Cops on Campus: The Racial Patterning of Police in Schools

Abstract

This article describes how the use of sworn law enforcement in American schools is patterned by school racial composition. Three distinct measures are constructed using data from the Civil Rights Data Collection and the School Survey on Crime and Safety: police prevalence, the degree of exposure that students have to police officers within their schools, and the roles of officers within those schools. Results show that police have become increasingly prevalent in schools with the largest shares of white students, especially at the elementary level. Yet youth in schools with the most Black, Latinx, and Native American students experience the highest exposure to police, and police in these schools are more frequently directed to carry out punitive tasks such as discipline. Student exposure to police is also relatively common in the whitest schools, but officers in these settings are more often used for tasks unrelated to punishment, such as teaching.

Law enforcement officers with the power of arrest were present in more than half of all U.S. public schools in the 2017–2018 school year (Diliberti et al. 2019). While movements such as Black Lives Matter have brought attention to the concerning nature of this practice, policing in American schools is hardly a new phenomenon. The first school to station a police officer was in Flint, Michigan in the 1950s, and the practice soon spread throughout large, urban school districts in response to concerns over school integration, juvenile delinquency, and a desire to control Black liberation movements. Partnerships between schools and the police continued to increase and spread to the suburbs in the 1990s amidst pushes for community policing and mounting fears of mass shootings (Brown 2006; French-Marcelin and Hinger 2017; McKenna, Martinez-Prather, and Bowman 2016; Sojoyner 2013). These officers were described in ways that highlighted their institution-spanning positions, as a “new type of public servant; a hybrid educational, correctional, and law enforcement officer” (Morris 2016:74).

Although the history of police in schools is documented, we lack a thorough understanding of which schools, and thus which students, are most exposed to these agents of the state today. The distribution of school policing matters because police interactions are linked to numerous detrimental outcomes for youth, including lowered academic achievement, decreased connection with school, worse mental health, and increased sleep difficulties (Geller 2017; Geller et al. 2014; Jackson et al. 2020; Legewie and Fagan 2019; Theriot 2016; Weisburst 2019). When police are in schools, chronic absenteeism increases and youth are more likely to experience exclusionary school discipline or be arrested in school (Fisher and Hennessy 2016; Gottfredson et al. 2020; Homer and Fisher 2020; Na and Gottfredson 2011; Owens 2017; Sorensen et al. 2021; Theriot 2009), which can have far-ranging negative impacts into adulthood (e.g., Aizer and Doyle 2015; Brayne 2014; Hirschfield 2009; Rumberger and Losen 2016). ¹ If children from different racial and ethnic backgrounds are differentially likely to attend schools with police and police presence is associated with negative student outcomes, then school policing may be one mechanism through which educational and criminal justice system inequalities are perpetuated.

Yet the case of school policing presents us with a paradox. The growth of school policing over the past seven decades occurred alongside schools’ increasing use of carceral tactics, such as metal detectors, invasive surveillance, and the adoption of zero-tolerance policies (Hirschfield 2010, 2018; Shedd 2015). School policing’s growth also paralleled the broader societal movement toward an increasingly “law and order” state. As practices such as mass incarceration and broken-windows policing disproportionately affect communities of color (e.g., Alexander 2010), it would be reasonable to assume that policing is also more common in schools serving many students of color. However, we might also suspect school policing in the contemporary United States to be more evenly distributed across racial lines than trends in other means of social control, given the vast attention devoted to deadly mass shootings in predominantly white schools (e.g., Triplett, Allen, and Lewis 2014:360; Walker 2019).

In this article, I adjudicate between these possibilities by characterizing how contemporary school police presence varies with school racial composition. Such descriptive work is critical for both research and policy efforts because “in order to know what types of interventions might be useful—what problems need to be solved—we must understand the landscape of needs and opportunities” (Loeb et al. 2017:1). This article contributes to public conversations about whether and which schools should use police (e.g., Kamenetz 2020) by thoroughly describing the landscape of school policing. Specifically, I address three research questions: (1) What is the relationship between schools’ racial composition and their use of police? (2) What is the relationship between schools’ racial composition and students’ level of exposure to school police? and (3) What tasks do police reportedly engage in while on campus, and how does this relate to school racial composition? I build on past research by examining school police beyond simply accounting for their presence or absence. I propose a new measure, the officer-to-student ratio (OSR), to approximate students’ level of exposure to officers within their school. This measure is especially valuable given that scholars theorize the negative outcomes associated with school policing arise, in part, through interactions between police and youth. In addition, I also consider how the roles officers fill within schools might vary across contexts. This renewed examination of school policing trends is called for given the ongoing social and political debates over the role of law enforcement in society.

Why Do Schools Choose to Use Police?

The decision to use police in schools is made primarily at the state or district level. Although federal grants provide funding that can be used to support school police programs (James and McCallion 2013), policies regarding the funding, training, and certification requirements for school-based law enforcement are decided primarily at the state level (Canady, James, and Nease 2012:35). Some states operate their own grant programs to incentivize the use of police in schools (French-Marcelin and Hinger 2017; Heintz 2014). More recently, some states, such as Kentucky, now require that schools have at least one sworn law enforcement officer present (Krauth 2020).

Absent a state mandate, the decision to contract with law enforcement typically lies with the school district. The structure of this relationship generally takes one of two forms. First, a district may forge a relationship with the local police department to assign officers to the position of school resource officer (SRO). Some SROs may be stationed in a single school as their “beat,” to which they report every day and where they remain for the entirety of the school day. Other locales may have “mobile” SROs, who are not assigned a single setting and instead rotate through the schools responding to calls (Rosiak 2020). The second general form is when a school district creates their own police department. Rather than being employed by the local police department, officers in a school police department are employees of the district itself and are generally responsible to the superintendent (McKenna et al. 2016). In both arrangements, decisions about whether to station officers in every school, assign officers to only some schools, or have officers respond on an as-needed basis are typically made at the district level. Schools may also employ security, though these officers do not hold the power of arrest and are therefore not included in my analyses.

School district decisions regarding whether and how to place police in schools rely heavily on key actors’ beliefs about the role police play within an educational environment (Turner and Beneke 2020; Viano, Curran, and Fisher 2021). Many of the common rationales for using school police would lead us to predict greater police presence in schools that serve predominantly non-white students. These rationales often rely on ideas about neighborhood crime and narratives of insufficient adult role models, both of which are deeply embedded in racialized stereotypes about communities of color. For instance, school leaders may be concerned about crime in the neighborhood surrounding the school and wish to prevent students and/or visitors from bringing weapons or drugs into the building (Lynch, Gainey, and Chappell 2016:531; Travis and Coon 2005:186–91). This may be a stronger motivating factor in secondary compared with elementary schools, as older students may be seen as more susceptible to perceived bad influences. Decision makers may also seek out police to serve in a mentorship role (Community Oriented Policing Services n.d.; Coon and Travis 2012). Administrators may believe that police complement perceived deficits in communities’ or students’ lives, framing officers as potential role-models for young people who they view as not getting enough support from adults at home (Higgins et al. 2020; Turner and Beneke 2020). In some cases, police are even positioned as necessary interventions to combat the racism students of color face in school (Turner and Beneke 2020), which might lead to increased police presence in the schools of non-white students.

Thus, the patterning of school police might mirror other trends toward increasingly carceral tactics used in the schools of children of color. The schools of more marginalized students tend to use harsher discipline and rely to a greater extent on suspension, expulsion, police referrals, and arrests (Ramey 2015; Welch and Payne 2010, 2018), engage in more intense and invasive surveillance practices such as metal detectors and drug-sniffing dogs (Hirschfield 2010; Kupchik and Ward 2014; Nolan 2011; Servoss and Finn 2014), and use fewer restorative alternatives to punitive discipline (Payne and Welch 2015). These trends are largely consistent with racial threat theory, which predicts more social control measures in contexts in which a greater proportion of the population is non-white (Blalock 1967). If the distribution of school police is similar to the distribution of these other security measures, and thus consistent with racial threat theory, we might understand policing as just another tactic in the “universal carceral apparatus” of schooling (Shedd 2015).

Yet the aforementioned rationales fail to account for one of the most common reasons why school leaders use police: the belief that police presence protects the schools from threats of mass shootings (Coon and Travis 2012; Madfis 2016). This may be particularly motivating for elementary school leaders following the 2012 Sandy Hook shooting (Viano et al. 2021) and because younger children could be perceived as more innocent and vulnerable. Furthermore, there is some evidence that, on average, students who are aware of an SRO at their school are more likely to report feeling safe rather than unsafe at school (Pentek and Eisenberg 2018). Given the high amounts of media coverage for mass shootings in predominantly white schools, schools with large shares of white students may be more apt to use police to improve students’ feelings of safety compared with places where students do not match the demographic stereotypes associated with school shootings. Beyond fears of mass shootings, there are further reasons why the distribution of school policing may not align with the predictions of racial threat theory and the patterning of other carceral trends. Policing can be a costly investment for schools (Nance 2016; Raymond 2010), and thus districts with more financial resources, which tend to have more white students (EdBuild 2019; Sosina and Weathers 2019), may be more apt to use this practice. Moreover, school decision-makers might use police because they believe SROs will help students form positive attitudes towards law enforcement (Kupchik et al. 2020; Viano et al. 2021). School police often embark on these “public relations campaign[s],” in which officers tell young people to come to them if they are in trouble and to consider becoming police informants in the future (Viano et al. 2021:15–21). Although decision makers might desire for police to be “goodwill ambassadors” across a variety of school settings, there may be less community pushback to officers’ presence for this reason in places where people hold more positive views of the police. These are likely to be whiter communities, as white residents tend to feel more confident in their local police compared with Black residents (Morin and Stepler 2016). Moreover, police leadership may be most enthusiastic to partner with schools in places where they anticipate students and families will be most receptive, in order to solidify loyalty among groups they perceive as sympathetic already.

As these scenarios suggest, school policing trends might not mirror those for other carceral tools like harsh discipline and invasive surveillance. If so, this would emphasize the importance of examining policing as a unique dimension of school practices, imbued with its own cultural meanings that are distinct from other components of the “school-to-prison pipeline.” Scholars have similarly noted the importance of analyzing school security measures independently, pointing out that some are more or less intrusive, and more or less easily used to single out specific students for intensified surveillance (Hirschfield 2010; Kupchik and Ward 2014). School policing may also be distinct from other security practices because there is arguably far greater variety in how people understand the purpose of school police compared with the purpose of, for instance, school metal detectors. These variable functions are one reason to suspect that school policing may not be best understood through racial threat theory, which would predict the greatest police presence in the schools with the highest shares of non-white students. Varying ideas about officers’ roles could mean that police in one setting might primarily conduct social control oriented tasks, while police in another setting are less overtly oriented toward social control. For instance, officers in some schools may be focused on “detecting criminal offenses and offenders” while officers may be “assimilated into the culture of the school” elsewhere (Kupchik and Ward 2014:337). Thus, it is crucially important to fully describe not only where and how many police are located in schools, but also the role(s) those officers fill.

The Many Roles of School Police

With a wide-ranging set of potential roles, it is possible that police presence in educational environments could be used to the detriment of some students and the advantage of others. According to the U.S. Department of Justice, SROs have four main roles, some carceral (law enforcer, emergency manager) and others ostensibly less so (counselor, educator) (Community Oriented Policing Services n.d.). In practice, officers’ roles are often ambiguous, and even school administrators and SROs sometimes struggle to coherently define the boundaries and priorities of officers’ work (Coon and Travis 2012; McKenna et al. 2016). Given this ambiguity and the varied possible reasons for their presence, we might wonder whether the daily work of school officers is shaped by the demographic characteristics of the students served.

A substantial body of literature demonstrates that police roles vary on the basis of the community they are embedded within. As Monica Bell (2020) describes “forms and qualities of policing are intimately linked with neighborhood characteristics, especially but not exclusively neighborhood racial composition and poverty levels” (p. 688). This manifests, in part, as variation in police priorities across different settings. Officers in low-income communities or communities of color focus on addressing violence and engage primarily in crime control–related tasks such as conducting “proactive” police stops. In contrast, officers in affluent, predominantly white spaces focus on relationship building and engage primarily in service-oriented tasks such as enforcing traffic safety and checking in with community members and local businesses (Bell 2020; Gordon 2020; Magliozzi 2018). Advantaged residents mobilize their one-on-one relationships to have an agent of the state at their beck and call, leading policing to become seen as an attractive “amenity.” Scholars also theorize that the form of policing in racially diverse neighborhoods is distinct from that in racially segregated places (Bell 2020:722–28), emphasizing that the fundamental tasks of policing vary widely across communities with different racial compositions.

Like policing outside of schools, the roles school-based officers take on are likely to vary based on students’ characteristics. In 2006, SROs in schools serving more disadvantaged students took on more law enforcement functions, whereas SROs in schools with more advantaged student populations participated in more education-related tasks (Lynch et al. 2016). ² Notably, this analysis did not explore how roles might vary by school racial composition, although school police officers do rely on racialized tropes to understand their own roles. Officers who work in schools with greater proportions of Black students tend to perceive potential threats as internal: coming from students themselves. In contrast, officers who work in schools with greater shares of white students view their role as protecting students from external threats, such as an intruder (Fisher et al. 2022). This difference in focus, with the perceived threat located internally versus externally, might correspond to officers engaging in more versus less crime control–related tasks within the school, respectively. Officers in schools with more students of color may be more frequently engaged in tasks that focus on students as a potential source of danger. This might involve conducting searches of students and their belongings, patrolling the hallways, or responding to student misbehavior.

If the perception of threat is coming from outside in schools with more white students, then officers’ patrol in these settings may primarily take place on the exterior of the building. Within the school, officers may engage in tasks unrelated to crime-monitoring. Forming close police-community relations, like those in advantaged neighborhoods that view the police as an amenity, could be an attractive feature of bringing police into schools when the students themselves are not viewed as threats. Having an officer in the school might be an opportunity for students, families, and staff members to develop a close personal relationship with an officer, one they can invoke in the future if necessary. Administrators in advantaged schools may rely on school-based officers in a similar way. School leaders may view officers as private consultants, available at a moment’s notice to help them solve problems, train staff members, and teach students.

The Present Study

This research builds upon literature about schools’ use of security measures by analyzing how three different dimensions of police presence—prevalence, exposure, and their roles—vary across school racial composition. Past work offers high-level insight into the relationship between school demographics and police presence. Consistent with racial threat theory, schools with large shares of Black students are generally subject to higher levels of security, including more frequent use of police (Finn and Servoss 2013; Hernandez 2017; Irwin, Davidson, and Hall-Sanchez 2013; Nolan 2011; Servoss and Finn 2014; Shedd 2015; Theriot 2009). Similarly, the percentage of schools using police increased for schools with progressively larger shares of non-white students during 2005–2006 and 2009–2010. Data from 2015–2016, however, suggest that the relationship between percentage non-white and police presence no longer maintains its monotonicity: schools serving 5 percent to 20 percent of students who are non-white had the highest share which used police (National Center for Education Statistics 2017). This past work also collapses racial composition into two, three, or four categories, which glosses over potential variation between schools with high and low proportions of non-white students and across schools with varying degrees of racial segregation.

Moreover, past studies have done little to examine whether there may be a nonlinear relationship between racial composition and police presence in schools. ³ Yet according to some conceptions of racial threat theory, if the proportion of whites in a space is low enough, then any “threat” posed by nonwhite people would be only unto themselves, and high levels of social control are not necessary (Rocque and Paternoster 2011). This benign neglect means that we might expect social control (i.e., police presence) to be greatest in schools with more even numbers of white and non-white students. Other mechanisms might also produce curvilinear trends. For instance, many school administrators cite lack of funding as a reason they did not have police in their school (Coon and Travis 2012). If districts serving mostly Black, Latinx, and Native American students have fewer financial resources than predominantly white districts, we may expect reduced police presence in the districts with the highest proportions of students of color.

Furthermore, past work has typically relied on a binary measure of whether a school uses police (e.g., Finn and Servoss 2013; Kupchik and Ward 2014; Lindsay, Lee, and Lloyd 2018; Servoss and Finn 2014), which overlooks potential variation in the intensity of police presence. For instance, students in a school with multiple full-time officers may encounter an agent of the state with the power of arrest more often than students in a school with only one part-time officer. Scholars have begun to note the importance of examining school policing beyond a binary measure of presence. For instance, Curran et al. (2021) theorized that the effects of school policing depend on “whether an SRO is present . . . but also, relatively less studied, how often and in what ways SROs interact with students” (p. 6). In this article, I capture each of these dimensions: “whether an SRO is present” through my prevalence measure, “how often” through my exposure measure, and “in what ways” through my analysis of the roles officers fulfill. A simple marker of whether a school uses police provides important information on how widespread the practice is. We gain a more complete, nuanced picture by also examining variation within the subset of schools using police, to document potentially disparate exposure to and forms of policing across educational environments.

Lastly, I expand our knowledge by using recent population-level data and conducting analyses for all U.S. elementary and secondary schools. Much past work on schools’ use of security measures focuses solely on secondary schools (e.g., Finn and Servoss 2013; Irwin et al. 2013; Nolan 2011; Servoss and Finn 2014; Shedd 2015). Yet a nontrivial share of elementary schools have police (National Center for Education Statistics 2017), and the social processes culminating in elementary school police presence likely vary meaningfully from those for secondary schools. For instance, administrators’ safety concerns vary across level and different community or national events might catalyze discussions about SRO adoption in elementary and secondary schools (Travis and Coon 2005; Viano et al. 2021). Prior work similarly suggests that the patterning of security practices by school racial and socioeconomic composition is not consistent across level (Kupchik and Ward 2014). Therefore, I conduct separate analyses for the patterning of police presence in elementary and secondary schools.

Data

Measures of police presence are calculated from two different data sets. First, I use the 2013–2014 and 2017–2018 public-use Civil Rights Data Collections (CRDC), which include the entire population of U.S. public schools. The 2017–2018 survey included a school-level count of the number of sworn law enforcement officers (in full-time equivalents [FTEs]), ⁴ which I use to construct both prevalence and exposure measures. In 2013–2014, the survey included only an indicator for whether any sworn law enforcement was assigned to the school; thus, I am only able to calculate prevalence, not exposure, in the population for this year.

In addition to population-level CRDC data I use five recent waves (2003–2004, 2005–2006, 2007–2008, 2009–2010, and 2015–2016) of restricted-use data from the National Center for Education Statistics School Survey on Crime and Safety (SSOCS). Each wave of this nationally representative survey contains information on the number of full- and part-time sworn law enforcement officers in a school. From this, I first replicate the prevalence and exposure trends from the CRDC. Then, I use the SSOCS to build on the CRDC analyses in two ways. First, the SSOCS includes a measure of how dangerous the school leader perceives the school’s location to be (high, medium, or low crime), offering important context that likely shapes leaders’ decision-making processes about police utilization. Second, the SSOCS includes information on the tasks officers reportedly engage in, which allows me to analyze whether officers’ roles vary systematically with school racial composition.

Data in both the CRDC and the SSOCS are linked to that year’s Common Core of Data (CCD), from which I obtain the total enrollment as well as the racial and socioeconomic composition of the school. Syntax for analyses are available on OSF (https://osf.io/dr6xa/?view_only=50e6d570bb70494e8d534828e9ad3459).

Measures

I use three theoretically distinct school-level measures for the use of police in educational settings: their prevalence across U.S. schools, students’ exposure to police within a school, and the roles of officers within a school. I do not include security in my conceptualization of school police for two central reasons. First, the distinction between security guards and law enforcement is meaningful to key actors, as the former are nonsworn personnel (Viano et al. 2021). Second, even if they work for a school district police department, security guards do not retain the power of arrest, which is a key driver of inequality (e.g., Aizer and Doyle 2015; Brayne 2014; Hirschfield 2009). Therefore, I limit my analyses in this study to only sworn law enforcement. I include all sworn law enforcement in my conception of police, regardless of whether they are armed, because they hold the power of arrest even in the absence of firearms.

Prevalence

I first consider which schools use police in any capacity. I refer to this as the prevalence of school policing because, in aggregate, it describes how common the use of police is across the entirety of or across subsets of schools (e.g., among schools of a specific demographic composition). Concretely, prevalence is a binary school-level measure indicating whether a school uses police. For the bivariate results, prevalence is calculated from the 2013–2014 and 2017–2018 CRDCs. The SSOCS is used to replicate the bivariate results and for the regression-adjusted estimates.

Exposure

The exposure measure captures the frequency with which students might expect to encounter officers with the power to arrest within their school. Exposure to school policing is more complex than simply the number of officers who report to a given school. Specifically, the following three factors determine the level of exposure: (1) the number of officers, (2) the amount of time those officers are present in the building, and (3) the number of students.

The number of officers and amount of time those officers are present will shape the likelihood that a student will encounter them in school. For example, high exposure to police could be identified in a school in which an officer is present every day during all school hours. An even higher level would exist in a school in which there were multiple officers present every day all day. A low, nonzero level might occur in a school in which one officer is dedicated to policing the entire district, and thus the officer may not be present daily. The number of students also affects how exposed a student is to school policing. The police presence in a school of 2,000 students with one officer may feel far lower than the police presence in a school of 200 with one officer. In the latter, presumably all 200 students would be aware of the officers’ presence and have a high chance of encountering them. In contrast, there could be many students in the former who would not know an officer is stationed in their school. ⁵

In this study, I operationalize student exposure to police through the officer-to-student ratio (OSR) in each school. Generally, I define this quantity as the ratio of number of officers, in FTEs, to the number of students in a school. In the 2017–2018 CRDC the numerator was collected directly. In the SSOCS, each responding school reported the number of full-time and number of part-time officers, from which I estimate the number of officers in FTEs. I assume that officers listed as part-time by the school follow the “mobile” SRO model and are shared within a district, therefore only spending a fraction of their time in each school (Rosiak 2020). Full details on the calculation of the exposure measure using SSOCS data are available in Appendix B. For ease of interpretation, I scale the ratio so that the OSR represents the number of officers per 1,000 students. For the bivariate results, exposure is calculated from the 2017–2018 CRDC. The SSOCS is used to replicate the bivariate results and for the regression-adjusted estimates.

Officer Roles

To investigate officers’ roles, I use a series of questions from the SSOCS. Survey respondents (most of whom are school administrators) selected the activities that officers participate in at their schools. Administrators’ understanding of police roles is important because it sets the tone for those officers’ engagement in the school, dictates the types of incidents they may call on officers to assist with, and affects the way they frame the presence of officers to other staff members, students, and families. I examine all tasks that appeared in each wave of the SSOCS between 2003–2004 and 2015–2016 as well as the indicator for whether officers carry a firearm, as being armed with a lethal weapon likely shapes understandings of officers’ roles on campus. For ease of interpretation, I conceptualize these tasks as falling within one of two broad categories, on the basis of existing literature about police roles (e.g., Bell 2020). The more crime control-oriented dimensions of the job are: carries a firearm, conducts security enforcement and patrol, coordinates with local police and emergency teams, and maintains school discipline. The tasks I conceptualize as more service-oriented are: trains teachers and staff members on safety or crime prevention, identifies problems in the school and proactively seeks solutions to those problems, mentors students, and teaches a law-related education course or trains students (e.g., drug-related education, criminal law, or crime prevention courses). Nonetheless, I analyze the tasks separately to allow each a unique relationship with school racial composition. Further details are available in Appendix B.

At the school level, each role is measured using a binary indicator for whether officers engage in that activity based on data from the SSOCS. Role analyses use the same subsample of schools as for the exposure analyses: all schools in the analytic sample with nonzero officer-to-student ratios.

Analytic Sample

Analyses are restricted to traditional public and public charter schools. Special education-specific, juvenile justice, and alternative schools are excluded because they may make school security decisions differently than traditional public schools (e.g., Flores 2016) and because they are not included in the SSOCS sampling frame. Moreover, very small schools (enrollment < 20) were excluded because they likely also make decisions around school security differently. Schools were categorized as elementary or secondary according to National Center for Education Statistics definitions. All schools falling outside these categories (e.g., K–12 schools) are excluded. Last, I excluded schools in the top 0.5 percent of the officer-to-student ratio (OSR > 30 in the CRDC and OSR > 19 in the SSOCS; see Appendix A).

The resulting analytic sample from the 2013–2014 CRDC covers 98 percent of the total traditional public and public charter elementary schools in that year and 99 percent of elementary students. It also covers 98 percent of the total public traditional and charter secondary schools, which served more than 99 percent of secondary students (National Center for Education Statistics 2018). Florida and New York City Public Schools are missing from the 2017–2018 CRDC analytic sample because of data anomalies (see Appendix A). Therefore, the analytic sample for this year covers a relatively smaller portion of the school and student population: 93 percent of the total traditional public and public charter elementary schools that year, which served more than 92 percent of elementary students, and 96 percent of the total public traditional and charter secondary schools, serving more than 94 percent of secondary students (National Center for Education Statistics 2019). The final analytic sample across the five SSOCS waves contains approximately 6,700 elementary and 5,600 secondary schools.

Empirical Strategy

As the primary aim of this study is to thoroughly characterize the relationship between school police presence and school racial composition, I focus first on descriptive, bivariate results. Specifically, I examine the relationship between racial composition and each police presence measure (prevalence, exposure, and the eight tasks) for elementary and for secondary schools, as different social processes may be driving patterns in outcomes for younger and older children. The literature on stereotypes of criminality suggests that Black, Latinx, and Native American students may be seen by adult decision-makers as the most “threatening” and may invoke the greatest amount of social control within their schools (e.g., Eberhardt et al. 2004; Rios 2017; Saperstein, Penner, and Kizer 2014). Therefore, racial composition is operationalized as the proportion of students in a school that are Black, Latinx (of any race), or Native American; descriptive results for each racial/ethnic category are presented in Appendix D. ⁶ To account for potential nonlinearity, I fit quadratic trend lines according to racial composition using linear probability models for binary outcomes (prevalence and tasks) and ordinary least squares regression models for the continuous outcome (exposure). When using the CRDC, neither means nor fitted line estimations are weighted because the CRDC contains population-level data. All estimates from SSOCS data are obtained using survey weights. For ease of interpretation, results from linear probability models are shown for the binary outcomes, though all findings are comparable using logistic regression (see Appendix F).

Next, I build on the descriptive analyses with regression-adjusted results. The purpose of these analyses is to investigate whether the descriptive patterns between school racial composition and police presence could reflect the mutually constituted nature of race and other social characteristics in America. Specifically, I interrogate whether racial patterning of school police remains after adjusting for two variables correlated with racial composition that may also influence school leaders’ decisions to use police: socioeconomic composition and perceived levels of crime. Individuals tend to vastly overestimate their actual risk for criminal victimization (Quillian and Pager 2010). Thus, risk perceptions, as opposed to an objective measure of risk, are likely more influential in leaders’ decision making. I use the SSOCS for the regression-adjusted results because it includes principal ratings of the level of crime near the school. If patterns between racial composition and police presence remain in these models, this suggests that school policing is racially patterned above and beyond the correlations between school racial and socioeconomic composition and between school racial composition and decision-makers’ perceptions of neighborhood safety.

To generate the regression-adjusted results, I run linear probability (prevalence and tasks) or ordinary least squares regression (exposure) models predicting police presence based on the racial composition of the school, controlling for the proportion of students who receive free or reduced-price lunch in the school and for the perceived level of crime near the school. ⁷ To account for unmeasured features that may vary across time and place, all models include year and region fixed effects. Some schools, by chance, show up in multiple waves of the SSOCS (n = 1,210), so standard errors are clustered at the school level. As with the bivariate, descriptive trends, linear probability models are used with binary outcomes for ease of interpretation, though results using logistic regression are comparable. All regression results are included in the appendices.

Results

Prevalence of School Policing

Figure 1 shows the proportion of schools with a given racial composition that use police. In general, the prevalence of police in U.S. elementary schools does not vary substantially based on school racial composition. In contrast, police presence varies considerably by school racial composition for secondary schools. Although the relationship between school racial composition and police presence is relatively flat for elementary schools, there is an inverted U-shaped relationship among secondary schools and the secondary schools with the fewest Black, Latinx, or Native American students were least likely to use police.

Figure 1.

Prevalence of police presence in U.S. schools, 2013–2014 and 2017–2018.

Since the 2013–2014 school year, the patterning of school policing has shifted significantly. ⁸ Between the 2013–2014 and 2017–2018 academic years, the prevalence of police in schools with the most Black, Latinx, and Native American students declined, whereas the prevalence in schools with the fewest racially marginalized students remained steady (in the case of secondary schools) or increased (elementary schools). In 2017–2018, just as in 2013–2014, the patterning of school police presence is roughly linear for elementary and an inverted U-shape for secondary schools. Yet the shift that took place over these years is similar at both levels: the prevalence of school policing is holding steady or even increasing in the schools with the most historically advantaged student populations and decreasing in the schools with the most historically disadvantaged student populations. Especially notable is the 2017–2018 trend for police prevalence among elementary schools: it is elementary schools with the fewest Black, Latinx, or Native American students that were most likely to use police, rather than the schools with the most or middling shares. At the elementary level in 2017–2018, schools with completely white student populations used police more often than schools with no white students (Figure D1).

Data from the nationally representative SSOCS provide somewhat higher estimates for the proportion of schools that use police (Table D3). However, the trends in police prevalence by school racial composition are similar across data sets (Figure D2), supporting the conclusion that the prevalence of school policing is more strongly patterned by school racial composition for secondary schools than for elementary schools. Comparing SSOCS estimates over time also reveals shifts in the patterning of police prevalence similar to population-level CRDC results.

The bivariate finding that school police prevalence is more strongly patterned according to school racial composition for secondary schools than elementary schools remains in the regression-adjusted analyses (Table C1). The prevalence of school policing in elementary schools is not significantly patterned by the combined proportion of students that are Black, Latinx, or Native American when controlling for socioeconomic composition, perceived crime, and year and region fixed effects. In contrast, the prevalence of school policing for secondary schools remains significantly patterned by school racial composition after including controls.

Overall, two main findings emerge with respect to the prevalence of school policing. First, the prevalence of policing according to school racial composition varies more for secondary schools than for elementary schools. Second, over time police presence may be decreasing in schools with the most Black, Latinx, and Native American students and increasing or holding steady in those schools with the greatest share of advantaged students.

Exposure to School Policing

The prevalence findings tell us which schools use police but say little about how often students in schools with police might encounter officers—encounters that likely drive the negative student outcomes associated with school policing. In this section, I focus on those schools that use police and use the officer-to-student ratio as a proxy for the likelihood of officer-student contact. Given that students attend a school that uses police, does their exposure to law enforcement vary by the racial composition of the school?

On average, the OSR among schools that used police in 2017–2018 was 1.46 officers per 1,000 students in elementary schools and 1.52 officers per 1,000 students in secondary schools. These are higher than the benchmark “best practice” set by the National Association of School Resource Officers of 1 officer for every 1,000 students (Canady 2018).

The exposure measure adds nuance to findings about police prevalence. As described previously, police were more prevalent in the elementary schools with the fewest Black, Latinx, and Native American students by the 2017–2018 academic year (Figure 1). In isolation, however, this finding obscures important variation in the potential exposure of students to police within schools that use them. When examining exposure, a different pattern emerges: it is generally those schools in which the largest shares of students are Black, Latinx, or Native American where more officers per student are used (Figure 2). Youth in schools where large shares of students are from historically criminalized groups are roughly twice as likely to be exposed to police within their educational environments as youth in schools with few students from these backgrounds.

Figure 2.

Student exposure to police in U.S. schools that have police.

For both elementary and secondary schools, the average OSR is relatively similar for schools with single-digit shares of Black, Latinx, and Native American students up through schools in which roughly 80 percent of students are from these groups. Students’ potential exposure to police climbs most substantially when more than 80 percent of students in a school are Black, Latinx, or Native American. One important exception is the average OSR in the schools with the very largest proportions of white students, which is similar to the OSR of schools with the highest shares of marginalized students (see also Figure D4). SSOCS data reveal a similar pattern, with the highest levels of exposure in those elementary and secondary schools with the greatest shares of Black, Latinx, or Native American students, followed by the schools with the fewest Black, Latinx, and Native American students (Figure D5). ⁹ The time trends for exposure are less stable year-to-year than the prevalence trends, which might occur if the number and part- versus full-time status of officers shifts year-to-year and student enrollment fluctuates.

Again, I turn to the SSOCS to examine whether the racial patterning of police exposure persists after accounting for school socioeconomic composition and the principal’s perception of the nearby crime. Among schools that use police, the OSR remains significantly patterned by the school racial composition net of these controls (Table C2). Among elementary and secondary schools with police, schools with the fewest and most Black, Latinx, and Native American students have the highest OSRs. For elementary schools, this finding is sensitive to the treatment of part-time officers. When part-time officers are excluded from consideration, there is no longer a statistically significant relationship between the racial composition and OSR, net of other factors (Figure F1); however, the U-shaped relationship is exaggerated and statistically significant when all officers (part- or full-time) are considered equally (Figure F2). For secondary schools, the net curvilinear relationship between racial composition and OSR remains statistically significant regardless of the treatment of part-time officers, though the curvilinear shape is exaggerated when part-time officers are not distinguished from full-time officers. This suggests that the racial patterning of student exposure to police in elementary schools, net of socioeconomic status composition and perceived crime, is driven by the use of part-time officers. For secondary schools, a large part, but not all, of the patterning in student exposure stems from the allocation of part-time officers.

Overall, students’ potential exposure in schools that use police is not the same across contexts, and this unequal distribution is robust to the inclusion of controls. Although the prevalence of school policing shows different patterns by racial composition for elementary and secondary schools, student exposure to police within these schools is patterned by the combined proportion of students who are Black, Latinx, or Native American in nearly identical ways for elementary and secondary schools. Specifically, young people who attend schools with the highest shares of students from historically criminalized racial groups face the highest levels of exposure to police within their school environments. Those schools with the very whitest student bodies also have relatively high rates of exposure. Although the prevalence of school policing appears to be shifting, it is not yet clear whether these changes have resulted in stable or substantial changes to the patterning of students’ exposure. If we were to examine only the binary measure of prevalence, we would conclude that it is those schools with the most racially advantaged student populations that are adding police. While this is true, we miss the equally important fact that students who attend schools with the highest shares of racially marginalized students face the highest levels of exposure to school police.

The Varying Roles of School Police

As I have shown, measuring exposure in addition to prevalence provides greater context to our understanding of where and in what quantity school police are used. Importantly, though, similar levels of exposure to police within schools still does not suggest equivalent content or effects of this exposure. The relatively vague role of school police means that they may be engaged in a wide variety of tasks. Thus, the exposure that students have to officers might be experienced differently depending on the activities officers are engaged in when they interact with students.

Figure 3 depicts how officers’ tasks vary according to school racial composition. In the left-hand column are more crime control–oriented aspects of the job. In the right-hand column are the more service-oriented tasks. Broadly, rates of engagement in crime control–related tasks are higher for officers in secondary schools compared with officers in elementary schools. Rates of engagement in tasks such as patrolling the school and coordinating with local police surpass 85 percent for secondary schools, regardless of racial composition. Officers in secondary schools also train staff members and problem-solve to a greater degree than their counterparts in elementary schools.

Figure 3.

Proportion of schools in which officers engage in given tasks, among schools that use police.

Generally there is greater variation across school context for officers’ tasks in elementary schools compared with secondary schools. For some tasks, this variation is significantly patterned according to school racial composition. For instance, officers in elementary schools in which most students are Black, Latinx, or Native American are less likely to carry a firearm but more likely to patrol the school compared with officers working in schools with no Black, Latinx, or Native American students (Figures 3A and 3B). This aligns with the qualitative finding that officers stationed in schools with higher proportions of nonwhite students are more likely to surveil students inside the building as opposed to focusing on potential mass shooters coming from outside the school (Fisher et al. 2022). Similar to elementary schools, officers in secondary schools with more Black, Latinx, or Native American students are also less likely to carry a firearm and more likely to patrol the school, though the magnitude of these differences across school racial composition is much smaller (Figures 3A and 3B).

Across all racial compositions, school administrators frequently report that their officers help maintain discipline. Nevertheless, this use is racially patterned, especially in elementary schools, where police involvement in discipline is more common in schools with larger shares of students from historically criminalized groups (Figure 3D). The trend is similar for secondary schools, though the magnitude of differences is smaller. The increasing reliance on officers to maintain discipline for schools with greater shares of students from historically criminalized groups stands in contrast to the patterning of officers’ involvement in the service-oriented task of teaching. The larger the share of students in a school that are Black, Latinx, or Native American, the less likely the school is to use the officer for instruction (Figure 3H). The pattern is similar for the schools of younger and older students.

For elementary schools, much of the racial patterning in police tasks can be explained by the socioeconomic composition and perceived crime near the school (Table C3). For secondary schools, many tasks remain significantly patterned by racial composition net of these controls. Specifically, after accounting for socioeconomic composition and perceived crime, officers in secondary schools with more Black, Latinx, or Native American students are more likely to patrol the school, coordinate with local police, engage in discipline, and help problem-solve while they are less likely to teach courses. Engagement in mentorship roles is highest in secondary schools with middling shares of students who are Black, Latinx, or Native American.

Although some of the variation in officers’ tasks can be statistically explained by other contextual variables, the mutually constituting nature of race, class, and perceptions of crime means that these factors cannot be disentangled in the daily lives of Americans. As such, I place a greater emphasis on the bivariate trends, which reveal substantively meaningful differences in officers’ activities for tasks that closely overlap with educators’ daily routines: maintaining school discipline and teaching courses. In elementary schools in which no students are Black, Latinx, or Native American, roughly half of elementary schools with police report using them for discipline. The proportion increases to roughly 70 percent in elementary schools in which all students are Black, Latinx, or Native American. In both elementary and secondary schools with no Black, Latinx, or Native American students, roughly 50 percent of officers teach, compared with roughly 30 percent of officers in schools with the most Black, Latinx, and Native American students.

Together with the findings in the previous section, we see that not only is students’ exposure to police patterned by the racial composition of the school, but the reported tasks of officers are also patterned by racial context. Officers are more often engaged in acts of punitive social control within the school (e.g., patrol, discipline) and less often engaged in educational activities in schools with more students from historically marginalized groups. In other words, both students’ potential for exposure to police within their schools and the ways in which they are likely to interact with those police vary by the racial composition of the school.

Discussion

In this article, I show how the prevalence, quantity, and form of school policing is patterned by racial context in contemporary American schools. The prevalence of police in American secondary schools is significantly patterned by racial composition in an inverted U-shape: there is an increasing relationship between the proportion of students that are Black, Latinx, or Native American in a school and the use of school police up to a threshold, after which the prevalence of police decreases. A similar inverted U-shape has been found for the relationship between school racial composition and punitive school discipline, which may reflect the “benign neglect” hypothesis of racial threat theory, in which social control decreases when there are few dominant group members “at risk” (Rocque and Paternoster 2011). The decline in prevalence seen here could also be explained if schools with the most students from historically marginalized groups have fewer financial resources with which to pay for policing. In recent years, this decline in police prevalence for schools with the largest shares of racially marginalized students became more pronounced. In contrast, the patterning of school police prevalence in elementary schools does not resemble this inverted U (with the greatest police prevalence in the most racially heterogeneous schools). Police prevalence for elementary schools varies less across racial contexts compared with secondary schools, and the prevalence of school policing shifted toward elementary schools with the largest shares of white students over time.

Despite the shift in policing prevalence toward schools with more white students, youth in schools with the highest shares of racially marginalized students face the greatest exposure to police within their educational environment. Surprisingly, student exposure is similarly high in the schools with the most racially advantaged students that use police. Specifically, student exposure in schools that use police follows a U-shaped trend for both elementary and secondary schools: schools with the most historically marginalized students have the highest levels of exposure, but schools with the most racially advantaged students are also quite likely to promote officer-student contact. This high exposure to policing at both ends of the spectrum does not conform to racial threat theory but does parallel policing in neighborhoods, which tends to operate with high intensity in the least and most advantaged communities (e.g., Bell 2020; Gordon 2020; Magliozzi 2018).

The form of high-intensity policing varies meaningfully across contexts. I find that even in schools with similarly high OSRs, the activities through which they may engage with police vary by the racial composition of the school. In schools with more Black, Latinx, and Native American students, officers are more often engaged in discipline and less often engaged in educational activities. Variation in officers’ tasks is especially pronounced at the elementary level. Because officers perceive younger students as more malleable (Kupchik et al. 2020), racialized stereotypes defining how those children need to be molded may play a greater role in determining the tactics officers use when engaging with young children. Although levels of exposure to police are similar in schools with the most and least racially advantaged student bodies, officers’ daily tasks, and thus the potential consequences of officer-student interactions, are not necessarily equivalent across context.

A few limitations are important to note. First, both the CRDC and SSOCS include data inconsistencies in the reporting of school law enforcement. For instance, some schools in both datasets report surprisingly large numbers of officers (e.g., 50 or 100), whereas other places are missing data altogether (e.g., New York City in the 2017–2018 CRDC; see Appendix A). Future data collection efforts should more carefully assess the quality of measures reporting the use of school police. Second, this study is focused on the roles of officers as defined by school leaders, in contrast to other work (i.e., Fisher et al. 2022) that examines how officers understand their own roles. Although school leaders’ ideas about officers’ tasks are important, administrators and officers don’t always describe officers’ roles in the same way (Coon and Travis 2012). Because school leaders generally report lower levels of police involvement than law enforcement leaders (Coon and Travis 2012), we can understand the estimates presented as a conservative case of how officers might report their own tasks. Nonetheless, trends observed here may not match trends in how officers’ reports their own activities, which in turn may not align to the tasks officers’ actually conduct on a daily basis (Jerolmack and Khan 2014). Last, the variables included in the models here account for only a small fraction of variation in which schools use police, suggesting that while school policing is significantly racially patterned in America, many other factors would need to be accounted for to build a full causal model explaining why some schools, and not others, use police.

Despite its limitations, the differences in school officers’ roles across racial contexts suggested by this study will be particularly important if police continue to become increasingly prevalent in elementary schools and in the schools with the most racially advantaged students. Heightened fear over school shootings is one likely explanation for the shifting patterns documented here, which occurred most notably in the mid-2010s. Following the 2012 Sandy Hook shooting, both Democrats and Republicans made efforts to increase the use of police officers in schools (French-Marcelin and Hinger 2017; McCluskey 2017; The White House n.d.). Schools serving high proportions of white students may have more readily responded to such calls, as these schools were less likely to already use police, tend to have more financial resources, and may have been particularly concerned about the potential for mass shootings given similar demographics to places such as Sandy Hook.

Although calls for increased school policing abound after mass shootings, social movements have also brought much-needed attention to the role of police in upholding racial inequality and activists such as the Black Organizing Project have worked for years to abolish the use of police in schools. This movement reached a key milestone in the summer of 2020, when the emergence of #PoliceFreeSchools and protests across the country catalyzed some districts to end their relationships with local police departments. In just the eight weeks following George Floyd’s murder, 15 school districts ended their partnerships with police. This is a stunningly rapid change compared with the usual process of educational policy reform. The school districts that ended their partnerships tended to be larger, urban districts that, on average, teach significantly higher proportions of Black, Latinx, and Asian students and significantly lower proportions of white students compared with all other U.S. districts (Table D4). Throughout the 2020–2021 academic year, debates over school policing continued in many locales (e.g., Rantala 2021; Zilis 2020). Could these early trends from the summer of 2020 continue and further shift the prevalence of school policing toward schools with the most white students? If so, fewer students from historically criminalized groups may be exposed to police within their educational environments and face the accompanying threats of hypersurveillance and arrest.

However, there are reasons to be concerned about policing in predominately white spaces. First, racial inequality in how adults respond to student behaviors exists in majority-white and in socioeconomically advantaged schools (Drake 2022; Lewis and Diamond 2015). In fact, school police officers in diverse and majority-white settings explicitly seek out students of color for interactions, perceiving that these are the students who have negative views of police and seeking to improve their opinions of law enforcement. This places students of color, regardless of the school racial composition, at greater risk for hypersurveillance and for the criminalization of what might otherwise be categorized as typical youth behavior (Kupchik et al. 2020). Students of color in predominately white settings may face particularly high risk, as I show that officers in these contexts are more likely to be armed with lethal weapons. Additionally, SROs cultivate relationships with children in the hopes that they become future informants (Viano et al. 2021). This not only extends the reach of the surveillance state but could do so in ways that exacerbate existing power dynamics. If police are successful in this work and are most prevalent in the whitest schools, then it is racially advantaged students that have the most access to forming personal relationships with police, using those police resources as an “amenity,” and to evading state punishment in the future. Future work must further explore the form of school policing “in practice” in the most advantaged spaces. For instance, I show that officers in these settings are more likely to teach. What are the explicit and implicit lessons conveyed in such courses? Policing in affluent neighborhoods serves to further protect the advantages of the most well-resourced (Bell 2020); ethnographic work on policing in the most advantaged schools would elucidate whether the same is true in educational environments.

The current moment represents an important potential inflection point in how police presence in schools is patterned by racial composition, primarily because many school districts only temporarily ended their contracts with police. For instance, Seattle Public Schools suspended their police contract for one year, and Sacramento’s 2020–2021 school district budget did not include a police contract. These changes occurred in a summer when, amid the coronavirus pandemic, it looked ever more likely that school districts would have to adapt their instructional methods for the upcoming year. It is possible that many districts did not deem police necessary during a year of what was likely to be disrupted, often virtual learning and—upon return to full capacity, in-person instruction—may eventually reinstate their use of police. Thus, ongoing work must continue to track the patterning of school police presence and its effects on students. Since school policing is currently unequally distributed across schools and is associated with potentially harmful and disparate effects on students even in the most advantaged settings, efforts to remove police from schools offer the promise of greater educational equity.

Supplemental Material

sj-pdf-1-srd-10.1177_23780231221108037 – Supplemental material for Cops on Campus: The Racial Patterning of Police in Schools

Supplemental material, sj-pdf-1-srd-10.1177_23780231221108037 for Cops on Campus: The Racial Patterning of Police in Schools by Rebecca D. Gleit in Socius