The Political Consequences of Policing: Evidence from New York City

Abstract

This paper explores the effect that municipal policing can exert on politics, and specifically investigates the effect that Stop, Question, and Frisk (SQF) policing has had on voter turnout and candidate choice in New York City. While extant studies of the American criminal justice system have found that mass incarceration and felon disenfranchisement negatively impact political participation and engagement, few have explicitly explored policing’s relationship with politics and fewer still consider its mobilizing potential. Mobilizing data from over 2.7 million geo-coded police stops and data from a series of national and municipal elections this paper uncovers a pattern of voter demobilization, voter mobilization, and candidate choice that cannot be anticipated from extant studies in the literature. Specifically, it finds that while concentrated policing was associated with reductions in voter turnout in the 2006 and 2010 midterm elections, it was associated with higher rates of turnout in the 2008 presidential election and 2013 Democratic primary and general mayor. Further analysis demonstrates that stopping intensity was strongly associated with candidate choice in the 2013 Democratic mayoral primary, such that higher rates of policing were positively associated with support for the candidate (John Liu) who advocated for eliminating SQF and less support for the candidate (William Thompson) who supported SQF. Together, these findings highlight the impact that policing can exert on political behavior, characterize the impact that harmful policing policies can play in instigating policing participation and engagement, and foreground the importance of considering local criminal justice policy and political action.

What effect, if any, do adversarial street-level interactions with police officers have on political behavior? Extant studies have explored the impact of the criminal justice system on American politics and have demonstrated that the US criminal justice system can significantly constrain an individual’s ability to participate, erode political efficacy and trust, and weaken the capacity of entire communities to engage in politics (Alexander 2012; Burch 2013; Manza and Uggen 2006; Lerman and Weaver 2014b; Uggen et al. 2016). Extant studies have emphasized that these negative political consequences are concentrated among black and Latino Americans, whose rates of incarceration and political disenfranchisement far outpace those of white Americans (Uggen et al. 2016).

Most extant studies, however, have focused on the political consequences of mass incarceration and felon disenfranchisement, to the relative exclusion of other criminal justice policies or forms of criminal justice contact. The omission of studies on police policy and police-citizen contact is perhaps the most striking, given policing’s status as the ‘frontline’ of the criminal justice system. The frequency of everyday police-citizen interactions, for example, far exceeds the frequency with which individuals are arraigned in court or the size of the incarcerated population (Eith and Durose 2011). Yet with the notable exception of Lerman and Weaver (2014b) and Weaver and Lerman (2010), few empirical studies have tested the impact that police contact or exposure may exert on political behavior.

The absence of studies focused on policing’s political consequences is surprising given evidence that police behavior—particularly violence and misconduct against black and Latino Americans—can instigate political action, as has been the case with political mobilization against police brutality in the 1960s and 1970s and the emergence of the Black Lives Matter movement (Escobar 1993; Felker-Kantor 2014 Goodman and Baker 2014; National Advisory Commission 1968). The contemporary literature largely omits these cases in its framing, however, and contemporary studies generally fail to provide a convincing explanation of how exposure to the criminal justice system can increase political participation and engagement. An important exception lies with recent studies which have characterized how the threat or harm constituted by the US criminal justice system and immigration enforcement policies can instigate mobilization (Walker 2014; White 2016). While criminal justice contact may demobilize individuals in many instances and contexts, findings from these studies suggest that criminal justice policies that are discriminatory, threatening, and do active harm can also instigate higher levels of political participation among those who are driven and able to combat them.

This study formally tests the political consequences of policing utilizing over 2.7 million geocoded police-initiated street stops conducted by the New York City Police Department under its Stop, Question, and Frisk (SQF) program, alongside precinct level voting data and data from the US Census Bureau. New York City’s SQF tactic is one among a larger set of policing strategies officially designed to reduce violent crime and improve the “quality of life” of New York City residents (Bratton and Knobler 1998; Vitale 2008). In practice, however, SQF policing resulted in millions of unnecessary police-citizen interactions (mostly of innocent young people of color), and likely had a negligible effect on crime (MacDonald et al. 2016; NYCLU 2017). Despite this, the program has been hailed as a success by its proponents, and versions of SQF policing have been implemented by police departments in many other American cities (Reisig and Kane 2014).

The results, drawn from binomial regression models and matching analysis, suggest that while policing can reduce political participation, it can also serve as a point of mobilization and engagement. Specifically, the results suggest that spatially concentrated policing reduced voter turnout in the 2006 and 2010 off year elections, most likely through the production of negative interpretive effects that can reduce an individual’s sense of political trust, efficacy, and standing, and thereby their political participation (Epp et al. 2014; Weaver and Lerman 2010). At the same time, however, the results demonstrate that policing is not uniformly demobilizing, and uncover instances where concentrated policing was associated with higher rates of voter turnout and engagement. More specifically, the results suggest that the spatial concentration of SQF policing was associated with higher voter turnout in the 2008 Presidential election, the 2013 mayoral election, and the 2013 Democratic primary for mayor—a pattern I argue is attributable to the mobilizing effect of Barack Obama’s presidential campaign and the political opportunity that the 2013 Democratic primary and general election presented voters to change policing policy in New York City. Indeed, further analysis suggests that policed communities were responsive to Democratic primary candidate positions on SQF, and that more intense SQF policing was associated with more support for the candidate (John Liu) who advocated for ending SQF policing and less support for the most prominent candidate (William Thompson) who supported the program.

Taken together, these findings suggest that everyday policing can significantly affect the behavior of American citizens and communities and foreground the political significance that millions of everyday police-citizen interactions have for American politics. While a set of this study’s findings are consistent with past studies that have characterized the negative impact of the US criminal justice system on political participation and engagement, additional findings offer evidence of mobilization and engagement to contest harmful and discriminatory policing policies. The results also importantly point to the potential for political mobilization directed towards formal electoral institutions, in addition to the protests and direct action identified in other work. Finally, the pattern of mobilization and demobilization identified in this analysis tentatively suggest that the likelihood of political mobilization against harmful criminal justice policies is contingent on political opportunities and context. This study’s identification of mobilization and voter engagement in New York City’s 2013 municipal elections suggests that political action against punitive and discriminatory criminal justice policies may be more likely at the local level, where most interactions with the criminal justice system are experienced, where political opportunities to affect policing policy may be more plentiful, and where the mobilization of negatively affected communities may be more likely to result in policy change.

Background: The Rise and Consequences of Punitive Policing

A broad literature has characterized the negative effect that mass incarceration and felon disenfranchisement exert on political participation and representation in the United States, particularly for black and Latino Americans. Over two million Americans are currently incarcerated in the United States and well over six million Americans—including 1 in every 13 black Americans—could not vote in the 2016 Presidential elections due to state level statutes prohibiting felons, ex-felons, parolees, and probationers from casting a ballot (Uggen et al. 2016). Extant studies have argued that these rates of incarceration and disenfranchisement may have significantly affected the outcome of the 2008 Presidential election, partisan control of state and federal legislatures in the 1990s and 2000s, and weakened the political representation of liberal, urban, and predominately minority districts relative to white, rural, and conservative districts (Gottschalk 2008; Lotke and Wagner 2003; Manza and Uggen 2006; Uggen et al. 2016. But see Burch 2012). Extant studies have also argued that individuals who have spent significant time in prison are far less likely to turnout to vote than individuals who have had no criminal justice contact (Weaver and Lerman 2010), and that those living in neighborhoods where large numbers of individuals have been incarcerated are significantly less likely to participate in politics, independent of their own experiences with the criminal justice system (Burch 2013).

Yet while the political significance of mass incarceration and felon disenfranchisement is now well established, the manner in which policing may uniquely and systematically affect political behavior is far less appreciated. Indeed, much of the extant literature characterizes policing as the entry point to other institutions (e.g. prisons) and processes (e.g. disenfranchisement) that do the “real work” of political marginalization or exclusion. This, I argue, is a significant oversight. Policing constitutes the single largest element of the criminal justice system and employs more personnel, commands a larger share of state and federal expenditures, and produces more contact with the criminal justice system than courts, prisons, or jails (Hughes 2006). Indeed, the absolute number of police-citizen interactions in a given year dwarfs the number of incarcerated individuals by a factor of ten: approximately forty-three million Americans experience some sort of police contact in a given year, including approximately 12% of all US drivers and millions of urban residents and pedestrians (Eith and Durose 2011; Lundman and Kaufman 2003; NYCLU 2017).Footnote 1

Existing sociological and criminological research has found that these instances of police contact can have significant consequences for attitudes and behaviors, with racial and ethnic minorities more likely to be negatively affected. For example, while individuals across all demographic groups rate contact with the police negatively, black and Latino Americans are more likely to report worse contact (Weitzer and Brunson 2009). Indeed, an expansive literature has characterized how black and Latino Americans (especially young men) are targeted to be stopped, detained, and searched by police officers—who may be more invested in maintaining “social order”, hierarchy, and inequality than in enforcing the law or reducing crime (Fassin 2013; Laniyonu 2017)—and the physical, verbal, emotional, and sexual abuse they can suffer in these interactions (Baumgartner et al. 2017; Epp et al. 2014; Fine et al. 2003; Gau and Brunson 2010; Geller et al. 2014; Peffley and Hurwitz 2010; Rios 2011; Stoudt et al. 2011; USDOJ 2015). Fine et al. (2003) and Stoudt et al. (2011), for example, characterized patterns of police harassment of black and Latino youth in New York City, and described how these stops can generate fear, alienation, distrust, resentment, and anger towards the police. Other studies of urban policing have suggested that frequent police stops can produce anxiety and symptoms consistent with post-traumatic stress disorder (Geller et al. 2014) and that policing can negatively affect test scores and opportunities for educational attainment (Rios 2011).

Extant studies have been very attentive to the consequences of that police stops can have for attitudes and perceptions of the legal system. Peffley and Hurwitz (2010), for example, argued that a large part of the gap between black and white attitudes towards the police and perceptions of structural bias in the criminal justice system can be explained by differences in individual and vicarious interactions with police officers (see also Carr et al. 2007). Research on police legitimacy—belief that police officers have the authority to dictate individual behavior and that citizens have an obligation to follow police orders—has similarly demonstrated that negative interactions with police officers can reduce an individual’s willingness to call, assist, or comply with law enforcement (Jackson et al. 2012; Kirk et al. 2012; Mazerolle et al. 2012; Sunshine and Tyler 2003). Stops produced by many contemporary policing practices—including SQF policing—may be uniquely corrosive to police legitimacy, however, as police legitimacy is strongly determined by the extent to which police treat members of the public fairly, with courtesy, and without bias (Geller et al. 2014; Tyler et al. 2014). As a broad literature on police-citizen interactions has described, however, stops generated by policies like SQF give officers broad discretion (which may appear like arbitrary and biased behavior), disproportionately target young racial and ethnic minorities, overwhelmingly fail in their legal state mission to recover drugs or weapons, and increase the opportunities for officers to engage in abuse (Fine et al. 2003; Geller et al. 2014; Stoudt et al. 2011; NYCLU 2017).

Police Contact and Political Demobilization

Based on this extant research, we may reasonably expect police contact or exposure to negatively affect political participation or engagement, arguably through four different mechanisms. First, policing practices may lead to the accumulation of legal financial obligations (LFOs), which in some states can lead to disenfranchisement. Meredith and Morse (2017), for example, have demonstrated that interactions with the criminal justice system can lead to the accumulation of debts stemming from court costs, fines, and victim restitution, which in some states must be paid back as a condition of voting or having ones right to vote restored as a consequence of incarceration (see also Ruback and Clark 2011). And while these authors specifically consider LFOs incurred following incarceration, predatory and revenue driven policing practices (like those practices in Ferguson, Missouri) can also lead to the systematic accumulation of LFOs and potentially to voting prohibitions (Carbado 2016; USDOJ 2015).

Second, policing practices can undermine the accumulation of skills and socioeconomic resources that facilitate political participation (Brady et al. 1995). As noted above, punitive interactions with officers can worsen education outcomes and lead to more serious criminal justice contact, especially for black and Latino youth (Owens 2017; Rios 2011). Policing can also negatively affect the economic resources that facilitate political action and voice, and can even affect an individual’s opportunity for employment, thereby affecting both economic resources and the development of politically useful skills (Brady et al. 1995). The US Department of Justice’s (2015) report on the predatory and revenue driven policing practices of the Ferguson Police Department, for example, characterized the negative effect that aggressive and racially discriminatory policing had on some resident’s ability to maintain employment, while other reports characterized the severe financial strain that tickets, fines, and LFOs put policed individuals and communities under (Carbado 2016).Footnote 2

Third, police stops may drive individuals to avoid state institutions, though a process called system avoidance (Brayne 2014). In an ethnography of a marginalized black neighborhood, for example, Goffman (2015) described how frequent direct and indirect interactions with police officers and other criminal justice institutions led some community residents to systematically avoided state and non-state institutions—including the police, courts, and even hospitals—which were perceived by residents as places where the surveillance and social control functions of the state would be exercised over them. In large test of this potential outcome, Brayne (2014) explored whether frequent and negative interactions with police and the criminal justice system led to a reticence among individuals to interact with institutions that keep or require formal records, and found evidence of it in a large, nationally representative survey. While neither author tested whether voting behavior was affected, we might predict that if aggressive proactive policing policies produce legal entanglements that result in system avoidance, then those targeted by these forms of policing may be less inclined to register or turnout to vote, since both processes require or produce records that can be tied back to the individual.

Finally, policing policies and practices may affect political behavior through their “interpretive effects” or the manner in which their implementation, enforcement, or symbolism can affect perceptions and socialization in politically meaningful ways (Mettler and Soss 2004; Lerman and Weaver 2014a). In a case study of welfare recipients, for example, Soss (1999) demonstrated how citizens who had frequent interactions with welfare agents that were invasive, humiliating, and disempowering formed negative attitudes about the government and their own socio-political standing based on their interactions with those agents. As additional scholarship has argued, these negative attitudes and perceptions can lead to lower levels of external political efficacy, trust in government, and thereby participation and engagement (Lerman and Weaver 2014a; Mettler and Soss 2004; Soss 1999). The ability of policies to frame impacted groups can also affect perceptions of deservingness and socio-political status. The framing of some needs-based welfare programs (such as Aid for Families with Dependent Children) may have sent signals to their recipients and to the larger society that program recipients were poor and undeserving members of the civic community, while the framing of other programs (such as aid to veterans) may have done the opposite (Mettler 2007; Mettler and Soss 2004).

A similar process of political socialization and signaling may be at work in policing, especially in the context of the frequent involuntary police stops produced by policies like SQF. Engagement with the police can constitute the most frequent interaction some people, particularly young adults, have with legal and government authority, so stopped individuals may form attitudes about the state based on interactions with the police (Tyler and Huo 2002). As noted, interactions with police are overwhelmingly poor or negative, and like the welfare agents considered by Soss (1999), police officers have a tremendous amount of discretion which they may abuse or manipulate in interactions that can produce anger, frustration, and distrust, resulting in lower levels of efficacy and trust. Police policies can also signal deservingness and social standing. Epp et al. (2014), for example, demonstrated that black drivers are aware that they are targeted by discriminatory police practices and argued that these practices signal to black drivers that they occupy a marginalized or second-class social position. Indeed, the authors argued that it is this signaling effect that does some of the most harm to black drivers’ perceptions of police fairness and legitimacy, independent of officer misconduct or abusive treatment,

And while extant studies on the interpretive effects mechanism principally describes a process that might connect individual contact with officers to political outcomes, the effects produced by interactions with police officers may extend past the individual stopped by the police and outwards towards the community where the interaction occurred, or to social networks of which the stopped individual is a member (Tyler et al. 2014; Walker 2014). Fagan et al. (2016, p. 210), for example, noted that police stops, “suggest [a] public discounting of worth,” and may be considered, “a form of public shaming,” which may affect individuals who identify with those being stopped or when a police policy targets a particular racial, ethnic group, or community (Epp et al. 2014). Rosenbaum et al. (2005) similarly found that the vicarious experiences individuals have with police-officers (e.g. experiences that they may have through their friends or family) strongly affect changes in attitude about the police. The ‘lessons learned’ about government and socio-political standing, therefore, may extend past the individual and to the larger community that they are a part of.

Police Contact and Political Mobilization

At the same time, however, demobilization may not be policing’s only political consequence. Indeed, police violence and misconduct has instigated well-known instances of collective political action, civil disobedience, political mobilization, rioting, and insurrection during the Civil Rights movement, the Black Power movement, Chicano student protests, and the Black Lives Matter movement (Escobar 1993; Bloom and Martin 2013; Goodman and Baker 2014; National Advisory Commission 1968; Predergast 2011; Smiley 2015; Williams 1993).

Research on policy threat offers a useful framework for understanding how criminal justice policies might induce participation. Policy threat characterizes the ability of threatening, harmful, or discriminatory policies and policy environments to catalyze political participation among the individuals and communities that are affected by them (Campbell 2003; Cho et al. 2006; Barreto et al. 2009, 2005; Walker 2016). For racial and ethnic minorities in particular, harmful policies or threatening policy environments may activate ethnic solidarity or racial group consciousness, which may further facilitate higher levels of political participation (Barreto et al. 2009; Chong and Rogers 2005; Dawson 1994). Barreto et al. (2009) and Pantoja and Segura (2003), for example, characterized how the threatening policy environment posed by punitive immigration policies spurred mobilization in Latino communities and resulted in higher levels of political engagement, participation, and perceptions of ethnic solidarity. Campbell (2003) similarly found higher levels letter writing to Congress among senior citizens following the passage of harmful and threatening Social Security reform measures. In both instances, communities that were or could have suffered from policy implementation mobilized politically to contest it.

Findings from a growing literature also suggest that criminal justice policies can exert a similarly mobilizing effect. As the extant literature makes clear, criminal justice policy can produce real and serious harm, and the policy threat literature would anticipate forms of political mobilization designed to contest it. Walker (2014) found precisely that and argued that when contact with the criminal justice system leads to a perception that it is systematically unjust and unfair, proximally affected individuals (e.g. family and friends) may be mobilized to change it. White (2016) similarly argued that the threat constituted by provisions for potentially discriminatory policing and immigration enforcement contained within the Secure Communities Act increased Latino American voter turnout in communities that implemented them. Simply put, rather than suffering the material, legal, or psychological effects of harmful policies, the policy threat literature demonstrates that affected groups occasionally mobilized to counter-act them.

Critically, the (counter)mobilizing potential of threatening or punitive policies may be contingent. Extant research suggests that such mobilization may first depend upon the presence of community-based organizations, local activists, or political elites who can facilitate participation by providing resources, fostering efficacy, coordinating action, or signaling to members of the community that institutional politics and policies are susceptible to change. Barreto et al. (2009), for example, attributed part of the success of the 2006 immigrant protests to the direct appeals made by elites and the organizational infrastructure they provided. Walker (2016) and White (2016) similarly point to the catalyzing role that community-based organizations and activists can play by providing material resources and information, coordinating action, and offering opportunities to participate.

Mobilization in the face of threatening policy may also depend on political opportunities and sites for political action: threat may only catalyze action when the perceived or objective possibility for affecting outcomes through participation exists. In their study of policy threat’s mobilizing effect on seniors, for example, Campbell (2003) found that the threat to seniors constituted by Social Security reform increased participation in an activity that could actually influence the policy (letter writing to Congress) but had no discernible effect on other forms of political participation. In a classic study of how violent police behavior can instigate protest participation, Eisinger (1973) similarly found that relatively “open” political opportunity structures—embodied in part by the presence of black elected officials—can signal to local community members that political participation (in that case protests) may lead to substantive policy change. Finally, Kriesi (1995) found that the success of once marginalized and disempowered left-libertarian movements in Europe could be explained in part by shifts in the constitution of political elites, competition among them, and attempts by some to court new political constituencies (see also Meyer 2004). The threatening policy environment posed by policing may therefore increase participation, but this mobilizing potential may be contingent on the presence of political elites and favorable political contexts.

The Case of New York City’s 2013 Democratic Primary and Mayoral Election

New York City’s 2013 Democratic primary and general election for mayor may have presented communities harmed by SQF policing with precisely this sort of electoral context or opportunity. Media reporting and the mobilization of unions and community-based organizations helped make police behavior and the SQF tactic a key campaign issue in both the Democratic primary and the general election, while the Democratic primary itself pitted several Democratic candidates with distinct position on SQF against one another (Leland and Moynihan 2012).Footnote 3 Among major candidates, Bill de Blasio and John Liu were the two candidates most opposed to existing SQF policing practices. Bill de Blasio was clear in his call for reforming SQF policing, replacing then Commissioner Ray Kelly, appointing a general inspector to oversee the NYPD, and allowing city residents to sue the city of New York if they believed they were targets of racial profiling. John Liu’s position went further, and he was the only major candidate to advocate for completely abolishing SQF policing (Gould 2013).

The opposition of these candidates to SQF stood in stark opposition to the position adopted by William Thompson, a leading candidate the race and an African American politician who won the Democratic party’s mayoral nomination in 2009 but lost to Michael Bloomberg in the general election. Thompson notably did not repudiate SQF and supported the program in televised debates (Barbaro 2013). His support of the program earned him the endorsement of former mayors Michael Bloomberg and Rudy Giuliani, as well as endorsements from unions representing police officers in New York City. It also earned him strong criticism from critics of the program and African American political leaders (Barbaro 2013).

I suggest that the electoral context presented by the 2013 Democratic primary and general election—specifically the opportunity to substantially change SQF policing—mobilized communities negatively affected by SQF policing, and further that these communities were engaged in the candidate selection process during the Democratic primary. Several candidates may have mobilized communities negatively affected by SQF by directly appealing to them; the presence of vocal critics of SQF among major candidates may have signaled an opening of the political or policy environment to policy change; and it was likely clear to community residents that political participation in the primary and general election could actually affect policing policy (Freelander 2013; Leland and Moynihan 2012). So, while intense community level policing likely reduces political participation in many (if not most) electoral contexts, conditions were such in 2013 that mobilization in response to harm could have produced favorable outcomes.

Review of Extent Empirical Literature

Empirically, the set of studies that test for mobilization or demobilization following any form of criminal justice contact is small. In one study, Weaver and Lerman (2010) mobilized data from two nationally representative surveys and explored whether a wide range of personal interactions with the criminal justice system (being questioned by the police, arrested, convicted of a crime, serving prison or jail time under a year, and serving time for longer than a year) were correlated with reduced levels trust in government, voter registration, voting, and civic participation. They found that being questioned by the police, arrested, convicted of a crime, and serving jail time of any length all reduced trust in government; that any jail time served was correlated with a reduced likelihood of being registered to vote and forms of civic participation; and that being arrested, convicted, and serving any jail time was correlated with a reduced likelihood of voting. In a different survey, however, Walker (2014) found a surprising positive association between criminal justice contact (direct or indirect) and political participation, particularly non-electoral forms of political participation.

Burch (2013) also considered the impact of criminal justice contact on political participation but focused on the effect that the intense spatial concentration of incarceration can have on community level political participation. Specifically, Burch (2013) tested whether the spatial concentration of incarcerated and politically disenfranchised individuals could affect community level political participation by weakening norms of political participation, undermining social capital and social networks, reducing the resources available for collective mobilization, and disincentivizing outside groups (like political parties and NGOs) from mobilizing community members. The analysis presented there did find convincing evidence for the latter three mechanisms.

Finally, Lerman and Weaver (2014b) investigated the effect that NYPD police stops have on local political engagement and tested this relationship at the neighborhood level, in an analysis that closely mirrors the one presented here. The authors investigated the effect that the intensity of SQF policing had on political participation, which they measured using community 311 call rates. Distinguishing between 311 calls that concern NYPD and those that did not, the authors tested whether frequent and negative contact with officers reduced citizens’ likelihood of engaging municipal bureaucracy related to the police or NYPD. Overall, they found that as the community-level intensity of policing increased, the frequency of 311 calls related to crime, public safety, or the NYPD decreased.

While important contributions to the literature, for scholars interested in the specific effect of policing on political participation, these studies have important limitations. Weaver and Lerman (2010) did not identify an effect of interactions with police officers on behavior in either survey, despite the broad set of dependent variables they tested. It may be that policing has no effect on political behavior, but as Lundman and Kaufman (2003) have suggested, self-reported police contact data may lack validity, and validity issues are well known to undermine data on self-reported political participation. Second, the authors explored the effect of police stops at the individual level. As Burch (2013) has demonstrated, however, the relationship between stopping and political engagement may have an important spatial component, and there may be political consequences that cannot be identified without accounting for police behavior in the community that an individual is a member of. And while Burch (2013) did explore aggregate level effects, that study focused on incarceration, and cannot provide substantial insight into the consequences of policing.

Finally, while Lerman and Weaver (2014b) do focus on the consequences of policing and did so at the aggregate level, the study contains important short comings as well. First, it is not clear what the underlying distribution or demand for 311 services is. Some sub-section of the population may constitute the core of 311 callers, whose sensitivity or exposure police stops may be confounded in their analysis. Second, it is not clear that 311 calls offer a strong proxy for political engagement. White and Trump (2016), for example, did not find a strong or consistent relationship between aggregate level 311 call rates and other aggregate level measures of political engagement such as Census response rates, voter turnout, or political donations. Minkoff (2016) has discussed the limitations of 311 data more generally and has argued that in order to be adequately mobilized as a proxy for civic engagement, researchers must simultaneously account for absolute demand for services.

The analysis presented below sidesteps these issues by utilizing administrative voter and policing data in New York City, which is not subject to the same reliability issues as self-reported data. It focuses specifically on policing to capture the consequences of this understudied dimension of the criminal justice system, and its use of voting data allows it to directly characterize the effect of policing on politics. By focusing on SQF policing in New York City, the analysis is also able to isolate some (but not all) of the potential mechanisms tying policing to political participation. New York, for example, does not require payment of LFO’s as a precondition for voting, so the systematic effect of policing on LFOs and thereby political participation is likely muted in this analysis. Similarly, SQF policing rarely results in more serious legal entanglements (such as a citation or summons to court) that might yield strategic avoidance of formal state record keeping agencies and practices that could result in arrest or surveillance. Finally, this analysis is able to control for the effect policing may exert of political participation through socioeconomic resources or status. The results presented below, likely therefore characterize the interpretive effects of policing on participation, although as I discuss in the conclusion, the specific mechanisms underlying the outcomes cannot be concretely identified in this study.

Hypotheses

On balance, the extant literature suggests that concentrated policing likely reduces political participation, although additional research suggests that mobilization to counter-act policing is possible. In the context of SQF policing in New York City, the 2013 Democratic primary and general election presented voters with an opportunity to support candidates who committed to reforming SQF policing and who actually had the power to do so. Conversely, the national level elections of 2006, 2008, and 2010 neither featured candidates who committed to changing SQF policy nor candidates who actually had the ability to do so. Phrased differently, while the spatial intensity of policing in a neighborhood may reduce the likelihood of turning out in national elections—where voters do not have the opportunity to affect policing policy and candidates generally do not run on police reform—the threatening and punitive environment constituted by policing policy may increase turnout at the local level where policing policy is set and in an election where police reform was a central campaign issue. I therefore hypothesize that:

H1a

Areas of New York City that experience higher levels of SQF policing will exhibit lower levels of voter turnout in the 2006, 2008, and 2010 national elections

H2a

Areas of New York City that experience higher levels of SQF policing will exhibit higher levels of voter turnout in the 2013 mayoral election and Democratic primary for mayor

New York’s 2013 municipal elections not only presented voters with an opportunity to vote for a candidate who might change SQF policy, but an election (the Democratic primary) where multiple competing candidates adopted distinct positions on SQF. Specifically, John Liu adopted the most radical position on SQF and advocated abolishing the program, De Blasio adopted a relatively progressive position and advocated significant reforms, and William Thompson adopted a moderate or conservative position, arguing that the program should remain relatively unchanged. If policy threat motivates affected communities to combat harmful and threatening policies, then we may not only expect higher levels of political participation among policed communities in an electoral context that might affect the policy, but higher levels of support for candidates who advocate for more reform. I therefore hypothesize that:

H2a

The spatial intensity of SQF policing will be positively associated with support for John Liu and Bill de Blasio

H2b

The spatial intensity of SQF policing will be negatively associated with support for William Thompson

The theory of interpretive effects further suggests that the quality of interactions may impact evaluations of self and government, and thereby likelihood to vote. Communities where officers frequently treat residents they stop in an aggressive or invasive manner may have worse experiences and may evince lower levels of participation than similarly situated areas. More explicitly I hypothesize that:

H3a

Areas where frisks and searches during a stop are more frequent will exhibit lower levels of voter turnout

H3b

Areas where the application of force during a stop is more frequent will exhibit lower levels of voter turnout

H3c

Areas where stops are based on the furtive movement of suspects will exhibit lower levels of voter turnoutFootnote 4

Finally, extant theory on police-citizen interactions suggests that black and Latino Americans experience police stops worse than white Americans and as a result, the magnitude of the hypothesized relationships may be stronger as the black and Latino share of a community increases. Hypothesis 4 states this expectation explicitly:

H4a

The demobilizing or mobilizing effect of police stops will increase as the minority share of the population increases

H4b

The affect of police stops on candidate support (for John Liu and Bill de Blasio and against William Thompson) will increase as minority share of the population increases

Data and Methods

Data for the analysis came from several sources.Footnote 5 Data on the number of police-initiated stops was pulled from the NYPD’s Stop, Question, and Frisk dataset. The NYPD records police-citizen interactions from the program on the UF-250 form, whose collection is governed by law and NYPD policy which mandate that all official stops (NYPD 2016).Footnote 6 Aggregate level voter turnout data for the 2006, 2008, and 2010 national elections came from the Harvard Election Data Archive (Ansolabehere et al. 2014). The Data Archive also contains GIS shape files, which present digital representations of the geographic boundaries of New York City’s voter tabulation districts (VTDs). VTDs are the smallest geographic unit that voter turnout and candidate choice data are available and serve as the primary unit of analysis for the study.

The generation of control values, drawn from Census data, was somewhat complicated since VTDs and Census tracts do no neatly overlap in many areas of New York City. To address this issue, the analysis took advantage of data collected during New York City’s 2010 redistricting process, from which the calculation of the fraction of each Census tract that falls within each VTD is relatively straight forward. Where a VTD was completely constituted by a single election district, control variables from the corresponding Census tract were assigned to that VTD. Where a VTD was constituted by multiple Census tracts, however, control values were calculated via an averaging procedure that weighted covariates based on the share of the population of the VTD that came from each constitutive Census tract. This procedure was used on data from the 2010 ACS to generate control values for the models run on 2010 turnout data. To calculate control variables for the 2006 and 2008 elections, the analysis assigned tract level control values from the 2000 Decennial Census to the 2010 voter tabulation map using Longitudinal Tract Database tool (Logan et al. 2014). Tract level control values for 2006 and 2008 were then generated by linear interpolation.

Data from the 2013 municipal elections were gathered from a different source and control variables were tabulated via a different procedure. Data on candidate choice was pulled from a publicly available data set hosted by the CUNY Graduate School of Journalism’s Election Atlas Project (New York City Primary Results 2013; Center for Urban Research 2013). Candidate choice data for these years was also collected at the VTD level but unlike data from the national level elections, they fit the 2013 map of voter tabulation districts rather than the 2010 map. To generate estimates of VTD level control values, the analysis first overlaid the 2013 map of New York City Census blocks over the 2013 VTD map, and assigned every Census block in the city to a VTD. It then collected control data from the 2013 ACS at the level of the block group, which like voter tabulation districts, are completely and exhaustively constituted by individual Census blocks. It then calculated block level control values using population size as a weight and with block level controls tabulated, aggregated these control values to the level of the voter tabulation district.

The analysis models aggregate level voter turnout in the national level elections as well as candidate choice in the 2013 mayoral Democratic primary election in a series of binomial regression models, which specifically model the total number of “successes” (citizens who turnout in the first set of analyses and number of supporters of a candidate in the second) out of a series of “trials” (an estimate of the voting eligible populationFootnote 7 and the number Democratic primary voters), as a function of an intensity parameter itself determined (through the logit link function) by covariates. In both cases (turnout and candidate support), the aggregate level data is the summation of discrete Bernoulli trials (e.g. an individual’s decision to turnout to vote or not, or to vote for a candidate or not) which can be represented in a binomial model.Footnote 8 Following standard practices (Gelman and Hill 2006), the analysis fits so called “quasibinomial” models, which depart from standard binomial models by allowing for variance terms that do not equal the mean.

Variables

Voter turnout is the focal dependent variable for the 2006, 2008, and 2010 national elections, for the 2013 Democratic primary election, and for the 2013 general election for mayor, and was operationalized as the number of voters out of the voter eligible population. Three models were run for candidate choice in the 2013 Democratic primary: (1) the number of Democratic primary voters who supported William Thompson Jr., (2) the number of Democratic primary voters who supported Bill de Blasio, and (3) the number of Democratic primary voters who supported John Liu, each out of the number of Democratic primary voters.

The focal independent variables for the models run on voter turnout were the per capita stopping rate; frisking and searching rate; use of force rate; and rate of stops based on furtive movements. For parsimony, the analysis of candidate choice in the 2013 Democratic primary election only explores the effect of the total stopping rate on candidate choice. The measure of the use of force relied on coding provided by the NYPD’s in their Stop, Question, and Frisk codebook, which characterizes the use of force as force applied via an officer’s hands, a wall, the ground, the drawing of an officer’s service weapon, the pointing of an officer’s service weapon, the use pepper spray, the use of a police baton, or handcuffing. Instances of searches, frisks, and stops based on furtive movements are categorically marked in the data set. Each type of stop was geolocated to the VTD in which it occurred and the aggregate number of each type of stop was calculated. This total count was then divided by the population of the VTD to get the stopping rate. To account for skew, this rate was logged.Footnote 9

Finally, a series of control variables were included to control for aggregate level factors that have been included in prior studies (see Geys 2006 for a review). The regressions for voter turnout in 2010, 2008, and 2006 include the share of the election district that is black American, Latino American, has been living in the same residence for the past 12 months, and is a homeowner. These regressions also control for population size, median household income, the share of single-mother households, and the share of the electoral district that turned out to vote in the last election. To control for skew, the population size and the median household income were logged. Due to data limitations, the 2006 regressions do not include a measure of past voter turnout.

The regressions from the 2013 elections include a similar set of controls: the share of the population that is black and Latino, the share that has lived in the same house for the past 12 months, rates of home ownership, the log of the population size and median household income, the share of households headed by single mothers and voter turnout in 2009. To control for additional unmeasured factors that explain candidate support, the candidate support models also include controls for past levels of electoral support that that candidate received. Thus, controls were introduced for the share of the election district that supported William Thompson in his 2009 mayoral run against Michael Bloomberg, the share of the electoral district that supported de Blasio in his 2009 campaign for public advocate of New York City, and the share of the district that supported Liu in his 2009 campaign for comptroller. Finally, given high rates of Democratic partisanship in New York City and the independent effect that party affiliation or membership might exert on the local municipal elections, all analyses run on data from 2013 include rates of Democratic party membership.

Results

Tables 1, 2, and 3 report the results of binomial regression models run on the relationship between stopping intensity and turnout in the 2010, 2008, and 2006 national elections, while Table 4 reports the results between stopping intensity and voter turnout in 2013. Each table reports the odds ratios, which give the effect of a unit switch in each covariate on the odds of turning out. They also report first differences for each covariate, which give the effect of switching the value of a covariate from the minimum value it obtains in the data to the maximum value it obtains, while holding all over covariate values in the VTD at their actual values.

Table 1 Effect of police stops on voter turnout: 2010
Table 2 Effect of police stops on voter turnout: 2008
Table 3 Effect of police stops on voter turnout: 2006
Table 4 Effect of Police Stops on Voter Turnout: 2013 Municipal Elections

Generally, such a switch is associated with a reduction in the share of the population that turns out to vote in the national level elections and an increase in the 2013 municipal elections, consistent with the hypotheses (H1a and H1b). Table 1 shows the results for the models run on turnout in the 2010 election. The first differences presented here suggest that a minimum to maximal shift in the overall stopping rate, frisk and searching rate, use of force rate, and furtive stopping rate is associated with a decrease of 9, 6, 4, and 7% in the share of the VTD that turns out to vote, respectively. Other coefficients largely behave as expected: increases in the black share of the district, the share of long-term residents, or the share of a district who are homeowners is associated with higher turnout, while an increase in the share of single mother households is associated with a reduction in turnout. Increases in the Latino and non-citizen share of the population is (surprisingly) associated with increases in turnout. Finally, and as might be expected, the turnout rate in 2008 is the single strongest predictor and the first difference of this coefficient is substantially larger than those generated from other coefficient estimates.

Table 2 presents an important exception to this pattern and suggests that higher rates of SQF policing are associated with higher rates of turnout in 2008 presidential election. Movement across the range of possible values for per capita stops, frisks and searches, use of force, and furtive movements is associated with a 9, 6, 5, and 7% increase in turnout, respectively. The relationship between policing and turnout shifts back to a negative relationship in the 2006 midterm elections however. Table 3 shows that the first differences for the overall stop rate, frisk and search rate, and furtive movement rate are associated with 7, 4, and 7% reductions in aggregate level turnout. No statistically significant result is identified in the relationship between the use of force per capita. In sum, higher rates of policing were associated with reduced turnout in two out of the three national elections, generally consistent with expectations.

Table 4 presents the results from analysis of the 2013 Democratic primary and general election and suggests that as policing intensity increased, so too did voter turnout. First difference values estimate that a shift from the minimum to maximum values of SQF are associated with a 2% increase in voter turnout in the Democratic primary and a 6% increase in voter turnout in the general election for mayor, supporting the notion that policing can constitute a force for political mobilization, conditional on elites and context. As might be expected in an off-year election, black and Latino share of the population is negatively associated with turnout in 2013, as is the population size and share of single mother households. Predictably, areas with long term residents evince higher rates of turnout in both the Democratic primary and the general election, and the share of registered Democrats in the VTD and the VTD’s past history of electoral participation were strongly associated with turnout rates. Income and home ownership are both, surprisingly, negatively associated with turnout.

I also hypothesized that more SQF policing would be associated with more support for candidates that endorsed reforming SQF (John Liu and Bill de Blasio) and less support for candidates who did not (William Thompson) (H2a, H2b). Results from the binomial regression models run on candidate support, presented in Table 5, are consistent with these hypotheses. Again, odds ratios and first differences giving the effect of a shift from the minimal to the maximal value of a coefficient are displayed. The results suggest that such a shift is associated with a 15% increase in support for John Liu and 12% reduction in support for Thompson. The estimate for SQF policing’s effect on support for Bill de Blasio is positive but not statistically distinguishable from zero.

Table 5 Effect of police stops on voter turnout: 2013 municipal elections

Tables 1, 2, and 3 also present a test of the hypothesis that the intensity of interactions that featured frisks and searches, the use of force, or stops based on the “furtive movements” of the suspect would be associated with stronger demobilization effects (H3a, H3b, H3c). The models uncover no evidence supporting these hypotheses.

The final set of hypotheses consider variation in these results across different types of communities. Given extant research that black and Latino communities experience more and worse SQF than white communities, H4a and H4b hypothesized that the magnitude of the effects hypothesized in H1a, H1b, H2a, and H2b—of demobilization in the national level elections, mobilization in the local elections, and candidate support for anti-SQF candidates in the 2013 Democratic primary—would increase as the black and Latino share of the VTD increases.

To test this contention, I ran an additional series of multiplicative interaction models, which adapt the baseline models displayed in Tables 1 through 5 but include as additional predictors multiplicative interaction terms which condition the effect of stopping intensity on turnout by the share of the voting district that is black or Latino. While interpretation of the substantive significance of the multiplicative interaction term is best communicated visually through marginal effects plots, given the number of potential models to test, I opt to display summary tables of these effects in the main text, and present results from the full interaction models in the Online Appendix.

Tables 6 and 7 summarize the results. The main results in these tables summarize whether the interaction effect (e.g. the marginal effect of black or Latino share of the VTD on the relationship between policing and turnout/candidate support) are positive or negative. In parentheses, the tables display whether the baseline relationship (e.g. the relationship between policing and turnout/candidate choice when black or Latino share of the population is zero) is positive or negative. Full interpretation of the interaction effect can be gained from considering the two together. For example, the first column of Table 6 suggests that the effect of black and Latino share of the population on the correlation between policing and turnout is negative. The “baseline relationship” is also negative, so the model estimates that the demobilizing relationship between policing and turnout gets stronger or “more negative” as black and Latino share of the population increases.

Table 6 Marginal effect of race and ethnicity on turnout (summary)
Table 7 Marginal effect of race and ethnicity on turnout (summary)

These results from 2010 are the only ones strongly consistent with expectation. Column 2 shows the results from 2008. The models estimate the marginal effect of black share of the population here to be positive, such that not only was policing positively associated with higher turnout in 2008, but this relationship gets stronger as black share of the population increases (no statistically significant effect was estimated for the Latino share of the population). Column 3 shows the results estimated for 2006, which also contradict expectation. While the baseline models estimate that the association between policing and turnout is negative in the 2006 mid-term election where black or Latino share is zero, it estimates that this negative association weakens as the share of black and Latino Americans residents increase. The results from the 2013 local elections are largely insignificant, with the exception of the positive marginal effect of the Latino share of the population on mobilization in the Democratic primary.

Table 7 presents the marginal effect of black and Latino share of the population on support for candidates in the Democratic primary, which again contradict Hypothesis 4. In every instance, the marginal effect of black and Latino share of the population runs counter to expectation. The baseline models, for example, estimate a negative relationship between William Thompson and policing intensity where no blacks or Latinos are present, but suggest that this relationship weakens as minority share of the population increases. The opposite is true of the marginal effect of black and Latino share of the population on the relationship between policing and support for John Liu and Bill de Blasio. Baseline modes estimates a positive relationship between support for these candidates and the spatial concentration of SQF policing where no blacks are Latinos are present, but that the magnitude of this association weakens as minority share of the tract increases. In sum, it does not appear as if an increase in the black or Latino share of the population was associated with a stronger relationship between SQF policing intensity and support for candidates who advocated for SQF reform, but the opposite.

Matching Analysis

As Lerman and Weaver (2014a) have noted, one weakness of cross-sectional models such as these lie in the potential endogeneity between neighborhood characteristics and levels of SQF policing, and the potential omission of variables that predict SQF policing levels. The authors, however, exploited the placement of communities into different police precincts, whose “policing styles and approach, training, force strength, leadership, and community relationships” may vary exogenous of community characteristics, to explore a potential casual effect (Weaver and Lerman 2014b, p. 213). Comparisons between socio-demographically similar districts, placed into starkly different police precincts may uncover a causal effect.

Following Lerman and Weaver (2014a), then, the following section offers supplementary evidence of policing’s effect on political participation, relying on a matching strategy to leverage potentially exogenous variation in levels of SQF policing across election districts as a function of the police precinct within which they reside. Like Lerman and Weaver (2014a), I identify police precincts that rank “high” in stops, searches and frisks, and stops based on furtive movements if they rank in the top 50% of all precincts on these metrics. I then match districts falling within “high” precincts to districts falling in “low” precincts based on all available covariates and explore differences in voter-turnout rates.

Figure 1 displays the results, which offer some confirmation of the associational results presented in Tables 1 through 4. Consistent with these original findings, point estimates of the treatment effects of SQF policing suggest that concentrated policing had a positive effect on voter turnout in the 2008 general election and a negative effect in the 2010 national election. In the first case, the results are statistically distinguishable from zero and are consistent across all three measures of treatment, and in the second case they are significant for two out of three measures of treatment (the exception being the measure of treatment generated from furtive stopping rates). The matching procedure also estimates higher levels of turnout in the 2013 Democratic primary for Mayor in two out of the three measures of treatment, but the results are not statistically significant. Estimates of the treatment effect of policing on turnout in the 2013 general election are not consistent and flip direction depending on the measure of treatment. Like results from the Democratic primary, however, these results are indistinguishable from zero. Finally, estimated treatment effects from 2006 run counter to expectation, as the matching procedure estimates a positive treatment effect, but again results are not statistically significant.

Fig. 1
figure1

Effect of SQF intensity on voter turnout rates. #p < 0.10; *p < 0.05; **p < 0.01; ***p < 0.001

Discussion and Conclusion

Overall, the results presented here suggest that while policing can negatively affect political participation, it may not always do so: in electoral contexts which present communities with an opportunity to address the threat or harm constituted by policing, policed communities may mobilize to counter-act that harm. Specifically, the results uncover a negative relationship between stopping intensity and aggregate level turnout in the 2010 and 2006 midterm elections in New York City, but also uncover a positive association between policing intensity and voter turnout in the 2013 Democratic primary and municipal elections—which fielded candidates who advocated for significant reforms to SQF policing and who may have directly appealed to affected communities—and in the 2008 presidential election.

Additional analysis of the 2013 Democratic primary election offers additional evidence that policing can increase participation and engagement. Here the results demonstrate a positive association between the intensity of policing and the share of the Democratic primary vote received by the Democratic candidate most critical of SQF (John Liu). The results also demonstrate a negative association between stopping intensity and support for William Thompson who largely supported SQF. These findings offer an important counterweight to the conclusions made by Lerman and Weaver (2014b), who have argued that policing drives people away from state services and from civic or political engagement. They also supplement findings made by Walker (2014), who argued that the most likely form of political mobilization against the harms of the criminal justice system will be non-electoral. The findings presented here suggest that while both may be true, policing can also be a driver of institutionally oriented electoral participation.

The finding that policed communities may mobilize to participate in electoral institutions echoes findings in the historical literature which have characterized the political strategies that groups have adopted to combat police abuse (see, for example Felker-Kantor’s (2014) analysis of black political mobilization against police brutality in Los Angeles and Baer’s (2015) analysis of mobilization against police torture in Chicago). The evidence presented here—that such contestation also happens in the domain of electoral politics and that policing was a relevant issue in New York’s 2013 Democratic primary for mayor—is consistent with their work. It is also consistent with reporting on the mobilizing strategy of the Black Lives Matter movement and other organizations oriented towards reducing police violence. These reports suggest that community groups have been active in several primary campaigns with significant implications for the criminal justice system at the local level, such as Anita Alvarez’s campaign for public prosecutor in Chicago in 2016 (Alcindor 2015; Lulay 2016). Interestingly, the findings here suggest that Democratic primary voters were responsive prior to campaigns centered on the issue of police reform prior to the mobilization of the Black Lives Matter movement in 2014.

The analysis encountered a surprising, positive relationship between policing intensity and turnout in the 2008 presidential election, despite the fact that that year did not feature a candidate who campaigned on police reform or could have directly affected SQF policing in New York City. I speculate that this positive association may be explained by the mobilizing effect that Barack Obama had on African American and potentially other marginalized communities (Fairdosi and Rogowski 2015; Merolla et al. 2013). Specifically, I speculate that Barack Obama’s candidacy could have conceivably raised political efficacy, trust, and engagement in policed communities or signaled changing opportunity structure. Even though it was unlikely that his election would lead to direct or immediate changes in SQF policing, policed communities may have viewed his race and/or his progressive policy platform as indicators that he would be a strong representative of their interests. At the same time, and consistent with the literature on political opportunity structure, it may also be the case that policed communities reasoned or believed that election of a progressive African American president figure could lead to shifts in political opportunities, yielding substantive policy change down the line, at the municipal level, and related to policing (Kriesi 1995).Footnote 10

The presence of Barack Obama on the ticket in 2008 may also explain the positive marginal effect of black share of the VTD on the positive relationship between policing and turnout in 2008. If Obama’s candidacy mobilized policed communities or was perceived as an opportunity to indirectly affect the odds of changes in SQF policing, then this effect is likely to increase as black share of the population increases. As literature on black empowerment theory demonstrates, these heavily policed black and Latino communities may have been especially mobilized by his campaign (Bobo and Gilliam 1990; Fairdosi and Rogowski 2015; West 2017).

The positive marginal effect of black and Latino share of the VTD on the relationship between policing on turnout in 2006 was similarly unanticipated. It may be the case, however, that the positive marginal effect of Latino share of the population on the relationship between policing and turnout in 2006 was a by-product of the large-scale immigration demonstrations and protests that occurred that year, which (as noted) increased levels of participation and engagement of Latinos (Barreto et al. 2009). Thus, while policing was negatively associated with turnout where no Latino’s were present, the higher rates of efficacy and engagement produced by those mobilizations in Latino communities upset policing’s demobilizing effect. What explains the positive marginal effect of black share of the population on the relationship between policing and turnout in 2006, however, remains a somewhat of a puzzle and may constitute an interesting site for future research.

The results giving the marginal relationship between race and candidate selection in the 2013 Democratic primary may be the most interesting and unanticipated. The results suggest a weakening in the relationship between stopping intensity and support for candidates with critical policy positions towards SQF as black and Latino share increases (and conversely, more support for William Thompson, a moderate on the issue). This result may nevertheless reflect recognition among African Americans that black elected officials can secure less punitive criminal justice policies and less discriminatory criminal justice outcomes relative to white elected officials (Sharp 2014). Thus, despite Thompsons articulated position on SQF and the criticism received from some black elites, heavily policed black and Latino communities in this instance simply may not have been convinced that Thompson would have perpetuated policing policies that harmed them, perhaps because of his race (Fairdosi and Rogowski 2015). Conversely, the credibility of De Blasio and Liu’s message to reform SQF may have been weaker in predominately African American and Latino communities, eroding the observed relationship between policing intensity and support for these candidates.

Limitations

This analysis of a single city is limited, and it remains to be seen whether the findings presented here generalize to other contexts in other cities and time periods. Future work may want to consider whether and to what extent policed communities evince political demobilization or mobilization in other political contexts, the various forms mobilization against harmful and discriminatory policing can take, and how they interact with one another. More generally, it remains relatively unknown how many electoral contests have featured candidates who campaigned on policing specifically, how frequently candidates have target policed communities for mobilization, and how communities that suffer the harm of policing seek to contest policing policy at the local level. Future work may be interested in addressing these questions.

The study is also limited in the extent to which it can adjudicate between different mechanism which might plausibly relate police contact to political participation, and which may be relevant in a larger cross-section of US municipalities. By mobilizing data from New York City (which does not disenfranchise individual based on LFO accumulated from criminal justice contact) and by considering SQF policing (which generally does not lead to more serious contact with the criminal justice system) the results of this analysis suggest that demobilization likely works through the interpretive effects mechanism. The analysis does not give proof positive of this potential mechanism, however, nor does it identify the mechanism(s) that produced higher rates of participation in 2008 and 2013. And, beyond New York City, policing may exert its most damaging political effect via its effects on income or by saddling individuals with LFOs, as observed in Ferguson, MO. Future research, perhaps taking individuals as the unit of analysis, may want to explore these different mechanisms and how they affect political behavior in locations where they might be at work.

Finally, the ability of the analysis to make causal claims is limited. While the analysis does present results from a supplementary matching strategy, such a strategy makes relatively strong assumptions, such as the assumption that police officers basically stick to beats contained within their police precinct or that discernible precinct level police cultures exist and drive SQF rates. As Lerman and Weaver (2014a) note, future work may also do well to explore other causal strategies, such as naturally occurring discontinuities or shocks to policing.

Notes

  1. 1.

    These high rates of police-citizen contact are attributable, in part, to the relatively recent implementation of aggressive and proactive policing strategies, such as order maintenance, zero-tolerance, and quality of life policing. See Vitale (2008) and Willis (2014).

  2. 2.

    Maciag (2014) has demonstrated that Ferguson, MO was not unique in this practice and many surrounding municipalities in St. Louis County engaged in similar policing practices. Other studies have more generally demonstrated that police departments engage in more revenue generating practices when their cities are under financial distress or their budgets contract (Baicker and Jacobson 2007; Park 2017) suggesting that this practice and potential mechanism may not be limited to Ferguson, MO.

  3. 3.

    No leading Republican candidates opposed the policy (Kaplan 2013).

  4. 4.

    The intuition for this last measure is that stops based on the ostensible furtive movements of the suspect are more likely to be discretionary, be perceived to be illegitimate, and therefore to produce negative interpretive effects. See, for example Goel et al. (2016).

  5. 5.

    Replication data and code are available at https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/AJYEDA

  6. 6.

    The set of interactions recorded in the database still may not constitute the universe of police-citizen interactions, as officers may not document all stops or may incorrectly record interactions (Ridgeway 2007). Nevertheless, the NYPD has significantly enhanced its auditing procedures to make the data more reliable, such that it is deemed reliable enough for use in legal procedures and other scientific analyses (Mummolo 2018; Gelman et al. 2007. See also, Goel et al. 2016).

  7. 7.

    Following McDonald and Popkin (2001), I estimate the voter-eligible population as the population of citizens over the age of 18 who are not incarcerated. New York State allows individuals on probation to vote and automatically restores the voting rights of all individuals who have completed parole, so no correction needs to be made for these populations. Unfortunately, the New York Board of Corrections does not release data that would allow for tract level estimation of the population currently completing parole, which may induce bias into the estimates.

  8. 8.

    The modeling specification is in this manner consistent with the data generating process. I present results from spatial model specifications in the Online Appendix. Results from these modeling specifications are consistent with the results presented in the main text.

  9. 9.

    Michener’s (2013) demonstration of a non-linear relationship between the perception of poor neighborhood quality and some forms of political participation suggests that there may be a non-linear relationship between policing, voter-turnout, and candidate choice. I present analysis of models run including a non-linear specification of the relationship in the Online Appendix. Results from the models are generally consistent with the results presented in the main text, though do suggest some variability in the relationship between policing and turnout across different communities.

  10. 10.

    Obama’s candidacy, for example, may have shifted white racial attitudes towards blacks (Goldman 2012).

References

  1. Alcindor, Y. (2015). Ferguson voters make history and increase turnout. USA TODAY. http://www.usatoday.com/story/news/2015/04/07/ferguson-voters-head-to-polls/25401037/.

  2. Alexander, M. (2012). The new Jim Crow: Mass incarceration in the age of colorblindness. New York: The New Press.

    Google Scholar 

  3. Ansolabehere, S., Palmer, M., & Lee, A. (2014). Precinct-level election data. https://www.dataverse.harvard.edu/dataset.xhtml?persistentId=hdl:1902.1/21919.

  4. Baer, A. (2015). From law and order to torture: Race and policing in de-industrial chicago. Evanston: Northwestern University.

    Google Scholar 

  5. Baicker, K., & Jacobson, M. (2007). Finders keepers: Forfeiture laws, policing incentives, and local budgets. Journal of Public Economics, 91(11), 2113–2136.

    Article  Google Scholar 

  6. Barbaro, M. (2013, May 29). William C. Thompson Jr. Takes moderate stand on police stops. The New York Times. http://www.nytimes.com/2013/05/30/nyregion/william-c-thompson-jr-takes-moderate-stand-on-police-stops.html.

  7. Barreto, M. A., Villarreal, M., & Woods, N. D. (2005). Metropolitan latino political behavior: Voter turnout and candidate preference in Los Angeles. Journal of Urban Affairs, 27(1), 71–91.

    Article  Google Scholar 

  8. Barreto, M. A., Manzano, S., Ramirez, R., & Rim, K. (2009). Mobilization, participation, and solidaridad Latino participation in the 2006 immigration protest rallies. Urban Affairs Review, 44(5), 736–764.

    Article  Google Scholar 

  9. Baumgartner, F. R., Epp, D. A., Shoub, K., & Love, B. (2017). Targeting young men of color for search and arrest during traffic stops: Evidence from North Carolina, 2002–2013. Politics, Groups, and Identities, 5(1), 107–131.

    Article  Google Scholar 

  10. Bloom, J., & Martin, W. E. (2013). Black against empire: The history and politics of the black panther party. Oakland: University of California Press.

    Google Scholar 

  11. Bobo, L., & Gilliam, F. D. (1990). Race, sociopolitical participation, and black empowerment. The American Political Science Review, 84(2), 377–393.

    Article  Google Scholar 

  12. Brady, H. E., Verba, S., & Schlozman, K. L. (1995). Beyond SES: A resource model of political participation. American Political Science Review, 89(02), 271–294.

    Article  Google Scholar 

  13. Bratton, W., & Knobler, P. (1998). Turnaround: How America’s top cop reversed the crime epidemic. New York: Random House Incorporated.

    Google Scholar 

  14. Brayne, S. (2014). Surveillance and system avoidance: Criminal justice contact and institutional attachment. American Sociological Review, 79(3), 367–391.

    Article  Google Scholar 

  15. Burch, T. (2012). Did disfranchisement laws help elect president bush? New evidence on the turnout rates and candidate preferences of Florida’s Ex-Felons. Political Behavior, 34(1), 1–26.

    Article  Google Scholar 

  16. Burch, T. (2013). Trading democracy for justice: Criminal convictions and the decline of neighborhood political participation. Chicago: University of Chicago Press.

    Google Scholar 

  17. Campbell, A. L. (2003). Participatory reactions to policy threats: Senior citizens and the defense of social security and medicare. Political Behavior, 25(1), 29–49.

    Article  Google Scholar 

  18. Carbado, D.W. (2016). Predatory policing. UMKC L. Rev. 85, 545–566.

    Google Scholar 

  19. Carr, P. J., Napolitano, L., & Keating, J. (2007). We never call the cops and here is why: A qualitative examination of legal cynicism in three Philadelphia neighborhoods. Criminology, 45(2), 445–480.

    Article  Google Scholar 

  20. Center for Urban Research. (2013). NYC election atlas—maps. http://www.nycelectionatlas.com/maps.html#!interactive.

  21. Cho, W. K. T., Gimpel, J. G., & Wu, T. (2006). Clarifying the role of SES in political participation: Policy threat and Arab American mobilization. Journal of Politics, 68(4), 977–991.

    Article  Google Scholar 

  22. Chong, D. & Rogers, R. (2005). Racial solidarity and political participation. Political Behavior, 27(4), 47–374.

    Article  Google Scholar 

  23. Dawson, M. C. (1994). Behind the mule-race and class in African-American politics. Princeton: Princeton University Press.

    Google Scholar 

  24. Eisinger, P. K. (1973). The conditions of protest behavior in American cities. American Political Science Review, 67(1), 11–28.

    Article  Google Scholar 

  25. Eith, C., & Durose, M. R. (2011). Contacts between police and the public, 2008. Washington, DC: US Department of Justice, Office of Justice Programs, Bureua of Justice Statistics. https://pdfs.semanticscholar.org/9532/21446c50f250ba96e76d38253d22a3c24c35.pdf.

  26. Epp, C. R., Maynard-Moody, S., & Haider-Markel, D. P. (2014). Pulled over: How police stops define race and citizenship. Chicago: University of Chicago Press.

    Google Scholar 

  27. Escobar, E. J. (1993). The dialectics of repression: The Los Angeles police department and the chicano movement, 1968-1971. The Journal of American History, 79(4), 1483–1514.

    Article  Google Scholar 

  28. Fagan, J., Tyler, T. R., & Meares, T. L. (2016). Street stops and police legitimacy in New York. In Comparing the democratic governance of police intelligence: New models of participation and expertise in the United States and Europe, p. 203.

  29. Fairdosi, A. S., & Rogowski, J. C. (2015). Candidate race, partisanship, and political participation. Political Research Quarterly, 68(2), 337–349.

    Article  Google Scholar 

  30. Fassin, D. (2013). Enforcing order: An ethnography of urban policing. Cambridge: Polity.

    Google Scholar 

  31. Felker-Kantor, M. (2014). Managing Marginalization from watts to rodney king: The struggle over policing and social control in Los Angeles, 1965-1992. Los Angeles: University of Southern California.

    Google Scholar 

  32. Fine, M., Freudenberg, N., Payne, Y., Perkins, T., Smith, K., & Wanzer, K. (2003). `Anything Can Happen with Police Around’: Urban youth evaluate strategies of surveillance in public places. Journal of Social Issues, 59(1), 141–158.

    Article  Google Scholar 

  33. Freedlander, D. (2013). Dante de Blasio’s killer ad may have won NYC primary for his dad. The Daily Beast. https://www.thedailybeast.com/articles/2013/09/14/dante-de-blasio-s-killer-ad-may-have-won-nyc-primary-for-his-dad.

  34. Gau, J. M., & Brunson, R. K. (2010). Procedural justice and order maintenance policing: A study of inner-city young men’s perceptions of police legitimacy. Justice Quarterly, 27(2), 255–279.

    Article  Google Scholar 

  35. Geller, A., Fagan, J., Tyler, T., & Link, B. G. (2014). Aggressive policing and the mental health of young urban men. American Journal of Public Health, 104(12), 2321–2327. https://doi.org/10.2105/AJPH.2014.302046.

    Article  Google Scholar 

  36. Gelman, A., & Hill, J. (2006). Data analysis using regression and multilevel/hierarchical models. Cambridge: Cambridge University Press.

    Google Scholar 

  37. Gelman, A., Fagan, J., & Kiss, A. (2007). An analysis of the new york city police department’s “Stop-and-Frisk” policy in the context of claims of racial bias. Journal of the American Statistical Association, 102(479), 813–823.

    Article  Google Scholar 

  38. Geys, B. (2006). Explaining voter turnout: A review of aggregate-level research. Electoral Studies, 25(4), 637–663.

    Article  Google Scholar 

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

    Article  Google Scholar 

  40. Goffman, A. (2015). On the run: Fugitive life in an American city. Basingstoke: Macmillan.

    Google Scholar 

  41. Goldman, S. K. (2012). Effects of the 2008 Obama presidential campaign on White racial prejudice. Public Opinion Quarterly, 76(4), 663–687.

    Article  Google Scholar 

  42. Goodman, J. D., & Baker, A. (2014). Wave of protests after grand jury doesn’t indict officer in Eric Garner chokehold case. The New York Times. http://1percent.info/wp-content/uploads/2014/12/2014-Wave_of_Protests_After_Grand_Jury_Doesn%E2%80%99t_Indict_Officer_in_pdf.

  43. Gottschalk, M. (2008). Hiding in plain sight: American politics and the carceral state. Annual Review of Political Science, 11, 235–260.

    Article  Google Scholar 

  44. Gould, J. (2013). The democratic mayoral candidates on stop-and-frisk and healthcare. WNYC. http://www.wnyc.org/story/313861-democratic-mayoral-candidates-stop-and-frisk-and-healthcare/?utmsource=sharedUrl&utm_medium=metatag&utm_campaign=sharedUrl.

  45. Hughes, K. A. (2006). Justice expenditure and employment in the United States, 2003. Bureau of Justice Statistics, NCJ212260. Washington, DC: US Department of Justice.

    Google Scholar 

  46. Jackson, J., Bradford, B., Hough, M., Myhill, A., Quinton, P., & Tyler, T. R. (2012). Why do people comply with the law? Legitimacy and the influence of legal institutions. British Journal of Criminology, 52(6), 1051–1071.

    Article  Google Scholar 

  47. Kaplan, T. (2013). Bristling Republican rivals discuss issues, and kittens, at debate - The New York Times. https://www.nytimes.com/2013/09/09/nyregion/at-debate-bristling-republican-rivals-discuss-issues-and-kittens.html.

  48. Kirk, D. S., Papachristos, A. V., Fagan, J., & Tyler, T. R. (2012). The paradox of law enforcement in immigrant communities. The Annals of the American Academy of Political and Social Science, 641(1), 79–98.

    Article  Google Scholar 

  49. Kriesi, H. (Ed). (1995). New social movements in Western Europe: A comparative analysis. University of Minnesota Press.

  50. Laniyonu, A. (2017). Coffee shops and street stops: Policing Practices in gentrifying neighborhoods. Urban Affairs Review. https://doi.org/10.1177/1078087416689728.

    Google Scholar 

  51. Leland, J., & Moynihan, C. (2012, June 17). Thousands March silently down fifth avenue to protest stop-and-frisk policies. The New York Times. https://www.nytimes.com/2012/06/18/nyregion/thousands-march-silently-to-protest-stop-and-frisk-policies.html.

  52. Lerman, A. E., & Weaver, V. M. (2014a). Arresting citizenship: The democratic consequences of american crime control. Chicago: University of Chicago Press.

    Google Scholar 

  53. Lerman, A. E., & Weaver, V. (2014b). Staying out of sight? Concentrated policing and local political action. The ANNALS of the American Academy of Political and Social Science, 651(1), 202–219.

    Article  Google Scholar 

  54. Logan, J. R., Xu, Z., & Stults, B. J. (2014). Interpolating US decennial census tract data from as early as 1970 to 2010: A longitudinal tract database. The Professional Geographer, 66(3), 412–420.

    Article  Google Scholar 

  55. Lotke, E., & Wagner, P. (2003). Prisoners of the census: Electoral and financial consequences of counting prisoners where they go, not where they come from. Pace Law Review, 24, 587.

    Google Scholar 

  56. Lulay. (2016). ‘Bye Anita’: How Chicago’s young black activists fought for Alvarez’s loss. DNAinfo Chicago. https://www.dnainfo.com/chicago/20160316/river-north/bye-anita-activists-celebrate-anita-alvarez-ouster-with-song-hashtag.

  57. Lundman, R. J., & Kaufman, R. L. (2003). Driving while black: Effects of race, ethnicity, and gender on citizen self-reports of traffic stops and police actions. Criminology, 41(1), 195–220.

    Article  Google Scholar 

  58. Maciag, M. (2014). Skyrocketing court fines are major revenue generator for Ferguson. Governing. http://www.governing.com/topics/public-justice-safety/gov-ferguson-missouri-court-fines-budget.html.

  59. Manza, J., & Uggen, C. (2006). Locked out: Felon disenfranchisement and American democracy. Oxford: Oxford University Press.

    Google Scholar 

  60. Mazerolle, L., Bennett, S., Antrobus, E., & Eggins, E. (2012). Procedural justice, routine encounters and citizen perceptions of police: Main findings from the queensland community engagement trial (QCET). Journal of Experimental Criminology, 8(4), 343–367.

    Article  Google Scholar 

  61. MacDonald, J., Fagan, J., & Geller, A. (2016).  The effects of local police surges on crime and arrests in New York city. PLos One, 11(6), e0157223.

    Article  Google Scholar 

  62. McDonald, M. P., & Popkin, S. L. (2001). The myth of the vanishing voter. American Political Science Review, 95(4), 963–974.

    Article  Google Scholar 

  63. Meredith, M., & Morse, M. (2017). Discretionary disenfranchisement: The case of legal financial obligations. Working Paper.

  64. Merolla, J. L., Sellers, A. H., & Fowler, D. J. (2013). Descriptive representation, political efficacy, and African Americans in the 2008 presidential election. Political Psychology, 34(6), 863–875.

    Article  Google Scholar 

  65. Mettler, S. (2007). Soldiers to citizens: The GI bill and the making of the greatest generation. Oxford: Oxford University Press.

    Google Scholar 

  66. Mettler, S., & Soss, J. (2004). The consequences of public policy for democratic citizenship: Bridging policy studies and mass politics. Perspectives on Politics, 2(01), 55–73.

    Article  Google Scholar 

  67. Meyer, D. S. (2004). Protest and political opportunities. Annual Review of Sociology, 30, 125–145.

    Article  Google Scholar 

  68. Michener, J. (2013). Neighborhood disorder and local participation: Examining the political relevance of “Broken Windows”. Political Behavior, 35(4), 777–806.

    Article  Google Scholar 

  69. Minkoff, S. L. (2016). NYC 311 a tract-level analysis of citizen-government contacting in New York city. Urban Affairs Review, 52(2), 211–246.

    Article  Google Scholar 

  70. Mummolo, J. (2018). Modern police tactics, police-citizen interactions, and the prospects for reform. The Journal of Politics, 80(1), 1–15.

    Article  Google Scholar 

  71. National Advisory Commission on Health. (1968). A report to the president. US Printing Office.

  72. New York City Primary Results. (2013). The New York Times. http://www.nytimes.com/projects/elections/2013/nyc-primary/mayor/exit-polls.html.

  73. NYCLU. (2017). Stop-and-frisk data. http://www.nyclu.org/content/stop-and-frisk-data.

  74. NYPD. (2016). NYPD patrol guide. New York City Police Department. https://www1.nyc.gov/site/ccrb/investigations/nypd-patrol-guide.page.

  75. Owens, E. G. (2017). Testing the school-to-prison pipeline. Journal of Policy Analysis and Management, 36(1), 11–37.

    Article  Google Scholar 

  76. Pantoja, A. D., & Segura, G. M. (2003). Fear and loathing in California: Contextual threat and political sophistication among Latino voters. Political Behavior, 25(3), 265–286.

    Article  Google Scholar 

  77. Park, S. (2017). Local revenue structure under economic hardship: Reliance on alternative revenue sources in California counties. Local Government Studies, 43(4), 645–667.

    Article  Google Scholar 

  78. Peffley, M., & Hurwitz, J. (2010). Justice in America: The separate realities of blacks and whites. Cambridge: Cambridge University Press.

    Google Scholar 

  79. Predergast, J. (2011). 2001 Riots led to top-down change for cincinnati police. USATODAY.COM. http://www.usatoday.com/news/nation/2011-04-03-cincinnati-riots-anniversary_N.htm.

  80. Reisig, M. D., & Kane, R. J (Eds). (2014). The Oxford handbook of police and policing. Oxford University Press.

  81. Ridgeway, G. (2007). Analysis of racial disparities in the New York police department’s stop, question, and frisk practices. Santa Monica: Rand Corporation.

    Google Scholar 

  82. Rios, V. M. (2011). Punished: Policing the lives of black and Latino boys. New York: NYU Press.

    Google Scholar 

  83. Rosenbaum, D. P., Schuck, A. M., Costello, S. K., Hawkins, D. F. & Ring, M. K. (2005). Attitudes toward the police: The effects of direct and vicarious experience. Police Quarterly, 8(3), 343–365.

    Article  Google Scholar 

  84. Ruback, R. B., & Clark, V. (2011). Economic sanctions in Pennsylvania: Complex and inconsistent. Duquesne Law Review, 49, 751.

    Google Scholar 

  85. Sharp, E. B. (2014). Minority representation and order maintenance policing: Toward a contingent view. Social Science Quarterly, 95(4), 1155–1171.

    Google Scholar 

  86. Smiley, D. (2015, May 16). McDuffie riots: Revisiting, retelling story—35 years later. The Miami Herald. http://www.miamiherald.com/news/local/community/miami-dade/article21178995.html.

  87. Soss, J. (1999). Lessons of welfare: Policy design, political learning, and political action. American Political Science Review, 93(02), 363–380.

    Article  Google Scholar 

  88. Stoudt, B. G., Fine, M., & Fox, M. (2011). Growing up policed in the age of aggressive policing policies. New York Law School Law Review, 56, 1331–1635.

    Google Scholar 

  89. Sunshine, J., & Tyler, T. R. (2003). The role of procedural justice and legitimacy in shaping public support for policing. Law & Society Review, 37(3), 513–548.

    Article  Google Scholar 

  90. Tyler, T. R., & Huo, Y. (2002). Trust in the law: Encouraging public cooperation with the police and courts through. New York: Russell Sage Foundation.

    Google Scholar 

  91. Tyler, T. R., Fagan, J., & Geller, A. (2014). Street stops and police legitimacy: Teachable moments in young urban men’s legal socialization. Journal of Empirical Legal Studies, 11(4), 751–785.

    Article  Google Scholar 

  92. Uggen, C., Larson, R., & Shannon, S. (2016). 6 million lost voters: State-level estimates of felony disenfranchisement. Washington, DC: The Sentencing Project.

    Google Scholar 

  93. United States Department of Justice. (2015). The Ferguson report: Department of justice investigation of the Ferguson police department.

  94. Vitale, A. S. (2008). City of disorder: How the quality of life campaign transformed New York politics. New York: NYU Press.

    Google Scholar 

  95. Walker, H. L. (2014). Extending the effects of the carceral state proximal contact, political participation, and race. Political Research Quarterly, 67(4), 809–822.

    Article  Google Scholar 

  96. Walker, H. L. (2016). Mobilized by injustice: Criminal justice contact, political participation and race. Seattle: University of Washington.

    Google Scholar 

  97. Weaver, V. M., & Lerman, A. E. (2010). Political consequences of the carceral state. American Political Science Review, 104(04), 817–833.

    Article  Google Scholar 

  98. Weitzer, R., & Brunson, R. K. (2009). Strategic responses to the police among inner-city youth. The Sociological Quarterly, 50(2), 235–256.

    Article  Google Scholar 

  99. West, E. A. (2017). Descriptive representation and political efficacy: Evidence from Obama and Clinton. The Journal of Politics, 79(1), 351–355.

    Article  Google Scholar 

  100. White, A. (2016). When threat mobilizes: Immigration enforcement and Latino voter turnout. Political Behavior, 38(2), 355–382.

    Article  Google Scholar 

  101. White, A., & Trump, K.-S. (2016). The promises and pitfalls of 311 data. Urban Affairs Review. https://doi.org/10.1177/1078087416673202.

    Google Scholar 

  102. Williams, R. G. (1993). Reading rodney king/reading urban uprising. New York: Routledge.

    Google Scholar 

  103. Willis, J. J. (2014). A recent history of the police. In M. D. Reisig & R. J. Kane (Eds.), The Oxford handbook of police and policing (pp. 3–33). Oxford: Oxford University Press.

    Google Scholar 

Download references

Author information

Affiliations

Authors

Corresponding author

Correspondence to Ayobami Laniyonu.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Supplementary material 1 (PDF 506 kb)

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Laniyonu, A. The Political Consequences of Policing: Evidence from New York City. Polit Behav 41, 527–558 (2019). https://doi.org/10.1007/s11109-018-9461-9

Download citation

Keywords

  • Policing
  • Stop and Frisk
  • Voter turnout
  • Carceral state
  • Policy threat