How can we make sense of international protests against police misconduct and anti-black racism in the summer of 2020 and the global diffusion of the Black Lives Matter (BLM) movement since its emergence in 2014? The importance of the US public's perception of local police is clear: evidence of systemic racism in law enforcement following the police killing of George Floyd in May 2020 in Minneapolis, Minnesota, for example, fueled widespread demonstrations, dramatically increased participation in the BLM movement, and may have reduced the willingness of members of the public to cooperate or comply with officers (Buchanan, Bui, and Patel Reference Buchanan, Bui and Patel2020; Jackson et al. Reference Jackson2012; Tyler and Fagan Reference Tyler and Fagan2008; Tyler and Huo Reference Tyler and Huo2002).
Although attitudes toward the police vary across social groups, black Americans typically have less trust and confidence in the police than white Americans, a demographic pattern that holds in some other national contexts, such as Canada and the UK (Bowling, Parmar, and Phillips Reference Bowling, Parmar, Phillips and Newburn2003; Brown and Benedict Reference Brown and Benedict2002; Wortley and Owusu-Bempah Reference Wortley and Owusu-Bempah2009). This is attributable to a range of factors, not least of which is the comparatively higher rates at which black persons are stopped by the police and the poorer quality of those interactions relative to white persons in these contexts (Ariza Reference Ariza2014; Tyler and Huo Reference Tyler and Huo2002; Wortley and Owusu-Bempah Reference Wortley and Owusu-Bempah2011).
While extant research tends to focus on the effect that officer behavior in direct interactions with civilians can have on attitudes toward the police, vicarious police contact—contact that individuals have through friends, family members, and neighbors—and media coverage of police violence can also powerfully affect the attitudes and behaviors of individuals and communities (Intravia, Wolff, and Piquero Reference Intravia, Wolff and Piquero2018; Walker Reference Walker2020; Weitzer and Tuch Reference Weitzer and Tuch2005). These encounters can provide evidence of systemic racism in policing, reverberate with negative personal experiences, and provoke anger, which can reduce satisfaction with, confidence in, and trust in policing. Indeed, research demonstrates that black Americans—whose attitudes toward a range of social and political policies are shaped by how they understand those policies to treat black Americans as a group—evince greater sensitivity to vicarious police interactions and coverage of police misconduct or abuse than white Americans (Epp, Maynard-Moody, and Haider-Markel Reference Epp, Maynard-Moody and Haider-Markel2014; Hurwitz, Peffley, and Mondak Reference Hurwitz, Peffley and Mondak2015; Weitzer Reference Weitzer2017).
In this article, I argue that police violence against black Americans and media coverage of that violence can affect the attitudes that black citizens in other national contexts hold toward police in their home countries. To a certain extent, this finding is anticipated by research on the black diaspora and black transnationalism, which describe persistent social and cultural linkages between Afro-descendent communities across national contexts (Gilroy Reference Gilroy1993; Patterson and Kelley Reference Patterson and Kelley2000). Importantly, this research also characterizes transnational linkages in black politics and describes how the circulation of people, ideals, and cultures across contexts can affect black politics across borders. Examples include the role that transnational black politics played in the implementation of land reforms in Columbia, affirmative action policies in Brazil, struggles against apartheid in South Africa, anti-racist activism in Canada, and the Civil Rights Movement in the United States (Paschel Reference Paschel2016; Waters Reference Waters2013; Williams Reference Williams2015). Collectively, it suggests that some black persons across national contexts view institutions of racial violence and subordination—and the experiences of black persons under them—as linked or related.
Nevertheless, the significance of police violence against black persons inside the United States for black persons outside of it is not entirely clear. Demonstrations against police violence and the emergence of BLM chapters in Canada, for example, have been characterized as simple expressions of support of black Americans in the United States, and mostly irrelevant for police–community relations in that context (Sandhu Reference Sandhu2018). Furthermore, research on transnational black politics suggests that it reached its apogee during the era of the American Civil Rights Movement, national independence movements in Africa, and the era of Black Power, and it is not clear the extent to which its matters in the 21st century.
In this article, I test the effect that police killings of unarmed black men in the United States can have on black British attitudes toward London's Metropolitan Police. I mobilize data from the Metropolitan Police Public Attitudes Survey (MPPAS), a survey measuring Londoners' trust and confidence in the police, focusing specifically on data collected in 2014 surrounding the killing of Eric Garner. Although US police officers killed hundreds of black Americans that year, I explore the particular effect of the police killing of Eric Garner, a widely covered event that accelerated social movement activism and political action against police violence (Freelon, McIlwain, and Clark Reference Freelon, McIlwain and Clark2016). I estimate the impact that this event had on black British attitudes toward the police using a regression discontinuity design (RDD) or unexpected event study design (UESD), which exploits randomization in the administration of a public attitudes survey around the police killing of Eric Garner to uncover the plausibly causal effect the killing had on attitudes (Muñoz, Falcó-Gimeno, and Hernández Reference Muñoz, Falcó-Gimeno and Hernández2020).
I find that Garner's killing had a clear and substantial effect on black Londoners' attitudes. Relative to a comparable group of black Londoners interviewed shortly before Garner's killing, black Londoners interviewed after his killing were less satisfied with the Metropolitan Police, less convinced that the police treat people fairly, and less convinced that officers understand local communities. Critically, I also find that Garner's killing had little measurable effect on the attitudes of white Londoners and no measurable effect on South Asian Londoners. At the same time, I do not uncover evidence that his death significantly affected any group's sense that the police were effective at addressing crime, responding to emergencies, or dealing with dangerous situations.
Altogether, these findings offer evidence that state violence against black persons in one national context can affect the attitudes of those in another. They contribute causal evidence of transnational linkages in a policy domain—policing—that is central to racial politics and racialization in the United States and elsewhere (Soss and Weaver Reference Soss and Weaver2017), and suggest that black participation in BLM protests outside the United States reflects attitude change toward local institutions that events in the United States precipitate. At the same time, the smaller but still significant decline in white Londoners' evaluations of police fairness suggests that while the transnational consequences of racialized police violence are larger and more significant among black persons, they are not exclusive to them, while the null effect among South Asians challenges notions of an essential non-white solidarity in Britain (Modood Reference Modood1994).
Determinants of Public Attitudes Toward the Police
Public perception of law enforcement exerts a clear and meaningful effect on behavior and attitudes. Public perception of systemic racism in law enforcement, for example, motivated widespread public demonstrations and calls for systemic reform in the United States in 2020 following the widely publicized police killing of George Floyd (Buchanan, Bui, and Patel Reference Buchanan, Bui and Patel2020). As Epp, Maynard-Moody, and Haider-Markel (Reference Epp, Maynard-Moody and Haider-Markel2014) demonstrate, police stops of black motorists can also raise racial group consciousness, particularly among those who believe police systematically discriminate against black persons. Finally, perceptions of police legitimacy can affect the likelihood that individuals will defer to police authority, cooperate with officers to solve crime, or comply with officers' orders (Tyler and Fagan Reference Tyler and Fagan2008; Tyler and Jackson Reference Tyler and Jackson2014).
Research conducted in several national contexts has uncovered a relatively consistent pattern of demographic differences in attitudes toward the police. The most well-known differences relate to race and ethnicity: while most surveyed individuals report having net positive attitudes toward the police, race and ethnic minority status consistently predict negative attitudes. In the United States, Canada, and the UK, for example, black or Afro-descendent persons report more negative perceptions of police officers than white persons (Bowling, Parmar, and Phillips Reference Bowling, Parmar, Phillips and Newburn2003; Bradford Reference Bradford2011; Brown and Benedict Reference Brown and Benedict2002; Weitzer and Tuch Reference Weitzer and Tuch2005; Wortley and Owusu-Bempah Reference Wortley and Owusu-Bempah2009).
Encounter-level factors help explain or mediate some of the relationship between race and attitudes toward the police. Encounter-level factors relate to the nature or quality of the interaction between police officers and the public, such as whether the encounter was initiated by the police officer or the citizen, whether the person received a ticket or was arrested, or whether force was used (Brown and Benedict Reference Brown and Benedict2002). Among encounter-level factors, few matter as much as procedural justice, or the extent to which police officers treat civilians with fairness, politeness, and neutrality (Sunshine and Tyler Reference Sunshine and Tyler2003; Tyler and Huo Reference Tyler and Huo2002). Irrespective of the outcome of an encounter, civilians want to be treated with dignity, competently, and without bias, and are more likely to view the police positively and with legitimacy when they believe that the police behave in this manner (Tyler and Jackson Reference Tyler and Jackson2014).
Observed racial differences in public perceptions toward the police are therefore partly a consequence of the disproportionate number of contacts with the police that black persons experience and the quality of those contacts. In the United States, Canada, and UK, black persons are far more likely to be subject to involuntary police contact and adverse contact (arrest, search, or use of force), controlling for their share of the population, rates of engagement in crime, distribution across neighborhoods, and other relevant factors (Gelman, Fagan, and Kiss Reference Gelman, Fagan and Kiss2007; Ontario Human Rights Commission 2018; Vomfell and Stewart Reference Vomfell and Stewart2021). In each of these countries, research also characterizes the poor quality of police contact with black persons, which can be more arbitrary, unprofessional, disrespectful, and abusive than the contacts that white persons have with the police (Bowling, Parmar, and Phillips Reference Bowling, Parmar, Phillips and Newburn2003; Tyler and Huo Reference Tyler and Huo2002; Wortley and Owusu-Bempah Reference Wortley and Owusu-Bempah2009).
However, these individual and encounter-level factors do not explain all of the variance in attitudes toward the police and do not totally mediate the effect of race on individual perceptions of the police. In other words, disparities in attitudes toward the police cannot totally be explained by the disproportionate contact that individual black persons have with the police or the fact that these contacts are qualitatively worse than those that white people have.
Vicarious contact with the police and depictions of police behavior in media also play an important role. Vicarious contact can affect an individual's attitudes through a variety of mechanisms: individuals can internalize the negative experiences that individuals in their social networks have with the police, may become angered by the treatment that people they care about receive, or learn about the police's function and role through the experiences of others or representation in media Walker (Reference Walker2020). Vicarious contacts can also reverberate or amplify an individual's own negative contact with the police (Rosenbaum et al. Reference Rosenbaum2005).
Individuals with friends or family members who have had contact with the police system tend to report significantly worse perceptions of the police (Rosenbaum et al. Reference Rosenbaum2005; Tyler, Fagan, and Geller Reference Tyler, Fagan and Geller2014; Weitzer and Tuch Reference Weitzer and Tuch2005). Weitzer and Tuch (Reference Weitzer and Tuch2004, Reference Weitzer and Tuch2005) show, for example, that vicarious police contact through friends and family members is associated with less overall satisfaction with the police and greater belief that the police are prejudiced and engage in racial profiling. Although vicarious contact through friends and family appears to be the most important, personal evaluation of how police treat members of the community also matters. Tyler, Fagan, and Geller (Reference Tyler, Fagan and Geller2014, 765), for example, show that individual-level perceptions of police legitimacy are powerfully shaped by evaluations of police behavior and procedural justice in their neighborhood, and that general judgments about how the police treat neighbors is almost as important to perceived legitimacy as personal experiences with the police.
Importantly, the negative effects of vicarious police contact on perceptions of the police are concentrated among black Americans. Since police disparately target black Americans for stops, concentrate proactive policing strategies in minority communities, and engage in qualitatively worse policing in those neighborhoods, black Americans are significantly more likely to report that a close family member or friend has been stopped by the police, and to know of or witness police use of force, misconduct, or harassment in their communities (Gelman, Fagan, and Kiss Reference Gelman, Fagan and Kiss2007; Tyler, Fagan, and Geller Reference Tyler, Fagan and Geller2014; Weitzer and Tuch Reference Weitzer and Tuch2006).
Black Americans are also more likely to frame the vicarious police contacts of friends, family, and neighbors in systemic terms. In other words, they are more likely to situate the contact that black friends, family, and neighbors have with the police within a framework that sees racial discrimination as endemic in US policing and as a manifestation of deep and pervasive racial inequality that spans a range of social institutions (Soss and Weaver Reference Soss and Weaver2017; Weitzer Reference Weitzer2017). Accordingly, stops of black friends, family members, and neighbors can serve as evidence of systemic anti-black racism or as examples of how they are likely to be treated by the police in potential future interactions. This research helps to explain why some empirical studies find that vicarious interactions are more pronounced or only appear in black Americans (Rosenbaum et al. Reference Rosenbaum2005; Weitzer and Tuch Reference Weitzer and Tuch2004; Weitzer and Tuch Reference Weitzer and Tuch2005).
Media coverage of police behavior can also powerfully shape attitudes and perceptions of the police, and for similar reasons. While research demonstrates that media generally affects the world view of those who consume it (Gerbner Reference Gerbner1998), individuals viewing news of the police may internalize the experiences of others, become angered by the treatment of others, or learn about the police's function. Weitzer (Reference Weitzer2002), for example, show that newspaper and television coverage of police brutality against Rodney King and others by the Los Angeles Police Department (LAPD) led to dramatic declines in trust and confidence lasting months or years. Consumption of news about police violence on the Internet, including through social media, can also lead to negative attitudes about the police (Gauthier and Graziano Reference Gauthier and Graziano2018; Intravia, Wolff, and Piquero Reference Intravia, Wolff and Piquero2018).
As with personal vicarious interactions, the negative effect of news coverage of policing is also concentrated among minority racial and ethnic groups (Graziano, Schuck, and Martin Reference Graziano, Schuck and Martin2010; Weitzer and Tuch Reference Weitzer and Tuch2004). This is likely attributable (at least in part) to the combination of racial disparities in police misconduct and violence against black Americans, and the greater affinity that black viewers of these stories have with the victims: black Americans viewing a preponderance of news stories in which they share similar demographic characteristics with victims of police violence and misconduct may fear becoming victims of police misconduct or violence themselves, or come to hold negative views of the police (Chiricos, Eschholz, and Gertz Reference Chiricos, Eschholz and Gertz1997). As with personal and vicarious interactions with the police, it is also likely that black Americans understand these news stories in systemic terms—as a consequence of severe and continued racial inequality—and as relevant to experiences they (or people they care about) may have with police officers in the future.
Transnational Consequences of Police Misconduct on Attitudes Toward the Police
This research helps frame the consequences that the high-profile police killings of unarmed black Americans has had on American attitudes toward the police (Taylor Reference Taylor2016). Starting first with Eric Garner's killing on July 17, 2014, and continuing with the police killings of Michael Brown on August 9, 2014, Ezell Ford on August 11, 2014, and many others since then, these incidents of police violence have had a sharp and deleterious effect on expressed satisfaction with police officers and perceptions of racial bias (Ekins Reference Ekins2016; Kochel Reference Kochel2019; Menasce Horowitz and Livingston Reference Menasce Horowitz and Livingston2017). As several surveys have demonstrated since then, however, the effects have been highly racialized, exerting a far stronger effect on black American attitudes toward the police than white American attitudes toward the police.
Yet, while the effect of these incidents in the United States is relatively well appreciated, little attention has been paid to the effect they can have elsewhere. I argue, however, that high-profile incidents of police violence in the United States have the potential to affect attitudes of Afro-descendent people in other national contexts for several, interconnected reasons.
First, past research on the black diaspora demonstrates that Afro-descendent persons across national contexts continue to maintain a shared cultural affinity and attachment to one another, such that black individuals in one national context identify with the experiences of black individuals in another (Edwards Reference Edwards2009; Gilroy Reference Gilroy1993; Narayan Reference Narayan2019; Patterson and Kelley Reference Patterson and Kelley2000). Research in this literature argues that the identities of Afro-descendent persons are not totally circumscribed by national community and that, at various points in time and to varying degrees, black intellectuals, political activists, and everyday citizens have viewed themselves as members of a larger international black community (Patterson and Kelley Reference Patterson and Kelley2000). Accordingly, black persons outside of the United States may identify with victims of police violence in the United States, and their attitudes toward the police may shift accordingly.
Some of this shared identity or cultural affinity has been characterized in survey research. Sizable and comparable portions of both the black American and black British population, for example, identify primarily as “black” rather than as a citizen of their country (British or American) (Laniyonu Reference Laniyonu2019). Surveys conducted in the United States similarly suggest that native-born black Americans express a sense of diasporic linked fate, or a sense that the fates of African and Afro-Caribbean immigrants in the United States have some bearing on what happens to them in their lives (Nunnally Reference Nunnally2010).
Second, black residents in many North American and European countries experience persistent discrimination, structural disadvantage, and mistreatment by the police. Vomfell and Stewart (Reference Vomfell and Stewart2021), for example, find that police officers in the UK are significantly more likely to stop and search black residents, even after controlling for the racial and ethnic composition of criminal suspects and neighborhoods. They demonstrate that this disparity in searches may be attributed both to individual-level bias among officers and to systemic racism, such as the over-patrolling of minority neighborhoods. As a consequence, black Britons are likely to have had negative contact with the police, to hold negative attitudes toward the police, and to believe that policing is systemically biased against black residents (Bowling, Parmar, and Phillips Reference Bowling, Parmar, Phillips and Newburn2003; Bradford, Jackson, and Stanko Reference Bradford, Jackson and Stanko2009). Given evidence that the negative effects of vicarious interactions with the police are stronger among individuals who have had frequent or poor interactions with the police themselves (Rosenbaum et al. Reference Rosenbaum2005), high-profile incidents of police violence and misconduct in the United States may therefore be more likely to affect black residents in other contexts specifically.
Finally, black residents in other national contexts may understand or frame police violence in the United States against black Americans as rooted in anti-black racism and shared marginalization that black citizens experience across many national contexts. Many black Americans understand and frame the police's role in political terms, and see the police as an institution that not only discriminates against them, but also actively maintains group-based inequality through violence, surveillance, and coercion (Soss and Weaver Reference Soss and Weaver2017; Weitzer Reference Weitzer2017). Critical for the present study, however, is research which demonstrates that moments of state violence against black people in one context—for example, by local police in the United States against non-violent protesters during the Civil Rights Movement or in South Africa in Sharpsville and Soweto—can galvanize and affect black people's attitudes and behavior in another (Waters Reference Waters2013; Williams Reference Williams2015).
At the same time, there are important differences in the experiences and attitudes of black persons across national contexts, and the strength of collective identification may not be as strong as suggested by diaspora studies. In a study on the effect of the BLM movement in Vancouver, Canada, for example, police officers suggested that police–community relations in Vancouver were significantly superior to those in the United States and that they did not believe that police killings in the United States exerted much, if any, effect on how local residents viewed the police (Sandhu Reference Sandhu2018). Empirical research in the UK also suggests that there are potential limits to the extent to which black Britons may identify with the experiences that black Americans have with the police, owing, in part, to ethnic and migratory differences among black Britons. Bradford (Reference Bradford2011), for example, demonstrates that while Afro-Caribbean Britons have worse attitudes toward the police compared to white Britons, African Britons have, on average, better attitudes toward the police than have both white and Afro-Caribbean Britons.
There are other reasons to doubt that police violence in the United States might affect attitudes toward the police in other contexts. In general, while black residents of such countries as Canada and the UK are disproportionately more likely to come into contact with the police and experience harm at the hands of the police relative to white residents, the per capita rate of the most severe forms of police contact are much higher in absolute terms for black Americans. The rate of fatal police shootings of black persons, for example, is several times higher in the United States than it is in Canada or the UK (Edwards, Lee, and Esposito Reference Edwards, Lee and Esposito2019; Independent Office for Police Conduct 2018; Marcoux and Nicholson Reference Marcoux and Nicholson2018).
Further, while black intellectuals and political activists were animated in much of the 20th century by conceptions that the political fate of black peoples is globally linked (for example, Pan-Africanism), it is not clear to what extent this affects attitudes and behavior today, particularly among the general public and in the “Global North.” For example, while Paschel (Reference Paschel2016) characterizes the importance that an international network of black intellectuals, activists, and nongovernmental organizations (NGOs) had in effecting the implementation of land reforms and affirmative action policies that aided Afro-Columbians and Afro-Brazilians, respectively, their study suggests that such influence is limited to countries in the “Global South.” Further, their work suggests that this sort of influence works principally on social elites, rather than everyday people.
The Current Study
On the whole, the extant literature suggests that high-profile incidents of police violence against black Americans can significantly erode their trust and confidence in the police. I argue, however, that the negative effects of vicarious police contact and portrayals of police violence in the media on black attitudes toward the police are not limited to black persons in the United States. In the current study, I mobilize data on the potential effect that police killings in the United States have on Londoners' attitudes toward their local police and hypothesize that:
Hypothesis 1: The police killing of Eric Garner will reduce black Londoners' evaluations of police.
Hypothesis 2: The police killing of Eric Garner will have no effect on white and South Asian Londoners' evaluations of the police.
Extant research suggests that trust and confidence in the police is composed of multiple components, including general satisfaction, perceptions of fairness, perceptions of community engagement, and perceptions of efficacy at solving crime (Jackson and Bradford Reference Jackson and Bradford2010). Public opinion, however, tends to be driven by general affect and impressions, so events that erode attitudes toward the police in general might affect all of these components. Empirical literature on public satisfaction with the police, however, suggests that major events do not move these components uniformly in direction or magnitude (Hohl, Stanko, and Newburn Reference Hohl, Stanko and Newburn2013). Thus, while overall satisfaction with the police and perceptions of fairness may be sensitive to police killings abroad, black Londoners' perceptions of police efficacy at solving crime or engaging with the community may not necessarily be affected, as violence against black persons may not clearly be connected to police performance in these domains. I therefore hypothesize that:
Hypothesis 3: Task-specific evaluations of police behavior and police work—such as effectiveness in fighting crime or engaging with the community—will be unaffected by police killings in the United States, while overall evaluations and perceptions of fairness will.
Analytic Strategy
I estimate the causal effect that police violence in the United States may have had on attitudes toward the police in London using an RDD or UESD, which can recover plausibly causal effects when experiments are unfeasible or undesirable (Jacob et al. Reference Jacob2012; Muñoz, Falcó-Gimeno, and Hernández Reference Muñoz, Falcó-Gimeno and Hernández2020). Studies of the political effects of police violence or misconduct may suffer from well-known issues of omitted variable bias, where some unmeasured factor determines both the chances than an individual receives some “treatment” and the outcome of interest. In this application, individuals in London may not be randomly exposed to images or news of police violence in the United States, and if such exposure is correlated with attitudes toward the police, then estimates of the effect of exposure to attitudes—such as those that might be generated from a cross-sectional survey—will be biased. Although some studies have randomized exposure to fictionalized police contact or violence in lab settings, and have usefully explored the emotional mechanisms through which this contact can affect some outcomes of interest (see, for example, Smith, Lopez, and Krishnamurthy Reference Smith, Lopez and Krishnamurthy2018), randomly exposing individuals to real-world instances of police violence is both impossible and unethical. Further, exposure done in a controlled lab setting may not approximate real-word exposure.
The identification strategy used here approximates as-if random exposure to police violence by comparing attitudes toward the police among respondents who were interviewed for a local survey in a narrow window before the killing of Eric Garner with those interviewed in a narrow window after it. This identification strategy assumes that who was interviewed and when they were interviewed for the survey are uncorrelated with attitudes toward the police, a highly plausible assumption given the sampling methodology used in the dataset used here (described in more detail later). In short, since interviewees were selected at random among residents of city and the timing of the interview was determined randomly, the date of the interview can be used to estimate the causal effect of police violence in the United States on attitudes in the UK.
I therefore estimate the effect that the killing of Eric Garner in the United States had on attitudes toward the police in an RDD where the date of interview defines the score, the date when Eric Garner was killed—on July 17, 2014—defines the cutoff, and units are said to be treated if they were interviewed after Eric Garner's death. I estimate the local average treatment effect of Garner's killing using the method of local linear approximation described in Cattaneo, Idrobo, and Titiunik (Reference Cattaneo, Idrobo and Titiunik2019), which fits a local linear regression to subjects interviewed before and after Garner's killing separately within a narrow window or bandwidth of the cutoff. Bandwidths used and presented in the analysis are mean-squared error (MSE) optimal bandwidths, also described in Cattaneo, Idrobo, and Titiunik (Reference Cattaneo, Idrobo and Titiunik2019).Footnote 1
A key assumption needed to interpret the results of the analysis as the causal effect of police killings in the United States on black British attitudes is that no other factor that could affect local attitudes toward the police changed systematically at the cut point (Muñoz, Falcó-Gimeno, and Hernández Reference Muñoz, Falcó-Gimeno and Hernández2020). I conducted a qualitative review of several major British newspaper issues published in the immediate aftermath of Garner's death and did not uncover the occurrence of major changes to police policy, criminal events, or local instances of misconduct or brutality that might otherwise explain the result. Interpretation also relies on the assumption that assignment to treatment and control conditions is independent of potential outcomes, the plausibility of which I test in several analyses later.
Data
To test these hypotheses, I mobilize data from the MPPAS. The MPPAS is a face-to-face survey that has been conducted annually since 1983 and has been in its current from—with the same target sample size of 12,8000 interviews—since 2011. The survey is a rolling survey and has been in the field continuously since 2011. It aims to provide local police and government with representative data on local attitudes toward the police.
The sampling frame used for the study is a list of residential addresses drawn from the Royal Mail's Postcode Address File, a database that contains all known delivery points and postcodes. This list of residential addresses was stratified by borough, and then, using a random starting point in the list, a one in n selection was made to randomly select addresses to be included in the sample. Addresses were randomly assigned months to be interviewed at the beginning of the survey period with equal probability using a similar one in n method. Each month and for each borough, approximately three times the number of addresses required were in circulation. Those addresses that were still valid (for example, those who had not refused to participate and those who had not been interviewed) at the end of one month were carried forward into the following month. Selected households were not recontacted if they refused to participate, and surveyors attempted to contact households three times before removing them from the list. Interviews could only be conducted with the addresses issued to the interviewers; interviewers could not replace any addresses by going next door or across the road, for example. At each household, respondents were selected randomly by identifying the person whose next birthday was closest to the date of the interviewer's visit.
Variable Coding
The MPPAS asks a range of questions concerning public attitudes toward the local police. While the MPPAS does not ask a direct question racial or ethnic discrimination, it does ask a range of questions that can nevertheless be used to evaluate trust and confidence. Following other studies that leverage the MPPAS to measure Londoners' attitudes toward the police, I construct a range of measures tapping public satisfaction, perception of police fairness, perception of police effectiveness, and community engagement (Hohl, Stanko, and Newburn Reference Hohl, Stanko and Newburn2013).
Public satisfaction is a standard, overall measure of public confidence in the ability of the police to execute their roles, with a question that asks: “Taking everything into account, how good a job do you think the police in London as a whole are doing?” This item was measured on a five-point scale ranging from “poor” to “excellent.” For interpretability, this outcome, along with all other index measures described later, were rescaled so that neutral response values (“fair”) equal 0, maximum positive values (“excellent”) equal 5 and maximum negative values (“very poor”) equal −5.
Perception of police fairness is measured using four items that measure the extent to which respondents agree that police officers: would treat the respondent with respect if they had contact with them for any reason; would treat people in the local community fairly regardless of who they are; are friendly and approachable; and are helpful. Individual responses were measured using a five-point Likert scale ranging from “strongly disagree” to “strongly agree.”
Perception of police effectiveness is measured using five items that ask respondents how well they believe the police deal with: tackling gun crime; supporting victims and witnesses of crimes; policing major events in London; tackling dangerous driving; and promptly responding to emergencies. Individual responses were measured using a seven-point Likert scale ranging from “not at all well” to “very well.”
Community engagement is measured using four items that ask respondents the extent to which they agree or disagree that the police: listen to the concerns of local community members; understand issues facing the local community; are dealing with issues that matter to the local community; and can be relied upon. Items were measured using five-point Likert scale ranging from “strongly disagree” to “strongly agree.”
The forcing variable—distance from the date of Eric Garner's death—was calculated as the distance between the interview date and July 17, 2014. Black respondents were coded as individuals who noted African, Caribbean, or any other Black, African, or Caribbean background as part of their ethnic ancestry. White respondents were individuals who identified as White English, Welsh, Scottish, Northern Irish, British, Irish, Gypsy or Irish Traveller, or any other White background. Asian respondents were individuals who identified as Indian, Pakistani, or Bangladeshi.
Results
Discontinuous Changes in Predetermined Covariates at the Cut Point
Estimation of the causal effect that Garner's killing had on black Londoners' attitudes relies on the assumption that assignment to treatment and control conditions is independent of potential outcomes, such that interviewees could not manipulate when they were interviewed and treatment and control groups are balanced on observed and unobserved covariates. While the design of the MPPAS suggests that this assumption is satisfied, this is not guaranteed. As a test, Cattaneo, Idrobo, and Titiunik (Reference Cattaneo, Idrobo and Titiunik2019) recommend testing for discontinuous shifts in predetermined covariates at the cut point using the same methods applied to the outcomes of interest.Footnote 2
Table 1 presents the results of such an analysis run on a set of predetermined covariates available in the MPPAS. The results do not give an indication of a discontinuous shift in any of the covariates following Garner's killing, with the exception of the share of 45- to 54-year-olds in the treated group, which increases significantly following the killing. While it is unlikely that this age cohort and no other sorted around the threshold, attitudes toward the police do vary systematically by age, with older persons typically reporting more favorable attitudes toward the police than younger persons. Accordingly, I include the share of 45- to 54-year-olds as a control variable in a secondary set of analyses.Footnote 3 Estimates of regression discontinuity effects on the neighborhood respondents live in—to test whether there is a discontinuous shift in the composition of the sample by neighborhood—are depicted in the Online Supplementary Appendix and reveal no such shifts among any racial group.Footnote 4
Table 1. Estimates of regression discontinuity effects on predetermined covariates, black Londoners
Notes: Covariates in bold are significant at α ≤ 0.05. Robust standard errors (SEs) and robust confidence intervals (CIs) are estimated using the methods described in Cattaneo, Idrobo, and Titiunik (Reference Cattaneo, Idrobo and Titiunik2019) and implemented in Calonico et al. (Reference Calonico2017). Standard errors are clustered at the borough level. Tables depicting estimates of the regression discontinuity effect on predetermined covariates among in the white and South Asian subsamples can be found in the Online Supplementary Appendix.
Consumption of American Police Shootings in the UK
For Garner's death to have affected black Londoners’ attitudes toward the police, they must have been aware of the shootings in the first place. A qualitative analysis of British media suggests that traditional print media in the UK extensively covered Garner's killing. For example, most of the of the major print publications in the country (for example, The Guardian, The Daily Telegraph, and The Times) and the British Broadcasting Corporation (BBC) carried stories of his death and subsequent protests soon after his killing, though not as extensively as they covered subsequent killings, such as the killing of Eric Garner or protests following the killing of Freddie Gray (Associated Press 2014; Swaine Reference Swaine2014; Wilson Reference Wilson2014). Coverage was far more widespread and immediate on digital and social media. Analysis of individual engagement with these killings on the Internet platform Twitter, for example, suggests that they were intense, particularly in major cities in the UK with large concentrations of black residents, and emerged in the immediate aftermath of the killings and the circulation of the videos online (Flynn Reference Flynn2014).
To provide further evidence that residents of London consumed information about the police killings in the United States, I display results from analyses of Google Trends data. Google Trends is a tool that provides information about trends and patterns in the search terms queried by users of their search engine. Google Trends analysis normalizes search term frequency over the period analyzed. As such, dates and locations where the frequency of searches for a particular term or set of terms were the highest receive a score of 100, and the frequencies of searching in other locations and dates are normalized relative to the maximum data point. In addition to this limitation, Google Trends analysis does not allow for trends to be disaggregated by county or city in the UK (which would allow the analysis to isolate London), so the following analysis summarizes trends for all of the UK.
Figure 1 displays trends in Google searchers for “Eric Garner” in the two months prior and two months following his killing on July 17, and compares searchers in the United States, given in the dashed line, and the UK, given in the solid line. As might be expected, searches for “Eric Garner” were highest in the immediate aftermath of his death, peaking three days following his shooting. Importantly, however, Figure 1 demonstrates that Britons also searched for news and information related to Eric Garner's killing: the overall trend in searches across time mimics the trend in the United States (importantly, spiking in the immediate aftermath of Garner's killing) and, in terms of volume, peaks at 28 per cent of the search volume in the United States. As we cannot be absolutely certain, however, that all persons in the sample were exposed to news of Garner's killing, the estimated effect should be interpreted as the intent to treat (ITT) effect (Muñoz, Falcó-Gimeno, and Hernández Reference Muñoz, Falcó-Gimeno and Hernández2020).
Fig. 1. Google searches for “Eric Garner” in the United States and the UK.
Effect of Eric Garner's Killing on Black Londoners' Attitudes Toward the Police
Figure 2 offers a descriptive depiction of changes in black Londoners' overall satisfaction with the Metropolitan Police before and after Eric Garner's killing. Figure 2 plots the average daily satisfaction provided by survey respondents as a function of their distance from Garner's killing, with one local linear regression fit to average daily police ratings among control respondents and another to police ratings among treated survey respondents in the two weeks before and after Garner's killing. Figure 2 clearly depicts a drop in outcome from before to after the event, though the apparent size of the effect appears somewhat moderate.
Fig. 2. Change in black Londoners' evaluations of the police following Garner's killing.
Table 2 presents estimates of the effect of Eric Garner's killing on the attitudes of black Londoners toward the Metropolitan Police. Consistent with Hypothesis 1, the results suggest that black Londoners' overall satisfaction of the Metropolitan Police did fall significantly following the killing of Eric Garner in New York. I estimate that overall satisfaction with the police fell by a value of 3.54 following the police killing of Eric Garner, overall perception of police fairness fell by 2.44, and perceptions of community engagement fell by 3.16. Evaluations of police effectiveness, however, were unchanged among the sample.
Table 2. Estimate of the ITT effect of Eric Garner's killing on Londoners' attitudes toward the Metropolitan Police
Notes: Data from the MPPAS. Robust SEs and robust CIs are estimated using the methods described in Cattaneo, Idrobo, and Titiunik (Reference Cattaneo, Idrobo and Titiunik2019) and implemented in Calonico et al. (Reference Calonico2017). Standard errors are clustered at the borough level. * Denotes statistically significant at α ≤ 0.05; ** denotes statistically significant at α ≤ 0.01.
Table 2 shows the results from white and South Asian Londoners as well. While the police killings of Eric Garner affected black Londoners, we would not expect the killings to affect white and South Asian Londoners to the same degree, if at all. The results presented in Tables 2 suggest that the effects of police killings on attitudes toward the police are largely—though not exclusively—limited to black Britons. Estimates of the ITT effect among South Asian Londoners across all outcomes were small and statistically insignificant, suggesting that Garner's killing had no effect on attitudes toward the police among this group. Results were similar among white Londoners, with one important exception: the killings do produce a comparatively smaller but statistically significant reduction in perception of police fairness.
Results with Covariates
Table 1 shows that the share of black Londoners in the treatment group between ages 45 and 54 rises discontinuously at the cut point. Not shown in the main text are similar tables for the white and South Asian subsamples, among whom the share of persons aged 35–44 and part-time workers drop at the cut point, respectively. Following Muñoz, Falcó-Gimeno, and Hernández (Reference Muñoz, Falcó-Gimeno and Hernández2020), I present results that condition on these variables in Table 3. Results are unchanged.
Table 3. Estimate of the ITT effect of Eric Garner's killing on Londoners' attitudes toward the Metropolitan Police with control values
Notes: Data from the MPPAS. Robust SEs and robust CIs are estimated using the methods described in Cattaneo, Idrobo, and Titiunik (Reference Cattaneo, Idrobo and Titiunik2019) and implemented in Calonico et al. (Reference Calonico2017). Standard errors are clustered at the borough level. Regressions control for those aged 45–54 among the black subsample, those aged 35–44 among the white subsample, and those working part-time among the South Asian subsample. * Denotes statistically significant at α ≤ 0.05; ** denotes statistically significant at α ≤ 0.01.
Conclusion and Discussion
This article presents evidence that police violence against black persons within the United States can negatively affect black persons' attitudes toward the police elsewhere. I find clear evidence that the police killing of Eric Garner reduced black Londoners' overall satisfaction with the Metropolitan Police, perceptions of police fairness, and sense of the quality of the Metropolitan Police's community engagement. Critically, I estimate that Garner's killing had no effect on South Asian Londoners' attitudes toward local police and a comparatively small effect on white Londoner's perceptions of police fairness. This instance of police violence in the US did not affect any group's perception of police effectiveness or ability to address crime.
These results help frame the emergence of BLM chapters and political demonstrations against police violence in other contexts, and suggest that they are not just expressive manifestations of cross-national solidarity. Police violence against black persons in the US likely resonates with the experiences of black persons in London (and elsewhere), and political actions taken in those contexts are aimed toward addressing systemic inequality there. The findings here are consistent with past research that has characterized the impact that US racial politics has had on civil rights struggles in Canada, Brazil, and the UK (Garbaye Reference Garbaye2005; Paschel Reference Paschel2016; Waters Reference Waters2013). Although US racial politics is not deterministic of racial politics elsewhere, black persons outside the United States do strategically adopt the strategies, tactics, and demands of black Americans to press for racial equality in their own contexts (Waters Reference Waters2013). US intellectuals, political activists, and NGOs also shape global discourse about minority rights (Paschel Reference Paschel2016). Finally, the United States is frequently treated as a foil against which racial inequality in other national contexts is assessed. Black activists in Canada, for example, have had success in advancing anti-discriminatory legislation by comparing Canada (unfavorably) to the US (Waters Reference Waters2013).
At the same time, the small drop in white Londoners' perceptions of police fairness suggests that while the transnational consequences of police violence in the United States are more pronounced among black Londoners, they are not exclusive to them. Furthermore, the null effect among South Asian Londoners' is consistent with past studies critical of assuming solidarity among minority ethnic Britons (Modood Reference Modood1994). Indeed, as the BLM movement has progressed since 2014, racial attitudes and support for the BLM movement among non-black Britons have shifted, with a majority of white Britons expressing support for BLM in the fall of 2020 (Ipsos MORI 2020). The results presented here, however, suggest that at the outset of movements such as BLM, black Americans will find the greatest levels of international support from black communities elsewhere.
Future Research
The policy implications of these findings are not entirely clear. Results clearly demonstrate that efforts by the Metropolitan Police to improve black Londoners' attitudes toward the police are affected by events beyond their control. Given the limited nature of the data, however, it is not possible to assess the extent to which these killings affected attitudes toward social or political policies more broadly. Given the documented influence that black political activists have had on black social movements and civil rights policies in Canada, the UK, and Brazil, tracing the impact that events in the United States have on policy preferences and political behaviors elsewhere constitutes an important site for future research.
This study is also unable to conclusively demonstrate that survey respondents were aware of the killings or consumed media concerning it. Black Londoners, for a variety of reasons, may have had greater exposure to the killings than white and South Asian Londoners. It may be that if all Londoners were equally exposed to news of the killings, we would have observed a similar treatment effect across groups. If, however, black Londoners purposefully sought out or were differentially more likely to be exposed to news on the killings because of their race, then the underling mechanism through which police killings in the US affects attitudes elsewhere may work through differential embeddedness in (transnational/diasporic) news or social media networks. This might be the case: survey data from the Office of Communications suggests that in 2014, minority racial and ethnic Britons: were less likely to read the news, listen to the radio, or read newspapers relative to white Britons; were no more likely to use social medial than white Britons; and generally turn to the same social medial websites for news. This is not definitive: the lumping together of all minority ethnic groups may mask differences and does not allow us to understand Londoners specifically. Generally, however, understanding the extent to which black persons outside the United States consume media on black American politics and culture, and how this affects their social and political attitudes, also constitute rich sites for future research.
The survey also does not allow the identification of potential differences in the effects that news coverage of the killings, circulation of images/video (on social or traditional media), and black American demonstrations against police violence had on attitudes. Since, however, police killings of black Americans precipitate all three, the analysis conducted here is representative of real-world events (see also Enos, Kaufman, and Sands Reference Enos, Kaufman and Sands2019). Nevertheless, differentiating between the effects that each can have on attitudes is important. Future work may consider addressing this as well.
Finally, while Garner's death was the first filmed and widely circulated police killing during the BLM movement, there have been many since. The analysis presented here is of a single, though important, case; the extent to which the findings generalize to other cases remains unknown. Future work may do well to consider how characteristics of the victim (such as gender) mediate the effect or changes in the effect over time. Future work may also do well to consider other diasporic or transnational communities or ethnic groups.
Supplementary Material
Online appendices are available at: https://doi.org/10.1017/S0007123421000557
Data Availability Statement
Replication data for this article can be found in Harvard Dataverse at: https://doi.org/10.7910/DVN/2BUYQV.
Acknowledgments
I would like to thank Dr Jesse Acevedo for his helpful comments on the draft manuscript and Dr Mark Sawyer for his support.
Financial Support
None.
Competing Interests
None.
