<<

The Effect of Race, Place, and Time on Police Use of Force:

How Social Context Influences Legal Decision-Making

Dissertation

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University

By

Charles Anthony LoFaso, M.A.

Graduate Program in Sociology

The Ohio State University

2020

Dissertation Committee:

Professor Ryan D. King, Advisor

Professor Paul E. Bellair

Associate Professor Hollie Nyseth Brehm

Copyright by

Charles Anthony LoFaso

2020

Abstract

Compared to the research investigating police use of force at the encounter-level, there are relatively fewer studies examining how neighborhood context influences the decision to use force. This dissertation adds to the research on neighborhood context through the investigation of two overarching research questions.

I first examine whether neighborhood racial composition and degree of disadvantage are associated with the frequency and severity of force after encounter-level variables are controlled. Second, I examine whether the frequency or severity of force change following incidents of violence by the police or against the police. Using an interrupted time series design for this question, the study analyzes whether the trajectory of force by

Rochester (N.Y.) officers was altered following the death of Michael Brown in Ferguson,

Missouri on August 9, 2014, and by the death of RPD officer Daryl Pierson on

September 3, 2014. As such, the dissertation examines the extent to which the risk of being subjected to police use of force is influenced by where and when a citizen encounters an officer. More broadly, the dissertation examines how social context influences legal decision-making, as well as the role that law, as governmental social control, plays in preserving social order.

. Findings indicate that both the frequency and severity of force are higher in neighborhoods with larger percentages of Black and Hispanic residents once encounter- level variables, degree of neighborhood disadvantage, and rates are controlled. A likely explanation for this finding is that police officers, like many Americans, including other actors in the criminal justice system, are acting on an implicit racial bias that

ii characterizes Black individuals as being prone to criminality and violence. Officers perceive that disadvantaged neighborhoods present elevated threats to officer safety because more Black individuals live there. Consequently, implicit racial bias results in more frequent and severe social control in these neighborhoods. While race and place are both contextual influences on the use of force, the effect of concentrated disadvantage is indirect and is mediated by neighborhood racial composition.

A primary finding of the interrupted time series analysis is that the frequency of force did not change during the 25-day period following Michael Brown’s death but did significantly increase in the 25-day period immediately after Officer Pierson’s death.

Officer Pierson’s death motivated officers to use force more often during confrontations with citizens, whether out of fear for their safety, in bounded solidarity, or as vicarious retribution for Pierson’s shooting. Thereafter, as negative and anti-police backlash increased in the U.S. following several more high-profile police shootings and in-custody deaths of unarmed Black males, officers significantly reduced their use of force through the end of 2015. However, officers were more punitive, as the odds of being subjected to more severe levels of force significantly increased even as the frequency of force declined. Thus, although officers entered a phase of “de-policing,” they likely remained fearful for their safety or retained a desire for retribution during citizen encounters and used excessive force to ensure a favorable outcome.

iii

Dedication

Over the past 144 years, fifteen officers of the Rochester Police Department were killed in the line of duty. Gunfire claimed the first, Patrolman Louis Gomenginger, on

July 3, 1876, and the fifteenth, Officer Daryl Pierson, on September 3, 2014. I had the privilege of knowing and working with Officer Pierson. Daryl was a highly productive police officer, and his work unquestionably made the Rochester community safer.

In addition to being an exceptional police officer, Daryl was also an exceptional young man. After high school, he joined the U.S. Army and served in Afghanistan before being honorably discharged. Upon returning to his hometown of East Rochester, New

York, Daryl married, started a family, and joined the Rochester Police Department.

Service was a theme of his tragically short life. Today, due to the advocacy of Sen.

Chuck Schumer and the late Rep. Louise Slaughter, and after President Obama granted their request, the post office in East Rochester is named in his honor.

In addition to his local post office, Daryl’s name has been added to a wall at the

National Police Memorial in Washington, D.C. There, Daryl’s name, and his End of

Watch date are etched alongside the names and EOWs of other fallen heroes who gave their lives for their communities across the United States. Every year during National

Police Week in May, officers from all over the world come to Washington, D.C. to gather at the National Police Memorial and pay tribute to the fallen. As a final tribute from his peers, the Major Unit’s conference room in the city of Rochester’s Public Safety

Building is named for Daryl. To honor his sacrifice for the citizens of Rochester, I add my own tribute by dedicating this dissertation to Daryl Pierson.

iv

Acknowledgements

I begin by thanking my advisor, Dr. Ryan King, for his mentorship and support during the four years since I arrived in Columbus, Ohio to begin graduate school. I am beyond grateful for the advice and encouragement that he generously provided during my graduate career. As my advisor, Ryan chaired my master’s committee, my candidacy exam committee, and now my dissertation committee. I readily acknowledge that his willingness to invest a significant amount of time in this role and his devotion to my success are the reasons I am completing the graduate program. Along the way, I appreciated his great sense of humor, as it made our interactions informal in a way that facilitated learning.

I next thank Dr. Paul Bellair. Although I never took a course from Paul, he has guided me during every milestone in the graduate program: as a member of my thesis committee, as a member of my candidacy exam committee, and now as a member of my dissertation committee. Paul also generously shared his time, advice, and encouragement.

As a student interested in neighborhood effects on homicide clearance and now police use of force, I have particularly benefitted from his expertise in neighborhood processes and social disorganization theory. His very substantive comments on this dissertation were thought provoking and improved the final product. I will be forever grateful to him for his investment in my research and willingness to invest his time and serve on my committees every time I asked.

I next thank Dr. Hollie Nyseth Brehm. I had the great fortune of taking two classes from Hollie, including Teaching Sociology, a course that made me a significantly better independent instructor. Hollie served on my candidacy exam committee and now my

v dissertation committee. Hollie also generously shares her time, an example of which was when she volunteered to review my practice answers to questions from previous candidacy exams. Her encouragement and constructive comments made me realize that I would pass and put me considerably at ease. Her detailed review of my draft dissertation chapters makes this dissertation a significantly better study.

The members of my committee helped me transition from “raw” graduate student to scholar. I learned from them how to evaluate and apply theory to generate research questions, use social science methods to investigate them, and interpret the findings to answer important questions that interest me. No one achieves any significant goal in life without having the assistance of caring people. I have achieved a life-long goal of earning a Ph.D. because of the assistance I received from the members of my committee and am grateful for the role they played in my scholarly development at Ohio State.

I also want to thank my non-Ohio State mentors, Dr. Jesenia Pizarro and Dr. William

Terrill, both of . Because of my interest in homicide research, I contacted Jesy when she and Bill were on the faculty at Michigan State University and I was a student in the online master’s program there. Jesy is a nationally recognized expert on homicide and embraced me, then a practitioner, as one of her academic “criminology kids.” I have always been honored to be one. She invited me to work on an article with her to enhance my prospects for acceptance in graduate school. With her encouragement and assistance, I collected 15 years of homicide data from the Rochester Police

Department, which formed the basis of my master’s thesis and the two articles that appear on my Vita page. Bill, who will tell you that his most important role is as Jesy’s husband and father to their son William, is a nationally recognized use of force expert.

vi

When in casual conversation I told Bill about how the RPD documents force in a form they call the “Subject Resistance Report,” he remarked that this data could form the basis of interesting research projects. It is thus fair to say he was the inspiration for my request to collect the use of force data that forms the basis of this dissertation. The astute reader will also note frequent references to Bill’s scholarly work in Chapters 2 and 3. I thank them for their acceptance of me, before I even applied to graduate school, as someone worthy of their time and advice. I am beyond fortunate that they did.

I next thank my wife, Lorrie. Her support over our 39 years of marriage has been unwavering, including when I left public accounting to attend law school and then when I made the shocking announcement that I wanted to give up a career in law to become a police officer. Lorrie readily embraced our move to Columbus and was excited about that new journey we would be undertaking together. It has been a wonderful life because of her unconditional love and support, which has made me a better person and helped me achieve goals that would not have been possible without her.

Finally, I want to thank the friends at Ohio State with whom I laughed and commiserated about the frustrations of graduate school. To Courtney DeRoche, Brandon

Moore, Joe Guzman, Jake Terrance, Laura Frizzell, Sadé Lindsay, Chris Kleps, Lora

Phillips, Amelia Li, Mary McKay, Nico Pinchak, Peter Choi, and Eric LaPlant, thanks for making the journey as much fun as finally reaching the destination. I appreciate the support and advice each of you provided along the way and wish you all much success in your scholarly careers.

vii

Vita

May 1978……………………A.B., Indiana University

August 1983…………………M.S., Northeastern University

May 1991……………………J.D., University at Buffalo, State University of New York

August 2013…………………M.S., Michigan State University

May 2018……………………M.A., The Ohio State University

1999-2016…………………...Recruit Officer, Police Officer, and Investigator,

Rochester Police Department, Rochester, New York

2016-2017……………………Recipient, the Mildred and Simon Dinitz Graduate

Fellowship in Criminology, Department of Sociology

The Ohio State University

2017 to 2020…………………Graduate Teaching Associate, Department of Sociology

The Ohio State University

Publications

LoFaso, Charles A. 2020. “Solving Homicides: The Influence of Neighborhood Characteristics and Investigator Caseload.” Criminal Justice Review, 45:84-103.

Pizarro, Jesenia, William Terrill, and Charles A. LoFaso. 2020. “The Impact of Investigative Strategies and Procedures on Homicide Clearance.” Homicide Studies, 24:3-24.

Fields of Study

Major Field: Sociology

viii

Table of Contents

Abstract ...... ii

Dedication ...... iv

Acknowledgements ...... v

Vita ...... viii

List of Tables...... x

List of Figures ...... xi

Chapter 1: Introduction ...... 1

Chapter 2: Use of Force: Theory, Legal Requirements, and the Continuum ...... 12

Chapter 3: How Neighborhoods Influence the Use of Force: Theory and Evidence ...... 41

Chapter 4: The Influence of Antecedent Events on Police Use of Force ...... 81

Chapter 5: Data, Methods, and Results ...... 110

Chapter 6: Discussion and Conclusions...... 144

References ...... 164

Endnotes ...... 180

ix

List of Tables

Table 1: Levels of Resistance and Associated Levels of Force…………………….40

Table 2: Factor Loadings………………………………………………………….136

Table 3: Neighborhood Racial Composition………………………………………136

Table 4: Descriptive Statistics……………………………………………………..137

Table 5: Negative Binomial Regression, Models 1, 2, and 3...……….……………138

Table 6: Multilevel Logistic Regression, Models 4 and 5…………………………139

Table 7: Multilevel Logistic Regression, Models 6 and 7…………………………140

Table 8: Interrupted Time Series Analysis, Models 8 and 9……………………….141

Table 9: Interrupted Time Series Analysis, Model 10……………………………..142

x

List of Figures

Figure 1: Age Range of Subjects…………………………………………………….143

Figure 2: 5-day Smoothed Average Force Counts……………………………….….143

xi

Chapter 1: Introduction

Through their mandate to enforce the law and ensure public safety, the police ideally represent order and justice in society (Wolfe & Piquero, 2011). As society’s most visible agents of formal social control, the police are also unique among governmental actors. By law, they are permitted to use situationally justified physical force, including lethal force, to fulfill their occupational mandate.

“Force” has no precise definition. Terrill and Reisig (2003, p. 299) define force as

“acts that threaten or inflict physical harm on suspects.” In contrast, the Rochester (N.Y.)

Police Department (“RPD”), the agency that provided the data for this study, defines force as any “intentional physical strength or energy exerted or brought to bear upon or against a person for the purpose of compulsion, constraint or restraint” (RPD General

Order (“G.O.”) 335. Regardless of how it is defined, Friedrich (1980, p. 82) notes that police use of force “is theoretically important because it involves the execution of perhaps the essential function of the state and practically important because it affects the public’s attitudes and behaviors toward the police and government more generally.” The authority to use coercion against citizens is thus central to the role of police officers and distinguishes policing from virtually every other profession (Bittner, 1970).

Despite its centrality to their occupational role, use of force by police officers is an infrequent event (Bayley & Garofalo, 1989; Eith & Durose, 2011; Terrill & Reisig, 2003;

Worden & Catlin, 2002). Police-citizen contacts are estimated to occur roughly 44 million times annually in the United States (Hyland, Langton, & Davis, 2015). Yet, in

2008, the use or verbal threat of force was estimated to have occurred in only 1.4% of those encounters with citizens age 16 or older, a percentage that was essentially

1 unchanged from 2002 or 2005 (Eith & Durose, 2011). Use of force is more probable when the police attempt to make an arrest and the subject resists, which results in use of force in 20% of all arrests (Hickman, Piquero, & Garner, 2008). Significantly, most force incidents are relatively minor and end quickly. For example, in 2008, 53.5% of police use of force cases involved the subject being pushed or grabbed (Eith & Durose, 2011).

Regardless of how infrequently it may occur, the use of force is still an important societal issue due to the potential for police to abuse their state-sanctioned authority and violate civil liberties. Use of force incidents now receive significant scrutiny inside and outside of police agencies. Exercise of the state’s enforcement power is now likely to be captured by cell phone or surveillance video, which may go “viral,” especially when the force employed inflicts serious injuries or loss of life. While millions may see these videos, most observers have neither the training nor experience to evaluate whether the force used is reasonable under the circumstances of the encounter. Nonetheless, police coercion engenders vociferous public condemnation when it is perceived, rightly or wrongly, as excessive, and hence legally indefensible.

Of concern to the public is who is being subjected to police use of force. A 2014

Gallup poll found that 68% of Blacks and 25% of whites believe Black individuals are treated unfairly by both the police and the criminal justice system (Gallop, 2014). The perception of unfairness is not surprising considering the history of contentious relations between the police and minority residents of disadvantaged communities. While research on the presence of racial disparities in the use of force is mixed (as discussed in Chapter

2), violence by the police is nonetheless “especially serious and unsettling when the citizens involved are minorities” (Bayley & Garofalo, 1989, p. 1).

2

Aggressive policing of disadvantaged communities generates negative public perceptions of the police, which can hinder citizen cooperation upon which police depend and thereby reduce police effectiveness (Freiburger, 2019; Sierra-Arévalo, 2019).

Moreover, citizens who intentionally refuse to defer to police authority make it more likely that a use of force will follow. As Sierra-Arévalo observes, the “misuse of force, in addition to physically harming the public, violates public expectations of fair impartial legal agents and reduces the legitimacy of police” (2019, p. 422).

Misuse of force can thus influence how residents view the law and their obligation to obey it (Anderson, 1999; Sampson & Bartusch, 1998). If people obey the law when they perceive its enforcement by the police and courts to be fair, impartial, and unbiased

(Tyler, 1990), targeting minorities for arrest or the use of force weakens support for law and creates within the community a cultural orientation called “legal cynicism” (Kirk &

Papachristos, 2011). Legal cynicism is defined as the belief that “the law and the agents of its enforcement, such as the police and courts, are … illegitimate, unresponsive, and ill equipped to ensure public safety” (Kirk & Papachristos, 2011, p. 1191). Simply stated, being subjected to coercive treatment because of who they are and where they reside, causes some minority residents of disadvantaged neighborhoods to perceive the police as biased, indifferent to their life circumstances, and unwilling or unable to respond to their problems and ensure their safety.

Legal cynicism has very real societal consequences. Because it constrains the legal choices available to residents in disadvantaged neighborhoods, legal cynicism can lead to the use of violence as a solution to everyday problems (Kirk & Papachristos, 2011,

3 pp. 1202-1203). When some residents of poor, segregated neighborhoods conclude that law is unavailable or unresponsive to their problems, they adapt and use violence as a specific situational response to interpersonal issues (Kirk & Papachristos, 2011, p. 1203).

In effect, violence is chosen out of necessity as a form of “self-help” because legal actors such as the police are perceived as unavailable and ineffective (Black, 1983; Kirk &

Papachristos, 2011, p. 1203).

In addition to fostering violence, legal cynicism has other detrimental societal consequences. Desmond, Papachristos, and Kirk (2016, p. 857) find that an unjustified beating of a Black citizen by the police led to a significant reduction in one of the most basic forms of citizen engagement: calling 911 for matters of personal and public safety.

Kirk and Matsuda (2011) also find that perceptions of the police as racially biased and unresponsive are associated with the conscious choice to allow crimes to go unreported to authorities and suspects to remain on the streets.

While non-lethal use of force against minorities is controversial and has negative societal consequences, police use of deadly force is arguably the most controversial aspect of their work, having emerged as a symbol of the gross inequities in the criminal justice system. Civil rights activists, as well as many members of the public, are particularly energized and enraged when force causes the death of an unarmed racial or ethnic minority. Those comparatively rare incidents receive widespread attention and condemnation, as was powerfully illustrated by some responses to highly publicized killings of Michael Brown, Eric Garner, and Freddie Gray. Peaceful public , calls for reform, property damage through violent civil unrest (in Ferguson and ), and even retributive violence against the police (in Dallas) resulted from these incidents.

4

It is thus undeniable that use of force is highly consequential for the public’s perception of the police, police-community relations, and police effectiveness. Scholars have identified police violence and its far-reaching consequences as a social problem and have generated a large body of research investigating the correlates of police-citizen interactions and use of force. In general, research at the encounter-level between police and citizens can be summarized around four overlapping domains: subject characteristics

(Engel, Klahm, & Tillyer, 2010; Engel, Sobol, &Worden, 2000); officer characteristics

(McCluskey, Terrill, & Paoline, 2005; McElvain & Kposowa, 2008; Paoline & Terrill,

2004; Paoline & Terrill, 2007; Rydberg & Terrill, 2010); situational characteristics

(Garner, Maxwell, & Heraux, 2002; Terrill, 2003; Terrill, 2005, Terrill & Mastrofski,

2002); and departmental characteristics, including police culture (Alpert & MacDonald,

2001; Paoline, 2004; Terrill, Leinfelt, & Kwak, 2008; Terrill, Paoline, & Manning, 2003;

Wolfe & Piquero, 2011). These studies examining use of force at the encounter-level have significantly increased our understanding of the demographic, situational, and organizational variables that are statistically associated with police use of force.

Klinger (1997) observes that most research analyzing police behavior has focused on situational characteristics while ignoring the ecological context. Indeed, compared to the voluminous studies examining use of force at the encounter-level, there are relatively fewer studies examining the ecological context in which suspects and officers are embedded and whether use of force is distributed differentially across neighborhoods and influenced by neighborhood context. Neighborhoods have criminological consequences, as studies utilizing a social disorganization framework establish that disadvantaged

5 neighborhoods tend to have higher rates of violent crimes, including homicide (Kirk &

Papachristos, 2011; Krivo & Peterson, 1996; Sampson, Raudenbush & Earls, 1997).

Police officers work daily in disadvantaged, high-crime neighborhoods. The behavioral choices they make while working in such neighborhoods, including decisions regarding the use of force, are likely influenced by the unique social-structural conditions found there. Klinger (1997) argues that understanding the ecological context in which police work could explain how and why police behavior varies across neighborhoods.

Thus, the sparseness of research on neighborhood context when examining police coercion is consequential because it inhibits a more thorough understanding of all correlates of police behavior.

Some researchers who have examined neighborhood context indeed find that it influences force outcomes. While the literature on neighborhood context will be discussed in greater detail in Chapter 3, it is worth mentioning at this juncture two prominent examples along the lines of race and place. Using data from the Police

Services Study, which collected observational data in Rochester, N.Y., St. Louis, M.O., and Tampa-St. Petersburgh, FL. in 1977, Smith (1986) finds that officers were significantly more likely to use force on Black suspects, but that this effect became non- significant when neighborhood context was considered. Smith (1986, p. 332) thus finds that “coercive authority is not influenced by the race of the individual suspect per se but rather the racial composition of the area in which the encounter occurs.”

In their research utilizing data from the Project on Policing Neighborhoods, an observational study of police-citizen encounters in Indianapolis and St. Petersburg,

Florida, Terrill and Reisig (2003) also incorporated neighborhood context, which in their

6 study included an index of census tract concentrated disadvantage and census tract homicide rates. Terrill and Reisig find that the effect of race became non-significant once concentrated disadvantage and area homicide rate were incorporated into a full model.

Their finding that race at the encounter level is confounded by neighborhood context is thus equivalent to what Smith (1986) finds using the Police Services data. However, while Smith finds racial composition of the neighborhood is significantly related to the use of force, Terrill and Reisig find poverty and crime are the significant contextual factors.

These findings suggest that police use of coercion is strongly related to the ways in which social context influences legal decision-making. Accordingly, sociological theories explaining the effects of race or place may also explain the decision to employ coercion.

However, a focus on race or place alone, while potentially theoretically relevant as an explanation for police behavior, ignores “the temporal embeddedness of social interactions” (Legewie, 2016, p. 380) and the possibility that antecedent events can trigger changes in the frequency and severity of force used against a particular group.

Events can increase perceptions of threat and reinforce preexisting implicit stereotypes and prejudices.

For example, Legewie (2013) finds that terrorist attacks negatively affect short-term perceptions of immigrants. He argues that events such as terrorist attacks foster the attitude that immigrants, as an out-group, are threatening, which adds a temporal component to group threat and conflict theories (Legewie, 2013). Applying the timing of events as a previously overlooked contextual factor in evaluating police use of force,

Legewie (2016) finds that police use of force against Blacks by the NYPD increased

7 significantly following two fatal shootings of police officers by Black suspects. He argues that racial bias in policing is not static but fluctuates. A significant event, such as the murder of a police officer, increases internal cohesion among police officers, provokes intergroup conflict and activates preexisting, implicit racial stereotypes that portray Black individuals as violent criminals. Specifically, Legewie suggests that officers may respond to “behavior that went unnoticed or without consequences before a certain event with the use of physical force after the event” (2016, p. 387) (emphasis in the original).

The literature on neighborhood context thus highlights the importance of race, place, and (while overlooked prior to Legewie’s study), time, as explanations for police behavior, including the use of force. Using data from the RPD in Rochester, New York, this dissertation adds to the growing body of research examining contextual explanations for police use of force. In Rochester, use of force at the encounter-level is documented in a Subject Resistance Report (“SRR”) completed by the RPD officer(s) who used force.

The study is based on 3,368 SRRs completed for incidents that occurred from October

2011 through June 2016. Each of the SRRs was merged with census tract data reflecting racial composition, degree of concentrated disadvantage, percent of residents on public assistance, percent of unemployed residents, and the average crime rate over the study period for the census tract in which the use of force occurred.

The study examines two overarching research questions. The first research question examines whether neighborhood context, as measured by racial composition and degree of economic disadvantage, is associated with the severity and frequency of force used by the police (after controlling for encounter-level variables). Individual uses of force are

8 nested within census tracts that serve as proxies for neighborhoods. The null hypothesis is that both the level of force and the number of uses of force are related only to encounter- level variables such as subject race, sex, age, degree of impairment, and level of resistance. Conversely, if, after controlling for these encounter-level variables, use of force is significantly related to a neighborhood’s racial composition or degree of disadvantage, it is arguable that officers are responding to individuals based not only on the nature of the encounter but also upon characteristics of the neighborhood in which they are encountered.

The second research question examines the contextual effect of time, specifically whether there is a change in the severity of force or frequency of force following a violent antecedent event by the police or against the police. Regarding the violent event by the police, the study uses as its factual predicate the fatal shooting of Michael Brown in Ferguson, MO on August 9, 2014. The violent event against the police is based on the fatal shooting of RPD officer Daryl Pierson on September 3, 2014. Using an interrupted time series design, the study analyzes whether there was a change in the frequency of force and the severity of force by RPD officers after the shootings of Michael Brown and

Officer Pierson. The dissertation thus makes a unique contribution to the use of force literature by examining how violent incidents against or by the police may influence police behavior in their aftermath.

Studies analyzing neighborhood variation in use of force typically draw on discrimination-based theories such as racial threat or racial bias (while controlling for disadvantage), or alternatively focus on the effect of structural factors such as measures of poverty (while controlling for race) to explain variation in use of force. Less common

9 are studies that test both racial and social disorganization perspectives (see, e.g.,

Lautenschlager & Omori, 2018, for one such study). Further, it appears that no studies combine the effects of race, place, and time. The dissertation thus uniquely adds to the use of force literature by testing not only racial and neighborhood structural explanations for use of force, but also the effect of time based on antecedent events. In short, the dissertation’s research questions examine the extent to which the risk of being subjected to higher levels of police force are influenced by where and when a citizen encounters an officer. By incorporating all three contextual factors, this study goes beyond previous research examining only one or two contexts to explain police use of force and extends previous work that focused only on where discrimination takes place to when discrimination occurs.

To examine the research questions, the study draws on theories of social control along lines of race, place, and time. Specifically, the effect of race is based on racial threat theory (see, e.g., Bentele & O’Brien, 2013; Blalock, 1967; Chamlin, 1989; Legewie &

Fagan, 2016) and Black’s morphology (i.e., the horizontal distance from the periphery of society), both of which will be discussed in detail in Chapter 3. The effect of place, which does not rely on minority threat arguments, draws on social disorganization theory

(specifically the effect of concentrated disadvantage) (see, e.g., Sampson & Groves,

1989; Sampson et al. 1997) and Werthman and Piliavin’s (1967) ecological contamination thesis. The contextual effect of time also relies on racial threat as well as

Portes and Sensenbrenner’s (1993) concept of “bounded solidarity” and a social- psychological model of intergroup aggression known as “vicarious retribution” (Lickel,

10

Miller, Stenstrom, Denson, & Schmader, 2006). In addition, the effect of fear on police behavior is explored as an explanation for how force is distributed.

The remainder of the dissertation is organized as follows. The second chapter addresses foundational issues regarding the use of force. In that chapter, theories explaining use of force are discussed. Legal guidelines under New York State law and

Supreme Court caselaw interpreting the U.S. Constitution are then examined. The chapter then reviews the use of force continuum and the nexus between force options and level of resistance encountered. Finally, how use of force is operationalized in the literature and in this dissertation is reviewed. Chapter 3 addresses theoretical foundations and the associated literature that generates the study’s first two hypotheses. The chapter discusses racial threat theory, the morphology dimension in Black’s Behavior of Law, social disorganization theory, and Werthman and Piliavin’s ecological contamination thesis.

Chapter 4 addresses the nexus between the timing of antecedent events, i.e., the lethal use of force by a police officer and the murder of a police officer, and subsequent uses of force by the police. The chapter draws on the racial threat theory discussed in Chapter 3, as well as Portes and Sensenbrenner’s concept of bounded solidarity and the social- psychological concept of vicarious retribution described by Lickel et al. (2006). Chapter

5 presents data, methods, and results. Chapter 6 discusses study findings and the conclusions to be drawn from the findings.

11

Chapter 2: Use of Force: Theory, Legal Guidelines, and the Continuum

Introduction: The Multiple Functions of Police and the Necessity for Coercion

Police officers fulfill multiple roles in society through different types of policing, all of which bring police in contact with citizens. These include “traditional policing,” which characterizes the police as crime fighters who react to calls for assistance from the public

(Greene, 2000, p. 310). In this role, the police have a “narrow range of interventions” and thus must enforce the law through the application of their primary legal sanction, i.e., arrest or the threat of arrest (Greene, 2000, p. 310). The police also provide order maintenance, which includes proactive activities such as “stop, question, and frisk” of citizens (Braga et al. 1999). Order maintenance policing, also known as “zero tolerance,” originates from Wilson and Kelling’s “broken windows” thesis (Greene, 2000). When fulfilling this role, police proactively address relatively minor order problems, such as street-level prostitution or drug use, to deter criminals from engaging in more serious criminal behavior (Greene, 2000).

A third role is fulfilled through problem-oriented policing, in which police focus on interventions to address neighborhood problems (Goldstein, 1990). Under this focus, “the problem, not the criminal law, becomes the defining characteristic of policing” (Eck,

1993, p. 63). The police utilize the “SARA” approach to problems such as crime hot spots, which requires them to scan, analyze, respond, and assess (Greene, 2000). A fourth, and more recent role, is community policing, in which the police partner with citizens to address multiple issues, including crime prevention, fear of crime, and quality of life issues (Greene, 2000, p. 312). Greene (2000, p. 315) notes that community

12 policing and problem-oriented police overlap in that both are engaged in “problem definition and discussions about interventions.”

Considering the multiple functions they fulfill in society, Manning (1978, p. 95) was no doubt correct when he noted that the police occupational mandate “embodies a vast and unmanageable social domain.” To accomplish their myriad functions, officers, as the most visible agents of formal social control, rely on the authority given them under state law. However, citizens will, at times, resist lawful police authority. For this reason, state law authorizes police to use reasonable physical force to overcome resistance by citizens, which can occur during any of the functions police fulfill. What follows are theoretical justifications for use of force, as well as legal guidelines stipulating when force may be used and what amount of force may lawfully be applied.

Theories Explaining Use of Force

As discussed above, force may be required periodically for the police to fulfill their various occupational roles. Three primary theoretical orientations have been proffered to explain the decision to use force: sociological, individual, and organizational (Friedrich,

1980; Terrill & Mastrofski, 2002). Sociological orientations are hypothesized in the literature to have the strongest influence on the use of force response, followed by individual and organizational perspectives. A fourth perspective – ecological/neighborhood context – is comparatively understudied and, as the subject of this dissertation, is discussed in Chapter 3.

A. The Sociological Orientation

The sociological orientation “is based on the premise that police behavior is influenced by the social dynamics of police-citizen encounters” (Worden, 1995, p. 32).

13

This orientation posits two distinct aspects of social dynamics (Terrill & Mastrofski,

2002). One of these is that force is related solely to what the citizen does, while the other suggests that force varies by subject traits, i.e., who the citizen is (Terrill & Mastrofski,

2002, p. 217). Legally (as discussed below), officers may only respond to what a citizen does, as reflected by situational factors at the encounter-level. Simply stated, police coercion must be applied only when situationally justified. Extra-legal factors – subject traits (or who the citizen is) – cannot legally justify the decision to use force.

Accordingly, the police are, at least legally, prohibited from taking subject traits (e.g., race) into account when using coercion against citizens.

Situational Factors: Research at the encounter-level has examined the situational factors that are likely to produce a force outcome. These include, among others, level of subject resistance, subject demeanor, degree of subject impairment, and subject possession of a weapon.

Resistance: Resistance during the police-citizen encounter, particularly when the police are attempting to make an arrest, is consistently associated with the decision to use force (Adams, 1999; Terrill & Mastrofski, 2002). Garner et al. (2002), for example, find that suspects who physically resist the police during an investigation or arrest increase the odds of being subjected to police force by 1800%.

Subject Demeanor/Judgmentally Impaired: Research examining the influence of subject demeanor on use of force is mixed, with some finding that disrespectful citizens were more likely to be subjected to force (Engel et al. 2000; Kaminski, Digiovanni, &

Downs, 2004), while others find no association between demeanor and use of force

(McCluskey & Terrill, 2005). Judgmentally impaired subjects (i.e., under the influence of

14 alcohol or drugs) are more likely to have force used against them (Bayley & Garofalo,

1989; Friedrich, 1980; Kaminski et al. 2004).

Weapons: A subject in possession of a weapon, particularly a firearm, is significantly more likely to be subjected to deadly force (e.g., White, 2002), while Kaminski and colleagues (2004) and Paoline and Terrill (2007) find that the presence of a weapon also increases the odds of being subjected to non-lethal force. Kaminski and colleagues

(2004), for example, find that the possession of a weapon by the subject increases the odds of high force rather relative to low force by 226%.

Subject Traits: Research at the encounter-level is mixed regarding the influence of citizen race on the decision to use force, but more consistent on the effects of sex and age.

Race/Ethnicity: Some empirical evidence finds that non-white citizens had a greater probability of being subjected to police use of force (Eith & Durose, 2011; Kaminski et al. 2004; Shuck, 2004; Terrill & Mastrofski, 2002), while other studies find that race or ethnicity is not significantly associated with use of force (Engel et al. 2000; McCluskey

& Terrill, 2005; Terrill, 2005).

Sex and Age: A consistent finding is that male suspects are significantly more likely than female suspects to be subjected to force during encounters with the police (Garner et al. 2002; McCluskey & Terrill, 2005; Terrill & Mastrofski, 2002; Terrill & Reisig, 2003).

Similarly, force is less likely to be used against older subjects relative to younger subjects

(McCluskey & Terrill, 2005; Paoline & Terrill, 2007; Terrill & Mastrofski, 2002; Terrill

& Reisig, 2003).

15

B. The Individual Orientation

The individual orientation examines variation among officers in their behavioral predispositions (Worden, 1995). Like the sociological orientation, the individual orientation also posits two distinct explanations for use of force (Terrill & Mastrofski,

2002). The first is based on officer characteristics and background, while the other is based on an officer’s attitudes and views (Terrill & Mastrofski, 2002, p. 218).

Collectively, these explanations seek to determine whether “there [is] something about particular officers that helps explain why they use force” (Terrill & Mastrofski, 2002, p.

218).

Officer Characteristics: Research at the encounter-level has examined whether officer characteristics are associated with the use of force. Officer characteristics include race/ethnicity, sex, age, experience, education, and attitudes toward police work.

Race/Ethnicity: Studies examining officer race/ethnicity consistently find no significant association between an officer’s race/ethnicity and the use of force

(Friederick, 1980; Garner, Schade, Hepburn & Buchanan, 1995; McCluskey & Terrill,

2005; Paoline & Terrill, 2004, 2007; Terrill & Mastrofski, 2002; Worden, 1995). An exception is Sun and Payne (2004), who found that Black officers were more likely than white officers to use force, although this association became nonsignificant once neighborhood characteristics were modeled.

Sex and Age: Similarly, studies have generally found no association between officer sex and the use of force (Kaminski et al., 2004; McCluskey & Terrill, 2005; Paoline &

Terrill, 2007; Sun & Payne, 2004; Terrill & Mastrofski, 2002; Worden, 1995). Studies finding the opposite include Garner et al. (2002) and Kop and Euwema (2001), who

16 found that male officers were more likely than female officers to use force despite there being no sex differences in attitudes toward the use of force. Research on the association of age and use of force is limited, although one would expect younger officers to be more willing to engage in a physical confrontation because “violence . . . is a young man’s game” (Adams, 1999, p. 503). Garner et al. (2002) find that older officers were less likely to use force as well as less severe levels of force, while McElvain and Kposowa (2008) find that the use of deadly force decreased as officers aged.

Experience and Education: On the issue of experience, studies report mixed findings, with some finding more experienced officers are less likely to use force, largely because they are more adept at reading situations and perceive that force should not be relied on to resolve common problems but used only as a last resort (Kop & Euwema,

2001; Paoline & Terrill, 2007; Terrill & Mastrofski, 2002). Others find no or an inconsistent association between experience and the use of force (Garner, Buchanan,

Schade, and Hepburn, 1996; Lawton, 2007; McCluskey & Terrill, 2005; Sun & Payne,

2004; Worden, 1995). Research on the effect of education is generally supportive of a negative association between educational attainment and the use of force (Paoline &

Terrill, 2007; Terrill & Mastrofski, 2002). In their study, McElvain and Kposowa (2008) find that college-educated officers were less likely to use deadly force than officers with no college education, while Hayden (1981) and Sherman and Blumberg (1981) find no association between educational attainment and use of deadly force. Worden (1995, p.

54) finds that officers with college degrees were somewhat more likely to use force, but that officers with bachelor’s degrees were somewhat less likely to use improper force.

17

The Effect of Attitudes: The second explanation under the individual orientation examines the effect of officer attitudes on the decision to use force. Two studies in this area are noteworthy.

Muir (1977) examined officer attitudes along two dimensions: their outlook on human nature and their moral attitudes regarding the use of coercive authority. Muir argues that officers who have little empathy and view the use of coercion as an appropriate social control mechanism are more likely to resort to the use of force in encounters with citizens.

Worden (1995, p. 34) posits that officers with attitudinal propensities to use force “(a) conceive the police role in narrow terms, limited to crime-fighting and law enforcement,

(b) believe that this role is more effectively carried out when officers may use force at their discretion, and (c) regard the citizenry as unappreciated at best and hostile and abusive at worst.” Using the Police Services Study data, Worden finds, like Muir, that officers with more negative attitudes toward citizens were significantly more likely to use both reasonable and unreasonable force (1995, p. 54). Officers’ attitudes toward the use of force were only moderately significant, such that “officers with more positive attitudes toward the use of force tend to use force improperly with greater frequency”

(1995, p. 54).

C. The Organizational Orientation

The organizational orientation posits that police use of force is influenced by the organizational setting, as reflected in departmental policy and the department’s informal organizational culture (Friedrich, 1980; Skolnick & Fyfe, 1993).

18

Departmental Policy: Research on departmental policy has focused primarily on the use of deadly force. Research consistently finds that administrative rules that set clear, restrictive boundaries for the use of deadly force as well as the effective review of deadly force incidents are associated with reductions in deadly force incidents (Meyer, 1980,

Sherman, 1983; Walker, 1993).

However, White (2001) illustrates how even restrictive department policies can be undermined by the personal policies of department leaders. Using an interrupted time series analysis of data from , White (2001) analyzed the effect on deadly force incidents following two significant administrative policy changes: the removal in

1974 of a restrictive deadly force policy and the reinstatement in 1980 of essentially the same prior restrictive policy abolished in 1974. White finds a significant increase in police shootings after removal of the policy, but that these increases were due to an increase in non-elective police shootings (those against assailants threatening the officer with a weapon). Significantly, White also finds a slight, non-significant decrease in elective shootings (those against either a physically assaultive or non-assaultive suspect).

White concluded from the pattern of elective shootings that officers were not following the restrictive policy prior to its removal, largely because the police commissioner’s personal policy and philosophy (i.e., “bust their heads”) permitted the use of deadly force freely and did not sanction officers despite the existence of a formal restrictive policy

(2001, p. 145). Notably, by 1980 department leadership had changed, and White finds a significant decrease in deadly force incidents of all types after implementation of the

1980 restrictive policy.

19

Using the Police Services Study data, Worden (1995) examined the effect of department characteristics on non-lethal force. Worden (1995, p. 57) finds that departments with higher levels of bureaucratization, measured by the number of full-time employees, levels of hierarchy or vertical differentiation, degree of specialization, and the extent to which the departments are civilianized, were significantly associated with the likelihood that reasonable, but not unreasonable force, will be used by officers working in those departments.

Department Culture: As discussed above, White (2001) finds that use of force within a police department may be guided by the informal organizational culture created by those holding power. Skolnick and Fyfe (1993) provide another example using the aggressive policing styles of Chiefs William Parker and Darryl Gates of the Los Angeles Police

Department, which created a subculture that condoned brutality and encouraged combative, proactive police work.

In addition to those holding power, there may exist within the rank and file an occupational culture separate and apart from formal administrative policies. Occupational culture has been defined as “the attitudes, values, and norms that are transmitted and shared among groups of individuals in an effort to collectively cope with the common problems and conditions members face” (Paoline & Terrill, 2014, p. 5). In the traditional, monolithic view of police culture, officers are depicted as “socially isolated, distrustful, and suspicious of their primary clientele” and approach police work in “crime fighting terms” (Paoline & Gau, 2018, p. 671). In this model, most officers embrace a “crime- fighting, warrior” role, in which they attempt to avoid punitive supervisors and trust only their peers with whom they share daily experiences on the street (Paoline & Gau, 2018,

20 pp. 671-672). Terrill et al. (2003) find that officers who embrace this crime-fighter, warrior orientation were more likely to use coercion than those officers who possessed nontraditional, more diverse cultural attitudes that included positive views of citizens and an approach to police work that does not emphasize aggressive crime fighting.

Terrill and colleagues’ findings have important implications for the association between informal cultural within police departments and the use of force. Other empirical research observes that police agencies have diversified demographically, and community policing has become a prominent feature of the philosophy of policing, two developments that are expected to produce cultural variation within police departments (Paoline, 2003).

Paoline (2004) finds that while a small number of officers continue to identify with

“traditionalist” attitudes, the majority possess more diverse cultural attitudes, including positive views of citizens. Their philosophy of policing emphasizes the role of police as guardians, not warriors. These developments could affect officer attitudes about the use of force, as “guardians” may be expected to use force less often than “warriors.”

To summarize, sociological, individual, and organizational orientations have been posited and empirically tested to explain why officers use force. Regardless of their motivation to use force, officers are bound by legal requirements imposed by both state law and the U.S. Constitution on when force can be used, as well as what amount of force is reasonable under the circumstances. What follows is a review of these legal requirements.

Legal Authority Provided Under State Law to Use Force When Necessary

State law across the United States authorizes police to use force when necessary to effect arrests and to defend themselves or others against physical aggression. In New

21

York, for example, Penal Law Article 35, § 35.30 (1),1 permits use of force by an officer when the officer reasonably believes force is necessary to effect an arrest. The officer must also reasonably believe that the subject upon whom force is used committed an

“offense,” which in New York is a traffic infraction, violation, misdemeanor, or felony

(Penal Law Article 10.00, § 10.00 (1)). Article 35, § 30.30 (1), also permits a police officer to use force against a person in self-defense or to defend a third person from what he or she reasonably believes to be the use or imminent use of unlawful physical force by that person.

The most extreme legal authorization for force is found in Article 35, § 35.15 (1) &

(2) (a), which permits an actor (including a police officer) to use deadly physical force2 against a person in self-defense or in defense of a third person if the actor reasonably believes the person is using or is about to use deadly force. “Deadly force” is defined as

“physical force which, under the circumstances in which it is used, is readily capable of causing death or other serious physical injury” (i.e., an injury that creates a substantial risk of death) (Penal Law § 10.00 (11)). An actor other than a police officer may not use deadly force if the actor knows that he or she can retreat from the encounter with complete personal safety (§ 35.15 (2) (a)). Thus, New York is not a “stand your ground” state such as Florida, which does not require a citizen to retreat where possible when confronted with deadly force.

New York also permits an officer to use deadly force to effect an arrest under certain circumstances, including if he or she reasonably believes the subject is resisting arrest for a felony and is armed with a firearm or deadly weapon (§ 35.30 (1) (b)), or the subject committed or attempted to commit kidnapping, , escape in the first degree, or

22 burglary in the first degree (§ 35.30 (1) (a) (ii)). It should be noted that RPD policy does not permit use of deadly force to effect an arrest under the conditions enumerated in

Article 35. Department policy dictates that RPD officers may use deadly force only when

“the use of deadly physical force is necessary to defend the member or another person from what the member reasonably believes to be the use or imminent use of deadly physical force” (RPD G.O. 340).

The Legal Authority to Use Force: The Reasonableness Requirement

The U.S. Supreme Court has held that restraint of a person by force constitutes a seizure of that person within the meaning of the 4th Amendment (Garner v. Tennessee,

475 U.S. 1, 1985). As discussed below, the Court has also held that when the seizure occurs must be reasonable under the circumstances (Garner, 475 U.S., at 8).

Consequently, all states mandate, pursuant to the U.S. Constitution, that the restraint of a person through the use force must be reasonable under the circumstances. This requirement is implemented in New York through Article 35. A common element to all

Article 35 statutes is the requirement that police officers “reasonably believe” that force, whether non-lethal or lethal, is necessary under the circumstances, including the necessity to defend themselves or a third person against what officers “reasonably believe” to be the use or imminent use of the same level of force.

There are thus two “reasonable belief” standards that must be satisfied to justify the use of force. The first (the subjective standard) requires a reasonable belief by the actor that force is necessary. This requirement is satisfied when he or she reasonably believes that force is being used or is about to be used by another person against the actor. The second (the objective standard) requires that any reasonable person in the same position

23 as the actor would evaluate the same facts and circumstances and draw the same conclusions, i.e., that force is necessary because another person is using or is about to use force against the actor. If, based on the facts, no reasonable person in the same circumstances could possibly conclude that some level of physical force was required, then the use of force was unlawful.

Article 35’s statutory scheme has been approved by the state’s highest court, the New

York Court of Appeals, particularly as applied to deadly force. In People v. Goetz, 68

N.Y.2d 96 (1986), the court affirmed that the necessity of deadly force must be objectively reasonable under the facts and circumstances of the incident. The Court noted that, in the absence of an objective standard, citizens could set their own standards for the permissible use of force, “no matter how aberrational or bizarre [their] thought patterns”

(68 N.Y.2d, at 111). In other words, interpreting the reasonableness standard as requiring that the actor’s belief be reasonable only to him and not to a reasonable person in the same situation “would also allow a legally competent defendant suffering from delusions to kill or perform acts of violence with impunity, contrary to fundamental principles of justice and criminal law” (68 N.Y.2d, at 111). Thus, for the use of physical force to be lawful, the actor’s subjective belief that force is necessary cannot be the sole predicate for the use of physical force. The reasonableness standard also requires consideration of the facts and circumstances that would persuade a reasonable person in the same situation that physical force was necessary.

To comply with constitutional and state law, the RPD provides the following guidance to its sworn officers concerning the existence of reasonable belief and the authority to use deadly force:

24

Reasonable belief exists when both of the following subjective and objective conditions are met:

1. The member reasonably believes another person is using or is about to use deadly physical force, and that it is necessary for the member to use deadly physical force to defend himself or another person, and

2. Evidence or information which appears reliable discloses facts circumstances [sic] which are collectively of such weight and persuasiveness as to convince a member of ordinary intelligence, training, judgment, and experience that another person is using or is about to use deadly physical force, and that it is necessary for the member to use deadly physical force to defend himself or another person.

(RPD G.O. 340).

Thus, a member’s subjective belief that another person is using or is about to use deadly force must be based on objective facts or circumstances that would also convince a reasonable officer in the same situation that a deadly force response was necessary.

U.S. Constitutional Standards for Determining the Appropriate Level of Force

New York State statutes provide standards from which a factfinder could evaluate whether the use of lethal or non-lethal physical force was necessary under the circumstances facing the actor. Notably, N.Y. statutes do not explicitly address how to determine the appropriate level of force. While officers are given the authority to use force, the force used must be reasonable under the circumstances (NIJ, 2019; Terrill,

2005). Consequentially, a determination under state law that the use of force was necessary under the circumstances facing the actor does not mean that the level of force used by the actor was also reasonable under the circumstances. An officer could use a level of force that exceeds what is objectively required under the circumstances facing the officer. Hence, it is important to define reasonable versus excessive force.

25

As Terrill and Mastrofski (2002) observe, excessive force is not easily defined. The

U.S. Commission on Civil Rights (2018, p. 9) agrees, stating that “there is no single, definitive answer for how much force is ‘necessary’ in any given situation.” Thus, a central issue in the use of force by police “is how to define the boundary between when force may be necessary or reasonable, versus when it becomes excessive” (U.S.

Commission on Civil Rights, 2018, p. 9). The U.S. Supreme Court has provided guidance on where this boundary may lie. In two landmark cases, Tennessee v. Garner, 471 U.S. 1

(1985) and Graham v. Connor, 490 U.S. 386 (1989), the Court concluded that the propriety of any seizure of a person by the police (which includes a use of force) must be made with reference to the Fourth Amendment’s reasonableness requirement, which, in part, guarantees citizens the right to be secure in their persons against unreasonable seizures.

A. Garner

At issue in Garner was the constitutionality of a Tennessee statute and a Memphis

Police Department policy permitting law enforcement officers to use deadly force to stop a fleeing felon. On the evening of October 3, 1974, Memphis police officers responded to a residence for a possible burglary in progress. In the backyard of the residence, the officers encountered Edward Garner, a Black, 15-year-old eighth grader who was crouched by a chain link fence at the edge of the yard. With the aid of a flashlight,

Officer Elton Hymon observed no weapon in Garner’s hands and was “reasonably sure” and “figured” that Garner was unarmed (471 U.S., at 3). Hymon yelled “police, halt” and took a few steps toward Garner, who then began to climb the fence to elude the police

(471 U.S., at 4). Garner, who was then shot by Hymon in the back of the head, later died

26 on a hospital operating table. Ten dollars and a purse taken from inside the house were located on his body (471 U.S., at 4).

Hymon was not charged for killing Garner as he acted pursuant to the Tennessee

“fleeing felon” statute and Memphis Police Department policy, which permitted the use of deadly force in cases of burglary. Garner’s father brought a civil action pursuant to 42

U.S.C. § 1983 in the U.S. District Court for the Western District of Tennessee. Section

1983 permits an injured party to bring a civil action for violations of Constitutional rights by any person acting under color of state law. Garner’s father argued that the murder of his son violated the Fourth, Fifth, Sixth, Eighth, and Fourteenth Amendments of the U.S.

Constitution (471 U.S., at 5).

Writing for the Court, Justice White began by noting a seizure under the Fourth

Amendment occurs whenever an officer restrains a person’s freedom to walk away. By extension, killing a fleeing suspect is also a “seizure’ under the Fourth Amendment and is therefore constitutional only if “reasonable” (471 U.S., at 7). Reasonableness under the

Fourth Amendment “depends on not only when a seizure is made, but also how it is carried out” (471 U.S., at 8). Justice White observed that the use of deadly force to prevent the escape of all felony suspects, even those who pose no immediate threat to the officer or others, is constitutionally unreasonable.

In this case, the facts were uncontroverted that Garner was unarmed and did not pose a significant threat of death or serious physical injury to the officers or others.

Accordingly, in a 6 to 3 decision, the Court held that the Tennessee statute was unconstitutional insofar as it permitted the use of deadly force in all circumstances. In this case, the use of deadly force against Garner, a racial minority who was unarmed and

27 posed no threat to the officers or others, violated his right under the Fourth Amendment to be free from an unreasonable seizure and therefore constituted “excessive” force. Note that neither the facts nor the Court’s decisions in Garner or Graham suggest that either subject/plaintiff was targeted because of his race. However, sociological theory and some empirical research suggest that use of force may be used disproportionately against minorities, particularly young Black males, because of racial stereotypes. This theory and research will be described in Chapter 3.

B. Graham

As noted, Garner settled the issue of when deadly force satisfies Constitutional requirements. Graham v. O’Connor addressed the constitutional reasonableness of non- lethal force, which in this case occurred during an investigatory stop. Together, Garner and Graham provide comprehensive guidance on when and how the state can use its coercive power against citizens.

Dethorne Graham, also a Black man as well as a diabetic whom the officers mistook as being drunk, sustained a broken foot, cuts on his wrists, a bruised forehead, and an injured shoulder from the force to which he was subjected (490 U.S., at 390). The significance of Graham is its holding that all claims of excessive force by police officers in the course of an arrest, investigatory stop, or other seizure of a person must be analyzed under the Fourth Amendment’s reasonableness standard rather than a substantive due process approach (490 U.S., at 395). In other words, because any use of force constitutes a seizure of the person against whom force is used, and because the

Fourth Amendment requires that seizure of a person must be reasonable, whether seizure of a person constitutes lawful government action must be judged with reference to the

28

Fourth Amendment’s reasonableness standard. As the Court observed, the Fourth

Amendment’s reasonableness standard, which was implicit in Garner’s analysis of when deadly force is lawful, is also the appropriate guide for analyzing excessive force claims because the Fourth Amendment provides “an explicit textual source of constitutional protection” against physically intrusive governmental conduct (490 U.S., at 395).

Significantly, Chief Justice Rehnquist concluded that reasonableness must be judged from the perspective of a reasonable officer on the scene and not with the benefit of 20/20 hindsight (490 U.S., at 396). Moreover, reasonableness takes into consideration that police officers “are often forced to make split-second judgments – in circumstances that are tense, uncertain, and rapidly evolving – about the amount of force that is necessary in a particular situation” (490 U.S., at 397). However, even after considering reasonableness from the perspective of a reasonable officer in a fluid encounter with a resistive subject, the proper standard to be applied under the Fourth Amendment is whether the force used was objectively reasonable considering the facts and circumstances confronting that officer (490 U.S., at 397). Facts and circumstances to be considered include the severity of the crime alleged, whether the suspect poses an immediate threat to the safety of the officers or others, and whether he is actively resisting arrest or attempting to evade flight

(490 U.S., at 396).

Scott v. Harris, 550 U.S. 372 (2007), illustrates how the Court applied the objective reasonableness standard to a use of force that seriously injured a motorist being chased by the police. In that case, a Georgia county deputy (Scott) rammed a motorist’s car from behind during a high-speed pursuit, causing the motorist’s vehicle to leave the road. The motorist, Harris, was left paralyzed by injuries he suffered when his vehicle left the road

29 and overturned. By ramming Harris’ vehicle from behind, Scott’s actions constituted a seizure of Harris that implicated the reasonableness standard of the Fourth Amendment.

At issue was whether it was objectively reasonable for Scott to place Harris at risk of serious injury or death to stop him from endangering other motorists and pedestrians.

A dashcam video established that Harris raced down narrow, two-lane roads in the dead of night at high speeds, swerving around more than a dozen other cars, cross the double-yellow line, and forced cars traveling in both directions to their respective shoulders to avoid being hit (550 U.S., at 379). Pursuing police cars were forced to speed to keep up and swerve around motorists to remain behind Harris. Writing for the Court,

Justice Scalia concluded that the chase initiated by Harris posed a substantial and immediate risk of serious injury to others. Under those circumstances, Scott’s use of force that posed a risk of serious injury or death was objectively reasonable.

To summarize, any use of force by the police must be objectively reasonable under the facts and circumstances of the encounter to satisfy the dictates of the Fourth

Amendment. Force that does not meet the standard established by the Court in Graham is excessive and violates the suspect’s right to be free from an unreasonable seizure.

However, the “boundary” between force that is reasonable or force that is excessive is not a bright line that is easily discerned. Rather, a claim of excessive force requires careful analysis of the facts and circumstances from the viewpoint of the officer involved and an objective determination as to whether his or her actions were reasonable under those circumstances.

30

Guidance for a Proportional Response to a Resistive Suspect: The Use of Force Continuum

To follow state law and U.S. Constitutional requirements for the use of force against citizens, officers must evaluate the circumstances confronting them in a situation and use only that amount of force that is necessary and objectively reasonable to achieve citizen compliance (Terrill, 2005). That determination requires a subjective evaluation by the officer of the tactical situation facing him or her, which includes the threat posed by the suspect and the level of resistance encountered. As Chief Justice Rehnquist noted in

Graham, the force response is often a split-second determination made “in circumstances that are tense, uncertain, and rapidly evolving” (490 U.S., at 397).

However, police officers do not decide in a vacuum on how to respond to a resistive suspect. In training that is typically referred to as “defensive tactics,” officers are taught multiple force options as well as how those options are related to levels of suspect resistance they may encounter (Terrill, 2005). In other words, since force must be proportional to the level of resistance encountered, officers should be trained to associate a tactic with the level of resistance to which it is proportional (Terrill, 2005).

For example, if, after being told he is under arrest and needs to put his hands behind his back, a suspect resists verbally (“I’m not going to jail”) or passively (by refusing to put his hands behind his back), an officer will have force options available that are proportional to these low levels of resistance. These could include relatively low levels of force such as a joint manipulation or pain compliance (pressure point) technique, as opposed to more severe (and in this case, not proportional) responses such as a baton, taser, or bean bag gun. Proportionality thus recognizes that a suspect’s verbal refusal

(e.g., I’m not going to jail) to obey an officer’s verbal command (e.g., put your hands

31 behind your back) presents a significantly lower threat to the officer than does an attack on the officer. Both suspect actions are forms of resisting arrest, “albeit at markedly different levels” (Terrill, 2005, p. 111), which accordingly justify different force responses. What results is an “interplay between suspect resistance and police use of force” that comprises the micro-process of the encounter (Terrill, 2005, p. 108).

To ingrain the concept of proportionality, officers are trained to think of force as lying along a “continuum,” ranging from verbal force on one end to deadly force on the other. The use of force continuum thus begins at the lowest possible level and escalates with increasing levels of severity to the highest possible level. As Terrill notes, the force continuum “is a policy framework used in numerous police departments that is explicitly norm-based, attempts to capture some of the subtleties of using coercion, and acknowledges that coercion in the police-suspect encounter is a dynamic process” (2005, p. 110).

In addition to proportionality, another principle inherent in the continuum structure is incrementalism (Terrill, 2005). Officers, in general, understand that the tactic they initially employ may not produce the desired result. When a tactic fails to gain compliance, officers may then proceed in small increments along the continuum, escalating the level of force until compliance is obtained. For example, a suspect may physically resist an officer’s attempt to manipulate the suspect’s arm behind his back for handcuffing. Defensive resistance can include tensing up, shaking violently from side to side, trying to pull away, or simply falling to the ground and kicking in the officer’s direction. Facing defensive resistance, an officer may escalate incrementally along the

32 continuum to a higher level of force, which in this case could be a distractionary punch or a knee strike.

To summarize, officers are trained to use force that is proportional to the resistance they encounter. The use of force continuum models force options and embodies the guiding principles of proportionality and incrementalism. Force can escalate or de- escalate along the continuum as the level of resistance dictates. Officers are taught that their ultimate objective is to use only that amount of force necessary to obtain compliance, i.e., to handcuff a suspect who is no longer resisting or thereafter ceases resisting. Even for those officers acting with the best of intentions and in good faith, it is not always easy to comply with legal guidelines, as a confrontation with a resistive or assaultive suspect who is not bound by any legal guidelines evolves quickly and unpredictably.

Operationalizing Force as a Dependent Variable

Research on police use of force operationalizes force as a dependent variable

(typically measured ordinally or as a dichotomous variable), representing the highest level of force used in the police-citizen encounter. Construction of the dependent variable incorporates some aspect of the use of force continuum. For example, in their studies based on the Project on Policing Neighborhoods observational data, which included encounters where no force was used, Terrill and Mastrofski (2002, p. 230) and Terrill and

Reisig (2003, p. 300) measured force as ranging from no force to verbal force to restraint techniques to impact methods. Using NYPD Stop, Question, and Frisk (“SQF”) data, which also includes encounters where no force was used, Lautenschlager and Omori

33

(2019, p. 10) measured force as it was recorded in the SQF data base along a 9-point

“severity” scale:

0 No Force 1 Hands 2 Frisked/Searched 3 Handcuffed 4 Other 5 Suspect Against Wall 6 Pepper Sprayed 7 Suspect on Ground 8 Baton 9 Weapon Drawn or Pointed

Using the same SQF data, Morrow, White, and Fradella (2017, p. 375) recoded the dependent variable into three categories: no force, physical/non-weapon force (which includes hands, suspect on ground, suspect against wall, and handcuffs), and weapon force (which includes weapon drawn, weapon pointed, baton, and pepper spray).

Ferrandino (2015, p. 216) collapsed the SQF data further into a dichotomous variable: no force or at least one level of force used.

Most studies, however, do not have the benefit of observations that include police- citizen encounters where no force was used. Like this study, these studies must rely instead on police use-of-force self-reports or other data that describe what took place after the decision to use force was made. Studies using self-report data generally operationalize the dependent variable using some or all categories described in the use of force reports. A few examples follow.

Using self-reports from an unidentified large municipal police agency located in the

Southeastern United States, Lersch and colleagues (2008, p. 289) recoded the categories on the use of force form into two categories: low force, encompassing handcuffing, leg restraints, pressure points, and transporter techniques; and high force, encompassing

34 countermeasures, total appendage restraint, use of chemical weapon, use of impact weapon, firearms pointed or fired at a subject, and bitten by police canine. Lawton (2007) used self-reports for the year 2002 from the Philadelphia Police Department. Using the nominal categories from the use of force form, Lawton (2007, p. 171) constructed an ordinal variable representing the highest level of force used in the encounter: (1) control hold, (2) strike or punch, (3) pepper spray or taser or (4) baton. Using use of force reports from the Austin, Texas Police Department over two years, Lee, Vaughn, and Lim

(2014, p. 494) modeled the highest level of force as a multinomial measure with soft empty hand control as the reference category. Force then ranges from hard empty hand control, pepper spray, taser, and impact weapon.

To summarize, research analyzing use of force issues incorporates the concept of a force continuum, although there is no universal agreement on how many categories of force are optimal. Indeed, Lee et al. (2010), using data from eight large police agencies including Dallas, Miami-Dade, and Phoenix, note that the agencies each have different types of force continuum scales, ranging from five to nine levels. In their effort to create a national use of force database, the International Association of Chiefs of Police (2001) decided to track uses of force according to a “street continuum” approach that ranked force from physical (the use of fists, feet, hands, etc.), to chemical (mace, pepper spray), to electronic (Taser or stun gun), to impact (use of a baton or other impact weapons), to lethal (any kind of firearm).

The RPD’s Force Continuum

The RPD has a four-level continuum it calls the “force matrix” (Defensive Tactics

Manual, 2005). Level 1 includes command presence and verbal force. Level 2 includes

35 pressure points, wrist/arm locks, and pepper spray. Level 3 includes impact techniques such as personal weapons (e.g., punch, forearm strikes, kick, and knee strikes), taser, baton, and beanbag gun. Finally, Level 4 includes deadly force options (e.g., handgun or long gun). According to Sgt. Andy MacPherson, the head of the RPD’s Defensive Tactics

Training Program, the RPD uses a “totality of the circumstances” test to evaluate whether force used by its officers was objectively reasonable. In making this evaluation, officer- suspect factors are considered, including gender, age, height, weight, level of exhaustion, injury to the officer or suspect, and special knowledge (e.g., suspect is a martial artist or boxer) (Defensive Tactics Manual, 2005). Officer-subject factors are documented on the

Subject Resistance Report (“SRR”) that officers must complete following a use of force.

The SRR is discussed in greater detail in Chapter 5, Data, Methods, and Results.

Sgt. MacPherson also stated that officers are trained to associate levels of force with the levels of resistance they may encounter. As enumerated on the SRR in use during the study period, these levels of resistance are: Verbal (i.e., failing to adhere to verbal commands), Passive Resistance (i.e., dead weight), Active Resistance (i.e., pulling away, striking or attempt ) and Armed Resistance (i.e., uses or attempts to use a weapon or dangerous instrument).

Notably, the current SRR in use no longer uses the term “active resistance.” Rather, two new categories are used that represent distinct types of resistance previously lumped together in the former active resistance category: avoiding custody and assaultive behavior (RPD G.O. 335). Thus, the current SRR documents the same levels of resistance documented on the SRR used during the study period but uses different nomenclature to more accurately reflect the distinct types of resistance encountered.

36

Avoiding custody may include running/walking away, bracing/tensing body, pulling away, locking arms under body, or holding onto a fixed object. Assaultive may include fighting stance, combative approach, punching, kicking, biting, tackling.

Categorizing levels of resistance by whether the subject is trying to avoid custody or actively attacking the officer is logical, according to Sgt. MacPherson, because RPD policy (which was also in use during the study period), stipulates that certain Level 3 options can be used only when the subject is exhibiting assaultive behavior (RPD SOP).

For example, while “personal weapons” such as punches or takedowns can be used against suspects who are avoiding custody or exhibiting assaultive behavior, and as such are considered lower levels of force, impact methods such as the taser, baton, beanbag gun or canine bite can only be used against assaultive resistance, unless special circumstances are present. Special circumstances may include proximity to weapons, special knowledge, injury/exhaustion of the officer, ground fighting, or number of participants involved (RPD SOP, p.2). Sgt. MacPherson noted that failure to gain compliance using lower levels of force against a suspect avoiding custody may also justify use of higher force options such as a taser or baton.

When asked about how officers are trained to use pepper spray,3 Sgt. MacPherson stated that pepper spray is a level of force that can be used against all levels of resistance, although obviously it would not be the first option used against a suspect armed with a knife, firearm or other deadly weapon. While pepper spray is included at Level 2 in the

RPD matrix, Sgt. MacPherson noted that pepper spray is more intrusive than other level 2 force options, as pepper spray contacts the subject’s skin, eyes, and respiratory tract, and requires treatment afterwards at an eyewash station (unless the suspect refuses treatment,

37 which must be witnessed by a supervisor). However, pepper spray is considered less intrusive than contact weapons such as the taser or the baton, which is why these options are generally restricted to assaultive or armed suspects.

Moreover, officers, consistent with policy, appear to view pepper spray as a lower force option than the taser, since it was used in 45.2% of encounters with females and only 32% of encounters with males, while the taser was used in only 5% of encounters with females versus 14% of encounters with males. For purposes of this study, then, pepper spray is categorized as a level of force option on par with non-weapon force options such as a punch or takedown.

The Study Dependent Variable

As discussed above, RPD’s current policy restricts the use of impact and deadly force options. Impact (weapon contact) and deadly force options can be used only when (1) an officer is confronted by assaultive or deadly force, (2) lower force techniques were ineffective in establishing control of the subject, or (3) special circumstances are present.

RPD policy thus provides appropriate guidance for how the dependent variable should be operationalized in this study. Accordingly, the dependent variable here is operationalized as a dichotomous variable constructed from the following force option categories:

38

1. Verbal/Soft Techniques: verbal force, nerve pressure points, nerve stuns, picking up or carrying a passive suspect

2. Non-Weapon Techniques: Punches/Jabs, Kicks/Knee strikes, Takedowns, Pepper Spray, and Other (which includes joint manipulation and techniques such as a 2- or 3-point landing on a suspect while on the ground)

3. Impact/Weapon Techniques (Taser Threat, Taser Use, Baton, Bean Bag, or Canine Bite)

4. Deadly Force (handgun, long gun)

These categories are collapsed into two categories: low force, comprised of verbal/soft and non-weapon techniques, and high force, comprised of impact/weapon techniques and deadly force. Like the coding scheme used by Lersch et al. (2008), these will be coded

“0” for low force and “1” for high force. Collapsing these categories into a dichotomous variable will also allow for easier and more logical interpretation of the model estimations.

Levels of suspect resistance are also constructed to reflect the types of resistance officers encountered during the study period, including the substantive difference between “avoiding custody” and “assaultive behaviors,” as follows:

1. Verbal Resistance 2. Passive Resistance (e.g., dead weight) 3. Avoiding Custody 4. Aggressive (i.e., assaultive behavior) 5. Armed Resistance

Consistent with the dependent variable, levels of resistance will be collapsed into a dichotomous variable reflecting low resistance (verbal, passive, and avoiding custody), and high resistance (aggressive/assaultive and armed). The study will thus control for levels of resistance officers encounter according to the current categories, which are more

39 reflective of actual levels of resistance. Table 1 displays the levels of resistance and associated levels of force used in this study:

Table 1: Levels of Resistance and Associated Levels of Force

Low Resistance: Low Force:

Verbal/Passive Verbal/Soft

Avoiding Custody Non-Weapon

High Resistance: High Force:

Aggressive/Assaultive Impact/Weapon

Armed Armed

40

Chapter 3: Theoretical Foundations for Neighborhood Influences on the Use of Force

Introduction

Use of force occurs during a micro-level encounter between police and citizens

(Terrill, 2003, 2005). However, use of force incidents are also embedded within macro- environments such as neighborhoods, which, in urban areas, typically vary by their racial and socioeconomic characteristics. Over 75 years ago, William Whyte (1943) observed that police have different standards of acceptable conduct depending on the nature of the neighborhood in which the conduct occurs. Conduct tolerated in one neighborhood may be addressed by police in another (Whyte, 1943).

Similarly, after observing interactions between police and youth in Oakland and San

Francisco, Werthman and Piliavin (1967) concluded that police enforcement is in part motivated by neighborhood variations in socioeconomic conditions, which provide cues regarding acceptable standards of conduct. Consequently, everyone, regardless of their personal characteristics or actions, encountered in neighborhoods the police designate as

“bad,” are tainted by the character of the neighborhood in a process they called

“ecological contamination.” Individuals may thus be viewed as being more suspicious or dangerous simply because of their location. Bayley and Mendelsohn (1969) also argued that police enforcement was higher in lower class neighborhoods where there was greater social distance between police and citizens, resulting in more coercion and arrests.

These findings suggest that police enforcement activities and use of coercion may vary considerably across neighborhoods. Thus, to fully understand police behavior and how force is used, it is necessary to consider the neighborhoods where police and citizens

41 are embedded and how community characteristics may shape outcomes between police and citizens. Specifically, is the level of force to which a citizen is subjected influenced not only by the situational aspects of the encounter but also by the nature of the neighborhood in which the encounter occurs? Stated another way, in distributing coercion as an extreme form of social control, do police consider not only what a citizen does, but also the racial and socioeconomic characteristics of where the encounter occurs?

To address neighborhood context, this chapter will devote a considerable amount of its focus on the effect of race. Race may be an especially salient factor when it intersects with sex and age (see discussion below of the perceived dangerousness of young Black males). The chapter draws heavily on the theoretical insights provided by racial threat theory and, to a lesser extent, by Black’s hypotheses about morphology and law. In tandem, and individually, these perspectives explain why the coercive power of the state may be used disproportionately in areas with higher percentages of racial and ethnic minorities.

The discussion of racial threat theory proceeds as follows. First an overview of

“classic” racial threat theory is presented. In its original form, racial threat theory describes how large or increasing percentages of minority residents in an area may present economic and political threats to the dominant group in power. A discussion follows as to why “classic” racial threat theory is likely not be applicable to the behavior of criminal justice actors, whose occupational mandate does not encompass economic or political threats. If racial threat is indeed a viable explanation for police use of force, it may be because officers harbor stereotypes and implicit biases about Black individuals.

To explain how racial threat theory is applicable to police officers, the chapter next

42 reviews the literature on racial threat and implicit bias, which associate Blacks with violence and criminality. The stereotype of as violent individuals, with a propensity for criminality requiring social control, is widespread in contemporary

American society, including among police officers and other criminal justice actors.

Research findings to support this conclusion are discussed. These findings suggest that police officers who harbor implicit biases may respond with more severe force or more frequent uses of force in areas with larger percentages of Blacks. Research finding that areas with larger minority populations do indeed have higher levels of police coercion is then presented.

As noted, there are other, non-racial perspectives that explain why higher levels of police coercion may be associated with economically disadvantaged neighborhoods.

These include social disorganization theory and Werthman and Piliavin’s (1967) ecological contamination thesis, both of which are discussed in this chapter. Research hypotheses derived from empirical tests of race-based and economic disadvantage theories are then presented. The chapter first begins with a discussion of the origins of policing in the U.S., which suggests that police have a long history of being used for the social control of citizens who threaten the economic and political interests of the dominant groups in society.

The Origins of the Police: Control of Disorder and the “Dangerous Classes”

Since police departments were created in the nineteenth century, one of their functions has been protecting the interests of dominant groups by controlling

“undesirables” in society. In late eighteenth and early nineteenth century London and

Paris, “undesirables” were rapidly multiplying numbers of the poor, who were perceived

43 as a threat to the social order (Silver, 1967, p. 3). In his essay on the origins of the modern police force, Silver (1967, p. 1) argues that a police force was created in London in response to a demand by the “peaceful and propertied people” for order and control of these “dangerous classes.” Reviewing literature from mid-nineteenth century England,

Silver notes it paints an image “of an unmanageable, volatile, and convulsively criminal class at the base of society” (1967, p. 3). Following their creation in 1829, the London police department’s mission was the prevention of crime as well as the control of the

“dangerous classes” and political agitation in the form of mobs or riots (1967, p. 7).

Policing in the U.S. followed a developmental trajectory like that of England (Potter,

2013). Policing began informally, with a system of night and day watches composed of community volunteers and constables paid for warrants they served (Potter, 2013). The first American police force, composed of full-time officers and not volunteers, was established in 1838 in Boston, followed by and other major cities (Potter,

2013). Potter argues that the modern police force emerged as urbanization was occurring at a rapid pace, but that there is no evidence police departments were created in response to crime. Rather, Potter contends that modern police forces emerged as a response to

“disorder” (Potter, 2013, p. 3). He argues that private mercantile powers had a greater interest in social control of disorder to insure a stable and orderly work force (Potter,

2013, p. 4).

Potter links exploitation of workers (through long hours, dangerous working conditions, and low pay) to the need to maintain order by preventing riots and strikes.

The modern police force “provided the illusion that this order was being maintained under the rule of law, not at the whim of those with economic power” (Potter, 2013, p. 4).

44

As in England, the political strategy of economic elites was to raise the specter of the

“dangerous classes,” comprised of the poor, immigrants, and free Blacks (Potter, 2013).

Painting these groups as an uneducated and unskilled underclass in society dovetailed with the focus of policing on bad individuals rather than “social and economic conditions that are criminogenic in their social outcomes” (Potter, 2013, p. 4).

Control of police departments by economic and political elites was easily accomplished, as the local political party ward leader (also a local businessman) appointed the police executive in charge of the ward (Potter, 2013). Police were tied directly to political patronage systems of that era and were thus associated with political machines and their dominance in politics (Greene, 2000). Consequently, police were

“primarily concerned with the maintenance of political, and often corrupt, relationships with those in power” (Greene, 2000, p. 305). For example, in the post-Civil War era, union organizing, and labor unrest became widespread throughout the U.S. and targeted specific companies for strikes. The police were then called in and used as strike breakers under the guise of addressing “riots” (Potter, 2013). As Potter notes, the “use of public employees against organizing workers was both cost-effective for manufacturing concerns and politically useful, in that it confused the issue of worker’s rights with the issue of crime” (Potter, 2013, p. 6).

As America was transformed from an agrarian/farm-based society to an urban/industrial society, power was redistributed “from the countryside to the cities and from wealthy landowners to the waves of Eastern European immigrants who gained political preeminence in cities” (Greene, 2000, p. 304). As European immigrants gained power, policing, along with other forms of municipal employment, became a primary

45 means for them to integrate into American society. Consequently, the ethnic composition of large city police departments such as “New York, Philadelphia, Boston, and , among others, can trace their heritage to major patterns of emigration from Ireland, Italy, and Germany” (Greene, 2000, p. 305). Muller (2012) similarly notes that police officers in the 1890s reflected the immigrants of a previous generation and included a disproportionately heavy percentage of Irish Americans. The irony is that European immigration initially created status competition with native-born whites. Consequently, political narratives of the mid-19th century associated the Irish, in particular, with disorder and rowdy, undisciplined behavior, resulting in their arrest and incarceration

(Muller, 2012).

Fortunately for the Irish, their integration into society through municipal employment occurred as roughly 3 million Blacks migrated from the South to the Northern United

States. Over the same period, the nonwhite incarceration rate rose sharply compared to a relatively flat white incarceration rate. Muller (2012) concludes that the increase in

Northern nonwhite incarceration rates was causally related to increases in the numbers of

Blacks arriving in the North from the South. Significantly, Muller presents strong evidence that it was not until white European immigrants, primarily the Irish, constituted a majority of northern police forces that Blacks began to be arrested and imprisoned at disproportionately higher rates than whites. The irony, of course, is that the Irish were themselves once vilified by native born whites as undesirables. Now, Irish Americans and other European immigrants employed in municipal police forces used their common whiteness as an obvious way to distinguish themselves from Blacks moving North and thereby align themselves with the objectives of economic and political elites.

46

To summarize, the history of policing reveals that police in the 19th and early 20th centuries were used by the dominant group to control those who threatened their economic and political interests. By the early 20th Century, the descendants of once vilified Irish immigrants now comprised the majority of urban police departments in the

Northeast. They aligned with those in power and used the mechanisms of formal social control to sanction the new “dangerous class,” i.e., Southern Blacks migrating to the

North.

In the remainder of this chapter, I discuss the racial and economic disadvantage theories explaining why police coercion is influenced by neighborhood context. These theories lead to essentially the same set of hypotheses, which are that in areas with higher percentages of Blacks or Hispanics, or, alternative, in economically disadvantaged areas, police use of force will be higher after controlling for encounter-level variables and neighborhood crime rates.

Racial Threat Theory

The early history of policing, as well as Muller’s empirical findings, indicate that dominant groups in society will use formal social control mechanisms to maintain their relative position over subordinate groups. Theoretical work by Blumer (1958) and

Blalock (1967), collectively known as “group threat theory” (or more recently as “racial threat theory” or “minority threat theory”), posits that threats to the dominant group’s economic and political power posed by the presence of large and increasing percentages of minorities will result in prejudice and discrimination against those minority groups.

Those seeking to preserve their power and dominance are threatened when minority populations increase and, presumably, will use resources at their disposal to take punitive

47 action against minorities who challenge them. For racial threat theory to be viable, then, there should be an observable empirical association between minority group size and punitive responses.

Certainly, an area where minorities could challenge political power is through their use of the ballot. Large and growing minority populations are seen by Republicans and conservatives as constituencies favorable to the Democratic party (Bentele & O’Brien,

2013). Bentele and O’Brien (2013) indeed find support for racial threat theory after examining proposed legislation, including requiring photo identification, proof of citizenship requirements, and reductions in early voting, among others, that was likely to restrict access for minority voters. The authors find that increases in restrictive legislation proposed between 2006 and 2011 were positively associated with the racial composition of the state proposing the legislation, as well as with increases in minority voter turnout in those states between the 2004 and 2008 presidential elections.

But is racial threat theory also applicable to social institutions other than the political and economic, such as the criminal justice system? While large and increasing geographic percentages of minorities may motivate legislative efforts to limit voting rights, do they also motivate criminal justice actors such as police officers and judges to take formal action against minorities? As applied to police behavior, research testing racial threat theory has associated percentages of Blacks people in the local population with the use of deadly force against them (Chamlin, 1989; Jacobs & Britt, 1979; Jacobs

& O’Brien, 1998), as well as with higher police expenditures (Jackson & Carroll, 1981).

However, recent research by Legewie and Fagan (2016) casts considerable doubt on the validity of associating police use of deadly force with just the size of the minority

48 population. They find that deadly force against Black individuals is not driven only by population shares, but by

a race-specific pattern in the ways in which the police respond to crime. Indeed, the black-on-white homicide rate is a significant predictor of officer-involved killings whereas black-on-black homicides and measures for political and economic threat are unrelated to police killings

(Legewie and Fagan, 2015, p. 34, emphasis in the original).

In other words, the use of deadly force against Black citizens increased as the rate of homicides committed against whites by Black suspects increased. Both homicides of

Black individuals committed by individuals and measures of political or economic threats posed by the size of the Black population had no effect on the use of police deadly force against Blacks. These findings are consistent with research attributing coercive treatment of minorities to “the high value attached to a white victim and the racial fears of authorities” associated with violence by minorities against the dominant group (Legewie and Fagan, 2016, p. 34).

Legewie and Fagan’s findings suggest that coercive treatment of minorities is related to the perceived threat they pose of criminality and violence, and not by economic or political threats. An implication of Legewie and Fagan’s findings is that the logic of classic racial threat theory may not be compatible with decision making by criminal justice professionals. Indeed, some scholars have made this argument. In evaluating the utility of racial threat theory as an explanation for sentencing disparities between Blacks and whites, King and Light (2019, pp. 404-405) observe that

it strains credulity to suggest that judges, in calculating fashion, are aware of demographic changes in the county and strategically use their power in the criminal justice system to stymie the advancement of racial minorities in the political realm, or that judges deliberately give longer prison sentences in order to put blacks at a competitive disadvantage relative to whites.

49

In addition to straining credulity, King and Light (2019, p. 405) note that the theory does not “square with available evidence” regarding sentencing outcomes and county demographics.

Applying the same logic to the police, it equally strains credulity to suggest that street cops observe demographic changes in their patrol areas and then associate minority citizens with economic or political threats, or view protecting majority group political and economic dominance as their occupational objective. As Holmes and Smith (2012, p.

347) observe, “the police constitute a distinct social group that possesses unique interests that do not always correspond to the interests of the dominant group.” Consequently, racial threat theory, specifically as it applies to threats to the existing power structure, does not appear to be a viable explanation for a police officer’s decision to arrest or use force against minorities.

Stereotypes, Implicit Bias, and Racial Threat

If racial threat theory is a viable explanation for the disproportionate use of coercion against minorities, it is likely because police officers, like other citizens (including other actors in the criminal justice system), harbor stereotypes or implicitly held biases about

Black individuals. In contemporary America, egalitarian norms largely denounce explicit communications of racial bias manifested in race-biased aggressive acts, verbal statements, and hiring procedures (Weisbuch, Pauker, & Ambady, 2009, p. 1711).

Consequently, “race biases are often communicated subtly via facial expressions and body language,” and “exposure to nonverbal race bias may undermine norm-driven correction processes and hence exert a social influence” (Weisbuch, et al. 2009, p.1711, emphasis in original). In other words, while verbal condemnation of an overtly racist

50 statement or act may chastise the actor and reinforce prevailing social norms, there is no similar correction process to offset the effect of nonverbal communication of race bias, thereby permitting the creation and spread of social stereotypes or implicitly-held biases.

A social stereotype “is a mental association between a social group or category and a trait” (Greenwald & Krieger, 2006, p.949). Stereotypes may be culturally shared and both positive and negative (Graham & Lowery, 2004). An implicit bias reflects discrimination based on implicit attitudes or stereotypes (Greenwald & Krieger, 2006). Actions may be taken pursuant to mental processes that are conscious (explicit) or unconscious (implicit)

(Graham & Lowery, 2004; Greenwald & Krieger, 2006). As Greenwald and Krieger observe, an “intention to act is conscious if the actor is aware of taking an action for a particular reason” (2006, p. 946, emphasis in the original). In contrast, an unconscious act reflects the fact that “actors do not always have conscious, intentional control over the processes of social perception, impression formation, and judgment that motivate their actions” (Greenwald & Krieger, 2006, p. 946). Unconscious acts are thus not planned responses, but are unintentional and involuntary, which causes them to occur in the presence of an environmental cue (Graham & Lowery, 2004).

Consequently, an individual may be exposed to bias without having a conscious awareness of what is occurring and may unconsciously act in accordance with that bias towards members of the stigmatized racial group. The transmission of bias can occur even during an otherwise innocuous activity such as watching television. Weisbuch,

Pauker, and Ambady (2009) find that Americans are exposed via television to nonverbal race bias, and that this exposure can transmit to viewers bias that may influence their race associations and self-reported racial attitudes.

51

The Association of Black Individuals with Violence and Criminality

The previous section explores how bias against one racial group, Black individuals, is widely transmitted nonverbally and influences race associations and self-reported attitudes. One such widely documented race-based association or attitude is the categorization of Black individuals as criminals, and hence more dangerous, than other groups. There is, indeed, a significant body of research finding that whites harbor stereotypes associating Black individuals, particularly young Black males, with violence and danger. These stereotypes have a long historical legacy. Scholars have observed that racial tensions in the U.S. have existed since the country’s founding (Mears et al. 2019;

Peffley & Hurwitz, 2010). In particular, scholars have argued that post-Civil War efforts to give Black citizens greater legal and civil rights created a foundation for racial animus and hostility that persisted through the Jim Crow era and into modern times (Gabriel &

Tolnay, 2017; Mears et al. 2019; Messner, Baller, & Zevenbergen, 2005).

A tangible manifestation of this racial animus and hostility during the late 19th and early 20th Centuries was lynchings of Blacks (Mears et al. 2019). There were approximately 2,000 documented lynchings of Black men, women, and children in southern U.S. states between 1882 and 1930 (Gabriel & Tolnay, 2017, p. 77). Mears and colleagues (2019) argue that that the entrenched legacy of lynchings still colors perceptions of Black citizens in areas where lynchings occurred. Using multilevel analyses of lynching and survey data on white individual views of Black citizens, the authors find that

52

whites who reside in communities with a greater historical legacy of racial animus and hostility, as reflected in and perpetuated by higher levels of lynchings, were more likely to view blacks as a threat. Specifically, whites from such areas were more likely to view blacks as criminal threats and to view blacks as more likely to commit crimes against whites.

(Mears et al. 2019, p. 509, emphasis in the original).

The legacy of lynching also influences contemporary events. King, Messner, and

Baller (2009) find that localities with past lynching and sizeable Black population are associated with suppression of police compliance with law, including the reporting of hate crime and the likelihood of prosecuting a hate crime incident. In negative binomial models, Messner, Baller, and Zevenbergen (2005) find that a measure of lynching was significantly positively associated with southern homicides, particularly white on Black homicides, in areas that experienced a high frequency of lynching in the past.

As the above discussion suggests, the stereotyping of Blacks as violent individuals, with a propensity for criminality requiring social control, has a long history, particularly in the South, that has not abated in modern times. In addition to historical influences on bias, research on the scientific foundations for bias indicates that stereotypes about

Blacks may be widespread and are conveyed by modern technology nonverbally, i.e., through television. Contemporary research examining these stereotypes nationally focuses on perceived fear of Blacks as violent individuals, and perceived fear of neighborhoods because they have larger percentages of Black residents.

Peffley and Hurwitz’s (1998) research provides an apt example of the association between the race of the individual and criminal propensity. Utilizing data from the

National Survey, Peffley and Hurwitz (1998) find that Black citizens and

53 white citizens differ significantly in their beliefs as to the causes of Black criminality.

Black individuals base their view of the criminal justice system on their lived experiences and are thus much more likely to attribute the higher Black crime rate to a discriminatory justice system (Peffley & Hurwitz, 1998, p. 1007). Conversely, whites attribute higher

Black arrest and imprisonment rates to the belief that Black people are more likely as individuals to commit more crimes and are thus more deserving of punishment (Peffley

& Hurwitz, 1998, p. 1007).

As discussed above, Weisbuch et al. (2009) find that nonverbal race bias of Blacks as individuals is transmitted via television. Research suggests that local television news is partly responsible for transmission of the type of stereotypes documented in Peffley and

Hurwitz’s research associating Blacks with criminality. Gilliam and Iyengar (2000) argue that crime reporting by local news follows a crime “script” with two core elements: crime is violent and, based on racial imagery contained in the story, perpetrators of crime are non-white males. Using a multi-method design, the authors find that this crime reporting script increases support for punitive approaches to crime and heightens negative attitudes about Blacks among white, but not Black, viewers. The authors observe that crime dominates local news “because its emphasis on vivid pictures and emotional personal accounts is believed to attract viewers” (2000, p. 572). As television is a visual medium, information about race is conveyed automatically, and thus the authors conclude that local news will continue to cultivate misperceptions and prejudice (2000, p. 572).

Similarly, Peffley, Shields, and Williams (1996) utilized a video experiment in which they manipulated only the visual image of the race of the suspect in a television news story of violent crime. The authors used actual news stories that they edited to produce

54 two identical videos, one with a visual image of a white suspect and the other with a visual image of a Black suspect. Participants were 95 first- and second-year white students in introductory political science classes at a large midwestern state university.

The authors find that even a brief visual image of a Black male suspect in the news story caused participants to view the Black suspect as more guilty, more deserving of punishment, more likely to commit future violence, and with more fear and loathing than a similarly portrayed white suspect (Peffley et al. 1996, p. 321).

Other research finds misperceptions of Blacks as individuals are widespread in contemporary American society. Utilizing a national random sample (N=885) of adults

18 years and older, Chiricos, Welch, and Gertz (2004) used measures of racial prejudice

(consisting of an index composed from survey questions), degree of political conservatism, fear of crime, and regional location to assess support for harsh punitive policies. The authors find that whites who are either less prejudiced (relative to whites who identified as being more racially prejudiced), not located in the South, conservative, and for whom crime salience is low, nonetheless believe violent crimes are typically committed by Blacks, and that this association is a significant predictor of punitiveness.

Similarly, Barkan and Cohn (1994), using 1990 General Social Survey data, find that whites who exhibited antipathy toward Blacks and stereotyped them as lazy, unintelligent, and having a preference for welfare and a lack of patriotism, were significantly more likely to support the death penalty. The authors conclude that white support for the death penalty is in part associated with prejudice against Blacks as individuals.

55

Correll, Park, Judd, and Wittenbrink (2007) tested racial bias using undergraduate subjects playing videogames. The videogames simulated encounters with potentially hostile Black and white targets, with and without guns. If the suspect had a gun, the participants were instructed to push a button and shoot the suspect; if the suspect did not have a gun, they were instructed to push a different button that signaled no shot was fired.

The authors experimentally manipulating the race-danger association by randomly exposing study participants to crime stories in which some saw the suspect as Black and others as white. They find that participants were more likely to mistakenly shoot an unarmed suspect when the suspect was Black than when the suspect was white. In a similar experiment involving police officers using a computer simulation, Plant and

Peruche (2005) find that police officers, like the undergraduates in Correll and colleague’s study, also were more likely to mistakenly shoot unarmed Black suspects than unarmed white suspects.

In a comprehensive review of theory and research examining Black criminal stereotypes and , Welch (2007, p. 276) also observes that “racial stereotyping of criminals has been an enduring and unfortunate feature of American culture.” This stereotype “overwhelmingly” presents the image of young Black males as violent and menacing (2007, p. 276). In her view, driving this stereotype is the disproportionate percentage of Blacks represented in the criminal justice system, which exceed their racial composition of the population (Welch, 2007). Crime statistics indicate that Blacks are arrested for violent crimes at a rate that far exceeds their percentage of the

U.S. population (Welch, 2007). Consequently, crime committed by Blacks may be especially salient for perpetuating stereotypes because “violent crimes tend to be the most

56 fearsome [and] are most disproportionately perpetrated by Black males” (Welch, 2007, p.

279).

Welch also cites the war on drugs of the 1980s, media portrayals of Blacks and crime, and the “racial ” as contributors to the perpetuation of these stereotypes. While powdered cocaine was used predominately by whites, the use of crack cocaine was associated with Blacks and helped promote punitive policies that resulted in the disproportionate arrest and conviction of Blacks for the sale and use of crack cocaine.

Like the research conducted by Gilliam and Iyengar (2000) discussed above, Welch notes that televised news coverage of violent crimes commonly portrays “accused Black criminals in scowling mug shots or in video clips being led in handcuffs by White police officers” (2007, p. 281), which only serves to reinforce the image of young Black males as violent criminals.

One of the most prominent examples cited by Welch was the use of Willie Horton’s mug shot in Bush campaign ads during the 1988 presidential election. Horton committed rape while on a work furlough. His image was used to both paint the democratic candidate, Gov. Michael Dukakis of Massachusetts, as too liberal and soft on crime, but also reinforced the image of Black males as violent and dangerous (Welch, 2007). The

“racial hoax” occurs when an actual criminal perpetrator attempts to deflect attention from him or herself by identifying the offender as a Black male. The most prominent example cited was Susan Smith, who killed her two children in 1994 but told police she had been carjacked by a young Black male while her children were in the car (Welch,

2007, p. 285). This nationally publicized story presented the image of an innocent white victim and a ruthless Black murderer.

57

As noted, the other branch of this research examines perceived fear of crime associated with the racial composition of neighborhoods. After controlling for crime rates and measures of neighborhood disorder, Quillian and Pager (2001) find that the percentage of young Black men in a neighborhood is significantly and positively associated with perceptions among neighbors of the neighborhood’s crime rate. The authors find evidence of the influence of race on perceptions, including that whites are averse to Black neighbors not because of the actual crime rate but rather the perceived crime rate and fear of crime in Black neighborhoods (2001, p. 748).

Examining fear of crime among Black and white individuals in relation to actual racial composition of neighborhoods, Chiricos, Hogan, and Gertz (1997) find that perceived racial composition, but not actual racial composition, elevates fear of crime among whites who perceive themselves to be in the racial minority. The authors specify that the effects of perceived racial composition on fear are indirect and mediated by perceived safety or risk of victimization (1997, p. 121). Significantly, Chiricos et al. note that perceptions of fear of crime among whites occurred in a “less intensely urban environment in a government and university town” in which the Black community involved substantial numbers of professionals and students (1997, pp. 123-124).

Racial Threat, the Violent Black Male, and the Punishment Cost

As discussed in the above sections, stereotypes or implicit biases that associate Black individuals with violence and criminality exist in contemporary U.S. society and have a long historical legacy. Stereotypes or implicit biases associating Black individuals, particularly young Black males, with criminality motivate formal social control responses by the criminal justice system that have very real, detrimental consequences for the

58 liberty of Black citizens. Scholars, political commentators, and civil rights activists have observed that race appears to be a factor in criminal justice outcomes, as Black defendants are significantly overrepresented in the criminal justice system and in rates of incarceration. Research documenting disparities in criminal sentencing outcomes indeed finds that criminal justice actors in the court system develop implicit biases against

Blacks defendants being more dangerous and more deserving of punishment.

Steffensmeier, Ulmer, and Kramer (1998), for example, find a “punishment cost” of being young, Black, and male. Specifically, they find that young Black males are more likely not only to be incarcerated, but also for longer periods of time than young white defendants. The authors conducted interviews with judges and found that young Black males are stereotyped as being less bonded to the community, more likely to recidivate, and hence more dangerous, thereby justifying higher rates of incarceration and longer sentences than their white counterparts. Similarly, Steen, Engen, and Gainey (2005) find that Black male drug dealers, with or without a criminal record, are viewed as dangerous and likely to be incarcerated. Johnson (2006) finds strong racial interactions between judges and offenders, such that Black and Hispanic offenders were significantly more likely to be incarcerated overall.

Entry into the criminal justice system begins with an arrest by a police officer, and there is no reason to believe that police officers, who actually work daily in poor, racially segregated, and crime-ridden neighborhoods, are any different from other citizens or criminal justice professionals who harbor stereotypes of Blacks as prone to criminality and hence more dangerous than other racial groups. Unlike courtroom professionals, who do not work in racially segregated neighborhoods where they will be exposed to danger,

59 police personally face potentially threatening and dangerous situations daily and experience threatening situations vicariously through incidents involving their colleagues.

As noted by Holmes and Smith (2012, p. 347), police “confront real and symbolic threats to their authority and personal safety in the course of street-level policing in disadvantaged areas, which has predictable consequences for police behavior.” Some experimental research involving police officers, such as the computer simulation by Plant and Peruche (2005), described above, suggests that police officers implicitly react to racial cues that they may interpret as threats, real or symbolic, to their safety.

Experimental research on implicit prejudice relies on the priming method, which “is an experimentally based method in which subjects are shown a word or image briefly before beginning a task – this display constitutes the prime” (Quillian, 2006, p. 315). Subjects are randomly primed with a racially linked word or image, while others are either not primed or primed with a neutral term (Quillian, 2006, p. 315).

In two studies demonstrating how implicit bias is activated, police officers were primed with a racially linked word (Graham & Lowery, 2004) or a racial image

(Eberhardt, Goff, Purdie, and Davies (2004). In their study, Graham and Lowery (2004) conducted an experiment with 105 police officers to determine how officers make decisions about juvenile suspects. The first part of the experiment involved the priming manipulation that exposed randomly assigned officers to a condition in which they were subliminally exposed to words related to the category Black (e.g., “homeboy”

“dreadlocks”) or to a condition that was neutral with respect to race or ethnicity (2004, p.

489). Each officer then read two vignettes written in the style of a crime report in which the suspect’s race or ethnicity was not disclosed. One vignette described a property crime

60 by a 12-year old boy with no criminal record and the other an assault by a 15-year old boy, although it wasn’t clear in the latter scenario whether the suspect was the aggressor or acted in self-defense. The authors “purposely chose relatively mild property and person offenses so that a range of options would be possible (e.g., let the suspect go with a warning vs. arrest on felony charges)” (2004, p. 490).

The authors find that police officers who were primed at the beginning of the experiment by the category “Black” were more likely to view the hypothetical adolescent offenders as more adult-like and judge them to have more generalized negative traits

(2004, p. 493). In addition, those officers judged the offenders to be more culpable, thereby justifying harsher sanctions (2004, p. 493). The authors concluded that the findings provide “partial support for our hypothesis that racial stereotypes could be unconsciously primed and that, once activated, those stereotypes would guide attributionally relevant judgments about the alleged offender and his crime” (2004, p.

494). In the other experiment involving priming, Eberhardt and colleagues (2004) recruited police officers and undergraduates to participate in five studies investigating the influence of stereotypic associations on visual processing. Across the five studies, the authors conclude that their results are consistent with other recent studies finding a strong association linking Black Americans and crime (2004, p. 889).

Research on the scientific foundations of bias thus suggests that racial bias “need not be explicit for policing to lead to racially disparate experiences,” and thus officers may not be consciously aware that their interactions with citizens differ by race (Kramer &

Remster, 2018, p. 966). Officers who possess social stereotypes or implicit biases may view the behavior of Blacks as suspicious and more dangerous than that of whites. Those

61 mental processes may then motivate their actions when interacting with Black citizens. In making decisions, including the use of force, officers “could, consciously, but without racial animus, take race into account” and “base their decision on beliefs about group criminality and who is most likely to be involved in crime” (Smith & Alpert, 2007, p.

1266). In other words, stereotyping Black individuals as more violent than white individuals may lead to biased interpretations of a Black subject’s behavior as more threatening, which in turn could lead to a more punitive use of force response when a

Black subject resists arrest.

Racial Threat = Fear?

The empirical literature discussed above suggests that officers may interpret the behavior of Black individuals as more threatening than that of white citizens. Officers who do so may also be motivated by fear for their safety in neighborhoods with greater percentages of Black residents. Research finds that fear motivates perceptions, not only of individuals, but of neighborhoods. Quillian and Pager (2001), for example, find that fear based on stereotypes may guide judgments and distort perceptions, including coloring perceptions of predominantly Black neighborhoods as areas of pervasive crime and violence. In particular, the authors find that the percentage of young Black males in a neighborhood is positively associated with white residents’ perceptions of crime rates in that neighborhood, regardless of actual crime rates or degree of physical deterioration within the neighborhood. Thus, residents base perceptions of the safety of their own neighborhoods, and their fear of victimization, on the race of their surrounding neighbors, particularly the number of young Black males.

62

Using data from the Bureau of Justice Statistics on murder cases adjudicated in 33 large urban counties in the United States in 1988, Baumer and Martin (2013) examine whether several features of community social organization and collective sentiment are associated with the prosecution, adjudication, and sentencing in those cases. Measures of community social organization and collective sentiment were aggregated from individual-level data in General Social Surveys from 1984 – 1988. One of the collective sentiment explanatory variables is fear of crime, measured by responses to the question of whether respondents were afraid to walk alone at night in their communities.

Employing multilevel regression models, the authors find that murder defendants receive longer sentences in jurisdictions with elevated levels of fear, as well as higher levels of social capital and greater endorsement of Christian fundamentalism. The authors speculate that judges, as elected officials, “may feel some pressure to take the public’s anxieties seriously and sentence more harshly in contexts of elevated fear[,]” where desire for severe punishment for murderers is particularly strong (2013, p. 171). Baumer and Martin conclude from their findings that legal outcomes are influenced by several features of the social environments in which cases are processed, including fear.

As this research makes clear, fear may influence both perceptions of crime as well as criminal justice outcomes. As applied to policing, officers who work in neighborhoods with large percentages of Black residents, particularly young Black males, may have greater concerns for their safety. In effect, racial threat and fear may operate in tandem to shape an officer’s perception of neighborhoods in which they patrol.

63

Neighborhood Racial Composition and the Use of Force

While research at the encounter-level regarding the effect of race on use of force is inconsistent (see Chapter 2), some empirical work that includes neighborhood-level variables indeed finds support for racial threat theory as an explanatory framework for the use of force. Those studies find a significant positive association between the racial and ethnic makeup of a community and the use of violence by the police. Consequently, police officers who associate Blacks as a group with criminality and danger may interpret the presence of a large and/or growing Black population in a neighborhood as more dangerous places, thereby presenting significant threats to their safety. This association may then influence the decision to use police coercion as well as the level of force against persons encountered in those neighborhoods.

Several studies examine the effect on police behavior of the racial and ethnic makeup at the city level. Liska and Yu (1992) used national homicide data collected from vital statistics and police records to test the hypothesis that the percent of nonwhites in a city positively affects the perceived threat they pose to the police, which in turn leads to a higher rate of police shootings. The authors find support for this interpretation of racial threat theory in that the percent nonwhite positively and linearly affects the overall rate of lethal force by police.

Using national data from municipal police departments in cities with a population greater than 150,000, Holmes (2000) finds support for racial threat theory in that the presence of “threatening people,” as measured by percent Black, percent Hispanic (in the

Southwest), and majority/minority income inequality at the city level, were positively related to average annual civil rights criminal complaints. Utilizing national data from

64 virtually all municipal police departments in cities with populations of 150,000 or more, which was compiled from more comprehensive sources than those used by Holmes

(2000), Smith and Holmes (2003) also find support for racial threat, in that percent Black and percent Hispanic (in the Southwest) were positively related to civil rights criminal complaints. In yet a third study using cities with populations of 100,000 or more, Smith and Holmes (2014) find that both percent Black and percent Hispanic were positively related to the incidence of excessive force complaints. In an analysis of police use of deadly force in all 50 states, Chamlin (1989) finds that percent Black and percent

Hispanic, but not economic inequality, significantly and positively predicted the rate of police killings.

Other studies have examined the effect of racial composition at the neighborhood or police precinct levels. Smith (1986) used the Police Services Study, composed of data from sixty neighborhoods located in three large U.S. cities, to examine the neighborhood context of police behavior. Significantly, Smith finds that the use of coercive authority is associated with the racial composition of neighborhoods, but not the race of the individuals confronted by police once neighborhood context was modeled.

Using data from a large, unnamed agency employing over 1000 sworn officers,

Lersch and colleagues (2008) find that use of force incidents were not randomly distributed throughout the city but concentrated in areas with larger numbers of minority residents. Utilizing the NYPD SQF data, Lautenschlager and Omori (2019) find that use of force is concentrated in Black neighborhoods and is more severe in those neighborhoods. Levchak (2017) also analyzed the NYPD SQF data and in a multilevel model finds that the percent of Blacks and the percent of Latinos in a precinct

65 significantly increases the odds of being subject to the use of force. Using the same data, however, Morrow et al. (2017) find that Black and Hispanic citizens were significantly more likely to experience non-weapon force than white citizens, but that precinct-level racial composition was not associated with levels of force. Similarly, Lawton (2007), using Philadelphia Police Department use of force reports, finds positive, but nonsignificant, effects of precinct level racial composition on the highest level of force used.

To summarize, police officers who perceive that Black citizens, particularly the stereotypical “young Black male,” are prone to criminality and present a threat to officer safety, may respond with higher levels of coercion in areas with relatively large and/or growing Black populations. In other words, police may associate the size of the Black population with the degree of dangerousness of the area. Some empirical research testing this aspect of racial threat theory indeed finds an association between racial and ethnic composition of an area and the police use of force within that area.

Black’s Behavior of Law: Social Periphery and the Use of Force

Donald Black’s Behavior of Law (1976) is a sociological explanation of how law behaves and varies across social space.4 In using the terms “behaves” and “varies,” Black is essentially describing how law functions in modern society. In any situation in which it is applied, law is the outcome variable, and is defined as governmental social control

(Black, 1976, p. 2).

Black characterizes law as a quantitative variable, as some situations may have more law than others (Black, 1976, p. 3). For example, in criminal matters, an arrest is significantly more intrusive than no arrest, and an indictment is more intrusive than no

66 indictment. Black would thus describe an arrest or an indictment as “more law” as compared to no arrest or no indictment, which are “less law” (Black, 1976, p. 3). The essence of Black’s theory is that social status is the critical determinant of how law/social control is distributed in society. Black thus contends that the quantity or amount of law/social control distributed by the state in any situation is determined by the relative social status of the parties, and not by the individual behavior or conduct at issue.

For example, Black’s theory (see below) predicts that a married, employed, white male would have higher social status than a poor, unemployed, Black male.

Consequently, in any dispute between the two, the quantity of social control delivered by the state would be based on their relative and quite different positions in social space, and not on the nature of the conduct complained of. In this instance, the law would favor the white male as a higher status person and disfavor the Black male as a lower status person.

Characterizing law as a quantitative variable means, in this instance, that the white male would receive “more” law relative to the Black male, who would receive “less” law, but would be treated more punitively by the law.

Social status is determined by five dimensions of social life. One of these dimensions, morphology, is “the horizontal aspect of social life” and reflects the degree of integration in society (Black, 1976, p. 37). Law varies directly with the degree of integration, such that a person integrated into society through barometers such as marriage, employment, home ownership, and organizational participation will receive more law relative to a person who is isolated within society. Those integrated into society are closer to the center of the social world while marginalized individuals are on its periphery.

67

Persons on the periphery of social life include racial and ethnic minorities living in poor, segregated neighborhoods, where one might also expect higher percentages of unemployed persons, persons on public assistance or living below the poverty level, and single-parent households. Individuals on the periphery of society are more vulnerable to law (Black, 1976, p. 52). As Black observes, “marginal people are disproportionately subject to law, and so, in this sense, are more deviant than other people” (Black, 1976, p.

54). As such, “a person without work, a family, or other involvements is more likely to get into trouble with the law” (Black, 1976, p. 55). Consequently, “[w]hatever goes wrong anywhere, those who are marginal to social life are more likely to be blamed . . .

[and] their conduct is more likely to be defined as deviant, and whatever they do is more serious” (Black, 1976, p. 59).

Black’s theory suggests that deviant behavior by those on the periphery of society is subject to a greater risk of severe sanctioning by the state. Police officers, particularly white police officers, will be closer to the center of the social world and, according to

Black’s theory, are more likely to view as deviant the conduct of minorities encountered in disadvantaged neighborhoods. It is thus arguable that application by the police of higher levels of coercion against minority suspects can be explained by their relative positions in social space. Simply stated, officers are more likely to view conduct by those at the periphery of society as deviant, more serious, and more deserving of harsh sanctioning than similar conduct by those more integrated and closer to the center of society. Although written before Black proposed his theory of law, Bayley and

Mendelsohn (1969) find that officers used force more aggressively and made more arrests in lower class and high-crime neighborhoods where greater social distance existed

68 between officers and citizens. Criminal sentencing research also finds support for Black’s morphology in sentencing decisions, including the decision to seek the death penalty.

Light, Massoglia, and King (2014) apply Black’s morphology dimension to disparities in incarceration of non-citizens, who, like racial and ethnic minorities, are also likely to be on the periphery of the social world. Judges are likely to see noncitizens as more deviant than citizens and consider their criminal conduct to be more serious.

Black’s theory thus suggests that noncitizens will be sentenced more harshly than citizens.

Using federal court sentencing data, the authors find that being a non-citizen is indeed a powerful determinant of punishment. In logistic regression models with citizenship status as their dependent variable (citizens as the reference category), non-citizens were over four times more likely to be incarcerated compared to citizens, including citizens of all racial and ethnic groups. When noncitizens were differentiated by legal immigrants and undocumented immigrants, both types of noncitizens were incarcerated more harshly than citizens, although undocumented noncitizens were incarcerated more often than noncitizens present in the U.S. legally. These effects held in the regression models analyzing sentence length, with noncitizens receiving significantly longer prison sentences.

Utilizing a data set of adult defendants indicted for capital murder in Harris County,

Texas, Phillips (2009) examines if victim social status, as measured by Black’s elements, including morphology, impacts the district attorney’s decision to seek the death penalty and the jury’s decision to impose a death sentence. Phillips finds that the decision to seek the death penalty was not related to vertical distance but was significantly more frequent

69 for more integrated victims (although that association became nonsignificant in the decision to sentence to death). Phillips also calculated predicted probabilities and found that the predicted probability of receiving a death sentence was roughly six times higher for a high-status victim compared to a low-status victim. Phillips then integrates the social status findings with his other research exploring the impact of defendant race and the type of legal counsel representing the defendant. He finds that the predicted probability of a death sentence for an indigent black defendant who killed a high-status victim was .42, compared to .02 for a white defendant who killed a low-status victim and could afford to hire legal counsel (2009, pp. 830-831).

Lawton (2007) tested whether Black’s theory predicted police use of force. Using officer force reports from the Philadelphia Police Department for one year, Lawton tested

Black’s theory in a multilevel model that included racial homogeneity at the police district level. Lawton finds no support for Black’s theory at the encounter level and only marginally significant support for race at the area level.

Summary and Research Hypotheses

To summarize, racial threat theory as applied to the behavior of police officers predicts that areas with higher percentages of racial and ethnic minorities will have higher levels of police use of force. Police officers, like other American citizens, may perceive Black individuals to be more violent and thus will conclude that areas with greater percentages of Black residents are more dangerous. More dangerous areas are perceived to pose more significant threats to their safety. Black’s dimension of morphology predicts that areas with higher percentages of marginalized persons, specifically racial and ethnic minorities, will have greater social distance between

70 residents and officers who patrol those areas, which may lead to higher levels of coercion. While both theories have been tested in separate studies, this study uniquely argues that the two perspectives are equally viable, complementary explanations for the disproportionate use of force against minority citizens.

The null hypothesis is that use of force is not related to community demographics, but only to encounter-level variables measured in this study, which include the number of officers, the subject’s race, sex, age, degree of impairment, level of resistance, and the officer’s race and sex. Theoretical perspectives explaining why use force may be related only to these encounter-level factors are discussed in Chapter 2 and are based on sociological, individual, and organizational factors. A use of force occurs during a micro- level encounter between an officer and a subject, and it may be that an officer’s actions are based only upon the salient aspects of the fast-moving encounter, including the subject’s demographics and behavior, the officer’s background and individual orientation, and the influence of organizational policies.

The alternative hypothesis, based on the research reviewed above, is that use of force is also influenced by the nature of the neighborhood where the encounter takes place.

Areas with higher percentages of racial and ethnic minorities may be perceived as not only more dangerous places, which can induce fear for personal safety, but also locations having a greater concentration of deviants, i.e., persons on the periphery of social life who are more deserving of social control. Accordingly, the two complementary theories of racial threat and Black’s Morphology generate the following research hypotheses:

Hypothesis 1a: After controlling for incident characteristics and the neighborhood crime rate, the number of uses of force will be higher in neighborhoods with larger percentages of Black and Hispanic residents.

71

Hypothesis 1b: After controlling for incident characteristics and the neighborhood crime rate, the severity of force will be higher in neighborhoods with larger percentages of Black and Hispanic residents.

Note another neighborhood influence on use of force may include its crime rate. Officers may respond with greater severity and frequency of force in high crime neighborhoods.

Accordingly, models testing the hypotheses in this Chapter will also control for census tract crime rates.

Disadvantaged Neighborhoods: The Influence of Place on Police Coercion

As Smith (1986, p. 337) observes, “[p]olice patrol both people and places” (emphasis added). Sociological research established long ago that places, particularly neighborhoods, have criminological consequences, including an association between poverty (typically an index of measures of disadvantage) and rates of violent crime (Kirk

& Papachristos, 2011; Krivo & Peterson, 1996; Sampson, Raudenbush & Earls, 1997).

Reviewing two meta-analyses covering the period from 1995 to 2006, Sampson and Bean

(2006, p.11) note that “concentrated disadvantage is the largest and most consistent predictor of violence across studies.”

The following discussion examines social disorganization theory and ecological contamination as independent but complimentary explanations for the influence of place.

This section assumes that place has a significant influence on police behavior that is independent of race. However, following the research hypotheses generated in tandem by theories of place, I discuss the possibility that race and place, because they are highly correlated, operate as joint influences on police use of force.

Social Disorganization Theory seeks to explain the association between neighborhood structure and crime. Social Disorganization Theory posits that neighborhoods with higher

72 levels of poverty, residential instability, and ethnic heterogeneity have weak social networks, which inhibit the neighborhood’s ability to exert informal social control over crime (Sampson and Groves, 1989; Sampson et al. 1997). Neighborhoods with weak social networks are unable to exercise adequate supervision of suspicious persons, including teenage peer groups hanging out on the streets. In their study using community- level data from the British Crime Surveys, Sampson and Groves (1989) find that unsupervised peer groups mediate the effect of neighborhood structural characteristics on crime and had the largest effects on victimization and rates of offending.

Police patrolling disadvantaged neighborhoods observe signs of disorder, which, in addition to unsupervised peer groups, include criminal gangs, drug , transient residents, boarded up structures, graffiti, and an ambience of despair. Some empirical research finds an association between neighborhood disadvantage and higher levels of police force (see below), although academics can only speculate about explanations for this association. For example, Parker and colleagues (2004) observe that visual signs of disorder in disadvantaged neighborhoods may provide cues to the police about the degree of incivility and crime within the neighborhood, which can increase proactive policing and escalate contact between residents. Herbert (1997) also speculates that location is crucial, because it may shape how officers view situations. In his view, “officers may understand an area to be characterized by a certain way of life that must be taken into consideration . . . [P]olice officers are acutely aware of the prevailing moral order in the areas they patrol and act in accordance more with that than with the law” (1997, pp. 35-

36).

In a similar observation, Kane states that,

73

[b]ecause many police officers are assigned to work in socially disorganized communities, and because they may respond to the same environmental cues to which members of the community respond, it might be that the same indicators of social disorganization that lead to increases in deviance among the general public also create contexts for

(2002, p. 872).

These hypotheses suggest that police officers patrolling disorganized neighborhoods may perceive visual signs of disorder and a higher rate of violent crime as factors that not only expose residents to higher risks of victimization, but also threaten their own safety while on patrol. Firearms are known to be more readily possessed and available in violent neighborhoods. A possible response to fear of violent victimization would be more proactive uses of force as well as higher levels of force.

In addition, officers working those neighborhoods day after day see the same people hanging out on the streets and routinely deal with the same offenders. Officers see people they arrest released almost immediately and back on the street by the start of their next shift. Thus, the police are reminded daily that deviance is intractable, and their arrest powers provide only a very temporary solution to routine disorder. Despite possessing the legal authority to address the quality of life and criminal issues endemic in poor, violent neighborhoods, the inability to effectuate meaningful change can make officers cynical and less confident in their moral authority. Trinkner, Kerrison, and Goff (2019, p.431) indeed find that “more cynical officers were less likely to see themselves as legitimate authorities, more likely to resist the department’s use of force policy, and less likely to endorse a procedurally just style of policing.” It would hardly be surprising, then, if fear of victimization, frustration, and cynicism motivate officers to respond to provocations in disorganized neighborhoods with higher levels of coercion.

74

As noted, several studies find an empirical association between neighborhood disadvantage and police coercion. Using the NYPD SQF data, Lautenschlager and Omori

(2018) find a significant positive association between concentrated disadvantage and frequency of force, but a negative associate between concentrated disadvantage and the severity of force. In a longitudinal study, Kane (2002) finds that officers patrolling poor, transient, and increasingly Latino neighborhoods across New York City police precincts incurred more frequent complaints of misconduct compared to officers in other neighborhoods. Sun, Payne, and Wu (2008) find that disorganized neighborhoods (as measured by concentrated disadvantage and percent senior) were more likely to experience police use of coercive force, which included nine items such as arrest, pain compliance, restraint, and drawing or discharging a firearm. Using data from the Project on Human Development in Chicago Neighborhoods, Hayes (2011) finds that residents of disorganized neighborhoods are victimized not only by the criminals who target their neighborhoods but also by the police through excessive use of force. Finally, Lersch

(1998) finds that complaints of excessive force were higher in in areas with lower median incomes and higher levels of poverty and unemployment.

One study suggests that the effect of disadvantage on use of force is more complex and is influenced by police organizational factors. Using official police use of force data collected in 73 large U.S. cities, Parker, MacDonald, Jennings, and Alpert (2005) find an indirect association between an index of concentrated disadvantage and police use of force. The authors find the effect of disadvantage is mediated by the presence of police unions, such that cities with higher levels of disadvantage were more likely to have police unions and higher rates of police use of force. Drawing from research, the authors posit

75 that “police departments in the cities that have union representation will have higher rates of force because the unions ensure legal representation to their members without providing much oversight or input into how they (the police) are to exercise their discretion” (2005, p. 70). Thus, police unions may increase use of force “through their impact on police culture and protecting individual officer discretion and autonomy”

(2005, p. 67).

As a body of research, empirical studies generally support the association between disadvantage and higher levels of police coercion. As one study conducted by Parker et al. (2005) suggests, this association may also be indirectly influenced by the protections afforded officers who are unionized. When officers are represented by a union and have collective bargaining rights, they may feel more emboldened to use force at higher rates when addressing problems in disadvantaged neighborhoods.

The RPD, the agency which is the subject of this study, has had a police union, the

Locust Club, since 1904. The RPD’s collective bargaining agreements with the city of

Rochester that were operable during the study period provide that officers accused of misconduct, which includes the inappropriate use of force, have due process rights

(Collective Bargaining Agreements expiring 6/30/12, 6/30/14, 6/30/16). These include notice, union representation, and the right to a hearing before an independent arbitrator before discipline can be imposed. Discipline in this context can range from up to 60 working days without pay or, the most drastic penalty, termination. Parker et al.’s research suggests that the protections afforded RPD officers, while unmeasured in this study, may indirectly influence how officers exercise discretion in using force in disadvantaged neighborhoods.

76

The disadvantaged neighborhoods described by social disorganization theory have the same ecological characteristics found in the work of Werthman and Piliavin’s (1967).

Werthman and Piliavin (1967) argue that police develop expectations about the nature of people they will encounter in poor neighborhoods in a process they call ecological contamination. They observed that

residence in a neighborhood is the most general indicator used by police to select a sample of potential law violators. Many local patrolmen tend to consider all residents of “bad” neighborhoods rather weakly committed to whatever moral order they make it their business to enforce, and this transforms most of the people who use the streets in these neighborhoods into good candidates for suspicion

(1967, p. 76, emphasis in original).

Consequently, a youth who is standing near a suspicious person in a disadvantaged neighborhood risks being contaminated by that association and drawn into his own encounter with the police (1967, p. 79). Bayley and Mendelsohn (1969) echoed the ecological contamination thesis, positing that an individual’s mere presence in a “bad” neighborhood (i.e., lower class and high-crime) increased the probability that the individual will be the recipient of police aggression, including being stopped, arrested and subjected to a use of force.

Analyzing observations of police-citizen interactions in Indianapolis and St.

Petersburg, Terrill and Reisig (2003) find that the level of force used was significantly influenced by the race and ethnicity of the suspect at the encounter level. However, this association became nonsignificant once neighborhood context (concentrated disadvantage and homicide rates) was introduced into the model. Arguing that their findings were supportive of the ecological contamination thesis, the authors observed,

77

[p]olice are significantly more likely to use higher levels of force when encountering criminal suspects in high-crime areas and neighborhoods with high levels of concentrated disadvantage independent of suspect behavior, officer, characteristics, and other statistical controls

(2003, p. 306).

Their findings indicate that race is confounded by neighborhood context and thus it is the degree of poverty and crime in a neighborhood, and not the suspect’s race, that influences the use of police coercion.

Summary and Research Hypotheses

Again, the null hypothesis is that use of force is unrelated to neighborhood characteristics and is determined solely by encounter-level variables. The research hypotheses are informed by the economic disadvantage literature discussed above. To summarize, social disorganization theory and the process of ecological contamination predict that officers will use higher levels of force in disadvantaged neighborhoods. The combination of high levels of poverty and crime causes officers to treat these neighborhoods as “bad” places populated by suspicious people. Officers may observe more threats to their safety in such neighborhoods and develop higher levels of frustration and cynicism from working there.

The two complementary theories of social disorganization and ecological contamination generate the following research hypotheses:

Hypothesis 2a: After controlling for incident characteristics, the number of uses of force will be higher in neighborhoods with higher levels of concentrated disadvantage and crime rates.

Hypothesis 2b: After controlling for incident characteristics, the severity of force will be higher in neighborhoods with higher levels of concentrated disadvantage and crime rates.

78

Concentrated Disadvantage and Racial Residential Segregation as Joint Influences on the Use of Force

As formulated in Hypotheses 1 and 2, the contextual effects of race and place are treated as being mutually exclusive. However, treating these effects as mutually exclusive ignores the reality of day to day patrol work. Disadvantaged neighborhoods are largely populated by Blacks. As Quillian and Pager observe, “the continuing severity of residential segregation remains a central feature of the African-American experience”

(2001, p. 717). A solid body of research has documented the racist and discriminatory housing practices that produced and still maintain residential segregation of Blacks in poor neighborhoods (see, e.g., Peterson & Krivo, 1999). Drawing on the National

Neighborhood Crime Survey, Peterson and Krivo (2010) find that 34% of Blacks live in neighborhoods that are 90% Black. Significantly, while only 1% of whites live in neighborhoods with extreme poverty, 15 to 25% of Blacks live in such neighborhoods

(Peterson & Krivo, 2010).

Peterson and Krivo (2010) refer to the different residential circumstances of Blacks and whites as a “racial-spatial divide.” Sampson and Bean (2006) similarly note that the ecological contexts in which Blacks and whites reside are quite different. They suggest that almost half of Blacks live in extreme poverty. Wilson (1987) also observes that poor

Blacks live in neighborhoods characterized by high levels of poverty while poor whites rarely live in such neighborhoods. Significantly, in a series of studies, Alba, Logan and their colleagues find that Blacks live in poorer and higher crime neighborhoods than whites at all levels of socioeconomic status (Alba, Logan, and Bellair, 1994; Logan and

Stults, 1999; Alba, Logan, & Stults, 2000).

79

The concentration of Black individuals in disadvantaged neighborhoods may be consequential for police decision-making. Officers who harbor implicit stereotypes of

Black individuals as violent criminals may view disadvantaged neighborhoods as dangerous places that present elevated threats to their safety because more Black individuals live and congregate there. If so, poverty may not represent a threat per se but instead presents a visual cue that poor neighborhoods will be dangerous areas because of their racial composition. Simply put, the contextual effect of concentrated disadvantage on police use of force may be indirect and mediated by the effect of race in segregated neighborhoods.

Summary and Research Hypotheses

The null hypothesis here is unchanged from the previous two sections, and states that use of force is influenced only by encounter-level variables. The research hypotheses are informed by the interactive effects of racial segregation and poverty as discussed above.

To account for the effect of poverty on racial composition, the research hypotheses now include concentrated disadvantage as a separate control. Accordingly, the interaction of disadvantage and race generates the following additional research hypotheses:

Hypothesis 3a: After controlling for incident characteristics, average crime rates, and concentrated disadvantage, the number of uses of force will be higher in neighborhoods with larger percentages of Black and Hispanic residents.

Hypothesis 3b: After controlling for incident characteristics, average crime rates, and concentrated disadvantage, the severity of force will be higher in neighborhoods with larger percentages of Black and Hispanic residents.

80

Chapter 4: The Influence of Antecedent Events on Police Use of Force

Introduction

The scholarship on police use of force has investigated the effect of racial demographics and, to a lesser extent, the influence of neighborhood characteristics, on the severity and amount of coercion to which citizens may be subjected. Research investigating neighborhood context suggests that police use of force may be significantly influenced by the racial composition and/or socioeconomic character of the neighborhood in which the encounter occurs.

While important, this research has largely ignored the temporal significance of the encounter. As Legewie observes,

research on the causal dynamics underlying racial profiling, the excessive use of police force, and other forms of discrimination has largely ignored an important aspect of the social environment in which discriminatory behavior occurs: the temporal embeddedness of social interactions and the potential importance of events

(2016, p. 380).

Legewie notes that the “focus on events helps us to extend previous work from where discrimination takes place to when it happens” (2016, p. 380, emphasis in original).

Two events that could potentially influence subsequent uses of force are a controversial, highly publicized use of lethal force by a police officer against an unarmed

Black citizen, or the murder of a white police officer in the line of duty by a Black citizen. This chapter uses as its factual predicates two such actual events: the shooting death of Michael Brown, an unarmed Black 18-year old male, on August 9, 2014 in

Ferguson, , and the shooting death of RPD Officer Daryl Pierson less than a

81 month later on September 3, 2014. Pierson, who was white, was on duty when shot by a

Black male recently released on parole.

The chapter proceeds as follows. The facts of each event are detailed, followed by theoretical arguments regarding how these tragic events could affect subsequent police uses of force. Empirical research relevant to these theoretical arguments is then discussed. Interrupted time series analysis of whether either event impacted subsequent uses of force by the RPD is presented in Chapter 5. However, I first discuss theoretical arguments explaining why any change in use of force may have a short duration.

Duration of Changes in the Use of Force

Research examining the effect of antecedent events on hate crimes (King & Sutton,

2013), attitudes toward immigrants following a terrorist attack (Legewie, 2013), and police use of force following the lethal shooting of an officer (Legewie, 2016), find that the effect of these events are temporary. For example, as noted above, Legewie (2016) finds that increases in the use of force against Blacks lasted about 10 days after the fatal shooting in 2011 and about 3.5 days after the fatal shooting in 2007 (2016, p. 399).

If the RPD increased use of force against Blacks following Officer Pierson’s shooting, it is likely that this effect also was temporary. As King and Sutton observe in the context of hate crimes, anger and prejudice are key emotions driving retaliatory hate crimes, and these emotions have a half-life, which could account for de-escalation (2013, p. 888). I similarly speculate that the effects of bounded solidarity, vicarious retribution, or racial threat/fear (theories discussed below) will abate over time, as they are also driven by anger, desire for retribution, and prejudice. While purely speculation, there is no reason to suspect that de-policing will also not be temporary. While officers may be

82 less proactive in the days immediately following events such as the high-profile shooting of an unarmed Black male or the death of a colleague, the impact of these events likely fade over time. The nature of the job does not change, and officers will continue to confront the same environmental conditions in the neighborhoods they patrol.

Violence by the Police: The Fatal Shooting of Michael Brown

A. Facts5

On August 9, 2014, just after noon, Ferguson, Missouri police officer Darren Wilson was on patrol on Canfield Drive in Ferguson. He observed Michael Brown and another

Black male walking eastbound in the middle of the street. Wilson had just heard a broadcast over his radio for a “stealing in progress” of several packages of cigarillos at a nearby convenience store. Wilson observed that Brown had cigarillos in his hand and that he and his friend matched the suspects’ description. Physical and forensic evidence and eyewitness testimony establishes the sequence of events that occurred over the next two minutes.

Wilson used his patrol vehicle to block the men’s path. As Wilson tried to exit the vehicle, Brown punched Wilson through the open window. Wilson withdrew his firearm and Brown allegedly grabbed it, which initiated a struggle to gain control of the weapon.

Wilson fired the gun twice, and one round struck Brown in the fleshy part of his hand.

Brown then ran more than 180 feet away from Wilson’s vehicle, as verified by bloodstains on the roadway that DNA testing matched to Brown. Blood evidence and witnesses confirm that Brown then ran back towards Wilson. Wilson stated he saw

Brown reach his right hand under his t-shirt into what appeared to be his waistband. No

83 evidence emerged upon which prosecutors could rely to disprove Wilson’s contention that he feared for his safety.

Wilson fired ten shots at Brown as he approached, killing him with a shot to the head.

Brown fell to the ground with his uninjured hand balled up by his waistband. Witness accounts and cell phone video establish that Wilson did not touch Brown’s body after he fired the final shot and Brown fell to the ground. Subsequent accounts by some individuals purporting to be witnesses stated that Brown had his hands raised in the air in a gesture of surrender when he was shot. However, these statements were inconsistent with physical and forensic evidence, the statements of other witnesses, and in some cases, with the witness’s own, prior statements to the police. Other witnesses recanted their original statements about Brown’s gesture of surrender, admitting that they did not actually witness the shooting. Based on the totality of the evidence, federal and state prosecutors declined to charge Wilson with murder or criminal civil rights violations.

B. The Effect of Ferguson and Its Aftermath on Perceptions of Police Legitimacy

News of Brown’s shooting quickly spread beyond Ferguson. As Nix and Pickett observe, “[a]lthough allegations of excessive force against minorities are certainly nothing new, the quickness with which news of the Ferguson shooting spread through social media and mainstream news outlets was unprecedented” (2017, p. 25). Citizens and media from outside the Ferguson community descended on the city immediately after the shooting. Initially peaceful protests/demonstrations turned violent, resulting in property damage and looting. A nightly curfew was imposed, and a state of emergency declared by the governor, who mobilized the Missouri State Highway Patrol and the

National Guard in Ferguson (Deuchar, Fallik & Crichlow, 2019).

84

Although Brown’s death was just the first of several high-profile deadly force incidents involving the police and unarmed Black males6, it was the catalyst for intense public and media scrutiny of the police and gave birth to the national activist movement known as “.” As Culhane, Bowman, and Schweitzer observe,

Ferguson “was clearly the spark that lit the match” (2016, p. 253). In the aftermath of the shooting, the narrative that Brown, an unarmed Black male shot while attempting to surrender to a white police officer, took hold and fueled social media posts and national demonstrations. The “hands up, don’t shoot” mantra shouted during marches was grounded in this narrative.

Weitzer (2015) observes that a series of deadly force incidents occurring within a compressed time period “can tarnish the image not only of the police in the cities where the incidents took place but can also damage the reputation of the police nationwide”

(2015, p. 475). Cell phone video of some of the lethal confrontations coupled with the widespread use of social media have moved these deadly force incidents from the

“jurisdictional bounds of the city” in which they occurred to the national spotlight (Wolfe

& Nix, 2016, p. 2). Consequently, these incidents gained massive traction in the local and national media. Activists and the media drew connections between these separate incidents, creating “contamination-by-association” (2015, p. 475). Weitzer observes that:

Time and again, we have seen protestors on the streets and commentators in the media (1) operating with a presumption of guilt toward the officer or officers involved, (2) asserting that misconduct is widespread and systemic, not confined to a few rogue cops, and (3) imputing racial animus as a motive

(2015, p. 475-476).

Deuchar and colleagues (2019) similarly observe that high profile officer-involved killings of unarmed Black males are no longer seen as an anomaly but as a pattern of

85 behavior. This shift in perception has in turn affected perceptions of police legitimacy.

The foundation of police legitimacy is the moral and legal authority that compels citizens to voluntarily comply with officer commands (Gau, Corsaro, Stewart, & Brunson, 2012).

As Tyler (2000) observes, police legitimacy reflects respect for the police that goes beyond mere fear of their authority. Legitimacy means that the public perceives justice is being administered fairly and without bias.

Recent research suggests, however, that public perceptions of police legitimacy are being negatively impacted by media coverage of these lethal force incidents. Deuchar and colleagues (2019) note that Gallup polls taken in 2014 and 2017 reported that less than

57% of respondents had a great deal of confidence in the police, compared to 88% in

2008 (2019, p. 1045).

Culhane and colleagues (2016) conducted two studies, one prior to Ferguson and one after, with participants from across the United States. Each study included between 400 and 500 participants, the majority of whom were white, female, and with an average age in their mid-thirties. In both studies, participants were told that they would be presented with an incident in which a white male named Eric Johnson was shot by a white police officer, who was responding to a hit-and-run investigation. The mode of the presentation was manipulated across three conditions: a video of an actual police shooting taken by a body-warn camera, just the audio from the body-worn camera, or the transcript of the audio. In the video, the officer is told by the suspect’s girlfriend that the suspect has a knife and may want to harm himself. After multiple attempts, the officer is finally able to call out the hit-and-run suspect, who walks toward the officer with a knife in his hand.

86

The suspect ignores multiple orders to drop the knife and is shot five times. Participants were asked whether the shooting was justifiable or unjustified.

Overall, the authors find that 39% of participants in both studies felt the shooting was justified, while 61% felt that it was unjustified. However, in the first study, participants were significantly more likely to perceive the shooting was justified if they could hear or see the event compared to those who just read a transcript. After Ferguson, the second study finds that participants who viewed the video were significantly more likely to feel the shooting was unjustified. While there was no significant difference between the views of non-white and white participants before Ferguson, after Ferguson non-white participants were significantly less likely to think the shooting was justified. The authors thus find that “there was a sudden and drastic changing of American citizens’ opinions after the Ferguson incident of the same police shooting as shown in Study 1 with the same experimental conditions” (Culhane et al. 2016, p. 267, emphasis in original).

As Culhane and colleague’s findings suggest, the apparent decline in law enforcement legitimacy is not evenly distributed by race. Racial and ethnic minorities are generally more aware of disparate outcomes by race when evaluating the police. Using a panel survey of St. Louis County, Missouri residents conducted before and after Michael

Brown’s shooting, Kochel (2019) finds that Black residents had significantly less trust in the police and perceived them to have less legitimacy after Ferguson, while white residents’ views remained stable. Gau and colleagues find that Black citizens’ negative views of the police are substantially mediated by area rates of concentrated disadvantage, which suggests that police lack legitimacy in neighborhoods where Black residents feel profiled and harassed (2012, p. 339).

87

C. The Impact of Post-Ferguson Publicity on Police Self-Legitimacy and Morale

As research suggests, negative publicity following Michael Brown’s death and its effect on perceptions of police legitimacy have had profound consequences for law enforcement. Indeed, Nix and Pickett (2017) argue that Brown’s death was an

“environmental jolt” to policing in the United States in that it was a “sudden and unprecedented event” that had a “disruptive and potentially inimical” impact

(2017, p. 25). Deuchar and colleagues similarly observe that protests and anti-police rhetoric “have taken a toll on law enforcement” (2019, p. 1045). This “toll” involves an impact on officers’ perception of self-legitimacy, which is the confidence they have in their authority (Nix & Wolfe, 2017).

Self-legitimacy has important influences on police behavior. For example, Tankebe and Mesko (2015) find that officers who had a greater sense of self-legitimacy were more likely to attempt to verbally diffuse a hypothetical scenario involving a drunk and agitated citizen as opposed to threatening to use force. Bradford and Quinton (2014) find that officers with higher levels of self-legitimacy were more likely to treat citizens fairly, i.e., with procedural justice. These findings suggest that officers who were confident in their authority may be more committed to using fair procedures when interacting with citizens, including a greater willingness to use their verbal skills as opposed to the use of force (Nix & Wolfe, 2017).

Empirical studies with police officers confirm that police self-legitimacy and morale have been negatively impacted by negative publicity. Using a survey of 567 sworn deputies serving a sheriff’s department in a southeastern U.S. city with 393 residents, Nix and Wolfe (2017) measured the participants’ sense of self-legitimacy. Participants were

88 asked to express agreement or disagreement on a 5-point Likert scale with five statements designed to measure self-legitimacy. The focal independent variable for the study was negative publicity, also measured along a 5-point Likert scale in response to three statements. The authors find that negative publicity, even if related to events occurring far from their jurisdiction, reduced officer motivation and was associated with lower feelings of self-legitimacy. Although unrelated to feelings of self-legitimacy, officers also felt “that negative publicity has made it more dangerous and difficult to work in law enforcement” (2017, p. 101).

Similarly, Deuchar and colleagues (2019) conducted semi-structured interviews with

20 law enforcement officers in two southern counties in the United States. Interviews were supplemented with participant observations during ride-a-longs. The authors find that officers experienced decreased morale and desire to engage in proactive policing, including uses of force, following Ferguson. As one officer stated, fear of being confronted with the ‘race card’ by a minority citizen causes hesitation to stop citizens but also to use any kind of force (2019, p. 1049). The officer added, “You know, I think instead of being primarily concerned with their own safety that hesitation comes into play as to ‘how am I going to be viewed?’” (2019, p. 1049).

Using data obtained from a survey of officers employed in a large police department in a major southeastern U.S. city with roughly 615,000 residents, conducted in the

Autumn of 2016, Nix and Pickett find that

officers who believed media coverage of the police had been more hostile in recent years were significantly more likely to believe that civilians had become more distrustful, resentful, and disrespectful of police. Such officers were also more likely to report greater fear of being falsely accused of wrongdoing by civilians.

89

(2017, p. 30).

Finally, Nix and Wolfe (2016) explore whether sensitivity to the Ferguson effect was related to the management culture of the agency in which officers were employed. In early 2015, the authors surveyed 510 sworn sheriff’s deputies in a police agency in a large southeastern U.S. city. Their focal independent variable was “organizational justice,” which is the perception by employees that their organization is fair. Factors affecting perceptions of organizational justice include whether salary and promotion decisions are based on merit, not “who you know,” the degree to which employees are treated with respect and politeness by supervisors, and whether organizational decisions are clearly explained, unbiased, and allow for employee input (Nix & Wolfe, 2016, p.

14).

The authors measured procedural justice with 18 study items on a 5-point Likert scale designed to capture perceptions of procedural, distributive, and interactional justice within the respondent’s agency. The authors find that officers who felt that their agency was procedurally fair “were less likely to report (1) being unmotivated, (2) that law enforcement has become more dangerous, (3) that their colleagues have been impacted by negative publicity, and (4) that citizens’ attitudes (both nationally and locally) toward the police have worsened” (2016, p. 17). The authors conclude that police supervisors whose management style incorporates the principles of organizational justice are more likely to create an atmosphere within their agency that insulates officers from the negative effects of post-Ferguson publicity.

In a similar study using data from a survey of 567 deputies in an agency in the southeastern U.S., Wolfe and Nix (2016) tested whether the Ferguson effect impacted

90 officers’ willingness to partner with community members. Research establishes that strong police-community partnerships lead to safer neighborhoods through fewer reports of disorder-related problems as well as higher levels of perceived safety (Reisig & Parks,

2004). In addition, police-community partnerships can increase citizen satisfaction with the police and increase (Mastrofski & Greene, 1994). The authors find that officers were, indeed, less willing to engage in community partnerships after

Ferguson. However, this effect became insignificant once organizational justice and self- legitimacy were modeled. The authors conclude that negative publicity has little effect on willingness to partner with community members when officers have confidence in their authority or perceive their agencies as fair.

D. The Ferguson Effect: De-policing

The research in Part C confirms that officers are aware of negative media coverage and anti-police rhetoric, which, without a supportive organizational culture or confidence in their authority, can influence their perceptions of citizens and heighten their fear of scandal (Nix & Pickett, 2017, p. 25). Scholars and commentators have questioned whether the effect of the “massive, Ferguson-driven backlash” to police across the United

States has led to de-policing. (Culhane et al. 2016, p. 252). “De-policing” is the act of

“retreating from active police work in reaction to the negative publicity that has been placed on police agencies across the county” (Shjarback, Pyrooz, Wolfe & Decker, 2017, p. 42). The practical effect of de-policing is that officers may be hesitant to be proactive while on patrol for fear of being subjected to media attention and allegations of excessive force and racial profiling (Pyrooz, Decker, Wolfe, & Shjarback, 2016, p. 1). The phenomenon of de-policing has been named, not surprisingly, the “Ferguson effect.”

91

As Oliver observes, if “the police are not engaging in crime fighting, but are rather disengaging, there is a reasonable belief that crime will increase because the police will have effectively abandoned certain communities out of fear of repercussions” (2015, p.

438). The issue then becomes whether de-policing is a real phenomenon and, if so, whether it has had consequences for the communities in which it is occurring. Empirical studies have attempted to determine whether changes in police activity have occurred in the post-Ferguson era, and whether there is an association between these changes and outcomes like crime rates or police shootings.

For example, using data from 118 of the 121 police departments serving jurisdictions of over 5000 residents in Missouri, Shjarback and colleagues (2017) examined changes in the rates of vehicle/traffic stops, searches, and arrests from 2014 to 2015, as well as changes in the quality of stops, as measured by “hit rates” from searches. The authors find mixed evidence of the quantity of de-policing, in that agencies made about 67,000 fewer vehicle/traffic stops in 2015 compared to 2014. However, while the number of searches and arrests also decreased in 2015, the decreases failed to reach statistical significance. Regarding the quality of policing, the contraband hit rate increased between

2014 and 2015, as items were confiscated 28.6% of the time in 2015 compared to 26.7% of the time in 2014 (2017, p. 46).

The authors conclude that officers were making fewer, but better stops and making searches that more consistently yielded contraband. Significantly, Shjarback and colleagues also find that agencies made fewer vehicle/traffic stops, searches, and arrests in 2015 compared to 2014 in jurisdictions with larger Black populations. The authors thus

92 argue that, post-Ferguson, “context – especially the racial composition of cities – shapes de-policing behavior” (2017, p. 50).

Pyrooz and colleagues (2016) examined whether there was a Ferguson effect on crime rates. Data for their study was based on aggregate and disaggregate monthly

Part 1 crime data gathered 12 months before and after August 2014 from police departments in 81 large U.S. cities with populations greater than 200,000. Specifically, the authors focused on violent offenses included in the Part 1 Uniform Crime Reports including criminal homicide, forcible rape, robbery, and aggravated assault, and measures of property crime, including burglary, larceny, and motor vehicle theft (2016, p. 3).

Using a discontinuous growth model to model within-city changes in crime trends,

Pyrooz and colleagues find no evidence to support a systematic change in overall, violent, and property crime trends. Based on disaggregated data, however, they did observe an increase in robbery rates, which had been declining before Ferguson. In addition, they observed increases in homicide in cities with much higher crime rates before Ferguson, such as Baltimore, St. Louis, Newark, New Orleans, ,

Washington, D.C., and Rochester, N.Y. (2016, p. 5). The authors conclude that any

Ferguson effect on crime rates is limited to “cities with historically high levels of violence, a large composition of black residents, and socioeconomic disadvantages,” as well as a greater number of police per capita (2016, p. 1). The magnitude of the effect was small, however, as the authors find that it would take almost two years to experience a one-unit increase in homicides, on average.

Campbell, Nix, and Maguire (2018) examined whether there was a change in the number of citizens fatally shot by police in the post-Ferguson era. The authors draw on

93 defiance theory as a theoretical explanation to explain why an increase in lethal shootings could occur. Sherman (1993) argued that unjust and excessive sanctions by government officials such as the police may result in defiance among citizens as opposed to exhibiting a deterrence effect. Defiance could result in more deadly force incidents “due to citizens’ increased willingness to defy and violently resist officers” (Campbell et al. 2018, p. 404).

Conversely, the authors speculate that de-policing in the wake of Ferguson may have led to less proactivity and greater fear of using force, which could translate into a reduction in the number of deadly force incidents (Campbell et al. 2018, p. 405).

The authors analyzed a data set of 3,055 civilians killed by the police between May 4,

2013 and December 25, 2015, obtained from the “Killed By Police” website, which uses news articles and crowdsourcing efforts to compile data on all citizens killed by the police. Using an interrupted time series design, the authors find that the use of lethal force by the police did not significantly change in the 16 months following the Michael

Brown shooting (2019, p. 411). As such, there was neither a decrease due to a Ferguson effect nor an increase in lethal shootings as the result of citizen defiance.

E. Summary and Research Hypothesis

Following the fatal shooting of an unarmed Black male by a white police officer, the null hypothesis is that use of force by the police will remain unchanged. Theoretical explanations for the null hypothesis are discussed in the next section. Research discussed above generates the alternate hypothesis, which is that negative publicity and its effects on police self-legitimacy lead to de-policing and a reduction in uses of force.

Accordingly, the research on de-policing generates the following research hypothesis:

94

Hypothesis 4a: Compared to the 20 months prior to Michael Brown’s shooting, the number of force incidents and severity of force used by RPD officers decreased in the days immediately after Brown’s shooting on August 9, 2014.

Violence Against the Police: The Fatal Shooting of Officer Pierson

In a study that appears to be the first of its kind, in that it specifically examines the effect of antecedent events on police use of force, Legewie finds support for the proposition that violence against the police motivates violence by the police. Using SQF data from the NYPD, Legewie (2016) finds that two lethal shootings of police officers by

Black suspects lead to substantial increases in use of force against Black subjects, but not other groups, stopped in the three days after each event. This effect was temporary, however, as the increase in force lasted about 10 days after the fatal shooting in 2011 and about 3.5 days after the fatal shooting in 2007. The RPD also experienced the sudden, violent death of a police officer, which occurred less than a month after Michael Brown’s shooting.

A. Facts

On September 3, 2014, Officer Daryl Pierson, an 8-year veteran, age 32, was shot and killed while on proactive patrol as a member of the RPD’s Tactical Unit. Riding as the passenger with Officer Mike DiPaola, Officer Pierson stepped out of his vehicle to question a person who was believed to have earlier bailed out of a suspicious vehicle driven by a warrant suspect. Upon seeing Officer Pierson, the suspect, who is Black, ran and Officer Pierson chased him. As Officer Pierson was within reach, the suspect turned and shot him in the throat. Officer DiPaola, who had followed in the patrol car, engaged the suspect on foot in a gun battle that left the suspect wounded.

95

Additional Tactical officers arrived, placed Officer Pierson in a patrol car, and took him to nearby Rochester General Hospital. Officer Pierson died in the emergency room.

The murder of any police officer is a tragic event for family, friends, fellow officers, and the community. However, Officer Pierson’s murder was especially traumatic because it was the first fatal shooting of a Rochester police officer since Officer Harold V. Shaw was fatally wounded on December 11, 1959 while investigating a commercial burglary.

Officer Pierson left behind a wife and two young children.

B. Applicable Theory

Theories that suggest pathways from an event such as Officer Pierson’s murder to an immediate increase in coercion, particularly against individuals of the same race as the assailant, are Bounded Solidarity, Vicarious Retribution, and Racial Threat, all discussed in the next section. Of course, it is also possible that the level of police coercion remains unchanged, or even decreases, after such an event. Theoretical explanations for why levels of force may remain unchanged or even decrease are discussed below.

C. Bounded Solidarity: Group-Oriented Responses to a Common Enemy

News that an officer has been shot spreads quickly throughout the officer’s agency.

Understandably, an attack on an officer generates anger and emotions in the police community (Legewie, 2016). As the Democrat & Chronicle, the local Rochester newspaper, reported two days after the shooting, “[e]motions – anger, sadness, despair – have been discussed since word of Pierson’s shooting began to spread among officers and the community about 9 p.m. on Wednesday” (Democrat & Chronicle, 9/5/14). Following any shooting of a police officer, officers, both on- and off-duty, respond to the hospital where the wounded officer has been taken and establish a vigil while waiting for

96 information on the officer’s condition.7 Officer Pierson’s shooting was no exception, as officers gathered at Rochester General Hospital to await word on his condition and commiserate amongst themselves. The Democrat & Chronicle (9/5/14) described how one off-duty officer went to the hospital “with everybody else” after he began receiving texts that an officer had been shot.

Papachristos’ (2009) work on the response of gangs to an act of violence against a group member is instructive for how police may feel and respond. Papachristos (2009) argues that acts of violence against a gang member are perceived as threats to all gang members, which then strengthens internal cohesion and solidifies group identity. Internal cohesion and a common group identity become a form of social capital, which Portes and

Sensenbrenner (1993) call bounded solidarity. Bounded solidarity emerges out of situations that produce “principled group-oriented behavior” (Portes & Sensenbrenner,

1993, p. 1324). They argue that its classic sources are exemplified by the rise of class consciousness and resulting class struggle described by Marx, in which the weapon of the working class was its internal solidarity that emerged out of common awareness of capitalist exploitation (1993, p. 1324).

The authors give several examples of how bounded solidarity was created among immigrant groups facing nativist prejudice upon arriving in America. For example,

Chinese immigrants to New York and San Francisco were subjected to discrimination but lacked the means to return home. Unable to find employment in factories and barred from bringing wives and other family members to America by the Chinese Exclusion Act,

Chinese immigrants banded together in tightly knit communities that grew into today’s

Chinatowns (1993, p. 1328). Outsiders noted the “clannishness” and “secretiveness” of

97 these communities, which represented solidarity born out of common adversity (1993. p.

1328). As such, Portes and Sensenbrenner argue that bounded solidarity is a “situational reaction of a class of people faced with common adversities. If sufficiently strong, this emergent sentiment will lead to the observance of norms of mutual support . . .” (1993, p.

1325).

Undoubtedly, the violent death of a group member is an event that generates mutual support among surviving group members, which can also focus the group’s attention on members of the offending group as a common enemy. Papachristos (2009) finds that an attack on a group member creates intergroup conflict that can trigger a violent response as a form of “self-help” against a member of the offending group. He observes that a

“group can be seen as cohesive and provide its members status and protection only if the group can continually assert its dominance by protecting its members and turf and thereby avoid subjugation” (2009, p. 119). Gangs seek to avoid defeat and subjugation in the present and in future interactions (Papachristos, 2009). Similarly, Black (1983, p. 34) argues that violent acts are a form of “self-help” that “express a grievance by one person or group against another.” Collectively, this research suggests that group members adopt the belief that an act against one is an act against all, which can produce feelings of group cohesion and solidarity.

As applied to the police, “bounded solidarity” among officers provides social capital upon which they can draw following a violent event against one of their own. Solidarity based on intergroup conflict may generate the collective belief that violence against an officer threatens the moral authority as well as the safety of all officers. It is thus conceivable that the police may respond to the murder of a colleague by acting

98 aggressively as their form of self-help, particularly towards members of the offending group. In the context of the murder of a police officer by a Black suspect, police may collectively respond by increasing use of force against Black suspects with whom they come in contact, who may be viewed as the common enemy. Collectively responding with higher levels of physical aggression against a common enemy symbolizes their solidarity and may serve as the primary means by which officers reassert their authority and avoid subjugation.

D. Use of Force as Vicarious Retribution

Using physical force against members of an offending group may not only reassert police authority and maintain internal solidarity but may also represent retaliation for the attack on a colleague. Responding with physical violence may be viewed by some officers as “street justice” following an attack on one of their own. In a similar but less threatening context, Robert Vargas (2016) illustrates why this is possible. In his ethnographic study of Little Village, a largely Hispanic neighborhood in Chicago, Vargas documents how Chicago police sometimes retaliated against residents who either wasted their time with frequent, frivolous 911 calls, gave officers a hard time, or disrespected officers. Officers retaliated by disclosing informants’ identities to the Latin Kings gang.

Officers referred to outing an informant as “street justice,” which would be administered by the Latin Kings against the informant for being a “snitch” for the police (Vargas,

2016).

If simply showing disrespect or wasting their time motivated officers to intentionally expose informants’ identities to a violent street gang, knowing that it would subject them to violence, it is more than likely officers would similarly retaliate against members of

99 the offending group following the murder of a police officer. Increasing the level of coercion against a common enemy may thus be characterized as a campaign of vicarious retribution by the slain officer’s colleagues.

Lickel and colleagues (2006) describe vicarious retribution as a social-psychological model of one major aspect of intergroup aggression. The authors define it as the commission of an act of aggression by a member of a group “toward members of an outgroup for an assault or provocation that had no personal consequences for him or her, but did harm a fellow ingroup member” (2006, pp. 372-373). The model predicts that members of an ingroup experience anger, sadness, and outrage following an assault against a colleague, which then motivates acts of retaliation against members of the outgroup (2006, p. 380).

The authors contend that vicarious retribution has cognitive and motivational roots.

The first of these is “event categorization.” An individual may perceive the intergroup nature of an act of aggression by a member of an “outgroup” against a member of their

“ingroup” as a way to cognitively make sense of the event (Lickel et al. 2006, p. 376).

When act identification occurs within an intergroup context, it is subject to biases in its interpretation. The authors observe that focusing on an ingroup – outgroup categorization to understand an event may cause individuals “to construe the intentions and actions of the individuals in the event in ways that cast ingroup members in a favorable light and outgroup members in an unfavorable light” (2006, pp. 376-377).

Once an event has been characterized as being relevant to an individual’s intergroup membership, that individual’s motivation to retaliate will be a function of several variables. These include 1) the individual’s degree of identification with other members

100 of the ingroup, including the degree of self-worth and self-esteem they feel by being a member of that ingroup; 2) the degree of empathy an individual experiences after witnessing another person’s “distressed condition or emotional state,” which may be more intense when an individual identifies with the person who has experienced the harm; and 3) when retaliation by members of the ingroup against members of the outgroup is perceived as being normatively appropriate, which can occur when retaliation on behalf of a harmed ingroup member goes unpunished or confers increased respect and status within the group (Lickel et al. 2006, pp. 377-378).

In summary then, group pride, empathy, and normative pressures motivate ingroup members to engage in a “revenge based vicarious retribution process” against members of the outgroup (2006, p. 378). This can occur even when the original “provocateur” of the outgroup is not available for retaliation. In this case, retaliation by members of the ingroup is more likely when the outgroup is perceived to be “a unified and coherent whole” and thus bonded together (2006, p. 378). This perception allows ingroup members to blame all members of the outgroup as having an indirect causal role in the individual outgroup member’s actions (2006, p. 379).

Arguably, vicarious retribution is a behavioral model that predicts how fellow officers, i.e., ingroup members, could react to the murder of a colleague, which is a highly traumatizing “event.” Police may be especially susceptible to the type of intergroup conflict described by Lickel and colleagues because they are socially isolated in both their work and personal environments (Griffin & Bernard, 2003). The nature of the job, with its element of danger, imposes social isolation and causes officers to seek social and emotional support from each other (Griffin & Bernard, 2003). Officers are

101 suspicious or distrustful of civilians and other actors in the criminal justice system, as well as of their superiors and command staff (Griffin & Bernard, 2003). Moreover, officers “are frequently introduced socially as police officers. Current and former officers alike can easily identify with fellow tales of being introduced as ‘my friend John, the cop’” (Griffin & Bernard, 2003, p. 14). Officers are thus continually reminded, both on- and off duty, that their profession separates them from other citizens.

Given the nature of their profession, and the social isolation, suspicion, and distrust it produces, officers understand that they belong to a unique “ingroup.” The murder of a colleague is a tragic event, signifying that a member of their ingroup was attacked by an outsider solely because of his/her membership in the group. It is likely that officers will identify and empathize with the slain officer, which can intensify their emotional response to the tragedy.

As the Democrat and Chronicle reported on September 5, 2014, two days after

Officer Pierson’s murder, officers experienced feelings of anger, shock, sadness, and despair. Lickel et al. (2006, p.380) observe that ingroup members often experience anger, sadness, and outrage following an assault on a colleague. Group pride conferred by membership in a unique profession, coupled with strong emotional responses to the murder, could produce an intense desire for retaliation against members of the outgroup.

Retaliation may be an appropriate normative response under the circumstances, particularly if officers voice their approval (among each other) of using force as “street justice.” Moreover, because the “provocateur” is most likely in custody and thus not available for retribution, members of the “outgroup” will then become targets.

102

When the officer is white and the suspect is Black, officers may implicitly blame all members of the subject’s race for the actions of one individual. This may be particularly true for those officers who hold implicit biases associating Blacks with criminality. These officers might then blame the perceived violent nature of Black individuals, and not the action of one individual, as the cause of the murder. If so, all Black people, particularly young Black males, may be blamed for having “an indirect causal role” in the action of their outgroup member.

Significantly, Lickel et al. (2006) argue that retaliation will be greater for a public event compared to a private event. The authors suggest that “a public context should promote stronger retaliation because it poses a stronger threat to group pride” (2006, p.

382). In addition, a public event has an audience, which “increases the salience of social identities and magnifies the humiliation associated with affronts” (2006, p. 382).

The murder of a police officer is quintessentially a public event, reported by the media locally and perhaps even nationally. The event conveys to the public the impression that police officers are vulnerable to attack by individuals who commit the ultimate act of “thumbing their nose” at police authority. It is not inconceivable that officers may view the death of a colleague as a “public humiliation,” which may strengthen their desire for revenge. In this context, then, a public affront that highlights the salience of group membership and produces strong emotions could result in increased physical aggression by the slain officer’s colleagues against members of the outgroup, be it racial, ethnic, or other (e.g., outlaw bikers), to which the suspect belongs.

King and Sutton (2013) draw on vicarious retribution as an explanation for the temporal clustering of hate crimes, which may be retaliatory in nature. The authors use

103 three types of events as antecedents for hate crimes: contentious criminal trials involving interracial crimes, lethal terrorist attacks, and appellate court decisions concerning same sex-marriage (2013, p. 871). Interrupted time series analyses find an association between contentious trial verdicts and lethal domestic terrorist attacks and a temporal increase in racially or religiously motivated hate crimes. However, the authors find less evidence for anti-gay hate crimes.

E. Racial Threat Activated by Implicit Racial Stereotypes

Collectively, the theoretical insights provided by “bounded solidarity” and “vicarious retribution” suggest that increasing the use of force after an attack on a police officer

“ensures social control, (re)asserts authority, and retaliates against the offending group”

(Legewie, 2016, p. 386). However, it is also possible that increases in force may not be the result of purposeful action. As discussed in Chapter 3, racial threat theory predicts that more punitive actions toward Black individuals will occur because of implicit racial stereotyping.

Experimental research establishes that some officers, like other citizens and criminal justice actors, have implicit biases that stereotype Black people as violent or dangerous.

This research also finds that these stereotypes or implicit biases can be activated, or primed, by relevant events, including crime events (Eberhardt et al. 2004; Graham and

Lowery, 2004; Legewie, 2013). In addition to Eberhardt et al. (2004) and Graham and

Lowery’s (2004) experimental research on priming effects, Legewie (2013) finds that actual events can activate biases. In his research, terrorist attacks have a profound, albeit short-term, effect on citizens’ perception of immigrants as dangerous.

104

Legewie’s subsequent study of the NYPD (2016) suggests that the murder of a white police officer by a Black perpetrator may similarly activate implicit racial stereotypes, which could increase concerns for officer safety and raise levels of fear when dealing with Black citizens. Consequently,

[p]olice officers might be more likely to interpret identical actions – implicitly or explicitly - as hostile or threatening and react to behavior that went unnoticed or without consequences before a certain event with the use of physical force after the event

(Legewie, 2016, p. 387).

As discussed in Chapter 3, perceptions of threat posed by Black individuals may influence the use of force in areas with larger percentages of Black residents even before the murder of a police officer. Officers may have heightened levels of fear for their safety in those areas. Still, common sense dictates that not all actions by Black citizens will be interpreted as threatening, and thus even those officers who possess racial stereotypes will not react punitively during most interactions with Black citizens that occur routinely during their shift.

However, following the murder of a colleague, officers are emotionally on edge and have a heightened sense of their vulnerability and mortality. Even those officers who do not possess racial stereotypes about Blacks will have heightened concern for their safety.

During a routine 911 call for service in a largely Black area, say for a generic family trouble, officers may quickly respond physically to a Black subject’s contentious behavior. A raised voice and emotional outbursts towards another family member (i.e., actions that do not directly threaten the officer), would likely have been interpreted before the murder because of frustration or stress due to life circumstances. Before the murder, officers would likely have tried to verbally diffuse the situation. After the murder

105 of a colleague, however, contentious behavior may be more likely to be interpreted as threatening and the transition from a verbal to a physical response may happen much more quickly.

Simply put, behavior in largely Black areas that went without consequences before the murder may now be viewed through a racially biased lens as challenging and threatening, requiring a punitive response. As Legewie observes, “events can increase perceptions of threat and trigger intergroup conflict[,]” which “highlights the temporal embeddedness of police-citizen interactions” (2016, p. 387).

F. Summary and Research Hypothesis

The null hypothesis applicable to the above discussion is that the use of force does not change following the murder of a colleague. Theoretical explanations for the null hypothesis are discussed in the next section. This chapter posits three theories that could explain an alternate hypothesis, which is that officers might respond to the violent murder of a colleague with higher levels of coercion, particularly against members of the perpetrator’s race. Collectively, these theories generate the following research hypothesis:

Hypothesis 4b: Compared to the 21 months prior to Officer Pierson’s shooting, the frequency and severity of force used by RPD officers increased in the days immediately after Officer Pierson’s shooting on September 3, 2014.

G. Discussion of the Null Hypothesis for Research Questions 4a and 4b.

Michael Brown

The fatal shooting of Michael Brown by a white police officer generated an unprecedented wave of negative publicity and anti-police backlash. Research suggests that negative publicity caused some officers, particularly those in agencies that did not

106 have a strong culture of organizational justice, to feel less confident in their authority.

Their morale was also negatively impacted. These developments were theorized to result in de-policing. The empirical research testing whether de-policing occurred in the wake of Ferguson is mixed, however. While one study finds a decrease in traffic/vehicle stops after Ferguson, particularly in areas with higher percentages of Black residents

(Shjarback et al. 2017), another study finds no effect of Ferguson on Part 1 crime rates, with the exception of robbery and homicide (Pyrooz et al. 2016). Another study finds no

Ferguson effect on the use of lethal force by the police (Campbell et al. 2018).

How did Michael Brown’s death impact use of force by the RPD? The null hypothesis is that there was no change in the use of force by RPD officers following

Michael Brown’s death. No change in the use of force could mean that officers remained confident in their authority and were not demoralized even during nationwide negative publicity and anti-police backlash. If self-legitimacy was not impacted, it could be because RPD supervisors created a supportive culture of organizational justice, which research shows can insulate officers from the negative effects of anti-police rhetoric.

Consequently, officers’ behavior was not impacted by Ferguson and there was no resulting effect on the decision to use force when involved in a contentious confrontation.

If use of force increased following Michael Brown’s death, it is possibly because officers remained confident in their authority but now had to respond to greater displays of citizen defiance. The contentious relationship that some residents of disadvantaged neighborhoods have with police were likely further strained following the shootings of

Brown and other unarmed Black males. As research discussed above indicates, Black citizens perceived Michael Brown’s shooting as unjust and an excessive sanction. Some

107

Black residents, when interacting with RPD officers post-Ferguson, may have felt emboldened to defy the law and its agents because of perceived unfairness and bias. As

Sherman observes, “[s]anctions provoke future defiance of the law (persistence, more frequent or more serious violations) to the extent offenders experience sanctioning conduct as illegitimate . . .” (1993, p. 448). Thus, RPD officers would have to respond to defiance of the law with more frequent uses of force and/or more severe force.

Daryl Pierson

The null hypothesis is that there was no change in the use of force by RPD officers following Officer Pierson’s death. If use of force was unchanged, it is likely because officers are already keenly aware of the potential dangers they face at work and can perform their duties regardless. Officer Pierson’s death, while tragic and sudden, was not an entirely unanticipated event, as officers know they or their colleagues could be assaulted or killed on any day they report to work. Consequently, officer behavior may not have changed following Officer Pierson’s death.

If use of force decreased, it may because of de-policing due to fear. While officers are aware of the dangers they face when on patrol, particularly in violent neighborhoods, those dangers are no longer hypothetical but starkly real after a colleague is seriously assaulted or murdered. The vicarious experience of a serious or fatal assault on a colleague could motivate officers to fear for their own safety. Consequently, they may choose to disengage from proactive encounters with citizens or allow conduct that may have resulted in an arrest (and potential use of force) prior to the event to go unchallenged after the event. Officers are human and when a colleague in the same agency is killed it is a human reaction to feel that “what happened to Daryl could happen

108 to me.”8 Fear can be a great motivator, and, in this case, it could motivate officers to “go on defense” and protect themselves by minimizing contact with citizens.

It is also possible that the effects of Ferguson and the other high profile in custody deaths of unarmed Black males cannot be disentangled from the effects of a colleague’s death in the line of duty, in that all could contribute to de-policing. This may be particularly true here as the shootings of Brown and Pierson occurred within 25 days of each other. Negative publicity coupled with the fatal shooting of an officer could motivate officers to retreat from proactive police work to avoid media attention and allegations of excessive force, as well as to protect themselves from assault, particularly when facing greater citizen defiance. Like the citizens who see no benefit from reporting crime in neighborhoods with higher levels of legal cynicism, and thus choose to allow offenses to go unreported (see Kirk & Matsuda, 2011), officers may see no benefit and only harmful consequences to their careers and reputations from proactively engaging citizens. Consequently, the joint effects of Ferguson and Officer Pierson’s death could contribute to de-policing and lower uses of force.

109

Chapter 5: Data, Methods, and Results

Study Objectives

Collectively, the theory guiding this dissertation and the empirical literature testing these theoretical arguments suggest that police use of force is influenced not only by encounter-level variables, but also by the neighborhood characteristics where the encounter occurs and by the timing of the encounter. This study contributes to the extant literature on use of force by examining two overarching research questions.

First, building on the work of Smith (1986) and Terrill and Reisig (2003), among others, the first research question examines whether the odds of being subjected to higher levels of police force, after controlling for encounter-level variables, is influenced by the nature of the neighborhood in which the incident occurs. Two neighborhood characteristics are of interest: race and degree of disadvantage. Race is measured by the census tract level percentages of Black and Hispanic residents, respectively. Degree of disadvantage is measured by an index of concentrated disadvantage, with additional controls for percent on public assistance and percent unemployed, also at the census tract level. The average Part 1 crime rate (explained below) at the census tract level is incorporated as a control variable in each model. In addition to examining whether these characteristics help us understand the use of severe force, I also test whether they are associated with the number of times force is used by officers.

The second research question builds on the work of Legewie (2016) and examines whether changes in (1) the level of force or (2) the number of force incidents occur after an antecedent event. The antecedent events in this case are the lethal shootings of

Michael Brown in Ferguson, MO and of Officer Daryl Pierson in Rochester, NY. The

110 dissertation thus fills a gap in the research by combining not only racial and socioeconomic contextual explanations for use of force, but also the effect of time.

Collectively, then, the dissertation examines the extent to which the risk of being subjected to police force are influenced by where and when a citizen encounters an officer. More broadly, the dissertation seeks to understand how social context influences legal decision-making by police officers.

Research Site

Rochester, New York, is the third largest city in New York State. As of July 1, 2018, the U.S. Census Bureau estimates Rochester’s population as 206,284, which represents a

2.1% decrease from the 2010 census population of 210,565. As of July 1, 2018, the

Census Bureau estimates Rochester’s population to be comprised of Non-Hispanic whites

(36.8%), Non-Hispanic Blacks (40.7%), Hispanics (17.8%), and other races (4.7%). In

2013, FBI statistics show that Rochester’s murder rate per capita of 19.9 per 100,000 residents exceeded that of both New York City (3.9) and the national rate (4.5) (FBI

Uniform Crime Reports, 2013). The U.S. Census Bureau estimates that Rochester’s median income was $30,875 in 2013, with 32.9% of Rochester residents living below the poverty level (U.S. Census, 2013). Hence, Rochester can fairly be described as a comparatively violent city with high levels of urban poverty. As such, Rochester, while smaller in population, is comparable in the salient factors affecting policing to locations where other use of force studies have been conducted, such as Philadelphia, Chicago, St.

Louis, and New York. Rochester, with a population of roughly 206,000, is also comparable to cities in national studies with populations greater than 150,000 (see, e.g.,

Holmes, 2000; Smith & Holmes, 2003).

111

Data: The SRR

Unlike observational data such as that collected by the Project on Policing

Neighborhoods, which is based on observations made by independent research assistants and includes police-citizen encounters that did not result in a use of force, the data for this study contain only encounters that resulted in a use of force. Thus, the study does not seek to evaluate why some encounters end with some degree of coercion while others do not. Rather, use of force is required for the incident to be in the data, as has been the case

(to my knowledge) with other studies analyzing officer self-reports of force (see, e.g.,

Lawton, 2007; Lee et al. 2014; Lersch et al. 2008). Each incident in this study thus involves behavior by the suspect that required (or was interpreted to require) physical restraint by an officer.

As noted, an observational study like the Project on Policing Neighborhoods has the benefit, which this study does not, of including police – citizen interactions that did not result in a use of force. Understanding the interactive dynamics that lead or do not lead to a use of force during a confrontation would provide, in general, a more complete picture of racial bias in policing, and, in the use of force. Do confrontations with white citizens end more often with no force compared to confrontations with Black citizens? Do confrontations in less disadvantaged neighborhoods or more racially mixed neighborhoods end more often with no force compared to poorer, racially segregated neighborhoods? I am unable to address these questions due to the inherent limitations of self-report data.

However, it is arguable that a more significant measure of racial bias in the use of force is the magnitude of force used. When using force, are officers more

112 coercive/punitive in non-white neighborhoods? The data here are well suited to address magnitude through the examination of how neighborhood racial composition influences the severity of force, which is one objective of this study. Thus, in assessing whether more severe force occurs in neighborhoods with larger percentages of Black and Hispanic residents, the study makes an important contribution to our understanding of one form of racial bias in the use of force.

The data for this study are contained in 3,368 Subject Resistance Reports (“SRRs”) completed by RPD officers between October 9, 2011 and June 30, 2016. The RPD has approximately 730 – 750 sworn officers at any given time (also known as members), and

RPD G.O 335 requires that an SRR must be completed by RPD members who use any force, whether on or off duty. G.O. 335 defines “force” as “any intentional physical strength or energy exerted or brought to bear upon or against a person for the purpose of compulsion, constraint or restraint.” An SRR does not have to be completed for mere handcuffing, blanketing (i.e., surrounding a subject), escorting, or application of a hobble or spit sock.9

For most analyses in this study, the unit of analysis is the use of force as documented on an SRR. An SRR is generated for each subject of a use of force under a crime report number unique to that subject. If more than one officer used force on the same subject, each is required to complete an SRR under that subject’s crime report number. During data coding (discussed below), data from multiple SRRs for the same subject were collapsed under the crime report number to the same row on an Excel spreadsheet. When coding was completed, the spreadsheet contained 3,368 unique rows/uses of force. Thus,

113 each of the 3,368 incidents, regardless of how many officers were involved or completed

SRRs, counts as only one use of force

The SRR captures levels of subject resistance, the type(s) of force used against that resistance, and whether each force technique was ineffective, moderately effective, or effective. A moderately effective technique enabled the officer to gain some, but not total, control over the subject, who continued to resist being taken into custody. An effective technique enabled the officer to take the subject into custody with no further resistance. The SRR also includes a written narrative from each officer who participated in the encounter that more fully describes the subject’s demeanor and actions, as well as the actions of the officer.

SRRs are stored under their crime report number in the computer system of the

Professional Standards Section (“PSS”), generically known as “internal affairs.” The computer system is password protected and access must be granted by the Chief’s Office.

I received permission from RPD Chief Michael Ciminelli to collect SRRs for research purposes and was granted access to the PSS computer system. Beginning in the late

Autumn of 2015 and ending in August 2016, I downloaded the SRRs used in this study and placed them on a secure drive. As noted above, the SRRs document use of force incidents that occurred from October 9, 2011 through June 30, 2016.

I created a code book comprised of variables taken directly from the SRR. These include: incident information, which includes the crime report number assigned to the incident, the car beat10 where the incident occurred, and the date, time, and location of the incident; subject demographics, which include race, sex, age, and degree of impairment; officer demographics, which include race, sex, and number of officers involved in the use

114 of force; situational variables, which include the types of subject resistance and the highest level of resistance encountered during the incident; and the types of force used during the incident, which includes the highest level of force used by any officer. Guided by the codebook, I coded variables from each SRR on an Excel spreadsheet. Excluded from the spreadsheet were “crowd control” SRRs documenting the use of pepper spray to disperse an unruly crowd, where no subject information was captured, and no arrest was made. Once coding was completed at the end of August 2019, the Excel spreadsheet containing all coded encounter-level data for 3,368 uses of force was imported into the

Stata 16 statistical program.

The negative binomial count analysis is based on this full data set of 3,368 uses of force, which are nested within 78 census tracts. However, the logistic regression analysis is based on 3,352 uses of force due to low cell counts for subject race and officer race. In logistic regression, low cell counts are problematic because they can cause very small or very large odds ratios. Low cell counts are revealed when a cross tabulation of the dependent variable and a predictor variable shows a cell with less than five cases. Here, a cross tabulation of the dependent variable (high v. low force) and subject race and officer race revealed less than 5 incidents for the high force category. There was only one “high” use of force against Asian subjects, and three uses of high force by Asian or Native

American officers. Accordingly, these 16 uses of force were eliminated from the logistic regression analysis, resulting in an N=3,352.

Research Questions 1a through 3b: Variables for Race and Disadvantage

All tables and figures depicting data and models are presented at the end of this chapter.

115

Analytic Strategy:

The first model is a negative binomial regression model with robust standard errors to predict whether use of force counts at the census tract level are influenced by the neighborhood characteristics where the use of force occurred (Research Hypotheses 1a,

2a, and 3a). As King and Sutton (2013, p. 881) observe, the negative binomial estimator

“is appropriate when analyzing non-negative integers in which overdispersion is present.”

The likelihood ratio generated by the comparison of the negative binomial model to a

Poisson model is significant at p<.001, which indicates that overdispersion is present in counts of police use of force, and thus negative binomial regression is appropriate.

As noted above, the count analysis is based on 3,368 uses of force and include only level-2 variables as the predictors of interest. The unit of analysis is thus the census tract, within which the 3,368 uses of force are nested. Thus, Research Hypothesis 1a incorporates only neighborhood racial variables and average Part 1 crime rates (explained below in the discussion of level-2 variables), 2a incorporates neighborhood disadvantage variables and crime rates, and 3a incorporates all neighborhood-level variables. Each model will also control for census tract population differences and yearly fluctuations in use of force.

The second model used is multilevel logistic regression, which is an appropriate analytical model when the dependent variable is a binary (and not a count) variable. Each use of force in the data set is nested within a census tract. Uses of force within census tracts are likely to share similarities with respect to their situational characteristics, and citizen and officer demographics, and thus cannot be readily treated as independent. A

116 simple regression model is not appropriate given the nested structure of the data and the likelihood of correlated errors. The structure of the data requires the estimation of multilevel models that account for the nonindependence of incidents within neighborhoods. The analysis is thus based on multilevel logistic regression models with random intercepts that include encounter-level (level-1) variables as well as the neighborhood (level-2) variables analyzed in the negative binomial models. After controlling for encounter-level variables, the objective is to test whether the odds of being subjected to high force, relative to low force, are influenced by the characteristics of the neighborhood where the use of force occurs.

For the logistic regression analysis, two models will be estimated to test research hypothesis 1b: one with controls for population and year and encounter-level variables, and a second model that incorporates neighborhood racial composition. Research hypothesis 2b will be tested with a model that controls for population and year, encounter-level variables, and neighborhood disadvantage variables. Research hypothesis

3b will be tested with a full model that includes all neighborhood-level variables.

Dependent Variables:

Negative Binomial Regression: The dependent variable for negative binomial regression analysis is number of force incidents by census tract (3,368 use of force incidents by 78 census tracts).

Logistic Regression: As discussed in Chapter 2, the dependent variable for the logistic regression analysis is a dichotomous measure of the highest level of force used by any officer involved in each incident. The dependent variable is coded 0 = Low Force

117

(verbal/soft, non-weapon force), and 1 = High Force (non-lethal weapon/impact instrument, lethal/deadly force).

Independent Variables:

The study controls for encounter-level variables that are commonly modeled in the use of force literature, broadly representing subject demographics, officer demographics, and situational variables (i.e., time of the incident, level of subject resistance, and the number of officers using force). The following describes the composition of these variables and how they were coded:

Encounter-Level Variables (level-1): Subject and Officer Demographics

Subject Race is coded as 1=White, 2=Black, and 3=Hispanic. Subject Sex is coded as

0=Female and 1=Male. Subject Age is a categorical variable coded 0=age 35 or less and

1=over age 35. I chose age 35 as the demarcation for subject age because approximately

75% of force incidents involved subjects age 35 or younger.11 Subject Age was recoded as a categorical variable due to low cell counts (less than five) for “high” force against subjects over age 50. Subject Degree of Impairment is coded as 1=Not Impaired,

2=Alcohol Influence, and 3=Drug Influence. Officer Race is coded as 1=Black, 2=White, and 3=Hispanic. Officer Sex is coded as 0=Female and 1=Male.

Encounter-Level Variables: Situational Controls

Subject Level of Resistance is coded as discussed in Chapter 2: 0 = Low Resistance

(verbal/passive, avoiding custody), and 1 = High Resistance (aggressive/assaultive behavior, deadly force). Time of the Incident is coded as 1=8 AM to 3:59 PM (reference category), 2=4 PM to 11:59 PM, and 3=12 AM to 7:59 AM. Number of Officers is coded as 0=one officer and 1=more than one officer.

118

Note that the periods for the time of the incident variable correspond to the shifts officers work and are not based on a day versus night dichotomy. The afternoon shift (4

PM to 11:59 PM) is typically significantly busier than the overnight shift and therefore has more officers assigned. Thus, a simple “day” versus “night” dichotomy to model time would not accurately reflect the working conditions officers encounter, including their likelihood of being involved in a use of force. Indeed, exactly 50% of uses of force occurred during the afternoon shift compared to 31% overnight.

Control Variables:

Census Tract Population, which is drawn from the Census Bureau’s American

Community Survey 2012–2016 five-year estimates, is included to control for census-tract population variances. As such, it is a level-2 variable. No transformation of this variable was necessary. The five-year estimates from 2012 - 2016 best match the temporal pattern of the data, as 88.4% of the observations occurred in the full years 2012 – 2015. Only

4.2% occurred in the last two months of 2011, while 7.4% occurred in the first six months of 2016. For the negative binomial model, it is possible that more populous census tracts will have higher numbers of force incidents. Accordingly, as recommended for negative binomial regression, an exposure term is added to census tract population to model the effect of population size on uses of force. Year is a continuous variable included to control for year to year variations in use of force.

Neighborhood Variables (level-2):

Neighborhood structural variables are also drawn from the Census Bureau’s

American Community Survey 2012-2016 five-year estimates. Consistent with prior urban research, census tracts are used as proxies for neighborhoods (see, e.g., Peterson, Krivo,

119

& Harris, 2000; Puckett & Lundman, 2003). The census tract is the primary geographic unit for which data (e.g., race and poverty statistics) are published by the U.S. Census

Bureau. As Sperling (2012) observes, census tract data “are highly accessible, relatively easy to use, and available across time and space” (2012, p. 219). The availability of data for neighborhood-level research is the primary advantage of using census tracts as proxies for neighborhoods.

However, it must be acknowledged that defining what constitutes a “neighborhood” is difficult and imprecise. As sociologist Roderick McKenzie (1923) observed in his study of Columbus, Ohio, “[p]robably no other term is used so loosely or with such changing content as the term neighborhood, and very few concepts are more difficult to define”

(1923, pp. 334–335). A concern is that census tracts are artificial representations of neighborhoods that may not reflect actual social realities. In their research on neighborhoods and violent crime in Columbus, Ohio, Peterson, Krivo, and Harris (2000) observe that “[c]ensus tracts do not necessarily correspond to neighborhoods in a socially meaningful sense” (1996, p. 623). Nonetheless, the authors used census tracts for their analysis after concluding that census tracts “are the best local areas for which the required data are available and they have been used in a number of prior analyses of urban crime”

(2000, p. 623).

As applied to use of force, an officer patrolling a census tract where 96.6% of the residents are Black and degree of disadvantage is high will undoubtedly experience different environmental conditions and answer different types of 911 calls (e.g., more violent , robberies, man with a gun, or shots fired) than an officer patrolling in a neighborhood where only 1.9% of the residents are Black and disadvantage is low. Based

120 on the theory discussed in Chapter 3, these environmental conditions may influence how an officer responds during an interaction with a citizen. Consequently, census tracts are arguably the best proxies of the local areas where police patrol in Rochester. To model neighborhood data, the address of each use of force incident, as documented on the SRR, was geocoded and assigned to the census tract encompassing that location for all level-1 data, which was then merged with level-2 data.

Consistent with social disorganization research and the approach of Terrill and Reisig

(2003), the neighborhood structural variables used to test the social disorganization/ ecological contamination research questions are modeled through the creation of an index of concentrated disadvantage, which represents the neighborhood poverty level.

Concentrated Disadvantage is composed of percent living below the poverty line, percent of female headed households, and percent children less than 18 years of age within each census tract. While other variables, such as percent on public assistance and percent unemployed are also included in the concentration disadvantage index in some studies, the three variables comprising the index used here had the highest factor loadings

(ranging from .75 to .97) and an eigenvalue of 2.36. As an index, these three variables explained the highest percentage of total variance at 79% (see Table 2). Percent on public assistance and percent unemployed are included in the social disorganization models and the combined model as separate neighborhood-level controls.

(Table 2 about here)

For the research questions testing racial threat/Black’s morphology, Percent Black and Percent Hispanic were drawn from the American Community Survey 2012-2016 five-year estimates to model racial composition of each census tract (see Table 3). The

121 average percentage of Black residents in a census tract was 45.8, with a range from 1.9 to

96.6. The average percentage of Hispanic residents in a census tract was 20.9, with a range from 1.5 to 51.7.

(Table 3 about here)

To control for neighborhood crime rates in all models, census tract Average Part I

Crime Rate from 2012-2016 per 1,000 residents was created with data from the Rochester

Police Department’s Open Data Portal Public Website, which was then merged with the

78 census tracts included in the RPD use of force data set. The variable was then transformed by 1/sqrt. Part 1 crimes are offenses known to law enforcement and collected by the FBI as part of its Uniform Crime Reporting Program. Part 1 crimes include violent offenses such as criminal homicide, forcible rape, robbery, and aggravated assault, and property offenses such as burglary, larceny-theft, and motor vehicle theft.12

Descriptive Statistics

Table 4 displays descriptive statistics. The majority (76.1%) of incidents involved low force. Subject resistance in most incidents (72.6%) also occurred at a low level. Most force incidents (50.1%) occurred between 4:00 p.m. and 11:59 p.m. and involved primarily Black subjects (66.8%) and males (76.9%). Most subjects involved in uses of force (54.7%) were not impaired by either alcohol or drugs. Regarding officer characteristics, most use of force incidents involved white males and more than one officer. However, as discussed above, a use of force involving more than one officer still counts as only one use of force for the analysis. Although not reflected in Table 4, the average number of force incidents across the 78 census tracts for the logistic regression analysis is 43, with a minimum of 4 and a maximum of 316. The majority of uses of

122 force, 316 or 9.4%, occurred in downtown Rochester, within the “inner loop” highway that circles the city. Many of these uses of force occurred at or near RPD headquarters, the downtown sports arena, or the Monroe County Jail during prisoner transports.

Approximately 80% of these incidents involved low force. The second largest number of force incidents, 175 or 5.2%, occurred in the East Avenue bar district just outside the inner loop. Approximately 75% of these incidents involved low force.

(Table 4 about here)

As depicted in Figure 1, the age of subjects is positively skewed. The average age is

29.4 years, with a median of 26 years and a range from age 9 to age 75.

(Figure 1 about here)

Results

Modeling results are displayed in the following section. The theoretical implications of these results will be discussed in Chapter 6.

A. Research Hypotheses 1a, 2a and 3a

Table 5 presents the results of three negative binomial models estimated to test

Research Hypotheses 1a, 2a, and 3a. Research Hypothesis 1a predicts that the number of uses of force will be influenced by neighborhood racial composition. Research

Hypothesis 2a predicts that number of uses of force will be influenced by neighborhood degree of disadvantage. Research Hypothesis 3a predicts that, after controlling for neighborhood degree of disadvantage and average Part 1 crime rate, the number of uses of force in a combined model will be influenced only by neighborhood racial composition. Each model contains an exposure term for population to control for the effect of census tract population on force counts.

123

(Table 5 about here)

Model 1 indicates that as the percentage of Black and Hispanic residents increases, uses of force significantly increase by 10.9% and 3.7%, respectively. Uses of force increase significantly, if modestly (by .2%) as average Part 1 crime rate increases (a result seen in all count models). Research Hypothesis 1a is supported by these findings.

In Model 2, uses of force increase significantly in neighborhoods with higher levels of concentrated disadvantage and percent on public assistance. Research Hypothesis 1b is supported by these findings. However, an unexpected finding is that uses of force decrease significantly by 25% in neighborhoods with higher levels of unemployed residents. Being on public assistance, being unemployed, and living below the poverty line (a component of the concentrated disadvantage index) are strongly correlated.

Disadvantaged neighborhoods should therefore have higher concentrations of all three characteristics. Based on the theory discussed in Chapter 3, officers should perceive poor neighborhoods with these characteristics as “bad” places populated with suspicious people. Accordingly, as Research Hypothesis 2a predicts, it is reasonable to expect that numbers of force will be higher in such neighborhoods.

The findings, however, indicate that neighborhoods with higher numbers of unemployed persons have lower numbers of force. A possible explanation is that being unemployed is not necessarily equivalent to being on public assistance or living below the poverty line. Not all unemployed persons receive public assistance or live in disadvantaged neighborhoods, typically because they live with and are supported by a relative or other person (a spouse, parent, domestic partner etc.) who has a self- supporting income. The findings in Model 2 suggest that neighborhoods with higher

124 percentages of unemployed persons do not always present the same visual cues (e.g., indicators of poverty or disorder) to officers as do highly disadvantaged neighborhoods.

Consequently, officers may perceive them to be less threatening and are less likely to become involved in a use force there.

In Model 3, the combined model, concentrated disadvantage is no longer significant as a predictor of force counts, while all other variables retain their levels of significance.

As predicted by Research Hypothesis 1c, after controlling for degree of disadvantage, uses of force are significantly higher in neighborhoods with greater percentages of Black residents (by 22.8%) and Hispanic residents (by 10.8%).

B. Research Hypotheses 1b, 2b, and 3b

Table 6 presents the results of two multilevel logistic regression models (Models 4 and 5) estimated to test Research Hypotheses 1b, which predicts that the odds of being subjected to high force relative to low force are influenced by the racial and ethnic composition of the neighborhood in which the incident occurs. Models 4 and 5 incorporate controls for population and year, and encounter-level variables.

(Table 6 about here)

At the encounter-level, the odds of being subjected to high force compared to low force are unrelated to subject race but increase significantly, on average, for males and those subjects under the influence of drugs. Not surprisingly, subjects who resist at a high level are significantly more likely, on average, to be subjected to high force. The odds of being subjected to high force also increase significantly, on average, during the

“graveyard” shift of midnight to 7:59 a.m., indicating that officers may view subjects who are out in public at a time when most people are sleeping as potentially more

125 dangerous or deviant. Subjects who are involved in a use of force with more than one officer are 2.3 times more likely, on average, to be subjected to high force. Surprisingly, and contrary to the research at the encounter-level discussed in Chapter 2, the odds of being subjected to high force increase significantly for subjects over age 35 relative to those age 35 or younger.

Model 5 incorporates neighborhood racial composition. All encounter-level variables remain at the same levels of significance. At the neighborhood-level, average part 1 crime rate is not significant. However, as predicted by Hypothesis 1b, the odds of being subjected to high force, relative to low force, increase by 11.5%, on average, as the neighborhood percentage of Black residents increase (p<.05) and by 13.8%, on average, as the percentage of Hispanic residents increase (p<.001). Since race at the encounter- level is nonsignificant, officers do not appear to be influenced by the race of the subject per se when using force, but by the racial composition of the neighborhood in which the encounter occurs. As discussed below, I test the possibility that degree of force is related to an interaction between race, age, and sex, such that the “young Black male” is subjected to higher levels of force at the encounter-level than other individuals.

Table 7 presents the results of Model 6, which is a multilevel logistic regression model estimated to test Research Hypothesis 2b, and Model 7, which is a multilevel logistic regression model estimated to test Research Hypothesis 3b. Research Hypothesis 2b predicts that the odds of being subjected to high force relative to low force are influenced by the level of disadvantage and crime rates within the neighborhood in which the incident occurs. Model 6 incorporates controls for population and year, encounter-level variables, and neighborhood-level disadvantage variables. As in Models 4 and 5, race of

126 the subject at the encounter-level is not a significant predictor of the odds of being subjected to high force. The significant finding in Model 6 is that the odds of being subjected to high force, relative to low force, increase by 16.9%, on average, as the level of neighborhood concentrated disadvantage increases. Average Part 1 crime rates, however, are not a significant predictor of the severity of force. Research Hypothesis 2b is thus partially supported by these findings.

(Table 7 about here)

The second model in Table 7, Model 7, is a multilevel logistic regression model estimated to test Research Hypotheses 3b. Research Hypothesis 3b predicts that the odds of being subjected to high force relative to low force, after controlling for neighborhood disadvantage and average Part 1 crime rates, are influenced by the racial composition of the neighborhood in which the incident occurs. Model 7 incorporates controls for population and year, encounter-level variables, and all neighborhood-level variables.

Again, race at the encounter-level is not a significant predictor of the odds of being subjected to high force. The significant finding in Model 7 is that the odds of being subjected to high force, relative to low force, increase by 17.1%, on average, as neighborhood percentage of Black residents increases and by 17.7%, on average, as neighborhood percentage of Hispanic residents increases. As in the count analysis, once all neighborhood variables are modeled, concentrated disadvantage becomes nonsignificant as a predictor while racial composition retains its significance on the odds of being subjected to high relative to low force. Research Hypothesis 3b is thus supported by this finding.

127

Interactions

Literature discussed in Chapter 3 suggests that many Americans, including police officers and other criminal justice actors, perceive Black individuals to be violent and prone to criminality. This same body of research suggests that the “young Black male” is perceived to be especially “dangerous” and hence a strong target of this form of racial bias. Thus, while race alone at the encounter-level is nonsignificant, it may be because race is a salient predictor of force only when it interacts with age and sex.

To model this interaction, I created a categorical variable operationalizing the combination of being young, Black, and male. Unlike race and sex, which typically are easily discerned, who officers consider to be “young” is a subjective determination.

Accordingly, I created three variables, each comprised of “Black” and “male” subjects but with different age ranges to represent the “young” category: one with age 29 (the average age of all subjects) or younger, one with age 25 or younger, and one with age 21 or younger. It is likely that there is no substantive difference in how officers react to male subjects in their late teens or early twenties, and thus age 21 or younger was chosen to represent the “young” category in the last variable.

To capture the effect of the interaction on severity of force, I estimated three models, each with a different “young Black male” variable. Each model included all encounter- level variables (other than race, sex, or age), all neighborhood-level variables, and the control variables for population and year. The “young Black male” variable was nonsignificant in each model. Accordingly, there was no association between severity of force and the subject’s race at the encounter-level, whether alone or in an interaction with age and sex.

128

Summary

The results of the negative binomial regression and logistic regression models suggest that officers are indeed conscious of the environmental conditions in which they work and, consistent with the theory discussed in Chapter 3, act accordingly. When modeled separately, degree of disadvantage and percentage of Black and Hispanic residents each significantly influence both numbers of force and severity of force. However, as predicted by Research Hypotheses 3a and 3b, when neighborhood racial composition and degree of economic disadvantage are combined in a full model, officers use force more often and at more severe levels only when interacting with citizens in neighborhoods with higher percentages of Black and Hispanic residents. The percent of residents on public assistance has a positive influence on numbers of force, but not severity of force. Thus, it appears that officers decide to go “hands on” with a citizen more frequently in neighborhoods with higher percentages of Black and Hispanic residents and use more severe levels of force in those neighborhoods.

Research Questions 4a and 4b: Analytic Strategy and Variables for Time

Interrupted time-series analysis (“ITSA”) is an appropriate method for assessing whether an antecedent event, in this case Michael Brown’s or Officer Pierson’s shootings, caused a change in the trajectory of force by the RPD following those incidents. Research by Legewie (2013, 2016) and King and Sutton (2013) suggests that if there is a change in use of force, it will only be temporary and will be followed by an eventual return to pre-event levels, a pattern which is known as a “pulse function.”

129

Data Set

For the interrupted time series analysis, the data set was restricted to the years 2013 to

2015. The antecedent incidents, which took place in August 2014 and September 2014, respectively, are roughly mid-way through this period. Thus, the beginning of 2013 to

August 2014 captures the immediate pre-event period and September 2014 to the end of

2015 captures the immediate post-event period. The data set thus has 156 weeks (52 x 3), with 83 weeks before Michael Brown’s death, 4 weeks from Brown’s death until Daryl

Pierson’s death, and 69 weeks after Pierson’s death. This period encompasses 61.1% of the total observations in the full data set and provides a sufficient period for the time series analysis. This is the same approach taken by Campbell and colleagues (2018), who examined whether the number of fatal shootings by the police changed following Michael

Brown’s death. Their data encompassed the period from May 4, 2013 to December 31,

2015, which had 138-weeks (66 before the shooting and 72 after).

Analytic Strategy

A concern in ITSA modeling is correlation between values of a time series and prior values of the same series, a condition known as serial correlation (King & Sutton, 2013, p. 881). To test for serial correlation, a Durbin-Watson test was run for each model.

Results indicated that there was no serial correlation in the time series. A second concern is seasonality, which refers to cyclical behavior in a time series. To control for daily and seasonal fluctuations in force (e.g., more uses of force on weekends, on holidays such as

New Year’s Eve, St. Patrick’s Day, or July 4th, or during the summer months), day of the week and season are added to each model. Season is coded 1 for Winter (January through

130

March), 2 for Spring (April through June), 3 for Summer (June through September), and

4 for Autumn (October through December).

Two ITSA models are estimated, each containing a constant term, a four-category dummy “period” variable (described below), and controls for day of the week and season.

I first estimate a negative binomial model with robust standard errors to test whether the daily number of uses of force changed after Brown’s and Pierson’s shootings. To test whether the odds of being subjected to high force, relative to low force, changed following the shootings of Michael Brown and Officer Pierson, I then estimate a multilevel logistic regression model.

Dependent Variables:

There were 2,058 uses of force in 2013 – 2015. The negative binomial regression dependent variable is a daily count of numbers of force for that period, which was created by collapsing the total 2,058 uses of force by the day they occurred. This results in 1,095

“count days” for the analysis. Daily force counts ranged from 213 days with 0 uses of force to one day having 10 uses of force (mean = 1.88 per day).

As in the analyses of Research Hypotheses 1b, 2b and 3b, the logistic regression dependent variable is operationalized using the same dichotomous categorizations (i.e.,

1=High Force, 0=Low Force). The logistic regression analysis is based on all 2,058 uses of force for 2013 – 2015. Because both interrupted time series models analyze the effect on force outcomes of the four-category dummy variable, and not encounter-level variables, it was not necessary to drop uses of force for low cell counts.

131

Focal Independent Variable:

To capture the immediate aftermath of each incident, a four-category dummy variable was constructed to model the period in days before each incident and the days after. The time periods reflect a one-day lag for days following Brown’s shooting, as there was likely no immediate effect on 8/9/14, the day of Brown’s shooting (day 0 in the Period variable).13

“Period” is thus coded as follows:

Time (in days) less than or equal to the day of the Brown shooting (day 0) = 1

Time (in days) greater than or equal to the day after Brown’s shooting (day 1) to

the day of the Pierson Shooting (25 days later) = 2

Time in days from the day after the Pierson shooting to 25 days later = 3

Time following the 25 days after the Pierson shooting = 4

Results

I begin the analysis by examining whether numbers of force changed following

Brown’s and Pierson’s deaths. Table 8 depicts Model 8, which is a negative binomial model with robust standard errors. This model tests the component of Research

Hypothesis 4a that predicts numbers of force will decrease following Michael Brown’s death, and of Research Hypothesis 4b that predicts numbers of force will increase following Officer Pierson’s death.

Model 8 indicates that there was no significant change in expected counts of force in the 25-day period after Brown’s death relative to the period before his death. However, in the 25-day period after Pierson’s death, relative to the period prior to Brown’s death,

132 expected force counts increased significantly, by 31.9 %. Thereafter, expected force counts decreased by 12.6%, on average, through the end of 2015.

(Table 8 about here)

To examine this pattern in more detail, Model 10 was estimated using only 2014 data and is depicted in Table 9. Restricting the data set to 2014 generated 365 daily counts of force. The results indicate that, again, there was no significant change in expected force counts following Brown’s death compared to the period before his death. As in the full model, expected force counts again increased significantly, this time by 58.3%, on average, in the 25-day period after Pierson’s death. Thereafter, uses of force decreased but were not significantly different compared to the period prior to Brown’s death.

(Table 9 about here)

Figure 2 illustrates the 2014 pattern. The diagram depicts 5-day smoothed average force counts on the y-axis and days since Daryl Pierson’s death on the x-axis. Note that the first vertical red line represents “0,” the day of Pierson’s death, and the second vertical red line represents the end of 25-day period following his death. The pattern between the red lines shows the apparent spike in uses of force in the 25-day period after

Pierson’s death, followed by a gradual decrease. Officers thus appear to have ramped up uses of force following the death of their colleague. Decreases from the end of the 25-day period following Pierson’s death were not significant (compared to the period prior to

Brown’s death) for the remainder of 2014 but became significant through the end of

2015. This suggests that officers gradually transitioned to a period of “de-policing” over the longer term, perhaps in response to negative publicity generated by other lethal shootings of unarmed Black males nationally (discussed further in Chapter 6).

133

(Figure 2 about here)

Transitioning to interrupted time series analysis of changes in the severity of force,

Table 8 also depicts Model 9. Model 9 is a multilevel logistic regression model testing the component of Research Hypotheses 4a that predicts severity of force will decrease following Michael Brown’s death and of Research Hypothesis 4b that predicts severity of force will increase following Officer Pierson’s death.

Contrary to expectations, results indicate that during the 25-day period following

Brown’s death until Pierson’s death, relative to the period prior to Brown’s death, the odds of being subjected to higher levels of force increased by 2.0 times, on average

(p<.01). While officers were no more likely to use force during this period compared to before Brown’s death, officers were significantly more likely to be punitive when they did use force, perhaps in response to a greater degree of “citizen defiance.”

There was no significant change in severity of force during the 25-day period following Pierson’s death relative to the period prior to Brown’s death. However, during the period more than 25 days following Pierson’s death through the end of 2015, the odds of being subjected to higher levels of force significantly increased by 46.4%, on average.14

Summary

The combined count and logistic analyses indicate that officers did not react to

Brown’s death by retreating from or increasing force responses relative to the period before his death. However, when they did use force, they engaged in a significantly more punitive response. With respect to Pierson’s death, officers immediately began using force more frequently during citizen encounters. However, officers did not resort to

134 higher levels of force. Thereafter, the opposite occurred. Officers appear to have avoided using force whenever possible but became more punitive when they did become involved in a use of force. In the next chapter, I discuss the important question of why?

135

Chapter 5 Tables

Table 2: Factor Loadings

Variable First Factor Loading Below poverty line .75 Female headed households .97 Percent children less than 18 .92

Table 3: Neighborhood Racial Composition (N=3,368)

Variables Mean Std. Dev. Minimum Maximum

Percent Black 45.8 18.8 1.9 96.6

Percent Hispanic 20.9 13.3 1.5 51.7

136

Table 4: Descriptive Statistics (N=3,368)

Dependent Variable Frequency Percent Level of Force: Low 2,564 76.1 High 804 23.9

Independent Variables Categorical Variables: Incident Time: 8 AM – 3:59 PM 623 18.5 4 PM – 11:59 PM 1,686 50.1 12 AM – 7:59 AM 1,059 31.4 Subject Race (N=3,360) White 728 22.7 Black 2,243 66.8 Hispanic 389 11.5 Subject Sex: Female 777 23.1 Male 2,591 76.9 Subject Age: Age 35 and under 2,516 74.7 Over Age 35 852 25.3 Degree of Impairment: Not Impaired 1,842 54.7 Alcohol Influence 1,270 37.7 Drug Influence 256 7.6 Subject Resistance: Low 2,445 72.6 High 923 27.4 Officer Race (N=3,360) Black 380 11.3 White 2,729 81.2 Hispanic 251 7.5 Officer Sex: Female 237 7.0 Male 3,131 93.0 Number of Officers: One 1,471 43.7 More Than One 1,897 56.3

Age as a Continuous Variable: Mean Std Dev Median Min Max

Subject Age 29.4 11.4 26 9 75

137

Table 5: Negative Binomial Regression Predicting Census Tract Use of Force Counts (N=78 Census Tracts)15 Research Question 1a (Model 1) Research Question 2a (Model 2) Research Question 3a (Model 3) (Standard Errors in Parentheses)

Model 1 Model 2 Model 3

B (SE) IRR B (SE) IRR B (SE) IRR Independent Variables

Population Exposure Exposure Exposure Year -.012 (.009) .988 -.005 (.008) .995 -.004 (.008) .996

Percent Black .104 (.011) 1.109*** .205 (.020) 1.228*** Percent Hispanic .037 (.010) 1.037*** .103 (.019) 1.108*** Avg. Part 1 Crime Rate .002 (.000) 1.002*** .005 (.000) 1.005*** .004 (.000) 1.004*** Concentrated Disadvantage .163 (.024) 1.177*** -.074 (.039) .928 Percent Public Assistance .099 (.022) 1.104*** .125 (.022) 1.133*** Percent Unemployed -.289 (.013) .749*** -.292 (.014) .747***

Constant .938 (.120)*** .637 (.114)*** .677 (.117)*** Model Chi-square 147.14*** 507.70*** 470.72***

*p<.05 **p<.01 ***p<.001

138

Table 6: Multilevel Logistic Regression Predicting High versus Low Force (N=3,352 Use of Force Incidents), Research Question 1b (Standard Errors in Parentheses)

Model 4 Model 5 Independent Variables B (SE) OR B (SE) OR Level 1: Population .000 (000) 1.000 .000 (.000) 1.000 Year .091 (.031) 1.095** .091 (.032) 1.095**

Subject Race: Ref: White Black .049 (.109) 1.050 -.064 (.115) .938 Hispanic .250 (.153) 1.284 .112 (.158) 1.119 Subject Sex: Ref: Female Male .604 (.114) 1.829*** .614 (.114) 1.848*** Subject Age: Ref: Age 35 or Under Over Age 35 .240 (.097) 1.272* .227 (.097) 1.255* Degree of Impairment: Ref: Not Impaired Alcohol Influence .160 (.101) 1.174 .195 (.101) 1.215 Drug Influence .773 (.150) 2.165*** .774 (.150) 2.167*** Officer Race: Ref: Black White -.104 (.136) .901 -.132 (.137) .876 Hispanic -.121 (.204) .886 -.157 (.205) .854 Officer Sex: Ref: Female Male .326 (.183) 1.385 .291 (.183) 1.338

Time of Incident: Ref: 8am – 3:59pm 4pm – 11:59pm .029 (.123) 1.029 .013 (.124) 1.013 12am – 7:59am .309 (.138) 1.362* .337 (.139) 1.401* Level of Resistance: Ref: Low High .828 (.090) 2.288*** .826 (.090) 2.285*** Number of Officers: Ref: One More Than One .839 (.092) 2.314*** .828 (.092) 2.289***

Level 2: Percent Black .108 (.051) 1.115* Percent Hispanic .129 (.045) 1.138** Avg. Part 1 Crime Rate 1.121 (1.358) 3.068

Constant -4.258 (.508)*** -3.345*** (.604) Model Chi-square 254.58*** 264.80***

*p<.05 **p<.01 ***p<.001

139

Table 7: Multilevel Logistic Regression Predicting High versus Low Force (N=3,352 Use of Force Incidents), Research Questions 2b (Model 6) and 3b (Model 7) (Standard Errors in Parentheses)

Model 6 Model 7 Independent Variables B (SE) OR B (SE) OR Level 1: Population .000 (000) 1.000 .000 (.000) 1.000 Year .091 (.031) 1.095** .093 (.032) 1.097**

Subject Race: Ref: White Black -.004 (.113) .996 -.049 (.116) .952 Hispanic .191 (.156) 1.210 .115 (.158) 1.122 Subject Sex: Ref: Female Male .610 (.114) 1.840*** .611 (.114) 1.842*** Subject Age: Age 35 or Under (Ref) Over Age 35 .236 (.097) 1.266* .231 (.098) 1.259* Degree of Impairment: Ref: Not Impaired Alcohol Influence .178 (.101) 1.195 .190 (.102) 1.209 Drug Influence .780 (.150) 2.180*** .779 (.150) 2.179*** Officer Race: Ref: Black White -.112 (.137) .894 -.122 (.137) .885 Hispanic -.129 (.204) .879 -.156 (.205) .856 Officer Sex: Ref: Female Male .318 (.183) 1.374 .287 (.183) 1.332

Time of Incident: Ref: 8am – 3:59pm 4pm – 11:59pm .018 (.124) 1.018 .018 (.124) 1.018 12am – 7:59am .318 (.139) 1.375* .331 (.139) 1.393* Level of Resistance: Ref: Low High .825 (.090) 2.281*** .827 (.091) 2.287*** Number of Officers: Ref: One More Than One .831 (.092) 2.295*** .830 (.092) 2.294***

Level 2: Concentrated Disadvantage .156 (.068) 1.169* -.023 (.090) .978 Percent Public Assistance -.053 (.065) .949 -.046 (.066) .955 Percent Unemployed -.021 (.064) .979 -.047 (.067) .954 Avg. Part 1 Crime Rate -.011 (1.633) .989 .191 (1.664) 1.210 Percent Black .158 (.068) 1.171* Percent Hispanic .163 (.057) 1.177**

Constant -3.231 (.606)*** -4.381*** Model Chi-square 259.41*** 266.47***

*p<.05 **p<.01 ***p<.001

140

Table 8: Interrupted Time Series Analysis, 2013 – 2015 Negative Binomial Regression, Daily Force Counts, Model 8 (N=1,095 Count Days) (Standard Errors in Parentheses)

Multilevel Logistic Regression, Severity of Force, Model 9 (N=2,058 Use of Force Incidents) (Standard Errors in Parentheses)

Model 8 Model 9 Independent Variables B (SE) IRR B (SE) OR

Period 1 (Reference) Period 2 -.164 (.160) .849 .708 (.347) 2.030* Period 3 .277 (.126) 1.319* .304 (.313) 1.355 Period 4 -.134 (.055) .874* .381 (.113) 1.464**

Sunday (Reference) Monday -.335 (.101) .716** -.241 (.206) .786 Tuesday -.362 (.101) .696*** .305 (.192) 1.356 Wednesday -.191 (.091) .826* -.093 (.192) .911 Thursday -.109 (.089) .897 -.231 (.191) .794 Friday -.032 (.091) .968 -.386 (.193) .608* Saturday .060 (.088) 1.062 .007 (.177) 1.007

Winter (Reference) Spring .134 (.075) 1.144 .122 (.149) 1.130 Summer .114 (.077) 1.120 -.117 (.162) .890 Autumn .010 (.077) 1.010 -.016 (.159) .984

Constant .741 (.083)*** -1.237 (.171)*** Chi Square 59.85*** 30.72**

*p<.05 **p<.01 ***p<.001 Key to Period Variable: Period 1 = days before Brown death on 8/9/14 Period 2= 25-day period after Brown death until Pierson death on 9/3/14 Period 3=25-day period after Pierson death Period 4=more than 25-days after Pierson death until December 31, 2015

141

Table 9: Interrupted Time Series Analysis, 2014 Only Negative Binomial Regression Predicting Daily Force Counts (N=365 Count Days) (Standard Errors in Parentheses) Model 10 Independent Variables B (SE) IRR

Period 1 (Reference) Period 2 -.026 (.234) 1.026 Period 3 .459 (.216) 1.583* Period 4 -1.017 (.678) .362

Sunday (Reference) Monday -.375 (.184) .688* Tuesday -.187 (.152) .829 Wednesday -.037 (.155) .964 Thursday -.279 (.142) .757 Friday .178 (.155) 1.195 Saturday .206 (.145) 1.229

Winter (Reference) Spring .229 (.118) 1.257 Summer .003 (.204) 1.003 Autumn .922 (.690) 1.515

Constant .593 (.136)*** Chi Square 45.98***

*p<.05 **p<.01 ***p<.001 Key to Period Variable: Period 1 = days before Brown death on 8/9/14 Period 2= 25-day period after Brown death until Pierson death on 9/3/14 Period 3=25-day period after Pierson death Period 4=more than 25-days after Pierson death until December 31, 2014

142

Figure 1: Age Range of Subjects

Figure 2: 5-day Smoothed Average Force Counts in 2014 Before and After Officer Pierson’s Death

143

Chapter 6: Discussion and Conclusions

Discussion Compared to the research investigating the correlates of police use of force at the encounter-level, there are relatively fewer studies examining how neighborhood context influences a police officer’s decision to use force. Several of the studies investigating neighborhood context indeed find an association between use of force and neighborhood characteristics such as racial composition, degree of disadvantage, and crime rates. This study adds to the growing body of research that incorporates these neighborhood characteristics into use of force models.

Using sociological theories and prior use of force research as a framework, this dissertation examined two research questions. First, drawing on theories of race (e.g., racial threat) and place (e.g., social disorganization), I investigate whether there are more frequent and severe uses of force in neighborhoods with higher percentages of Black and

Hispanic residents and greater degrees of disadvantage.

Second, using theories of group oriented processes (e.g., bounded solidarity and vicarious retribution) and Legewie’s 2016 study of the NYPD, I examine whether use of force is influenced by the timing of two antecedent events: violence by the police and violence against the police. Violence by the police is based on the fatal shooting of

Michael Brown on August 9, 2014, while violence against the police is based on the fatal shooting of Officer Daryl Pierson on September 3, 2014. In investigating these two research questions in the same study, this dissertation uniquely adds to the extant research by combining not only racial and socioeconomic contextual explanations for use of force, but also the effect of time.

144

The first significant finding in Chapter 5 is that a subject’s race, at the encounter- level, is unrelated to severity of force. Rather, both the frequency and severity of force are influenced by neighborhood context. Racial composition and concentrated disadvantage, when modeled separately, are each a significant predictor of the frequency and severity of force. However, after all neighborhood variables are modeled, concentrated disadvantage is no longer significant. Rather, both the frequency and severity of force are higher in neighborhoods with larger percentages of Black and

Hispanic residents. This finding supports Research Hypothesis 3b, as well as racial threat and Black’s thesis in the Behavior of Law as theoretical explanations for officer behavior when using force.

As applied to police officers, racial threat theory predicts that some officers perceive

Black individuals to be violent and prone to criminality. Disadvantaged neighborhoods are largely populated by Black residents. According to Donald Black’s theory, residents of these neighborhoods will be perceived by the criminal justice system as more deviant and thus more deserving of social control than higher status individuals. Consequently, both theories predict that the concentration of Black individuals (and to a lesser extent,

Hispanics) in disadvantaged neighborhoods is consequential for police decision-making.

Officers may perceive neighborhoods with greater degrees of poverty and disorder as areas that present elevated threats to their safety because more Black individuals live and congregate there. As such, these neighborhoods are perceived to require more frequent and severe social control.

Notably, when using force, officers are reacting not to the race of the individual subject, but to the racial composition of the neighborhood where the encounter occurs.

145

Thus, as an explanatory variable, race at the neighborhood-level is of paramount importance when analyzing uses of force. I therefore conclude that race and place are both contextual influences on the use of force, although the effect of concentrated disadvantage is indirect and is mediated by neighborhood racial composition.

The second significant finding in Chapter 5 is that violence by the police and violence against the police, particularly when they occur within a short time of each other, jointly influence subsequent uses of force. During the 25-day period following Michael Brown’s death, the trajectory of use of force incidents did not change significantly. However, the severity of force increased significantly compared to the period prior to Brown’s death. It is likely that officers remained confident in their authority and were not demoralized even during an intense period of nationwide negative publicity and anti-police backlash following Brown’s death. However, officers may have been confronted with more citizen defiance as a form of protest following Brown’s death. In response, officers increased the severity of force during citizen confrontations.

As predicted by Research Hypothesis 4b, the frequency of force significantly increased in the 25-day period immediately after Officer Pierson’s death. Officer

Pierson’s death motivated officers to use force more often during confrontations with citizens, whether out of fear for their safety, in bounded solidarity, or as vicarious retribution for Pierson’s shooting. The results of the interrupted time series analysis support all three theories as explanations for the spike in force following Officer

Pierson’s death.

Thereafter, frequency of force decreased significantly in Rochester through the end of

2015. One plausible interpretation for this finding is that officers were reacting to the

146 negative publicity, public protests, and anti-police backlash that accompanied several more high-profile police shootings or in-custody deaths of unarmed Black males. These include Eric Garner in New York City on July 17, 2014, Laquan McDonald in Chicago on October 20, 2014, Tamir Rice in Cleveland on November 22, 2014, Walter Scott in

North Charleston, S.C. on April 4, 2015, and Freddie Gray in Baltimore on April 21,

2015. Research in other jurisdictions (discussed in Chapter 4) indicates that officers were aware of these events and the negative publicity and allegations of racism leveled at the officers involved in those incidents. Rochester officers were also likely aware of these incidents and the public’s response to them. In Rochester, as elsewhere, officers likely avoided using force whenever possible to avoid negative publicity and allegations of racism. Avoiding using force whenever possible can be interpreted as a form of “de- policing.” Officers likely realized that there were no benefits and only harmful consequences to their careers and reputations from proactively engaging citizens.

Nonetheless, when a use of force could not be avoided, officers were more punitive.

The odds of being subjected to more severe force significantly increased even as the frequency of force declined during the 15 months following Officer Pierson’s death. Why would officers react more punitively during this period? While the frequency of force increased significantly only during the 25-days immediately after Officer Pierson’s death, it is unlikely that officers simply forgot about his shooting beyond those 25-days. Rather, it is highly probable that his shooting remained fresh in the minds of all Rochester officers for some time.

While it is unlikely they could forget even if they wanted to, officers were nonetheless periodically reminded that one of their own had been killed. In the fifteen

147 months following Officer Pierson’s shooting, there were several fundraisers for his family that were widely reported in local media.16 The one year anniversary (September

3, 2015) of his death was also widely reported.17 In addition, it was announced on

November 6, 2015 that the post office in Officer Pierson’s hometown of East Rochester,

New York would be named in his honor.18 Thus, a possible explanation for why severity of force increased more than 25-days after Officer Pierson’s death is that Rochester officers remained cognizant of his shooting when having to use force. They likely retained some fear for their own safety or, perhaps, a desire for retribution. Consequently, when use of force could not be avoided, officers responded by using high levels to increase the odds of a favorable outcome.

I conclude from the findings of the interrupted time series analysis that (1) officers, as predicted, initially responded to Officer Pierson’s death by significantly increasing the frequency, if not the severity, of force for the theoretical reasons previously discussed; and (2) the effects of Michael Brown’s death, the deaths of other unarmed Black males nationwide, and Officer Pierson’s death were joint influences on both the frequency and severity of force by Rochester officers in the 15 months from September 2014 through the end of 2015.

The findings from both research questions illustrate how social context influences legal decision making. Police officers work in different social contexts, which include urban, suburban, and rural communities. The racial and socioeconomic demographics of suburban and rural communities can be significantly different from urban areas. For example, as discussed in Chapter 5, a plurality (40.7%) of the city of Rochester’s residents in 2018 were Black. In comparison, the Town of Perinton, New York, a

148

Rochester suburb patrolled by the Monroe County Sheriff’s Office, had 46,462 residents in 2010, 92.7% of whom were white and only 1.7% of whom were Black (U.S. Census

Bureau). Wayne County, which is a rural county contiguous to Monroe County, had

93,772 residents in 2010, of whom 93.5% were white and only 3.1% of whom were

Black (U.S. Census Bureau). Compared to a Rochester police officer, a sheriff’s deputy in Wayne County or a sheriff’s deputy patrolling Perinton will likely deal with racial and ethnic minorities less often and be dispatched to fewer 911 calls involving interpersonal violence.

In rural communities like Wayne County and largely white suburbs like Perinton, where neighborhoods are more homogeneous (i.e., largely comprised of white working or middle-class residents), use of force is significantly more likely to be related only to encounter-level variables and not neighborhood characteristics. Minorities will be encountered less frequently in these areas. However, there is no reason to believe that officers working in largely white communities are immune to racial stereotypes about

Black individuals. Consequently, when minorities are encountered, they may be dealt with more punitively than white subjects. In these homogeneous communities, race at the encounter-level may be more salient as a predictor of police use of force than race at the neighborhood-level.

The findings of this dissertation indicate that the opposite is true in urban environments. The character of the neighborhoods in Rochester include many poor, racially segregated neighborhoods, and fewer racially mixed, less disadvantaged neighborhoods. The concentrated disadvantage index for Rochester neighborhoods ranged from a low of 24.7 to a high of 189.5, with a mean of 97.9 and a median of 100.3,

149 which indicates that the distribution is skewed toward higher degrees of disadvantage.

Similarly, neighborhood percentage of Black residents ranged from 1.9% to 96.6%, with a mean of 45.8% and a median of 44.2%, which indicates that half of Rochester’s neighborhoods have at least 44% or more Black residents.

These demographics and the statistical findings of Chapter 5 support the conclusion that Rochester police officers are cognizant of the vastly different racial and socioeconomic characteristics of the Rochester neighborhoods they patrol. Use of force is more frequent and severe against anyone encountered in neighborhoods with higher percentages of Black and Hispanic residents. In this sense, the findings also support

Werthman and Piliavin’s ecological contamination thesis in that anyone, regardless of race, encountered in a neighborhood that the police perceive to be “bad” runs a higher risk of a punitive response when interacting with a police officer. And although the object of the interrupted time series analysis was the influence of the timing of events and not neighborhood context, the results indicate that officers are also keenly aware of both national and local events that impact their authority and safety. In short, social context, whether it be based on race, place, or time, significantly influences legal decision making by police officers regarding use of force.

More broadly, the dissertation’s findings illustrate the role law as governmental social control plays in preserving the social order. Officers, like other Americans who associate the size of a neighborhood’s Black population with its crime rate (see e.g., Quillian &

Pager, 2001), may fear an elevated risk of violent victimization in predominantly Black neighborhoods. In addition, officers, who are overwhelmingly white and male, may believe that culturally dissimilar racial and ethnic minorities threaten the social order. In

150

Black’s terms, racial and ethnic minorities, who are culturally distant by virtue of being on the periphery of society, become vulnerable to law (Black, 1976, p. 52).

Consequently, the use of force is one form of law employed by the state to coercively control deviant individuals, particularly those who are perceived to threaten the dominant social order by spreading crime and violence.

Racial threat theory fared particularly well as a theoretical framework to explain why the police apply law disproportionately to preserve the social order. Racial threat theory and this dissertation’s findings imply that other legal mechanisms of social control may also be applied disproportionately against minorities. Some studies indeed find racial bias in pedestrian and vehicle stops by the police (e.g., Fagan, 2002; Fagan & Davies, 2000;

Fagan, Zimring, & Kim, 1997). For example, researchers found racial bias in the NYPD’s

“stop, question, and frisk” (“SQF”) program as Black and Hispanic pedestrians were stopped more often than whites, even after controlling for social and economic factors, precinct crime rates, and neighborhood racial composition (Fagan, Conyers, & Ayers,

2014; Gelman, Fagan & Kiss, 2007; Morrow et al 2017). In 2013, a federal judge held in

Floyd v. City of New York that the NYPD violated the Fourth Amendment by engaging in a pattern and practice of racial profiling and unconstitutional stops. Testimony during the trial documented the “disproportionate, harmful impact of SQF on minority citizens in

New York’s low-income, high-crime neighborhoods” (Morrow et al. 2017, p. 371).

Racial threat as a theoretical framework can also be applied to explain vigilantism by white citizens against Black citizens, particularly those whose only “threat” to the social order is their presence in a white neighborhood. A prominent example is the shooting death of in Florida in February 2012.19 Martin, a teenager, was walking

151 in a white neighborhood when he was confronted by a “neighborhood watch” captain,

George Zimmerman. Zimmerman had called 911 to report a “suspicious person” in his neighborhood and was instructed not to exit his vehicle or confront the person.

Zimmerman disregarded those instructions, confronted Martin (who was unarmed), and became involved in a physical altercation during which he shot Martin to death.

Zimmerman, who claimed he was in fear for his life, was found by a jury to have used justifiable self-defense and was acquitted of murder charges.

More recently, on February 23, 2020, a white father and son followed a Black 25-year old man jogging through their neighborhood in rural Brunswick, Georgia.20 Believing him to be a “burglary suspect,” the men, one of whom carried a shotgun, confronted the unarmed jogger. As in Martin’s case, the confrontation led to a physical altercation in which the jogger, Ahmaud Arbery, was shot to death. The father and son were arrested for murder almost three months later after a video of the confrontation was leaked to the media. If Zimmerman’s case is any guide, the men will allege that Arbery precipitated his death by initiating a physical confrontation that caused them to fear for their lives and act in self-defense. Ironically, the attorney who leaked the video, Alan Tucker, is friends with the defendants and mistakenly believed the video would help them. As he told

“Inside Edition,” Arbery would still be alive if he just “froze” and did not confront his friends.21 The incidents involving Martin and Arbery illustrate that social control of individuals who are perceived to threaten the social order can occur without state intervention.

152

Implications for Police Training

The dissertation’s findings have implications for officer training. The RPD already has a rigorous hiring process that seeks to identify and disqualify candidates who have exhibited racial bias in their past, are emotionally immature, and have low self-control.

Recruit candidates are subjected to psychological testing, a thorough background check, and a polygraph examination. Academy training includes multicultural awareness and community policing courses. However, once they complete academy training, officers

“hit the street” and are paired with a series of field training officers before being certified to work alone. They quickly learn that field training provides a significantly different educational experience than classroom training. Everything, every day, is real and not simulated. They are taught that survival and “going home safe” at the end of the shift are the only objectives that matter.

Most officers who join the RPD are white, male, from middle-class backgrounds, and have little life experience outside of school. While some served in the military or have prior work experience after college, most do not. Some of these new officers now find themselves working in poor, violent, racially segregated neighborhoods that could not be more different from those in which they grew up. In addition, they are now interacting daily with racial and ethnic minorities whose lived experiences are vastly different from theirs. In contrast, other new officers are assigned to more racially mixed or predominantly white neighborhoods with middle-class incomes and lower levels of disadvantage.

This dissertation’s findings suggest that all officers, regardless of the characteristics of the neighborhoods to which they are assigned, adapt to their different working

153 conditions. Those patrolling in neighborhoods with higher percentages of Black and

Hispanic residents adapt by using force more frequently and at more severe levels than their counterparts patrolling less disadvantaged neighborhoods that have more white residents. Research discussed in Chapter 3 and these findings strongly suggest that implicit racial bias is responsible for neighborhood variation in use of force.

Other cities have confronted this issue through implicit bias training. A foundational idea underlying this training is that class, gender, and racial biases are the products of everyday life in the U.S. (The Atlantic, 12/23/17)22. As discussed in Chapter 3, implicit bias is often conveyed nonverbally through television, particularly news programs.

Participants in implicit bias training programs learn that even people who embrace fairness and equality may draw on unconscious stereotypes when evaluating people or situations (The Atlantic, 12/23/17). Thus, even the best of intentions cannot prevent bias from surfacing.

One example of this training was reported by on July 15, 2018.

The Times reported that the NYPD instituted a $4.5 million implicit bias training program for all sworn officers.23 The program was conducted by Fair and Impartial

Policing, a Florida company that the Times described as a leading provider of such training. Trainers included retired law enforcement officers such as Barbara Leffler, who observed that officers initially exhibit defensive or hostile body language. However, she notes, once they realize that implicit bias is a human issue and not a law enforcement issue, officers become open to learning how bias seeps into the subconscious mind beginning in childhood. In other words, once it becomes clear that the trainer is not painting police officers as racists, but as human beings who have been exposed to bias

154 and stereotypes their entire lives, they are more willing to learn how to become aware of bias and manage it when it surfaces.

The Atlantic noted that implicit bias training is spreading to departments throughout the U.S. However, an issue with the training is the lack of standards regarding content, the most effective way to teach the material, and how to measure its effectiveness (The

Atlantic, 12/23/17). There is also a lack of data indicating whether this training will be effective over the long-term. But considering this dissertation’s findings and their implications for police legitimacy and police-citizen relations, this training may be the only way to make officers aware of their bias and learn strategies for managing it. While

RPD’s rigorous employment screening may be effective in detecting persons with a history of explicit racial bias, it appears to be ineffective at detecting implicit bias.

Accordingly, a recommendation for RPD leaders is to provide implicit bias training to all sworn officers and incorporate this training into the academy curriculum.

As noted, most RPD officers are white and male. As long ago as 1967, the Kerner

Commission on Civil Disorders, which was formed by President Johnson to investigate the causes of race riots in 1967, suggested that tensions could be diffused with greater recruitment of Black officers by urban police departments (Weitzer, 2000). Subsequent government commissions have echoed this view (e.g., The U.S. Commission on Civil

Rights, 1981; President’s Advisory Board, 1998). Contemporary public policy assumes that police officers should reflect the demographics of the communities they serve to promote positive police-community relations (Weitzer, 2000). Consequently, the movement to change police department demographics has been a major tenet of police reform (Brown & Frank, 2006).

155

The RPD has actively attempted to recruit female police candidates as well as candidates of color. Indeed, the RPD is required under a U.S. Justice Department Consent

Decree to do so. As such, the RPD maintains separate civil service lists of potential hires and must hire one minority candidate for every two majority candidates. Female officers and officers of color are displayed prominently on all recruitment posters. The findings of this dissertation, however, indicate that there is no statistical difference, on average, between white, Black, and Hispanic officers, or male and female officers, in the severity of force used at the encounter-level. While the presence of more minority officers may enhance police legitimacy and inspire greater citizen confidence in some neighborhoods, it does not appear that minority or female officers are behaving any differently than their white male counterparts when using force.

Police academy training socializes recruits into police culture and prepares them to encounter the myriad, sometimes life-threatening, situations they will experience on the street. As they are socialized, they undergo a metamorphosis from recruit to officer in field training to trained officer handling 911 calls for service. The findings here suggest that this socialization process causes officers, regardless of their diverse backgrounds, values, and beliefs, to become similar in their orientations about police work.

An academic debate exists as to whether there is a single, homogeneous occupational culture emphasizing a “crime-fighting, warrior” role of police versus cultural heterogeneity emphasizing the role of police as guardians. Paoline (2004), for example, finds that while some officers continue to identify with “traditionalist” attitudes associated with a monolithic view of police culture, the majority possess a “guardian mentality,” which includes holding more diverse cultural attitudes, positive views of

156 citizens, and an approach to police work that does not emphasize aggressive crime fighting.

Based on the findings of this dissertation, however, training has not sufficiently emphasized the guardian mentality as the role of police in the 21st Century. Indeed, while coding SRRs, I observed a disturbing number of incidents that, in my view, might have been resolved without the use of force if the officer had simply continued talking. For example, the SRR’s narrative suggested that, while trying to resolve a dispute in public, the officer became irritated at the use of profanity and quickly decided to make a arrest. Had the officer ignored the profanity and continued trying to resolve the issue, a use of force could potentially have been avoided.

I therefore recommend that academy training include more courses emphasizing the role of police as guardian, which includes the need for resolving confrontations with verbal skills rather than force wherever possible. When reviewing SRRs, supervisors should be alert for situations, like that describe above, where force, even if used appropriately, might have been avoided if the officer continued talking. Counseling of the involved officers should then follow. It must be acknowledged that officer behavior may be difficult to change, particularly if academy training fades as the officer is socialized on the street. But what is taught now does not appear to be working and therefore this dissertation’s findings suggest that the training process needs revamping.

In addition, the interrupted time series findings have implications for police management following the murder of an officer in the line of duty. The murder of a police officer generates strong emotions, including anger, sadness, and a sense of vulnerability and loss. Police leadership should communicate to their officers that they

157 should mourn their colleague and acknowledge these legitimate emotions. However, they should also remind officers to not engage in a campaign of retribution against citizens, particularly citizens of the same race as the offender, as to do so would blame all members of that race for the criminal actions of one person.

Limitations

This study is not without limitations. One concern is the reliance on officer self-report data. Officers, whether on or off duty, are required to complete an SRR whenever they are involved in a use of force. Under RPD policy, a use of force can be something as minor as shoving a verbally-resistance subject against a wall. It is entirely possible that a minor use of force, which may not be accompanied by an arrest or is not recorded on video, may not be documented.

Alternatively, the use of more severe force, such as striking a subject with a baton or threatening the use of a taser or handgun when it is not justified by the situation or RPD policy, may also go unreported.24 It is thus not possible to know how many uses of force, whether low or high, were unreported during the study period.

On the positive side, SRRs, once generated, go through several layers of review. The review chain includes the officer’s immediate supervisor, that supervisor’s supervisor, the commander of the patrol unit, and then the head of the defensive tactics program, who is assigned to the Professional Development Section. It is not unusual for deficient SRRs to be “kicked back” somewhere along this chain of review to the patrol officer, who is then required to provide more detail regarding that officer’s actions.

In addition, while reviewing and downloading SRRs, I noted an occasional SRR that included an accompanying counseling memorandum documenting and informing the

158 officer of failure to adhere to RPD policy. These actions, which occasionally resulted in disciplinary action and/or retraining of the officer, ranged from failing to have a supervisor witness a refusal for medical attention to more serious infractions such as using an inappropriate technique or an approved technique that was not justified by the situational aspects of the encounter. This suggests that officers, who know their actions will be scrutinized and may even have been captured on video, are documenting, sometimes to their detriment, the actual force they used.

Moreover, the findings indicate that the frequency and severity of force are significantly higher in neighborhoods with larger percentages of Black and Hispanic residents. If officers were intentionally underreporting some uses of force, most omissions are also likely to have occurred in those neighborhoods. The question is why?

It is likely because any use of force invites scrutiny not only from an officer’s direct supervisor, but also from RPD command personnel. Even before Michael Brown’s death, use of force in minority communities was a sensitive political issue. More frequent uses of force in minority neighborhoods can be problematic politically for police management, the city council, and the mayor. An officer, even if using force in good faith, cannot be

100% confident that the review process will determine it to have been appropriate under the circumstances. If a documented use of force in a minority neighborhood or against a minority suspect is later determined to be inappropriate, the officer may face disciplinary consequences such as unpaid time off. When a use of force occurs at night, without independent witnesses, and without causing an injury (e.g., in the backyard of a vacant house), the path of least resistance might be to not document or report it. If the citizen files a complaint with the Professional Standards Section, and there are no independent

159 witnesses or corroborating evidence, the case will be closed, and the officer will face no disciplinary consequences. If bias in reporting occurred during the study period, I expect that the motive would be to avoid scrutiny and the risk of discipline. Consequently, the findings here may be conservative. Frequency and severity of force may be higher in minority neighborhoods than the results suggest.

While the findings indicate that frequency of force decreased after Officer Pierson’s shooting, another limitation is that we can only speculate as to the reason why. I suggest that de-policing due to the influence of national events is the likely cause. This theory would be strengthened if we had data measuring proactive police stops and arrest rates before and after Officer Pierson’s death. If de-policing were indeed the cause, I would expect to also see a decrease in proactive police stops and/or arrests.

While de-policing remains the leading theory to explain decreases in the frequency of force, another possibility is the contributing effect of declining crime rates. Part 1 crimes reported to the RPD decreased by 4.5% from July 2014 to December 2015 (NYDCJS;

Rochester Open Data Portal). Part 1 crimes are considered “reported” to the RPD when an officer generates a crime report for a Part 1 crime. An actual reduction in Part 1 crimes should lead to a decline in uses of force as there would be an accompanying reduction in police-citizen interactions with potential suspects. Another possibility is that Part 1 crimes did not actually decrease but went unreported to 911 due to fear of retaliation or the effect of legal cynicism (Desmond et al. 2016), which would also lead to fewer police-citizen interactions and uses of force.

Another limitation is not having an independent variable measuring the degree of disrespect, if any, displayed towards the officer prior to the use of force. Research

160 regarding the effect of subject demeanor on use of force is mixed (see, e.g., Sun & Payne,

2004, who find that officers were more likely use force to resolve a dispute involving disrespectful citizens; McCluskey & Terrill, 2005, who find no relationship between subject demeanor and use of force). Nonetheless, the study would have benefitted from a measure of subject demeanor at the encounter-level to estimate whether it is a significant predictor of force in Rochester. The study would have also benefitted in analysis of force at the encounter-level from measures of officer age, experience, and education. Without such measures, the study was unable to determine whether younger officers, with less experience, use force more often or more severely than older, more experienced officers.

In addition, it would been possible to determine whether officers with some degree of college or graduate education use force differently than less educated officers.

A final limitation is not having the dispatch data the officer received prior to responding to a 911 call. The nature of the call for service, for example shots fired or an armed robbery, which typically occur in more violent, racially segregated neighborhoods, may predispose officers to use force when confronting a suspect. Thus, the ecological pattern of force may also be partially driven by where 911 calls for service involving violence are distributed.

On the other hand, the research discussed in Chapter 3 and the findings suggest that officers already perceive neighborhoods with larger percentages of Black and Hispanic residents to be a more significant threat to their safety. This perception is driven primarily by the implicit racial bias that characterizes Black individuals as more violent.

Consequently, it is possible that the nature of the call for service just reinforces the bias officers already possess. Officers likely expect more calls for service involving violence

161 to occur in neighborhoods that they already perceive to be violent because of their racial composition.

Directions for Future Research

Officers in Rochester, as in other jurisdictions, are now required to wear body cameras when interacting with citizens. The use of body worn cameras is a relatively new area of research. An ideal future project in Rochester would collect individual uses of force recorded on body cameras over a 5-year period. Research could then explore whether neighborhood context is still significantly associated with use of force. Does neighborhood context become less salient than encounter-level variables when every use of force is recorded and available for supervisory review? Or do officers continue to act more aggressively in Black and Hispanic neighborhoods because they perceive them to be more dangerous areas?

A study that combines use of force data with individual survey data measuring legal cynicism would also provide a fruitful research project. At the encounter-level, individual uses of force and resident measures of legal cynicism would be geocoded to the census tracts where they occur. This data would then be merged with the neighborhood-level predictors included in this and other multilevel studies. Multilevel analysis could then uncover whether uses of force are still higher or more severe in neighborhoods with higher percentages of Black and Hispanic residents and whether legal cynicism at the encounter-level is also a significant predictor of severity of force. A cross-level interaction of legal cynicism at the encounter-level and racial composition at the neighborhood-level could also be modeled to determine whether severity of force is influenced by the interaction between cynical views of the law and race.

162

Conclusion

The use of physical coercion to enforce the law and ensure public safety is an essential function of the state. However, the police cannot use force indiscriminately or be influenced by extralegal factors such as race. The U.S. Constitution and state law require that force must be situationally justified and reasonably applied without regard to personal characteristics, as citizens have a right to be free from an unreasonable seizure by their government. Use of force continues to be an important societal issue due to the potential for abuse of one of our most cherished rights as Americans.

This dissertation contributes to the scholarly literature on police behavior by investigating two understudied areas: how neighborhood characteristics, particularly neighborhood racial composition and degree of disadvantage, and the timing of antecedent events influence both the frequency and severity of force used by the police.

Findings indicate that police use of force is influenced not by the subject’s race at the encounter-level but by a different extralegal factor: neighborhood racial composition. In addition, antecedent events may also be extralegal factors that influence the trajectory of subsequent uses of force. More broadly, these findings illustrate that law may be applied disproportionately against those with the least power and status in society, but who nonetheless threaten the established social order.

163

References

Adams, K. 1999. “What We Know About Police Use of Force.” In J. Reno, R. Fisher, L. Robinson, and N. Brennan (Eds.). Use of Force by Police: Overview of National and Local Data (pp. 1-14). Washington, DC: National Institute of Justice and Bureau of Justice Statistics.

Alba, R.D., J.R. Logan, and P.E. Bellair. 1994. “Living with Crime: The Implications of Racial/Ethnic Differences in Suburban Location.” Social Forces 73:395-434.

Alba, R.D., J.R. Logan, and B.J. Stults. 2000. “The Changing Neighborhood Contexts of the Immigrant Metropolis.” Social Forces 73:57-621.

Alpert, G.P., and J.M. MacDonald. 2001. “Police Use of Force: An Analysis of Organizational Characteristics.” Justice Quarterly 18:393-409.

Anderson, E. 1999. Code of the Street: Decency, Violence and the Moral Life of the Inner City New York: Norton.

Barkan, S.E., and S.F. Cohn. 1994. “Racial Prejudice and Support for the Death Penalty by Whites.” Journal of Research in Crime & Delinquency 31:202-209.

Baumer, E.P., and K.H. Martin. 2013. “Social Organization, Collective Sentiment, and Legal Sanctions in Murder Cases.” American Journal of Sociology 119:131-182.

Bayley, D.H., and J. Garofalo. 1989. “The Management of Violence by Police Patrol Officers.” Criminology 27:1-27.

Bayley, D.H., and H. Mendelsohn. 1969. Minorities and the Police: Confrontation in America New York: Free Press.

Bentele, K.G., and E.E. O’Brien. 2013. “Jim Crow 2.0? Why States Consider and Adopt Restrictive Voter Access Policies.” Perspectives on Politics 11:1088-1116.

Bittner, E. 1970. The Functions of the Police in Modern Society. Washington, D.C.: Government Printing Office.

Black, D. 1976. The Behavior of Law. San Diego, CA: Academic Press.

Black, D. 1983. “Crime as Social Control.” American Sociological Review 48:34-45.

Blalock, H.M. 1967. Toward a Theory of Minority-Group Relations. New York: Capricorn.

164

Blumer, H. 1958. “Race Prejudice as a Sense of Group Position.” Pacific Sociological Review 1:3-7.

Bradford, B., and P. Quinton. 2014. “Self-Legitimacy, Police Culture, and Support for Democratic Policing in an English Constabulary.” British Journal of Criminology 54:1023-1046.

Braga, A.A., D.L. Weisburd, E.J. Waring, L.G. Mazerolle, W. Spelman, and F. Gajewski. 1999. “Problem-Oriented Policing in Violence Crime Places: A Randomized Controlled Experiment.” Criminology 37:541-580.

Brown, Robert A., and James Frank. 2006. “Race and Officer Decision Making: Examining Differences in Arrest Outcomes between Black and White Officers.” Justice Quarterly 23:96-126.

Campbell, B.A., J. Nix, and E. R. Maguire. 2018. “Is the Number of Citizens Fatally Shot by Police Increasing in the Post-Ferguson Era?” Crime & Delinquency 64:398-420.

Chamlin, M.B. 1989. “Conflict Theory and Police Killings.” Deviant Behavior 10:353- 368.

Chiricos, T., M. Hogan, and M. Gertz. 1997. “Racial Composition of Neighborhood and Fear of Crime.” Criminology 35:107-131.

Chiricos, T., K. Welch, and M. Gertz. 2004. “Racial Typification of Crime and Support for Punitive Measures.” Criminology 42:359-389.

Collective Bargaining Agreements Between the city of Rochester and the Rochester Police Locust Club, Expiring 6/30/12, 6/30/14, 6/30/16.

Correll, J., B. Park, C.M. Judd, and B. Wittenbrink. 2007. “The Influence of Stereotypes on Decisions to Shoot.” European Journal of Social Psychology 37:1102-1117.

Culhane, S.E., J.H. Boman IV, and K. Schweitzer. 2016. “Public Perceptions of the Justifiability of Police Shootings: The Role of Body Cameras in a Pre- and Post – Ferguson Experiment.” Police Quarterly 19:251-274.

Democrat & Chronicle. 9/5/14. “’Shock, disbelief’ in days after Officer Daryl Pierson’s Death.” Retrieved from https://www.democratandchronicle.com/story/news/2014/09/05/hundreds-attend-special- mass-police-officer-daryl-pierson/15119199/

165

Democrat & Chronicle, Deadly officer shooting: Updates as it happened, Published 10:14 p.m. ET Sept. 3, 2014, retrieved from https://www.democratandchronicle.com/story/news/2014/09/03/breaking-police- investigate-shooting/15050817/

Department of Justice, March 4, 2015. Department of Justice Report on the Criminal Investigation into the Shooting Death of Michael Brown by Ferguson, Missouri\ Police Officer Darren Wilson Retrieved 2/25/20 from: https://www.justice.gov/sites/default/files/opa/press- releases/attachments/2015/03/04/doj_report_on_shooting_of_michael_brown_1.pdf

Desmond, M., A. V. Papachristos, and D.S. Kirk. 2016. “Police Violence and Citizen Crime Reporting in the Black Community.” American Sociological Review 81(5):857-876.

Deuchar, R., S. Wyatt Fallik, and V.J. Crichlow. 2019. “Despondent Officer Narratives and the ‘Post Ferguson’ Effect: Exploring Law Enforcement Perspectives and Strategies in a Southern American State.” Policing and Society 29:1042-1057.

Eck, J.E. 1993. “Alternative Futures for Policing.” In Police Innovation and the Control of the Police, edited by D. Weisburd and C. Uchida. New York: Springer-Verlag.

Eberhardt, J.L., P.A. Goff, V.J. Purdie, and P.G. Davies. 2004. “Seeing Black: Race, Crime, and Visual Processing.” Journal of Personality and Social Psychology 87:876-893.

Eith, C., & M.R. Durose (2011). Contacts Between Police and the Public, 2008. Washington, D.C.: U.S. Department of Justice. Retrieved from https://www.bjs.gov/content/pub/pdf/cpp08.pdf

Engel, R.S., C.F. Klahm, and R. Tillyer. 2010. “Citizens’ Demeanor, Race, and Traffic Stops.” In S. Rice & M.D. White (Eds.), Race, Ethnicity, and Policing: New and Essential Readings. New York, NY: New York University Press.

Engle, R., J. Sobol, and R. Worden. 2000. “Further Exploration of the Demeanor Hypothesis: The Interaction Effects of Suspects’ Characteristics and Demeanor on Police Behavior.” Justice Quarterly 17:235-258.

Fagan, J. 2002. “Law, Social Science, and Racial Profiling.” Justice Research and Policy 4:103-130.

Fagan, J., G. Conyers, and I. Ayres. 2014. No Runs, Few Hits and Many Errors: Street Stops, Bias and Proactive Policing. Berkeley: Conferences on Empirical Legal Studies.

166

Fagan, J., and G. Davies. 2000. “Street Stops and Broken Windows: Terry, Race and Disorder in New York City.” Fordham Urban Law Journal 28:457-504.

Fagan, J., F.E. Zimring, and J. Kim. 1997. “Declining Homicide in New York City: A Tale of Two Trends.” Journal of Criminal Law and Criminology 88:1277-1324.

Federal Bureau of Investigation. 2013. Crime in the United States, 2013. Retrieved from https://ucr.fbi.gov/crime-in-the-u.s/2013/crime-in-the-u.s.-2013.

Ferrandino, J. 2015. “Minority Threat Hypothesis and NYPD Stop and Frisk Policy.” Criminal Justice Review 40:209-229.

FindLaw, Thomson Reuters, retrieved 9/30/19 from https://codes.findlaw.com/ny/penal- law/pen

Floyd v. New York City, 959 F. Supp. 2nd 540 (SDNY 2013).

Freiburger, Tina, L. 2019. “Improving Youths’ Attitudes About the Police: Results from an Experimental Design.” Criminal Justice Review 44:413-430.

Friedrich, R.J. 1980. “Police Use of Force: Individuals, Situations, and Organizations.” The Annals of the American Academy of Political and Social Science 452:82-97.

Gabriel, R., and S. Tolnay. 2017. “The Legacy of Lynching? An Empirical Replication and Conceptual Extension.” Sociological Spectrum 37:77-96.

Gallop Editors. (6/12/2014). Gallop Review: Black and White Differences in Views on Race. Gallop. Retrieved from: https://news.gallup.com/poll/180107/gallup-review-black-white-differences-views-race.aspx

Garner, J., C.D. Maxwell, and C. Heraux. 2002. “Characteristics Associated with the Prevalence and Severity of Force Used by the Police.” Justice Quarterly 19:705-746.

Garner, J.H., J. Buchanan, T. Schade, and J. Hepburn. 1996. Understanding the Use of Force by and Against the Police. Washington, DC: National Institute of Justice.

Garner, J.H., T. Schade, J. Hepburn, and J. Buchanan. 1995. “Measuring the Continuum of Force Used by and Against the Police.” Criminal Justice Review 20:146-168.

Gau, J.M., N. Corsaro, E.A. Steward, and R. Brunson. 2012. “Examining Macro-Level Impacts on Procedural Justice and Police Legitimacy.” Journal of Criminal Justice 40:333-343.

Gelman, A., J. Fagan, and A. Kiss. 2007. “An Analysis of the New York City Police Department’s ‘Stop-and-Frisk’ Policy in the Context of Claims of Racial Bias.” Journal of the American Statistical Association 102:813-823.

167

Gilliam, Jr., F.D., and S. Iyengar. 2000. “Prime Suspects: The Influence of Local Television News on the Viewing Public.” American Journal of Political Science 44:560-573.

Goldstein, H. 1990. Problem Oriented Policing. New York: McGraw-Hill.

Gottfredson, M.R., and M.J. Hindelang. 1979. “A Study of the Behavior of Law.” American Sociological Review 44:3-18.

Graham v. Connor, 490 U.S. 386 (1989).

Graham, S., and B. S. Lowery. 2004. “Priming Unconscious Racial Stereotypes About Adolescent Offenders.” Law and Human Behavior 28:483-504.

Greenberg, D.F. 1983. “Donald Black’s Sociology of Law: A Critique.” Law & Society Review 17:337-368.

Greene, J. 2000. “Community Policing in America: Changing the Nature, Structure, and Function of the Police.” Criminal Justice 3:299-370. Downloaded 9/30/19 from https://www.ncjrs.gov/criminal_justice2000/vol_3/03g.pdf.

Greenwald, A.G., and L. H. Krieger. 2006. “Implicit Bias: Scientific Foundations.” California Law Review 94:945-967.

Griffin, S.P., and T.J. Bernard. 2003. “Angry Aggression Among Police Officers.” Police Quarterly 6:3-21.

Hayden, G.A. 1981. “Police Discretion in the Use of Deadly Force: An Empirical Study of the Information Usage in Deadly Force Decision Making.” Journal of Police Science and Administration 9:102-107.

Hays, Z.R. 2011. Police Use of Excessive Force in Disorganized Neighborhoods. El Paso: LFB Scholarly Publishing LLC.

Herbert, S. 1997. Policing Space: Territoriality and the Los Angeles Police Department. Minneapolis, MN: University of Minnesota.

Hickman, M.J., A.R. Piquero, and J.H. Garner. 2008. “Toward a National Estimate of Police Use of Nonlethal Force.” Criminology & Public Policy 7:563-604.

Holmes, M.D. 2000. “Minority Threat and : Determinants of Civil Rights Criminal Complaints in U.S. Municipalities.” Criminology 38:343-367.

Holmes, M.D., and B.W. Smith. 2012. “Intergroup Dynamics of Extra-Legal Police Aggression: An Integrated Theory of Race and Place.” Aggression and Violent Behavior 17:344-353.

168

Hyland, S., L. Langton, and E. Davis. 2015. Police Use of Nonfatal Force, 2002-11. Bureau of Justice Statistics Special Report. Retrieved on 12/12/19 from https://www.bjs.gov/content/pub/pdf/punf0211.pdf

International Association of Chiefs of Police. 2001. Police Use of Force in America 2001 Retrieved from https://www.theiacp.org/sites/default/files/2018-08/2001useofforce.pdf

Jackson, P.I., and L. Carroll. 1981. “Race and the War on Crime.” American Sociological Review 46:290-305.

Jacobs, D., and D.W. Britt. 1979. “Inequality and the Police Use of Deadly Force.” Social Problems 26:403-412.

Jacobs, D., and R.M. O’Brien. 1998. “The Determinants of Deadly Force: A Structural Analysis of Police Violence.” American Journal of Sociology 103:837-862.

Johnson, B.D. “The Multilevel Context of Criminal Sentencing: Integrating Judge- and County-Level Influences.” Criminology 44:259-298.

Kaminski, R.J., C. Digiovanni, and R. Downs. 2004. “The Use of Force Between the Police and Persons with Impaired Judgment.” Police Quarterly 7:311-338.

Kane, R.J. 2002. “The Social Ecology of Police Misconduct.” Criminology 40:867-896.

King, R.D., and M.T. Light. 2019. “Have Racial and Ethnic Disparities in Sentencing Declined?” Crime and Justice 48:365-437.

King, R.D., S.F. Messner, and R.D. Baller. 2009. “Contemporary Hate Crimes, Law Enforcement, and the Legacy of Racial Violence.” American Sociological Review 74:29-315.

King, R.D., and G.M. Sutton. 2013. “High Times for Hate Crimes: Explaining the Temporal Clustering of Hate-Motivated Offending.” Criminology 51:871-894.

Kirk, D.S., and M. Matsuda. 2010. “Legally Cynicism, Collective Efficacy, and the Ecology of Arrest.” Criminology 49:443-472.

Kirk, D.S., and A.V. Papachristos. 2011. “Cultural Mechanisms and the Persistence of Neighborhood Violence.” American Journal of Sociology 116:1190-1233.

Klinger, D.A. 1997. “Negotiating Order in Patrol Work: An Ecological Theory of Police Response to Deviance.” Criminology 35:277-306.

169

Kochel, T.M. 2019. “Explaining Racial Differences in Ferguson’s Impact on Local Residents’ Trust and Perceived Legitimacy: Policy Implications for Police.” Criminal Justice Policy Review 30:374-405.

Kop, N., and M. Euwema. 2001. “Occupational Stress and the Use of Force by Dutch Police Officers.” Criminal Justice and Behavior 28:631-652.

Kramer, R., and B. Remster. 2018. “Stop, Frisk, and Assault? Racial Disparities in Police Use of Force During Investigatory Stops.” Law & Society Review 52:960-993.

Krivo, L.J., and R.D. Peterson. 1996. “Extremely Disadvantaged Neighborhoods and Urban Crime.” Social Forces 75:619-648.

Lautenschlager, R., and M. Omori. 2019. “Racial Threat, Social (Dis)organization, and the Ecology of Police: Towards a Macro-level Understanding of Police Use-of-force in Communities of Color.” Justice Quarterly 36:1050-1071.

Lawton, B.A. 2007. “Levels of Nonlethal Force: An Examination of Individual, Situational, and Contextual Factors.” Journal of Research in Crime and Delinquency 44:163:184.

Lee, H., J. Hyunseok, I. Yun, H. Lim, and D.W. Tushaus. 2010. “An Examination of Police Use of Force Utilizing Police Training and Neighborhood Contextual Factors.” Policing: An International Journal of Police Strategies and Management 33:681-702.

Lee, H., M.S. Vaughn, and H. Lim. 2014. “The Impact of Neighborhood Crime Levels on Police Use of Force: An Examination at Micro and Meso Levels.” Journal of Criminal Justice 42:491-499.

Legewie, J. 2013. “Terrorist Events and Attitudes toward Immigrants: A Natural Experiment.” American Journal of Sociology 118:1199-1245.

Legewie, J. 2016. “Racial Profiling and Use of Force in Police Stops: How Local Events Trigger Periods of Increased Discrimination.” American Journal of Sociology 122:379-424.

Legewie, J., and J. Fagan. 2016. “Group Threat, Police Officer Diversity, and the Deadly Use of Police Force.” Columbia Public Law Research Paper no. 14-215. https://ssrn.com/abstract=2778692

Lersch, K.M. 1998. “Police Misconduct and Malpractice: A Critical Analysis of Citizens’ Complaints.” Policing: An International Journal of Police Strategies & Management 21:80-96.

170

Lersch, K.M., T. Bazley, T. Mieczkowski, and K. Childs. 2008. “Police Use of Force and Neighbourhood Characteristics: An Examination of Structural Disadvantage, Crime, And Resistance.” Policing & Society, 18:282-300.

Levchak, P.J. 2017. “Do Precinct Characteristics Influence Stop-and-Frisk in New York City? A Multi-Level Analysis of Post-Stop Outcomes.” Justice Quarterly 34:377-406.

Lickel, B., N. Miller, D.M. Stenstrom, T.F. Denson, and T. Schmader. 2006. “Vicarious Retribution: The Role of Collective Blame in Intergroup Aggression.” Personality And Social Psychology Review 10:372-390.

Light, M.T., M. Massoglia, and R.D. King. 2014. “Citizenship and Punishment: The Salience of National Membership in U.S. Criminal Courts.” American Sociological Review 79:825-847.

Liska, A., and J. Yu. 1992. “Specifying and Testing the Threat Hypothesis: Police Use of Deadly Force.” in Social Threat and Social Control, A. Liska (Ed.) Albany: State University of New York Press.

Logan, J.R., and B.J. Stults. 1999. “Racial Differences in Exposure to Crime: The City and Suburbs of Cleveland in 1990.” Criminology 37:251-276.

Manning, P. 1978. “The Police: Mandate, Strategies, and Appearances.” In P. Manning and J. Van Maanen (Eds.) Policing: A View from the Street Chicago, IL: Goodyear Publishing Company.

Mastrofski, S., and J. Greene. 1994. “Community Policing and the Rule of Law.” In A. Weisburd & C. Uchida (Eds.), The Changing Focus of Police Innovation: Problems of Law, Order, and Community. New York, NY: Doubleday.

McCluskey, J.D., and W. Terrill. 2005. “Departmental and Citizen Complaints as Predictors of Police Coercion.” Policing: An International Journal of Police Strategies and Management 28:513-529.

McCluskey, J., W. Terrill, and E. Paoline III. 2005. “Peer Group Aggressiveness and the Use of Coercion in Police-Suspect Encounters.” Police Practice and Research: An International Journal 6:19-37.

McElvain, J.P., and A.J. Kposowa. 2008. “Police Officer Characteristics and the Likelihood of Using Deadly Force.” Criminal Justice and Behavior 35:505-521.

McKenzie, R. 1923. The Neighborhood: A Study of Local Life in the City of Columbus, Ohio Chicago: The University of Chicago Press.

171

Mears, D.P., P.Y. Warren, A.N. Arnio, E.A. Stewart, and M.O. Craig. 2019. “A Legacy of Lynchings: Perceived Black Criminal Threat Among Whites.” Law & Society Review 53:487-517.

Messner, S.F., R.D. Baller, and M.P. Zevenbergen. 2005. “The Legacy of Lynching and Southern Homicide.” American Sociological Review 70:633-655.

Meyer, M.W. 1980. “Police Shootings of Minorities: The Case of Los Angeles.” Annals of the American Academy of Political and Social Science 452:98-110.

Morrow, W.J., M.D. White, and H.F. Fradella. 2017. “After the Stop: Exploring the Racial/Ethnic Disparities in Police Use of Force During Terry Stops.” Police Quarterly 220:367-396.

Muir, W.K., Jr. 1977. Police: Streetcorner Politicians. Chicago: University of Chicago Press.

Muller, C. 2012. “Northward Migration and the Rise of Racial Inequality in American Incarceration.” American Journal of Sociology 118:281-326.

National Institute of Justice. 5/21/19. Overview of Police Use of Force. U.S. Department of Justice, Office of Justice Programs. Washington, D.C.

National Institute of Justice. March 1994. Oleoresin Capsicum: Pepper Spray as a Force as a Force Alternative. U.S. Department of Justice, Office of Justice Programs. Washington, D.C.

New York State Division of Criminal Justice Services: Criminal Justice Statistics Retrieved from https://www.criminaljustice.ny.gov/crimnet/ojsa/stats.htm

Nix, J., and J.T. Pickett. 2017. “Third-Person Perceptions, Hostile Media Effects, and Policing: Developing a Theoretical Framework for Assessing the Ferguson Effect.” Journal of Criminal Justice 51:24-33.

Nix, J., and S.E. Wolfe. 2016. “Sensitivity to the Ferguson Effect: The Role of Organizational Justice.” Journal of Criminal Justice 47:12-20.

Nix, J., and S.E. Wolfe. 2017. “The Impact of Negative Publicity on Police Self-Legitimacy.” Justice Quarterly 34:84-108.

Oliver, W.M. 2015. “Depolicing: Rhetoric or Reality?” Criminal Justice Policy Review 28:437-461.

172

Paoline, Eugene A. III. 2003. “Toward a Richer Understanding of Police Culture.” Journal of Criminal Justice 31:199-214.

Paoline, E.A. 2004. “Shedding Light on Police Culture: An Examination of Officers’ Occupational Attitudes.” Police Quarterly 7:205-236.

Paoline III., E.A., and J.M. Gau. 2018. “Police Occupational Culture: Testing the Monolithic Model.” Justice Quarterly 35:670-698.

Paoline III, E., and W. Terrill. 2004. “Women Police Officers and the Use of Coercion.” Women and Criminal Justice 15:97-119.

Paoline III, E., and W. Terrill. 2007. “Police Education, Experience, and the Use of Force.” Criminal Justice and Behavior 34:179-196.

Paoline III., E.A., and W. Terrill. 2014. “Police Culture.” In G. Bruinsma & D. Weisburd (Eds.) Encyclopedia of Criminology and Criminal Justice (pp.81-88) New York, NY: Springer.

Papachristos, A.V. 2009. “Murder by Structure: Dominance Relations and the Social Structure of Gang Homicide.” American Journal of Sociology 115:74-128.

Parker, K.F., J.M. MacDonald, G. P. Alpert, M.R. Smith, and A.R. Piquero. 2004. “A Contextual Study of Racial Profiling.” American Behavioral Scientist 47:943-962.

Parker, K.F., J.M. MacDonald, W.G. Jennings, and G.P. Alpert. 2005. “Racial Threat, Urban Conditions and Police Use of Force: Assessing the Direct and Indirect Linkages Across Multiple Urban Areas.” Justice Research and Policy 7:53-79.

Peffley, M., and J. Hurwitz. 1998. “ and Resistance: Race and the Death Penalty in America.” American Journal of Political Science 51:996-1012.

Peffley, M., T. Shields, and B. Williams. 1996. “The Intersection of Race and Crime in Television News Stories: An Experimental Study.” Political Communication 13:309-327.

People v. Goetz, 68 N.Y.2d 96 (1986).

Peterson, R.D., and L.J. Krivo. 1999. “Racial Segregation, the Concentration of Disadvantage, and Black and White Homicide Victimization.” Sociological Forum 14:465-493.

Peterson, R.D., and L.J. Krivo. 2010. Divergent Social Worlds: Neighborhood Crime and the Racial-Spatial Divide. Russell Sage Foundation: New York, NY.

173

Peterson, R.D., L.J. Krivo, and M.A. Harris. 2000. “Disadvantage and Neighborhood Violent Crime: Do Local Institutions Matter?” Journal of Research in Crime and Delinquency 37:31-63.

Phillips, S. 2009. “Status Disparities in the Capital of Capital Punishment.” Law & Society Review 47:803-838.

Plant, E.A., and B.M. Peruche. 2005. “The Consequences of Race for Police Officers’ Responses to Criminal Suspects.” Psychological Science 16:180-183.

Portes, A., and J. Sensenbrenner. 1993. “Embeddedness and Immigration: Notes on the Social Determinants of Economic Action.” American Journal of Sociology 98:1320-1350.

Potter, G. 2013. The History of Policing in the United States. EKU Online, retrieved 10/5/19 from https://plsonline.eku.edu/insidelook/history-policing-united-states-part-1

Police Use of Force: An Examination of Modern Policing Practices. 2018. U.S. Commission on Civil Rights, Washington, D.C.

President’s Advisory Board. 1998. One America in the 21st Century. Washington, D.C.: President’s Initiative on Race.

President’s Commission on Law Enforcement and Administration of Justice. 1967. Task Force Report: The Police Washington, D.C. U.S. Government Printing Office.

Puckett, J.L., and R.L. Lundman. 2003. “Factors Affecting Homicide Clearances: Multivariate Analysis of a More Conceptual Framework.” Journal of Research in Crime & Delinquency 40:171-193.

Pyrooz, D.C., S.H. Decker, S.E. Wolfe, and J.A. Shjarback. 2016. “Was There a Ferguson Effect on Crime Rates in Large U.S. Cities?” Journal of Criminal Justice 46:1-8.

Quillian, L. 2006. “New Approaches to Understanding Racial Prejudice and Discrimination.” Annual Review of Sociology 32:299-328.

Quillian, L., and D. Pager. 2001. “Black Neighbors, Higher Crime? The Role of Racial Stereotypes in Evaluations of Neighborhood Crime.” American Journal of Sociology 107:717-767.

Reisig, M.D., and R.B. Parks. 2004. “Can Community Policing Help the Truly Disadvantaged?” Crime & Delinquency 50:139-167.

Rochester Police Department, Defensive Tactics Instructor Manual for Public Safety Officers. 2005.

174

Rochester Police Department, General Order 335 effective 7/12/17, retrieved 10/2/19 from the RPD Open Data Portal, https://data-rpdny.opendata.arcgis.com/datasets/go-335-subject-resistance-report

Rochester Police Department, General Order 340 effective 8/29/17, retrieved 9/30/19 from the RPD Open Data Portal, https://data-rpdny.opendata.arcgis.com/datasets/go-340-deadly-physical-force

Rochester Police Department, Standard Operating Procedure. 6/29/17. Taser Conducted Electrical Weapons (CEW).

Rydberg, J., and W. Terrill. 2010. “The Effect of Higher Education on Police Behavior.” Police Quarterly 13:92:120.

Sampson, R.J., and D. Jeglum Bartusch. 1998. “Legal Cynicism and (Subcultural?) Tolerance of Deviance: The Neighborhood Context of Racial Differences.” Law and Society Review 32:777-804.

Sampson, R.J., and L. Bean. 2006. “Cultural Mechanisms and Killing Fields: A Revised Theory of Community-Level Racial Inequality.” In R. Peterson, L. Krivo, and J. Hagan (Eds.), The Many Colors of Crime (pp. 8-36). New York, N.Y.: NYU Press.

Sampson, R.J., and W.B. Groves. 1989. “Community Structure and Crime: Testing Social-Disorganization Theory.” American Journal of Sociology 94:774-802.

Sampson, R.J., S.W. Raudenbush, and F. Earls. 1997. “Neighborhoods and Violent Crime: A Multilevel Study of Collective Efficacy.” Science 277:918-924.

Scott v. Harris, 550 U.S. 372 (2007)

Shjarback, J.A., D.C. Pyrooz, S.E. Wolfe, and S.H. Decker. 2017. “De-policing and Crime in the Wake of Ferguson: Racialized Changes in the Quantity and Quality of Policing Among Missouri Police Departments.” Journal of Criminal Justice 50:42-52.

Sherman, L.W. 1983. “Reducing Police Gun Use: Critical Events, Administrative Policy, and Organizational Change.” In Maurice Punch (ed.) Control in the Police Organization. Cambridge, MA: MIT Press.

Sherman, L. W. 1993. “Defiance, Deterrence, and Irrelevance: A Theory of the Criminal Sanction.” Journal of Research in Crime & Delinquency 30:445-473.

Sherman, L.W. and M. Blumberg. 1981. “Higher Education and Police Use of Deadly Force.” Journal of Criminal Justice 9:317-331.

175

Shuck, A.M. 2004. “The Masking of Racial and Ethnic Disparity in Police Use of Physical Force: The Effects of Gender and Custody Status.” Journal of Criminal Justice 32:557-564.

Sierra-Arévalo, M. 2019. “Technological Innovation and Police Officers’ Understanding and Use of Force.” Law & Society Review 53:420-451.

Silver, Allan. 1967. “The Demand for Order in Civil Society. A Review of Some Themes in The History of Urban Crime.” (pp. 1-24). In David Bordua (ed.), The Police: Six Sociological Essays. New York: John Wiley.

Skolnick, J.H., and J.J. Fyfe. 1993. Above the Law: Police and the Excessive Use of Force. New York, NY: Free Press.

Smith, D.A. 1986. “The Neighborhood Context of Police Behavior.” In A.J. Reiss, Jr. and M. Tonry, eds. Community and Crime Chicago: University of Chicago Press, 313-341.

Smith, M.R., and G.P. Alpert. 2007. “Explaining Police Bias: A Theory of Social Conditioning and Illusory Correlation.” Criminal Justice and Behavior 34:1262-1283.

Smith, B.W., and M.D. Holmes. 2003. “Community Accountability, Minority Threat, and Police Brutality: An Examination of Civil Rights Criminal Complaints.” Criminology 41:1035-1063.

Smith, B.W., and M.D. Holmes. 2014. “Police Use of Excessive Force in Minority Communities: A Test of the Minority Threat, Place, and Community Accountability Hypotheses.” Social Problems 61:83-104.

Sperling, J. 2012. “The Tyranny of Census Geography: Small-Area Data and Neighborhood Statistics.” Cityscape: A Journal of Policy Development and Research 14: 219-223.

Steen, S., R.L. Engen, and R.R. Gainey. 2005. “Images of Danger and Culpability: Racial Stereotyping, Case Processing, and Criminal Sentencing.” Criminology43:435-468.

Steffensmeier, D., J. Ulmer, and J. Kramer. 1998. “The Interaction of Race, Gender, and Age in Criminal Sentencing: The Punishment Cost of Being Young, Black, and Male.” Criminology 36:763-798.

Sun, I.Y., and B.K. Payne. 2004. “Racial Differences in Resolving Conflicts: A Comparison Between Black and White Police Officers.” Crime and Delinquency 50:516-541.

176

Sun, I.Y., B.K. Payne, and Y. Wu. 2008. “The Impact of Situational Factors, Officer Characteristics, and Neighborhood Context on Police Behavior: A Multilevel Analysis.” Journal of Criminal Justice 36:22-32.

Tankebe, J., and G. Mesko. 2015. “Police Self-Legitimacy, Use of Force, and Pro-Organizational Behavior in Slovenia.” In G. Mesko & J. Tankebe (Eds.), Trust and Legitimacy in Criminal Justice (pp. 261 – 270). New York, NY: Springer.

Tennessee v. Garner, 471 U.S. 1 (1985).

Terrill, W. 2001. Police Coercion: Application of the Force Continuum. New York: LFB Scholarly.

Terrill, W. 2003. “Police Use of Force and Suspect Resistance: The Micro-Process of the Police-Suspect Encounter.” Police Quarterly 6:51-83.

Terrill, W. 2005. “Police Use of Force: A Transactional Approach.” Justice Quarterly 22:107-138.

Terrill, W., F. Leinfelt, & D. Kwak. 2008. “Examining Police Use of Force: A Smaller Agency Perspective.” Policing: An International Journal of Police Strategies & Management 31:57-76.

Terrill, W., and S. Mastrofski. 2002. “Situational and Officer Based Determinants of Police Coercion.” Justice Quarterly 19:215-248.

Terrill, W., E. Paoline III, and P. Manning. 2003. “Police Culture and Coercion.” Criminology 41:1003-1034.

Terrill, W., and M. Reisig. 2003. “Neighborhood Context and Police Use of Force.” Journal of Research in Crime and Delinquency 40:291-321.

Trinkner, R., E.M. Kerrison, and P. Atiba Goff. 2019. “The Force of Fear: Police Stereotype, Threat, Self-Legitimacy, and Support for Excessive Force.” Law and Human Behavior 43:421-435.

Tyler, T.R. 1990. Why People Obey the Law: Procedural Justice, Legitimacy, and Compliance. New Haven, CT: Yale University Press.

U.S. Census Bureau. 2013. American Community Survey (ASC). Retrieved from https://www.census.gov/programs-surveys/acs/guidance/comparing- acsdata/2013.html.

U.S. Census Bureau, Quick Facts, 2010, 2018. Retrieved from https://www.census.gov/quickfacts/

177

U.S. Commission on Civil Rights. 1981. Who’s Guarding the Guardians? A Report on Police Practices: Washington, D.C.: Commission.

Vargas, R. 2014. Wounded City: Violent Turf Wars in a Chicago Barrio. New York: Oxford University Press.

Weisbuch, M., K. Pauker, and N. Ambady. 2009. “The Subtle Transmission of Race Bias via Televised Nonverbal Behavior.” Science 326:1711-1714.

Weitzer, R. 2000. “White, Black, or Blue Cops? Race and Citizen Assessments of Police Officers.” Journal of Criminal Justice 28:313-324.

Weitzer, R. 2015. “American Policing Under Fire: Misconduct and Reform.” Society 52:475-480.

Walker, S. 1993. Taming the System – The Control of Discretion in Criminal Justice, 1950-1990. New York, NY: Oxford University Press.

Welch, K. 2007. “Black Criminal Stereotypes and Racial Profiling.” Journal of Contemporary Criminal Justice 23:276-288.

Werthman, C., and I. Piliavin. 1967. “Gang Members and the Police.” (pp. 56-98). In In David Bordua (ed.), The Police: Six Sociological Essays. New York: John Wiley.

White, M.D. 2001. “Controlling Police Decisions to Use Deadly Force: Reexamining the Importance of Administrative Policy.” Crime & Delinquency 47:131-151.

White, M.D. 2002. “Identifying Situational Predictors of Police Shooting Using Multivariate Analysis.” Policing: An International Journal of Police Strategies & Management 25:728-754

Whyte, W.F. 1943. Street Corner Society, Chicago: University of Chicago Press.

Wilson, O.W. 1968. Varieties of Police Behavior: The Management of Law and Order in Right Communities. Cambridge, MA: Press.

Wilson, W.J. 1987. The Truly Disadvantaged: The Inner City, the Underclass, and Public Policy. Chicago: University of Chicago Press.

Wolfe, S.E., and J. Nix. 2016. “The Alleged ‘Ferguson Effect and Police Willingness to Engage in Community Partnership.” Law and Human Behavior 40:1-10.

Wolfe, S.E., and A.R. Piquero. 2011. “Organizational Justice and Police Misconduct.” Criminal Justice and Behavior 38:332-353.

178

Worden, R.E. 1995. “The ‘Causes’ of Police Brutality: Theory and Evidence on Police Use of Force.” In W.A. Geller & H. Toch (Eds.). And Justice for All: Understanding And Controlling Police Abuse of Force, (pp.31-60). Washington, DC: Police Executive Research Forum.

Worden, R.E., and S.E. Catlin. 2002. “The Use and Abuse of Force by Police.” In K. M. Lersch, ed. Policing and Misconduct. Upper Saddle River, N.J.: Prentice Hall, 85-120.

179

Endnotes

1 All references to Article 35 of the New York State Penal Law were obtained online from FindLaw, retrieved 9/30/19 from https://codes.findlaw.com/ny/penal-law/pen

2 Although Article 35 consistently uses the phrase “deadly physical force,” the study hereinafter uses the shorthand “deadly force.”

3 “Pepper spray” is a generic name for what is known as “Oleoresin Capsicum spray,” or “OC.” OC is naturally occurring, found in the oily resin of cayenne and other varieties of hot peppers. Contact with OC particles, whether sprayed in a mist or stream, usually (but not always) induces a burning sensation of the skin, eyes, and respiratory tract (if it is inhaled) (NIJ, 1994). If effective, OC causes the subject to focus on the immediate discomfort they are feeling and cease fighting or resisting. The effects of OC are temporary and are alleviated by the application of cold water to the face and eyes. The generic name pepper spray will be used throughout the remainder of this study.

4 Black’s theory is not without its critics. For example, Greenberg (1983) argues that the logic of Black’s theory is incoherent and empirical evidence does not support his propositions (see also Gottfredson & Hindelang, 1979).

5 The facts detailing the shooting death of Michael Brown are taken from the Department of Justice’s official report on the shooting dated March 4, 2015.

6 These include the deaths of Freddie Gray in Baltimore, Tamir Rice in Cleveland, Laquan McDonald in Chicago, Walter Scott in North Charleston, S.C., Philandro Castile in St. Anthony Village, MN, and Eric Garner in New York City. These incidents occurred between October 20, 2014 and July 6, 2016, or within less than two years after Michael Brown’s death. All were widely reported nationally.

7 News coverage of the immediate aftermath of Officer Pierson’s shooting similarly reported that RPD officers established a vigil at Rochester General Hospital as they awaited word on his condition. See, e.g., Democrat & Chronicle, Deadly officer shooting: Updates as it happened, Published 10:14 p.m. ET Sept. 3, 2014, retrieved from https://www.democratandchronicle.com/story/news/2014/09/03/breaking-police-investigate- shooting/15050817/

8 A New York Times story reporting the funeral of slain Jersey City, New Jersey Detective Joseph Seals commented on the heightened awareness of danger that occurs after the death of a colleague, as reflected in this quote from an officer attending the funeral: “We’re here to pay respects,” said Lt. Frank Todd of the Edison Police Department in New Jersey. “This is something we all face. Any day, it could be any one of us.” New York Times, 12/17/19, “’It Could Be Any One of Us:’ Hundreds Mourn Slain Jersey City Officer.”

180

9 A “hobble” is a restraint device used to secure a kicking subject’s legs. A “spit sock” is a mesh hood that is placed over a spitting subject’s head. The spit sock allows the subject to see and breathe, but restrains his ability to spit.

10 A “car beat” is a patrol area to which one officer is assigned each shift. Each car beat, on average, has the same number of 911 calls for service regardless of its geographic size. Thus, a large car beat that is spread out over a less populated area will typically receive the same number of 911 calls for service as a much smaller car beat in a densely populated area.

11 For comparison purposes, models were estimated using age 26 (the median age) and age 30 as the demarcation points. Estimating models with these categorical age variables did not change the results.

12 As reported in FBI, Uniform Crime Reporting Statistics, https://www.ucrdatatool.gov/offenses.cfm

13 Brown was shot just after 12:00 p.m. on August 9th, which was before officers working 4:00 p.m. to 11:59 p.m. began their shift. Although social media and mainstream news outlets began reporting Brown’s shooting that afternoon (Nix & Pickett, 2017), it is highly likely that this news had no immediate effect on the behavior of officers starting work only three hours after the event. These officers would also not be watching television or listening to commercial radio during the eight hours they were on duty. However, officers starting work at midnight or later on August 10th may have been exposed to more media coverage and thus a one-day lag in measuring the effect of Brown’s death is appropriate.

14 A model with just 2014 data was run for comparison purposes but would not converge.

15 For Model 5, the unit of analysis is the census tract. The analysis is based on 3,368 use of force incidents nested within 78 census tracts.

16 See, e.g., “Brockport Fundraiser Organized for Pierson Family,” Westside News & Greece News, westsidenewsny.com, November 17, 2014; “Pierson Family Fundraiser Continues,” Democrat & Chronicle, democratandchronicle.com, January 2, 2015; “Fundraiser for Officer Pierson’s Family to Attend Washington, D.C. Ceremony,” Spectrum News, spectrumlocalnews.com, April 21, 2015.

17 See, e.g., “Vigil to Mark Anniversary of Officer Pierson’s Death,” Democrat & Chronicle, democratandchronicle.com, August 31, 2015; “Memorial Painting Unveiled in Honor of Officer Daryl Pierson,” RochesterFirst.com, September 2, 2015.

18 “Obama Signs Bill for East Rochester Post Office to be Named After Fallen Officer Pierson,” Spectrum News, spectrumnewsny.com, November 6, 2015.

181

19 Facts regarding the Treyvon Martin shooting were taken from “Trayvon Martin Fast Facts,” CNN, February 20, 2020, retrieved from https://www.cnn.com/2013/06/05/us/trayvon-martin-shooting-fast-facts/index.html

20 Facts are taken from “White Father, Son, Charged with Murder in Ahmaud Arbery Case,” ABC News, May 8, 2020, retrieved from https://abcnews.go.com/US/wireStory/father-son-charged-killing-black-man-ahmaud-arbery- 70575513

21 Daily Mail.com, May 12, 2020, retrieved from https://www.dailymail.co.uk/news/article- 8302641/Attorney-leaked-Ahmaud-Arbery-lynching-video-thought-CLEAR-friends.html

22 “Can Cops Unlearn Their Unconscious Biases?” The Atlantic, 12/23/17, retrieved from https://www.theatlantic.com/politics/archive/2017/12/implicit-bias-training-salt-lake/548996/

23 “Confronting Implicit Bias in the New York Police Department.” New York Times, 7/15/18, retrieved from https://www.nytimes.com/2018/07/15/nyregion/bias-training- police.html

24 Note that the tasers carried by officers record and retain data generated from every use. This includes how many times the taser was used and the duration of each use. This data is required to be downloaded and provided for review. It is thus highly unlikely that the actual use of a taser would go unreported.

182