPSS format.
In addition, data representing the political participation of United States citizens were
obtained from Congressional Quarterly Voting and Elections Collection. The data are available
in Microsoft Excel formal for the year 2000. This allows for the creation of the relevant
independent variable representing voter participation to be created and used in analyses.
Specifically, this dissertation uses the number of persons per county voting in the 2000
presidential election to measure political participation.
The dependent variables for this dissertation are based on the frequency of bias crime
occurrence in United States counties. In addition to including the aggregate figure of bias crime
occurrence, dependent variables based on the motivation of the offender are included as well.

99

Each of these presents the frequencies of bias crime occurrence in each county. Specifically, the
three most commonly occurring motivations are examined: anti-race, anti-sexual orientation, and
anti-religion. Many of the independent variables in this dissertation are based on the traditional
indicators of social disorganization theory. This allows for analyses to be conducted that explore
the relevance of social disorganization theory as applied to bias crime occurrence at the county
level. In order to verify that the models are answering the research questions accurately,
numerous control variables are included. Each of the datasets were downloaded and aggregated
in an SPSS file. Moreover, the research conducted for this dissertation is classified as “exempt”
by Michigan State University’s Institutional Review Board3 and is therefore cleared for analyses.
Counties as the Unit of Analysis
The unit of analysis for this study is counties – or each state’s equivalent – in the United
States. For instance, the state of Louisiana signifies their sub-state political units as parishes. In
contrast, the rest of the United States uses the term county to represent its sub-state units. In the
case of this dissertation, counties are used because they allow for a more precise level of
measurement than larger political units such as states. In addition, using counties as the unit of
analysis can help to control for variation between the states (Maltz & Targonski, 2002). In
essence, using counties allows for a greater number of communities to be studied. Moreover, as
observed in Porter, Rader, and Cossmann (2012), ecological data measured at the county level
are appropriate for testing social disorganization and the relevant theory. Given social
disorganization theory is a macro-level theoretical approach to understanding crime, measuring it
at a larger political level, such as at the county level, is appropriate. The FBI does report bias
crime data at the county level and the city level, but as mentioned - county level data provide

3

IRB #x13-1200e, Exempt #i045116

100

greater precision (Maltz & Targonski, 2002). However, a limitation is the fact that the majority
of counties do not, in fact, report having bias crimes occur.
In order to achieve a large enough sample size to work toward statistical conclusion
validity, each county and parish in the United States is included in the study. Moreover,
Washington, DC is included as well. However, counties and parishes that do not report bias
crimes are only included in the main model. Table 2 presents basic characteristics of all 3,141
United States counties.
Table 4.1: Characteristics of United States Counties by State
State
Number of Counties Population (2000)
Alabama
67
4,447,100
Alaska
27
626,932
Arizona
15
5,130,632
Arkansas
74
2,673,400
California
58
33,871,648
Colorado
64
4,301,261
Connecticut
8
3,405,565
Delaware
3
783,600
District of Columbia
1
572,059
Florida
67
15,982,378
Georgia
159
8,186,453
Hawaii
4
1,211,537
Idaho
44
1,293,953
Illinois
102
12,419,293
Indiana
92
6,080,485
Iowa
99
2,926,324
Kansas
105
2,688,418
Kentucky
120
4,041,769
Louisiana
64
4,468,976
Maine
16
1,274,923
Maryland
24
5,296,486
Massachusetts
14
6,349,097
Michigan
83
9,938,444
Minnesota
87
4,919,479
Mississippi
82
2,844,658
Missouri
115
5,595,211
Montana
56
902,195
Nebraska
93
1,711,263
Nevada
17
1,998,257
New Hampshire
10
1,235,786
101

Table 4.1 (Cont’d)
New Jersey
New Mexico
New York
North Carolina
North Dakota
Ohio
Oklahoma
Oregon
Pennsylvania
Rhode Island
South Carolina
South Dakota
Tennessee
Texas
Utah
Vermont
Virginia
Washington
West Virginia
Wisconsin
Wyoming
Total

21
33
62
100
53
88
77
36
67
5
46
66
95
254
29
14
136
39
55
72
23
3,141

8,414,350
1,819,046
18,976,457
8,049,313
642,200
11,353,140
3,450,654
3,421,399
12,281,054
1,048,319
4,012,012
754,844
5,689,283
20,851,820
2,233,169
608,827
7,078,515
5,894,121
1,808,344
5,363,675
493,782
281,421,906

As is evident from the Table 4.1, the majority of states have dozens of counties within
their borders – some more than others. Containing 254 counties, Texas is the state with the
highest number of counties; with three, Delaware has the fewest. The District of Columbia has
one county, the County of Washington. California is the most populous state as it has
33,871,648 residents, as of 2000. Wyoming has a population of 493,782 individuals, making it
the least populous state in the country. Nationwide, as of the 2000 Census, the United States had
a total of 281,421,906 residents.
Given that so many counties do not report bias crime to the FBI, it is important to account
for this fact in the primary model. This allows for a more inclusive analysis that can speak to all
United States counties. In contrast, subsequent models include only counties in each state that
reports bias criminality to the FBI. As such, counties and parishes that report zero bias crimes

102

are included as well. The subanalyses that exclude non-reports allow greater precision in
understanding the impacts of bias criminality. Essentially, given they only include the counties
that do in fact report incidents as having occurred, the analyses are able to speak to actual bias
crime occurrence without potentially being moderated by non-reports.
The total N is 3,141 - as this figure represents the total number of counties and parishes in
the United States. However, the FBI data indicate that the majority of counties in the United
States report zero hate crimes per year. For this dissertation, UCR Hate Crime data from 20002009 and U.S. Census data from 2000 are used. As mentioned, although U.S. Census data are
available for a variety of years, relevant independent variables are constructed from the 2000
U.S. Census data as to provide indicators that speak to the years 2000-2009. In addition, ARDA
data from 2000 are used for the creation of religion-related independent variables. Data from
Congressional Quarterly are used to construct an independent variable representing political
participation in 2000.
Operationalization of Dependent and Independent Variables
Numerous variables are used to analyze the posed research questions. Moreover, the
operationalization of each variable is explained and the data source for each variable is identified
and justified. Each variable is constructed in a manner so as to measure or support the
measurement of traditional indicators of social disorganization theory. Control variables are also
used to ensure the findings of the analyses speak directly to the specific research questions asked.
Operationalization of Dependent Variables
The goal of this dissertation is to determine the likelihood of bias crime occurrence in
United States counties. Given this, the concept this dissertation seeks to measure is the presence
of bias crimes across the United States. As previously discussed, bias crime is defined as a

103

criminal act motivated by animus toward the victim’s affiliation with an oppressed social group
(Gerstenfeld, 2004). Considering this definition, the operationalization of this concept also
requires understanding what is not considered a bias crime for the purposes of this study.
More specifically, this dissertation uses the FBI’s definition of a hate crime to operationalize the
concept. The FBI uses the following definition to categorize hate crimes in the United States:
“The Hate Crime Statistics Program of the FBI’s Uniform Crime Reporting (UCR) program
collects data regarding criminal offenses that are motivated, in whole or in part, by the offender’s
bias against a race, religion, sexual orientation, ethnicity/national origin, or disability and are
committed against persons, property, or society” (Federal Bureau of Investigation, 2008). It is
important to note that offender motivation is the primary difference between the FBI’s
designation of hate crimes and that of other index offenses. Certainly, motivation is difficult to
determine, but the FBI relies on local police departments’ designations of motive and the
existence of bias criminality. This means that the data the FBI collects and maintains are only as
valid and reliable as are the local police departments reporting the offenses. There are likely
numerous bias crimes that go unreported each year. However, the FBI cannot possibly
determine the existence of these crimes unless they are reported as such.
Bias crime is quantitatively measured by examining the reported figure of bias crimes by
United States law enforcement agencies. This is indicated by considering the bias crimes known
to the FBI. The dependent variables for this dissertation are constructed from the frequencies of
bias crime offenses – as reported – to the FBI for each of the United States law enforcement
jurisdictions. These data are available for the range of years 2000-2009 and are continuous in
nature.

104

The FBI’s Hate Crime data were accessed and downloaded from the Inter-University
Consortium for Political and Social Research (ICPSR) at the University of Michigan. These data
are publicly available and provided in datasets organized by year. UCR Hate Crime data are
available for all years 2000-2009. These UCR files provide data for hate crime incident per
motivation and quarter for each year. Moreover, they include the various states and agencies
from which the hate crime report originated. This informs the creation of the dependent
variables - essentially, the total number of bias crimes in each United States county by
motivation type. It is also necessary to provides statistics for each state and agency that did not
report data for the respective year. These data are used to consider the research questions posed
in Chapter Two.
Creation of Dependent Variables
In order to determine the variation that may exist between different types of bias
criminality, several dependent variables are used. First, all reported bias crime incidents are
collected for each United States county or parish. Therefore, the total number of bias crimes that
occur in each United States county are used to construct the dependent variables – which are
continuous in nature. Even with a dependent variable that is continuous in nature, it is important
to acknowledge the number of zeroes present in the data. Given this, appropriate analytical
techniques are considered and utilized. Moreover, these data are available at the county level
and are disaggregated to build three other dependent variables. These variables are constructed
in a manner that allows this dissertation to examine the nature of anti-sexual orientation, antirace, and anti-religion motivated bias crime. Moreover, this dissertation seeks to examine the
differences between the various types of motivation that affect bias crime occurrence.

105

The following identify the type and source of data used to construct the dependent variables in
this dissertation:
1) Bias crimes reported (by county)


Data source: UCR (2000-2009)

2) Anti-sexual orientation bias crimes reported (by county)


Data source: UCR (2000-2009)

3) Anti-race bias crimes reported (by county)


Data source: UCR (2000-2009)

4) Anti-religion bias crimes reported (by county)


Data source: UCR (2000-2009)
In order to understand the nature of bias crime occurrence in the United States, it is

necessary to examine the frequencies of the three most commonly occurring bias crime. Given
the data represent bias crime reported to the FBI for the years 2000-2009, importance lies in
breaking the data down into useful segments. First, Table 4.2 presents the number of all bias
crimes reported in the United States, organized by state and federal district. The data are
presented for all years 2000-2009. Second, Table 4.3 displays the characteristics of bias crime
identified by the following motivations: anti-race, anti-sexual orientation, and anti-religion.
Although the unit of analysis for this dissertation is United States counties, descriptive data are
organized by state as this organization is more amenable to presentation in table format. The
3,141 counties and parishes that exist in the United States are too numerous, and therefore too
cumbersome, to present in a single table.

106

Table 4.2: Reported Bias Crime Totals by State
State
2000
2001
2002
2003
2004
AL
0
0
2
1
3
AK
0
0
0
0
0
AZ
248
395
244
253
197
AR
4
3
0
176
92
CA
1,988 2,250 1,651 1,476 1,164
CO
109
135
105
89
60
CT
169
181
135
152
119
DE
35
23
13
17
26
DC
5
11
14
31
49
FL
216
266
240
216
220
GA
35
39
31
28
17
HI
0
0
0
0
0
ID
49
35
44
21
33
IL
192
298
172
222
159
IN
111
79
79
49
50
IA
33
33
48
46
23
KS
43
72
58
50
40
KY
73
83
79
81
61
LA
12
10
16
7
21
ME
29
33
40
87
37
MD
210
227
199
234
216
MA
465
605
435
422
286
MI
439
456
420
439
500
MN
170
210
203
215
205
MS
2
3
3
1
2
MO
74
69
47
51
56
MT
21
14
14
5
53
NE
18
56
76
47
54
NV
97
101
63
99
72
NH
33
27
27
39
36
NJ
652
766
570
594
714

2005
0
0
152
135
1,427
132
111
49
48
204
17
0
33
175
54
39
78
42
20
60
188
380
663
206
0
65
69
71
77
34
738

2006
1
0
161
118
1,376
150
143
54
57
199
13
0
23
159
40
34
113
60
23
62
205
386
666
137
0
75
27
57
127
35
755
107

2007
6
0
167
35
1,532
170
128
52
41
153
13
0
42
173
41
29
118
46
33
78
141
359
639
157
0
96
22
45
63
45
748

2008
11
0
194
92
1,507
164
156
63
42
135
10
0
30
124
61
33
118
62
75
65
97
333
582
164
4
87
25
12
97
46
744

2009
9
0
226
78
1,119
223
192
39
37
115
11
0
35
137
55
17
130
154
17
51
96
328
327
153
2
120
29
78
48
26
549

Total
33
0
2,237
733
15,490
1,337
1,486
371
335
1,964
214
0
345
1,811
619
335
820
741
234
542
1,813
3,999
5,131
1,820
17
740
279
514
844
348
6,830

Table 4.2 (Cont’d)
NM
16
20
NY
209
399
NC
33
90
ND
5
17
OH
247
388
OK
83
51
OR
142
223
PA
147
138
RI
46
63
SC
34
42
SD
7
5
TN
222
337
TX
305
468
UT
81
71
VT
20
17
VA
98
177
WA
250
293
WV
69
41
WI
49
62
WY
9
10
Total
7,604 9,392

15
413
65
18
285
49
61
94
38
72
4
126
367
58
23
105
175
46
33
5
7,080

11
313
88
19
235
43
96
122
46
56
10
134
308
64
28
112
233
33
31
12
7,142

18
261
62
9
325
51
126
89
22
91
5
118
267
57
24
129
154
28
27
4
6,432

18
251
93
12
185
42
138
120
15
104
9
127
274
50
35
146
181
49
49
3
7,168

20
250
107
16
324
54
147
99
20
113
78
175
254
39
21
174
187
34
89
5
7,462

108

14
199
79
14
328
33
179
89
47
132
37
234
255
59
21
175
203
45
72
21
7,408

10
311
139
18
360
51
199
70
36
159
42
264
258
41
20
145
264
46
97
7
7,670

15
356
106
14
313
66
136
38
36
114
43
170
168
50
25
67
222
28
56
8
6,432

157
2,962
862
142
2,990
523
1,447
1,006
369
917
240
1,907
2,924
570
234
1,328
2,162
419
565
84
73,790

As is evident in Table 4.2, California reliably reports the highest number of bias crimes in
the United States. In the ten year period from 2000-2009, California reported 15,490 bias crimes
as having occurred within its boundaries. Of the ten years, 2001 had the highest number of
reported bias crimes in California, with 2,250 incidents reported. The lowest number of bias
crimes reported was in 2009, when the state had 1,119 incidents. In addition, the state that
reports the lowest number of bias crimes is Mississippi, with 17 incidents reported between 2000
and 2009. The year of 2008 held the most reported bias crimes in Mississippi. That year, the
state reported four bias crime incidents as having occurred.
Alaska and Hawaii are excluded from analyses in this dissertation. Although Alaska does
report bias crime to the FBI, its political structure is divided into organized and unorganized
boroughs instead of counties. In the available UCR Hate Crime data, Federal Information
Processing Standard (FIPS) are absent. This precludes analyses due to the lack of availability of
assigned codes. The District of Columbia suffers from the same issue; however, since the
district falls into only one political unit, Washington, the data can be accurately utilized in this
dissertation. Any data that may exist for Hawaii are unavailable from the UCR and therefore not
included in the dataset.
Table 4.3 presents basic statistics surrounding the context of bias crime occurrence.
Specifically, the total number of bias crimes in each state is presented. In addition, those
reported crimes are presented by motivation type. The research questions of this dissertation
address the differences between anti-race, anti-sexual orientation, and anti-religion motivated
bias crime. Although other types do occur, they are excluded from analyses in this dissertation.

109

Table 4.3: Bias Crime Motivations (2000-2009)
State
Total Anti%
Antirace
sex.
orient.
Alabama
33
25 75.76
5
Alaska
0
0
0
Arizona
2,237 1,446 64.64
318
Arkansas
733
590 80.49
79
California
15,490 9,718 62.74
2,798
Colorado
1,337
849 63.50
172
Connecticut
1,486
867 58.34
171
Delaware
371
232 62.53
53
District of
Columbia
335
90 26.87
218
Florida
1,964 1,371 69.81
283
Georgia
214
134 62.62
58
Hawaii
0
0
0
Idaho
345
223 64.64
37
Illinois
1,811 1,174 64.83
288
Indiana
619
427 68.98
98
Iowa
335
223 66.57
50
Kansas
820
582 70.98
121
Kentucky
741
558 75.30
112
Louisiana
234
171 73.08
17
Maine
542
289 53.32
152
Maryland
1,813 1,251 69.00
120
Massachusetts 3,999 2,437 60.94
711
Michigan
5,131 3,876 75.54
526
Minnesota
1,820 1,335 73.35
265
Mississippi
17
15 88.24
0
Missouri
740
542 73.24
73
Montana
279
153 54.84
49
Nebraska
514
369 71.79
58
Nevada
844
556 65.88
123
New
Hampshire
348
169 48.56
86
New Jersey
6,830 3,819 55.92
494
New Mexico
157
103 65.61
34
New York
2,962 1,391 46.96
275
North
Carolina
862
647 75.06
84
North Dakota
142
113 79.58
15
Ohio
2,990 2,081 69.60
348
Oklahoma
523
381 72.85
41
Oregon
1,447
906 62.61
287
Pennsylvania
1,006
685 68.09
82
110

%

Antireligion.

15.15
14.22
10.78
18.06
12.86
11.51
14.29

3
0
377
32
2,362
207
282
55

65.07
14.41
27.10
10.72
15.90
15.83
14.93
14.76
15.11
7.26
28.04
6.62
17.78
10.25
14.56
0.00
9.86
17.56
11.28
14.57

%

9.09

Other

%

16.85
4.37
15.25
15.48
18.98
14.82

0
0
96
32
612
109
166
31

0.00
4.29
4.37
3.95
8.15
11.17
8.36

26
309
16
0
73
250
72
35
71
55
22
65
431
742
541
204
2
90
41
71
134

7.76
15.73
7.48
21.16
13.80
11.63
10.45
8.66
7.42
9.40
11.99
23.77
18.55
10.54
11.21
11.76
12.16
14.70
13.81
15.88

1
1
6
0
12
99
22
27
46
16
24
36
11
109
188
16
0
35
36
16
31

0.30
0.05
2.80
3.48
5.47
3.55
8.06
5.61
2.16
10.26
6.64
0.61
2.73
3.66
0.88
0.00
4.73
12.90
3.11
3.67

24.71
7.23
21.66
9.28

72
2,493
17
1,265

20.69
36.50
10.83
42.71

21
24
3
31

6.03
0.35
1.91
1.05

9.74
10.56
11.64
7.84
19.83
8.15

70
10
273
65
208
201

8.12
7.04
9.13
12.43
14.37
19.98

61
4
288
36
46
38

7.08
2.82
9.63
6.88
3.18
3.78

Table 4.3 (Cont’d)
Rhode Island
369
181
South
Carolina
917
586
South Dakota
240
182
Tennessee
1,907 1,294
Texas
2,924 1,983
Utah
570
334
Vermont
234
113
Virginia
1,328
891
Washington
2,162 1,450
West Virginia
419
291
Wisconsin
565
374
Wyoming
84
48
Total
73,790 47525

49.05

108

29.27

75

20.33

5

1.36

63.90
75.83
67.86
67.82
58.60
48.29
67.09
67.07
69.45
66.19
57.14
64.41

106
27
310
458
75
66
119
336
71
107
16
10,500

11.56
11.25
16.26
15.66
13.16
28.21
8.96
15.54
16.95
18.94
19.05
14.23

116
27
205
323
117
39
210
250
29
59
19
12,711

12.65
11.25
10.75
11.05
20.53
16.67
15.81
11.56
6.92
10.44
22.62
17.23

109
4
98
160
44
16
108
126
28
25
1
3,054

11.89
1.67
5.14
5.47
7.72
6.84
8.13
5.83
6.68
4.42
1.19
4.14

Table 4.3 presents the breakdown of bias criminality by motivation for the years 2000
through 2009, combined. From this, it is evident that out of the 73,790 reported bias crimes
occurring in the United States over the ten year period, 64.41% were motivated by animus
toward the victim’s race. Second to race as a motivation for bias crime offending is religion. It
is anti-religion based animus that accounts for 17.23% of bias criminality. Bias crimes
motivated by anti-sexual orientation based animus are the third most frequently occurring type of
bias crime. From 2000-2009, 14.23% of bias crimes in the United States were motivated by antisexual orientation hostility.
Operationalization of Independent Variables
Social disorganization theory has long been discussed as a leading theoretical perspective
explaining the incidence of crime in communities. As such, concepts that are used to test social
disorganization theory are introduced in Chapter Three. The following section explains the
operationalization of these concepts into feasible independent variables. This dissertation uses
American FactFinder from the United States Census Bureau in order to construct appropriate
independent variables – both those that capture traditional indicators of social disorganization
111

theory and appropriate control variables. Sampson and Groves (1989) identified four traditional
indicators of social disorganization theory. It is Sampson and Groves’ (1989) study upon which
the social disorganization indicators are based in this dissertation.
Low economic status, social heterogeneity, residential mobility, and social cohesion are
commonly considered to represent factors of social disorganization theory and its ability to
predict crime and violence in communities. Therefore, these four concepts are used to test social
disorganization theory and the theory’s ability to predict bias crime occurrence at the county
level in the United States. Although Sampson and Groves’s (1989) indicators are used, this
dissertation adds numerous variables to measure and test social disorganization theory. The
majority of these variables are informed by the more recent work of Freilich, Adamczyk,
Chermak, Boyd, and Parkin (2013). It is from this study that various indicators of social
disorganization theory as well as control variables are informed.
Low Economic Status
One traditional indicator of social disorganization theory is the concept of low economic
status (Shaw & McKay, 1942). Kornhauser (1978) suggests that low economic status is perhaps
the most salient indicator of social disorganization theory. Kornhauser (1978), in effect, suggests
that communities with lower economic capabilities have fewer resources to discourage
criminality. In order to operationalize this concept for the purposes of this dissertation, three
concepts are considered: percent unemployed, percent living below the poverty line, and the
percent of citizens without a college degree.
Based on Sampson and Groves’ (1989) study, the occupational status of persons in each
county is considered. Specifically, this includes examining the percentage of individuals
employed. First, the percentage of unemployed persons in each United States county or parish is

112

used to measure low economic status. These data are taken from the United States Census
Bureau’s American FactFinder. Second, the percentage of individuals living below the poverty
line in each county is used to measure low economic status. Similar to percent unemployed,
these figures are available from American FactFinder for each United States county or parish.
Third, the percentage of individuals residing in each county without a college degree is used to
measure low economic status. These data are available and are sourced from the American
FactFinder. Each of these estimations of unemployment is available for all United States
counties or parishes for 2000.
Social Heterogeneity
Social heterogeneity – which also can be defined as diversity - is yet another factor that is
important when considering the social organization or disorganization of a community. The
social heterogeneity of a county is likely an important factor in understanding the distribution of
crime. Kornhauser (1978) also suggests that racial heterogeneity is an adequate measure of
social disorganization theory as more diverse communities have greater difficulty organizing.
Even Suttles in his (1972) study suggested that heterogeneity increases delinquency and weakens
the various aspects of social organization. The identification of the following serve to
operationalize social heterogeneity: percentage of non-white persons, percentage of foreign-born
persons, percentage identifying as Jewish, and the percentage of citizens identifying as Muslim.
First, the percentage of non-white persons living in each United States county or parish is
used to measure social heterogeneity. This is constructed by attaining available data from
American FactFinder. Certainly, given the overwhelming number if white-identifying
individuals in the United States, it is important to consider the experiences of individuals who are
non-white. Second, social heterogeneity is operationalized by identifying the percentage of

113

foreign-born persons present in each county. This indicator is used to represent the level of
immigration each United States county experiences. Third, the diversity of religious population
in each county is considered as well. In order to determine religious diversity, the percentage of
individuals who are Jewish, Muslim, and Christian are identified. Specifically, because the
overwhelming majority of individuals in the United States identify with Christianity in some
manner, the percentage of Catholic citizens and the percentage of Protestant citizens are
considered as a control variable. Although Christianity can be separated into numerous sects,
only Catholicism and Protestant denominations in the aggregate are considered. Often this
separation is into “Catholic” and “other Christian” given the nature of how Christianity has
historically evolved. Therefore, “other Christian” includes the entire range of sects from
mainline denominations to evangelical sects of Christianity. Each of these estimations of social
heterogeneity is available for all United States counties or parishes for 2000.
Social Cohesion
As is appropriate when measuring aspects of social disorganization theory, considering a
community’s level of social cohesion is necessary. As scholars have noted, the level of
disorganization (or lack of social cohesion) that a community experiences may be related to the
level of criminality it experiences. Therefore, in the case of this dissertation, social cohesion is
measured by examining each county’s crime rate. Additionally, the divorce rate is another
measure of social cohesion used in this dissertation.
The United States Census Bureau’s American FactFinder is used to capture the
percentage of divorced persons living in each American county or parish. Historically, Sampson
(1987) has suggested that family disruption can lead to criminal activity. As such, the
percentage of divorced persons residing in each county or parish is used to measure family

114

disruption. Specifically, divorce is used to represent the concept of social disorganization given
it represents the disintegration of families in American communities. Certainly, this definition
operates under the assumption that a family is such because a married heterosexual couple heads
the household. Although there are numerous ways a family can be identified and structured, this
dissertation is limited by the available data. It was not until 2010 that President Barack Obama
instructed the United States Census Bureau to collect information on households that may
include same-sex couples. Therefore, even with its current and inherit inequality, the term
marriage, as of 2000 - only included heterosexual couples. As such, only heterosexual couples
are included in the definition of family in this dissertation.
Although bias criminality is certainly included in the FBI’s overview of American crime,
it is important to discern the relative level of violent criminality present in each county or parish.
This allows this dissertation to measure social cohesion in yet another manner. For instance, if a
county with numerous incidents of bias criminality also has a high rate of other violent crime,
this fact could be significant. In the event that the county has a low rate of violent crime
otherwise, but a relatively high rate of bias crime, this could be a finding worthy of attention.
This variable is operationalized by identifying and using UCR data for each American county or
parish. The frequency of violent crime occurrence is used to measure violent (index) crime in
American counties and parishes. Each of these estimations of social cohesion is available for all
United States counties or parishes for 2000. In addition, political participation data are provided
by Congressional Quarterly’s Voting and Elections Collection. Specifically, these data allow for
county-level measurement of percentage voting in the 2000 presidential election. Each of these
variables is available for all United States counties for the year 2000 and is measured in a
continuous manner.

115

Residential Mobility
In order to capture the concept of residential mobility, the percentage of home-owners in
each United States county or parish is used. This is constructed in order to accommodate
Brooks-Gunn et al.’s (1997) assertion that there are correlations between residential instability
and poverty – two clear indicators of social disorganization. Sampson (2002) also suggests that
there is much stratification in the way that neighborhoods interact with criminality. Moreover,
Kasarda and Janowitz (1974) purport that the integration of new residents in a neighborhood
does not always result in extensive friendship networks; rather, it supports the residential
mobility that is theoretically problematic. As such, the data for this dissertation are available
from the United States Census Bureau’s American FactFinder for each United States county.
Given home owners are likely to be more permanent fixtures in their communities than are
renters, the percentage of those who own homes in a county is used to measure residential
mobility. Moreover, the percentage of unoccupied housing in each county or parish is
considered as a measure of residential mobility. This measure is different from the percentage of
citizens owning homes because it accounts for rented properties as well. When considering
whether a community is social organized, it is necessary to understand the percentage of
individuals actually living in the community. Each of these estimations of residential mobility is
available for all United States counties or parishes for 2000.
Control Variables
In addition to the dependent variables and the social disorganization-related independent
variables, this study employs several control variables. In order to determine whether a
relationship exists between two or more variables, it is important to use control variables as a
constant. This allows the researcher to know whether the study is actually measuring what it set

116

out to measure. In the case of this dissertation, several attributes of American counties and
parishes are examined in an effort to obtain a constant during multivariate analyses.
In addition, the population of each American county or parish is included as a control
variable in this dissertation. It is important to be able to discern the effect a county’s population
has not only on bias criminality, but also each of the traditional indicators of social
disorganization theory. The population of each American county or parish is available from the
United States Census Bureau for 2000. The population for each county is constructed from 2000
Census data in order to maintain consistency with the other data used in the study. Each
county’s population is available from the United States Census Bureau’s American FactFinder.
Other control variables used are taken from the American Factfinder provided by the
United States Census Bureau. They are as follows: percentage of individuals in each county
where the language spoken is other than English, median age of residents, percent of veterans
residing in each county, and percent Christian (Catholic and Protestant).
Moreover, drawing from Lyons’s (2007) study, anti-white bias crimes are included as
controls as well. In this dissertation, bias crime committed against superordinate groups is not
considered directly; however, with regard to race, anti-white bias crimes are considered as a
control variable. The existence of hate crime legislation in each state in 2000 is used a control
variable as well. This is structured as a dichotomous variable and taken adopted from a
collection of hate crime laws compiled by the Southern Poverty Law Center. Additional related
variables break down the hate crime laws by the existence of the following protections: race,
religion, and ethnicity; sexual orientation; gender; gender identity; and disability. Each of these
variables is coded in a dichotomous manner as well. Although not directly related to traditional

117

indicators of social disorganization theory, the preceding control variables provide greater
information about the relationships present in this study.
Codebooks
As is appropriate in any study, the codebooks for relevant variables used in this
dissertation are presented. The codebooks include not only the dependent variables, but also the
independent variables necessary for this dissertation’s analyses. Both continuous and discrete
variables are included. Codebook 1 provides the variables necessary for measuring United States
counties. Codebook 2 presents the variables used to represent social disorganization theory.
Codebook 3 displays the control variables relevant to this dissertation. As follows, Table 4.4
presents the variables as are appropriate for considering the analyses in this dissertation.

118

Table 4.4: Codebooks for County, Social Disorganization, and Control Variables
Codebook 1. County
Codebook 2. Social Disorganization Codebook 3. Control
Variables (2000-2009) Theory Variables (2000)
Variables (2000)
County Population
Low Economic Status
Median Age of Residents
% Unemployed
Bias Crime Total
% below Poverty Line
% of Anti-White Hate
% without Bachelor’s
Crime Rate
Motivation
Degree
Anti-Race
Hate Crime Legislation
Anti-Sexual
Social Heterogeneity
No=0
Orientation
% Non-White
Yes=1
Anti-Religion
% Muslim
% Jewish
Race, Religion, Ethnicity
Bias Crime Event
% Foreign-born
Legislation
No=0
No=0
Yes=1
Social Cohesion
Yes=1
% Divorced
Crime Rate
Sexual Orientation
Legislation
Residential Mobility
No=0
% Owning Homes
Yes=1
% Unoccupied
Houses
Gender Legislation
No=0
Yes=1
Gender Identity Legislation
No=0
Yes=1
Disability Legislation
No=0
Yes=1
% Voting in 2000
Presidential Election
% Christian
% Veterans
% Language Spoken –
Not English

119

Data Analyses
In Chapter Five, univariate (descriptive) statistics for each of the dependent and
independent variables are presented. This allows for an explanation of such statistics such as the
frequency distributions, measures of central tendency, and measures of dispersion – as
appropriate. These analyses allow the reader to understand the nature and range of each variable.
Moreover, it is from these analyses that the proper multivariate statistical methods can be
determined.
Additionally, appropriate bivariate analyses are conducted to examine the relationships
between the independent and dependent variables. The statistical techniques are determined by
the appropriate and respective level of measurement. An effort was made to attempt parametric
analyses. As such, correlations analyses are considered in order to determine the nature of
relationship between each dependent variable and each independent variable.
In order to determine the relationship between numerous independent variables and the
dependent variable, appropriate multivariate statistical methods are used. This is primarily
determined by the nature of the dependent variables and how they are measured. In the case of
this dissertation, the primary model is analyzed using Poisson-based negative binomial
regression due to the large number of zeroes present in the dataset. Given so many zeroes are
present in the FBI’s UCR Hate Crime data, it is necessary to be reflective – and be inclusive of that fact in the main model. Therefore, negative binomial regression is used to analyze the
relationship between the independent variables and the continuous dependent variable. Since so
many counties register zero incidents of bias criminality throughout the years 2000-2009, it is
most appropriate to account for the over-dispersion in the data. As such, this means that the
dependent variable for the main model in this dissertation is created using a continuous measure.

120

Additionally, in order to measure the relationships between the counties that do, in fact,
report hate crimes and the traditional indicators of social disorganization theory; several subanalyses are performed using OLS regression. Given that the proposed dependent variables in
the sub-analyses are continuous in nature (number of bias crimes in United States counties);
multiple regression is an appropriate method for the analysis of the counties that do indeed report
bias crime occurrence. Removing the zeroes allows for a more accurate analysis of the counties
that actually do report bias crimes each year. Given the sample size of this study is 3,141; it is
important to use a technique that operates well with large samples. Therefore, ordinary least
squares (OLS) regression is used as the statistical method of analysis for the identified subanalyses.
Primary Models
As mentioned, four models are included in this dissertation. First, this dissertation
examines the ability of social disorganization theory to predict the likelihood (or risk) of bias
criminality in United States counties. Given this, the following discussion of Poisson-based
negative binomial regression is appropriate for understanding the analyses appropriate when
numerous zeroes are present in the data. Three subanalyses are included as well as they speak to
the counties that do in fact report bias criminality. As such, following discussions of OLS
regression explain the methods used to explore the relationships between traditional indicators of
social disorganization theory and the incidence of bias criminality in United States counties
between 2000 and 2009.
Negative Binomial Regression
In an effort to predict bias criminality in United States counties, negative binomial
regression can be a helpful tool with regard to choosing a multivariate analysis. According to

121

Osgood (2000), Poisson-based regression models are appropriate as they are based on
assumptions about error distributions. This is consistent with the character of count data. Due to
the high number of counties in the United States do not, in fact, report bias crime occurrence –
the result is a significant number of zeroes present in the bias crime data.
From the current dataset, it is unknown whether the counties with zero recorded bias
crimes are a result of failure to report such crimes or the actual lack of bias criminality occurring
in each United States county. However, the lack of zeroes indicates the need for a multivariate
model that can account for presence of this issue. Although logistic regression can be viewed as
an appropriate method for evaluating the relationships present in the main model, logistic
regression can reduce statistical power (Gardner, Mulvey, & Shaw, 1995). Given multiple bias
crime incidents are compressed into one figure in logistic regression, the method may undercount
the total number of bias crimes (Piza, 2012). In order to avoid lowering statistical power and
thereby falsely accepting a null hypothesis, logistic regression is not used in this dissertation.
Poisson-based negative binomial regression is appropriate due the uneven distribution of
the dependent variables (Agresti, 2002). Relevant data justifying the use of negative binomial
regression are presented in Chapter Five. In the case of this dissertation, the overdispersion is
apparent by the presence of numerous zeroes in the data. As such, this dissertation uses negative
binomial regression as a multivariate method to account for the greater variability of the
dependent variable. This allows the model to predict bias crime occurrence with only a few
counts, in nonlinear models (Gardner, Mulvey, & Shaw, 1995). Moreover, negative binomial
regression allows for the analysis of the relationships between the numerous independent
variables and the binary dependent variable present in this study. Thus, it allows for an analysis
of relationships between variables that could not otherwise be analyzed properly using logistic or

122

OLS regression. While logistic regression can undercount rare events, OLS regression assumes
adequate variance in data and requires it to maintain validity (Gardner, Mulvey, & Shaw, 1995).
OLS Regression
In addition, sub-analyses are conducted using OLS regression given the more standard
variability of the involved counties. In essence, given that the appropriate dependent variables
are continuous in nature (number of bias crimes per county), linear regression is appropriate. For
instance, multivariate subanalyses are conducted that include each law enforcement entity does
indeed report bias crime events to the FBI for each respective county. Given this, these research
questions are explored using OLS Regression as the statistical method, given the dependent
variables are continuous in nature. Ordinary Least Squares regression is a frequently used
statistical method for measuring the effects of several independent variables on a dependent
variable concurrently (Schroeder, Sjoquist, & Stephan, 1986). Consequently, multiple regression
is often noted as the dominant paradigm in social science research (Freedman, 1991).
The robust nature of multiple regression makes it a particularly useful and an
advantageous methodological tool for social scientists (Allen, 1997). As noted, there are several
assumptions of multiple regression; however, multiple regression is often used even when one or
more of these assumptions is violated as well (Berry & Feldman, 1985). Moreover, there are
numerous methods sometimes known as diagnostics which are used to detect and correct for the
various violations of multiple regression assumptions. Therefore, the flexibility of multiple
regression allows researchers to use the method in situations where assumptions are violated.
For example, although an assumption of multiple regression is that all variables are interval/ratio,
it is possible to use multiple regression even when one or more of the independent variables is

123

categorical (Allen, 1997). This allows researchers a much more broad selection of independent
variables from which to choose.
Multiple regression is also advantageous in that the analytical method allows researchers
to accomplish more than with simple linear regression. In many cases, social science researchers
are interested in examining the effect that multiple independent variables have on the dependent
variable. By design, multiple regression allows researchers to disentangle the effects of two or
more independent variables on a dependent variable (Allen, 1997). Additionally, multiple
regression allows researchers to determine the accuracy of the relationship with two or more
independent variables and compare it to the accuracy of the relationship with one independent
variable (Allen, 1997).
Although there are many advantages to using multiple regression, there are a number of
researchers that argue that the method has its disadvantages. For instance, Freedman (1991)
suggests that regression models are not always the best way of determining causality since
common violations of assumptions rely on knowledge that the researcher does not have.
However, in the case of this dissertation, the assumptions of multiple regression are not violated.
One concern in many social sciences studies is missing data. In particular, regression models
could be more meaningful if greater attention was paid to an appropriate research question, links
to theory, quality of the data, and the inclusion of appropriate variables (1991). In essence, it is
not necessarily the method of multiple regression that is criticized; but rather the carelessness
with which it is often utilized. Since multiple regression is generally viewed as an extremely
robust and flexible statistical tool, researchers may become less rigorous in their research design.
As a result, some researchers view multiple regression as a method with limited or no ability to

124

provide causal inference (1991). After all, a researcher cannot correct a weak research design
with strong data analysis (Berk, 1991).
Summary
This dissertation seeks to explore the relationship between community-level factors
related to social disorganization theory and bias crime occurrence in American counties. The
data used are provided by the FBI’s Uniform Crime Report, the United States Census Bureau,
Congressional Quarterly’s Voting and Elections Collection, and the Association of Religious
Data Archives. Given the relatively sparse literature on bias crime and American counties, this
dissertation seeks to make a several contributions. Primarily, this study provides insight into the
numerous indicators of social disorganization theory and its ability to inform bias crime
occurrence in United States counties. Secondarily, this dissertation seeks to examine the
differences between the various motivations of bias crime offending at the county level. The
results of the analyses used to answer the research questions and to address the hypotheses are
presented in Chapters Five and Six, and Seven.

125

CHAPTER 5: THE CONTEXT OF BIAS CRIME IN UNITED STATES COUNTIES
The purpose of Chapter Five is to examine the context of bias criminality in United States
counties. In order to understand the full context of bias criminality and its effect on United
States counties, Chapter Five considers the descriptive statistics for contextual factors associated
with bias criminality occurring between 2000 and 2009. First, this chapter presents the
descriptive findings appropriate for the study. The dependent and independent variables are
described to assess normality and the nature of missing data. These analyses present the data
necessary to understand the nature of bias crime occurrence in United States counties.
The Nature of Counties in the United States: Descriptive Findings
In order to understand the context surrounding the nature of bias criminality in American
counties, it is important to consider descriptive findings. Table 5.1 presents the number of
counties within each state that reports at least one bias crime for each year, 2000-2009. In
addition, the number of counties that report zero bias crimes are presented as well. This figure
does include the counties that do not report bias crimes – whether they occur or not - to the FBI.
Table 5.1: Bias Crime Reporting by State
State
Number of Number of Counties
Counties
Reporting at Least
One Hate Crime
Alabama
Alaska
Arizona
Arkansas
California
Colorado
Connecticut
Delaware
District of
Columbia
Florida
Georgia

67
27
15
74
58
64
8
3

14
0
13
67
56
46
8
3

20.90
0.00
86.67
90.54
96.55
71.88
100.00
100.00

Number
Reporting
Zero Hate
Crimes
53
27
2
7
2
18
0
0

1
67
159

1
54
24

100.00
80.60
15.09

0
13
135

126

%

%

79.10
100.00
13.33
9.46
3.45
28.13
0.00
0.00
0.00
19.40
84.91

Table 5.1 (Cont’d)
Hawaii
Idaho
Illinois
Indiana
Iowa
Kansas
Kentucky
Louisiana
Maine
Maryland
Massachusetts
Michigan
Minnesota
Mississippi
Missouri
Montana
Nebraska
Nevada
New
Hampshire
New Jersey
New Mexico
New York
North
Carolina
North Dakota
Ohio
Oklahoma
Oregon
Pennsylvania
Rhode Island
South
Carolina
South Dakota
Tennessee
Texas
Utah
Vermont
Virginia
Washington
West Virginia
Wisconsin
Wyoming
Total

4
44
102
92
99
105
120
64
16
24
14
83
87
82
115
56
93
17

0
31
64
41
42
67
75
32
16
21
14
79
60
12
63
30
34
11

0.00
70.45
62.75
44.57
42.42
63.81
62.50
50.00
100.00
87.50
100.00
95.18
68.97
14.63
54.78
53.57
36.56
64.71

4
13
38
51
57
38
45
32
0
3
0
4
27
70
52
26
59
6

100.00
29.55
37.25
55.43
57.58
36.19
37.50
50.00
0.00
12.50
0.00
4.82
31.03
85.37
45.22
46.43
63.44
35.29

10
21
33
62

10
21
11
53

100.00
100.00
33.33
85.48

0
0
22
9

0.00
0.00
66.67
14.52

100
53
88
77
36
67
5

62
16
71
55
24
51
5

62.00
30.19
80.68
71.43
66.67
76.12
100.00

38
37
17
22
12
16
0

38.00
69.81
19.32
28.57
33.33
23.88
0.00

46
66
95
254
29
14
136
39
55
72
23
3,141

45
26
82
115
11
13
79
35
36
49
14
1,862

97.83
39.39
86.32
45.28
37.93
92.86
58.09
89.74
65.45
68.06
60.87
59.28

1
40
13
139
18
1
57
4
19
23
9
1,279

2.17
60.61
13.68
54.72
62.07
7.14
41.91
10.26
34.55
31.94
39.13
40.72

127

From Table 5.1, it is evident that 59.28% of United States counties do in fact report bias
crimes as having occurred within their boundaries. Certainly, this does not mean that bias crimes
have not occurred in the nearly 40 percent of other counties – it simply indicates that bias crimes
were not reported, either to the police or later, to the FBI. There are numerous states that report
bias crimes as having occurred in 100 percent of counties. Those states are as follows:
Connecticut, Delaware, District of Columbia, Maine, Massachusetts, New Hampshire, New
Jersey and Rhode Island. Other states with the high percentages of counties reporting one or
more bias crimes are California (96.55%), Michigan (95.18%), South Carolina (97.83%), and
Vermont (92.86%).
Of particular importance for this dissertation is the acknowledgement of the high number
of counties that report zero bias crimes. This is relevant as this distribution affects the
effectiveness of the multivariate methods used to assess relationships. As mentioned, numerous
states have a low percentage of counties reporting bias crimes; however, the following states are
worthy of noting due to their particularly low percentages: Alabama (20.90%), Mississippi
(14.63%), North Dakota (30.19%), and New Mexico (33.33%). Taking each of these
percentages into account, it is important to note that overall 40.72 percent of United States
counties do not report any bias crimes as having occurred between 2000 and 2009.

128

Given there are 3,141 counties in the United States, providing descriptions of bias crime
reporting of all counties is rather excessive. Table 5.2 provides a brief survey of the ten most
and least populous counties in the United States. Loving County, Texas reports the smallest
population of any United States county, with 67 persons residing there. Los Angeles County,
California has the largest population, with 9,519,338 persons living in the county. Table 5.3
provides further descriptive statistics about United States counties in 2000.
Table 5.2: Survey of County Populations across the U.S.
County/State
Population (2000)
10 least populous counties*
1. Loving County, Texas
67
2. King County, Texas
356
3. Kenedy County, Texas
414
4. Arthur County, Nebraska
444
5. Petroleum County, Montana
493
6. McPherson County, Nebraska
533
7. San Juan, County, Colorado
558
8. Blaine County, Nebraska
583
9. Loup County, Nebraska
712
10. Thomas County, Nebraska
729
10 most populous counties
1. Los Angeles County, California
9,519,338
2. Cook County, Illinois
5,376,741
3. Harris County, Texas
3,400,578
4. Maricopa County, Arizona
3,072,149
5. Orange County, California
2,846,289
6. San Diego County, California
2,813,833
7. Kings County, New York
2,465,326
8. Miami-Dade County, Florida
2,253,362
9. Queens County, New York
2,229,379
10. Dallas County, Texas
2,218,899
*The USCB reports a population of zero for Broomfield County, CO
and South Boston County, VA.

129

Table 5.3: Descriptive Statistics for U.S. County Population (2000)
U.S. Counties
N (Counties)
3,141.00
Mean
89,593.29
Median
24,595.00
Mode*
10,155.00
SD
292,463.10
Range
9,519,338.00
Minimum
0.00
Maximum
9,519,338.00
*More than one mode; lowest listed.
As mentioned, there are 3,141 counties in the United States. Their populations vary
drastically – with a range of 9,519,338 residents. The mean number of county residents is
89,593.29. Although there are numerous modes, the lowest is 10,155 and the standard deviation
is 292,463.10. Certainly, there is great variation in United States counties’ populations.
In order to present additional useful descriptive statistics with regard to bias crime occurrence in
U.S. counties, Table 5.4 presents the counties with the highest percentages of bias crime
reporting for each state. This allows the name of each county reporting highest percentages of
bias criminality to be identified. Additionally, the number of bias crimes per capita in each of
the 50 counties is presented in order to account for each the variation in each county’s
population.

130

Table 5.4: Characteristics of United States Counties (2000-2009)
State
Number of
County with Most Number Population Bias
Counties
Bias Crime
of Bias (2000)
Crimes
Crimes
Per
Capita
(100,000)
Alabama
67
Jefferson
13
66,2047
1.96
Alaska*
27
----Arizona
15
Maricopa
1,679 3,072,149
54.65
Arkansas
74
Pulaski
61
361,474
16.88
California
58
Los Angeles
5,936 9,519,338
62.36
Colorado
64
Arapahoe
208
487,967
42.63
Connecticut
8
Hartford
390
857,183
45.50
Delaware
3
New Castle
216
500,265
43.18
District of
1
Washington
335
572,059
58.56
Columbia
Florida
67
Broward
314 1,623,018
19.35
Georgia
159
Fulton
114
816,006
13.97
Hawaii**
4
----Idaho
44
Ada
95
300,904
31.57
Illinois
102
Cook
768 5,376,741
14.28
Indiana
92
Monroe
184
120,563
152.62
Iowa
99
Scott
57
158,668
35.92
Kansas
105
Sedgwick
301
452,869
66.47
Kentucky
120
Fayette
94
260,512
36.08
Louisiana
64
Evangeline
66
35,434
186.26
Maine
16
Cumberland
202
265,612
76.05
Maryland
24
Baltimore
519
754,292
68.81
Massachusetts
14
Suffolk
1,754
689,807
254.27
Michigan
83
Wayne
921 2,061,162
44.68
Minnesota
87
Hennepin
751 1,116,200
67.28
Mississippi
82
De Soto
4
107,199
3.73
Missouri
115
Jackson
225
654,880
34.36
Montana
56
Missoula
57
95,802
59.50
Nebraska
93
Lancaster
218
250,291
87.10
Nevada
17
Clark
714 1,375,765
51.90
New
10
Rockingham
102
277,359
36.80
Hampshire
New Jersey
21
Monmouth
1,230
615,301
199.90
New Mexico
33
Bernalillo
112
55,6678
20.12
New York
62
Suffolk
997 1,419,369
70.24
North
100
Mecklenburg
111
695,454
15.96
Carolina
North Dakota
53
Morton
40
25,303
158.08
Ohio
88
Franklin
845 1,068,978
79.05
Oklahoma
77
Oklahoma
113
660,448
17.11
131

Table 5.4 (Cont’d)
Oregon
36
Multnomah
581
660,486
88.00
Pennsylvania
67
Philadelphia
326 1,517,550
21.48
Rhode Island
5
Providence
277
621,602
44.56
South
46
Richland
69
320,677
21.52
Carolina
South Dakota
66
Minnehaha
124
148,281
83.63
Tennessee
95
Shelby
397
897,472
44.24
Texas
254
Harris
633 3,400,578
18.61
Utah
29
Salt Lake
342
898,387
38.07
Vermont
14
Chittenden
102
146,571
69.59
Virginia
136
Fairfax
192
969,749
19.80
Washington
39
King
714 1,737,034
41.10
West Virginia
55
Cabell
106
96,784
109.52
Wisconsin
72
Milwaukee
163
940,164
17.34
Wyoming
23
Natrona
19
66,533
28.58
Total
3,141
-23,791 50,288,965
-*Alaska reports bias crime by borough; not compatible with county-based data.
**Hawaii does not report bias crime to the FBI.
In Table 5.4, several characteristics of United States counties are presented. Due to the
fact that there are 3,141 counties across the country, this table lists the county and its population
in each state with the greatest number of bias crimes occurring. The top five most populous
counties with the greatest number of bias crimes occurring between 2000 and 2009 are as
follows: Los Angeles County, California (9,519,338), Cook County, Illinois (5,376,741), Harris
County, Texas (3,400,578), Maricopa County, Arizona (3,072,149), and Wayne County,
Michigan (2,061,162). In addition, to account for variation in county population, the bias crimes
per capita are listed as well. Taking population into consideration changes the counties with the
highest rate of bias crimes occurring. The county with the highest incidences of bias criminality
per capita is Suffolk, Massachusetts with 254.27 bias crimes occurring per 100,000 persons. The
four other highest figures of bias criminality per capita are as follows: Evangeline County,
Louisiana (186.26); Monmouth, New Jersey (199.90); Morton, North Dakota (158.08); and
Monroe County, Indiana (152.62).

132

Description of Independent Variables
This dissertation presents the relevant descriptive statistics for each independent variable
as well. Predictor variables used to measure social disorganization theory are grouped
conceptually - by economic deprivation, social heterogeneity, social cohesion, and residential
mobility. In addition, the relevant descriptive statistics for all control variables are presented.
Economic Deprivation
Appendix A presents descriptive statistics for two measures of economic deprivation.
First, descriptions of the employment status of residents of U.S. states are displayed.
In addition, this table includes figures for the population and the population for individuals age
16 and older. This is presented as it is necessary in order to calculate the percent of unemployed
people in each county. Using the total population would provide misleading employment
figures. Second, Appendix A also displays the number and percent of individuals living below
the poverty line in each county. The United States Census Bureau measures poverty with a
weighted threshold that, in 2000, ranged from $8,794 for one person household to $35,060 for a
nine person household (USCB, 2000). Table 5.5 presents appropriate measures of central
tendency and dispersion for percent unemployed and percent living below the poverty line.
Appendix A presents descriptive findings for economic measures of United States
counties in 2000. With a total population of 281,412,514, the United States has a vast range of
residents living in each state. Wyoming is the least populous state with a total population of
493,782. California has the greatest population in the United States. In 2000, 33,871,648
individuals were residing in the state. In 2000, the percentage of unemployed Americans was a
rather low 3.6%. The following five states had the lowest percentages of unemployed persons:
Iowa (2.85%), Kansas (2.84%), Nebraska (2.45%), New Hampshire (2.65%), and Virginia

133

(2.73%). The five states with the highest percentages of unemployed persons were Alaska
(6.10%), the District of Columbia (6.79%), California (4.34%), Mississippi (4.34%), and New
Mexico (4.41%).
Moreover, in 2000 nearly 34 million (33,898,201) individuals lived below the poverty
line in the United States. This amounts to 12.05 percent of the country’s population living in
poverty. Of course, this is poverty only as defined by the government – so there are likely many
others that live just above the U.S. Census Bureau’s poverty line threshold. Five states with the
lowest percentage of residents living below the poverty line are as follows: New Hampshire
(6.35%), Minnesota (7.73%), Wisconsin (8.42%), New Jersey (8.32%), and Maryland (8.28%).
In addition, the five states reporting the highest percentage of individuals living below the
poverty line were: West Virginia (17.46%), New Mexico (18.08%), Mississippi (19.27%),
Louisiana (19.04%), and the District of Columbia (19.14%).
Table 5.5: Descriptive Statistics for Measures of Economic Deprivation (2000)
Number Unemployed
Number Living below Poverty
Line
N
3,139.00
3,139.00
(Counties)
Missing
2.00
2.00
Mean
2,531.71
10,799.04
Median
659.00
3,243.00
Mode*
23.00
713.00
SD
9,862.43
43,982.13
Range
354,347
1,674,599
Minimum
0.00
0.00
Maximum
354,347
1,674,599
*More than one mode; lowest listed
Table 5.5 describes the two independent variables representing economic deprivation: the
number of unemployed persons in each county and the number of persons living below the
poverty level. These data were available for 3,139 counties; they were missing for two. The
values range from 0 to 357,347 and mean number of unemployed persons in each American
134

county is 2,531.71. The median is 659 and the mode is 23. The standard deviation is 9,862.43,
indicating that there is a significant level of dispersion from the mean. The mean of numbers of
persons living below the poverty line is 10,799.04 for United States counties. The values range
from 0 to 1,674,599 and the median is 3,243. The mode is 713 and the standard deviation is
43,982.13, indicating great variation from the mean.
Social Heterogeneity
Appendix B displays the descriptive characteristics for several measures of social
heterogeneity, or diversity. In order to understand how social heterogeneity affects the
occurrence of bias criminality, several measures are examined. First, the number of non-white
residents in each county is considered for analysis. This variable allows for the representation of
racial and ethnic diversity. Second, the number of naturalized and non-naturalized foreign-born
persons residing in each United States county introduces yet another measurement of a
community’s heterogeneity. Third, as indicators of religious diversity, the number of Jewish and
the number of Muslim persons residing in each county are included in the analyses of this
dissertation.
Appendix B presents the descriptive statistics for all measures of social heterogeneity.
Nearly 25 percent (24.86) of the United States is populated by non-white persons. The following
states have the highest percentages of non-white residents: California (40.45%), the District of
Columbia (69.22%), Louisiana (36.09%), Maryland (35.97%), and Mississippi (38.62%). The
states with the lowest percentages of non-white residents are Iowa (6.07%), Maine (3.05%), New
Hampshire (3.96%), Vermont (3.22%), and West Virginia (4.95%).
Approximately ten percent (9.91%) of United States residents identify as being foreignborn. This figure represents all foreign-born persons, both naturalized and non-naturalized. The

135

five states with the highest percentages of foreign-born persons are: California (23.60%), Florida
(15.28%), Hawaii (16.22%), New Jersey (15.94%), and New York (18.64%). In addition, the
states that report the lowest percentages of foreign-born persons are as follows: Mississippi
(1.26%), Montana (1.63%), North Dakota (1.60%), South Dakota (1.48%), and West Virginia
(0.98%).
Across the United States, only 2.18% of residents are Jewish. The states with the highest
populations of Jewish persons are all on the east coast and are the District of Columbia (4.46%),
Maryland (4.08%), Massachusetts (4.33%), New Jersey (5.56%), and New York (8.72%). States
with the lowest percentage of Jewish persons residing within their borders are: Arkansas
(0.06%), Idaho (0.08%), Mississippi (0.05%), South Dakota (0.05%), and Montana and
Wyoming (both with 0.09%).
With even fewer adherents in the United States than Judaism is Islam, comprising only
0.55% of the population. Certainly, worthy of discussion is the District of Columbia, which is an
extreme outlier. Approximately ten percent (10.57%) of residents in the District of Columbia
identify as Muslim. This percentage is far higher than the next states with the highest
percentages of Muslims residing there. Those states are as follows: Illinois (1.01%), Maryland
(1.00%), New Jersey (1.43%), and New York (1.18%). No extreme outliers exist with regard to
states with the lowest percentages of Muslim residents. However, the following states report the
lowest percentages of those practicing Islam in the United States: Hawaii (0.05%), Idaho
(0.03%), South Dakota (0.01%), Vermont (0.02%), and Wyoming (0.05%). Table 5.6 presents
additional descriptive statistics for measures of social heterogeneity.

136

Table 5.6: Descriptive Statistics for Measures of Social Heterogeneity (2000)
Non-White
Foreign-Born
Jewish Residents
Muslim
Residents
Residents
Residents
N
3,139
3,138
3,138
3,138
(Counties)
Missing
2.00
3.00
3.00
3.00
Mean
22,287.59
8,891.47
1,957.08
496.91
Median
2,514.00
360.50
0.00
0.00
Mode*
19.00
23.00
0.00
0.00
SD
126,261.45
74,089.18
17,478.39
3,832.42
Range
4,882,274.00
3,140,052.00
564,700
95,623
Minimum
2.00
0.00
0.00
0.00
Maximum
4,882,276.00
3,140,052.00
564,700
95623
* More than one mode; lowest listed
Table 5.6 describes the four independent variables representing social heterogeneity: the
number non-white, foreign-born, Jewish, and Muslim residents in each United States county.
For the measure of non-white residents, these data were available for 3,139 counties and they
were missing for two. The values range from 0 to 4,882,276 and the mean number of non-white
persons in each American county is 22,287. The median is 2,514 and the mode is 19. The
standard deviation is 126,261.45, indicating that there is a substantial level of dispersion from the
mean. For the measure of foreign-born residents, the mean in United States counties is 8,891.47
and the values range from 0 to 3,140,052. Data are available for 3,138 counties; three counties
have missing values. The median is 360.50 and the mode is 23. In addition, the standard
deviation is 74,089.18, indicating great variation from the mean. Moreover, the independent
variable representing the number of Jewish residents in each American county has a mean of
1,957.08. Data are available for all but three United States counties. The median and mode are
both 0. The data have a range of 0 to 564,700 and a standard deviation of 17,478.39. Similarly,
the mean number of Muslim residents in American counties is particularly low at 496.91. The
data are available for 3,138 counties and the values range from 0 to 95,623. Similar to the

137

measure of the Jewish population, the median and the mode are both 0. The standard deviation
is 3,832.42.
Social Cohesion
Appendix C presents relevant descriptive statistics for the measures of social cohesion in
United States counties. The number of divorced persons and the number of index offenses in
each county are presented. In addition, this table includes the crime rate (based on the FBI’s
reporting of index offenses in 2000).
From the table in Appendix C, several findings emerged. First, the mean percentage of
individuals who are divorced in the United States is 7.66 percent. The top five states with the
highest percentages of divorced persons are as follows: Maine (9.25%), Wyoming (9.21%),
Nevada (10.81%), Oregon (9.26%), and Florida (9.40%). The five states with the lowest
percentages are New Jersey (5.95%), Utah (5.92%), North Dakota (6.20%), New York (6.18%),
and Massachusetts (6.66%).
With regard to each states crime rate per capita, five states emerged as having the highest
level of reported index crimes per capita. Interestingly, Montana had the highest crime rate per
100,000 persons, with a figure of 7,040.94. Other states with high rates of crime per capita are:
New Mexico (6,262.68), Arizona (5,854.25), Florida (5,625.88), and South Carolina (5,823.29).
In contrast, New Hampshire reported the lowest crime rate, with 1,473.34 index crimes occurring
per 100,000 persons residing in the state. Additional states reporting low levels of crime per
capita are: Kentucky (1,473.34), Rhode Island (1,956.47), Illinois (2,024.97), and Maine
(2,290.88). Table 5.7 presents additional descriptive statistics appropriate for continuous
independent variables.

138

Table 5.7: Descriptive Statistics for Measures of Social Cohesion (2000)
Number of
Number of Index Offenses
Divorced
Persons
N (Counties)
3,139.00
3,141.00
Missing
2.00
0.00
Mean
6,868.28
3,471.96
Median
1,882.00
477.00
Mode*
958.00
00.00
SD
21,201.16
13,595.26
Range
613,661.00
383,722.00
Minimum
6.00
0.00
Maximum
613,667.00
383,722.00
* More than one mode; lowest listed
Table 5.7 displays descriptive statistics for two measures of social cohesion used in this
dissertation. In terms of the divorced persons in each county, 3,139 of the counties were
included in the analysis. Two counties had missing data. The mean number of divorced persons
across the United States was 6,868.28. The median was 1,882 and the mode was 958. The data
ranged from values of six to 613,667 and the standard deviation was 21,201.16. Additionally,
the mean number of index offenses occurring in United States counties was 3,471.96. The
median was 477 and the mode was zero. The values ranged from zero to 383,722 and had a
standard deviation of 13,595.26, indicating great variation.
Residential Mobility
Appendix D provides descriptive statistics necessary for understanding the nature of
residential mobility measures in 2000. In this dissertation, the number of owner occupied
housing units is used to represent the number of homeowners in each county. Moreover, the
number of unoccupied housing units represents abandoned houses. Both measures represent
residential mobility in United States counties.
According to the preceding tables, as of 2000, there were 115,899,421 total housing units
in the United States. Approximately 60% (60.24%) of those housing units were occupied by the
139

homeowner. This figure varied slightly across the country, with the District of Columbia
reporting the lowest total number of owner occupied housing units. As of 2000, only 36.83% of
the housing units were occupied by the homeowner. Other states with low percentages of owner
occupied housing units were as follows: Hawaii (49.50%), Nevada (55.26%), Rhode Island
(55.74%), and Alaska (53.07%). The five states with the highest percentages of owner occupied
housing units are: Minnesota (68.39%), Iowa (67.46%), Michigan (65.96%), Indiana (65.91%),
and West Virginia (65.56%).
With regard to the percentage of unoccupied housing units in each state in 2000, the
mean value was nearly 9% (8.99%). Maine reported the highest percentage of unoccupied
housing, with 20.51% of total housing units remaining unoccupied. Several other states reported
rather high percentages as well. Those states are: Vermont (18.26%), Alaska (15.09%),
Wyoming (13.51%), and New Hampshire (13.24%). In contrast, California reports the lowest
percentage of unoccupied housing units in 2000, with a figure of 5.83%. Four additional states
with figures well below the mean are: Massachusetts (6.80%), Iowa (6.75%), Connecticut
(6.08%), and Illinois (6.01%). Table 5.8 provides additional descriptive statistics for the
measure of residential mobility.
Table 5.8: Descriptive Statistics for Measures of Residential Mobility (2000)
Number of
Number of
Owner Occupied Housing Units
Unoccupied Housing Units
N (Counties)
3,141.00
3,141.00
Missing
0.00
0.00
Mean
22,226.24
3,318.44
Median
6,995.00
1,320.00
Mode*
948.00
690.00
SD
59,472.87
7,606.58
Range
1,499,744.00
137,135.00
Minimum
0.00
0.00
Maximum
1,499,744
137,135.00
* More than one mode; lowest listed

140

As presented in Table 5.8, several additional descriptive statistics for two measures of
residential mobility are used in this dissertation. With regard to the number of owner occupied
housing units in United States counties, the data included values for all 3,141 counties. There
were no missing values. However, the mean value was 22,226.24, the median was 6,995, and
the mode was 948. The figures ranged from zero to 1,499,744 and the standard deviation was
59,472.87. Similarly, data for all counties were available for the number of unoccupied housing
units. For this independent variable, the mean was 3,318.44 and the median was 1,320. The
mode was 690 and the values ranged from zero to 137,135. Showing marked variability, the
standard deviation was 7,606.58.
Table 5.9: Context of Bias Crime Laws in the United States (2000)
State
Any
Protects
Protects
Protects Protects
Law
Race,
Sexual
Gender Gender
Ethnicity, Orientation
Identity
Religion
(RRE)
Alabama
Yes
Yes
No
No
No
Alaska
Yes
Yes
No
Yes
No
Arizona
Yes
Yes
Yes
Yes
No
Arkansas
No
No
No
No
No
California
Yes
Yes
Yes
Yes
Yes
Colorado
Yes
Yes
Yes
No
Yes
Connecticut
Yes
Yes
Yes
Yes
Yes
Delaware
Yes
Yes
No
No
No
District of
Yes
Yes
Yes
Yes
Yes
Columbia
Florida
Yes
Yes
Yes
No
No
Georgia*
Yes
No
No
No
No
Hawaii
Yes
Yes
Yes
Yes
Yes
Idaho
Yes
Yes
No
No
No
Illinois
Yes
Yes
Yes
Yes
No
Indiana
No
No
No
No
No
Iowa
Yes
Yes
Yes
Yes
No
Kansas
Yes
Yes
Yes
No
No
Kentucky
Yes
Yes
Yes
No
No
Louisiana
Yes
Yes
Yes
Yes
No
Maine
Yes
Yes
Yes
Yes
No
Maryland
Yes
Yes
No
No
No
141

Protects
Disability

Yes
Yes
Yes
No
Yes
Yes
Yes
Yes
Yes
Yes
No
Yes
No
Yes
No
Yes
Yes
Yes
Yes
Yes
No

Table 5.9 (Cont’d)
Massachusetts Yes
Yes
Yes
Michigan
Yes
Yes
No
Minnesota
Yes
Yes
Yes
Mississippi
Yes
Yes
No
Missouri
Yes
Yes
Yes
Montana
Yes
Yes
No
Nebraska
Yes
Yes
Yes
Nevada
Yes
Yes
Yes
New
Yes
Yes
Yes
Hampshire
New Jersey
Yes
Yes
Yes
New Mexico Yes
Yes
Yes
New York
Yes
Yes
Yes
North
Yes
Yes
No
Carolina
North Dakota Yes
Yes
No
Ohio
Yes
Yes
No
Oklahoma
Yes
Yes
No
Oregon
Yes
Yes
Yes
Pennsylvania Yes
Yes
No
Rhode Island Yes
Yes
Yes
South
No
No
No
Carolina
South Dakota Yes
Yes
No
Tennessee
Yes
Yes
Yes
Texas
Yes
Yes
Yes
Utah
No
No
No
Vermont
Yes
Yes
Yes
Virginia
Yes
Yes
No
Washington
Yes
Yes
Yes
West Virginia Yes
Yes
No
Wisconsin
Yes
Yes
Yes
Wyoming
No
No
No
Total ‘Yes’
45
44
28
*Hate crime law struck down by U.S. Supreme Court

Yes
Yes
Yes
Yes
Yes
No
Yes
No
Yes

No
No
Yes
No
Yes
No
No
No
No

Yes
Yes
Yes
No
Yes
No
Yes
Yes
Yes

Yes
Yes
Yes
Yes

Yes
Yes
No
No

Yes
Yes
Yes
No

Yes
No
No
No
No
Yes
No

No
No
No
Yes
No
No
No

No
No
Yes
No
No
Yes
No

No
Yes
Yes
No
Yes
No
Yes
Yes
No
No
28

No
No
No
No
Yes
No
Yes
No
No
No
12

No
Yes
Yes
No
Yes
No
Yes
No
Yes
No
38

From Table 5.9, it is evident that the majority of states did, in fact, have hate crime laws
in effect as of 2000. Only five states deemed such protections from bias criminality unnecessary.
They are as follows: Wyoming, Arkansas, Indiana, South Carolina, and Utah. In contrast,
Washington, Hawaii, Vermont, New Jersey, New Mexico, Missouri, Minnesota, the District of

142

Columbia, California, and Connecticut provide protection from bias criminality for all the
identified categories. Georgia did pass a law protecting its citizens from bias crimes, only to
have it struck down in 2000 by the United States Supreme Court on account of its vague nature
and lack of specification for protected groups (Henry, 2009).
Figure 1: Percentage of States with Hate Crime Laws
100
90
80
70
60
50
Number

40

Percent

30
20

10
0

Any Hate Protects
Crime Law RRE

Protects Protects
Sexual
Gender
Orientation

Protects Protects
Gender Disability
Identity

As presented in Figure 1, 90% of states have a hate crime law of some variety. Although
there is quite a bit of variation in the identities that are protected by state hate crime statutes, the
majority (88%) do protect their citizens from bias crimes motivated by animus toward the
victim’s race, religion, and ethnicity. Bias crimes committed against persons with disabilities are
also addressed by law in 76% of states. In addition, 56% of states protect victims based on
gender and sexual orientation. Only a minority of states (24%) list statutory protections for those
who are victimized based on their actual or perceived gender identity.

143

Control Variables
Appendix E and Table 5.10 present the descriptive statistics for the control variables used
in this dissertation. The number of non-English speaking persons, the median age, the number of
veterans, and the number of persons with a bachelor’s degree are used to ensure the models are
measuring the concepts of social disorganization theory properly. In addition, the number of
Christians residing in each county and the number of persons participating in the 2000
presidential election are included. With the exception of median age, the overall number and
percent are available for each control variable. Moreover, additional descriptive statistics such
as measures of central tendency and measures of dispersion are analyzed and presented.

144

Table 5.10: Additional Descriptive Statistics for Control Variables (2000)
Number
Number of
Number with a
Non-English
Veterans
Bachelor’s Degree
Speaking
N (Counties)
3,141.00
3,139.00
3,139.00
Missing
0.00
2.00
2.00
Mean
14,947.84
8,411.12
9,021.17
Median
1,094.00
2,460.00
1,432.00
Mode*
116.00*
919.00
315.00*
SD
111,340.62
21,706.02
33,045.30
Range
4,758,482.00
510,707
945,634.00
Minimum
0.00
5.00
0.00
Maximum
4,758,482.00
510,712.00
945,634.00
* More than one mode; lowest listed

145

Number of
Christian Residents
3,138.00
3.00
45,050.05
12,432.50
3,280.00*
158,426.51
5,528,796.00
18.00
5,528,814.00

Number
Participating in 2000
Presidential Election
3,035.00
106.00
33,393.99
9,307.00
1,519.00*
97,515.46
2,695,017.00
137.00
2,695,154.00

Appendix E and Table 5.10 display appropriate descriptive statistics for relevant control
variables used in this dissertation. For each of the control variables, the overwhelming majority
of counties are represented. With regard to the number of persons voting in the 2000 presidential
election, 3,035 counties are represented with data; 106 counties are missing.
For the number of Christian residents, data are missing for three counties. For the number of
veterans and the number of residents with a bachelor’s degree, only two counties each are
missing. Data are available for all 3,141 counties with regard to the number of non-English
speaking residents.
The mean number of non-English speaking residents in United States counties is
14,947.84. This accounts for the high range in values, ranging from zero to 4,758,482. The
median figure for number of non-English speaking persons is 1,094 and the mode is 116. In
addition, the standard deviation is 111,340.62.
The nature of the number of veterans ranges from five to 510,712. The mean for this
control variable was 8,411.12 and the median was 2,460. The mode was 919 veterans per United
States county in 2000. The standard deviation was rather large, with a value of 21,706. The
range of persons holding bachelor’s degrees across United States counties varies from zero to
945,634 with a mean of 9,021.17. In addition, the median figure for those having earned a
bachelor’s degree is 1,432 and the mode is 315. There is great variation across counties as the
standard deviation is 33,045.30.
In terms of adherents to Christianity, the mean is 45,050.05. The median is 12,432 and
the mode is 3,280. The figures range from 18 to 5,528,814 and the standard deviation is
158,426.51. With regard to political participation in the 2000 presidential election, the mean is
33,393. The median value is 9,307 and the mode is 1,519. The figures ranged from 137 persons

146

voting in the election to 2,695,154 persons participating in the election. The standard deviation
is 97,515.46.
In Appendix F, the number of anti-white bias crimes in United States counties is
displayed. Although the main models examine differences between groups, it is important to
examine and note the anti-white bias crimes that occurred in 2000. The results are located in
Table 5.11. In the main models, anti-white bias crime incidents are excluded from the dependent
variable named “anti-race bias crime.” This is done because this dissertation seeks to examine
bias criminality that has occurred against subordinate social groups – racial, religious, and sexual
minorities. Although anti-white bias crimes are certainly worthy of investigation, they comprise
10.10% of all bias crimes and are not viewed as violent representations of oppression. After all,
in order to oppress a social group must possess the power to maintain structural inequality – an
advantage white Americans have and persons of color do not (Bell, 2013). The five states that
report the highest number of anti-white bias crimes are as follows: Michigan (20.96%), Virginia
(25.51%), South Carolina (29.41%), Tennessee (37.84%), and Florida (16.67%). The following
tables provide additional descriptive statistics for anti-white bias crime.
Table 5.11: Descriptive Statistics for Measures of Anti-White Bias Occurrence (2000)
Number of Anti-Race Bias
Number of Anti-White Bias Crimes
Crimes
N (Counties)
737.00
738.00
Missing
2,401.00
2,403.00
Mean
1.04
6.67
Median
0.00
2.00
Mode
0.00
1.00
SD
3.35
24.10
Range
57.00
555.00
Minimum
0.00
0.00
Maximum
57.00
555.00

147

Table 5.11 compares the measures of central tendency and the measures of dispersion for
both the dependent variable (number of anti-race bias crimes in 2000) and the control variable
(number of anti-white bias crimes in 2000). Most important is the acknowledgement that only
738 counties in the United States reported a bias crime as having occurred in 2000. The
overwhelming majority of counties (2,403) did not report any bias crimes as having happened
within their borders. The mean is 6.67, the median is two, and the mode is one. Certainly, antiwhite bias crimes are not the most frequently occurring bias crime. The range was zero to 555
anti-white bias crimes per county and the standard deviation was a 24.10.
Summary
The purpose of Chapter Five was to examine the nature and context of United States
counties and the bias crimes that have occurred within them from 2000 to 2009. This allows for
a description of that data prior to the answering of the research questions. As such, appropriate
descriptive statistics were displayed for the dependent and independent variables used in this
dissertation. More specifically, the descriptive statistics for the measures of social
disorganization theory are presented. Chapter Six discusses the appropriate bivariate analyses
for this dissertation.

148

CHAPTER 6: CONSIDERING BIAS CRIME OCCURRENCE IN UNITED STATES
COUNTIES
The purpose of Chapter Six is to present the results of bivariate analyses used to support
the investigation of the stated research questions in this dissertation. Numerous statistical
methods are used to evaluate the existence of bivariate relationships; each is appropriate for the
type of variables examined. These selection of these statistical techniques are informed by the
nature of the dependent and independent variables. The bivariate analyses are organized and
presented according to each of the main models in this dissertation. First, the nature of the
relationships between the categorical dependent variable and each of the independent variables is
presented and discussed. Second, the appropriate bivariate analyses for the subsequent models
are presented and discussed.
Bivariate Analyses
Model One
Table 6.1 displays the results of chi-square tests for independence between the
categorical dependent variable and the predictor variables for Model One. In this case, the
dependent variable is a continuous measure (0 = no bias crimes occurring; 1 = at least one bias
crime occurring) of bias criminality across United States counties. Chi-square analyses are used
given the data are non-parametric.

149

Table 6.1: Model One Chi-Square Tests for Independence (Anti-race)
Predictor
N
χ²
p
ɸ
Cramer’s V
Population in 2000
3,141 664.394 .000
-.460
Any hate crime law (HCL)
3,141 7.555* .006
.050
-in 2000
HCL RRE**
3,141 26.356* .000 -.093
-HCL sexual orientation
3,141 27.249* .000 -.094
-HCL gender
3,141
.001* .981
.001
-HCL gender identity
3,141 28.567* .000 -.096
-HCL disability
3,141 25.335* .000 -.090
-*Yates’s Continuity Correction
**Race, Religion, Ethnicity

df
2
1
1
1
1
1
1

From Table 6.1, numerous findings are apparent. A chi-square test for independence
indicated a significant relationship between United States counties’ populations and the
occurrence of bias crime, χ² (2, N-3,141) = 664.294, p=.000. A Cramer’s V value of .460
indicated a moderately strong association between the two measures. Additional chi-square tests
for independence were conducted for each state’s hate crime laws and their relationships to the
occurrence of bias criminality in United States counties.
First, the relationship between the occurrence of bias criminality and the presence of any
hate crime law in each state was explored. Since the tables were 2 by 2 for the remaining
variables, Yates’s correction for continuity was used to account for any overestimations in the
chi-square analyses. As such, the Yates score was 7.555 and was significant at the .05 level
(.006). In addition, the phi value was .050, indicating a weak association between the variables.
Second, the types of legislation were separated by the groups that were protected. The
relationship between the occurrence of bias criminality and states that have laws protecting
against racially, religiously, and ethnically motivated bias crime was significant (.000). The
Yates’s chi-square value was 26.356 and the phi was -.093, indicating a dependent, yet weak
association. Similarly, the relationship between the occurrence of bias criminality and states that
have laws protecting against anti-sexual orientation-motivated bias crime was also significant at
150

the .05 level (.000). The Yates’s chi-square was 27.249 with a phi of -.094. In contrast, the
Yates’s value for laws protecting gender was .001 and was not significant (.981). The phi was
.001. Given these findings, it was evident that the occurrence of bias criminality in United States
counties and the presence of hate crime laws protecting gender were independent of one another.
Moreover, the Yates’s score for laws protecting gender identity was 28.567 and was significant,
with a value of .000. The reported phi was -.096. The relationship between the occurrence of
bias crime and laws protecting disability had a Yates’s score of 25.335 and was significant at the
.05 level. The phi was -.090, which indicated a weak association between the two variables.
The following tables present the findings from several simple logistic regression
analyses. Simple logistic regression was used as the data are non-parametric in nature. The
tables are organized by traditional indicators of social disorganization theory. A separate table
presents finding for simple logistic regression findings for the control variables. All analyses
were bivariate in nature and conducted with the dependent variable used in Model One, a
dichotomous indicator of bias crime occurrence in United States counties.
Table 6.2: Logistic Regression on Measures of Low Economic Status
Predictor
β
Wald χ²
p
Exp(B)* Cox & Snell
R²
Percent unemployed
.027
.797 .372
1.028
.000
Percent below poverty
-.081
161.172 .000
.922
.056
line
Percent without a
-.166
145.677 .000
.847
.053
Bachelor’s
degree
*Confidence Interval (95.0%)

Nagelkerke
R²
.000
.075
.072

Table 6.2 displays the results of three simple logistic regression analyses. Each model
included one independent variable and one dependent variable. Any relationship between the
percentage of unemployed persons in United States counties and the presence of bias crime was
not significant. However, the logistic regression analysis did indicate a significant relationship
151

between the percentage of persons living below the poverty line in each county and bias crime
occurrence. The model explained between 5.6% and 7.5% of the variance in the dependent
variable. The B value is -.081, which indicated a negative relationship between the variables.
The odds ratio for the percentage of persons living below the poverty line in United States
counties was .992. This suggested that for each additional one unit increase in unemployment,
United States counties was .922 times less likely to experience the occurrence of bias criminality.
In addition, the relationship between the percentage of persons without a bachelor’s degree
residing in each county and the presence of bias criminality was significant at the .05 level. The
model explained between 5.3 percent and 7.2 percent of the variance in the dependent variable.
The B value was -.166. This indicated that for every one percent increase in persons without
Bachelor’s degrees residing in each county, bias criminality was .847% less likely to occur.
Table 6.3: Logistic Regression on Measures of Social Heterogeneity
Predictor
β
Wald
p
Exp(B)*
Cox & Snell
χ²
R²
Percent non-White
-.012 29.929 .000
.988
.010
residents
Percent foreign-born
.078 49.394 .000
1.082
.019
Percent Jewish
.339 27.853 .000
1.404
.015
Percent Muslim
1.041
31.07 .000
2.833
.015
*Confidence Interval (95.0%)

Nagelkerke
R²
.013
.026
.021
.020

In Table 6.3, the results for simple logistic regression analyses on four measures of social
heterogeneity are presented. The logistic regression analysis did indicate a significant
relationship between the percentage of non-White persons residing in a county and the existence
of bias criminality. The model explained between 1.0% and 1.3% of the variance in the
dependent variable. The B-value was -.012, indicating the presence of a negative relationship.
Therefore, the odds ratio suggested that for every one percent increase in non-White residents,
the likelihood of bias criminality occurring decreased by .988%.

152

In addition, the relationships between the remaining three measures of social
heterogeneity were all significant and the .05 level. First, the percentage of foreign-born persons
had a positive relationship the occurrence of bias criminality in United States counties, with a B
value of .078. The model explained between 1.9% and 2.6% of the variance in the dependent
variable. The odds ratio was 1.082, indicating that for every one percent increase in the
population of foreign-born persons, the likelihood of bias criminality occurring increased by
1.082%. Second, the relationship between the percentage of Jewish residents and the occurrence
of bias criminality was yet stronger, with a B value of .339. This model explained between 1.5%
and 2.1% of the variance in the dependent variable. As such, for every one percentage increase
in the Jewish population in United States counties, the likelihood of bias criminality occurring
increased by 1.404%. Third, the strongest relationship existed with the percentage of Muslim
residents in each county. The B value was 1.041, indicating a positive relationship.
Additionally, the model explained 1.5% to 2.0% of the variance in the dependent variable. An
odds ratio of 2.833 indicated that for every one percent increase in Muslim residents that existed
in United States counties, the likelihood of bias crime occurring increased by 2.833%.
Table 6.4: Logistic Regression on Measures of Social Cohesion
Predictor
β
Wald
p
Exp(B)* Cox & Snell
χ²
R²
Percent divorced
.207 75.618 .000
1.230
.025
Crime rate
.000 10.531 .001
1.000
.006
Percent voting in 2000
-.012
6.142 .013
.988
.002
Presidential election
*Confidence Interval (95.0%)

Nagelkerke
R²
.034
.009
.003

Table 6.4 presents the findings of two simple logistic regression analyses representing
measures of social cohesion. The percentage of divorced persons residing in United States
counties had a positive relationship with the occurrence of bias criminality. The B value was
.207. This model explained between 2.5% and 3.4% of variance in the dependent variable. The
153

odds ratio was 1.230, indicating that with every 1% increase in divorced persons in United States
counties, the likelihood of bias crimes occurring increased by 1.230%. In addition, the crime
rate had a positive, yet weak, relationship with the occurrence of bias criminality in United States
counties. The B value was .000 and the model explained between .6% and .9% of the variance in
the dependent variable. The odds ratio indicated that as the crime rate increased by one unit in
United States counties, the likelihood of the occurrence of bias criminality increased by 1.00%.
The percentage of county residents voting in the 2000 Presidential election was statistically
significant. With a B value of -.012, the relationship was negative and weak. However, the
model did predict between .2% and .3% of the variance in the dependent variable. With every
one percent increase in voter participation, the likelihood of bias criminality occurring was
reduced by .988%.
Table 6.5: Logistic Regression on Residential Mobility
Predictor
β
Wald
p
Exp(B)* Cox & Snell Nagelkerke
χ²
R²
R²
Percent owner-occupied
.011
7.384 .007
1.011
.002
.003
housing units
Percent unoccupied
-.041
104.198 .000
.959
.036
.048
housing units
*Confidence Interval (95.0%)
Table 6.5 presents findings from logistic regression analyses conducted on measures of
residential mobility. Both models were statistically significant at the .05 level. The percentage
of owner-occupied housing units had a positive relationship with bias crime occurrence in United
States counties. The B value was .011 and the model explains between .2% and .3% of the
variance in the dependent variable. Moreover, for each one percent increase in owner-occupied
housing units in United States counties, the likelihood of the occurrence of bias criminality
increased by 1.011%. In contrast, the percentage of unoccupied housing units present in United
States counties had a negative relationship with bias crime occurrence, as is indicated by a B
154

value of -.041. The model explained between 3.6% and 4.8% of the variance in the dependent
variable. As such, the model suggested that for every one percent increase in unoccupied
housing units, the counties were .959% less likely to experience bias criminality.
The results of simple logistic regression analyses conducted on the continuous control
variables are found in the following table. All but the percentage of non-English speaking
residents in United States counties had a statistically significant relationship with the occurrence
of bias criminality.
Table 6.6: Logistic Regression on Control Variables
Predictor
β
Wald
p
Exp(B)* Cox & Snell Nagelkerke
χ²
R²
R²
Percent Christian
-.017 74.979 .000
.983
.024
.033
Median Age
-.057 36.681 .000
.945
.012
.016
Percent non-English -.004
1.362
speaking
Percent veterans
.084 25.897
*Confidence Interval (95.0%)

.243

.996

.000

1.088

.000
.008

.0
01
.011

Table 6.6 presents the results of five simple logistic regression analyses. Each one of the
analyses was conducted with the dichotomous dependent variable. First, the relationship
between the percentage of any denomination of Christians and the occurrence of bias crime was
significant and negative. The B value was -.017. The model explains 2.4% to 3.3% of the
variance in the dependent variable. The odds ratio suggested that for every one percent increase
in Christian residents, bias crimes were .983% less likely to occur. Second, median age also had
a negative, yet significant, relationship with bias crime occurrence. The B value was -.057. In
addition, between 1.2% and 1.6% of the variance in the dependent variable was explained by this
model. As such, as the median age increased by one unit, bias crime was .945% less likely to
occur in United States counties. Third, the percentage of non-English speaking persons residing
in each county was not statistically significant in its relationship with bias crime occurrence.
155

Fourth, the percentage of veterans residing in United States counties had a positive and
significant relationship with the occurrence of bias criminality. The B value was .084, indicating
the presence of a relationship that was rather weak. Between .8% and .11% of the variance in
the dependent variable was explained by the model. Additionally, the odds ratio suggested that
for every one percent increase in the veteran population in United States counties, the likelihood
of bias criminality occurring increased by 1.088%.
Models Two, Three and Four
In order to determine the nature of relationships between the continuous dependent
variables used in Models Two, Three and Four and the categorical independent variables, the
Kruskal-Wallis Test was used. This test is a non-parametric alternative to ANOVA. Model Two
represents the number of bias crimes that were motivated by animus toward race between 2000
and 2009 in United States counties. This is a continuous variable and it is used as the dependent
variable for Model Two. Similarly, Model Three represents the number of anti-sexual
orientation bias crimes that occurred during the same time period. Anti-religion bias crimes are
represented in Model Four. Moreover Spearman rho analyses were conducted to measure the
relationship between the continuous independent variables and the continuous dependent
variables present in Models 1-4. Table 6.7 presents the findings from the Kruskal-Wallis tests,
including the effect size. Table 6.8 displays the findings from Spearman rho analyses.

156

Table 6.7: Kruskal-Wallis Test for Models Two, Three, and Four
Model/Variable
χ²
df
p
Model 2 (Anti-Race)
Population in 2000
1,222.030
2 .000
Any hate crime law in 2000
11.715
1 .001
HCL RRE
15.498
1 .000
HCL Sexual Orientation
20.418
1 .000
HCL Gender
1.858
1 .173
HCL Gender Identity
50.109
1 .000
HCL Disability
20.459
1 .000
Model 3 (Anti-Sexual Orientation)
Population in 2000
992.960
2 .000
Any hate crime law in 2000
.966
1 .326
HCL RRE
15.563
1 .000
HCL Sexual Orientation
38.136
1 .000
HCL Gender
14.277
1 .000
HCL Gender Identity
58.488
1 .000
HCL Disability
38.105
1 .000
Model 4 (Anti-Religion)
Population in 2000
999.951
2 .000
Any hate crime law (HCL) in 2000
.046
1 .831
HCL RRE
18.387
1 .000
HCL Sexual Orientation
16.247
1 .000
HCL Gender
2.752
1 .097
HCL Gender Identity
39.165
1 .000
HCL Disability
14.754
1 .000

ES
.389
.004
.005
.007
-.016
.007
.316
-.005
.012
.005
.019
.012
.318
-.006
.005
-.012
.005

Model Two
A Kruskal-Wallis analysis of variance (ANOVA) revealed that county population varied
significantly across counties that reported the occurrence of anti-race motivated bias crime, χ² (2)
= 1,222.030, p = .000. In addition, the presence of any hate crime law in United States counties
varied significantly across anti-race bias crimes in United States counties, χ² (1) = 11.715, p =
.001. The following independent variables also had a statistically significant (p = .000) variation
across anti-race bias crimes in United States counties: HCL RRE, χ² (1) = 15.498; HCL Sexual
Orientation, χ² (1) = 20.418; HCL Gender Identity, χ² (1) = 50.109; and HCL Disability, χ² (1) =
20.459. These findings indicate that there was a significant difference in the dependent variable

157

across each of these groups. Hate crime laws protecting gender did not have a statistically
significant relationship with the number anti-race motivated bias crimes that occurred in United
States counties.
Model Three
For the third model, a Kruskal-Wallis Test revealed several findings. The county
populations varied significantly across the number of anti-sexual orientation bias crime events in
United States counties, χ² (2) = 992.960, p = .000. Additionally, the following independent
variables varied significantly (.000) across the dependent variable: HCL RRE, χ² (1) = 15.563;
HCL Sexual Orientation, χ² (1) = 38.136; HCL Gender, χ² (1) = 14.277; HCL Gender Identity, χ²
(1) = 58.488; and HCL Disability, χ² (1) = 38.105. The independent variable representing the
dichotomous measure of the presence of hate crime law in the aggregate was not statistically
significant.
Model Four
Kruskal-Wallis (ANOVA) analyses uncovered several results for Model Four. The
county populations varied across number of anti-religion bias crimes that occurred in United
States counties between 2000 and 2009. χ² (2) = 999.951, p = .000. Also significant at the .05
level were several other relationships. The following variables varied significantly across the
dependent variable: HCL RRE, χ² (1) = 18.387; HCL Sexual Orientation, χ² (1) = 16.247, HCL
Gender Identity, χ² (1) = 39.165; and HCL Disability, χ² (2) = 14.754. The independent variable
representing the dichotomous measure of the presence of hate crime law in the aggregate was not
statistically significant. Additionally, the independent measure (HCL Gender) was not
significant.

158

Bivariate Analyses for Continuous Independent Variables
The following tables display the results of Spearman’s rho correlation analyses conducted
between the continuous dependent variables of Models Two, Three, and Four and the continuous
independent variables. Each of the three tables is organized by specific model.
Table 6.8: Spearman’s Rho Correlations for Model Two (Anti-Sexual Orient.)
Independent Variable
N
rs
P
Low Economic Status
Percent unemployed
3,139
.080* .000
Percent below poverty line
3,139 -.259* .000
Percent without bachelor’s degree
3,139 -.340* .000
Social Heterogeneity
Percent non-white residents
3,139
.125* .000
Percent foreign-born
3,138
.364* .000
Percent Jewish
3,138
.468* .000
Percent Muslim
3,138
.405* .000
Social Cohesion
Percent divorced
3,139
.171* .000
Crime rate
3,139 -.137* .000
Percent voting in 2000 presidential election
3,035
-.57* .002
Residential Mobility
Percent owner-occupied housing units
3,141
-.013 .480
Percent unoccupied housing units
3,141 -.372* .000
Control
Percent Christian
3,138 -.183* .000
Median age
3,139 -.219* .000
Percent non-English speaking
3,141
.234* .000
Percent veterans
3,139
.026 .147
Percent of hate crimes that are anti-white
737
.368* .000
*Significant at the .01 level
According to Table 6.8, there are numerous statistically significant findings for Model
Two. For the measures of low economic status, each of the independent variables was
significant at the .01 level. The Spearman’s rho correlation between the percentage of persons
who were unemployed in 2000 and the occurrence of anti-race bias crimes had a value of .080.
This indicated a positive, though very small relationship. For the percentage of persons living
below the poverty line in 2000, the Spearman’s rho correlation indicated a value of -.259. This

159

indicated a negative relationship, though it is of small strength. The percentage of persons
holding a bachelor’s degree, however, had a value of -.340 and is negative. This relationship
with the occurrence of anti-race bias criminality was medium in strength.
The measures for social heterogeneity all proved to be statistically significant at the .001
level in their Spearman’s rho correlations with the dependent variable. The percentage of nonWhite residents present in each United States county did have a positive relationship with the
occurrence of anti-race bias criminality, though the strength of the relationship was small. The
Spearman’s rho correlation value was .125. The percentage of foreign-born persons in each
county also had a positive relationship with the occurrence of anti-race bias criminality. The
Spearman’s rho value was .364, which indicated a medium strength in the relationship. Both the
percentage of Jewish persons and the percentage of Muslim persons residing in each county had
a positive and significant relationship to the occurrence of anti-race bias criminality. The
Spearman’s rho correlation for the percentage of Jewish residents was .468, suggesting a
medium strength in the relationship. For the percentage of Muslim residents in each county, the
Spearman’s rho value was .405, providing evidence that a medium strength relationship existed
between the variables.
Measures of social cohesion also exhibited statistically significant relationships at the .01
level with the dependent variable. The percentage of divorced persons residing in each county
and the occurrence of anti-race bias criminality had a positive, yet small relationship. The
Spearman’s rho value was .171. In addition, the crime rate in 2000 had a negative and small
relationship with the dependent variable. The Spearman’s rho value was -.137. The percentage
of persons voting in the 2000 presidential election was negatively correlated with the dependent

160

variable. The Spearman’s rho value was -.57, which was evidence of a small strength in the
relationship.
The percentage of owner-occupied housing units in each county was not statistically
significant in its relationship to the dependent variable. However, the percentage of unoccupied
housing units present in each county was statistically significant at the .01 level in its relationship
with the occurrence of anti-race bias criminality. The Spearman’s rho value was -.57, indicating
a negative, yet small strength in the relationship.
Numerous significant relationships were found in Spearman’s rho correlation analyses
between control variables and the dependent variable in Model Two. The percentage of
Christian residents (any denomination) was negatively correlated with the occurrence of antirace bias criminality in United States counties. The Spearman’s rho value was -.183, indicating a
small and weak relationship to the dependent variable. Median age also had a negative and small
relationship to the dependent variable in Model Two. The Spearman’s rho value was -.219. The
percentage of non-English speaking individuals in each county was positively correlated with the
occurrence of anti-race bias criminality. The Spearman’s rho value was .234, which indicated a
small strength of bivariate relationship. Moreover, the percentage of total bias crimes that were
reported as anti-White was positively correlated to the occurrence of anti-race bias criminality.
The dependent variable excludes anti-White bias crime incidents so as not to create an
interaction effect. The Spearman’s rho value was .368, indicating a medium strength in
relationship with the dependent variable. The percentage of veterans residing in each county was
not correlated to the occurrence of anti-race bias criminality.

161

Table 6.9: Spearman’s Rho Correlations for Model Three (Anti-Religion)
Independent Variable
N
rs
P
Low Economic Status
Percent unemployed
3,139
.081* .000
Percent below poverty line
3,139 -.215* .000
Percent without bachelor’s degree
3,139 -.367* .000
Social Heterogeneity
Percent non-white residents
3,139
.117* .000
Percent foreign-born
3,138
.337* .000
Percent Jewish
3,138
.518* .000
Percent Muslim
3,138
.453* .000
Social Cohesion
Percent divorced
3,139
.145* .000
Crime rate
3,139 -.132* .000
Percent voting in 2000 presidential election
3,035
.012 .506
Residential Mobility
Percent owner-occupied housing units
3,141 -.075* .000
Percent unoccupied housing units
3,141 -.320* .000
Control
Percent Christian
3,138 -.155* .000
Median age
3,139 -.203* .000
Percent non-English speaking
3,141
.265* .000
Percent veterans
3,139
-.021 .248
Percent of hate crimes that are anti-white
737
.285* .000
*Significant at the .01 level
As is apparent in Table 6.9, numerous statistically significant findings emerged from the
Spearman’s rho correlation analyses. Several measures represented low economic status in the
bivariate Spearman’s rho correlations. Each was statistically significant at the .01 level. The
percentage of unemployed persons in each United States county in 2000 was positively
correlated with the occurrence of anti-sexual orientation bias criminality. However, this
relationship was small in strength as its Spearman’s rho value was .081. The percentage of
persons living below the poverty line had a small, yet negative, relationship to the occurrence of
anti-sexual orientation bias criminality. The Spearman’s rho value was -.215. In addition, the
percentage of residents without a bachelor’s degree in 2000 was negatively related to the

162

occurrence of bias criminality in United States counties. The Spearman’s rho value was -.367,
which indicated a medium strength of relationship.
Each of the measures representing social heterogeneity was statistically significant in its
bivariate correlational analysis with the dependent variable. A Spearman’s rho correlation
analysis revealed that there was a small, yet positive relationship between the percentage of nonWhite residents in United States counties in 2000 and the occurrence of anti-sexual orientation
bias criminality. The test statistic for this analysis was .117, indicating a small strength of
relationship. The percentage of foreign-born persons residing in American counties was also
positively correlated with the occurrence of anti-sexual orientation bias criminality. The
spearman’s rho value was .337, revealing a medium strength of relationship between the two
variables. In addition, the percentage of Jewish residents in American counties was significantly
correlated with the occurrence of anti-sexual orientation bias criminality, with a reported test
statistic of .518. This indicated a positive and large strength of relationship between the two
variables. The percentage of Muslim persons residing in United States counties had a
Spearman’s rho value of .453. This indicated a positive, yet medium strength in the relationship
between the percentage of Muslim residents and the occurrence of anti-sexual orientation bias
criminality.
Two of the three measures for social cohesion were statistically significant at the .01
level. First, the percentage of divorced persons residing in American counties was positively
correlated with the occurrence of anti-sexual orientation bias criminality. The Spearman’s rho
test statistic was .145, indicating a small strength of relationship between the two variables.
Second, the crime rate in 2000 had a negative relationship with the dependent variable. The
Spearman’s rho value was -.132. This relationship was small in strength. The percentage of

163

persons in each county voting in the 2000 presidential election was not statistically significant in
its relationship with the occurrence of anti-sexual orientation bias criminality.
Both measures of residential mobility had a statistically significant relationship with the
dependent variable at the .01 level. First, the percentage of owner-occupied housing units had a
Spearman’s rho value of -.075. This indicated a very small strength of relationship that was
negative in direction. Second, the percentage of unoccupied housing units in American counties
in 2000 had a negative, yet medium strength, relationship with the occurrence of anti-sexual
orientation bias criminality. The test statistic was -.320.
All but one control variable had a statistically significant relationship at the .01 level with
the dependent variable for Model Three. The percentage of Christians (any denomination)
residing in American counties was negatively correlated with the occurrence of anti-sexual
orientation bias criminality. The Spearman’s rho value was -.155, indicating that the strength of
the relationship between the two variables was small. In addition, the median age of residents of
American counties had a negative relationship with the dependent variable. The test statistic was
-.203, suggesting that the relationship between the two variables was small as well. The
percentage of non-English speaking residents was positively correlated to the occurrence of antisexual orientation bias criminality. The Spearman’s rho value was .265, indicating the strength
of the relationship was small. Moreover, the percentage of bias crimes that were anti-White in
American counties had a statistically significant and positive relationship with the dependent
variable. With a Spearman’s rho value of .285, the relationship was also small in strength. The
percentage of veterans residing in each American county had no statistically significant
correlation with the occurrence of anti-sexual orientation bias criminality.

164

Table 6.10: Spearman’s Rho Correlations for Model Four
Independent Variable
N
Low Economic Status
Percent unemployed
3,139
Percent below poverty line
3,139
Percent without bachelor’s degree
3,139
Social Heterogeneity
Percent non-white residents
3,139
Percent foreign-born
3,138
Percent Jewish
3,138
Percent Muslim
3,138
Social Cohesion
Percent divorced
3,139
Crime rate
3,139
Percent voting in 2000 presidential election
3,035
Residential Mobility
Percent owner-occupied housing units
3,141
Percent unoccupied housing units
3,141
Control
Percent Christian
3,138
Median age
3,139
Percent non-English speaking
3,141
Percent veterans
3,139
Percent of hate crimes that are anti-white
737
*Significant at the .01 level

rs

p

.060* .001
-.238* .000
-.385* .000
.121*
.380*
.537*
.491*

.000
.000
.000
.000

.105* .000
-.117* .000
.007 .707
-.072* .000
-.327* .000
-.157*
-.209*
.278*
-.031
.290*

.000
.000
.000
.082
.000

Table 6.10 presents the results of several bivariate Spearman’s rho correlation analyses.
Model Four represents the occurrence of anti-religion bias criminality as the dependent variable
across all analyses. All three measures of low economic status were statistically significant at
the .01 level for Model Four. Specifically, the percentage of unemployed persons residing in
American counties had a positive, yet very small strength, relationship with the occurrence of
anti-religion bias criminality. The Spearman’s rho value was .060. Additionally, the percentage
of persons living below the poverty line was negatively correlated with the dependent variable.
The test statistic was -.238, indicating a small strength between the two variables. The
percentage of person’s without a bachelor’s degree in 2000 was negatively correlated to the

165

occurrence of anti-religion bias criminality. The Spearman’s rho value was -.385. This figure
suggested a medium strength relationship between the variables.
Each of the four measures of social heterogeneity and its relationship with the dependent
variable was significant at the .01 level. First, the percentage of non-white residents in American
counties was positively correlated with the occurrence of anti-religion bias criminality. The
Spearman’s rho value was .121, indicating the strength of the relationship was small. Second,
the percentage of foreign-born persons was positively correlated with the dependent variable.
However, the strength of the relationship was medium in nature, with -.380 reported as the test
statistic. Third, the percentage of Jewish persons in United States counties was significantly and
positively correlated with the incidence of anti-religion bias criminality. The Spearman’s rho
was .537, indicating a large strength in the relationship between the two variables. Fourth, the
percentage of Muslims residing in American counties was positively correlated with the
dependent variable. The Spearman’s rho value was .491. This statistic suggests that there was a
medium strength to the relationship between the two variables.
Two of the three measures of social cohesion had statistically significant relationships
with the dependent variable. The percentage of divorced persons in American counties was
positively correlated with the incidence of anti-religion bias criminality. The Spearman’s rho
statistic was .105, indicating that the strength of the relationship is small. In addition, the crime
rate is negatively correlated with the dependent variable. With a test statistic of -.117, the size of
this relationship was small as well. The percentage of persons voting in the 2000 presidential
election did not have a statistically significant relationship with the dependent variable.
Two independent variables were used to represent residential mobility. Both were
statistically significant at the .01 level in their relationships with the dependent variable. The

166

percentage of owner-occupied housing units was negatively correlated with the occurrence of
anti-religion bias criminality. The Spearman’s rho value was -.072, indicating a negative, yet
very small strength of relationship. The percentage of unoccupied housing units in American
counties was also negatively correlated with the occurrence of anti-religion bias criminality. The
Spearman’s rho value was -.327. This figure suggested a medium strength relationship between
the two variables.
All but one of the control variables had a statistically significant relationship with the
dependent variable. First, the percentage of Christian (any denomination) residents in American
counties was negatively correlated with the incidence of anti-religion bias criminality. The
Spearman’s rho value was -.157, indicating not only small strength, but also a negative
relationship. The median age of United States county residents was also negatively correlated
with the dependent variable. In this case, the test statistic was -.209. This suggested a
relationship that is small in strength. In addition, the percentage of non-English speaking
persons residing in American counties was positively correlated with the occurrence of antireligion bias criminality. The Spearman’s rho value was .278, indicating another small strength
relationship. Similarly, the percentage of bias crimes that were anti-White by motivation was
positively correlated with the dependent variable. The test statistic was .290, indicating small
strength. The percentage of veterans residing in United States counties did not have a
statistically significant relationship with the incidence of anti-religion bias criminality.
Key Findings
From the numerous bivariate analyses presented in this chapter, several notable findings
emerged. For the main model (Model 1), it is apparent that several of the independent variables
were good predictors of bias crime at the county level. The population of a county was

167

significantly related to bias crime occurrence. This is relevant given the nature of both
criminality and bias criminality in the United States. Although bias crimes do occur across the
country, it is likely that they are not distributed uniformly. By considering this in the
multivariate model, even more information will be apparent for bias crime occurrence in rural
areas, urban clusters, and urbanized areas. Although several types of hate crime legislation had
significant relationships with bias crime occurrence in American counties, the strength of the
relationships were not particularly high. This indicates that there is not a strong relationship
between hate crime laws and bias crime occurrence – one direction or another. Essentially, the
existence of hate crime legislation (of any type) in a location is not a good predictor of bias crime
occurrence.
Several measures of social disorganization theory are good predictors of bias criminality
at the county level. The percentage of persons living below the poverty line was a useful
measure of economic deprivation as it predicts the likelihood of bias crime occurrence. All
measures of social heterogeneity were good predictors of bias criminality. Certainly, the
percentage of Muslim residents in American counties was the strongest predictor of bias
criminality. The measures of social cohesion did have a statistical significance in their
relationships with bias crime occurrence; however, none of the relationships was particularly
strong. Of the measures, the percent of divorced persons residing in each United States county
was the strongest predictor of bias crime occurrence. It seems that the percentage of divorced
persons residing in each county, although significant, is not a strong predictor of bias criminality.
In addition, the measures of residential mobility were not good indicators of bias crime
occurrence. Although the relationships were statistically significant, they were not strong

168

enough to warrant the expectation of adequate prediction. Moreover, none of the control
variables were good predictors of bias crime occurrence.
For Models 2, 3, and 4, there were several variables that seemed to be good predictors of
bias criminality. The strongest predictors in Model 2 were those variables that measured social
heterogeneity. Specifically, the percentage of persons who were foreign-born, the percentage of
residents who were Jewish, and the percentage of persons who were Muslim were several of the
strongest predictors of anti-race motivated bias criminality. However, the percentage of
unoccupied housing units in each county, as a measure of residential mobility, was a good
predictor of anti-race motivated bias crime occurrence. Models 3 and 4 indicated similar
findings, with measures of social heterogeneity being the strongest predictors of anti-sexual
orientation and anti-religion motivated bias crime occurrence.
Conclusion
The purpose of Chapter Six was to examine the bivariate relationships present between
the dependent variables and the independent variables. This was accomplished by examining the
variables used in each of the four models in order to inform their usage in the multivariate
models. In Model 1, a chi square analysis and logistic regression were used to analyze nonparametric data. In Models 2, 3, and 4, the Kruskal-Wallis test and Spearman’s rho correlations
were conducted to analyze data that was non-parametric in nature. Chapter Seven presents the
findings from the multivariate analyses.

169

CHAPTER 7: USING SOCIAL DISORGANIZATION THEORY TO PREDICT BIAS
CRIME AT THE COUNTY LEVEL
The purpose of Chapter Seven is to present and discuss the findings from the multivariate
analyses of the four main models. First, negative binomial regression is used to analyze the main
model (Model 1). Given the large number of zeroes present in the data, and the resultant overdispersion, this method allows for a more accurate examination of count data. Second, in
Models 2, 3, and 4, Ordinary Least Squares regression (OLS) is used to determine the nature of
the relationships between individual three motivations of bias criminality. In these models, only
the counties that actually report bias criminality as having occurred are included in the analyses.
These specific analyses allow for the exploration of the research questions. The research
questions guide the order of presentation and the discussions address the relevance of the models
to bias crime at the county level.
Question 1:
What are the relationships between traditional indicators (low economic status, social
heterogeneity, family disruption, residential mobility, and violent crime rate) of social
disorganization concepts (as measured by US Census Data, ARDA, and Congressional
Quarterly) and bias crime at the county level?
Table 7.1 presents the findings from negative binomial regression analysis used on Model
1. Each of the continuous and dichotomous independent variables was loaded into the model.
This model examines the relationship between the incidents of bias criminality during the decade
spanning from 2000 to 2009. Traditional indicators of social disorganization theory comprise the
majority of the variables, giving this model the ability to inform their predictive abilities on bias
crime occurrence at the county level.

170

Table 7.1: Model 1 – Negative Binomial Regression
Predictor
B
SE
P
VIF
Constant4
8.589 1.974 .000
Low Economic Status
Percent unemployed
-.067 .069 .329 2.230
Percent below poverty line
-.025 .016 .117 3.087
Percent without bachelor’s degree
-.037 .021 .081 3.414
Social Heterogeneity
Percent non-white
.017** .005 .000 2.129
Percent foreign-born
.065** .022 .004 9.917
Percent Jewish
.190** .036 .000 1.720
Percent Muslim
.361** .124 .004 1.653
Social Cohesion
Percent divorced
.158** .038 .000 2.151
Crime rate
.000 .000 .266 1.057
Percent voting in 2000 presidential election
.012 .011 .302 3.541
Residential Mobility
Percent owner-occupied housing units
-.013 .010 .181 4.418
Percent unoccupied housing units
-.018 .009 .055 4.964
Control
Population in 20005
1.800
1 = < 2,500 (Rural)
-1.652** .162 .000
2 = 2,500 – 49,999 (Urban cluster)
-.963** .125 .000
3 = > 50,000 (Urbanized area)
0
--Hate Crime Law (HCL) in 2000
1.380* .582 .018 12.004
HCL – Race, religion, ethnicity
-1.375* .568 .015 12.233
HCL – Sexual orientation
.521** .139 .000 3.176
HCL – Gender
-.305** .107 .004 1.712
HCL – Gender identity
-.321** .117 .006 1.649
HCL –Disability
-.443 .157 .005 3.483
Percent Christian
-.002 .004 .599 1.630
Median age
-.046* .022 .033 4.284
Percent non-English speaking
-.014 .011 .181 7.855
Percent veterans
-.007 .033 .837 3.274
Percent of bias crimes that are anti-white
.004* .002 .017 1.061
Log Likelihood Chi-Square
1,283.11
Probability > Chi-Square
.000
*p ≤ .05, **p ≤ .01
Table 7.1 presents the findings from one negative binomial regression predicting the
incidence of bias criminality at the county level. In order to assess the fit for Model 1, the
4

In a comparative OLS regression model, only the percent of foreign-born persons (B=27.206) and the percent of
non-English speaking persons (B=-6.123) had statistically significant results at the .01 level.
5
Using population as a continuous variable, the negative binomial regression coefficient is .000 and is significant at
the .01 level.

171

likelihood ratio test was conducted. The value was 1,283.11 with a 99% confidence interval.
Therefore, it is evident that the model fits the data quite well.
With regard the independent variables related to measuring social disorganization theory,
each of the indicators of social heterogeneity was statistically significant at the .01 level. Not
one of the measures of low economic status was statistically significant. However, from Model
1, it is expected that bias crime occurrence would increase by 2% for each percentage increase in
non-White persons residing in each county. Therefore, Hypothesis 2 is supported. Similarly, the
percent of foreign-born persons in each county was positively related to the occurrence of bias
criminality from 2000-2009 in American counties. For each one percent increase in foreign-born
persons residing in United States counties, it is expected that bias crime occurrence would
increase by 7%. This finding supports Hypothesis 3. In addition, it is expected that bias
criminality would increase by 19% for each one percent increase in Jewish persons residing in
each county. Similarly, the percent of Muslim residents in American counties was positively
related to the likelihood of bias crimes occurring. For each one percent increase in Muslim
residents, it is expected that bias criminality would be 36% higher. One measure of social
cohesion had a positive relationship with bias crime occurrence at the .01 significance level. It is
expected that for each one percent increase in divorced persons residing in each American
county that bias criminality would be 16% higher. Thus, Hypothesis 4 is supported by this
finding. The crime rate and the percent of persons voting in the 2000 Presidential election were
not statistically significant. Moreover, neither measure of residential mobility was statistically
significant in its relationship with bias criminality in American counties.
Also relevant are the relationships the control variables have with the dependent variable.
From the negative binomial regression analysis, it appears that bias criminality and county

172

population were negatively related. For each one percent decrease in population between >
50,000 and 49,999 (urbanized area to urban cluster), it is expected that bias crime occurrence
would be decreased by 96.3%. Additionally, as the county population decreased one percent for
all counties with a population under 50,000, (urban cluster to rural), bias criminality would be
165.2% less likely to occur. This finding suggests that that controlling for all other variables, the
larger the population, the greater likely a bias crime is predicted to occur. Hypotheses 1, 5, and 6
are not supported by the findings of Model 1.
It is important to examine the interaction effects that might exist between the independent
variables. Model 1 included collinearity diagnostics so as to test for the potential of
multicollinearity being present in the model. The Variance Inflation Factor (VIF) for each
independent variable indicates the potential for issues involving multicollinearity. All VIF
values above 10 were identified. It appears that both HCL 2000 and HCL – RRE 2000 both had
VIF values above 10, indicating caution should be used when including both variables in the
model. The VIF score for the independent variable indicating the presence of any hate crime law
in 2000 was 12.004. For hate crime laws that specifically protect race, religion, and ethnicity,
the VIF score was 12.233. In Model 1, it is likely that the two independent variables were
interacting. Given there is some intersection in the concept these variables attempt to measure, it
would be reasonable to remove the second measure from any future analyses. Additionally, with
VIF scores just below 10, the percent of foreign-born persons and the percent of non-English
speaking persons residing in each American county were also likely interacting with one another.
Although each variable is different in what it measures, it is entirely possible that a person who is
foreign-born would also speak a language other than English. In future analyses, removing the

173

control variable that measures the percent of persons speaking a language other than English
would likely produce stronger results.
The impact of hate crime legislation on bias criminality was varied. Controlling for all
other variables, the measure of hate crime legislation that includes any type of law had a positive
relationship with bias crime occurrence. Specifically, for each one unit increase in counties with
hate crime legislation, it is expected that there is a 138% increase in bias criminality. However,
when the independent variables representing hate crime legislation are isolated by motivation
type, the findings differ. The most common type of hate crime legislation protects its citizens
from crimes committed based on animus toward their race, religion, or ethnicity. The aggregate
measure of these three types of bias crime motives had a negative relationship with bias crime
occurrence at the county level. For each one unit increase in counties with race, religion, and
ethnicity protected by legislation, it is expected that bias crime will decrease by 137.5%. When
the law protects citizens based on their sexual orientation, the variable had a positive relationship
with the occurrence of bias criminality. Specifically, for each one unit increase in counties with
sexual orientation-based laws, it is expected that the occurrence of bias criminality rises by
52.1%. This relationship was significant at the .01 level. Conversely, the independent variables
gender and gender identity had negative relationships with the occurrence of bias crime
occurrence. For each one unit increase in legislation at the county level, counties can expect
30.5% fewer gender-motivated bias crimes and 32.1% fewer gender identity-motivated bias
crimes to occur. Each of these relationships was significant at the .01 level.
Two additional independent variables had statistically significant relationships with bias
crime occurrence at the .05 level. First, median age had a negative relationship with the
dependent variable in Model 1. For each one unit increase in median age in United States

174

counties, it is expected that bias criminality will increase by 4.6%. Second, the percent of all
bias crimes that were anti-White, by motivation, has a positive relationship with bias crime
occurrence in United States counties. For each one percent increase in anti-White bias crime
incidence, total bias crime levels can be expected to increase by .4%.
Question 2:
What are the relationships between traditional indicators (low economic status, social
heterogeneity, family disruption, residential mobility, and violent crime rate) of social
disorganization concepts (as measured by US Census Data, ARDA, and Congressional
Quarterly) and anti-race motivated bias crime at the county level?
In order to explain differences between the three most frequently occurring motivation
types, OLS regression is used for three separate models. Model 2 regresses the independent
variables on anti-race motivated bias crime counts. Models 3 and 4 regress the independent
variables on anti-sexual orientation motivation and anti-religion motivation, respectively. Table
7.2 presents the findings of an OLS regression analysis conducted for Model 2. The aim of this
sub-analysis is to determine the predictive abilities of measures of social disorganization theory
on the occurrence of anti-race motivated bias crime.

175

Table 7.2: Model 2 – OLS Regression Analysis (Anti-Race)
Predictor
B
Constant
Low Economic Status
Percent unemployed
-12.824
Percent below poverty line
-.276
Percent without bachelor’s degree
5.292
Social Heterogeneity
Percent non-white
1.482*
Percent foreign-born
16.561*
Percent Jewish
6.548
Percent Muslim
33.143*
Social Cohesion
Percent divorced
9.885
Crime rate
.000
Percent voting in 2000 presidential election
2.167
Residential Mobility
Percent owner-occupied housing units
-1.501
Percent unoccupied housing units
-1.004
Control
Population in 2000
7.023
Hate Crime Law (HCL) in 2000
-56.917
HCL – Race, religion, ethnicity
70.505
HCL – Sexual orientation
-44.285*
HCL – Gender
21.047
HCL – Gender Identity
26.612
HCL – Disability
20.682
Percent Christian
.697
Median age
-2.091
Percent non-English speaking
-3.711*
Percent veterans
-2.251
Percent of bias crimes that are anti-white
.058
F
9.495**
R²
.247
*p ≤ .05, **p ≤ .01

SE

β

VIF

10.156
2.169
2.966

-.062
-.007
.109

2.230
3.087
3.414

.692
3.348
3.732
15.612

.103
.513
.076
.090

2.129
9.917
1.720
1.653

5.901
.000
1.761

.081
-.029
.076

2.151
1.057
3.541

1.421
1.407

-.073
-.052

4.418
4.964

11.160
85.555
82.936
22.018
15.711
17.819
24.546
.500
3.105
1.548
4.585
.246

.028 1.800
-.076 12.004
.098 12.233
-.118 3.176
.058 1.712
.063 1.649
.052 3.483
.059 1.630
-.046 4.284
-.221 7.855
-.029 3.274
.008 1.061

Table 7.2 presents the findings from the OLS regression analysis conducted for Model 2.
The R-square is .247, indicating that 24.7% of the variation in anti-race motivated bias crime was
explained by all the independent variables considered together. In addition, the F value is 9.495
and is significant at the .01 level, indicating the model had a sample size large enough to hold

176

predictive capabilities. With findings similar to Model 1, several indicators of social
heterogeneity were significant at the .05 level.
Model 2 indicated that percent of non-White persons living in a county had a statistically
significant relationship with the occurrence of anti-race motivated bias criminality. The
estimated regression coefficient was 1.482. This figure suggests that as the percent of non-White
persons residing in an American county increased by one unit, the likelihood of anti-race
motivated bias criminality increased by 1.482 units. Moreover, the estimated regression
coefficient was significant at the .05 level. The standardized beta coefficient similarly indicated
that for every one standard deviation increase in percent non-White residents, anti-race
motivated bias criminality increased by .103 standard deviations. The standard error was .692.
Model 2 also indicated that the percent of foreign-born persons residing in United States
counties had a statistically relationship to the incidence of anti-race motivated bias crime. The
regression coefficient was 16.561 and the standard error was 3.348. This figure suggests that for
each percent increase in foreign-born persons living in American counties, the likelihood of antirace bias criminality increased by 16.561 units. The standardized coefficient suggests that for
every one standard deviation increase in foreign-born persons residing in American counties,
anti-race motivated bias crime occurrence increased by .513 standard deviations.
The percent of Muslim residents was also statistically significant in its relationship with
the likelihood of anti-race motivated bias criminality at the county level. Model 2 indicated that
regression coefficient was 33.143, which was quite larger than the relationships the dependent
variables had with both percent foreign-born and the percent non-White. The B value suggested
that for each one percent increase in Muslim residents per American county, the likelihood of
anti-race bias crime occurrence increased by 33.143 units. Essentially, as the standardized

177

coefficient suggests, for every one standard deviation increase in Muslim residents, American
counties were likely to experience a .090 standard deviation increase in anti-race motivated bias
criminality. The standard error was 15.612.
In total, two control variables had statistically significant relationships with the dependent
variable in Model 2. First, counties that had hate crime legislation protecting sexual orientation
had a statistically significant relationship with the occurrence of anti-race motivated bias crimes.
The unstandardized regression coefficient was -44.285. This value indicated that for each
additional county that provides hate crime legislation protecting sexual orientation, the likelihood
of anti-race motivated bias criminality decreased by 44.285 units. The standard error was
22.018. The standardized coefficient was -.118, indicating a .118 standard deviation decrease in
bias criminality for every one standard deviation increase in sexual orientation hate crime laws in
American counties. Second, the percent of non-English speaking individuals residing in each
county had a statistically significant relationship with the occurrence of anti-race motivated bias
crime. For every one percent increase in non-English speaking persons in American counties,
the likelihood of anti-race motivated bias crime decreased by 3.711 units. The standardized
coefficient was -.221, indicating that for each one standard deviation increase in non-English
speaking persons residing in each United States county, there was a .221 standard deviation
decrease in anti-race motivated bias crime occurrence. The standard deviation was 1.548.
Importance lies in examining the interaction effects that exist between the independent
variables. Model 2 included collinearity diagnostics in order to test for the possibility of
multicollinearity in the model. The Variance Inflation Factor (VIF) for each independent
variable indicates the potential for problems with multicollinearity. For the purposes of this
dissertation, any VIF values above 10 were identified. It appears that both HCL 2000 and HCL

178

– RRE 2000 both had VIF values above 10, warranting caution in their inclusion in the model.
The VIF score for the independent variable indicating the presence of any hate crime law in 2000
was 12.004. For hate crime laws that specifically protect race, religion, and ethnicity, the VIF
score was 12.233. It is likely that the two independent variables were interacting in this model.
Given there is some overlapping in the concept these variables measure, it would be reasonable
to remove the second measure from any future analyses. In addition, with VIF scores just below
10, the percent of foreign-born persons and the percent of non-English speaking persons residing
in each American county were likely interacting. Although different in what they measure, it is
probably that a person who identifies as foreign-born would also speak a language other than
English. In future analyses, removing the control variable that measures the percent of persons
speaking a language other than English would likely produce stronger results.
In summary, Model 2 suggests statistically significant relationships between three
theoretically important variables. It seems that the more diverse a county becomes, the greater
likelihood it will have anti-race motivated bias crime occurrence. The strongest relationships
occurred when considering the percent of foreign-born persons and the percent of Muslim
persons residing in each American county. Traditional indicators of low economic status, social
cohesion, and residential mobility did not have statistically significant relationships with the
occurrence of anti-race motivated bias crime.
Question 3:
What are the relationships between traditional indicators (low economic status, social
heterogeneity, family disruption, residential mobility, and violent crime rate) of social
disorganization concepts (as measured by US Census Data, ARDA, and Congressional
Quarterly) and anti-sexual orientation motivated bias crime at the county level?

179

Model 3 involves an OLS regression analysis with anti-sexual orientation motivated bias
crime occurrence. Each of the theoretically relevant variables and control variables from Models
1 and 2 are included in this analysis as well. The findings are presented in Table 7.3. Discussion
of the findings and their relevance to the third research question follows the table. The aim of
this sub-analysis is to determine whether traditional indicators of social disorganization theory
have predictive abilities for the occurrence of anti-sexual orientation motivated bias crime.

180

Table 7.3: Model 3 – OLS Regression Analysis (Anti-Sexual Orientation)
Predictor
B
SE
β
Constant
Low Economic Status
Percent unemployed
-3.635 2.417 -.073
Percent below poverty line
.087
.516
.010
Percent without bachelor’s degree
1.316
.706
.113
Social Heterogeneity
Percent non-white
.221
.165
.064
Percent foreign-born
4.326**
.797
.559
Percent Jewish
.471
.888
.023
Percent Muslim
5.888 3.715
.067
Social Cohesion
Percent divorced
2.626 1.404
.090
Crime rate
.000
.000 -.022
Percent voting in 2000 presidential election
.717
.419
.105
Residential Mobility
Percent owner-occupied housing units
-.947**
.338 -.193
Percent unoccupied housing units
-.652
.335 -.142
Control
Population in 2000
-.329 2.656 -.005
Hate Crime Law (HCL) in 2000
-5.408 20.358 -.030
HCL – Race, religion, ethnicity
4.160 19.735
.024
HCL – Sexual orientation
-5.595 5.239 -.062
HCL – Gender
5.834 3.739
.067
HCL – Gender Identity
2.344 4.240
.023
HCL – Disability
1.847 5.841
.019
Percent Christian
.138
.119
.049
Median age
-.165
.739 -.015
Percent non-English speaking
-.861*
.368 -.214
Percent veterans
-.730 1.091 -.040
Percent of bias crimes that are anti-white
.007
.059
.004
F
10.048**
R²
.258
*p ≤ .05, **p ≤ .01

VIF

2.230
3.087
3.414
2.129
9.917
1.720
1.653
2.151
1.057
3.541
4.418
4.964
1.800
12.004
12.233
3.176
1.712
1.649
3.483
1.630
4.284
7.855
3.274
1.061

Table 7.3 presents the findings from the OLS regression conducted for Model 3. This
sub-analysis is aimed at determining the predictive abilities of social disorganization theory for
anti-sexual orientation motivated bias crime occurrence at the county level in the United States.
The dependent variable for this analysis is continuous in nature and includes anti-LGB bias
crime offenses that were reported as having occurred during the decade of 2000-2009.

181

The R-square was .258, indicating that nearly 26% of the variance in anti-sexual
orientation bias crime at the county level was explained by the independent variables together.
The F value was 10.048 and was significant at the .01 level. This indicates that the sample size
was large enough for the model to hold predictive capabilities.
Compared to Model 2, Model 3 had fewer statistically significant findings. However,
like the other models, social heterogeneity did have one significant relationship between the
dependent variable and one of its measures. The percent of foreign-born persons residing in
United States counties had a statistically significant relationship with the occurrence of antisexual orientation motivated bias crimes. The unstandardized regression coefficient for the
relationship between percent foreign-born and anti-sexual orientation motivated bias crime
occurrence was 4.326. This was a positive relationship indicating that for every one percent
increase in foreign-born persons that reside in American counties, the likelihood of anti-sexual
orientation motivated bias criminality increased by 4.326 units. The standard error was .797.
The standardized regression coefficient indicated that for standard deviation increase in percent
of foreign-born persons residing in American counties, there was a .559 standard deviation
increase in anti-sexual orientation motivated bias crime occurrence.
In addition, one measure of residential mobility had a statistically significant relationship
with the dependent variable. The percent of owner-occupied housing units in American counties
had a negative relationship with the occurrence of anti-sexual orientation motivated bias crimes.
The regression coefficient was -.947, indicating that for every one percent increase in owneroccupied housing units in United States counties, there was a .947 unit decrease in anti-sexual
orientation motivated bias crimes. By examining the standardized coefficient, it is apparent that
for every one standard deviation increase in owner-occupied housing units present in American

182

counties, the likelihood of anti-sexual orientation bias criminality decreased by .193 standard
deviations. The standard error was .338.
Only one control variable had a statistically significant relationship with the dependent
variable in Model 3. This relationship was significant at the .05 level. The relationship between
the percent of non-English speaking individuals and the occurrence of anti-sexual orientation
motivated bias criminality had a B value of -.861. This figure indicated that for every one
percent increase in the control variable, there was a .861 unit decrease in anti-sexual orientation
bias crime occurrence. The standardized regression score was -.214 and the standard error was
.368. The standardized score indicated that as the percent of non-English speaking persons
residing in each American county increased by one standard deviation, the likelihood of antisexual orientation motivated bias crime occurrence decreased by .214 standard deviations.
Model 3 experienced similar interaction effects that exist between the independent
variables as Model 2. Both HCL 2000 and HCL – RRE 2000 both had VIF values above 10,
warranting caution when using these variables in the model. It is likely that the two independent
variables were interacting in this model. Given there was some overlapping in the concept these
variables measure, it would be reasonable to remove the second measure from any future
analyses. In addition, with VIF scores just below 10, the percent of foreign-born persons and the
percent of non-English speaking persons residing in each American county were likely
interacting. Although still different in what they measure, it is likely that a person who identifies
as foreign-born would also speak a language other than English. In future analyses, removing
the control variable that measures the percent of persons speaking a language other than English
would likely produce stronger results.

183

In summary, Model 3 explored the relationships between traditional indicators of social
disorganization theory and anti-sexual orientation motivated bias crime occurrence in American
counties. Only three independent variables had significant relationships with the outcome
variable. Similar to Models 1 and 2, social heterogeneity had one indicator, percent foreignborn, that was related to the occurrence of anti-sexual orientation bias criminality. In addition,
one predictor measuring residential mobility had a significant relationship as well. The percent
of owner-occupied housing units in United States counties had a relationship with the outcome
variable. The percent of non-English speaking persons residing in American counties was also
significantly related to the likelihood of anti-sexual orientation motivated bias crime events. The
indicators of social cohesion and low economic status were not significantly related to the
dependent variable in Model 3.
Question 4:
What are the relationships between traditional indicators (low economic status, social
heterogeneity, family disruption, residential mobility, and violent crime rate) of social
disorganization concepts (as measured by US Census Data, ARDA, and Congressional
Quarterly) and anti-religion motivated bias crime at the county level?

The purpose of Model 4 is to examine the relationships between the independent
variables measuring traditional indicators of social disorganization theory and anti-religion
motivated bias crime occurrence in American counties from 2000-2009. The control variables
used in Model 4 are identical to those used in previous models. Table 7.4 presents the findings
from an OLS regression analysis. The dependent variable is continuous in nature and consists of
reported anti-religion motivated bias crime incidents.

184

Table 7.4: Model 4 – OLS Regression Analysis (Anti-Religion)
Predictor
B
SE
Constant
Low Economic Status
Percent unemployed
-2.024 3.049
Percent below poverty line
-.158
.651
Percent without bachelor’s degree
1.162
.890
Social Heterogeneity
Percent non-white
.161
.208
Percent foreign-born
5.164** 1.005
Percent Jewish
8.214** 1.120
Percent Muslim
13.623** 4.686
Social Cohesion
Percent divorced
-.951 1.771
Crime rate
.000
.000
Percent voting in 2000 presidential election
-.103
.529
Residential Mobility
Percent owner-occupied housing units
.266
.427
Percent unoccupied housing units
.403
.422
Control
Population in 2000
1.127 3.350
Hate Crime Law (HCL) in 2000
-35.204 25.681
HCL – Race, religion, ethnicity
35.636 24.895
HCL – Sexual orientation
-3.772 6.609
HCL – Gender
6.741 4.716
HCL – Gender Identity
10.710* 5.349
HCL – Disability
.393 7.368
Percent Christian
.283
.150
Median age
-.297
.932
Percent non-English speaking
-1.376**
.465
Percent veterans
-.055 1.376
Percent of bias crimes that are anti-white
.002
.074
F
15.627**
R²
.351
*p ≤ .05, **p ≤ .01

Β

VIF

-.030
-.013
.074

2.230
3.087
3.414

.035
.495
.294
.114

2.129
9.917
1.720
1.653

-.024
-.031
-.011

2.151
1.057
3.541

.040
.065

4.418
4.964

.014 1.800
-.145 12.004
.153 12.233
-.031 3.176
.057 1.712
.079 1.649
.003 3.483
.074 1.630
-.020 4.284
-.254 7.855
-.002 3.274
.001 1.061

According to Table 7.4, the R-square was .351, indicating that nearly 35% of the variance
in anti-religion bias crime at the county level was explained by Model 4. The F value was
15.627 and was significant at the .01 level. This indicates that the sample size was large enough
for the model to hold predictive capabilities.

185

It is evident from an OLS sub-analysis that there were several statistically significant
findings. First, similar to the previous models, measures of social heterogeneity had statistically
significant results. The percent of foreign-born persons residing in each American county had a
regression coefficient of 5.164. This figure was statistically significant at the .01 level and was
positive. This finding indicates that a one percent increase in foreign-born persons residing in
American counties was related to a 5.164 unit increase in anti-religion motivated bias crime
occurrence. The standard error was 1.005 and the standardized regression coefficient was .495.
Using the standardized beta value to evaluate the relationship, it is evident that for each one
standard deviation increase in the percent of foreign-born residents, the dependent variable
increased by .495 standard deviations.
With an even stronger relationship to anti-religion based bias criminality, the percent of
Jewish persons living in counties had a statistically significant relationship with the dependent
variable in Model 4. The regression coefficient was 8.214 and the standard error was 1.120. For
each one percent increase in Jewish residents, bias crime was expected to increase by 8.214
units. In order to compare across independent variables, however, it is important to examine the
standardized coefficient as well. This value was .294, which indicates that for a one standard
deviation increase in the independent variable, there was a .294 standard deviation increase in
anti-religion motivated bias crime events.
The percent of residents that were Muslim was also statistically significantly related to
the dependent variable at the .05 level. The B value was 13.623 and the standard error was
4.686. This regression coefficient indicated that each one percent increase in Muslim residents
in American counties resulted in bias criminality increasing by 13.623 units. The standardized
score was .114, indicating that for each one standard deviation increase in Muslim residents,

186

there was a .114 standard deviation increase in bias crime occurrence in American counties from
2000-2009.
In addition, Model 4 indicated statistically significant relationships between two control
variables and the dependent variable. First, in counties that report having hate crime legislation
protecting gender identity, there was a significant relationship present with anti-religion bias
crime events. This relationship was significant at the .05 level. The regression coefficient was
10.710 and the standard error was 5.349. This value indicated that for every additional county
that includes gender identity laws, anti-religion bias crimes increased by 10.710 units. The
standardized coefficient was .079. Therefore, it is evident that for every one standard deviation
increase in hate crime laws that include gender identity, there was a .079 standard deviation
increase in anti-religion motivated bias criminality. Second, the percent of non-English speaking
persons residing in American counties had a statistically significant (at the .01 level) relationship
with the dependent variable. This relationship was also negative as the B value was -1.376. For
each one percent increase in non-English speaking persons residing in American counties, it can
be expected that anti-religion motivated bias crime will decrease by 1.376 units. The standard
error was .465. The standardized regression coefficient indicated that for every one standard
deviation increase in the independent variable, there was a .254 standard deviation decrease in
anti-religion motivated bias criminality in United States counties.
Model 4 resulted in similar findings for multicollinearity. It is evident that both HCL
2000 and HCL – RRE 2000 both had VIF scores above 10, creating question about the
appropriateness of their inclusion in the model. It is probable that the two independent variables
are interacting in Model 4. Given there is some overlapping in the concept these variables
measure, it would be reasonable to remove the second measure from any future analyses. In

187

addition, with VIF scores just below 10, the percent of foreign-born persons and the percent of
non-English speaking persons residing in each American county were likely interacting with one
another. Although categorically different in what they measure, it is probably the case that a
person who identifies as foreign-born would also speak a language other than English. In future
analyses, removing the control variable that measures the percent of persons speaking a language
other than English would likely produce stronger results for the entire model.
To summarize Model 4, there were five statistically significant findings. The percent of
foreign-born persons, the percent of Jewish residents, and the percent of Muslim residents in
each county represented significant findings for measures of social heterogeneity. In addition,
two control variables were statistically significant – the presence of hate crime laws protecting
gender identity and the percent of non-English speakers present in each county. Similar to the
preceding models, Model 4 provides insight into how the social heterogeneity aspect of social
disorganization theory affects bias criminality at the county level. The traditional indicators of
low economic status, social cohesion, and residential mobility did not have statistically
significant relationships with anti-religion motivated bias crime from 2000-2009.
Question 5:
What are the differences between anti-race, anti-religion, and anti-sexual orientation bias crime
occurrences at the county level?
Table 7.5 presents a summary of multivariate findings in this dissertation. Each of the
independent variables is listed in the first column. Each of the statistically significant findings
from Models 1, 2, 3, and 4 are presented together. For Model 1, negative binomial regression
was used for the analysis. In Models 2, 3, and 4, OLS regression is used for the sub-analyses.
The unstandardized and standardized regression coefficients are presented in order to allow

188

comparisons of the independent variables effects on the dependent variables through different
models.
Table 7.5: Summary of Multivariate Findings/Regression Coefficients (Standardized)
Model 1 Model 2 Model 3 Model 4
Low Economic Status
Percent unemployed
Percent below poverty line
Percent without bachelor’s degree
Social Heterogeneity
Percent non-white
.017**
1.482*
(1.018)
(.103)
Percent foreign-born
.065**
16.561* 4.326**
5.164**
(1.067)
(.513)
(.559)
(.495)
Percent Jewish
.190**
8.214**
(1.209)
(.294)
Percent Muslim
.361**
33.143*
13.623**
(1.434)
(.090)
(.114)
Social Cohesion
Percent divorced
.158**
(1.171)
Crime rate
Percent voting in 2000
presidential election
Residential Mobility
Percent owner-occupied
-.947**
housing units
(-.193)
Percent unoccupied housing units

189

Table 7.5 (Cont’d)
Control
Population in 2000
Rural
Urban cluster
Urbanized area
Hate Crime Law (HCL) in 2000
HCL – Race, religion, ethnicity
HCL – Sexual orientation
HCL – Gender
HCL – Gender identity
HCL – Disability
Percent Christian
Median age

-1.652**
(.192)
-.963**
(.382)
1.380*
(3.976)
-1.375*
(.253)
.521** -44.285*
(1.684)
(-.118)
-.305**
(.737)
-.321**
(.726)

-.046*
(.955)

Percent non-English speaking
Percent veterans
Percent of bias crimes
that are anti-white
*p ≤ .05, **p ≤ .01

10.710*
(.079)

-3.711*
(-.221)

-.861*
(-.214)

-1.376**
(-.254)

.004*
(1.004)

Social Heterogeneity
Table 7.5 presents the summary of the multivariate findings from Models 1, 2, 3, and 4.
It is clear that across all four models, the measures of social heterogeneity had the most
consistent findings. In the main negative binomial regression model, all four indicators of social
heterogeneity produced significant findings. This is the most indicators of social disorganization
theory to have significant findings in the model. Specifically, the percent of persons that were
foreign-born in each county seemed to have the greatest effect on the likelihood of bias
criminality occurring. This was consistent across Models 2, 3, and 4 as well. Model 2 did report

190

the strongest relationship between the variables, with a regression coefficient of 16.561. Other
measures of social heterogeneity did not exceed 5.164 across all models. Therefore, of all
findings, the percent of foreign-born persons residing in each county was the strongest predictor
of bias criminality in the aggregate, and bias criminality when motivated by anti-race, anti-sexual
orientation, and anti-religion animus.
Social Cohesion
The only measure of social cohesion that had any relationship with bias criminality at the
county level was the percent of divorced persons residing in each county. This variable was only
statistically significant in the main model – Model 1. The regression coefficient was .158 and
was significant at the .01 level. However, this measure of social cohesion lost its statistical
significance when regressed upon the three specific motivation types.
Residential Mobility
Similar to social cohesion, only one measure of residential mobility had a statistically
significant relationship with the dependent variable. However, in this case, the relationship was
negative and rather weak in Model 3. The regression coefficient was -.947 and was significant at
the .01 level. This indicates that as the percent of owner-occupied housing units increases, the
likelihood of anti-sexual orientation bias crimes goes down. This relationship is only significant
in Model 3.
Population
County population also had a statistically significant relationship with bias criminality in
Model 1. The findings here indicate negative, yet weak relationships. According to Model 1, it
seems that as the population decreases in a county, the likelihood of bias criminality decreased as
well. This was consistent across each of the three U.S. Census-defined cut points: rural, urban

191

cluster, and urbanized area. The regression coefficient for urban cluster was -.963 and was
significant at the .01 level. Even stronger, the regression coefficient for urbanized area was .1.652. This figure was also statistically significant at the .01 level. However, these
relationships disappeared for Models 2, 3, and 4.
Hate Crime Legislation
Overall, the presence of hate crime legislation had mixed results in its ability to predict
bias crime occurrence in American counties. For Model 1, the measure of a county having any
hate crime law in existence as of 2000 had a positive and statistically significant relationship
with bias crime occurrence. The regression coefficient value was 1.380. This indicated that the
presence of hate crime legislation resulted in the increase of bias crime offenses in American
counties. Models 2, 3, and 4 did not produce statistically significant results for this variable.
However, when examined by specific type of law, the results changed direction in several cases.
For instance, hate crime laws protecting for race, religion, and ethnicity in 2000 had a
negative relationship with bias criminality. The regression coefficient was -.1.375. This
indicated that the presence of hate crime legislation protecting for race, religion, and ethnicity
does in fact decrease the likelihood of bias criminality. Only Model 1 produced a statistically
significant relationship between these variables.
However, hate crime laws that protect its citizens based on sexual orientation produced
statistically significant results in Models 1 and 2. In Model 1, the regression coefficient was .521
and was significant at the .01 level. This indicates, that overall, the presence of hate crime laws
results in an increase of anti-sexual orientation motivated bias crime. That result changed,
however, when examined in Model 2. Here, there was a rather strong, negative relationship
between the presence of hate crime legislation protecting for sexual orientation and the

192

occurrence of anti-race motivated bias criminality. The regression coefficient for this analysis
was -44.285.
Both types of laws, protecting for gender and for gender identity, also had negative and
statistically significant relationships with the dependent variable in Model 1. This indicates that
with the added presence of these laws, bias crime occurrence decreases in American counties.
The regression coefficient for gender was -.305 and for gender identity, the coefficient was -.321.
Both relationships are statistically significant at the .01 level. In addition to Model 1, laws
protecting for gender identity had a stronger effect on anti-religion motivated bias crime
occurrence in Model 3. Here, there regression coefficient was 10.710 and was statistically
significant at the .05 level. This finding indicates that as counties gain laws protecting its
residents from bias crimes motivated by anti-gender identity animus, anti-religion motivated bias
criminality increased.
Additional Control Variables
First, in Model 2, median age had a statistically significant relationship with the
occurrence of bias crime occurrence from 2000-2009. The regression coefficient was -.046,
indicating a weak relationship. However, it seems from Model 1 that as the median age in a
county rises, the likelihood of bias criminality falls.
Second, in Models 2, 3, and 4, the percent of non-English speaking persons residing in
American counties had negative and statistically significant relationships with the dependent
variables. These findings are considered with caution given the multicollinearity findings
discussed in earlier in the chapter. It is likely that the percent of non-English speaking persons
residing in each county interacts with the percent of foreign-born persons living in each
American county. However, the regression coefficients indicate that as the percent of non-

193

English speak persons increase in each county, the risk of bias crime occurrence decreases. The
direction of the relationship is different than the relationship percent foreign-born has on the
dependent variable. In Model 2, the regression coefficient was -3.711. Models 3 and 4 reported
relationships with coefficients of -.861 and -1.376, respectively.
Third, the percent of bias crimes that are anti-White does seem to have a small, yet
positive relationship with bias crime occurrence in Model 1. The regression coefficient was .004
and it was statistically significant at the .05 level. This indicates that as the percent of anti-White
bias crimes increase in American counties, the likelihood of bias crime minimally increases
overall.
Summary
The purpose of Chapter Seven was to present and discuss the findings from several
multivariate analyses. Negative binomial regression was used for Model 1 – the main model.
Since bias crimes are rare events, and numerous counties have zero reported bias crime incidents,
negative binomial regression was chosen to account for the over-dispersed nature of the
dependent variable. Theoretically relevant independent variables representing social
disorganization theory were regressed on the dependent variable, which was the total recorded
number of bias crime events per each United States county. Models 2, 3, and 4 used Ordinary
Least Squares regression to analyze the effects of social disorganization theory driven variables
on bias crime occurrence by motivation. Moreover, this chapter presented diagnostic statistics to
assess the presence of multicollinearity. Chapter Eight includes a summary of the dissertation
and a discussion of policy implications. Additionally limitations of the research are discussed as
well as the future direction of bias crime research.

194

CHAPTER 8: DISCUSSION AND CONCLUSION
The purpose of Chapter Eight is to provide an overview and discussion of this
dissertation. Much information has been assembled during the course of this study. Therefore,
Chapter Eight includes a review of the dissertation and its relevant findings. In addition,
attention is paid to the many limitations to this dissertation and to bias crime research in general.
In sociology and criminal justice, direct applications to theory testing are often important in
order to inform public policy. Since bias crime is generally addressed by the public sector (i.e.
the criminal justice system), several implications for public policy are presented and discussed.
Moreover, adding to the knowledge base in the social sciences is imperative to solving questions
involving human behavior. As such, this dissertation presents and discusses several avenues for
future research. There is much to do in terms of social science research on bias crime
occurrence. Although this study certainly contributes to the discussion of bias criminality,
additional research is needed to uncover the factors present when considering the context of biasmotivated crimes.
A Review of the Dissertation and Notable Findings
Chapter One provided an overview of bias criminality and the problems the phenomenon
creates. In addition, the chapter introduced the notion that bias crime research is lacking overall,
and in need of further investigation. Chapter Two provided a detailed discussed about what is
known about bias criminality in the social science research. This chapter introduced descriptive
information about bias crime occurrence by motivation type, as well as how society has
responded to these specific crimes. Additionally, Chapter Two provided an overview of the
social science theoretical perspectives most frequently considered when examining bias crime
research. Chapter Three introduced social disorganization theory as the theory to be tested by

195

the research questions and data in this dissertation. An overview of social disorganization theory
was presented, including an explanation of the historical development of the perspective since its
inception in the Chicago School. Chapter Four introduced the data and the methods used for the
analyses. This chapter also presented each of the dependent and independent variables, and
explained the theoretical relevance of each. Chapter Five presented the context of bias
criminality as found in the data. This resulted in the display of descriptive findings for both bias
crimes as they occurred in American counties and descriptive findings for each of the motivation
types. Chapter Six included the various bivariate analyses and the resultant findings. These
analyses examined the relationships between each of the dependent variables and each of the
independent variables. Numerous statistical methods were used as were appropriate for each
variable’s level of measurement. Chapter Seven presented the main models for the dissertation.
Model 1 used negative binomial regression to analyze the relationships between measures of
social disorganization and bias crime occurrence in United States counties. Models 2, 3, and 4
used OLS regression to analyze the relationships between indicators of social disorganization
theory and bias crime occurrence by motivation type.
Notable Findings
Measures of Social Disorganization Theory
From the four multivariate models, several notable findings emerged. The aim of this
dissertation was to examine the traditional indicators of social disorganization theory and their
effect on the likelihood of bias crime occurrence at the county level. Overall, each of the models
provided indication that measures of social heterogeneity were helpful in predicting bias crime
occurrence. In Model 1, all four measures of social heterogeneity had importance in predicting
bias criminality at the county level. The percentage of non-White residents in American counties

196

and the percentage of foreign-born residents both were notable in their ability to aid in the
predictive ability of the regression model. In addition, the percentage of religious minorities that
lived in American counties contributed to the model’s predictive ability. However, the strongest
finding involved the percentage of persons living in American counties who identified as
Muslim. For Model 1, these measures of social heterogeneity seemed to be predictive of bias
crime occurrence at the county level from 2000 – 2009. This is not particularly surprising as the
increase in diversity of a community also represents the increased opportunity for interaction,
perhaps strained, with people who are part of marginalized populations. Essentially, this refers
to the situation where a white, American, Christian person encounters someone who is viewed as
“the other.” This seemed most likely when the member of a superordinate group interacted
negatively with a Muslim resident. However, interactions with non-White persons and Jewish
persons were also statistically significant. Model 1 suggested that this “exposure” to diversity
was predictive of bias crime occurrence in United States counties.
Across Models 2, 3, and 4, findings were similar. In each of these subanalyses, the
percentage of foreign-born persons residing in American counties contributed toward the
prediction of bias crime occurrence. This is the only statistically significant finding that was
consistent across all four models. Models 2, 3, and 4 have different dependent variables which
are based on the specific motivation type. Therefore, it is evident that the percentage of foreignborn persons is predictive of the occurrence of bias criminality. In addition, the percentage of
foreign-born persons residing in American counties also aids in the prediction of anti-race, antisexual orientation, and anti-religion motivated crimes. Certainly, there are different ways to
view this finding. However, it should be noted that it is highly unlikely that the minority group
is responsible for the commission of the bias crimes. In this case, foreign-born persons do not

197

have the social standing (power) to be able to maintain a hegemonic type of violence. It is more
likely that in counties that have a higher percentage of foreign-born persons, the members of
superordinate groups might feel threatened and are acting accordingly.
Although measures of social heterogeneity were the prominent findings related to social
disorganization theory, two other measures are important as well. First, in Model 1, examining
the percentage of persons divorced aids in the predictive ability of the model. This is not a
particularly strong relationship, but it is statistically significant at the .01 level. This is a cautious
indication that the decrease of social cohesion is worth noting in understanding the predictors of
bias criminality. This finding only surfaced in Model 1 and was rather weak. Second, one
measure of residential mobility was statistically significant in Model 3. Specifically, this
indicates that the percentage of owner-occupied housing units that exist in American counties is
indeed predictive of bias crime occurrence. Essentially, as more residents purchase their homes
and are considered to be permanent fixtures, the likelihood of bias criminality decreases. This is
a weak relationship, but it is statistically significant at the .01 level.
The findings of this dissertation are evidence that the use of social disorganization theory,
a macro-level theory - to explain bias crime is helpful. Traditional indicators of diversity, social
cohesion and residential mobility had varying influence on Model 1’s ability to predict bias
criminality in American counties. However, measures of economic deprivation did not lend
significant findings. This is an area that should be explored in future research.
Additional Variables
The population of American counties was relevant in the understanding of bias crime
occurrence in this dissertation. From Model 1, it is evident that as the population decreases, so
does the likelihood of bias criminality. In essence, as established in the literature, bias

198

criminality can be expected to occur in greater frequency in urban areas. When considering the
characteristics of the three U.S. Census defined population areas, this finding is expected. In
many situations, urban areas have greater social diversity. With this diversity, as explained,
comes the likelihood of bias crime occurrence. As the population decreases to urban clusters, the
likelihood of bias criminality decreases as well. In rural areas, where the communities are rather
homogenous, the likelihood of bias crime occurrence decreased yet again. Population is a
relevant predictor of bias crime occurrence only in Model 1.
In addition, hate crime legislation had a rather inconsistent influence on the likelihood of
bias crime occurrence in American counties. In Model 1, it seems as though the presence of hate
crime legislation in general actually increases the chance of bias criminality. This is likely due
to increased attention, and therefore better reporting that is the result of new laws. However,
there is a difference that is worth nothing when the laws are broken down into specific protection
types. For instance, when laws protecting race, religion, and ethnicity are present, bias
criminality actually decreased. Again, this is likely because of increased attention, but there also
may be a deterrence factor involved. In Model 1, this finding applies to laws protecting gender
and gender identity as well. However, when the law protects sexual orientation, it seems that the
occurrence of bias criminality increases across American counties. Although this finding is not
strong enough to suggest that the laws are problematic in that they increased anti-sexual
orientation bias criminality, it should be noted that the direction of this relationship is different
from the other types of laws in Model 1. It very well may be that there is backlash toward the
LGB community when counties do in fact have laws on the books protecting anti-sexual
orientation bias criminality. However, Model 2 suggests yet another type of relationship is
present. In Model 2, laws protecting race, religion, and sexual orientation are strongly associated

199

with the decrease in anti-sexual orientation bias criminality. The cause of this is unknown;
however, it would be worthy of examination in future studies.
Like sexual orientation, laws protecting a person’s gender identity also had different
findings in Models 1 and 4. In the main model, it seems that there is a relatively weak predictive
ability of laws protecting for gender identity to affect the likelihood of bias criminality. This
relationship was negative in Model 1, but was positive in Model 4. This suggests a presence of
laws protecting gender identity actually increased the likelihood that an anti-religion based bias
crime would occur. This finding is curious as gender identity and religion are seemingly
unlinked. However, it is worth examining the specific motivation for anti-religion based bias
crimes.
Limitations
Unfortunately, there are numerous limitations to this study. Each of the current
limitations to this dissertation is discussed; however, there are certainly limitations to bias crime
research that are beyond the scope of this small section in a dissertation. These limitations
inform the future research implications presented toward the end of the chapter. Bias criminality
is certainly an important and complex issue social scientists need to continue exploring.
Perhaps one of the major limitations of this study – and much of empirical research on
bias criminality – is the reliance on official data. Given that so many counties, cities, and states
do not report bias crimes to the FBI, any research from data that do exist should be viewed with
caution. Certainly, each piece of research that investigates bias crime occurrence is worthwhile;
however, the data that do exist are limited in nature. In addition, the reliance on using counties
as the unit of analysis is a limitation as well. Given some communities spread across numerous
counties, it would be appropriate to measure these communities collectively – or as cities.

200

However, due to the limited availability of county-level data, an exhaustive analysis is not
possible.
A major limitation to this dissertation is political ideology and its effect on the quality of
data. There are many ideological barriers to proper investigation of the context of bias
criminality. First, many individuals see bias crime to be an example of identity politics (Jacobs
& Potter, 1998). Generally, conservatives oppose an overreaching government. Perhaps this
informs the reason many local governments are unwilling to report bias crime occurrence to the
FBI. Of course, this is a matter of perspective, as there are many options for how crimes are
classified in different locales. A bias crime that occurs in New York might be identified and
labeled as a bias crime, but the same crime in Mississippi it is labeled as an assault. Certainly,
this results in differential reporting. In addition, many conservatives may support having greater
police presence in communities in the name of crime control; however, when the United States
government attempts to enforce laws or collect information about bias criminality, this is
considered by the same people as government intrusion.
The lack of cooperation by some conservatives at the national and local level is a serious
impediment to a greater understanding of the problem. Since bias crime disproportionately
affects minority communities, it comes as no surprise that some conservatives would oppose the
identification and labeling of bias crimes as such. To do so would acknowledge the relationships
that may occur between superordinate and subordinate communities, power, and hegemonic
forms of oppression. The result of this lack of political cooperation is that it creates a situation
where researchers are required to examine areas that are already known.
Another limitation relates to using the county as the unit of analysis. As collected and
reported by the FBI, some bias crime incidents are reported as having occurred in multiple

201

counties. Because the data are collected by city and state, there is not always an opportunity to
accurately identify the county in which the crime occurred. For instance, some agencies exist
within two or more counties. Therefore, when a bias crime is reported, two or more county
names are associated with it. In the case of this dissertation, this leads to the exclusion of bias
crime events that occur in many large cities. In the future, a similar analysis could be conducted
using cities as the units of analysis. That would account for this limitation.
In addition, this dissertation is solely a quantitative effort. In future studies, qualitative
inquiries could be included to enrich the understanding and depth of the research questions.
Perhaps by talking with the gatekeepers of the criminal justice system – the police – additional
findings would be uncovered. The detail gained by such analyses could shed light on the nature
of the communities that most experience bias crimes, and perhaps explain effects of social
disorganization theory from a different perspective.
Implications for Policy
There are numerous policy implications of this dissertation. First, this dissertation adds
to the scholarly discussion on bias criminality as it affects the United States – particularly at the
local level. This can be used to promote increased awareness for police officers, the government
official most likely to interact with the effects bias crimes have on society. However, this
information is pertinent across the criminal justice system and into the political arena as well. As
laws and policies are debated and enacted, knowing what works and what does not if an effective
way of informing policy. Better bias crime policy will allow researchers to add benefit to society
by uncovering the “dark figure of crime” as it relates specifically to bias crime.
As mentioned in the discussion of limitations, much of the resistance to bias crime policy
change comes from conservative ideology and the notion of bias crimes being reduced to mere

202

identity politics. However, part of this debate may center on notions of federalism. The FBI is a
federal agency, and therefore its existence is antithetical to many conservatives’ positions on the
proper role of government. Given conservatives favor local control, perhaps scholars and
progressive politicians who recognize the importance of bias crime policy could make inroads by
appealing to notions of local control.
Aside from the macro-level issues associated with bias crime occurrence, there are also
policy implications for individuals in American communities. This aligns well with the major
findings in this dissertation – the relationship between social heterogeneity and bias crime
occurrence. Although it is not a new concept, increased diversity trainings – or sensitivity
trainings – for criminal justice systems actors might be beneficial in creating a local government
culture that is informed and equipped to tackle the specific issues many minority communities
face with regard to bias criminality. Certainly, this can exist by way of inclusion into more
police academy curricula and department trainings. In addition, the increased concentration on
the recruitment of minority police officers could certainly allow police departments to service
communities in a more complete fashion. After all, given police departments have historically
been comprised of white, heterosexual males, there is room for most departments to diversify its
force.
Implications for Future Research
There is much room for additional research related to bias crime, its effects on society,
and the relevant governmental policy. This dissertation examined disorganization theory and its
relationship to bias crime at the local level. Although findings were somewhat inconclusive, it
would be appropriate to test other sociological theoretical perspectives as they relate to bias
crime in the United States. Given this dissertation’s main results were related to aspects of social

203

heterogeneity – or diversity – theoretical perspectives that speak to social difference in
communities might be enlightening. For instance Green’s (1998) defended neighborhoods
perspective may be appropriate in order to understand bias criminality as it exists related to the
in-migration of minority groups in various spaces. Moreover, examining the masculinity
literature and applying Messerschmidt’s structured action theory may produce findings that
explain how “doing gender” affect the occurrence of bias criminality across the United States.
Continuing to study social disorganization theory as it relates to bias crime still has much
relevance in the social sciences. Examining the effect the same independent variables have on a
dependent variable built from different data could be useful. Although the data on bias crime
occurrence in the United States is indeed limited, many social advocacy groups collect and
maintain records of officially reported and unreported bias-based incidents. The AntiDefamation League, the Southern Poverty Law Center, and the Human Rights Campaign are just
a few organizations that could be contacted in order to gain access to the rich data that are
available.
Moreover, continuing to study social disorganization theory and its effect on bias
criminality is possible by identifying and collecting data to create different measures of
traditional indicators of the theory. This dissertation primarily used United States Census data;
however, other measures were created from ARDA and Congressional Quarterly data.
Given this dissertation identified social heterogeneity as a major factor associated with
bias crime occurrence, further studies on diversity-related policy could prove to be quite useful.
An examination of the literature related to diversity programming in both the public and private
sectors would add a layer of detail that could be of particular importance when considering how
policy informs practice.

204

Overall, researchers are largely unable to make conclusive statements about the
predictors of bias criminality. Currently, most of the extant research is descriptive in nature.
This prevents researchers from knowing great detail about bias criminality both at the macro and
micro levels. Given there is little empirical research in existence, any study that adds to the
discussion and understanding of bias crime is worthwhile.
Conclusion
This goal of this dissertation was to identify predictors of bias criminality at the county
level in the United States. This was accomplished through the lens of social disorganization
theory. Although the findings are not resoundingly supportive of the application of social
disorganization theory to the understanding of bias criminality, there are remarkable conclusions
nonetheless. First, measures of social heterogeneity – or diversity – seem to yield the most
conclusive evidence toward predicting the risk of bias crime occurrence. Certainly, these
findings from various measures of social heterogeneity must be met with caution – as the data
should not be used to promote or sustain socially oppressive structures. Second, the influence of
hate crime legislation must be considered in depth. Although views on hate crime laws are
highly politically polarized to date, this dissertation presents evidence that additional exploration
as to their efficacy is necessary. The increased exposure to social difference and attention to hate
crime legislation would, perhaps, promote a depoliticized atmosphere in which policy makers
could consider the problem of bias criminality more accurately – and work toward mitigating
bias-based offending in the United States. This dissertation is a contribution toward that end.

205

APPENDICES

206

APPENDIX A
Table A.1: Characteristics of Economic Deprivation (2000)
State
Population
Population
Number
16+
Unemployment
Alabama
Alaska
Arizona
Arkansas
California
Colorado
Connecticut
Delaware
District of
Columbia
Florida
Georgia
Hawaii
Idaho
Illinois
Indiana
Iowa
Kansas
Kentucky
Louisiana
Maine
Maryland
Massachusetts
Michigan
Minnesota
Mississippi
Missouri
Montana
Nebraska
Nevada
New
Hampshire
New Jersey
New Mexico
New York
North
Carolina
North Dakota
Ohio

4,447,100
626,932
5,130,632
2,664,155
33,871,648
4,301,261
3,405,565
783,600

3,450,542
458,054
3,907,229
2,064,753
25,596,144
3,325,197
2,652,316
610,289

126,911
27,953
133,368
75,894
1,110,274
99,260
92,668
20,549

3.68
6.10
3.41
3.68
4.34
2.99
3.49
3.37

Poverty
line
Number
698,097
57,602
698,669
410,225
4,706,130
388,952
259,514
69,901

572,059
15,982,378
8,186,453
1,211,390
1,293,953
12,419,293
6,080,485
2,926,324
2,688,418
4,041,769
4,468,976
1,274,923
5,296,486
6,349,097
9,938,444
4,919,479
2,844,658
5,595,211
902,195
1,711,263
1,998,257

469,041
12,744,825
6,250,687
949,908
969,872
9,530,946
4,683,717
2,281,274
2,059,160
3,161,542
3,394,546
1,010,318
4,085,942
5,010,241
7,630,645
3,781,756
2,158,941
4,331,369
701,168
1,315,715
1,538,516

31,844
412,411
223,052
35,886
36,784
375,412
152,723
64,906
58,415
109,350
146,218
31,165
128,902
150,952
284,992
109,069
93,778
148,794
28,710
32,287
61,920

6.79
3.24
3.57
3.78
3.79
3.94
3.26
2.85
2.84
3.46
4.31
3.08
3.15
3.01
3.73
2.88
4.34
3.44
4.09
2.45
4.02

109,500
1,952,629
1,033,793
126,095
148,732
1,291,958
559,484
258,008
257,829
621,096
851,113
135,501
438,676
573,421
1,021,605
380,476
548,079
637,891
128,355
161,269
205,685

19.14
12.22
12.63
10.41
11.49
10.40
9.20
8.82
9.59
15.37
19.04
10.63
8.28
9.03
10.28
7.73
19.27
11.40
14.23
9.42
10.29

1,235,786
8,414,350
1,819,046
18,976,457

960,498
6,546,155
1,369,176
14,805,912

25,500
243,116
60,324
640,108

2.65
3.71
4.41
4.32

78,530
699,668
328,933
2,692,202

6.35
8.32
18.08
14.19

8,049,313
642,200
11,353,140

6,290,618
502,306
8,788,494

214,991
15,257
282,615

3.42
3.04
3.22

958,667
73,457
1,170,698

11.91
11.44
10.31

207

%

%

15.70
9.19
13.62
15.40
13.89
9.04
7.62
8.92

Table A.1 (Cont’d)
Oklahoma
3,450,654
Oregon
3,421,399
Pennsylvania
12,281,054
Rhode Island
1,048,319
South
Carolina
4,012,012
South Dakota
754,844
Tennessee
5,689,283
Texas
20,851,820
Utah
2,233,169
Vermont
608,827
Virginia
7,078,515
Washington
5,894,121
West Virginia
1,808,344
Wisconsin
5,363,675
Wyoming
493,782
Total
281,412,514

2,666,724
2,673,782
9,693,040
827,797

86,832
112,529
339,386
29,859

3,114,016
577,129
4,445,909
15,617,373
1,600,279
479,140
5,529,980
4,553,591
1,455,101
4,157,030
381,912
217,160,615

113,495
17,221
153,596
596,187
54,561
13,997
151,125
186,102
58,021
134,311
13,453
7,947,033

208

3.26
4.21
3.50
3.61

491,235
388,740
1,304,117
120,548

14.24
11.36
10.62
11.50

3.64
547,869
2.98
95,900
3.45
746,789
3.82 3,117,609
3.41
206,328
2.92
55,506
2.73
656,641
4.09
612,370
3.99
315,794
3.23
451,538
3.52
54,777
3.66 33,898,201

13.66
12.70
13.13
14.95
9.24
9.12
9.28
10.39
17.46
8.42
11.09
12.05

APPENDIX B
Table B.1: Characteristics of Social Heterogeneity (2000)
State
Population
Number of
%
Number of
Non-White
ForeignPersons
Born
Persons
Alabama
4,447,100
1,284,292 28.88
77,478
Alaska
626,932
192,398 30.69
32,905
Arizona
5,130,632
1,257,021 24.50
564,906
Arkansas
2,664,155
534,379 20.06
63,030
California
33,871,648
13,701,589 40.45
7,993,001
Colorado
4,301,261
741,256 17.23
319,974
Connecticut
3,405,565
625,210 18.36
336,357
Delaware
783,600
198,827 25.37
39,935
District of
572,059
395,958 69.22
67,693
Columbia
Florida
15,982,378
3,517,349 22.01
2,441,330
Georgia
8,186,453
2,859,172 34.93
504,287
Hawaii
1,211,390
917,326 75.73
196,430
Idaho
1,293,953
116,649
9.01
54,290
Illinois
12,419,293
3,293,822 26.52
1,372,426
Indiana
6,080,485
760,463 12.51
165,285
Iowa
2,926,324
177,684
6.07
77,047
Kansas
2,688,418
374,474 13.93
115,892
Kentucky
4,041,769
400,880
9.92
70,072
Louisiana
4,468,976
1,612,815 36.09
106,749
Maine
1,274,923
38,909
3.05
33,501
Maryland
5,296,486
1,905,178 35.97
467,016
Massachusetts
6,349,097
981,811 15.46
706,096
Michigan
9,938,444
1,972,391 19.85
464,429
Minnesota
4,919,479
519,197 10.55
215,941

209

%

Number of
Jewish
Persons

%

Number of
Muslim
Persons

%

1.74
5.25
11.01
2.37
23.60
7.44
9.88
5.10
11.83

9,100
3,525
81,675
1,600
994,000
72,000
108,280
13,500
25,500

0.20
0.56
1.59
0.06
2.93
1.67
3.18
1.72
4.46

7,670
1,381
11,857
2,044
259,762
14,855
29,647
3,691
60,479

0.17
0.22
0.23
0.08
0.77
0.35
0.87
0.47
10.57

15.28
6.16
16.22
4.20
11.05
2.72
2.63
4.31
1.73
2.39
2.63
8.82
11.12
4.67
4.39

628,485
93,500
7,000
1,050
270,000
18,000
6,400
14,500
11,350
16,500
8,290
216,000
275,000
110,000
42,000

3.93
1.14
0.58
0.08
2.17
0.30
0.22
0.54
0.28
0.37
0.65
4.08
4.33
1.11
0.85

31,661
38,882
609
363
125,203
11,002
4,717
3,470
4,696
13,050
809
52,867
41,497
80,515
12,305

0.20
0.47
0.05
0.03
1.01
0.18
0.16
0.13
0.12
0.29
0.06
1.00
0.65
0.81
0.25

Table B.1 (Cont’d)
Mississippi
Missouri
Montana
Nebraska
Nevada
New
Hampshire
New Jersey
New Mexico
New York
North Carolina
North Dakota
Ohio
Oklahoma
Oregon
Pennsylvania
Rhode Island
South Carolina
South Dakota
Tennessee
Texas
Utah
Vermont
Virginia
Washington
West Virginia
Wisconsin
Wyoming
Total

2,844,658
5,595,211
902,195
1,711,263
1,998,257
1,235,786

1,098,559
847,128
84,966
178,002
496,371
48,935

38.62
15.14
9.42
10.40
24.84
3.96

35,724
132,099
14,699
63,221
281,775
48,767

1.26
2.36
1.63
3.69
14.10
3.95

1,400
62,315
850
7,100
77,100
10,020

0.05
1.11
0.09
0.41
3.86
0.81

3,919
19,359
614
3,115
2,291
3,782

0.14
0.35
0.07
0.18
0.11
0.31

8,414,350
1,819,046
18,976,457
8,049,313
642,200
11,353,140
3,450,654
3,421,399
12,281,054
1,048,319
4,012,012
754,844
5,689,283
20,851,820
2,233,169
608,827
7,078,515
5,894,121
1,808,344
5,363,675
493,782
281,412,514

2,309,645
604,793
6,082,768
2,244,657
49,019
1,707,687
822,220
459,776
1,796,851
157,128
1,316,452
85,440
1,125,973
6,052,315
240,194
19,619
1,958,405
1,072,298
89,567
593,818
39,112
69,960,748

27.45
33.25
32.05
27.89
7.63
15.04
23.83
13.44
14.63
14.99
32.81
11.32
19.79
29.03
10.76
3.22
27.67
18.19
4.95
11.07
7.92
24.86

1,340,911
130,259
3,538,160
373,156
10,297
307,295
115,420
252,418
457,661
108,202
104,013
11,172
139,100
2,549,836
135,441
20,689
512,409
538,423
17,774
166,309
10,134
27,901,434

15.94
7.16
18.64
4.64
1.60
2.71
3.34
7.38
3.73
10.32
2.59
1.48
2.44
12.23
6.06
3.40
7.24
9.13
0.98
3.10
2.05
9.91

468,000
10,500
1,653,870
25,545
730
142,255
5,050
31,625
283,000
16,100
11,000
350
18,250
128,000
4,500
5,810
76,140
43,500
2,400
28,230
430
6,141,325

5.56
0.58
8.72
0.32
0.11
1.25
0.15
0.92
2.30
1.54
0.27
0.05
0.32
0.61
0.20
0.95
1.08
0.74
0.13
0.53
0.09
2.18

120,724
2,604
223,968
20,137
902
41,281
6,145
5,224
71,190
1,827
5,761
50
18,464
114,999
3,645
100
51,021
15,553
1,528
7,796
263
1,559,294

1.43
0.14
1.18
0.25
0.14
0.36
0.18
0.15
0.58
0.17
0.14
0.01
0.32
0.55
0.16
0.02
0.72
0.26
0.08
0.15
0.05
0.55

210

APPENDIX C
Table C.1: Characteristics of Social Cohesion in States (2000)
State
Number of Divorced
%
Number of Index
Persons
Offenses*
Alabama
Alaska
Arizona
Arkansas
California
Colorado
Connecticut
Delaware
District of
Columbia
Florida
Georgia
Hawaii
Idaho
Illinois
Indiana
Iowa
Kansas
Kentucky
Louisiana
Maine
Maryland
Massachusetts
Michigan
Minnesota
Mississippi
Missouri
Montana
Nebraska
Nevada
New
Hampshire
New Jersey
New Mexico
New York
North
Carolina
North Dakota
Ohio

371,218
54,650
440,890
232,296
2,474,567
373,606
249,707
60,943
46,044

8.35
8.72
8.59
8.72
7.31
8.69
7.33
7.78
8.05

169,728
24,474
300,360
110,134
1,277,054
166,199
103,131
37,516
5,403

Crime
Rate
(100,000)
3,816.60
3,903.77
5,854.25
4,133.92
3,770.27
3,863.96
3,028.31
4,787.65
944.48

1,502,561
656,777
87,188
105,546
867,414
518,243
210,767
211,653
353,637
354,047
117,936
366,483
422,950
799,643
336,648
221,798
475,341
78,079
121,224
216,097
102,723

9.40
8.02
7.20
8.16
6.98
8.52
7.20
7.87
8.75
7.92
9.25
6.92
6.66
8.05
6.84
7.80
8.50
8.65
7.08
10.81
8.31

899,149
382,028
56,033
48,685
251,487
194,076
92,667
95,969
59,549
231,832
29,207
195,198
240,118
424,952
174,228
97,663
188,719
63,523
75,707
84,250
17,112

5,625.88
4,666.59
4,625.51
3,762.50
2,024.97
3,191.78
3,166.67
3,569.72
1,473.34
5,187.59
2,290.88
3,685.42
3,781.92
4,275.84
3,541.59
3,433.21
3,372.87
7,040.94
4,424.04
4,216.17
1,384.71

500,848
162,589
1,173,090
576,718

5.95
8.94
6.18
7.16

247,448
113,921
515,531
407,771

2,940.79
6,262.68
2,716.69
5,065.91

39,836
951,705

6.20
8.38

17,475
373,875

2,721.11
3,293.14

211

Table C.1 (Cont’d)
Oklahoma
315,452
9.14
160,668
4,656.16
Oregon
316,981
9.26
150,490
4,398.49
Pennsylvania
799,339
6.51
356,555
2,903.29
Rhode Island
79,251
7.56
20,510
1,956.47
South
290,037
7.23
233,631
5,823.29
Carolina
South Dakota
51,773
6.86
25,709
3,405.87
Tennessee
511,189
8.99
274,200
4,819.59
Texas
1,559,049
7.48
1,046,189
5,017.26
Utah
132,270
5.92
89,520
4,008.65
Vermont
51,570
8.47
24,991
4,104.78
Virginia
504,369
7.13
198,432
2,803.30
Washington
529,708
8.99
296,674
5,033.39
West Virginia
154,363
8.54
67,497
3,732.53
Wisconsin
383,233
7.14
165,751
3,090.25
Wyoming
45,481
9.21
22,443
4,545.12
Total
21,559,527
7.66
10,905,432
3,875.25
*As defined by the FBI, index offenses include: murder and non-negligent manslaughter,
forcible rape, robbery, aggravated assault, arson, burglary, larceny, and motor vehicle theft (FBI,
2010).

212

APPENDIX D
Table D.1: Characteristics of Residential Mobility in States (2000)
State
Total
Number of
%
Number of
Housing
Owner
Unoccupied
Units
Occupied
Housing Units
Housing Units
Alabama
1,963,711
1,258,705 64.10
226,631
Alaska
260,978
138,509 53.07
39,378
Arizona
2,189,189
1,293,556 59.09
287,862
Arkansas
1,167,995
720,399 61.68
129,084
California
12,214,549
6,546,334 53.59
711,679
Colorado
1,808,037
1,116,137 61.73
149,799
Connecticut
1,385,975
869,729 62.75
84,305
Delaware
343,072
216,038 62.97
44,336
District of
274,845
101,214 36.83
26,507
Columbia
Florida
7,302,947
4,441,799 60.82
965,018
Georgia
3,281,737
2,029,154 61.83
275,368
Hawaii
460,370
227,888 49.50
57,245
Idaho
527,824
339,960 64.41
58,179
Illinois
4,885,615
3,088,884 63.22
293,836
Indiana
2,532,319
1,669,162 65.91
196,013
Iowa
1,232,511
831,419 67.46
83,235
Kansas
1,131,200
718,703 63.53
93,309
Kentucky
1,750,927
1,125,397 64.27
160,280
Louisiana
1,847,181
1,125,135 60.91
191,128
Maine
651,901
370,905 56.90
133,701
Maryland
2,145,283
1,341,751 62.54
164,424
Massachusetts
2,621,989
1,508,052 57.52
178,409
Michigan
4,234,279
2,793,124 65.96
448,618
Minnesota
2,065,946
1,412,865 68.39
170,819
Mississippi
1,161,953
756,967 65.15
115,519
Missouri
2,442,017
1,542,149 63.15
247,423
Montana
412,633
247,723 60.03
53,966
Nebraska
722,668
449,317 62.17
56,484
Nevada
827,457
457,247 55.26
76,292
New
547,024
330,700 60.45
72,418
Hampshire
New Jersey
3,310,275
2,011,473 60.76
245,630
New Mexico
780,579
474,445 60.78
102,608
New York
7,679,307
3,739,166 48.69
622,447
North
3,523,944
2,172,355 61.65
391,931
Carolina
North Dakota
289,677
171,299 59.13
32,525
Ohio
4,783,051
3,072,522 64.24
337,278
213

%

11.54
15.09
13.15
11.05
5.83
8.29
6.08
12.92
9.64
13.21
8.39
12.43
11.02
6.01
7.74
6.75
8.25
9.15
10.35
20.51
7.66
6.80
10.59
8.27
9.94
10.13
13.08
7.82
9.22
13.24
7.42
13.15
8.11
11.12
11.23
7.05

Table D.1 (Cont’d)
Oklahoma
1,514,400
Oregon
1,452,709
Pennsylvania
5,249,750
Rhode Island
439,837
South
1,753,670
Carolina
South Dakota
323,208
Tennessee
2,439,443
Texas
8,157,575
Utah
768,594
Vermont
294,382
Virginia
2,904,192
Washington
2,451,075
West Virginia
844,623
Wisconsin
2,321,144
Wyoming
223,854
Total
115,899,421

918,259
856,951
3,406,337
245,156
1,107,617

60.64
58.99
64.89
55.74
63.16

172,107
118,986
472,747
31,413
219,816

11.36
8.19
9.01
7.14
12.53

197,940
1,561,363
4,716,959
501,547
169,784
1,837,939
1,467,009
553,699
1,426,361
135,514
69,812,617

61.24
64.00
57.82
65.26
57.67
63.29
59.85
65.56
61.45
60.54
60.24

32,963
206,538
764,221
67,313
53,748
205,019
179,677
108,142
236,600
30,246
10,423,220

10.20
8.47
9.37
8.76
18.26
7.06
7.33
12.80
10.19
13.51
8.99

214

APPENDIX E
Table E.1: Descriptive Statistics for Control Variables
State
Number
%
Median Number of
%
NonAge
Veterans
English
Speaking
AL
AK
AZ
AR
CA
CO
CT
DE
DC
FL
GA
HI
ID
IL
IN
IA
KS
KY
LA
ME
MD
MA
MI
MN

162,483
82,758
1,229,237
123,399
12,401,756
604,019
583,913
69,533
90,417
3,473,864
751,438
302,066
111,879
2,220,719
362,082
160,022
218,655
148,473
382,364
93,966
622,714
1,115,570
781,381
389,988

3.65
13.20
23.96
4.63
36.61
14.04
17.15
8.87
15.81
21.74
9.18
24.94
8.65
17.88
5.95
5.47
8.13
3.67
8.56
7.37
11.76
17.57
7.86
7.93

35.80
32.40
34.20
36.00
33.30
34.30
37.40
36.00
34.60
38.70
33.40
36.20
33.20
34.70
35.20
36.60
35.20
35.90
34.00
38.60
36.00
36.50
35.50
35.40

447,397
71,552
562,916
280,510
2,569,340
446,385
310,069
84,289
44,484
1,875,597
768,675
120,587
136,584
1,003,572
590,476
292,020
267,452
380,618
392,486
154,590
524,230
558,933
913,573
464,968

10.06
11.41
10.97
10.53
7.59
10.38
9.10
10.76
7.78
11.74
9.39
9.95
10.56
8.08
9.71
9.98
9.95
9.42
8.78
12.13
9.90
8.80
9.19
9.45

Number
with a
Bachelor’s
Degree
351,772
61,196
493,419
190,090
3,640,157
599,028
416,751
80,376
69,496
1,573,121
829,873
142,478
116,901
1,317,182
475,247
278,350
290,271
271,418
339,711
129,992
629,304
834,554
878,680
605,210

215

%

Number of
Christian
Residents

%

7.91
9.76
9.62
7.14
10.75
13.93
12.24
10.26
12.15
9.84
10.14
11.76
9.03
10.61
7.82
9.51
10.80
6.72
7.60
10.20
11.88
13.14
8.84
12.30

2,435,373
215,223
2,048,023
1,522,072
15,614,071
1,697,259
1,971,102
318,122
418,543
6,576,205
3,665,550
438,675
627,262
6,867,625
2,608,882
1,711,426
1,327,235
2,159,541
2,627,028
463,541
2,291,896
4,069,606
4,158,134
3,035,510

54.76
34.33
39.92
57.13
46.10
39.46
57.88
40.60
73.16
41.15
44.78
36.21
48.48
55.30
42.91
58.48
49.37
53.43
58.78
36.36
43.27
64.10
41.84
61.70

Number
Partic. in
2000
Presidential
Election
1,621,897
-1,532,016
909,546
10,965,856
1,741,368
1,459,525
327,622
201,894
5,814,074
2,376,665
367,871
501,621
4,601,446
2,088,046
1,308,519
1,072,218
1,544,187
1,521,377
651,817
1,686,895
2,702,984
4,142,171
2,331,221

%

36.47
-29.86
34.14
32.37
40.49
42.86
41.81
35.29
36.38
29.03
30.37
38.77
37.05
34.34
44.72
39.88
38.21
34.04
51.13
31.85
42.57
41.68
47.39

Table E.1 (Cont’d)
MS
95,522
MO
264,281
MT
44,331
NE
125,654
NV
427,972
NH
96,088
NJ
2,001,690
NM
616,964
NY
4,962,921
NC
603,517
ND
37,976
OH
648,493
OK
238,532
OR
388,669
PA
972,484
RI
196,624
SC
196,429
SD
45,575
TN
256,516
TX
6,010,753
UT
253,249
VT
34,075
VA
735,191
WA
770,886
WV
45,895
WI
368,712
WY
29,485
Total 46,951,180

3.36
4.72
4.91
7.34
21.42
7.78
23.79
33.92
26.15
7.50
5.91
5.71
6.91
11.36
7.92
18.76
4.90
6.04
4.51
28.83
11.34
5.60
10.39
13.08
2.54
6.87
5.97
16.68

33.80
36.10
37.50
35.30
35.00
37.10
36.70
34.60
35.90
35.30
36.20
36.20
35.50
36.30
38.00
36.70
35.40
35.60
35.90
32.30
27.10
37.70
35.70
35.30
38.90
36.00
36.20
--

249,431
592,271
108,476
173,189
238,128
139,038
672,217
190,718
1,361,164
792,646
61,365
1,144,007
376,062
388,990
1,280,788
102,494
420,971
79,370
560,141
1,754,809
161,351
62,809
786,359
670,628
201,701
514,213
57,860
26,402,499

8.77
10.59
12.02
10.12
11.92
11.25
7.99
10.48
7.17
9.85
9.56
10.08
10.90
11.37
10.43
9.78
10.49
10.51
9.85
8.42
7.23
10.32
11.11
11.38
11.15
9.59
11.72
9.38

194,325
507,892
100,758
179,181
158,078
153,874
1,063,665
154,372
1,954,242
808,070
67,551
1,016,256
297,082
369,252
1,153,383
110,175
351,526
73,563
478,463
1,996,250
213,959
74,124
835,011
704,826
109,651
530,268
47,066
28,317,440

216

6.83
9.08
11.17
10.47
7.91
12.45
12.64
8.49
10.30
10.04
10.52
8.95
8.61
10.79
9.39
10.51
8.76
9.75
8.41
9.57
9.58
12.17
11.80
11.96
6.06
9.89
9.53
10.06

1,554,483
2,893,159
403,492
1,006,860
685,119
589,022
4,858,756
1,057,828
11,461,411
3,651,416
470,112
5,102,269
2,096,476
1,071,287
7,116,698
665,170
1,908,638
511,886
2,905,619
11,573,549
1,668,851
238,251
2,943,551
1,942,850
650,016
3,241,659
230,725
141,367,057

54.65
51.71
44.72
58.84
34.29
47.66
57.74
58.15
60.40
45.36
73.20
44.94
60.76
31.31
57.95
63.45
47.57
67.81
51.07
55.50
74.73
39.13
41.58
32.96
35.95
60.44
46.73
50.23

994,184
1,588,679
410,997
697,019
608,970
569,081
3,187,226
598,605
6,782,220
2,911,262
285,788
4,701,998
1,234,229
1,533,968
4,912,185
408,783
1,127,003
316,269
2,076,181
6,407,637
770,754
294,308
1,540,003
2,487,433
648,124
2,568,653
218,351
101,350,746

34.95
28.39
45.56
40.73
30.48
46.05
37.88
32.91
35.74
36.17
44.50
41.42
35.77
44.83
40.00
38.99
28.09
41.90
36.49
30.73
34.51
48.34
21.76
42.20
35.84
47.89
44.22
36.02

APPENDIX F
Table F.1: Controlling for Anti-White Bias Crime (2000)
State
Total Bias Number of Number of Percent
Percent of
crime
Anti-Race Anti-White Anti-White Total
Alabama
0
0
0
0.00
0.00
Alaska
0
0
0
0.00
0.00
Arizona
248
165
11
6.67
4.44
Arkansas
4
3
0
0.00
0.00
California
1,988
1234
145
11.75
7.29
Colorado
109
74
6
8.11
5.50
Connecticut
169
104
14
13.46
8.28
Delaware
35
18
2
11.11
5.71
District of
5
3
0
0.00
0.00
Columbia
Florida
216
149
36
24.16
16.67
Georgia
35
23
1
4.35
2.86
Hawaii
0
0
0.00
0.00
Idaho
49
28
4
14.29
8.16
Illinois
192
121
15
12.40
7.81
Indiana
111
76
12
15.79
10.81
Iowa
33
25
1
4.00
3.03
Kansas
43
32
4
12.50
9.30
Kentucky
73
62
12
19.35
16.44
Louisiana
12
8
1
12.50
8.33
Maine
29
12
0
0.00
0.00
Maryland
210
150
22
14.67
10.48
Massachusetts
465
282
44
15.60
9.46
Michigan
439
327
92
28.13
20.96
Minnesota
170
122
24
19.67
14.12
Mississippi
2
2
0
0.00
0.00
Missouri
74
61
14
22.95
18.92
Montana
21
9
0
0.00
0.00
Nebraska
18
12
1
8.33
5.56
Nevada
97
48
7
14.58
7.22
New
33
11
1
9.09
3.03
Hampshire
New Jersey
652
401
36
8.98
5.52
New Mexico
16
5
1
20.00
6.25
New York
209
83
12
14.46
5.74
North
33
28
4
14.29
12.12
Carolina
North Dakota
5
5
0
0.00
0.00
Ohio
247
181
36
19.89
14.57
Oklahoma
83
55
9
16.36
10.84

217

Table F.1 (Cont’d)
Oregon
Pennsylvania
Rhode Island
South
Carolina
South Dakota
Tennessee
Texas
Utah
Vermont
Virginia
Washington
West Virginia
Wisconsin
Wyoming
Total

142
147
46
34

93
109
24
33

5
19
0
10

5.38
17.43
0.00
30.30

3.52
12.93
0.00
29.41

7
222
305
81
20
98
250
69
49
9
7,604

5
168
189
46
7
75
191
37
23
7
4,926

1
84
24
5
1
25
16
9
2
0
768

20.00
50.00
12.70
10.87
14.29
33.33
8.38
24.32
8.70
0.00
15.59

14.29
37.84
7.87
6.17
5.00
25.51
6.40
13.04
4.08
0.00
10.10

218

REFERENCES

219

REFERENCES

Mirriam-Webster's collegiate dictionary (9th ed.). (1991). Springfield, MA: Mirriam-Webster.
Adorno, T. W., Frenkel-Brunswik, E., Levinson, D. J., & Sanford, R. N. (1950). The
authoritarian personality. New York: Harper.
Agresti, A. (2002). Categorical data analysis. Hoboken, NJ: Wiley.
Allport, G. W. (1954). The nature of prejudice. Reading, MA: Addison-Wesley.
Altschiller, D. (1999). Hate crimes: A reference handbook. Santa Barbara, CA: ABC-CLIO.
Anderson, E. (1999). Code of the street: Decency, violence, and the moral life of the inner city.
New York: WW Norton.
Anderson, J. F., Dyson, L., & Brooks, W. (2002). Preventing hate crime and profiling hate crime
offenders. Western Journal of Black Studies, 26(3), 140.
Anti-Defamation League. (2001). 2001 audit of anti-Semitic incidents. Washington, DC: ADL.
Barnes, A., & Ephross, P. H. (1994). The impact of hate violence on victims: Emotional and
behavioral responses to attacks. Social Work, 39(3), 247.
Beck, A. T. (1999). Prisoners of hate: The cognitive basis of anger, hostility, and violence. New
York: Harper-Collins.
Beirne, P., & Messerschmidt, J. (2000). Criminology. Boulder, CO: Westview.
Bell, L. A. (2013). Theoretical foundations. In M. Adams, W. Blumenfeld, C. Castaneda, H.
Hackman, M. Peters, & X. Zuniga, Readings for diversity and social justice (3rd ed., pp.
21-26). New York, NY: Routledge.
Berrill, K. T. (1992). Anti-gay violence and victimization in the United States: An overview. In
G. M. Herek, & K. T. Berrill, Hate crimes: Confronting violence against lesbians and
gay men (pp. 19-45).
Blau, J. R., & Blau, P. M. (1982). The cost of inequality: Metropolitan structure and violent
crime. American Sociological Review, 47(1), 114-129.
Blumer, H. (1958). Race prejudice as a sense of group position. The Pacific Sociological Review,
1(1), 3-7.
Boekmann, R. J., & Turpin-Petrosino, C. (2002). Understanding the harm of hate. Journal of
Social Issues, 207-225.
220

Boyle, K. (2005). When the KKK ruled Detroit. Journal of Blacks in Higher Education, 47, 100.
Bufkin, J. L. (1999). Bias crime as gendered behavior. Social Justice, 26(1), 155.
Bursik, R. (1988). Social disorganization and theories of crime and delinquency: Problems and
prospects. Criminology, 26(4), 519-545.
Bursik, R., & Grasmick, H. (1993). Neighborhoods and crime: The dimemsions of effective
community control. New York: Lexington Books.
Byers, B. D., & Cryder, B. W. (2002). Hate crimes against the Amish: A qualitative analysis of
bias motivation using routine activities theory. Deviant Behavior, 23(2), 115-148.
Byers, B., Crider, B. W., & Biggers, G. K. (1999). Bias crime motivation: A study of hate crime
and offender neutralization techniques used against the Amish. Journal of Contemporary
Criminal Justice, 15, 78-96.
Campbell, D. (1965). Ethnocentric and other altruistics motives. In D. L. (Ed.), Nebraska
symposium on motivation (pp. 283-311). Lincoln, NE: University of Nebraska Press.
Chakraborti, N., & Garland, J. (2009). Hate crime: Impact, causes, and responses. Thousand
Oaks, CA: Sage.
Chen, T. (2000). Hate violence as border patrol: An Asian American theory of hate violence.
Asian Law Journal, 7, 69-101.
Cogan, J. C. (2001). Hate crime as a crime category worthy of policy attention. The American
Behavioral Scientist, 46(1), 173.
Coleman, J. (1988). Social capital in the creation of human capital. American Journal of
Sociology, 94, 95-120.
Comstock, G. D. (1991). Violence against lesbians and gay men. New York: Columbia
University Press.
Craig, K. M. (2002). Examining hate-motivated aggression: A review of the social psychological
literature on hate crimes as a distinct form of aggression. Aggression and Violence
Behavior, 7, 85-101.
Cullen, F., & Agnew, R. (2003). Criminological theory: Past to present. Los Angeles: Roxbury.
Disha, I., Cavendish, J. C., & King, R. D. (2011). Historical events and spaces of hate: Hate
crimes against Arabs and Muslims in post-9/11 America. Social Problems, 58(1), 21-46.
Dunworth, T., & Mills, G. (1999). National evaluation of Weed and Seed. U.S. Department of
Justice. Washington, DC: National Institute of Justice.
221

Ehrlich, H. J., Larcom, E. K., & Purvis, R. D. (1994). The traumatic effects of ethnoviolence.
Towson, MD: The Prejudice Institute.
Ezekiel, R. (1995). The racist mind: Portraits of American Neo-Nazis and Klansmen. New York:
Penguin.
Federal Bureau of Investigation. (1996). Uniform Crime Reports: Hate crime reporting statistics.
Washington, D.C.: Department of Justice.
Federal Bureau of Investigation. (2000). Uniform Crime Reports: Hate crime reporting statistics.
Washington, D.C.: Department of Justice.
Federal Bureau of Investigation. (2004). Uniform Crime Reports: Hate crime reporting statistics.
Washington, D.C.: Department of Justice.
Federal Bureau of Investigation. (2007). Uniform Crime Reports: Hate crime reporting statistics.
Washington, D.C.: Department of Justice.
Federal Bureau of Investigation. (2008). Uniform Crime Reports: Hate crime reporting statistics.
Washington, D.C.: Department of Justice.
Federal Bureau of Investigation. (2010). Uniform Crime Reports: Hate crime reporting statistics.
Washington, D.C.: Department of Justice.
Franklin, K. (2000). Antigay behaviors among young adults. Journal of Interpersonal Violence,
15(4), 339-362.
Franklin, K. (2002). Good intentions: The enforcement of hate crime penalty-enhancement
statutes. The American Behavioral Scientist, 46(1), 154.
Freilich, J. D., & Chermak, S. M. (2013). Hate Crimes. U.S. Department of Justice. Washington,
D.C.: Office of Community Oriented Policing Services.
Freilich, J. D., Adamczyk, A., Chermak, S. M., Boyd, K., & Parkin, W. S. (2013).
Disorganization, diversity and deadly far-right ideological violence: A county-level
analysis. nternational Forum on Contemporary Criminal Law: Contemporary
Punishment Reform . Beijing.
Gardner, W., Mulvey, E. P., & Shaw, E. C. (1995). Regression analyses of counts and rates:
Poisson, overdispersed Poisson, and negative binomial models. Psychological Bulletin,
118(3).
Gerstenfeld, P. (2002). A time to hate: Situational antecedents of intergroup bias. Analyses of
Social Issues and Public Policy, 61-67.

222

Gerstenfeld, P. (2004). Hate crimes: Causes, controls, and controversies. Thousand Oaks, CA:
Sage.
Gerstenfeld, P. (2011). Hate crimes: Causes, controls, and controversies (2 ed.). Thousand Oaks,
CA: Sage.
Green, D. P. (1998). White supremacist activity and crossburnings in North Carolina. Journal of
Quantitative Criminology, 14(3), 263-282.
Green, D. P., Adelson, R. P., & Garnett, M. (1999). The distinctive views of hate-crime
perpetrators and white supremacists. In D. Prentice, & D. Miller, Cultural divides:
Understanding and overcoming group conflict (pp. 429-464). New York: Rusell Sage
Foundation.
Green, D. P., Strolovitch, D. Z., & Wong, J. S. (1998). Defended neighborhoods, integration, and
racially motivated crime. The American Journal of Sociology, 104(2), 372.
Green, D., McFalls, L. H., & Smith, J. K. (2001). Hate crime: An emergent research agenda.
Annual Review of Sociology, 27, 479.
Gruenewald, J. (2011). A comparative examination of homicides perpetrated by far-right
extremists. Homicide Studies, 15(2), 177-203.
Haider-Markel, D. P. (2002). Regulating hate: State and local influences on hate crime law
enforcement. State Politics & Policy Quarterly, 2(2), 126.
Hamm, M. (1994). Hate crime: International perspectives in causes and control. Cincinnati,
OH: Anderson.
Henry, J. S. (2009). Hate crime laws: A critical assessment. In B. Perry, & B. Levin, Hate
crimes: Understanding and defining hate crime (pp. 183-203). Westport, CT: Praeger.
Herek, G. M. (1989). Hate crimes against lesbians and gay men: Issues for research and policy.
American Psychologist, 44(6), 948-955.
Herek, G. M., & Berrill, K. T. (1992). Hate crimes: Confronting violence against lesbians and
gay men. Newbury Park, CA: Sage Publications.
Herek, G. M., Cogan, J. C., & Gillis, J. R. (2002). Victim experiences in hate crimes based on
sexual orientation. Journal of Social Issues, 58, 319-339.
Hindelang, M. J., Hirschi, T., & Weis, J. G. (1979). Correlated of delinquency: The illusion of
discrepancy between self-reports and official measures. American Sociological Review,
44(6), 995-1014.

223

Hovland, C., & Sears, R. (1940). Minor studies in aggression: Correlations of lynchings with
economic indices. Journal of Psychology, 9, 301-310.
Iganski, P. (2001). Hate crimes hurt more. The American Behavioral Scientist, 45(4), 626.
Jackson, K. (1967). The Ku Klux Klan in the city: 1915-1930. New York: Oxford University
Press.
Jacobs, D., & Wood, K. (1999). Interracial conflict and interracial homicide: Do political and
economic rivalries explain white killings og blacks or black killings of whites? American
Journal of Sociology, 105, 157-190.
Jacobs, J. (1998). The emergence and implications of American hate crime jurisprudence. In R.
K. (Eds.), Hate crime: The global politics of polarization (pp. 150-176). Carbondale, IL:
Southern Illinois University Press.
Jacobs, J. B., & Potter, K. A. (1997). Hate crimes: A critical perspective. Crime and Justice, 22,
1-50.
Jacobs, J. B., & Potter, K. A. (1998). Hate crimes: Criminal law & identity politics. Oxford:
Oxford University Press.
Jacobs, J., & Henry, J. (1996). The social construction of a hate crime epidemic. The Journal of
Criminal Law and Criminology, 86(2), 366-391.
Jenness, V., & Broad, K. (1997). Hate crimes: New social movements and the politics of
violence. New York: Aldine deGruyter.
Karapin, R. (1996). Explaining the surge in right-wing violence by German youth. New York:
Hunter College.
Karsada, J., & Janowitz, M. (1974). Community attachments in mass society. American
Sociological Review, 39, 328-339.
Kitsuse, J., & Cicourel, A. (1963). A note on the uses of official statistics. Social Problems, 11,
131-139.
Kleck, G. (1982). On the use of self-report data to determine the class distribution of criminal
and delinquent behavior. American Sociological Review, 47, 427-433.
Koopmans, R. (1996). Explaining the rise of racist and extreme right violence in Western
Europe: Grievances or opportunities? European Journal of Political Research, 30, 185216.
Kornhauser, R. R. (1978). Social sources of delinquency. Chicago: The University of Chicago
Press.
224

Levin, J., & McDevitt, J. (1993). Hate crimes: The rising tide of bigotry and bloodshed. New
York: Plenum.
Levine, R. A., & Campbell, D. T. (1972). Ethnocentrism. New York: Wiley.
Lieberman, J. D., Arndt, J., Personius, J., & Cook, A. (2001). Vicarious annihilation: The effect
of mortality salience on perceptions of hate crimes. Law and Human Behavior, 25(6),
547.
Lipsky, M. (1980). Street-level bureaucracy: The dilemmas of the individual in the public
services. New York: Russell Sage Foundation.
Lowencamp, C., Cullen, F., & Pratt, T. (2003). Replicating Sampson and Groves's test of social
disorganization theory: Revisiting a criminological classic. Journal of Research in Crime
and Delinquency, 40(4), 351-373.
Lyons, C. J. (2007). Community (dis)organization and racially motivated crime. American
Journal of Sociology, 113(3), 815-863.
Maltz, M. D., & Targonski, J. (2002). A note on the use of county-level UCR data. Journal of
Quantitative Criminology, 18(3), 298-318.
Maxfield, M. G., Weiler, B. L., & Widom, C. S. (2000). Comparing self-reports and official
records of arrests. Journal of Quantitative Criminology, 16(1), 87-110.
Maxfield, M., & Babbie, E. (2011). Research methods for criminal justice and criminology.
Belmont, CA: Wadsworth.
McDevitt, J., & Williamson, J. (2003). Hate crimes directed at gay, lesbian, bisexual, and
transgendered victims. In W. Heitmeyer, & J. Hagan, International handbook of violence
research (pp. 801-815). Boston: Kluwer.
McDevitt, J., Balboni, J. M., Bennett, S., Weiss, J. C., Orchowsky, S., & Walbolt, L. (2000).
Improving the quality and accuracy of bias crime statistics nationally: An assessment of
the first ten years of crime crime data collection. Boston: Northeastern University.
McDevitt, J., Balboni, J., Garcia, L., & Gu, J. (2001). Consequences for victims: A comparison
of bias and non-bias-motivated assaults. American Behavioral Scientist, 45(4), 697-711.
McDevitt, J., Levin, J., & Bennett, S. (2002). Hate crime offenders: An expanded typology.
Journal of Social Issues, 58(2), 303-317.
McPhail, B. A., & DiNitto, D. M. (2005). Prosecutorial perspectives on gender-bias hate crimes.
Violence against Women, 11(9), 1162-1185.

225

McVeigh, R., Bjarnason, T., & Welch, M. (2003). Hate crime reporting as a successful social
movement outcome. American Sociological Review, 68(6), 843.
Messner, S. F., McHugh, S., & Felson, R. B. (2004). Distinctive characteristics of assaults
motivated by bias. Criminology, 42(3), 585.
Michigan Legislature. (2011). Retrieved 2011 йил 26-November from The Michigan Penal
Code: www.legislature.mi.gov
Miller, B., & Humphries, L. (1980). Lifestyles and violence: Homosexual victims of assault and
murder. 3(3), 169-185.
Morenoff, J. D., Sampson, R. J., & Raudenbush, S. W. (2001). Neighborhood inequality,
collective efficacy, and the statial dynamics of urban violence. Criminology, 39, 517-560.
National Gay and Lesbian Task Force. (2009). Retrieved 2012 йил 5-March from Hate crime
law in the U.S.:
http://www.thetaskforce.org/downloads/reports/issue_maps/hate_crimes_7_09_color.pdf
Noelle, M. (2001). The ripple effect of the Matthew Shepard murder: Impact on the assumptive
worlds of members of targeted groups. The American Behavioral Scientist, 46(1), 27.
Nolan, J. J., Akiyama, Y., & Berhanu, S. (2002). The Hate Crime Statistics Act of 1990:
Developing a method for measuring the occurrence of hate violence. The American
Behavioral Scientist, 46(1), 136.
Osgood, D. W. (2000). Poisson-based regression analysis of aggregate crime rates. Journal of
Quantitative Criminology, 16(1), 21-43.
Park, R. (1952). Human communities. Glencoe, IL: The Free Press.
Park, R. E., & Burgess, E. W. (1925). The city. Chicago: The University of Chicago Press.
Perry, B. (2001). In the name of hate: Understanding hate crime. New York: Routledge.
Perry, B. (2003b). Where do we go from here? Researching hate crime. Internet Journal of
Criminology.
Pew Forum on Religion and Public Life. (2008). U.S. religious landscape survey. Washington,
DC: Author.
Pinkney, A. (1994). Lest we forget: White hate crimes. Chicago, IL: Third World Press.
Piquero, A. R., MacIntosh, R., & Hickman, M. (2002). The validity of a self-reported
delinquency scale: Comparisons across gender, age, race, and place of residence.
Sociological Methods and Research, 30(4), 492-529.
226

Piza, E. L. (2012). Rutgers Center on Public Security. Retrieved 2014 йил 1-April from Using
Poisson and negative binomial regression models to measure the influence of risk on
crime incident counts: www.riskterrainmodeling.com
Porter, J. R., Rader, N. E., & Cossman, J. S. (2012). Social disorganization and neighborhood
fear: Examining the intersection of individual, community, and county characteristics.
American Journal of Criminal Justice, 37, 229-245.
Pratt, T., & Cullen, F. (2005). Assessing macro-level predictors and theories of crime: A metaanalysis. Crime and Justice, 32, 373.
Rayburn, N. R., & Davison, D. C. (2002). Articulated thoughts about antigay hate crimes.
Cognitive Therapy and Research, 26(4), 431-447.
Rose, S. M., & Mechanic, M. B. (2002). Psychological distress, crime features, and help-seeking
behaviors related to homophobic bias incidents. The American Behavioral Scientist,
46(1), 14.
Sampson, R. (1986). Neighborhood family structure and the risk of personal victimization. In R.
Sampson, & J. Byrne, The social ecology of crime. New York: Springer-Verlag.
Sampson, R. (2003). The neighborhood context of well-being. Perpsective in Biology and
Medicine, 46(3), S53-S64.
Sampson, R. (2003). The neighborhood context of well-bring. Perspective in Biology and
Medicine, 46(3), S53-S64.
Sampson, R. (2012). Great American city: Chicago and the enduring neighborhood effect.
Chicago: The University of Chicago Press.
Sampson, R. J., & Groves, W. B. (1989). Community structure and crime: Testing socialdisorganization theory. The American Journal of Sociology, 94(4), 774-802.
Sampson, R. J., & Wilson, W. J. (1990). Toward a theory of race, crime, and urban inequality. In
J. Hagan, & R. Peterson, Crime and inequality. Palo Alto, CA: Stanford University Press.
Sampson, R. J., Raudenbush, S., & Earls, F. (1997). Neighborhoods and violence crime: A
multilevel study of collective efficacy. Science, 277, 918-924.
Schuman, H., Steeh, C., Bobo, L., & Krysan, M. (1997). Racial attitudes in America: Trends and
interpretations (rev. ed.). Cambridge, MA: Harvard University Press.
Shaw, C., & McKay, H. (1942). Juvenile delinquency in urban areas. Chicago: The University
of Chicago Press.

227

Sidanius, J., & Pratto, F. (1999). Social dominance: An intergroup theory of social hierarchy and
oppression. New York: Cambridge University Press.
Skogan, W. (1986). Fear of crime and neighborhood change. In A. J. Reiss, & M. Tonry,
Communities and crime. Chicago: The University of Chicago Press.
Skogan, W. (2006). Police and community in Chicago: A tale of three cities. New York: Oxford
University Press.
Stanton, B. (1992). Klanwatch: Bringing the Ku Klux Klan to justice. New York: Mentor Books.
Steinberg, A., Brooks, J., & Remtulla, T. (2003). Youth hate crimes: Identification, prevention,
and intervention. The American Journal of Psychiatry, 160, 979-989.
Sternberg, R. J., & Sternberg, K. (2008). The nature of hate. New York: Cambridge University
Press.
Sugrue, T. J. (2005). The origins of the urban crisis: Race and inequality in postwar Detroit.
Princeton, NJ: Princeton University Press.
Suttles, G. D. (1972). The social construction of communities. Chicago: The University of
Chicago Press.
Torres, S. (1999). Hate crimes against African Americans: The extent of the problem. Journal of
Contemporary Criminal Justice, 15(1), 48-63.
Tsesis, A. (2000). The empirical shortcomings of first amendment jurisprudence: A historical
perspective on the power of hate speech. Santa Clara Law Review, 40, 729-780.
Tsesis, A. (2003). Conceptualizing bias crimes: A social and theoretical perspective. Jaw &
Society, 28(1), 315.
Vito, G., Maahs, J. R., & Holmes, R. (2006). Criminology: Theory, research and policy. Boston:
Jones & Bartlett.
Wang, L. (1994). Hate crimes laws. New York: Clark, Boardman, Callaghan.
Warner, B. (2003). The role of attenuated culture in social disorganization theory. Criminology,
41(1), 73-97.
Widick, B. (1972). Detroit: City of race and class violence. Chicago: Quadrangle Books.
Wilson, J. Q., & Kelling, G. (1982 йил March). Broken windows. The Atlantic Monthy, pp. 2939.

228

Wilson, M. S., & Ruback, R. B. (2003). Hate crimes in Pennsylvania, 1984-99: Case
characteristics and police responses. Justice Quarterly, 20(2), 373.
Yang, A. (1997). Trends: Attitudes toward homosexuality. Public Opinion Quarterly, 61, 477507.

229