School-based law enforcement strategies to reduce crime,increase perceptions of safety,and improve learning outcomes in primary and secondary schools: A systematic review

Systematic review evidence supports the criticism that school-based law enforcement criminalizes students and schools

School-based law enforcement (SBLE) includes a variety of forms of sworn law enforcement officers stationed in schools on at least a part-time basis. Although SBLE is intended to maintain school safety, critics claim it is ineffective in this role and has negative consequences for students. The existing literature does not find that SBLE makes schools safer. The evidence suggests that SBLE has detrimental consequences, particularly for exclusionary discipline.

What is this review about?

The past several decades has witnessed a steady growth of the presence of SBLE around the world. There is no standard definition for SBLE, nor is there a common training, job description, or chain of authority. In the USA, the two most common approaches are to contract officers from a local law enforcement agency to work in the school, and school district police departments, in which the local school or district employs its own sworn police force.

There is debate about whether SBLE is effective or harmful.

This study synthesizes the empirical literature that examined the relationship between SBLE presence and school-related outcomes including (a) crime and behavior problems; (b) perceptions of school; and (c) student learning.

What is the aim of this review?

The objective of this Campbell systematic review is to examine the relationship between school-based law enforcement (SBLE) presence and school-related outcomes.

What studies are included?

This review includes quantitative studies that examine any outcomes related to crime and behavior, perceptions of safety, and learning outcomes. Included studies must have compared outcomes in units with SBLE to units without SBLE (including different dosages of SBLE, such as increasing SBLE presence).

What are the main findings of this review?

Contrary to what is intended, schools with SBLE had higher rates of crime and behavior problems than schools without SBLE. This was primarily driven by higher rates of exclusionary discipline, with no detectable improvements to school crime or violence.

We also found that students in schools with SBLE tended to feel safer at school, although this finding is less trustworthy because it is based on very little data.

None of our other analyses showed that SBLE had beneficial effects.

These results suggest that SBLE is an ineffective practice for keeping schools safe, and even has detrimental consequences.

The findings are based on studies that vary in the extent to which we can infer causal relationships between SBLE and the outcomes of interest, so we urge caution in interpreting this as a cause-and-effect relationship. Still, the study's findings are consistent across studies with different methodological characteristics.

What do the findings of the review mean?

These findings suggest that SBLE is not an effective strategy for maintaining school safety and lead to harm for students in the form of exclusionary discipline. However, limitations in the strength of the evidence affect the ability to claim that these relationships are causal.

How up-to-date is this review?

The final electronic search for the literature included in this review was carried out on 17 July 2020.

The problem, condition, or issue

The past several decades has witnessed a steady growth of the presence of school-based law enforcement (SBLE) around the world. SBLE includes a variety of forms of sworn law enforcement officers stationed in schools on at least a part-time basis, and often on a full-time basis. SBLE might include local police officers, sheriff's deputies, officers from school-based police agencies, or a variety of other possible arrangements. There is no standard definition for SBLE, nor is there a common training, job description, or chain of authority. In the United States, the two most common approaches are to contract officers from a local law enforcement agency to work in the school, and school district police departments, in which the local school or district employs its own sworn police force. Although data on the number of schools using police are elusive, some estimates are available. The 2019 Indicators of School Crime and Safety (Musu et al., 2019) report showed that according to principal self-reports, 42% of public schools had at least one SBLE officer present at least 1 day a week during the 2015-2016 academic year.

Although similar figures are not available in other countries, the use of SBLE is nevertheless widespread internationally. For example, schools and school districts in Canada (Broll & Howells, 2019), Great Britain (Henshall, 2018), and Australia (Mazerolle, Bennett, et al., 2017) regularly use SBLE in one form or another. Until recently, support for police in schools—at least in the United States—has crossed political lines, with SBLE receiving political and financial support at the federal level from both major political parties (Blad, 2020; Pilkington, 2013). This support has begun to wane since the summer of 2020, following the murder of George Floyd by the Minneapolis Police Department, when school districts nationwide began to remove SBLE out of concerns about racially disparate policing practices and the negative consequences for Black students in particular.

From a law enforcement perspective, the primary intended purpose of SBLE is to maintain school safety (Community Oriented Policing Services,  n.d.). That is, SBLE should be able to prevent, mitigate, and respond to incidents of school-based crime and violence. As law enforcement officers, they are trained to recognize and respond to threats and to conduct investigations; stationing officers with this training in school is intended to make schools safer places. This is a particular concern among the public in the United States, where high-profile incidents of gun violence in schools have demanded responses from policymakers and school leaders alike. SBLE has often been a politically expedient response to these high-profile incidents (Madfis, 2016; Viano et al., 2021), even while silencing or ignoring perspectives that suggest SBLE might have harmful consequences (Koon, 2020; Nolan, 2011; Turner & Beneke, 2020).

In spite of the steady growth of SBLE, criticisms of SBLE have grown louder and more pointed in recent years (Counseling Not Criminalization in Schools Act, S.2125, 2021–2022). Where it once was taken for granted that SBLE would be present in schools, many large school districts have begun the process of removing SBLE. Notable examples of this are in Toronto, Canada, and several major US cities including Minneapolis, Denver, Seattle, Oakland, and others. Calls to remove SBLE have largely focused on two criticisms: that SBLE (a) does not make schools safer, and (b) has disproportionately negative impacts on students who may already experience marginalization in school settings, including non-White students, students with disabilities, queer students, and others. SBLE have been implicated in the so-called “school-to-prison pipeline,” a process by which students of color are funneled from schools into carceral settings. Although the metaphor of a “pipeline” has its shortcomings (Crawley & Hirschfield, 2018; McGrew, 2016; Simmons, 2017), the concerns about SBLE nevertheless reflect the increasing interconnectedness of education and the criminal legal system.

How the intervention might work

A general theory of change for SBLE strategies is provided in Figure 1. As Figure 1 denotes, there are a number of interventions that fall under these strategies, including various police patrol strategies, school resource officer allocation, police-school liaisons, safe routes/corridor programs, and community and problem-oriented policing strategies specifically targeting the schools.

OPEN IN VIEWER

Figure 1

General theory of change for school-based law enforcement programs.

One expectation for how SBLE might work is grounded in the logic of crime deterrence (Becker, 1968; Cohen & Felson, 1979), in which crime is less likely to occur in situations where a potential offender perceives a high risk that they will be caught. In this vein, the underlying theory of SBLE programs is that they will increase police surveillance activities and interaction with students and staff at the schools. In turn, this is expected to reduce criminal behavior and non-criminal student misconduct, increase perceptions of safety by students, staff, and parents (Theriot & Orme, 2016), and lead to more positive relationships between police with young persons and with the schools (Kupchik et al., 2020). This may also have secondary consequences that result in increased attendance, test scores and grades, and improved perceptions of the school and its climate.

However, there is possibility of unintended negative consequences that police presence leads to overuse of legal responses to behavior normally resolved by school administration (Hirschfield, 2008; Kupchik & Monahan, 2006; Simon, 2007), leading to more punishment of students (Kupchik, 2010; Nolan, 2011; Theriot, 2009), particularly among already marginalized students including racial and ethnic minorities and youth with disabilities (Homer & Fisher, 2020; Morris, 2016; Weisburst, 2019). SBLE programs could also lead to increased fear of crime if the presence of the officer sensitizes students to the possibility that violence may occur (Curran et al., 2021). Additionally, due to increased police scrutiny, it could lead to objective measures of misbehavior increasing simply because the police are on site and able to detect more student offending. These increases in surveillance, fear, punishment, and behavior problems could have a radiating effect on the school environment in which they foster negative perceptions of the school and harm learning outcomes such as grades, test scores, and attendance.

Why it is important to do this review

The use of SBLE is met with contrasting expectations in terms of theories of change, empirical findings, and political arguments alike. Proponents of SBLE point to the potential crime deterrent effect of SBLE and suggest that stationing officers in schools can help repair the strained relationship between law enforcement and the public. Opponents, on the other hand, contend that SBLE has a criminalizing effect on students and schools, leading to students’ increased contact with the criminal legal system. Both of these perspectives can cite theories to justify their stances and can cite empirical research that will back their claims. To help inform this debate that is occurring both in academia and the public sphere (Counseling Not Criminalization in Schools Act, S.2125, 2021–2022), this review will summarize the current body of research on SBLE, providing a quantitative synthesis of this literature. For policymakers and practitioners, this summary will provide guidance for real-world decisions about whether and how to use SBLE. For researchers, it will corral a topic of study that spans multiple academic disciplines including Education, Criminal Justice, Sociology, and others.

Potential positive effects

(1)

Examine the effects of policing strategies in primary and secondary schools (K-12 in USA) on measures of crime and non-criminal student misbehavior (e.g., certain levels of bullying, harassment, etc.) in and around schools

Potential negative consequences

(1)

Examine the effects of policing strategies in primary and secondary schools on arrests, formal sanctions, and exclusionary discipline of students (e.g., suspensions, expulsions)

Criteria for considering studies for this review

This study has a registered protocol (Petrosino et al., 2014), although some changes were made as described below.

Search methods for identification of studies

This review benefited from an earlier systematic search process conducted by the WestEd Justice & Prevention Research Center for George Mason University's Center for Evidence-based Crime Policy. The CEBCP received funding from the UK National Police Improvement Agency for a series of systematic reviews, and our proposed review of “policing schools” was selected. For that project, we searched all available years of 64 databases through December of 2009 (see Supporting Information: Appendix A), contacted 25 researchers and experts in school safety and/or school policing, and reviewed references in over 100 prior reviews and meta-analyses (Petrosino et al., 2012).

The Campbell Collaboration review was initiated, following title registration and then protocol approval in 2014. For our searches, we examined 18 databases. Given our exhaustive searches through 2009 for the earlier review, we only examined abstracts available during 2005–2015 to provide some overlap with the earlier search process but not to be completely duplicative with those efforts. Unfortunately, many of the databases that we had access to in wave 1 were not available to us during wave 2 so there is not perfect symmetry between databases.

Because an inordinate amount of time had elapsed before we could finish the C2 editorial process, we decided to update the searches to cover more recent years to see if we found newly published studies. Those searches covered 35 databases across the years 2015–2020 (see Supporting Information: Appendix A). Again, there is not 100% symmetry as the databases we had access to during this last stage changed.

Data collection and analysis

Selection of studies

Studies were screened for inclusion using a multi-stage process. The first stage was a title and abstract screening, in which members of the research team read the titles and abstracts, excluding any reports that were clearly off-topic. Next, team members conducted a full-text screening, verifying that each report met all the eligibility criteria for inclusion, excluding any reports that did not meet these criteria (see Figure 2). All reports reviewed in the full-text screening were reviewed by two team members, and any disagreements were resolved by reaching a consensus through discussion. Because consensus was the goal, we did not conduct interrater reliability.

OPEN IN VIEWER

Figure 2

PRSIMA study search and screening flow diagram.

Description of studies

Risk of bias in included studies

The risk of bias in the included studies was high. There were no studies that used random assignment, so our assessment of risk of bias was based on the methodological features of the quasi-experiments, looking in particular for features of the study design or analyses that would improve both internal and external validity. We extracted effect size estimates that were derived from a variety of designs, ranging from cross-sectional studies without any controls to difference-in-difference designs with a robust set of controls to help mitigate the potential impact of selection bias. Several studies used rigorous quasi-experimental designs to estimate the causal impact of SBLE on outcomes of interest. For example, two studies (Owens, 2017; Weisburst, 2019) used instrumental variable approaches with credibly exogenous variation in the implementation of SBLE. Another study (Gottfredson et al., 2020) used a comparative interrupted time series design with observations at 43 time points. Additionally, five studies (Anderson, 2018; Devlin & Santos, & Gottfredson, 2018; Owens, 2017; Scanlon, 2016; Sorensen et al., 2021) used a difference-in-differences framework to examine change over time relative to a comparison group. Another seven studies (Barnes, 2008; Devlin & Fisher, 2021; Devlin & Gottfredson, 2018; Dohy & Banks, 2018; Fisher, 2016; Fisher & Devlin, 2020; Na & Gottfredson, 2013) used either student fixed effects with student-level data or school fixed effects with school-level data to examine within-unit change over time, or controlled for a lagged measure of the outcome. Additionally, four other studies (Brady et al., 2007; Homer & Fisher, 2020; Swartz et al., 2016; Zhang, 2019) applied matching or weighting techniques to reduce selection bias. The remaining studies did not include elements of either their design or analysis that permitted strong causal inferences.

There was also variability in the external validity of the included studies. Some were very strong, including three studies that used nationally representative data from the United States (Boman & Mowen, under review; Homer & Fisher, 2020; Swartz et al., 2016). Seven other studies with particularly strong external validity used statewide data from a single US state (Anderson, 2018; Barnes, 2008; Fisher, 2016; Sorensen et al., 2021; Sullivan, 2013; Weisburst, 2019; Zhang, 2019). Another six studies drew on data that were nationally representative in their original form, but used a subset of data that made the data no longer nationally representative (Devlin & Fisher, 2021; Devlin & Gottfredson, 2018; Devlin & Santos, & Gottfredson, 2018; Fisher & Devlin, 2020; Na & Gottfredson, 2013; Owens, 2017). The remaining studies had weak external validity, focusing on a set of schools that was not evidently representative of a broader population.

Synthesis of results

The results from the series of meta-analytic models are presented in Table 2 As shown, for each set of outcomes, models were estimated separately for effect sizes that were based on studies that used either students or schools as the unit of analysis.

Summary of main results

This study's findings indicate that SBLE programs were associated with higher levels of crime and behavior problems in studies where schools were the unit of analysis, but studies with students as the unit of analysis yielded no statistically significant differences for crime and behavior outcomes. This is particularly noteworthy given that the presumed purpose of SBLE is to deter crime and behavior problems; this study's findings provide no such evidence, and even suggest that the opposite may be true—that crime and behavior problems increase with SBLE. In examining more specific measures of crime and behavior, schools with SBLE had more discipline-related outcomes in studies with both schools and students as the unit of analysis. There were no statistically significant differences between schools with and without SBLE for criminal justice system contact, violence, substance-related outcomes, or weapon-related outcomes.

Given this set of findings pertaining to outcomes related to crime and behavior, it seems that the association between SBLE and more crime and behavior problems is largely driven by outcomes related to schools’ use of exclusionary discipline such as suspensions. This is important because exclusionary discipline is not a direct measure of student behavior, but is instead a measure of schools’ responses to student behaviors. In fact, although one might assume that suspensions are reserved for only the most severe offenses, research has demonstrated that a substantial number of suspensions are administered for highly subjective offenses (e.g., disorderly conduct) or repetitive low-level offenses (e.g., dress code violations). Moreover, research has demonstrated that schools’ use of exclusionary discipline is not a corrective for students, but puts them on a pathway toward future punishment and behavior problems (Gerlinger et al., 2021), even with radiating negative effects on classmates who were not excluded (Perry & Morris, 2014).

Although we were unable to empirically assess the potential mediating pathways linking SBLE to increased exclusionary discipline, there are multiple theoretical pathways that merit further examination. First, SBLE may increase the detection of student behaviors without affecting either how students behave or how schools respond to student behaviors (Na & Gottfredson, 2013). Second, SBLE may become involved in the process of administering student disciplinary sanctions, advising school principals and disciplinarians on how they ought to punish students (Kupchik, 2010). Third, SBLE may shift the overall climate of the school to one that is more punitive (Devine, 1996), relying on more formal responses to student behavior (e.g., suspension) rather than informal ones (e.g., corrective conversations with teachers). Fourth, SBLE may cause students to engage in more problem behaviors if students interpret their presence as a sign that behavior problems are normative. Of course, more than one of these explanations may occur simultaneously, and there may be other explanations not listed here.

Additionally, although examining differences in the impacts of SBLE by student race was not a part of this meta-analysis, an emerging body of evidence continues to point to increases in exclusionary discipline associated with SBLE particularly for Black students (Crosse et al., 2022; Sorensen et al., 2023). Again, given that exclusionary discipline is largely considered a harmful practice for students (Gerlinger et al., 2021), perhaps particularly so for Black students (Fisher et al., 2022), more attention should be given to the racially disproportionate impact of SBLE on the exclusionary discipline of Black students. There is the potential that the increased exclusionary discipline leads to increased likelihood of criminal justice system involvement, which in turn leads to a suite of negative economic, health-related, and other consequences during adulthood (CITE).

In terms of perception outcomes, overall perceptions of school were not significantly different with or without SBLE. However, among studies that used schools as the unit of analysis, there were greater perceptions of safety among schools with SBLE (although this was based on a model with only seven effect sizes from two studies). None of the models synthesizing learning outcomes were statistically significant, providing no evidence that the presence of SBLE was related to learning outcomes.

Overall completeness and applicability of evidence

The evidence base examining the crime and behavior outcomes associated with SBLE is large and appears to be still growing. This is particularly true of studies that use schools as the unit of analysis. Although there are numerous studies, many of them lack methodological rigor (Petrosino et al., 2012) and there were notably no randomized control trials identified for inclusion in the meta-analysis. Some of the more recent studies used study designs that permit stronger causal inferences such as interrupted time-series, instrumental variables, fixed effects, and difference-in-differences; this is a notable improvement to the literature that provides more credible estimates than were available even 5 years ago. It is worth noting that several of these studies were based on data from the School Survey on Crime and Safety, and although the effect size estimates derived from these studies were not found to be different from those using other data sources, it would benefit this body of research to have more sources of data that have additional measures of crime and behavior and draw on a different sample of schools. Additionally, there were a limited number of studies that used students as the unit of analysis. Future research may benefit from examining student-level differences, particularly with an eye toward understanding the experiences of young people who are traditionally marginalized in both education and the criminal legal system, including but not limited to Black, Latino, disabled, and LGBTQ+ students.

Whereas the literature on crime and behavior outcomes was quite robust, the same was not the case for perception and learning outcomes. There were relatively few studies that examined these outcomes, and they were often of poorer methodological quality than those that examined crime and behavior outcomes. This lack of studies makes sense given that perception and learning outcomes are generally considered secondary to the presumed primary outcomes of crime and behavior. Still, if SBLE remain present in schools, it would be useful to better understand their impact on these secondary outcomes. Of particular relevance may be the impact on other aspects of perceived school climate besides feelings of safety. Ethnographic work has linked SBLE to feelings of isolation among students (Nolan, 2011), and also deteriorating student–teacher relationships (Devine, 1996). A few quantitative studies have examined related phenomena (e.g., Fisher et al., 2019; Theriot, 2016), but this evidence base would benefit from more attention to school climate outcomes.

One major limitation of this body of evidence is a general lack of detail about the programmatic elements of SBLE. Very few studies provided specific information about the goals and activities of SBLE programs, making it difficult to assess the extent to which certain elements of SBLE programs may contribute more or less to shaping a variety of outcomes. Although we intended to examine differences across SBLE programs of different types, not enough data about the programming was present in the literature to permit such analyses. Given this limitation in the primary studies, this meta-analysis can be understood as an analysis of the impacts of the mere presence of SBLE.

Still, it is unclear how and the extent to which different programmatic elements of SBLE programs would shape outcomes differently. Some evidence suggests that as SBLE take on more of a law enforcer role, they respond to student behavior problems with more formal sanctions as opposed to SBLE who identify more as educators and mentors (McKenna & White, 2018). That said, law enforcement activities are often among the most common activities of SBLE (Coon & Travis, 2012). Moreover, research has begun to question the extent to which law enforcement and non-law enforcement roles can really be separated; one qualitative study found that even as officers describe their non-law enforcement roles, they articulated them in terms of law enforcement goals and logics (Higgins et al., 2022). A set of studies included in this review examined different arrangements of roles of SBLE, assessing the extent to which officers’ engagement in various activities might differently shape outcomes related to crime, behavior, and responses to student behavior (Devlin & Fisher, 2021; Devlin & Gottfredson, 2018; Fisher & Devlin, 2020). Although these studies found somewhat different patterns of outcomes associated with different SBLE roles, it was not clear that one approach was obviously superior to the others in terms of both reducing problem behaviors and student punishment. So, although it is true that the evidence synthesized in this review lacked important detail about the programmatic elements of SBLE implementation and use, evidence from other research calls into question how important these differences might be.

In terms of generalizability, the vast majority of the studies were set in the United States, so the findings may not generalize beyond that country. Other countries have different cultures around both education and law enforcement, and comparative studies may help elucidate the extent to which these differences shape outcomes associated with SBLE. Additionally, these findings pertain to secondary schools, and the findings may not generalize to preschools, elementary schools, or institutions of higher education. Finally, the measures of the dependent variables included in the analysis (i.e., crime and behavior, perception, and learning outcomes) are based on prior researchers’ measurement and should not be taken to encompass other constructs that may be included within any of those three categories.

Quality of the evidence

One of the major limitations of this review is the quality of the evidence. Although studies in recent years have used increasingly sophisticated study designs and identification strategies that permit stronger causal inferences, many of the studies included here do not have sufficient methodological rigor to permit strong causal inferences. As such, the findings presented here should be interpreted as correlational rather than causal. However, it is noteworthy that in the moderator analyses that examined methodological differences, no significant differences were found, perhaps indicating that the less rigorous studies have come to similar conclusions as the more rigorous ones.

It is particularly noteworthy that there were no randomized control trials included in this review. This is likely because the use of SBLE is highly politicized (Nolan, 2015; Turner & Beneke, 2020) and emotional (Madfis, 2016), and may be difficult and impractical to gain support from practitioners to randomize schools or districts to receive SBLE. Researchers studying SBLE would benefit from continuing to use some of the more rigorous study designs used in the set of studies reviewed here, including time-series designs (Gottfredson et al., 2020), instrumental variable approaches (Owens, 2017; Weisburst, 2019), and difference-in-differences designs that incorporate multiple measures before and after SBLE implementation. Although there were very few studies that met these criteria in the current review, evidence about the impacts of SBLE based on rigorous study designs continues to accrue, with at least two studies becoming available since the we conducted final search for this review (Crosse et al., 2022; Sorensen et al., 2023). Although this meta-analysis provides evidence from studies with a range of methodological rigor—and finds general consistency when examining differences based on methodological characteristics—these more rigorous studies permit stronger causal inferences than those with weaker designs. Moreover, in a political climate that has recently seen more widespread critiques of SBLE, even calling for their removal, it will be important to use rigorous methods to investigate what happens to school and student outcomes when SBLE is removed.

Additionally, the finding about SBLE being associated with increased perceptions of safety should be interpreted with caution. Although the finding was statistically significant, it was based on a meta-analysis of only two studies that provided a total of seven effect sizes. Moreover, these two studies (McKay et al., 2006; Stokes et al., 1996) used fairly weak study designs that do not permit strong causal inferences. Future studies would benefit from applying more rigorous methods to examining the impact of SBLE on students’ feelings of safety at school.

Potential biases in the review process

As with most empirical research, this review included many researcher choices that may have introduced bias into the review process. For example, we chose to aggregate all effect sizes that fell under each of the three broad topics (i.e., crime and behavior, perception, and learning outcomes) and identify common groupings within each category. Other researchers may have made different decisions about what effect sizes should and should not be grouped together, and this may shape the findings and conclusions of the study.

We also opted to include a wide range of studies rather than imposing strict methodological criteria that would have led to only those studies with the strongest internal validity to be included. As such, some effect sizes are likely more trustworthy than others not in terms of the precision of the estimate, but because of the design of the study from which they were derived. Although we tested for moderating effects of multiple study design features, other researchers may have made different choices in this vein.

Additionally, as noted, because of personnel limitations associated with the multiple literature search and screening processes, not all screening and data extraction was completed by two coders. This could have introduced bias through human error that might have been detected if another coder was available.

Finally, we as a research team acknowledge our own cultural embeddedness that shaped a variety of choices we made during the process of conducting this review, ranging from the conception of the study to particular analytic choices to the perspectives we chose to highlight in this narrative. Other research teams with different enculturation and perspectives may have conducted this review differently, and we encourage replication by future researchers.

Agreements and disagreements with other studies or reviews

To our knowledge, there have been two prior meta-analyses related to SBLE. The findings from this study have some similarities and differences with these two meta-analyses. For instance, one prior meta-analysis of longitudinal research found that high schools with school resource officers—one type of SBLE—experienced increases in exclusionary discipline associated with the implementation of the officers (although, notably, the model using cross-sectional studies showed null results; Fisher & Hennessy, 2016). This aligns closely with the current study's finding that associates SBLE with increased exclusionary discipline.

Additionally, this study's findings align with another meta-analysis that found no significant relationship between the presence of a law enforcement officer or security guard at school and school violence and victimization (Turanovic et al., 2019). This is similar to the current study's findings that do not find a significant link between SBLE and measures of violence, or weapon-related outcomes. It is notable that the current study's finding linking SBLE to greater crime and behavior problems was driven largely by the increase in school discipline, a measure that was not included in the Turanovic et al. (2019) study.

Implications for practice

Most schools in the United States—and many schools in other countries where data are not systematically collected—use some form of SBLE, at least implicitly to make the school safer. However, this study's findings indicate that SBLE presence is linked with an increase in school crime and behavior problems. Moreover, when investigating more specific types of crime and behavior problems, this study found no impact on outcomes related to criminal justice system contact, violence, to substance use or possession, or weapon use or possession. If the intent of SBLE is to prevent crime and delinquency in schools, this study found no evidence that this intent is being accomplished. Instead, consistent with a more critical literature that focuses on the harms of SBLE, this study found that SBLE presence was associated with increases in school discipline. This is particularly noteworthy given the robust literature that has linked school discipline to a variety of adverse consequences, including increased delinquency and contact with the criminal justice system (Gerlinger et al., 2021).

Given these findings, practitioners are likely to benefit from reconsidering their use of SBLE. Momentum for removing SBLE has been growing in the United States, and this study's findings imply that removing SBLE is likely to reduce schools’ use of exclusionary discipline, and perhaps even decrease the amount of crime that happens in school, although it is difficult to disentangle the amount of crime that happens from the amount of crime that is detected. The benefits associated with this are unclear—although it may be valuable to detect more crimes, if those are highly subjective offenses like disorderly conduct that are recategorized and responded to as crimes (rather than school disciplinary events), it may have harmful consequences for students (Theriot, 2009). Still, given the financial costs of maintaining SBLE presence, the lack of clear benefits identified in this meta-analysis suggests that money is likely better spent elsewhere. Research has not examined in any depth the extent to which other school personnel may reduce school crime, but this may be a fruitful line of inquiry.

For schools that do choose to retain SBLE, it is unclear whether there are best practices that will enhance their effectiveness while limiting their negative consequences. Certainly, there are recommended strategies based on experience (e.g., see Petrosino et al., 2020). A few studies have investigated how implementing SBLE with different arrangements of roles is related to crime and behavior outcomes at school, but these studies’ findings are inconsistent and point to no real definitive conclusion about whether and how the roles of SBLE might make a difference (Devlin & Fisher, 2021; Devlin & Gottfredson, 2018; Fisher & Devlin, 2020). Although some scholars and practitioners have pointed to the importance of choosing the “right” officer, having memoranda of understanding in place, giving formal training, and not having officers be part of school discipline (Finn et al., 2005; Rosiak, 2016) these strategies have not been thoroughly evaluated, and studies show these efforts may be difficult to achieve and even counterproductive (Bolger et al., 2019; Curran et al., 2021). We advise those jurisdictions using SBLE to study their programs to ensure they are not harming the students and staff and are having the appropriate impact on the school community. Schools and communities that want to employ SBLE should look to innovate with current SBLE models and then test to see if the new type of strategy had desired impacts.

If schools choose to divest from SBLE they are likely to benefit from using those funds on promising evidence-based practices for reducing crime and violence in schools. One promising approach is to invest in school climate improvement. For example, increasing students’ sense of belonging at school and feelings of connection to peers and adults in the school is an effective strategy for preventing problematic behaviors (Gregory et al., 2012; Hirschi, 1969; Payne et al., 2003). Research in the last decade has increasingly focused on an authoritative school climate, which prioritizes both support and structure in the form of (a) strong interpersonal relationships between students and school personnel, and (b) clear, fair, and consistently enforced rules. This approach has shown benefits for a variety of outcomes related to crime and behavior (Berg & Cornell, 2016; Cornell et al., 2015; Crowley et al., 2019; Fisher et al., 2017; Gerlinger & Wo, 2016; Huang et al., 2021).

Implications for research

The findings from this meta-analysis have implications for research on SBLE. First, as noted, methodologically rigorous studies are needed to address the issue of selection bias that is common in this literature. Using the receipt of grant funding for SBLE as an instrumental variable has been a promising approach, as has estimating pre-trends using many time points. Further research in this vein will be a useful contribution to this literature.

Second, given the pessimistic findings from this study and the growing movement toward removing SBLE, research is needed on what happens when SBLE is removed. Studies of SBLE removal should focus on measurable outcomes such as school crime, discipline, and perceptions of safety, but also procedural and cultural outcomes that may be better addressed through qualitative research. How do school administrators compensate for their loss of a resource in the school? What new strategies or processes emerge for handling crime or misbehavior? How does the relationship between schools and police change as a result? Although this study's findings are suggestive that removing SBLE may have benefits for students and schools, this is only one step among many that schools might take, and it is critical to understand both the process of and outcomes associated with removing SBLE.

Third, research should focus more on issues of equity associated with SBLE implementation, use, and removal. There is both theoretical reasoning (Hirschfield, 2008; Kupchik & Monahan, 2006) and empirical evidence in the US setting (Crosse et al., 2022; Fisher et al., 2020; Homer & Fisher, 2020; Weisburst, 2019) to suggest that SBLE have disproportionate impacts on traditionally marginalized students—especially Black students. More evidence is needed to better understand the nature of this problem, and to expand the analysis to other traditionally marginalized groups of students including students with disabilities, LGBTQ+ students, undocumented students, and others.

Fourth, a greater focus on the role of context is likely to be important. There is evidence from the United States that the roles and perspectives of SBLE are contingent on their school's context (Fisher et al., 2020; Lynch et al., 2016). However, relatively little research in this body of literature has focused explicitly on the role that characteristics of school and community contexts might play in the impacts of SBLE. Further attention to this would likely benefit research, policy, and practice alike.

Internal support

We appreciate in-kind support from WestEd.

External support

This study was funded in part by the Campbell Collaboration Education Group (University of Arkansas).