Protocol for a Systematic Review: Policing Schools Strategies to Reduce Crime, Increase Perceptions of Safety, and Improve Learning Outcomes in Primary and Secondary Schools

The presence of gangs, drugs, bullying, weapons, and other criminal activity is associated with negative effects on school culture and student learning (e.g., Henry, 2000). Schools and school districts have long turned to police interventions to address these and other challenges to improve school safety (Shaw, 2004). In turn, police departments have also recognized the importance of developing strategies in collaboration with schools, such as special truant units or police-school liaisons, to combat school-based crime and disorder or to develop more positive relationships with youth (Shaw, 2004). Although school-based police interventions are controversial (see below), American educators and the public are increasingly open to their implementation (Cook, Gottfredson, & Na, 2009). Police presence on school grounds has grown dramatically following several high-profile shootings (Brown, 2006). In particular, the recent massacre at the Sandy Hook Elementary School in December, 2012, has renewed calls for deploying armed police in schools, a proposal offered by both President Obama and the National Rifle Association. Increased police presence and intervention at schools has also been reported in the United Kingdom and other European nations (Smith, 2003). Research suggests that juvenile violence has been on the decline in the United States since the mid-1990s (Gottesman & Schwarz, 2011), and schools remain safe places for students (relative to their time outside of the school setting). However, a considerable amount of crime and misbehavior has always occurred in educational settings—a result of placing large numbers of children together in one setting. For example, the U.S. Departments of Education and Justice report that during the 2009–2010 school year, 85% of public schools reported that at least one violent crime, theft, or other crime occurred at their school (Robers, Zhang, & Truman, 2012). Over half of administrators surveyed in the 2009–2010 School Survey on Crime and Safety reported at least one student threat to physically attack another person, with or without a weapon (Neiman, 2011). Moreover, in 2007, 8% of students in high school reported being threatened or injured with a weapon during the previous 12 months (Dinkes, Kemp, & Baum, 2009). The problem of school violence, crime, and disorder is not just an American issue. Estimates of school-related crime from other countries are more difficult to obtain, but the European Observatory on School Violence (EOSV) has been collating data from nations such as the U.K., Germany, France, and Spain (Debarbieux, 2003). For example, the EOSV has found that nearly 6% of U.K. teachers report being threatened or attacked at school, 20% of Spanish students report that they have committed some form of vandalism at school, between 15–46% of German students report carrying a weapon to school, and over 25% of French students report being involved in a physical assault on school property (Debarbieux, 2003). These international estimates dovetail with a recent World Health Organization (WHO) report that listed violence as the world's priority public health issue (Krug, Mercy, Dahlberg, & Ziwi, 2002). The WHO report has generated a renewed focus on youth violence and, more specifically, on school violence, in such organizations as the United Nations Economic, Social and Cultural Organization (UNESCO, 2007). The European Observatory's offspring, the International Observatory on Violence in the School Environment, was inspired in part by growing interest in this issue by nations around the globe. Given the high levels of crime and disorder present at schools in the U.S. and in other nations, it is not a surprise that police and educators would collaborate in order to combat it. Shaw's (2004) international review of police-school practices categorized such strategies into three groups: (1) school-based police officer programs; (2) police as 'teachers' (i.e., curriculum approaches such as the police officer teaching the students about drug, gang, or sexual abuse prevention); and (3) comprehensive or broad-based liaison programs in which police and other social service agencies are involved with the schools. Shaw reported that non-curricular policing programs (e.g., police patrol strategies, problem-oriented or community policing tactics, assigning school resource officers or police-school liaison offices, or other measures) have been implemented in a wide range of countries, including the U.S., U.K., Canada, Germany, Denmark, Australia, the Netherlands, Belgium, Estonia, Poland, and South Africa, at least through 2003. Shaw's review was globally comprehensive but not systematic; she concluded her synthesis (2004, p. 26) with a future research agenda that includes "a more comprehensive, comparative review of police-school interventions" and "the systematic collection and documentation of good practice models and tools." To our knowledge, there has not previously been a systematic review of evaluative evidence specific to the "policing schools" area, with the exception of police-led curriculum efforts such as Drug Abuse Resistance Education (D.A.R.E) (Ennett, Tobler, Ringwalt, & Flewelling, 1994; West & O'Neal, 2004). In other words, while police departments maintain a significant presence in many schools, there is not yet a research-based consensus on whether School Resource Officer or special police-school liaison programs, or other policing schools strategies make schools safer and reduce criminal or delinquent behavior and non-criminal student misbehavior (such as bullying and harassment). There is also no agreement on whether the presence of police enhances perceptions of safety by students or staff. Critics of school-based policing argue that police in schools represent an unnecessary expense and may have unintended negative consequences for students, teachers, and staff (e.g., Petteruti, 2011). For example, police presence in the schools could lead to behavior normally handled informally by schools to be criminalized by arrest and formal charges (e.g., Petteruti, 2011). The National Association of School Psychologists (2013) recently issued a brief that summarized a number of potential harms from increasing police and security at schools, including sending a signal to students that their campus is unsafe, increasing fear, and undermining school climate. In 2012, Petrosino, Guckenburg and Fronius reported on a scoping study using systematic review techniques to assess the state of the evaluation evidence in this area. They located 11 quasi-experiments available that had a comparison group. Only one would have met the minimum evidentiary standards of the What Works Clearinghouse in education, and most of the studies would have been scored a "3" by the Maryland Scale of Scientific Methods: a common approach to rating evaluations in justice. However, they did not attempt any meta-analysis, and their review covered the literature through 2009. This proposed systematic search is designed to build upon Shaw's (2004) work and recommendations, and the earlier Petrosino, Guckenburg and Fronius (2012) scoping study, to provide a systematic review of research on the effects of policing schools strategies. A general theory of change for policing schools strategies is provided in the Figure in Appendix A. As the Figure denotes, there are a number of anticipated interventions that we expect fall under the "policing schools" 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. Programs that solely involve the police officer teaching a specific prevention curriculum such as D.A.R.E., G.R.E.A.T., or sexual abuse prevention are not eligible, though interventions that combine curricular and non-curricular strategies would be included. Policing programs that involve providing information on school crime or data to schools would be included, if the police are not teaching a curriculum or lesson plan. These programs and strategies are expected to increase police presence activities and interaction with students and staff at the schools. Police presence serves as a deterrent to crime and misbehavior, and police establish relationships that allow them to gather information to prevent or intervene with students who may be planning a crime. In turn, this should reduce criminal behavior and non-criminal student misconduct; increase perceptions of safety by students, staff, and parents; and lead to more positive relationships between police with young persons and with the schools. This should also result in increased attendance, test scores, and grades, and improved perceptions of school climate. However, there is the possibility of unintended negative consequences that police presence leads to heavy surveillance, resulting in police coming across more misbehavior by youth, leading to overuse of arrest for behavior normally handled informally by schools and to more push-outs of students (disproportionately affecting racial and ethnic minorities, and youth with disabilities). Police presence could also lead to increased fear of crime and misbehavior by sending a signal to the school community that the building is unsafe. The types of learning outcomes we anticipate are grades, attendance/absenteeism, dropout, test scores, and perceptions of school climate. The types of non-educational outcomes we anticipate are measures of safety such as police reports, calls for service, arrests, school disciplinary reports, self-reports of student behavior, self-reports by staff of safety and fear, self-reports by students of safety and fear, and measures of parent fear for their youth at school. The objectives are to systematically review the research to answer three research questions relevant to potential positive effects of police-school programs, and two questions relevant to a potential unintended negative consequence. For this project, we will only include those studies that have the following characteristics: A quasi-experimental evaluation that would meet the study inclusion criteria was reported in the United Kingdom (Bhabra, Hill, & Ghate, 2004). The study evaluated the location of one police officer in each of 11 schools located within ten designated "crime hot spots" in the United Kingdom. The results were compared to two schools that were also in high crime areas that did not have a police officer assigned. Esbensen, Osgood, Taylor, Peterson, & Freng (2001) reported on a quasi-experimental evaluation of a police-taught curriculum designed to prevent adolescent gang involvement, Gang Resistance Education Awareness Training (G.R.E.A.T.). Because this study involves police teaching a curriculum in the schools, it would not be included in this review. We will rely on five major strategies to identify eligible evaluations published through 2013. These are: The databases listed in Appendix B can be somewhat idiosyncratic. Our approach is to conduct pilot searches of terms, working iteratively until the yield of citations and abstracts is as relevant as possible. In other words, we want to maximize sensitivity (getting as many citations and abstracts as possible) and specificity (making sure that as many as possible are relevant to the project). Our planned search strategy combines three types of keywords. The first set of keywords (and their derivatives) targets outcome studies: e.g., "random," "experiment," "control," "evaluate," "trial," "impact," "effect," and "outcome." A second set of keywords focuses the search on schools and education institutions: e.g., "student," "school," "district," "classroom," "academy," "," "campus," "teacher," "principal," "faculty," "bully," "truancy," and "superintendent." The third set of keywords focuses the search on policing, including keywords such as "police," "patrol," "sheriff," "constable," "enforcement," "officer," and "security." When the database does not permit extensive lists of keywords to be combined, simpler searches involving words such as "police" and "schools" will be used. For example, a proposed search in Criminal Justice Abstracts would be: Query: AB=(Columbine or school* or student* or campus* or teacher* or truan* or vandal* or classroom*) and AB=(experiment* or evaluat* or assess* or impact* or outcome* or effect* or randomly or randomize* or "comparison group" or "control group" or controls or comparisons or "control condition" or "comparison condition" or "time series") and AB=(police* or "law enforcement" or patrol* or policing or "security guard" or "security measure*" or "crime prevention" or sheriff* or "cop" or "cops" or constabl* or detective* or undercover or "school crime" or "school violence" or "school safety" or "school security" or "resource officer*) or "problem-oriented"). Search methods will usually result in a large number of citations and abstracts. The review author conducting the search will save the results from the searches described earlier and mark those for which full-text should be retrieved. This will be examined by at least one other co-author to ensure that there is agreement on the selections. Since all three authors are experienced in conducting reviews of this type, no training is necessary. Many of these will be easily excluded as not being relevant to the proposed review. In some cases, however, they will identify potentially eligible studies. The full text documents of those potentially eligible studies will be retrieved and screened by the co-authors before the study can be formally included. We expect to retrieve the full text of most articles electronically. We also have access to interlibrary loan through Bridgewater State College. When a full text report is received, two of the review co-authors will read it to ensure that it meets the eligibility criteria. If there is no agreement on the inclusion of a particular study, it will be excluded and the exclusion documented in the final report. We will also establish a bibliographic reference database to maintain a log of all included and excluded studies. The log includes a field that allows the research team to document the reason for exclusion. We have designed a preliminary instrument to guide us in recording information from each study (see Appendix C). Although the instrument contains several open-ended items, these will be collapsed when appropriate into a smaller number of categories to permit further analysis. For example, items such as "how equating was performed" can be collapsed into three or four larger categories representing the most frequent responses (e.g., discontinuity, covariate matching, propensity score, post-hoc statistical matching) and an "other" category that captures all those responses that do not fit into the most common methods of equating in this set of studies. The instrument has items in the following areas: Study reports can be used to provide information about the publication and characteristics about the study and its context. For example, we will extract data about the type of publication and the setting in which the study was conducted. If the documents provide more detail on the setting and political/social context in which the study takes place, we will also include it. We will extract information about the randomization, quasi-experimental assignment, and other methodological aspects of the evaluation. The level of assignment and whether the study included multiple analyses at different levels will also be coded. It is especially critical that information about three key issues in the implementation be extracted from each study report: These items will solicit detailed descriptions of the intervention and control conditions, including the "dosage" of the treatment being implemented, and the number of participants assigned to each group. The types of interventions that we expect include School Resource Officer Programs, police-school liaison programs, problem-oriented policing strategies, and police patrol/surveillance at the school to deter crime within or around the school, or to ensure safe routes for children as they walk to school. We anticipate that the evaluations in this review sample will be comprised of a single intervention and a single control group. When this is not the case, we will select the most policy-relevant groups to compute our experimental versus control condition contrast. In most cases, it will be the groups that experience the greatest contrast between conditions, i.e., the most intensive intervention condition versus the least intensive control condition. We recognize the importance of documenting these decisions for full transparency. These items solicit detail about the type of participants in the trials, including information on the country where the study took place, the nationality of the participants, the age and school level targeted, gender, whether an urban or rural setting was involved, and the socioeconomic status of the students. For each eligible study (each study will have, at minimum, one outcome measure meeting the eligibility criteria described earlier), we will extract information on reported outcomes of crime (e.g., specific delinquent or criminal acts, arrests, calls for service, referrals to police, disciplinaries, etc.) and other behavior (bullying, aggression, etc. that does not rise to "criminal" act but represents misbehavior), perceptions of safety/fear, and learning outcomes. We will also code any other outputs or data on key "mechanisms" that would provide clues as to why the intervention did or did not have its that may several articles on the Our of is the evaluation and not the research and it is to extract information from all documents to the instrument for one When reports on the study we will a number of strategies, including the for study will be by a single outcome to prevent the from being by from one Although some evaluations may report just a single outcome at one time it is possible that other evaluation reports will include analyses at various time and may various that school safety and learning outcomes. decisions have to be about outcome will represent the for that For this we will the outcomes we will learning and potential unintended negative consequences We do not as of yet such outcomes will be i.e., will they be measures of groups that or or measures for some outcome of such as the number of reports If results are and include and we will the to report these and etc.) and make such decisions in our review. We expect to a range of measures used to assess each of these four types of and we to these measures into For example, within learning and may be common Although we expect that many reports will only include an at a single in we to report two different analyses to the studies that report outcome data at various time We will report at first first time and the for the If one time (e.g., is reported in the study, it will be used in both If estimates are reported for the experimental versus control we will rely on them for any synthesis since they reduce statistical that may have from or the of implemented or the of studies report analyses at multiple levels (e.g., i.e., for schools or and for students. Our is to this information at the If or assignment is at the level (i.e., school or we will to into this was a we used in a for the (see Petrosino, & 2012). studies in which it is not possible to for at the level will be To ensure that we good we will have two of the read and information from all reports if the yield of studies is and from a randomly of the studies if the yield is more We will assess (i.e., by using the of agreement for each before (see below), a This will measures with study characteristics that agreement of with those that do with of agreement will be to the for The two and will meet with the lead co-author to and will their for they and if there is consensus the three the will be We will those items from our database in which could not be as as items that will be into the We will to combine results from the We will report which is to but it includes a sample for for each analysis. is a and many are available to it from information reported in evaluation articles (e.g., statistical test The will be when are available in to do will be used to the results from the The will for each study, the and The will also the for the across studies. that all analyses will be reported a effects and the will be by the When results in the we will report the the and whether the that the result is that a positive that the strategy had a on the outcome (e.g., a in crime or an in a negative a for intervention there is no in of the outcome measure between treatment and control Because of the in and we will effects models in our which to be more the effects For our we will conduct for to if the is a good of the sample of studies being used in the analysis. We anticipate the will be the in intervention type, sample and the in these studies. we will in each each outcome at each of the two time first and through the which is reported as a of the of across studies in the We will also report and as on However, analyses have to be as they are on numbers of studies and such analyses can be significant by if large numbers of are on We anticipate a number of as a of by the a effects for the following potential We are open to including other if appropriate to conduct (i.e., there are studies to such as at the between the characteristics of the schools and outcomes. As we will evaluations and Our is to them in a and in particular, to them to the and challenges to policing schools It is of our focus on and and our focus on outcomes that can be into an that we will However, we will code the presence or of the studies and the data from the and quasi-experiments will be used to three particular areas: (1) the context for the (2) the theory or by which the is to the (3) the and of the intervention and comparison and potential issues that may be to and in to the outcome or We will report on any data included in the studies that are included in the review. Our experience in conducting reviews is that data are included, but it to This includes information on the of the any of the of the intervention (e.g., the and studies (e.g., the and of the It is that this information be in some to the outcome studies that it can be retrieved during the for What was the document was What was the type of many documents were in this What state or country did the study What was the setting for the What other information was provided on the context for the can include the for the more about the to us more about the any criteria for the units of in the study this be in the What were the relationships of the authors to the implementation of the assignment used to level was was the specifically any (e.g., did with If assignment was not quasi-experimental was used to (e.g., comparison post-hoc statistical of propensity etc.) any with assignment If were did with assignment did comparison level was assignment any in of problem from assigned sample was with by of groups in the for intervention and control contrast if multiple groups: excluded study groups with brief the intervention with particular to the "dosage" of the there any implementation described by detail theory for why it should include any post-hoc information from study on why the or did not the control or comparison condition "dosage" if of school of participants that were of participants that were data on all data on treatment and including sample used in the statistical whether or (and if controls were statistical and detail all effects any information on data in the detail all information review Education the of The authors a systematic search of the literature for a scoping study in for the U.K. National Police for The results were published in the of 2012). of those however, any about the of policing school they on the of the evaluation The review will be in The review will be within three it is this for and the review in with The will provide as as possible to with the of the review. A review be to the relevant within two of If are not before the or if we are to for an the relevant has the to the or the to The also has the to or the if it does not meet the standards of the the for the review in of and and other and the review at least five if for the review to as with the The of the in review is upon agreement to the and in the The places no on publication of the of a systematic review in a more form as a before or the publication of the in however, have that publication of that have or will reported and authors publication in such a should be of possible with publication of the in in a publication or in status in should the and include a to it. that systematic reviews published in and with the may have or for authors for meeting any the to a and to in the on of the

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.