Risky Situations: Sources of Racial Disparity in Police Behavior

Swencionis & Goff identified five situations that tend to increase the likelihood that an individual police officer may behave in a racially disparate way: discretion, inexperience, salience of crime, cognitive demand, and identity threat. This article applies their framework to the realities of police work, identifying situations and assignments in which these factors are likely to influence officers’ behavior. These insights may identify opportunities for further empirical research into racial disparities in such contexts and may highlight institutional reforms and policy changes that could reduce officers’ vulnerability to risks that can result in racially unjust actions.


INTRODUCTION
What are the sources of the persistent racial disparities that plague American policing? To date, social psychologists have relied largely on experimental studies to infer the reasons for racial disparities in policing (Eberhardt et al. 2004, Goff & Godsil 2016, Swencionis & Goff 2017. Relatively few empirical studies have analyzed the causes of racially disparate police behavior in the field (Goff & Kahn 2012). Economists, meanwhile, aim to assess discrimination in policing by estimating the racial disparity in police encounters that can be predicted by yield rates, crime rates, and neighborhood differences in income and demography, attributing any remaining unexplained disparity to an idiosyncratic "taste for discrimination" in the individual police officer (Gelman et al. 2007, Knowles et al. 2001. None of these empirical approaches, though, can predict when or where disparities in policing outcomes will emerge. To generate testable predictions about the sources of disparate behavior by police, empirical scholars need to know which situational risk factors predict biased behaviors. They need to know where to look.
For decades, criminal justice researchers have used risk assessments to aggregate situational risk factors and develop risk profiles (Andrews 1983;Andrews et al. 1984;Campbell et al. 2009;Feeley & Simon 1992Haggerty & Ericson 2000;O'Malley 2004O'Malley , 2010. Most of the prior research in criminality focuses on the risk of unlawful or violent behavior by residents (Andrews et al. 2006, Byrne & Pattavina 2006, Douglas et al. 1999, Goddard & Myers 2016, Hanson & Morton-Bourgon 2009, O'Malley 2010, Rigakos 1999. To date, very little research has addressed factors that may increase the risk of unjust, unlawful, or violent behavior by police officers (Goff & Rau 2020). We know from the social psychology literature that attitudes are weak predictors of behavior (Ajzen 1988, Deutscher 1966, LaPierre 1934, Wicker 1969, including racially disparate behavior (LaPierre 1934, Wicker 1969. Situational factors can serve as moderators, transforming internal biases into overt actions (Ajzen 1988, Schwartz & Alwin 1971. The common thread in existing risk assessments is that they aim to capture the predictive factors of certain situations, or situational risk factors. In this article, we extend this tradition to include biased and burdensome policing. Social psychology has long established that individual predispositions, such as implicit and explicit bias, are weak predictors of racially disparate behavior: External situations influence human behavior much more strongly than individual biases do (Crandall et al. 2002, Dovidio & Gaertner 2000, Krieger & Fiske 2006. Thus, racially disparate policing practices are unlikely to arise solely (or even mainly) from police officers' individual biases: Depending on context, norms, and situational cues, less-biased officers may behave in racially disparate ways, and more-biased officers may behave fairly (Martin et al. 2013, Swencionis & Goff 2017, Trinkner et al. 2016. For example, in 2013, in the face of sustained public criticism and constitutional litigation over racial disparity in its stop-and-frisk encounters, the New York City Police Department (NYPD) issued several directives that discouraged racial profiling and increased the reporting requirements for such encounters. Although it seems unlikely that the directives would have instantly transformed any beliefs or biases held by individual NYPD officers, the directives produced an immediate, steep decline in the number of stop-and-frisk encounters, and by 2015 the racial disparity in such encounters had been eliminated (MacDonald & Braga 2019). If we are looking to reduce racial disparity in stops, searches, and use of force by police, researchers must identify, and policy makers must address, the contextual factors that may frame officers' judgments and behaviors. Swencionis & Goff (2017) identified five situational risk factors that can arise in policing and have been shown to heighten the risk of discriminatory behavior: discretion, inexperience, salience of crime, cognitive demand, and identity threat. The disparate behaviors that can result from these and other factors can undermine community perceptions of police legitimacy, fray police-community relations, and erode the safety of residents who may fear and mistrust police officers they see behaving in racially disparate ways. The risk factors Swencionis & Goff (2017) identified may offer a research agenda for empirical scholars to examine how basic psychological science translates to this domain, and for police executives and other policy makers to identify policy changes that could mitigate racial injustice. This article considers the application of these five factors to police officers' behavior on the job.

SITUATIONAL RISK FACTORS FOR DISPARATE BEHAVIOR
Social scientists conceptualize human social behavior as resulting from a combination of individual dispositions (such as implicit and explicit prejudice, authoritarianism, and social dominance orientation) and situational factors (such as rules, social norms, and circumstances). This section identifies aspects of policing activity that may be affected by the five factors Swencionis & Goff noted-discretion, inexperience, salience of crime, cognitive demand, and identity threat-and considers how each of them might contribute to racially disparate behaviors by officers on the job.

Discretion
Situational ambiguity-the absence of clear-cut rules that prescribe an appropriate course of action-can increase the likelihood of racially disparate behavior. Where decisions or behaviors are governed by clear-cut, binding norms, a decision maker is more likely to apply them fairly, without racial favoritism (Dovidio & Gaertner 2000). But where rules or norms are vague, nonexistent, or not clearly applicable, people, including police officers, are more likely to act in ways that favor White persons and disfavor non-Whites (Dovidio & Gaertner 2000, Gaertner et al. 1982. Thus, the existence of broad or untrammeled discretion tends to exacerbate the risk that a person will behave in racially disparate ways. Police officers, of course, enjoy broad discretion in the course of their work. In the field, officers are often not directly observable by their supervisors (Walker & Katz 2017). Although the introduction of body-worn cameras and the ubiquity of cellphone recordings of police encounters may expose officers' behavior to increased public scrutiny, it is not clear that the availability of video recordings has systematically increased accountability for police misconduct (Buchanan & Goff 2019).
Police officers necessarily exercise a great deal of discretion in the course of their work: They must continually make judgment calls about the meaning of others' behavior and about how to weigh facts, circumstances, and evidence. Even where an officer's behavior is governed by specific departmental rules or procedures, direct supervision of their conduct in the streets is seldom possible. Moreover, in reviewing police incidents, supervisors tend to defer to field officers' judgment, even when their behavior deviates from departmental norms (Wachtel 1992). In the absence of binding directives or established protocols, officers' reliance on discretion or personal judgment can allow their prejudices to influence their decisions (Bloche 2001, Bodenhausen 1988, Devine 1989, Dovidio 2001, Dovidio et al. 2008, Smith & Levinson 2011, Wachtel 1992. Furthermore, officers' discretionary exercise of their powers to stop, search, and use force against civilians is not meaningfully constrained by the courts. Although courts characterize the Fourth Amendment standard they apply to stops, searches, and police use of force as an objective one [Graham v. Connor (1989)], the objective "reasonableness" of an action-that is, its constitutionality-is determined "from the perspective of a reasonable officer on the scene," making "allowance for the fact that officers are often forced to make split-second judgments in circumstances that are tense, uncertain, and rapidly evolving" [Graham v. Connor (1989)]. "Not every push or shove" will violate the Fourth Amendment, "even if it may later seem unnecessary in the peace of a judge's chambers" [ Johnson v. Glick (1973)]. In practice, courts are extremely deferential to officers' discretionary decisions to stop members of the public and to use force against them. The federal courts' generous interpretation of qualified immunity, moreover, "gives ample room for mistaken judgments" by protecting "all but the plainly incompetent or those who knowingly violate the law" [Hunter v. Bryant (1991), Malley v. Briggs (1986)]. As a result, courts tend not to hold officers accountable for decisions to use force, even where their use of force was unlawful [Davis v. Scherer (1984)], contrary to departmental policy [Davis v. Scherer (1984), Mullenix v. Luna (2015)], in contravention of their training [City & County of San Francisco v. Sheehan (2015)], or in defiance of a supervisor's direct order [Mullenix v. Luna (2015)]. Indeed, courts have invoked the existence of departmental procedure manuals designed by police leadership to govern police officers' exercise of discretion-"a plethora of rules, often so voluminous, ambiguous, and contradictory, and in such flux, that officials can only comply with or enforce them selectively" [Davis v. Scherer (1984)]-as reason to grant wide discretion to officers whose conduct diverges from those rules. The courts permit officers to "act swiftly and firmly at the risk that action deferred will be futile or constitute virtual abdication of office" [Scheuer v. Rhodes (1974)]. In other words, they rarely impose meaningful constraints on police officers' exercise of discretion.
The risk of discrimination posed by overly broad discretion can arise in nearly any area of policing. This section addresses those risks for officers in three types of assignments: undercover investigation of sex work and criminal gangs, homicide investigations, and school resource officers (SROs).
Police officers are frequently assigned to operate undercover to investigate sex work, sex trafficking, or gang activity. The discretion afforded to officers working on such assignments poses acute risks of discrimination and abuse, both because of the racialized and gendered vulnerability of the people being investigated and because the disguised presence of law enforcement invades what the people being investigated think are private interpersonal interactions (Buchanan & Goff 2019, Fijnaut & Marx 1995, Girodo 1991, Joh 2009, Marx 1989, Wachtel 1992. For example, in gang investigations, the absence of standards for what counts as a criminal "gang" and what kind of acquaintance counts as gang "membership" allows (and almost requires) officers to rely on stereotypes to fill in the gaps. Such ambiguity has empowered gang investigators in police departments across the country to mislabel thousands of people as "gang members" on bases that are difficult to account for except as race discrimination (Barrows & Huff 2009, Decker & Kempf-Leonard 1991, Dumke 2018, Katz 2003, Sweeney & Fry 2018. People who are identified by police-rightly or wrongly-as gang members face harsh treatment at each stage of the criminal justice system, from higher rates of police use of force (Garner et al. 1995) to longer mandatory sentences (Klein 1997, Wright 2006) to more onerous supervision requirements upon release (Spergel 1995).
Without clear definitions of the problem they are supposed to address, gang investigators are left to rely on their assumptions about race, neighborhood, clothing, friendships, and appearance, which can give rise to unwarranted suspicion of community members who live in neighborhoods where gangs operate (Dumke 2018, Hagedorn 1990, Huff 1990, Katz 2003, Sweeney 2018, Zatz 1987. Moreover, some scholars have found that law enforcement agencies may need to demonstrate a gang problem to receive federal funds, creating an incentive to err on the side of inclusion when identifying potential gang members (Bursik & Grasmick 1995).

4.4
Pryor •  Although between 11% and 27% of self-identified gang members are estimated to be White, and White gang members report levels of criminal offending equal to those of their non-White counterparts, a powerful racial stereotype associates gang membership with Latinx and especially Black boys and young men (Esbensen & Carson 2012, Esbensen et al. 2008, Natl. Gang Cent. 2014). Thus, for example, over the past 20 years in Chicago, police officers recorded more than 128,000 adults and 33,000 children and youth in a database of gang members and "gang-affiliated" persons (Sweeney 2018, Sweeney & Fry 2018. Nearly all were non-White, with 95% of adults logged as Black or Hispanic (Dumke 2018). The criteria used to identify people as gang members or affiliates were not made public. Despite evidence that the "gang" listings were riddled with obvious errors (Dumke 2018, Sweeney & Fry 2018, the department had no procedure for removing inaccurate or outdated records from the database. Police officers acknowledged that some people had their names added to the database "based solely on where they live" (Dumke 2018). Yet the database was used for criminal investigations, background checks, immigration enforcement, and criminal sentencing (Dumke 2018).
Other undercover units, such as those assigned to investigate prostitution and sex trafficking, also enjoy wide latitude for discretion that can generate racialized and gendered abuses against marginalized populations that are already vulnerable to discrimination and violence (Buchanan & Goff 2019, Int. Assoc. Chiefs Police 2018, Shaver 1994, US Dep. Justice 2016, Williamson et al. 2007). In the United States, non-White groups (including Asian American, Black, Latinx, and Indigenous persons) and transgender people are overrepresented in the sex trade (Fitzgerald et al. 2015, James et al. 2016. As Buchanan & Goff (2019) note in a recent essay, police officers often exercise their investigative discretion in ways that harm persons involved in sex work or sex trafficking. For example, police enjoy broad discretion to believe or disbelieve reports of crimes, and further discretion as to whether to investigate and file criminal charges if the report is believed. Although sex workers face elevated risks of sexual victimization, both at work and in their personal lives, the US Department of Justice and other observers have noted a systematic practice by which police departments tend to "unfound" or refuse to investigate reports of sexual assault when the person reporting the assault is a sex worker (O'Neal et al. 2015, US Dep. Justice 2016. Worse, many jurisdictions afford police officers discretion to sexually touch persons they suspect of being sex workers. The rationale for allowing such conduct is to prevent the suspected sex worker from asking a new client to touch them sexually to prove he is not a cop. Conferring discretion on police officers to have sexual contact with persons they are investigating creates an obvious risk of sexual exploitation and abuse (Buchanan & Goff 2019, Int. Assoc. Chiefs Police 2018, Williamson et al. 2007).
Thus, the broad discretion enjoyed by police officers investigating sex work and sex trafficking creates ample opportunities for race and sex discrimination: Officers face few rules requiring them to act fairly; the populations they police are vulnerable; and, because of their marginalized status, those who do report abuse are less likely to be taken seriously.
Police leadership and senior officers such as detectives who are assigned to investigate homicides or other serious crimes enjoy broad discretion as to whether, when, and how to investigate criminal offenses. They can decide whether to treat a case as high or low priority or not to investigate it at all (Corsianos 2003). For example, police decision makers may allocate more resources for the investigation of crimes against wealthy, prominent, or high-status victims, while neglecting or ignoring crimes against poor, non-White, or otherwise marginalized victims (Baldus et al. 1983, Buchanan & Goff 2019, Corsianos 2003, Spohn 2015. SROs, too, enjoy broad discretion that can be exercised in racially disparate ways. They often work alone or in small teams, patrolling school grounds with little supervision by more senior officers (Fisher et al. 2017 are not accountable to them. Furthermore, although agreements between schools and police departments offer some guidelines as to the scope of their authority, the conflict between their law enforcement role and their placement in an educational institution for children can create ambiguity about whether and when to treat student misbehavior as a violation of criminal law (Fisher et al. 2017, May et al. 2018, McKenna et al. 2016, Schlosser 2014. Officers typically receive little guidance as to how to balance these roles (Brown 2006). Furthermore, student misbehaviors that are developmentally normal, such as insubordination, fighting, disruptive conduct, and certain drug or weapons violations, may be treated as crimes if an SRO is present. Although these undesirable behaviors have traditionally been addressed through school discipline, they are more likely to result in arrest if an SRO is present (Hirschfield 2008, Na & Gottfredson 2013. The presence of SROs is associated with increased rates of schoolbased arrest for minor criminal offenses such as simple assault and disorderly conduct (Brown 2006, Hudson et al. 2019, May et al. 2018, Na & Gottfredson 2013, Skiba et al. 2014, Theriot 2009).
Because misbehaviors like fighting and disruptive conduct are commonplace in high schools, the absence of clear guidelines or binding rules allows powerful cultural stereotypes of Black and non-White criminality to influence decisions by SROs (and school officials) about whether and when to punish student misbehavior as a crime. This may contribute to the severe racial disparities in surveillance and criminalization by which Black, Indigenous, and Latinx students face heightened risk of being arrested at school (Education Week 2014, Hirschfield 2008, Hudson et al. 2019, Morris et al. 2016. In Connecticut, for example, a recent study found that Latinx students were six times more likely to be arrested or referred to law enforcement at schools that had SROs than at schools that did not (Hudson et al. 2019). Students with disabilities are at especially high risk of school-based arrest, particularly if they are non-White (Hudson et al. 2019, Morris et al. 2016, Rausch & Skiba 2006, Theriot 2009, US Dep. Educ. Off. Civ. Rights 2014. Although training documents produced by the National Association of School Resource Officers has developed instructional materials that aim to address racial bias (NASRO n.d.), use of these curricula is not mandatory, and their effectiveness in reducing discriminatory behaviors has not been tested.

Inexperience
As Swencionis & Goff (2017) have pointed out, psychological literature on task expertise confirms the adage that practice makes perfect: Repetition of a task increases accuracy in task performance and reduces the influence of bias (Kawakami et al. 2005, MacLeod 1998. For example, clinical studies of weapon identification and shooting decisions have found that, in laboratory simulations, both civilians and police officers are more likely to shoot at an image of a Black person than a White one, more likely to misidentify an object carried by a Black person as a weapon, and quicker to make a "don't shoot" decision about a White person than a Black one (Correll et al. 2002(Correll et al. , 2007Payne 2001;Sim et al. 2013). This pattern of error, of course, is consistent with widespread social stereotypes that associate Black racial identity with crime (Eberhardt & Goff 2005;Eberhardt et al. 2004Eberhardt et al. , 2006Goff et al. 2008a).
But training and practice seem to mitigate the effects of this racial bias. Even though police officers on average have higher levels of social-dominance orientation (SDO)-that is, a commitment to the legitimacy of social hierarchies-than most community members have, and their levels of SDO tend to increase with number of years as a police officer (Sidanius & Pratto 1999, Teahan 1975, police officers are still less likely than community members to make these racialized errors on shooting tasks (Correll et al. 2007). It seems likely that police officers' accuracy results from their training and expertise in firearms use and their familiarity with departmental use-of-force guidelines about whether and when to shoot (see discussion of discretion, above). In the laboratory, community members who receive training and practice in the shooter task show increased accuracy and diminished behavioral effects of racial bias (Correll et al. 2007, Sim et al. 2013. Even among police officers, practice with the simulated shooting task tended to reduce racialized errors . Furthermore, skills training can reduce bias only if the program does not itself reinforce racial bias (Sim et al. 2013). Davies (2015) notes that police training simulations can themselves prime racial discrimination if, for example, Black persons are presented as more menacing than White persons or are more likely to be shown with guns in the simulation task.
It seems likely, then, that officers who are new to police service, or who receive a new assignment (e.g., a patrol officer who is reassigned to a drug enforcement task force or homicide investigation), might be particularly susceptible to the racialized error that can arise from inexperience. New and inexperienced officers may also be especially vulnerable to pressure to fit in with a departmental culture and prove their suitability for promotion in a way that more senior officers are not (Gau et al. 2013). They may also lack context for the crime-control initiatives to which they are assigned, leaving them to rely on limited cues, such as community members' race, gender, or presence in a high-crime neighborhood, to determine whether to suspect a person of lawbreaking (Willis et al. 2010).
Moreover, newer officers are more likely to be assigned to rotating or night shifts, which are linked to sleep disturbances (Violanti & Aron 1995) and which independently increase the likelihood of discriminatory behavior (see discussion of cognitive demand, below). Compared with officers who are over 40 years old, younger officers and officers with fewer than five years' experience use more force and are 2.5 times more likely to be investigated for use of force (McElvain & Kposowa 2004, Terrill & Mastrofski 2002. Officers who are 25 years old or younger are more than three times more likely to be investigated for use of force compared with officers over the age of 40 (McElvain & Kposowa 2004).

Salience of Crime
A powerful social stereotype associates Black people with crime and crime with Black people (Eberhardt et al. 2004). Experimental studies have shown that when primed with Black faces, police officers and other experimental subjects are more likely to think of crime, and when primed with thoughts about crime, they are more likely to discriminate against Black people (Eberhardt et al. 2004). Moreover, Eberhardt et al. (2004) have also observed that individuals who are primed to think of the stereotypical association between Blackness and crime are more likely to behave in racially disparate ways. When police officers are asked to identify a suspect (seen in a video) from a photo lineup, those who have been primed with this stereotype tend to make identification errors that are stereotype consistent (that is, if they have observed a Black suspect commit a crime, their erroneous lineup identification is likely to misidentify a photograph of a Black man whose appearance is more racially stereotypical) (Eberhardt et al. 2004).
The stereotypical association between Blackness and crime is so influential that, when individuals are primed with words associated with Black identity, participants are more likely to view the target's actions as hostile, even where the target's ethnicity is not known (Devine 1989). Similarly, experimental subjects who are primed with images of Black faces are more likely to perceive guns (whether or not the objects they see are actually guns) and less likely to identify harmless objects as harmless. The opposite effects are seen when subjects are primed with images of White faces (Bishara & Payne 2009, Govorun & Payne 2006, Payne 2001 White (Correll et al. 2002). Moreover, as mentioned above, when the video shows an unarmed man, police are quicker to decide "don't shoot" for White than for Black targets (Correll et al. 2002(Correll et al. , 2006. These findings were not affected by the racial identity of the shooter: Both Black and White officers were quicker to shoot Black targets and quicker to decide "don't shoot" for White ones. The greater the individual's existing prejudice, the greater these effects (Correll et al. 2002(Correll et al. , 2006Payne 2001). These racial effects were exacerbated when the study participants were forced to make the shoot/don't-shoot decision quickly (Correll et al. 2002). In practice, police officers make the decision of whether to shoot in approximately 0.39 seconds (Blair et al. 2011). It seems likely, then, that these racial effects might be more acute in the field than in the laboratory.
Police officers, of course, are assigned to focus on crime: Crime is chronically salient in their full-time work. Performance metrics such as COMPSTAT, which typically evaluate officer performance by its effects on crime rates without measuring racial disparity or community effects, tend to incentivize a focus on crime to the exclusion of considerations of justice , Swencionis & Goff 2017. Because thoughts of crime tend to evoke thoughts about Black people (Eberhardt et al. 2004), this risk factor may inhere in most types of police work (Swencionis & Goff 2017). It seems likely, then, that the cultural salience of the Black-criminality stereotype may increase the likelihood of racially disparate behavior that harms Black people (Eberhardt & Goff 2005;Eberhardt et al. 2004Eberhardt et al. , 2006Goff et al. 2008a).
Where a police officer's work assignment involves the possibility that persons they interact with may be armed or dangerous, these research findings highlight the risk that unarmed Black people may be exposed to unwarranted police shootings, while White persons may not be shot even if they are armed and dangerous. Indeed, some commentators have observed anecdotally that despite the proliferation of high-profile police killings of unarmed Black persons-such as Eric Garner, Walter Scott, Michael Brown, Tamir Rice, and Stephon Clark-White male murderers seem to be routinely taken into custody without injury, even after shooting at police officers (Harriot 2016, Woolington 2013).

Cognitive Demand
Police officers' work is notoriously difficult: In the face of multiple simultaneous demands upon their attention, they have to make life-altering decisions under severe time constraints, often under stressful or dangerous conditions (Swencionis & Goff 2017). These elements, coupled with the physical and mental exhaustion from shift work described above, can deplete the cognitive resources police officers may need to process information, exercise sound judgment, and inhibit the expression of racial bias. Police officers, like all other people, have limited capacity for executive function (Broadbent 1958, Kool et al. 2010, Navon & Gopher 1979, Pashler 1994. In a stressful interpersonal interaction, an officer must control the expression of bias, inhibit other undesired actions, direct their attention, avoid distraction, plan, choose, reason, and hold multiple thoughts in working memory. Each of these efforts, Swencionis & Goff (2017, p. 402) point out, "is more prone to error when the available [cognitive] resources are relatively scarce." Police officers, like everyone else, tend not to make their best decisions while exhausted or overwhelmed (Anderson et al. 2005, Macrae et al. 1994, Miller 1999, Robinson et al. 2010. The shift work required of junior officers and the late nights and odd hours that investigations and undercover work entail have been shown to reduce the quality of officers' decision making and problem solving (Burke & Mikkelsen 2006). When officers experience anxiety, anger, or fear, their high levels of arousal can affect their "ability to attend to vital information being presented in the environment" (Anderson et al. 2005, p. 45). Time pressure tends to exacerbate these effects (Dror 2007, 4. 8 Pryor •  ARjats.cls May 6, 2020 9:0 Dror et al. 1999). Wheeler & Fiske (2005) found that prejudice is exacerbated by time pressure, the need for closure, and other cognitive demands. Faced with cognitive overload, most people, including police officers, will resort to cognitive shortcuts, such as the racial stereotypes that associate Black people with crime (Bodenhausen 1988, Eberhardt et al. 2004, Govorun & Payne 2006, Swencionis & Goff 2017. Given the salience of crime and the dearth of rules constraining officers' exercise of discretion, these cognitive demands may result in racially disparate behavioral outcomes (Eberhardt et al. 2004, Kahn & Davies 2011, Kahn et al. 2016). These effects may be more acute in cross-racial interactions because interracial contact can itself act as an additional stressor, especially for individuals who score high on measures of implicit bias , Salvatore & Shelton 2007, Trawalter & Richeson 2006. Richeson & Shelton (2003) found that, among individuals who scored high on measures of prejudice, interracial interactions diminished their performance on subsequent tests of executive function, a result that the researchers later attributed to cognitive resource depletion . In police departments that are racially diverse, police officers may routinely work with different-race colleagues under conditions of equal status and shared goals, which tends to reduce prejudice and the stress associated with interracial interactions among police officers (Pettigrew & Tropp 2006). Officers' interracial interactions with witnesses, suspects, and victims of crime, however, do not share these prejudice-mitigating features.
Under these taxing conditions, officers may routinely act upon conscious or unconscious racial stereotypes without noticing or seeking evidence to counteract their preconceptions. For example, homicide investigators may experience high levels of stress, fear, sadness, or outrage at the violence they witness and the distress of survivors and their families. When investigating a new homicide, homicide investigators may find themselves working nearly straight through the first 48 hours of an investigation. Their fatigue may combine with time pressure, emotional distress, and demands to identify a suspect, resulting in "case errors, shortcuts, difficulty in working relationships, and severe domestic problems" for the officers (Sewell 1994, p. 572). These pressures are also likely to exacerbate the risk of bias.
Studies have found that, while moderate anxiety can lead to optimal performance in use-offorce situations (Andersen et al. 2016), extreme anxiety is detrimental to performance (Siddle 1995). Susceptibility to these effects varies between individuals, and expertise can insulate from some of the worst effects of high-pressure situations (LeBlanc et al. 2005, Regehr et al. 2008. The performance decrements associated with excessive anxiety may compromise an officer's ability to inhibit the influence of racial bias, increasing the likelihood of discrimination against non-White people. Officer performance can also be affected by organizational stressors arising from the institutional structure of a police department, the department's policies, and officers' perceptions of those factors (Shane 2010). When officers expect that their superiors will second-guess them, punish them for minor infractions, and fail to reward them for positive actions, they experience greater workplace stress (Shane 2010). Combined with the cognitive overload of fieldwork, workplace stressors such as these can result in diminished job performance (Shane 2010).
Even judges-who enjoy higher status, greater decisional autonomy, and more time to make decisions than police officers do-are susceptible to the effects of cognitive depletion. A foundational study of parole judges in Israel found that, immediately after a food break, judges would grant approximately 65% of parole applications, a percentage that would decline to near zero over the course of a three-hour session, then rebound to approximately 65% after the judge's next break (Danziger et al. 2011). These results were not attributable to factors relevant to the criteria for granting parole decisions, leading the authors to conclude that cognitive depletion was the culprit.

Identity Threats
Most people value their racial, ethnic, and gender ascriptions as important aspects of who they are. Police officers, like anyone else, have and value such identities; they also value their identities as police officers (Bolton & Feagin 2004, Loftus 2010. Stereotype threat can arise when an individual experiences concern that they may be judged in accordance with unfavorable stereotypes about an important aspect of their social identity. Any social identity can be the site of stereotype threat if it is subject to adverse stereotyping and is important to the individual's sense of self (Branscombe et al. 1999, Steele & Aronson 1995, Steele et al. 2002. A police officer, like anyone else, may experience "identity threat" when their situation evokes concern that they may be stereotyped (Swencionis & Goff 2017).
In particular, police officers may experience stereotype threats associated with their racial identities. A Black officer might fear being judged in accordance with a stereotypical expectation of Black criminality (Bergsieker et al. 2010, Bolton & Feagin 2004, Wilkins & Williams 2008. A White officer might fear being judged in accordance with a stereotype that White people are racist (Bergsieker et al. 2010, Goff et al. 2008b, Richeson & Shelton 2003. They may also experience stereotype threats about their professional identity: A police officer of any race or gender may be concerned that members of the community may stereotype police officers as racist (Goff 2016). Research conducted in the southwestern United States found that both officers and residents felt less positively toward law enforcement when faced with the prospect of enforcing laws they felt would be seen as racist, and officers feared for their safety as a result (Epstein & Goff 2011. These identity threats may be associated with harsh behavior toward non-White community residents. Some Black officers may be tempted to arrest non-White community members or treat them harshly in an effort to impress White peers or fit into the organizational culture (Bolton & Feagin 2004, Brown & Frank 2006, Wilkins & Williams 2008. And officers who experience concern about being judged racist are more likely to behave in racially disparate ways (Goff et al. 2008b, Martin et al. 2013, Richeson & Shelton 2003, Trinkner et al. 2019. A 2012 study in San Jose, California, found that officers were more likely to use force if they perceived that community members stereotyped police officers as racist (Martin et al. 2013). Whatever the race of the officer, the harmful behaviors engendered by racialized stereotype threats are likely to be visited on non-White people.
Officers can also experience gendered stereotype threats: Women police officers may experience concern that their authority will be questioned because of their gender, while male police officers may experience concern that they will be judged as unmanly (Garcia 2005). Racial stereotypes may also interact with masculinity norms; officers may respond to either masculinity threat or race discrimination (or both) by redoubling their commitment to hegemonic norms of masculinity as violence and physical strength , Liang et al. 2011.
Officers' work assignments may occasion masculinity threats. Officers who identify closely with masculine gender norms or with a particular work assignment may experience a combination of masculinity threat and professional threat if they believe that their work group is seen as unconventional, desk-bound, or soft on crime (Garcia 2005, Lurigio & Skogan 1994, Rosenbaum 2000, Walker & Katz 2017. Officers assigned to less-confrontational roles, such as SROs or officers assigned to work in community-oriented policing, crisis intervention teams, or youth education programs, may find that their work is devalued as feminized, in contrast to more stereotypically masculine assignments, such as beat patrol, homicide investigation, gang investigation, SWAT teams, or vice squads, which typically enjoy more prestige within police departments (Garcia 2005). At least one-third of officers assigned to community-oriented policing reported experiencing stigma 4. 10 Pryor •  and status challenge over the perception of a less-masculine assignment (Garcia 2005). Even in work groups that may be stereotyped as hypermasculine, such as SWAT teams and drug enforcement, officers may experience masculinity threat from people who use or threaten violence against them or verbally question their masculinity. Unfortunately, when gender-conforming men experience masculinity threat, they often seek to restore the threatened identity by using violence against people, such as women, effeminate men, or belligerent men, who might threaten that masculine identity (Cohn et al. 2009, Garcia 2005, Kalish & Kimmel 2010, Kimmel & Mahler 2003, Poteat et al. 2011. Racial disparities in use of force may be shaped in part by the interaction of racialized and gendered stereotype threats.

CONCLUSION
In their day-to-day work, police officers are routinely exposed to the five situational risk factors Swencionis & Goff (2017) identified-discretion, inexperience, salience of crime, cognitive demand, and identity threats-often simultaneously. Police officers, some of whom are inexperienced, are assigned to detect and respond to crime and interact with individuals and communities who may resent them and challenge their identities, all under stressful conditions of sleep deprivation and cognitive overload that tax their executive function. Each of these factors alone, and especially in combination, can aggravate the risk that officers will exercise their judgment in discriminatory ways. The circumstances of officers' work tend to exacerbate the risk of discrimination while minimizing accountability for it, and they enjoy almost unlimited discretion to make mistakes. Yet police officers are not the ones who create this high-risk situation for racially disparate behavior: Their circumstances are framed by policy choices made by courts, lawmakers, and policing executives.
The sources of disparity identified in this article, then, may identify promising sites not only for future empirical research into how and why racial disparities arise but also for institutional changes that could alter the circumstances that foster racially unjust decision making. Although it might be neither possible nor desirable to completely eliminate all policing situations that can exacerbate the risk of discrimination, many of them can be mitigated. It is appropriate and inevitable that police officers turn their attention to crime prevention, but police departments and training academies could review their curricula and training programs to ensure that they do not reinforce the stereotypical association between Blackness and crime. Although some officers will always have to work night shifts and all officers will face on-the-job stress, cognitive demand can be reduced with shorter and more regular shifts and by administrative processes that are perceived by officers as fair. Identity threats could be addressed through training for officers so that they can learn to manage their experience of, for example, masculinity threat or racist stereotype threat, minimizing the risk that they will react to such threats by engaging in discrimination or violence. And the risks posed by untrammeled discretion can be mitigated by systems of accountability for discrimination and inappropriate use of force.
Many of the situational risks for discriminatory police behavior may be institutionally entrenched. By identifying them, we hope to empower researchers, policy makers, and police leadership to take steps to enhance their officers' ability to protect public safety through just, wise, and rational decision making. By addressing the chronic situational risks of disparate policing outcomes, departments can protect officers and communities from their worst days and prevent the kinds of historical tragedies that have occurred too often between non-White communities and law enforcement.