needing help with this work

Journal of Social Issues, Vol. 73, No. 4, 2017, pp. 744--767 doi: 10.1111/josi.12246

This article is part of the Special Issue “What Social Science Research Says About Police Violence Against Racial and Ethnic Minorities: Understanding the Antecedents and Consequences,” Kristin N. Dukes, and Kimberly B. Kahn (Special Issue Editors). For a full listing of Special Issue papers, see: http://onlinelibrary.wiley.com/doi/10.1111/josi.2017.73.issue-4/issuetoc.

Understanding Racial Disparities in Police Use of Lethal Force: Lessons from Fatal Police-on-Police Shootings

Amanda Charbonneau∗, Katherine Spencer, and Jack Glaser University of California, Berkeley

During a 29-year period studied by a government task force, 10 off-duty police officers were found to have been mistaken for civilians and fatally shot by another police officer. Eight of these officers were Black, one was Hispanic, and one was White. Given that at least 75% of U.S. police officers in this period were White, we estimate that there is a roughly one in one million chance that this disparity reflects a random deviation from a condition in which Black and White officers faced the same risk of being fatally shot by another officer while off duty. Estimates of the magnitude of this racial disparity must be interpreted cautiously, but the increased risk faced by Black officers while off duty compared to their White counterparts appears to be even larger than the racial disparity among civilians killed by police officers. The disparity is much less pronounced in mistaken-identity fatal shootings of on-duty officers, of which two were Black, one was Hispanic, and 12 were White. These incidents are rare, but they comprise an important subset of all police interactions because they are known errors that involve a misperception of threat, and because the differential patterns of racial disparities suggest that there are situational factors that vary systematically and contribute to the observed outcomes. We examine fatal, mistaken identity police-on-police shootings and explore potential explanations for the dramatic racial disparity among officers killed while off duty.

Unwarranted fatal shootings by police are rare occurrences, but they are dev- astating to the affected victims, families, communities, officers, and departments. When the victim is from a minority group, which happens disproportionately, the

∗Correspondence concerning this article should be addressed to Amanda Charbonneau, Goldman School of Public Policy, University of California, Berkeley, CA 94720 [email: [email protected]].

The authors are grateful to Christine Haworth for research assistance.

744

C© 2017 The Society for the Psychological Study of Social Issues

Police-on-Police Shootings 745

effect is particularly destabilizing because the event may exacerbate longstand- ing distrust between law enforcement and non-White communities. Data on fatal shootings of off-duty police officers by on-duty officers provide a striking example of racial disparities in the use of lethal force; an example that might shed light on the causes of the disparities we see in cases with civilian victims. In 2010, The New York State Task Force on Police-on-Police Shootings (hereafter, “task force”) identified a troubling trend among the rare, but tragic fatalities that occurred when police officers mistook other officers for crime perpetrators and fatally shot them. Among the 10 off-duty officers who were found to have met this fate during the period studied (1981–2009), eight were Black, one was Hispanic,1 and one was White (Stone, Carter, & Belfiore, 2010). This disparity is much less pronounced in fatal shootings of on-duty officers, of which two were Black, one was Hispanic, and 12 were White (Stone et al., 2010).2

Although they are a very small subset of all police interactions, police-on- police shootings can offer some insight into police use-of-force more generally because we know with near certainty that all 26 fatalities over this time period were errors (i.e., the confronting officer’s perception of threat from the victim was wrong); if a victim officer managed to successfully identify him- or herself, the confronting officer would not deliberately shoot the victim.3 All of the cases involved the misidentification of the victim as a civilian offender, followed by a deliberate shooting. In other words, this is a unique subset of cases where at least one specific error (the misidentification) is a known factor in the incorrect decision to shoot. In addition, the difference between the racial disparities in the on- and off-duty fatalities suggests that there may be other important situational factors affecting the observed outcomes. In sum, this is a small, but very informative set of cases that holds promise to yield considerable understanding of the causes and potential solutions to wrongful officer-involved shootings and the racial disparities therein. We extend the task force’s work by estimating the probabilistic magnitude of the racial disparity in fatal shootings of off-duty officers and by comparing cases along the dimension of tactical circumstances (i.e., planned operations vs. crimes in progress). Considering tactical circumstances further highlights the role of ambiguity as a potential driver of racial disparities. The pattern of events revealed in these cases raises questions: Why is there such a dramatic racial disparity

1 We use the same labels as the task force report, which treats “Hispanic” as a distinct racial category. We note some of the limitations associated with these labels in the Data subsection of the limitations section.

2 The duty status of one victim officer, a White police chief who was assisting another agency in a planned operation, is unknown.

3 Note that “friendly fire” and crossfire incidents are accidental, do not involve a mistaken identity, and are not included here or in the task force report.

746 Charbonneau, Spencer, and Glaser

among mistaken identity fatalities for off-duty officers while the disparity is much smaller among those killed while on duty? How do these disparities compare and relate to the racial differences among civilians killed by police officers? Given the limitations of the existing data, we can only begin to explore these questions. However, this is an important avenue of inquiry not only because of the gravity of these events, but because the patterns among these rare events may offer insight into the broader phenomenon of disparate outcomes found in nonlethal force and other policing decisions (e.g., stop, search, and arrest).

We begin by describing the 26 fatal police-on-police shootings that were doc- umented by the task force for the period 1981–2009, then estimate the likelihood that the observed racial disparity among officers killed by other officers while off duty would have occurred by chance if Black and White officers in fact faced the same risk of being involved in and killed during these encounters, and describe the limitations associated with these analyses. We leverage the unique characteristics of the population and situations to explore some of the possible explanations for racial disparities in police use of force generally and their specific relevance to off-duty officers. Finally, we describe policies and procedures that might reduce police shootings and use of force generally, as well as racial disparities among these events.

Background

Racial Disparities in Police Use-of-Force

As is true in many other domains, racial disparities are evident in a number of law enforcement outcomes (for a review, see Kahn & Martin, 2016), including police stops and searches (see, e.g., Glaser, 2015; Jones-Brown, Gill, & Trone, 2010), and in the use of force (see, e.g., Goff, Lloyd, Geller, Raphael, & Glaser, 2016). These patterns are largely consistent with the disparities observed in fatal shootings of civilians by police officers, which have been documented and thor- oughly investigated in recent years by The Washington Post (2017). While the civilians armed with lethal weapons who were fatally shot by police in 2015 and 2016 were predominantly White, unarmed civilians fatally shot by police during that period were more likely to be Black. Figure 1 compares the proportions of unarmed civilians that were identified as belonging to several major racial and ethnic categories, compared to those that were armed with lethal weapons.

Among the unarmed civilians fatally shot by police over these 2 years, 39% were Black and 38% were White. This is strikingly different from those who were armed with a lethal weapon, among whom 25% were Black and 51% were White. Because Black people represent only about 12% of the U.S. population, they are overrepresented among fatal police shooting victims in general, but especially so among unarmed victims. We cannot necessarily infer that the disproportion

Police-on-Police Shootings 747

0%

10%

20%

30%

40%

50%

60%

White Black Hispanic Other Unknown

Unarmed (n = Lethal Weapon (n =

= 142)

= 1,384)

Fig. 1. Racial/Ethnic breakdown of unarmed versus lethally armed civilians who were fatally shot by police in the United States, 2015–2016. Source: The Washington Post (2017). Notes: The “Lethal Weapon” category includes civilians who were armed with guns, knives, and similar weapons. It does not include civilians who were armed with vehicles, toy weapons, or makeshift weapons, nor does it include civilians for whom it is unknown whether they were armed. A total of 428 fatal shootings are therefore not included.

relative to population is the result of racial bias (it could be contributed to by other factors), but that the disparity is greater among unarmed victims relative to armed victims is compelling evidence that victim race directly influences the use of deadly force. Figure 2 focuses on the proportion of unarmed civilians among all civilians fatally shot by police, by racial and ethnic category and by year.

Both the number of unarmed civilians shot by police and the racial disparities among them appear to be declining. As we discuss further below, it is often, though not always, the case that racial disparities decline when the total number of events is reduced.

One analysis of the 2015 Washington Post data found that the racial disparity among unarmed civilians remained after statistically controlling for several fairly general variables, including U.S. regional crime rates and approximate size of the city and agency (Nix, Campbell, Byers, & Alpert, 2017), but much more research is needed to reveal the causal determinants of these disparities. ProPublica’s analysis of fatal officer-involved shootings of civilians indicates that young Black males are 21 times as likely to be fatally shot by police relative to their White peers (Gabrielson, Jones, & Sagara, 2014). This differential is so much higher than those that we describe in the Washington Post data because we compare the race/ethnicity only among civilians who were fatally shot, while the ProPublica analysis compares the racial distribution of those fatally shot to that of all young, male Black and White civilians, encompassing all of the compounding influences

748 Charbonneau, Spencer, and Glaser

n

0%

2%

4%

6%

8%

10%

12%

14%

16%

= 495

n = 465

White

n = 258

n = 233

Black

n = 172

n = 160

Hispanic

n = 38

n = 42

Other

n = 28 n = 63

Unknowwn

201

201

15

16

Fig. 2. Percent of each racial/ethnic group of fatal police shooting victims that were unarmed, by year. Source: The Washington Post (2017). Notes: The n provided over each bar represents the total number of victims from that racial/ethnic group in that year. A similar pattern of fewer shootings of unarmed civilians accompanied by reductions in racial disparities appears to be holding steady in the first half of 2017, but we do not report those data here because a relatively higher proportion of civilians is not yet categorized by race.

that affect the likelihood of the outcome (e.g., rates of initial contact with the police). The multiple and interacting determinants of these troubling patterns are difficult to study because policing decisions occur within a broader social context rife with racial disparities and because the number and racial distributions of peo- ple in the population who pose a true and immediate threat to the lives of officers or other citizens are unknown.

Mistaken identity shootings of law enforcement officers by other officers are a very small subset of all incidents involving the use of force by police officers, and interactions involving force are a small subset of all encounters between officers and civilians. Nevertheless, these cases offer an opportunity for gaining understanding of the disparities seen in the broader sets of encounter types. The misidentification of the victim officer, and resulting misperception of the level of threat posed by the victim officer is not entirely surprising. Only one victim officer was in uniform,4 nearly all were wielding weapons, the situations required quick decisions under conditions of significant ambiguity, and many of the victim officers reportedly failed to comply with the confronting officer’s verbal commands. As the task force pointed out, some of the lessons learned from studying the circumstances of these tragic shootings, and from comparing the on- and off-duty cases, may extend to police–civilian encounters. The next

4 The victim officer in uniform was leading a SWAT team and detonated a smoke grenade just before he was mistaken for a suspect and killed by a supervisor from the same department.

Police-on-Police Shootings 749

Table 1. Fatal, Mistaken-Identity Police-on-Police Shootings, 1981–2009

Victim Officer Race/Ethnicity

Victim officer status Black Hispanic White Total

On duty 2 (13.3%) 1 (6.7%) 12 (80%) 15 (100%) Off duty 8 (80%) 1 (10%) 1 (10%) 10 (100%) Unknown 0 0 1 (100%) 1 (100%)

Source: Stone et al. (2010).

section describes the cases identified by the task force in order to afford a deeper understanding of their nature before turning to discussion of the role of race and the relevance to cases with civilian victims.

Fatal, Mistaken Identity Police-on-Police Shootings, 1981–2009

The task force’s final report (Stone et al., 2010) identifies 26 incidents in which an officer was fatally shot by another police officer due to mistaken identity from 1981 through 2009. These cases, the details of which are summarized in the Supporting Information, Appendix A, appear in the Law Enforcement Officers Killed or Assaulted (LEOKA) data set maintained by the Federal Bureau of Investigation’s Uniform Crime Reporting program, with the exception of three additional cases that were identified by the task force through their review of media reports. Table 1 summarizes these fatalities by duty status and race of the victim officer.

In all of these fatal shootings, the confronting officer mistook the victim officer for a civilian. In 24 of the 26 cases, the victim officer was in possession of a drawn firearm during the incident, and in one case the officer was holding knives he had confiscated from a civilian. In the remaining case, the officer was driving his vehicle and presumed by the confronting officer to be an armed suspect. Most (62%) of the mistaken-identity shooting fatalities involved victim officers who were on duty during the confrontation, which may not be surprising, given the greater likelihood that an officer will be armed and in a situation (e.g., on a planned operation, called to a crime scene) to draw a weapon while serving in an official capacity.

On-Duty Victim Officers

The victim officer was serving in plainclothes or undercover in nearly all of the 15 on-duty cases,5 and was participating in a planned team operation in 11

5 One White, on-duty officer was in uniform and shot after the detonation of a smoke grenade, and one White police chief was assisting another agency in a planned operation but whether he was on

750 Charbonneau, Spencer, and Glaser

(73%) of the cases (in the other five cases the victim officer came upon a crime- in-progress while on duty). In six (40%) of the mistaken-identity shootings of on-duty officers, the victim and confronting officers served in different agencies. Ten (66.7%) of the incidents are known to have involved drug law enforcement. The rate of occurrence of these incidents declines over the time period examined.

Off-Duty Victim Officers

Among the 10 off-duty officers who were mistakenly killed by other officers during this time period, five were coming to the assistance of civilians or other officers during a crime-in-progress, three were involved in disputes, and two were victims of crimes. In three cases, the victim and confronting officers served different agencies. Eight of the victim officers were Black, one was Hispanic, and one was White. We calculate estimates of the likelihood of observing a racial disparity of this magnitude due to chance in the next section.

Likelihood Estimates

Risk Disparity

Eight of the 10 off-duty officers who were misidentified and fatally shot by other officers between 1981 and 2009 were Black (Stone et al., 2010). To get a sense of proportionality, we sought statistics on the demographics of U.S. police in the relevant period. Among all police officers in 1987 and 2007,6 respectively, 9.3% and 11.9% were Black (Bureau of Justice Statistics, 2008). Figure 3 compares the race/ethnicity of all officers in 2007 to those of the victim officers.

The racial distribution of officers who were fatally shot while on duty is simi- lar to the racial distribution of all officers in 2007. Based on the more conservative 2007 benchmark, we estimate that the likelihood of a Black off-duty officer en- countering, being misidentified as a civilian offender, and being fatally shot by another officer was more than 50 times as high as the likelihood that a White off-duty officer would meet the same circumstances and fate (see the Supporting Information, Appendix B, I-V for details and sensitivity analyses).

What is the likelihood that a disparity of this magnitude would occur by chance if Black and White officers faced the same level of risk while off duty?

duty is unknown. These are not considered “friendly fire” or crossfire cases because the confronting officer is reported to have mistaken the victim officer for a civilian and deliberately fired at the victim officer.

6 1987 and 2007 are the years closest to the end points of the period documented by the task force (1981–2009) for which data on officer race and ethnicity are available from the Bureau of Justice Statistics.

Police-on-Police Shootings 751

0% 10% 20% 30% 40% 50% 60% 70% 80% 90%

100%

All Officers 2007

On-Duty Fatalities

1981–2009

Off-Duty Fatalities

1981–2009

Other/Multiple Races

Black

Hispanic and/or Latino

White

Fig. 3. Officer race/ethnicity among all officers (in 2007) compared to on-duty and off-duty fatalities. Sources: Bureau of Justice Statistics (2008), Stone et al. (2010).

Simple probability estimates provide a blunt answer to this question. These 10 deaths constitute the entire known population of occurrences between 1981 and 2009 (Federal Bureau of Investigation, 2013; Stone et al., 2010), so concerns about sampling variation and statistical tests for the extrapolation from a sample to a population (e.g., null hypothesis significance tests) do not apply. This analysis assumes that the demographic distribution of officers who were off duty was, on average, similar to that of all sworn officers as represented by the BJS census data. We use 2007 data as a benchmark, which most closely represents the end of the time period examined by the task force. A smaller (larger) proportion of officers serving in 2007 were White (Black) than during most of the 1981–2009 period. For example, while 75% of police officers were White in 2007, 86% were White in 1987 (Bureau of Justice Statistics, 2008). Consequently, the probabilities provided here represent a conservative estimate of the likelihood that the disparity is not due to chance.

In a world where the underlying binomial probability distribution was the same for officers of all races, the likelihood of 8 or more of the ten officers killed being Black is less than two in one million (0.0000014). In order for this likelihood to reach 0.05 (the conventional statistical threshold for an “unlikely” outcome), there would have to be roughly 25 unreported cases of off-duty White officers (and no additional unreported cases with Black victims) being fatally shot by on-duty officers added. It is therefore extremely unlikely that Black and White officers face the same risk (or underlying probability) while off duty. These estimates describe the improbability of observing a disparity of this magnitude or greater under conditions where Black and White officers in fact face the same level of risk of encountering, being misidentified as a civilian offender, and fatally shot by another officer. They do not account for the many potential causes of the observed

752 Charbonneau, Spencer, and Glaser

disparity, which we consider in the discussion section below. Regardless of its causes, the racial disparity is massive under any reasonable assumption about the racial composition of law enforcement agencies during this period.

Other Patterns

Of the 15 officers mistaken for civilian offenders and fatally shot by other offi- cers while on duty, 2 were Black, 1 was Hispanic, and 12 were White. Compared to officers killed while off duty, the racial and ethnic distributions of the officers killed while on duty are much closer to those of police officers generally. However, we are not aware of existing data on the race and ethnicity of officers serving undercover or in plainclothes that could serve to generate a benchmark for exploring whether there was a racial disparity in on-duty fatalities. While, for the purposes of the preceding analyses, we were not able to identify reasons for concern that the racial distribution of police officers overall would be representative of officers when they are off duty, there are plausible reasons to question whether officers on plainclothes assignments would be racially distributed in the same manner. For example, plain- clothes assignments may be more common in particular ranks, and race is not proportionally distributed across ranks. As a consequence, inferences about the racial proportionality of on-duty victims need to be made with particular caution. There are, nevertheless, a few patterns among the situational factors worth noting.

As noted above, most of the officers who were killed while on duty were participating in planned team operations (e.g., drug raids), but five cases involved officers responding to a crime in progress. Two of the three non-White officers who were killed while on duty were responding to crimes in progress. In other words, during planned team operations, the racial distribution of officer victims closely reflects the racial distribution of the officers in the United States. During the more spontaneous situations of crimes and disputes7 in progress, non-White officers are overrepresented among victims. This distinction may be informative with regard to both racial disparities and potential policy implications. Table 2 summarizes all of the mistaken identity fatalities along this dimension as well as duty status of the victim officer.

Separating the mistaken identity fatalities into these two groups of tactical circumstances (planned operations vs. events in progress) as we have done in Table 2 directs our attention to contextual factors, other than the duty status of the victim officer, that could be important. Planned team operations involve an organized tactical objective in a selected location, and the team may have

7 Three officers (two Black officers and one Hispanic officer) were involved in disputes when they were killed. One involved an armed civilian, the second involved a struggle for the officer’s weapon, and in the third case the victim officer was assaulted, mistook a bystander for his assailant, and was killed by the confronting officer while holding the bystander at gunpoint.

Police-on-Police Shootings 753

Table 2. Mistaken Identity Fatalities by Tactical Circumstances, 1981–2009

Planned team operations Crimes and disputes in progress

Black Hispanic White Black Hispanic White Total

On duty 1 0 10 1 1 2 15 Off duty 0 0 0 8 1 1 10 Unknown 0 0 0 0 0 1 1 Total 1 0 10 9 2 4 26

information about the presence of undercover and plainclothes officers. Most of the on-duty fatalities occurred as part of these planned operations. In contrast, the times and locations of crimes and disputes are inherently less predictable, and responding officers are less likely to know whether, or expect that other officers are present at the scene. All of the off-duty fatalities occurred in the context of crimes or disputes in progress. We describe the potential implications of these differences in situational factors in the discussion section.

Limitations

While analyses of patterns and trends in these shootings can point to important observations and questions about the use of force in policing, our descriptive analyses and the discussion that follows are limited in several important ways.

Data. There are three important limitations of the source data. First, LEOKA data are voluntarily reported by law enforcement agencies, although an FBI field office often initiates the reports by notifying the Uniform Crime Reporting program of an officer fatality. In searching for news coverage and ad- ditional documentation regarding the mistaken identity shootings listed in the LEOKA data, the task force identified three additional cases that did not appear in the LEOKA data set and acknowledges that there could be others (Stone et al., 2010). That said, inaccuracy in general homicide reporting tends to be low and these are high profile, rare events. The task force did not specify which three cases were identified outside of the LEOKA data set. Second, as is frequently the case with racial categorization and especially the label of “Hispanic,” an officer’s race and/or ethnicity as identified by LEOKA and media reports may be inconsistent with present-day labels and/or the officer’s own self-identification. In addition, the publically available LEOKA data do not include “Hispanic” or any related labels as categories. Finally, the circumstances of each case are based on LEOKA and media reports, which rely heavily on the narratives from the officers present at

754 Charbonneau, Spencer, and Glaser

the scene and are subject to any number of reporting biases. The high profile and tragic nature of these incidents, as well as the civil and criminal legal implications, could produce strong incentives to selectively report specific details. The general facts (e.g., whether the situation involved a planned team operation) tend to be corroborated by multiple sources. In some cases, however, the accounts of officers and witnesses about the details of the confronting and victim officers’ actions (e.g., whether commands were issued and how the victim officer responded) differ.

Quantitative analyses. There are only 26 known cases over a roughly 30-year period, a small number with which it is only possible to observe basic patterns. This issue should not be conflated, however, with problems associated with small samples, i.e., the limited ability to generalize observed effects to a represented population; the data we have analyzed are population level.

The estimates of the racial disproportionality among off-duty fatalities are approximate; they average risk over time, space, and officers, using the best avail- able benchmarks. Given the rarity of these events, risk comparisons are relatively insensitive to assumptions about the total number of officers who served over the entire period. They are, however, sensitive to assumptions about the proportion of officers who were Black during this period. Nevertheless, under any reason- able set of assumptions about the demographics of the underlying population, the racial disparity in off-duty fatalities is large and would be exceedingly unlikely to occur due to chance variation if Black and White officers faced the same risk. Calculations and sensitivity analyses are included in the Supporting Information, Appendix B. The estimates reported in the text probably understate the magnitude of the disparity because they utilize the conservative benchmark of White officer representation from 2007, which was lower than during most of the studied period.

As is common in law enforcement data (see, e.g., Withrow, 2006), valid benchmarks are not available for a number of the analyses that would be infor- mative. For example, we have no information about the encounters during which officers were mistaken for civilian offenders but were not fatally shot. Due to this limitation, we are not able to estimate disparities in the decision to shoot condi- tional on the occurrence of a police-on-police encounter. Instead, we estimate the racial disparity in the probability of officers encountering, being misidentified as a civilian offender, and fatally shot by another officer.

We calculated the likelihood of observing the off-duty racial disparity by chance in a world where all officers face the same risk by making the assumption that the demographic characteristics of officers who were off duty were, on aver- age, similar to those of all officers during this time period. We cannot, however, make this assumption about officers serving in plainclothes or undercover, nor in planned team operations generally. When the mistaken identity fatalities are di- vided by the duty status of the victim officer, or by the context of the incident (i.e., planned team operation vs. crimes in progress), there are large differences in the

Police-on-Police Shootings 755

reported racial and ethnic distributions of officers killed. However, we cannot make quantitative comparisons of these patterns without additional data (specifically, the racial distribution of patterns of assignments to plainclothes duties).

Discussion

Potential Explanations of Racial Disparities

Although estimates of the magnitude of the racial disparity among officers killed while off duty must be interpreted cautiously because of the rarity of the events and limitations of the existing data, the increased risk for Black officers compared to their White counterparts appears to be even greater than the racial disparity among civilians killed by police officers. Racial disparities in police use of force can be attributed to multiple causes (Goel, Rao, & Shroff, 2016; Goff et al., 2016; Schuck, 2004), many of which are mutually reinforcing. In an effort to understand the factors contributing to the racial disparity among off-duty officers mistakenly killed by on-duty officers, we begin with the victim officers.

Behavior and social context: victim officers. Although they were perceived as threatening civilians at the time of the shooting, we know that none of the victim officers were criminal offenders and have a high degree of confidence that they would not have been killed if the confronting officers had correctly perceived them as police officers. We know with near certainty that the victim officers were very unlikely to have harmed the confronting officer or other civilians. Even in the cases where the off-duty officer was involved in a dispute, these officers pulled their weapons in self-defense, in an attempt to gain control of a situation, or to detain a civilian. Nevertheless, all of the victim officers were armed, and several reportedly either ignored or failed to comply with commands from the confronting officers.

The dramatic disproportions in the composition of victims beg several ques- tions. Are Black officers more likely than White officers to carry or draw their weapons while off duty, or in response to crimes in progress? Are they less likely to obey commands from challenging officers? Unfortunately, we are not aware of any empirical studies investigating these questions. However, anecdotal evidence suggests that the opposite may be true. Many prominent Black Americans, includ- ing policing executives, have had direct experience with other officers perceiving them as a threat or with greater suspicion (Harris, 1999, 2002; Russell-Brown, 1998). The present authors know several high-level policing professionals who give their Black children “the talk” about the risks of interacting with police of- ficers. Anecdotal reports, including one described by the task force, indicate that at least some minority officers are reluctant to carry their weapons while off duty, suggesting that it is at least possible that non-White officers are less likely to carry

756 Charbonneau, Spencer, and Glaser

or draw their firearm and would be more cautious than White officers when taking police action while off duty.

It seems more plausible that the disparity can be at least partially explained by the possibility that Black officers may be more likely than White officers to encounter crimes in progress or other situations requiring them to draw their weapons while off duty. If Black officers spend their off-duty time in neighbor- hoods that experience more crime and/or more policing, they may be more likely than White officers to find themselves in the situations that lead to mistaken identity shootings. Although we know that people of color, on average, live in neighborhoods that experience higher rates of reported crime and policing, we do not know that this is true of police officers specifically. Without fairly granular data on where officers spend their off-duty time, this explanation of the racial disparity is difficult to investigate. Nevertheless, we can draw on some labor statistics to help determine the likelihood that socioeconomic differences explain the fatality disparity. In 2010, the Bureau of Labor Statistics (2010) reported that the median weekly earnings for Black men working in protective services, which includes people working in law enforcement and similar fields, was 71% of that of White men in the same field. This earnings gap suggests that there are differences in so- cioeconomic status between Black and White officers, though it does not capture disparities in wealth, education, housing, or other factors that could have an effect on the likelihood that an off-duty officer would encounter a crime.

We would expect socioeconomic differences to be smaller on average between Black and White police officers than they are between Black and White civilians. In fact, the median household income for Black families was just 58% that of White families in 2010 (U.S. Census Bureau, 2010), for example. Similarly, we would expect that Black and White officers live in neighborhoods that are more comparable on average than the neighborhoods of average Black and White civil- ians, if only because they share employment in a specific field. In light of these factors, if anything, we would expect a smaller racial disparity among mistaken identity shootings of officers compared to the racial disparity among civilians shot by police. The pattern appears to be the opposite, although it is not possible to compare the two disparities directly because reliable data on lethal force against civilians is available only from a much more recent period. In the next section, we examine the psychological bases of racially biased perceptions of threat and automatic tendencies in decisions to shoot that could contribute to the confronting officers’ decisions.

The psychology of racial bias and law enforcement. Police officers must decide whether to use lethal force under time pressure and psychological stress, and with limited information. In all of the mistaken identity shootings, the confronting officers decided whether a person they believed to be a civilian was a threat, involved in a crime, and/or armed, then issued commands (in most cases) and then

Police-on-Police Shootings 757

decided to shoot, all within a matter of seconds. Psychological research indicates that intergroup biases like stereotypes, even, and perhaps especially those stored implicitly and operating automatically (i.e., outside of conscious awareness or control) can play an important role in such perceptions and decisions, even in the presence of consciously held egalitarian goals (Devine, 1989). A confronting officer need not hold explicitly racist beliefs to be influenced by the race of the victim officer, especially in ambiguous circumstances. Implicit racial biases likely affect many aspects of policing (Spencer, Charbonneau, & Glaser, 2016), including the perception of weapons (Eberhardt, Goff, Purdie, & Davies, 2004) and the decision to shoot (Correll, Park, Judd, & Wittenbrink, 2002; Glaser & Knowles, 2008), and we tend to associate Black people with criminality and threat (Eberhardt et al., 2004; Park & Banaji, 2000; Walsh, Banaji, & Greenwald, 1995). All of these effects are potential contributors to the racial disparity in mistaken identity shootings of officers.

Perceiving weapons. In the cases identified by the task force, all of the off-duty officers fatally shot by on-duty officers were holding weapons. Merely holding a weapon, even a firearm, however, does not necessitate that a suspect will be shot. The likelihood that the weapon will be rapidly perceived, perhaps faster than mitigating information, or that a threat response will be evoked is influenced by target race. Eberhardt et al. (2004) found that police officers and lay people identified weapons and crime-related objects more quickly after subliminal exposure to images of Black faces and that they were slower to identify the same objects after exposure to White faces, relative to a race-neutral control condition. Other findings indicated that people were quicker to categorize weapons and to incorrectly categorize tools as weapons when primed with Black faces (e.g., Judd, Blair, & Chapleau, 2004; Payne, 2001). These findings have important implications for police use of force, and especially for encounters involving weapons. Fractions of seconds matter during these encounters, and the speed of perceptual processing could affect the outcomes.

Shooter bias. The research on “shooter bias” is directly relevant to these incidents. Police officers are faster to shoot armed Black suspects than armed White suspects in laboratory-based simulations (Correll et al., 2007; Peruche & Plant, 2006; Plant & Peruche, 2005). This bias has been shown in civilian samples to be predicted by knowledge of cultural stereotypes (Correll et al., 2002) and implicit associations between weapons and Black people (Glaser & Knowles, 2008). These biases in reaction time and the perception of weapons are particularly relevant to mistaken identity shootings because all of the victim officers were armed, but none would have been killed if they were correctly identified. In the case of police-on- police encounters, additional time might allow the confronting officer to process more information about the situation or the confronted officer to identify him- or herself.

758 Charbonneau, Spencer, and Glaser

Decisions to use lethal force are complex and difficult to simulate in a labora- tory setting, and studies that have employed different procedures have yielded interestingly conflicting results. Two types of experiments simulate different elements of real-world conditions: Studies presenting rapidly sequenced trials of still photographs simulate time constraints and ambiguity of objects, while studies employing extended video scenarios simulate realistic sights and sounds. Neither type of study can fully replicate the stress and fear that officers face in the field. In a set of studies comparing photo and video stimuli, officers made very few errors overall and showed a reversal of the latency bias in the video simulations (Cox, Devine, Plant, & Schwartz, 2014). Officers were faster to shoot armed Black men compared to armed White men when responding to photo stimuli, and were slower to shoot armed Black men when responding to video stimuli. In a study utilizing high fidelity, computerized training simulators, officers took longer to shoot Black suspects, were more likely to shoot unarmed White suspects, and more likely to fail to shoot armed Black suspects (James, Vila, & Daratha, 2013). The video simulation studies raise important questions about the activation and control of biases and may be more applicable to situations where officers have more time to carefully consider the situation, including the race of the suspect and whether or not force will be necessary. The findings suggest that, under the sim- ulated conditions, officers are averse to making discriminatory errors. In contrast, when deliberation and control are relatively precluded, as in the rapidly sequenced still photo simulations, automatic activation of implicit associations (e.g., race- weapon) appears to be more influential. However, without realistically evoking the intense fear inherent to these confrontations, it is unclear that the video or photo simulations can capture the real-world response tendencies. The research on weapon perception and shooter bias lacks external validity to the extent that the decisions occur in response to a still image, but they may better operationalize the spontaneity of the decision to shoot in real cases. They do, in fact, comport better to the racial disparities seen in the real-world data.

Criminality, danger, and threat. Mental associations between Black peo- ple, criminality, and danger more generally could also contribute to the confronting officers’ decisions. In all of these situations, the confronting officer reported mis- perceiving the victim officer as a civilian offender posing a great deal of threat. Implicit biases that link social groups with traits related to crime and violence are particularly relevant in this context. Eberhardt and colleagues (2004) demon- strated that police officers, like most other Americans, exhibit implicit biases linking Black men to crime. As a function of their professional role, law enforce- ment officers will think about crime and related concepts very frequently, and these studies suggest that even the abstract idea of crime will lead them to pay more attention to Black people they encounter, and to be more likely to presume those Black people pose a threat. Stereotypes linking Black people to crime have

Police-on-Police Shootings 759

been demonstrated using multiple methodologies (e.g., Hurwitz & Peffley, 1997; Park & Banaji, 2000). Implicit associations could facilitate an officer’s misper- ception of a Black officer as a dangerous criminal in addition to facilitating the identification of a weapon, and could even influence the perception of a White person holding a gun as a potential officer, or at least not a criminal.

Body language and behavior can be especially ambiguous in police encounters involving weapons. For example, experts consulted by the task force describe “reflexive spin,” where officers and civilians alike tend to turn quickly toward an officer who is shouting, “Police! Don’t move!” and are shot by an officer who perceives this motion as a threat (Stone et al., 2010). People tend to judge the ambiguous behavior of a Black person as more aggressive than the identical behavior of a White person (Duncan, 1976; Sagar & Schofield, 1980). Other studies have uncovered pervasive implicit and explicit stereotypes of Black people as dangerous, violent, and hostile (Bargh, Chen, & Burrows, 1996; Devine, 1989; Devine & Baker, 1991; Dovidio, Evans, & Tyler, 1986; Jackson, Lewandowski, Ingram, & Hodge, 1997; Schaller, Park, & Mueller, 2003). These stereotypes could contribute to an officer’s quick assessment of whether the individual s/he encounters poses a threat, even when the officer consciously wants to avoid making biased decisions.

Taken together, this body of research on stereotypes and biases that are relevant to high-stress policing decisions provides useful insights into the potential causes of racial disparities in mistaken identity shootings of police officers. None of the officers would have been killed if correctly identified, and they were generally perceived to be posing a level of threat that was greater than the actual threat. These are important distinctions from cases involving civilian victims, where the culpability of the civilian is central to the discussion, and it can be difficult or impossible to know whether the person posed a true threat. Obviating the need to determine the intentions of the individual person killed allows for a more focused discussion of the troubling pattern of racial disparities in police use of force. The psychological research can help us understand why these tragic errors occur more often for civilians and off-duty officers of color. When encountering a Black person, most officers, like most other Americans, are more likely to perceive that person as a threat, and will be faster to identify (or misidentify) a weapon and shoot. The opposite will be true when they encounter a White person. In fact, Eberhardt et al. (2004) found that police officers who were subliminally primed with White people’s faces were slower to identify crime-related objects than were those in the neutral, no-prime control condition. The inhibiting effect of encountering a White person could, in fact, put officers at risk of not defending themselves adequately. Given that the Washington Post data on fatal officer-involved shootings indicate that the overwhelming plurality of victims is armed Whites, this is not merely a theoretical concern.

760 Charbonneau, Spencer, and Glaser

Variation in ambiguity. Without additional data, we cannot be certain that the difference in racial disparities between on- and off-duty fatalities, or between situations involving planned team operations and crimes in progress, is not in fact due to a difference between the demographic characteristics of the two underlying populations. However, the patterns suggest that there may be multiple factors to consider and warrant qualitative investigation. All 26 of these fatalities represent errors in the decision to shoot. Planned team operations are carried out by specialized teams in a specified location, incorporate a number of tactical strategies, and there are typically policies and procedures in place that aim to protect officers serving undercover or in plainclothes. Departments often adopt practices specific to officers serving out of uniform, such as a clothing color and/or code word of the day and maintaining radio contact. In addition, dispatchers may notify responding officers of the presence of a plainclothes officer in the vicinity of a response location. Therefore, the fatalities observed during planned operations represent at least two errors: A failure of procedures intended to protect officers serving out of uniform and the mistaken perception of the victim officer as a suspect. It is likely that the protective policies and procedures in place greatly reduce the potential number of misidentifications of officers. In contrast, officers responding to crimes and disputes in progress are reacting to circumstances as they unfold, and there are few procedures8 in place to protect out-of-uniform (whether on- or off duty) officers who step in.

To summarize, planned team operations involve less ambiguity about the identities of participating officers than do crimes in progress. Given that heuris- tics and biases often influence judgments made under conditions of uncertainty (Tversky & Kahneman, 1974), a reduction in ambiguity could decrease the like- lihood that a confronting officer will be influenced by racial stereotypes when sizing up another officer, particularly one who is out of uniform. Put another way, in addition to the specific information helping to identify out-of-uniform officers during planned operations, the relative predictability of these operations may ren- der them less uncertain situations, and therefore less prone to bias-driven errors. We might therefore observe more errors driven by stereotypes in the inherently more ambiguous situations involving crimes in progress. This line of reasoning is consistent with the observed racial distribution of officers killed during planned team operations, which closely resembles that of all officers who served during this time period, and the racial disparity among officers killed during disputes and crimes in progress. A compelling parallel case comes from U.S. Customs. In 1999, the Customs Director instructed agents to change the set of criteria used to trigger searches, from 43 down to 6 (Gladwell, 2006). Under these conditions, search rates declined roughly 75%, but hit rates (e.g., contraband yields) quadrupled

8 One exception is radio contact via a dispatcher, although this is not always possible.

Police-on-Police Shootings 761

while previously dramatic ethnic disparities in search rates and hit rates nearly disappeared (Ramirez, Hoopes, & Quinlan, 2003). A likely explanation for this set of outcomes is that the honing of attention to specific, relatively reliable indicators of suspiciousness reduced the relative influence of group-based stereotypes. Sim- ilar patterns (increased yield rates, reduced racial disparities) have been observed in administrative data of police departments that have reduced pedestrian stop and search rates (see e.g., Chauhan et al., 2015, figure 18).

Summary and Lessons

The racial disparity in fatal police shootings of off-duty officers is likely caused by a number of interacting factors, as is true for racial disparities in many domains. Black officers may be more likely to spend time in higher crime or more heavily policed neighborhoods compared to White officers, but if this difference exists we would expect it to be smaller than that among civilians. Although we are not aware of any empirical studies examining differences by race in the propensity of officers to carry or draw their weapon while off duty, anecdotal reports indicate that some officers of color are apprehensive about and avoid carrying or using their weapons while off duty. In contrast, the empirical evidence from psychological science affords less ambiguous explanations. Decades of research on implicit and explicit racial biases provide insight into why Black civilians and officers alike might face a higher risk of being mistaken for a criminal. Here, we discuss critical factors that should be considered in future research, and in the development and implementation of policies and practices to address wrongful officer-involved shootings and group-based disparities therein.

Reducing misidentification. Fatal officer-on-officer shootings, though rare, may simply be the most visible among many other situations where officers of color are misidentified. Many agencies seek to increase the diversity of sworn officers, as recommended by The President’s Task Force on 21st Century Policing (2015), in a potentially valuable effort to shift department culture and improve community relations. To address a potential unintended consequence, these agen- cies may also need to consider adopting policies and practices that protect officers against misidentification. The New York State task force report made a number of recommendations, including the implementation of specific protocols for officers taking action while out of uniform (e.g., serve as a witness instead of interven- ing), confronted officers (e.g., avoid “reflexive spin”), and officers challenging an armed individual (e.g., take cover) (Stone et al., 2010, p. 55). These protocols could certainly decrease the likelihood that a confronted officer will be misiden- tified and/or subjected to force. However, the first instruction in the protocol for confronting officers: “Don’t Stereotype” (Stone et al., 2010, p. 55) is likely to be ineffective given the potential for stereotypes to operate outside of conscious awareness and control, particularly in high stress, ambiguous situations.

762 Charbonneau, Spencer, and Glaser

Addressing racial bias. Law enforcement agencies and leaders are increas- ingly aware of, and attempting to address, the effects of racial bias. In the case of mistaken-identity shootings of officers, doing so could increase officer safety and prevent events that are deeply damaging to departments. The task force on police-on-police shootings (Stone et al., 2010), The President’s Task Force on 21st Century Policing (2015), and others (e.g., U.S. Department of Justice, 2016) rec- ommend that police officers receive training on implicit bias. However, although there is strong evidence that police officers, like all people, are affected by implicit biases, law enforcement leaders, policymakers, and researchers should not focus so narrowly on implicit bias that other important factors are ignored, such as the potential effects of social inequities, the reliance on policing in place of other so- cial services, segregation, police deployment policies, institutional discrimination, and explicit bias on policing decisions and related outcomes. In addition, there is no evidence to suggest that implicit biases can be reduced through education alone. Given that this particular form of bias operates largely outside of conscious awareness and control, there is no reason to have confidence that education would be effective. In fact, a large number of carefully designed interventions that move beyond basic education have failed to lastingly reduce implicit biases (Lai et al., 2016) (but see Devine, Forscher, Austin, & Cox, 2012; Forscher, Mitamura, Dix, Cox, & Devine, 2017 for an intensive, multifaceted intervention that lastingly pro- motes processes that reduce biased outcomes, if not lasting reduction in implicit biases themselves). Instead of focusing purely on raising awareness, trainings should emphasize practices that reduce the potential for biases to affect officers’ judgments and decisions regarding the people they encounter, whether civilians or other officers. Agencies can incorporate these strategies into their policies and procedures; we describe a few examples below. Finally, any training program should be rigorously evaluated to ensure that it does not have ironic effects and to identify promising practices.

Reducing uncertainty and ambiguity. Because we lack information about the racial distribution of officers serving in plainclothes and/or planned team operations, we cannot make definitive comparisons between the magnitudes of racial disparities among fatalities that occurred during planned team operations as opposed to during crimes in progress. It is possible, however, that the rela- tively low ambiguity in the former makes for fewer potential opportunities for stereotype-driven errors, and this is consistent with research on decision-making under conditions of uncertainty. Policies, procedures, and trainings that reduce un- certainty and ambiguity are likely to reduce racial disparities in policing decisions. In the present case, that could involve improving methods for officers identifying themselves and discouraging officers from spontaneously engaging with others’ operations.

Police-on-Police Shootings 763

Reducing use of lethal force. Another avenue to reducing disparities is to reduce the total number of incidents where lethal force is used. To this end, a number of use-of-force policies and procedures show promise for reducing the use of lethal force generally, decreasing ambiguity, and lowering racial disparities. For example, in Richmond, CA, a city notorious for high crime and racial segregation, the police department has drastically reduced its use of lethal force. The command staff and commentators have pointed to an increase in the scenario-based trainings that incorporate live actors, streamlined crisis deployment strategies, an emphasis on de-escalation and nonlethal force, and thorough reviews of situations involving force which are later incorporated into training efforts (Rogers & DeBolt, 2014). Other departments have changed their vehicle (Alpert & Lum, 2014) and foot pursuit (Goff, 2016) policies in an effort to reduce the number of incidents that escalate into the use of force, in part by slowing the course of events.

If we extend the lessons from the weapon perception and shooter bias research, tactics that increase the amount of time before an officer fires his or her weapon (e.g., taking cover and putting distance between the officer and suspect), without compromising safety, may allow for more deliberative cognitive processing that is less prone to the effects of racial biases. Furthermore, any intervention that reduces the overall rate of a behavior that tends to have a discriminatory pattern is likely to reduce the sheer number of biased events and even to reduce disparities between groups (Glaser, 2006). In fact, trends in the Washington Post fatal officer- involved shootings data indicate that this is the case, as numbers of unarmed civilian victims and the disparities associated with such events have declined over the recorded period from 2015 through mid-2017. Many departments have increased their training on de-escalation strategies and other methods to reduce unnecessary enforcement contact and use of force. In cases where the person does not pose a true threat, such as situations where an intervening officer is mistaken for a criminal or an object is mistaken for a weapon, these tactics can save lives. That said, trainings on tactics meant to increase the time an officer takes to resolve dangerous situations must also be carefully calibrated to consider the nuances of protecting the safety of officers and the public. Many departments employ strategies to increase time, cover, and distance in situations involving armed or potentially armed suspects, and these procedures should be evaluated to ensure that they do not increase the likelihood of officer or civilian fatalities.

Conclusion

Mistaken identity police-on-police shooting fatalities involve a misperception of threat, and the patterns of racial disparities across operational configurations suggest that there are situational factors that vary systematically and contribute to the observed outcomes. The dramatic racial disproportion in off-duty officer fatalities in police-on-police shootings is likely caused by multiple interacting

764 Charbonneau, Spencer, and Glaser

factors, and unfortunately comes as no great surprise to those familiar with the psychological science on implicit and automatic stereotyping and discrimination, particularly “shooter bias.” Although inferences need to be made with caution because the underlying demographics of officers serving in plainclothes or under- cover is unknown, the less pronounced racial disparities in fatalities that occurred during planned team operations suggests that race-neutral policies and proce- dures that assist officers in decision-making may reduce the potential for racially disparate outcomes. It may be that the procedures in place for on-duty officers reduce ambiguity, thereby decreasing the influence of stereotypic associations. Although procedures like “the color of the day” cannot be directly applied to civil- ians, they can guide researchers, practitioners, and policymakers toward practices that reduce ambiguity or safely increase the amount of time that an officer has to make a decision, so that policing decisions can be based to a greater degree on actual, valid indicators of suspiciousness and threat.

Supporting information

Additional Supporting Information may be found in the online version of this article at the publisher’s website:

Appendix A. Fatal, Mistaken Identity Police-on-Police Shootings 1981-2009. Appendix B. Methods and Results for Probability of Disparity Analyses.

References

Alpert, G. P., & Lum, C. (2014). Police pursuits: A complex policy arena. In Police pursuit driving (pp. 1–12). New York, New York: Springer.

Bargh, J. A., Chen, M., & Burrows, L. (1996). Automaticity of social behavior: Direct effects of trait construct and stereotype activation on action. Journal of Personality and Social Psychology, 71(2), 230.

BLS. (2010). Labor Force Characteristics by Race and Ethnicity, 2009 (No. 1026). Washington, DC: U.S. Bureau of Labor Statistics, U.S. Department of Labor.

Bureau of Justice Statistics. (2008). Census of state and local law enforcement agencies. Washington DC: Office of Justice Programs.

Chauhan, P., Warner, T., Fera, A., Balazon, E., Lu, O., Welsh, M., & Travis, J. (2015). Tracking enforcement rates in New York City, 2003–2014. Report Presented to the Citizens Crime Commission.

Correll, J., Park, B., Judd, C. M., & Wittenbrink, B. (2002). The police officer’s dilemma: Using ethnicity to disambiguate potentially threatening individuals. Journal of Personality and Social Psychology, 83(6), 1314–1329.

Correll, J., Park, B., Judd, C. M., Wittenbrink, B., Sadler, M. S., & Keesee, T. (2007). Across the thin blue line: police officers and racial bias in the decision to shoot. Journal of Personality and Social Psychology, 92(6), 1006.

Cox, W. T., Devine, P. G., Plant, E. A., & Schwartz, L. L. (2014). Toward a Comprehensive Under- standing of Officers’ Shooting Decisions: No Simple Answers to This Complex Problem. Basic and Applied Social Psychology, 36(4), 356–364.

Devine, P. G. (1989). Stereotypes and prejudice: their automatic and controlled components. Journal of Personality and Social Psychology, 56(1), 5.

Police-on-Police Shootings 765

Devine, P. G., & Baker, S. M. (1991). Measurement of racial stereotype subtyping. Personality and Social Psychology Bulletin, 17(1), 44–50.

Devine, P. G., Forscher, P. S., Austin, A. J., & Cox, W. T. (2012). Long-term reduction in implicit race bias: A prejudice habit-breaking intervention. Journal of Experimental Social Psychology, 48(6), 1267–1278.

Dovidio, J. F., Evans, N., & Tyler, R. B. (1986). Racial stereotypes: The contents of their cognitive representations. Journal of Experimental Social Psychology, 22(1), 22–37.

Duncan, B. L. (1976). Differential social perception and attribution of intergroup violence: Testing the lower limits of stereotyping of Blacks. Journal of Personality and Social Psychology, 34(4), 590.

Eberhardt, J. L., Goff, P. A., Purdie, V. J., & Davies, P. G. (2004). Seeing black: Race, crime, and visual processing. Journal of Personality and Social Psychology, 87(6), 876–893.

Federal Bureau of Investigation. (2013). United States Department of Justice. Uniform Crime Reporting Program Data: Police Employee. Law Enfrocement Officers Killed or Feloniously Assaulted.

Forscher, P. S., Mitamura, C., Dix, E. L., Cox, W. T., & Devine, P. G. (2017). Breaking the prejudice habit: Mechanisms, timecourse, and longevity. Journal of Experimental Social Psychology 72, 133–146.

Gabrielson, R., Jones, R. G., & Sagara, E. (2014). Deadly Force, in Black and White. Retrieved February 17, 2015, from http://www.propublica.org/article/deadly-force-in-black-and-white

Gladwell, M. (2006). The tipping point: How little things can make a big difference. New York, New York: Little, Brown and Company.

Glaser, J. (2006). The efficacy and effect of racial profiling: A mathematical simulation approach. Journal of Policy Analysis and Management, 25(2), 395–416.

Glaser, J. (2015). Suspect race: Causes and consequences of racial profiling. USA: Oxford University Press.

Glaser, J., & Knowles, E. D. (2008). Implicit motivation to control prejudice. Journal of Experimental Social Psychology, 44(1), 164–172.

Goel, S., Rao, J. M., Shroff, R. (2016). Precinct or prejudice? Understanding racial disparities in New York City’s stop-and-frisk policy. Annals of Applied Statistics, 10(1), 365–394.

Goff, P. A. (2016). Identity traps: How to think about race & policing. Behavioral Science & Policy, 2(2), 10–22.

Goff, P. A., Lloyd, T., Geller, A., Raphael, S., & Glaser, J. (2016). The science of justice: Race, arrests, and police use of force. Los Angeles, CA: Center for Policing Equity.

Harris, D. A. (1999). The stories, the statistics, and the law: Why driving while Black matters. University of Minnesota Law Review, 84, 265–326.

Harris, D. A. (2002). Profiles in injustice: Why racial profiling cannot work. New York: The New Press.

Hurwitz, J., & Peffley, M. (1997). Public perceptions of race and crime: The role of racial stereotypes. American Journal of Political Science, 375–401.

Jackson, L. A., Lewandowski, D. A., Ingram, J. M., & Hodge, C. N. (1997). Group stereotypes: Content, gender specificity, and affect associated with typical group members. Journal of Social Behavior & Personality.

James, L., Vila, B., & Daratha, K. (2013). Results from experimental trials testing participant responses to White, Hispanic and Black suspects in high-fidelity deadly force judgment and decision- making simulations. Journal of Experimental Criminology, 9(2), 189–212.

Jones-Brown, D. D., Gill, J., & Trone, J. (2010). Stop, question & frisk policing practices in New York City: A primer. New York: Center on Race, Crime and Justice, John Jay College of Criminal Justice.

Judd, C. M., Blair, I. V., & Chapleau, K. M. (2004). Automatic stereotypes vs. automatic preju- dice: Sorting out the possibilities in the weapon paradigm. Journal of Experimental Social Psychology, 40(1), 75–81.

Kahn, K. B., & Martin, K. D. (2016). Policing and race: Disparate treatment, perceptions, and policy responses. Social Issues and Policy Review, 10(1), 82–121.

Lai, C. K., Skinner, A. L., Cooley, E., Murrar, S., Brauer, M., Devos, T., Calanchini, J., Xiao, Y. J., Pedram, C., Marshburn, C. K., Simon, S., Blanchar, J. C., Joy-Gaba, J. A., Conway, J., Redford,

766 Charbonneau, Spencer, and Glaser

L., Klein, R. A., Roussos, G., Schellhaas, F. M. H., Burns, M., Hu, X., McLean, M. C., Axt, J. R. Asgari, S., Schmidt, K., Rubinstein, R., Marini, M., Rubichi, S., Shin, J.-E. L. Nosek, B. A. (2016). Reducing implicit racial preferences: II. Intervention effectiveness across time. Journal of Experimental Psychology: General, 145(8), 1001–1016.

Nix, J., Campbell, B. A., Byers, E. H., & Alpert, G. P. (2017). A bird’s eye view of civilians killed by police in 2015: Further evidence of implicit bias. Criminology & Public Policy, 16(1), 309–340.

Park, J., & Banaji, M. R. (2000). Mood and heuristics: The influence of happy and sad states on sensitivity and bias in stereotyping. Journal of Personality and Social Psychology, 78(6), 1005–1023.

Payne, B. K. (2001). Prejudice and perception: the role of automatic and controlled processes in misperceiving a weapon. Journal of Personality and Social Psychology, 81(2), 181.

Peruche, B. M., & Plant, E. A. (2006). The correlates of law enforcement officers’ automatic and controlled race-based responses to criminal suspects. Basic and Applied Social Psychology, 28(2), 193–199.

Plant, E. A., & Peruche, B. M. (2005). The consequences of race for police officers’ responses to criminal suspects. Psychological Science, 16(3), 180–183.

President’s Task Force on 21st Century Policing. (2015). Final Report of the President’s Task Force on 21st Century Policing. Washington, DC: Community Oriented Policing Services.

Ramirez, D. A., Hoopes, J., & Quinlan, T. L. (2003). Defining racial profiling in a post-September 11 world. Am. Crim. L. Rev., 40, 1195.

Rogers, R., & DeBolt, D. (2014). Use of deadly force by police disappears on Rich- mond streets. Retrieved February 17, 2015, from http://www.contracostatimes.com/ news/ci_26482775/use-deadly-force-by-police-disappears-richmond-streets

Russell-Brown, K. (1998). The color of crime. New York: NYU Press. Sagar, H. A., & Schofield, J. W. (1980). Racial and behavioral cues in black and white children’s

perceptions of ambiguously aggressive acts. Journal of Personality and Social Psychology, 39(4), 590.

Schaller, M., Park, J. H., & Mueller, A. (2003). Fear of the dark: Interactive effects of beliefs about danger and ambient darkness on ethnic stereotypes. Personality and Social Psychology Bulletin, 29(5), 637–649.

Schuck, 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(6), 557–564.

Spencer, K. B., Charbonneau, A. K., & Glaser, J. (2016). Implicit bias and policing. Social and Personality Psychology Compass, 10(1), 50–63.

Stone, C., Carter, Z., & Belfiore, T. (2010). Reducing inherent danger: Report of the task force on police-on-police shootings. New York: New York State Task Force on Police-on-Police Shootings.

The Washington Post. (2017). Fatal force. Retrieved from https://github.com/washingtonpost/ data-police-shootings.

Tversky, A., & Kahneman, D. (1974). Judgment under uncertainty: Heuristics and biases. Science, 185(4157), 1124–1131.

U.S. Census Bureau. (2010). Annual Social and Economic Supplements (Current Population Survey No. P60-239).

U.S. Department of Justice. (2016, June 27). Implicit bias training. Retrieved from https://www.justice. gov/opa/file/871116/download.

Walsh, W., Banaji, M. R., & Greenwald, A. G. (1995). A failure to eliminate race bias in judgments of criminals. 7th Annual Convention of the American Psychological Society, New York.

Withrow, B. L. (2006). Racial profiling: From rhetoric to reason. USA: Prentice Hall.

AMANDA CHARBONNEAU is a PhD student at UC Berkeley’s Goldman School of Public Policy. She studies the law and psychology of suspicion and analyzes law enforcement policies and practices.

Police-on-Police Shootings 767

KATHERINE SPENCER is a Social/Personality Psychology graduate student at the University of California, Berkeley. Her research focuses on the relationship between implicit bias and behavior, with special attention on the control of implicit bias. She is particularly interested in studying this relationship in public policy- relevant domains such as policing behavior. She is currently examining the role of egalitarian motivation in controlling implicit bias, with a view toward practical applications of this knowledge, such as interventions to boost motivation and training to reduce the effects of implicit bias.

JACK GLASER received his PhD in psychology from Yale University in 1998 and is a professor at the Goldman School of Public Policy at the University of California, Berkeley. He conducts research on stereotyping, prejudice, and dis- crimination, examining phenomena ranging from unconscious thoughts, feelings, and motives to discriminatory behaviors like racial profiling and hate crimes.

Copyright of Journal of Social Issues is the property of Wiley-Blackwell and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.