Shades of Blue: Racial Compositions of Police and Attitudes toward Policing

1 Introduction

On May 25 of 2020, George Floyd, an unarmed 46-year-old African American man, was choked to death in Minneapolis, Minnesota while a white local police officer knelt on his neck for over nine minutes. This blatant instance of political brutality, along with multiple police-involved homicides under intense public scrutiny, brings about a national debate on how police and policing behavior should be constrained. Similar to many other issues in American politics, a racial cleavage emerges in the public’s response to police violence. Compared to people of color, especially African Americans, whites are in general more approving of police performance, and more likely to regard fatal encounters with police as isolated incidents rather than indications of larger problems (Morin and Stepler 2016; Desilver, Lipka, and Fahmy 2020). In terms of proactive moves to address police brutality, whites are also less likely to support the Black Lives Matter movement or defunding the police to better finance other public programs (Thomas and Horowitz 2020). Behind this attitudinal gap on police is an extensive division between whites and blacks regarding their perception of the criminal justice system (Sigelman et al. 1997; Hurwitz and Peffley 2005; Peffley and Hurwitz 2010).

Nonetheless, popular media’s impact on racial animus is theoretically and empirically ambiguous. On the one hand, correlations between media consumption and racial prejudice may simply reflect self-selection among individuals predisposed to those beliefs. In line with this, Dahl and DellaVigna (2009) examine modern violent film releases and find a reduction in same-day assaults due to the “voluntary incapacitation” of would-be offenders. Our analysis is motivated by the fact that race and ethnicity alone may be relatively crude indicators of how officers relate to civilians or behave on the job. Why do people of distinct racial identities come to divergent responses to the same institution? Some studies have emphasized the role of lived experiences of blacks and whites, with the former far more intensely scrutinized and confronted by law enforcement agencies during their life course (Peffley and Hurwitz 2010; Alvarado 2020). Prior information and expectations established from these experiences set apart the ways black and white citizens process events of police violence (Jefferson, Neuner, and Pasek 2021). Attention is also called to the possibility that cultural or social processes can drive blacks and whites to different conclusions even in the face of similar information. For example, racial differences in attribution between structure versus agency may matter: whereas blacks tend to see police violence as product of institutional ills, whites are more likely to blame the persons killed for their wrongdoings, leaving views on police intact (Streeter 2019; Israel-Trummel and Streeter 2022).

Jefferson, Neuner, and Pasek (2021) focus on how the incidents of police violence are themselves racialized in the public mind: indeed, all cases of fatal police encounters that received intense scrutiny from the public in 2020, like that of Jacob Blake and George Floyd, involve a black victim and a white officer. With these racial images in mind, whites and blacks engage in “race-based motivated reasoning”, predisposed to blame black victims and white officers respectively as “an emotional or affective commitment” to defending their racial in-groups (Jefferson, Neuner, and Pasek 2021, 1166). In this sense, the racial gap in perceiving police exemplify the imbalance of racial dynamics within the practice of policing itself. Whites’ imagination of blacks as criminal and violent is well documented, and is shown to affect whites’ preferences on law enforcement (Hurwitz and Peffley 1997; Payne 2001; Peffley and Hurwitz 2007). Less examined, however, is the extent to which police officers are perceived by the public as a white profession.

Will a “whiter” police receive higher approval among the white population it serves? More importantly, are whites less willing to update their beliefs in the face of olice violence when such brutality is perpetrated by a police body in which whites are excessively represented? Built upon the framework of group-based motivated reasoning proposed by Jefferson, Neuner, and Pasek (2021), this study approaches the racial division on policing through the lens of racial imagery of police. Specifically, I examine whether whites’ perceptions of police are shaped by how whites are represented in local police workforce. Using the 2016 Law Enforcement Management and Administrative Statistics (LEMAS), I measure the white imagery of a place’s police by looking at how much whites’ share in police employment exceeds their percentage in local population, and link this measure to 2020 Cooperative Election Study data where a wide range of policing attitudes are documented for a large sample.

The results show that in places where whites are overly represented in police workforce than in overall population, whites present more favorable views on police performance and have a larger gap with blacks regarding perceptions of police. They are also more tolerant of police in the wake of fatal police encounters, and this effect is most salient when local police violence is racialized in terms of the racial groups it victimizes. In a word, racial imagination of police and police brutality is of great influence in shaping the current racial divide on policing.

2 Race, Representation, and Police

Beyond its substantive influence on policy outcomes, descriptive representation regarding race is also of great symbolic, intangible value (Hayes and Hibbing 2017). In theory, a descriptive linkage between public officials and their constituents should foster feelings of trust and inclusion, and thereby enhance the perceived legitimacy of related institutions (Mansbridge 1999; Phillips 2003). Either by taking demographic traits as ideological shortcuts (Sen 2017) , or influenced by other cultural forces set in motion (Dawson 1995), descriptively represented constituents are more likely to contact the elected officials (Gay 2002), and express more favorable perceptions of their performance (Gay 2002; Jones 2016). Though most studies on this topic are conducted within the context of elected or judiciary offices (Kaslovsky, Rogowski, and Stone 2021), the symbolic power of descriptive representation should be expected in more grassroots agencies like police. Just like the way voters can perceive the social imagery of more high-profile figures through visual medium (Mutz 2015) , street-level bureaucracy, with their discretion of and proximity to people’s everyday life (Lipsky 1980), can bring demographic similarity between the servants and the served to the front stage of the public mind.

Outside specific discussions on political representation, a more extensive literature of racial and ethnic politics also suggests that white attitudes on policing can be shaped by the racial appearances of police. In American politics, some policy domains are strongly associated with distinct racial groups in people’s mind (Mendelberg 2001; Valentino, Hutchings, and White 2002; Edsall and Edsall 1992), largely due to racial imageries that mass media choose in reference to social groups at stake (Gilliam and Iyengar 2000; Dixon and Linz 2000; Gilens 2009) . For example, racial minorities are perceived to be unfairly advantaged by social welfare programs, and African Americans in particular are frequently linked to the issue of crime or “law and order” (Winter 2008; Haney-López 2015).

Such racialized imageries of ostensibly non-racial categories open the door for racialized thinking to enter into a wide range of policy preferences (Valentino, Hutchings, and White 2002; Winter 2008; Tesler 2012, 2016), allow politicians to strategically exploit racial sentiments using subtle cues (Mendelberg 2001), and breed political cleavages along racial lines (Kinder and Winter 2001; Winter 2008). Prevalent as they are in the public perception, the attitudinal influence of racial imageries can still be moderated depending on whether the racial expressions conforms to or contradict the stereotypical beliefs that people have in store (Valentino, Hutchings, and White 2002; Ahler and Sood 2018). Given the power of race-group linkages in shaping public opinion, it would be a reasonable expectation that when a police agency has a greater presence of white officers in its overall workforce, thereby cultivating a “white imagery” of the profession, white residents within its jurisdiction will present more positive feelings toward the police’s performance. In combination with the insights from studies on the symbolic value of descriptive representation, I hypothesize a positive link between white presence in police and white perceptions of policing:

H1 (Perception Hypothesis)

More representation of whites in local police workforce should lead to more favorable perceptions of police among white residents.

Racial dynamics underlying whites’ favorable feelings for police are especially exemplified in the context of police brutality. Evoking huge waves of protests and debates around the nation, major cases of police violence during 2020 all involved white officers as direct perpetrators and civilians of color, primarily African Americans, as victims. This racial contrast that overlaps with the police-civilian tensions leads Jefferson, Neuner, and Pasek (2021) to more closely investigate how whites and blacks may approach the events of police violence as “racial partisans” in commit to defending their racial in-groups. With images of white police officers and black victims in mind, they argue, whites and blacks engage in “race-based motivated reasoning” where people bias their processing of information to achieve a more favorable conclusion of their own race. For whites, this means to discount information of the white officers’ misconduct and concentrate on the decedent’s wrongdoings, which, in turn, will render whites less willing to update their beliefs in accordance with the reality of police violence. Extending this framework from the scope of specific cases to a more general context, we would expect that the in-group defense in white response to police violence is contingent upon the degree to which police workforce are perceived as white:

H2 (Defense Hypothesis)

When local police officers present a stronger white imagery, the occurrence of police violence will be less effective in persuading whites into more skeptical attitudes toward police.

Racial contrast in the occurrence of police brutality is conditioned by not only the racial compositions of police officers but also racial imagery on the end of victims. More importantly, far from immutable entities, the awareness and practice of racial identities are fluid social constructs contingent upon specific contexts (Sen and Wasow 2016), better understood as a perspective than a standalone consideration (Cramer 2020). A sharper comparison of races between police officers and victims can bring race into the conversation of police violence and altogether enhance the weight that racial considerations have in shaping people’s views. Therefore, if the explanation centering on the role of racial imagery is valid for whites’ tolerance of police brutality, then such racially motivated in-group defense should be most overt when the overall situation is most racialized:

H2a (Racialization Hypothesis)

The moderation effect of police’s white imagery on white response to police violence will be strongest when the cases of such violence are racialized regarding their victims.

3 Empirical Strategy

3.1 Attitudes on Policing

To examine how racial imagery of local police shapes whites’ perception of policing and their attitudinal reactions to police violence, a data source regarding the outcome variable needs to meet two requirements: first, a diverse range of questions should be included that pertains to the respondent’s attitudes on the performance and the role of police. Second, the place of the respondent’s residence must be disclosed in the dataset so that I can merge their policing attitudes to the local context of police representation and police violence. The 2020 Cooperative Election Study (CES) meets both conditions. Unlike other common public opinion surveys like American National Election Studies or General Social Survey where a respondent’s geographic information below the state level is highly restricted for public access, CES of every year discloses its respondent’s place of residence as detailed as down to the zip code level. Since the majority of police departments in the United States are funded and operated on a municipality basis, the precision of residence reported in CES allows us to match respondents with the local context most immediate to their living experiences. On the other hand, in the wake of the murder of George Floyd murder in the June of 2020, the 2020 CES added to its questionnaire a battery of questions regarding the respondent’s perception of policing.

Figure 1: Attitudes toward Police by Race. All respondents in 2020 CES. Black error bars at the top of columns indicate the 95% confidence interval of the estimated attitude. Within each attitudinal outcome, the difference between non-Hispanic whites and other respondents is significant at the level of 0.05.

On a binary scale of yes or no, the 2020 CES asks the respondents whether the presence of police makes them feel safe (“police felt as safe”), whether they support increasing police funding at the expense of sacrificing some budget for other public services (“increase police”), and whether they are for decreasing police funding to better support other public programs (“decrease police”). While the first question tries to directly measure one’s given perception of local police, the latter two capture the respondent’s attitudes on how police should be maintained or regulated. To ease our interpretation of later analyses, I linearly coerced the original responses so that for each question, 1 indicates an affirmative response and 0 a negative one.

Figure 1 shows descriptive statistics of attitudes toward police by racial groups. A clear racial divide stands out: Compared to people of color, non-Hispanic white respondents in 2020 CES are more likely to feel police as an addition to their safety, more likely to support increasing the presence of police, and less likely to endorse reducing the number of police officers. Consistent with our perception in the real world, such differences on police attitudes are all significant at the level of 0.05.

3.2 Racial Imagery of Local Police

By the time when this study was conducted, there is no data source that captures the universe of racial compositions of all police agencies in the United States. To measure representation of different racial groups in local police agencies, this study needs the racial breakdown of police officers of a given police agency on the one hand, and racial structure of the locality it serves on the other. For the first requirement, I choose the 2016 Law Enforcement Management and Administrative Statistics (LEMAS) that were collected by the Bureau of Justice Statistics under the U.S. Department of Justice. Started in 1987, LEMAS periodically collects data from more than 3000 state and local law enforcement agencies, including all those that employ over 100 sworn officers as well as a nationally representative sample of smaller police agencies. Among the agency-level characteristics surveyed by LEMAS are the demographic structure of their employees, the place they serve, and their address in terms of zip code. Figure 2 shows the county-level coverage of 2016 LEMAS sampling, where counties are colored blue if at least one police department within its jurisdiction is surveyed. Though apparently having sampled more police departments in more populated areas, the 2016 LEMAS in general has a geographically comprehensive coverage. Since most municipalities in the United States are served by only one police agency, it is reasonable to infer the police’s racial imagery of a place by looking at the racial compositions of its local police employment. I extract from the 2016 LEMAS data the total number of sworn officers in a police agency and the respective numbers of white, Hispanic, and Black officers, and then calculates the share of each racial group in the agency’ employment.

Figure 2: Geographic Coverage of 2016 LEMAS at the County Level. Counties are colored blue where at least one police department within its jurisdiction is surveyed in 2016 LEMAS.

Given that the share of a racial group in police employment is a general reflection of its percentage in local population, we can hardly distinguish the effect of demographic structure from that of racial composition of police officers if simply looking at the absolute value of racial shares. A more precise approach would be to measure racial representation in police by measuring the extent to which the presence of a racial group exceeds or falls short of its presence in the local population. This study hence measures racial representations in police using the gap between the shares of a racial group in police workforce and in the total population of a place. On a positive-negative spectrum, a large value of this difference for a racial group means that in this place, a racial group is better represented in a police agency. For example, if white Americans count for 70 percent of total population of a city but over 90 percent of the city’s police officers are white, then black representation in police of the city would be 0.20, meaning an excess of representation for whites in the police workforce and thereby a “whiter” appearance of police officers.

Figure 3: Distribution of Racial Presence among Police Departments Surveyed in LEMAS 2016. Racial imagery is measured by the difference between a racial group’s share in police officers and in local population. A positive value indicates that the corresponding racial group is excessively represented in local police departments, and a negative value the otherwise.

Figure 3 shows the distribution of racial representations among the police departments surveyed in 2016 LEMAS. For most police departments, whites are excessively represented in its workforce, whose share notably exceeds their percentage in the overall population. The median level of white representation is greater than 0 and its distribution is apparently right-skewed. For a considerable number of departments, the level of excessive representation is beyond 0.2, creating a strongly whiter imagery of police officers. Representations for African and Hispanic Americans, on the other hand, present a opposite pattern. A dominant number of police departments employ a smaller share of black or Hispanic police officers than their presence in the population. The descriptive statistics found in 2016 LEMAS are consistent with the journalist and anecdotal evidence that police employment is not representative of the population it serves, forging a white imagery.

3.3 Local Police Violence

There have been no successful efforts by governmental agencies at either local or federal level to systematically document police-involved homicides. So existing literature addressing police violence has usually turned to data voluntarily collected by advocacy groups, the most used of which is that of Mapping Police Violence project. Drawing data points from other reliable sources of police-involved shootings and collecting its unique data through diverse channels including social media, local newspaper accounts, and police reports, the Mapping Police Violence project has by far the most detailed and accurate record of fatal shootings by police since 2013.

Mapping Police Violence project reports where every police-involved fatal shootings took place at zip code level along with what police agency is responsible for the shooting. This allows this study to connect CES respondents with their local context of police violence using zip code and police agency information. To capture the occurrence of police violence that are most immediate to a respondent’s memory, I focus on the those police-involved fatal shootings that took place in 2020 before September 29th, the date of the first surveying of 2020 CES. The vast majority of the places where police violence happened saw only one incident of such during the range of my observation, and outliners with more than two police-involved homicides are rare, counting for less than one percent of all observations. For this reason, I measure the level of police violence on a binary basis, with 1 that there is at least one case of police violence recorded in a place.

4 Results

4.1 White Attitudes on Policing

We first examine the Perception Hypothesis, that is, more representation of whites in local police workforce should lead to more favorable perceptions of police among white residents. Figure 4 shows the estimates on the relationship between racialized imagery of police workforce and whites’ attitudes on policing. Based on an individual regression among white respondents of 2020 CES, each point represents the estimated effect of the related racial imagery (indicated by the horizontal axis) on one’s policing attitudes. As whites’ presence in local police workforce increases relative to its share in local population, whereby a “whiter” imagery of police officers emerges, white Americans are more likely to claim that police make them feel safe, and they are less willing to decrease the funding for police to support other public services. No significant effect of white imagery was found for the attitude on increasing police funding, but the positive sign of the point estimate is consistent with our expectation. In contrast, increased representation of African or Hispanic Americans, which results in a police team that looks less like white people, is associated with more negative feelings toward policing among white Respondents. In either case, whites are less likely to perceive police as safe or support an increase in police funding, and are more likely to support decreasing police resources.

Figure 4: The Estiamted Relationship between Racial Imagery of Local Police and White Attitudes on Policing. Each point indicates the coefficient estimate of the realted racial imagery on policing attitudes out of an invididual regression among white respondents in 2020 CES. 95% confidence intervals are shown by the range. The strip of each panel indicates the outcome variable of interest. Positive estimates are colored blue, and significant estimates (p < 0.05) are marked in bold text. White percentage in local population is controlled for.

Analyses in Table 1 further looks into the extent to which overall whiter imagery of police workforce contributes to the current racial divide on policing attitudes. Here respondents of all racial identities are included in the regressions. Taking the value of the respondent’s white identity, the term Racial Divide captures the systemic difference between whites and people of color regarding their attitudes on policing. Further, the interaction term between racial divide and police’s white imagery, when combined with the standalone term of racial divide, estimates how racial imagery of police workforce moderates the racial gap. Results indicate that regardless of the level of police’s white imagery, whites in general are more favorable of police across the three attitudinal indicators. The racial gap on feeling police as safe is exacerbated by a whiter imagery of police, while the gaps on increasing or decreasing police funding are not.

Table 1: Racial Imagery of Local Police Moderates Racial Divides on Policing Attitudes

	Police Felt as Safe	Increase Police	Decrease Police
Racial Divide	0.130***	0.033**	−0.055***
	(0.007)	(0.010)	(0.010)
White Imagery of Police	−0.090***	−0.006	−0.010
	(0.019)	(0.028)	(0.029)
Racial Divide × White Imagery	0.120***	0.020	−0.041
	(0.023)	(0.035)	(0.036)
Observations	39551	39597	39589
R squared	0.061	0.001	0.005
Note: All respondents in 2020 CES. The term Racial Divide takes the value of the respondents white identity, 1 if white and 0 if not white, thereby capturing the difference of whites on policing attitudes as compared to people of color. Robust standard errors in parentheses. + p

But when we are concerned about not the continuous variation of racial gap but whether there is a gap at all, a different landscape shows up. Building upon the OLS results in Table 1, Figure 5 presents whether the racial gap on policing existed as a function of white imagery of police. Ranges where the effect of white identity is not significantly from 0 at the level of 0.95 are colored red and indicated by blue reference lines. As the white imagery of local police decreases, we see that the racial gap in feeling police as safe notably declines and eventually diminished when the police workforce are least white. For the attitudes on increasing or decreasing police funding, even though less variation across levels of racial imagery is found, it is still clear that when whites become insufficiently represented in police workforce, the racial gap slightly declines and are eventually not significant. This means that in places where police officers appear less white, whites and people of color are generally not so divided or not divided at all with regard to their perceptions of and visions for policing.

Figure 5: Racial Imagery of Local Police Moderates Racial Divide regarding Attitudes on Policing. Each plot shows the racial divide on the outcome policing attitude as the white imagery of local police increases. Ranges of racial imagery where racial divide diminishes are colored red and indicated by blue reference lines. The thicker black bar represents the observed range of white imagery of police in our data.

The results from the analyses above build up our confidence that racial imagery of police workforce, specifically racial compositions of local police officers relative to the racial structure of local population, is a shaping force on white people’s attitudes on policing. As the police team gets “whiter”, whites are more favorable of the role of police. Racial imagery of police also affects the racial gap on policing attitudes, places with higher representation of whites in police employment seeing a larger racial divide. Based on these finding, in the next part we turn to how racial imagery of police moderates white people’s attitudinal responses in the after of police-involved fatal shootings.

4.2 White Reaction to Police Violence

Now we turn to the Defense Hypothesis that expects the occurrence of police violence will be less effective in persuading whites into more skeptical attitudes toward police when local police officers present a stronger white imagery. Table 2 shows the moderating effect of racial imagery of police on white people’s reaction to local police violence. Quite intuitively, the occurrence of police violence during has a negative impact on white’s perception of policing, making them less likely to feel police as safe or support increasing police funding, and more likely to consider cutting police funding in support of other public services. But this attitudinal reaction of whites to police violence is strongly moderated by the racial imagery of local police, with the interaction term significantly going against the sign of the main term. As local police present more of a white imagery, the happening of police violence is less effective in pushing whites toward more negative views on policing.

Plotting linear predicted values of policing attitudes based on results in Table 2, Figure 6 communicates this moderating effect of racial imagery of police in more straightforward way. At each given level of white imagery of police, the distance between the black point and the green point indicates the impact of police violence occurrence on white people’s policing attitudes. We can see that across the three outcome variables, the happening of police violence results in the greatest shift of white attitudes when whites are insufficiently represented in police workforce (-0.1). As the representation level of whites in police rises to 0.2, thereby producing a “whiter” imagery of police, the attitudinal shift is weakened but still existing. When whites are overwhelmingly over-represented in police, however, almost no difference is observed between those facing local police violence and those who are not, indicating a diminishing of the persuasion effect of police violence occurrence. Overall, greater white presence in police workforce renders whites of a place more resistant to updating their attitudes on policing even in the face of the occurrence of police-involved fatal shootings.

Table 2: Racial Imagery of Local Police Moderates Racial Reaction to Police Violence

	Police Felt as Safe	Increase Police	Decrease Police
Any Police Violence in 2020	−0.047***	−0.057***	0.074***
	(0.008)	(0.013)	(0.013)
White Imagery of Police	0.006	−0.009	−0.014
	(0.018)	(0.029)	(0.029)
Police Violence × White Imagery	0.087**	0.078+	−0.142**
	(0.028)	(0.045)	(0.045)
Observations	26016	26039	26036
R squared	0.004	0.002	0.003
Note: Non-Hispanic white respondents only, 2020 CES. Robust standard errors in parentheses. + p

Figure 6: Racial Imagery of Local Police Moderates Whites’ Attitudinal Response to Police Violence. Based on the previous OLS analyses with interaction terms, points here represent linear predicted values of policing attitudes at different levels of police violence occurrence and racial imagery of local police. At the each level of racial imagery, whether there police violence happended in 2020 is indicated by the black-green scheme.

4.3 Examining the Racial Component

So far we have observed that as white presence in police workforce increases, whites will present more favorable views toward policing, widen their gap on policing attitudes compared to people of color, and are more resistant to updating their attitudes in the aftermath of police violence. But how much of this association be attributed to the racial in-group favoritism fostered among by whites by a “whiter” imagery of police officers? This brings us to the empirical test of the Racialization Hypothesis, which contends that the moderation effect of police’s white imagery on white response to police violence will be strongest when the cases of such violence are racialized regarding their victims. Since 2020 CES does not include in its questionnaire in-depth questions regarding one’s racial attitudes or racial consciousness, this study is unable to directly validate the mechanism. But if it can be shown that the effect of police’s white imagery is strongest in situations where racial dynamics are emphasized or policing itself is scrutinized racially, we can have some confidence that racial consideration is to some extent what drives whites to favor police at the presence of a “whiter” police workforce.

Table 3: Racial Imagery of Local Police Moderates Whites’ Attitudinal Response to Police Violence

	Police Felt as Safe	Increase Police	Decrease Police
PV Whites	−0.043***	−0.039*	0.056***
	(0.010)	(0.016)	(0.016)
PV POC	−0.051***	−0.077***	0.093***
	(0.010)	(0.016)	(0.016)
White Imagery of Police	0.005	−0.013	−0.010
	(0.018)	(0.029)	(0.029)
White Imagery × PV Whites	0.065+	−0.003	−0.075
	(0.036)	(0.057)	(0.057)
White Imagery × PV POC	0.109**	0.162**	−0.210***
	(0.035)	(0.058)	(0.058)
Observations	26016	26039	26036
R squared	0.004	0.002	0.003
Note: Non-Hispanic white respondents only, 2020 CES. PV Whites means at least one victim of local police violence during 2020 is white. PV POC means that all victims of local police violence during 2020 are people of color. Robust standard errors in parentheses. + p

Based upon analyses into how racial imagery of police moderates white’s attitudinal response to police violence, this study further separates the respondents into three categories based on whether police violence took place in their locality during 2020, and the racial groups impacted by such police violence. The first group of respondents is those who do not have police-involved fatal shootings in their place of residence (“No PV”). The second group live in places where at least one case of police violence happened, but at least one victim impacted by such violence is white (“PV Whites”). Similar to the second group, the third group also sees at least one incident of police violence during 2020, but all racially identifiable victims are people of color (“PV POC”). Police violence are more likely to activate concerns of racial justice when predominantly involving people of color. Therefore, though both affected by police violence, the third group differs from the second in the sense that racial consciousness may be more prominent to one’s thinking in a more racialized context of police violence. If there is a racial component in the relationship between police’s racial imagery and white attitudes on policing, we would expect that the moderation effect of racialized police imagery may be stronger among the third group than the second.

Figure 7: Moderating Effect of Racial Imagery Depends upon Racial Groups Victimized by Police Violence. Based on the previous OLS analyses with interaction terms, points here represent linear predicted values of policing attitudes at different types of police violence victimization and levels of racial imagery of local police. At the each level of racial imagery, the racial groups impacted by local police violence in 2020 are indicated by the black-to-yellow scheme.

Table 3 shows the results of our analyses. Whites’ attitudes on policing become less favorable both when police violence impacts purely people of color and when such violence involves at least one white victim. But the moderation effect of white imagery of police is strongest or only present in the case of highly racialized police violence. For the outcome variable of the respondent feeling police as safe, whites are less likely to negatively update their views in the face of police violence if police workforce sees better representation for whites. This countering effect of racial imagery is stronger when police violence purely impacts people of color than when it involves one white victim. For the attitudes on increasing or decreasing police, however, the moderation effect of police’s white imagery vanishes if at least one white victim is involved in police violence.

Figure 7 illustrates more clearly the heterogeneity of moderation effect of racial imagery contingent upon the racialization of police violence. Plotting linear predicted values of policing attitudes based on results in Table 3, Figure 7 shows at given levels of white imagery of police, how white attitudes change in response to different types of police violence, which are indicated by a group of points with different colors. When white imagery of police violence is low (0), we see that police violence has great persuasion effect regardless of their victims, resulting in a notable distance from the baseline points of no police violence. As white imagery of police violence rises to 0.5, thereby fostering the perception of police officers as predominantly white, we see that the attitudinal shit caused by police violence is smaller. When police violence involves at least one white victim, the shift, though small, is still observable in terms of point estimates. But when police violence purely victimizes people of color, we see that the occurrence of such violence not only fails to shift policing attitudes negatively, but even pushes the point estimates toward more favorable direction. To summarize, the fact that the moderation effect of racialized police imagery is contingent upon the racialization of police violence lends confidence to our hypothesis that a racial component lies underneath the association between racial imagery of police and white attitudes on policing.

5 Conclusion

Looking into how racial imagery of local police moderates the way white respondents perceive police and respond to police violence, this study offers an identity-oriented approach to disentangling the racial divide featuring debates on police. This study shows that whites’ positive perceptions of police are associated with better representation of whites in police workforce that creates a white imagery of the profession. In places with more of a white imagery in police, a larger racial divide on policing attitudes emerges. White imagery of police workforce also repels whites from updating their beliefs about police in the face of police brutality events, especially when such events are racialized by the contrast between a white officer and non-white victim. In a word, racial imagination of police plays a powerful role in shaping the current racial divide on policing.

Since most data sources that capture detailed feelings and experiences with police forces fail to disclose residence information of their respondents, the author is rather constrained to investigate in depth those psychological processes linking racial representation to one’s perception of local police. Even though the explanation based on racial imagery is largely self-containing, more specific questions remain including what factors influence one’s processing of such imagery, and what considerations may outweigh the moderation of racial representation when one is understanding police violence. Besides, because the number of Black and Hispanic respondents is too small to generate enough statistical power after geographic matching in 2020 CES, this study does not discuss how people of color may also approach police violence through a lens of racial representation. People of color are in general better informed about the reality of police brutality, but will they, like in the case of whites, perceive police more favorably if local police is more representative of non-whites? Similarly, will people of color be less triggered by the occurrence of police brutality if the victim is a white killed by a non-white officer? Empirical examination of these questions is important, and will be possible in the future with more detailed and comprehensive data.

This study also generates practical implications for addressing police brutality in real life. Since people’s perceptions of police and police violence are so strongly conditioned by whether they see local police as representative of themselves, it would help if we first repair the representation gap in police employment. This helps foster a non-racial imagination of police workforce so that people can approach the issue of police violence through a more pragmatic and less racialized perspective. Behind the veil of ignorance where police are not perceived as more of a agent of one race than another (Rawls 1999), it is more likely that people with different racial identities can have a meaningful and honest conversation in full emancipation from racial tribalism.

6 Data Descriptions

References

Ahler, Douglas J., and Gaurav Sood. 2018. “The Parties in Our Heads: Misperceptions about Party Composition and Their Consequences.” The Journal of Politics 80 (3): 964–81. https://doi.org/10.1086/697253.

Alvarado, Steven Elías. 2020. “The Complexities of Race and Place: Childhood Neighborhood Disadvantage and Adult Incarceration for Whites, Blacks, and Latinos.” Socius: Sociological Research for a Dynamic World 6 (January): 1–16. https://doi.org/10.1177/2378023120927154.

Cramer, Katherine. 2020. “Understanding the Role of Racism in Contemporary US Public Opinion.” Annual Review of Political Science 23 (1): 153–69. https://doi.org/10.1146/annurev-polisci-060418-042842.

Dawson, Michael C. 1995. Behind the mule: race and class in African-American politics. 1. paperback print. Princeton, NJ: Princeton Univ. Press.

Desilver, Drew, Michael Lipka, and Dalia Fahmy. 2020. “10 Things We Know about Race and Policing in the u.s.” https://www.pewresearch.org/fact-tank/2020/06/03/10-things-we-know-about-race-and-policing-in-the-u-s/.

Dixon, Travis L., and Daniel Linz. 2000. “Overrepresentation and Underrepresentation of African Americans and Latinos as Lawbreakers on Television News.” Journal of Communication 50 (2): 131–54. https://doi.org/10.1111/j.1460-2466.2000.tb02845.x.

Edsall, Thomas Byrne, and Mary D. Edsall. 1992. Chain reaction: the impact of race, rights, and taxes on American politics; with a new afterword. 1. publ. New York, NY: Norton.

Gay, Claudine. 2002. “Spirals of Trust? The Effect of Descriptive Representation on the Relationship Between Citizens and Their Government.” American Journal of Political Science 46 (4): 717. https://doi.org/10.2307/3088429.

Gilens, Martin. 2009. Why Americans hate welfare: race, media, and the politics of antipoverty policy. Chicago: University of Chicago Press.

Gilliam, Franklin D., and Shanto Iyengar. 2000. “Prime Suspects: The Influence of Local Television News on the Viewing Public.” American Journal of Political Science 44 (3): 560–73. https://doi.org/10.2307/2669264.

Haney-López, Ian. 2015. Dog whistle politics: how coded racial appeals have reinvented racism and wrecked the middle class. Oxford; New York: Oxford University Press.

Hayes, Matthew, and Matthew V. Hibbing. 2017. “The Symbolic Benefits of Descriptive and Substantive Representation.” Political Behavior 39 (1): 31–50. https://doi.org/10.1007/s11109-016-9345-9.

Hurwitz, Jon, and Mark Peffley. 1997. “Public Perceptions of Race and Crime: The Role of Racial Stereotypes.” American Journal of Political Science 41 (2): 375. https://doi.org/10.2307/2111769.

———. 2005. “Explaining the Great Racial Divide: Perceptions of Fairness in the U.S. Criminal Justice System.” The Journal of Politics 67 (3): 762–83. https://doi.org/10.1111/j.1468-2508.2005.00338.x.

Israel-Trummel, Mackenzie, and Shea Streeter. 2022. “Police Abuse or Just Deserts? Deservingness Perceptions and State Violence.” Public Opinion Quarterly 86 (S1): 499–522. https://doi.org/10.1093/poq/nfac017.

Jefferson, Hakeem, Fabian G. Neuner, and Josh Pasek. 2021. “Seeing Blue in Black and White: Race and Perceptions of Officer-Involved Shootings.” Perspectives on Politics 19 (4): 1165–83. https://doi.org/10.1017/S1537592720003618.

Jones, Philip Edward. 2016. “Constituents’ Responses to Descriptive and Substantive Representation in Congress*: Constituents’ Responses to Descriptive and Substantive Representation.” Social Science Quarterly 97 (3): 682–98. https://doi.org/10.1111/ssqu.12243.

Kaslovsky, Jaclyn, Jon C. Rogowski, and Andrew R. Stone. 2021. “Descriptive Representation and Public Support for Supreme Court Nominees.” Political Science Research and Methods 9 (3): 583–98. https://doi.org/10.1017/psrm.2019.59.

Kinder, Donald R., and Nicholas Winter. 2001. “Exploring the Racial Divide: Blacks, Whites, and Opinion on National Policy.” American Journal of Political Science 45 (2): 439. https://doi.org/10.2307/2669351.

Lipsky, Michael. 1980. Street-level bureaucracy: dilemmas of the individual in public services. New York: Russell Sage Foundation.

Mansbridge, Jane. 1999. “Should Blacks Represent Blacks and Women Represent Women? A Contingent "Yes".” The Journal of Politics 61 (3): 628–57. https://doi.org/10.2307/2647821.

Mendelberg, Tali. 2001. The Race Card: Campaign Strategy, Implicit Messages, and the Norm of Equality. Princeton, N.J: Princeton University Press.

Morin, Rich, and Renee Stepler. 2016. “The Racial Confidence Gap in Police Performance.” https://www.pewresearch.org/social-trends/2016/09/29/the-racial-confidence-gap-in-police-performance/.

Mutz, Diana Carole. 2015. In-Your-Face Politics: The Consequences of Uncivil Media. Princeton, New Jersey: Princeton University Press.

Payne, B. Keith. 2001. “Prejudice and Perception: The Role of Automatic and Controlled Processes in Misperceiving a Weapon.” Journal of Personality and Social Psychology 81 (2): 181–92. https://doi.org/10.1037/0022-3514.81.2.181.

Peffley, Mark, and Jon Hurwitz. 2007. “Persuasion and Resistance: Race and the Death Penalty in America.” American Journal of Political Science 51 (4): 996–1012. https://doi.org/10.1111/j.1540-5907.2007.00293.x.

———. 2010. Justice in America: The Separate Realities of Blacks and Whites. Cambridge Studies in Public Opinion and Political Psychology. Cambridge ; New York: Cambridge University Press.

Phillips, Anne. 2003. The politics of presence: the political representation of gender, ethnicity, and race. Repr. 2003. Oxford political theory. Oxford: Oxford Univ. Press.

Rawls, John. 1999. A Theory of Justice. Rev. ed. Cambridge, Mass: Belknap Press of Harvard University Press.

Sen, Maya. 2017. “How Political Signals Affect Public Support for Judicial Nominations: Evidence from a Conjoint Experiment.” Political Research Quarterly 70 (2): 374–93. https://doi.org/10.1177/1065912917695229.

Sen, Maya, and Omar Wasow. 2016. “Race as a Bundle of Sticks: Designs That Estimate Effects of Seemingly Immutable Characteristics.” Annual Review of Political Science 19 (1): 499–522. https://doi.org/10.1146/annurev-polisci-032015-010015.

Sigelman, Lee, Susan Welch, Timothy Bledsoe, and Michael Combs. 1997. “Police Brutality and Public Perceptions of Racial Discrimination: A Tale of Two Beatings.” Political Research Quarterly 50 (4): 777–91. https://doi.org/10.1177/106591299705000403.

Streeter, Shea. 2019. “The Racial Politics of Police Violence in the United States.” PhD thesis, Palo Alto, California. https://purl.stanford.edu/kf447bj0520.

Tesler, Michael. 2012. “The Spillover of Racialization into Health Care: How President Obama Polarized Public Opinion by Racial Attitudes and Race.” American Journal of Political Science 56 (3): 690–704. https://doi.org/10.1111/j.1540-5907.2011.00577.x.

———. 2016. Post-Racial or Most-Racial? Race and Politics in the Obama Era. Chicago Studies in American Politics. Chicago ; London: University of Chicago Press.

Thomas, Deja, and Juliana Menasce Horowitz. 2020. “Support for Black Lives Matter Has Decreased Since June but Remains Strong Among Black Americans.” https://www.pewresearch.org/fact-tank/2020/09/16/support-for-black-lives-matter-has-decreased-since-june-but-remains-strong-among-black-americans/.

Valentino, Nicholas A., Vincent L. Hutchings, and Ismail K. White. 2002. “Cues That Matter: How Political Ads Prime Racial Attitudes During Campaigns.” American Political Science Review 96 (1): 75–90. https://doi.org/10.1017/S0003055402004240.

Winter, Nicholas J. G. 2008. Dangerous Frames: How Ideas about Race and Gender Shape Public Opinion. Chicago: University of Chicago Press.

Footnotes

1. Chaoyue R. Wang (chyrwang@gmail.com, 86-177-2387-5369) is a senior student of Philosophy, Politics and Economics at Yuanpei College of Peking University, Beijing, China 100871. I sincerely thank Professor Yuchen Liu of the School of Government for his constructive criticism during the early stage of this study. Data, codes, and other documents necessary to reproduce the results reported in this study are openly available at https://github.com/raphaellie/racialized-police.

2. Chaoyue R. Wang (chyrwang@gmail.com, 86-177-2387-5369) is a senior student of Philosophy, Politics and Economics at Yuanpei College of Peking University, Beijing, China 100871. I sincerely thank Professor Yuchen Liu of the School of Government for his constructive criticism during the early stage of this study. Data, codes, and other documents necessary to reproduce the results reported in this study are openly available at https://github.com/raphaellie/racialized-police.

3. Chaoyue R. Wang (chyrwang@gmail.com, 86-177-2387-5369) is a senior student of Philosophy, Politics and Economics at Yuanpei College of Peking University, Beijing, China 100871. I sincerely thank Professor Yuchen Liu of the School of Government for his constructive criticism during the early stage of this study. Data, codes, and other documents necessary to reproduce the results reported in this study are openly available at https://github.com/raphaellie/racialized-police.