Academic and Behavioral Consequences of Visible Security Measures in Schools
Co-Principal Investigator: Mark Lipsey
Purpose: In an attempt to create safe environments for students, many schools have turned to visible security measures such as security cameras, metal detectors, and law enforcement officers. This project will explore the ways that these security measures are used in schools; how they are related to middle- and high-school students' perceived school safety; academic and behavioral outcomes; and in what contexts those associations are strongest. Despite increased use of such costly security measures, there is a lack of systematic research on whether and when these security measures are associated with student outcomes. Theories of deterrence and criminalization suggest there may be substantial variation in the effects of visible security measures on student outcomes, possibly even negative effects in certain environments. Understanding how visible security measures are associated with student outcomes, and whether certain combinations of security measures are more influential than others in certain school contexts, is important for identifying malleable factors that enhance or inhibit learning outcomes. Thus, this exploratory study will help identify those school security approaches worth further refinement and empirical investigation.Project Activities: The proposed project will use two national datasets, the School Crime Supplement to the National Crime Victimization Survey and the School Survey on Crime & Safety, to explore the patterns of visible security measures utilized in schools and their association with middle- and high-school students' outcomes. Researchers will first explore the different patterns of visible security measure utilization in U.S. middle- and high-schools. After identifying different security use patterns, researchers will then examine how schools' use of such security measures are associated with student outcomes—namely, perceived school safety, attendance, achievement, educational expectations, antisocial behavior, weapon use, and drug use at school. By identifying patterns of visible security measure utilization, researchers will examine whether various combinations of visible security measures (e.g., security personnel and metal detectors vs. security personnel only) are differentially associated with student outcomes. Finally, researchers will explore the conditions under which school security measures may have the largest effects, examining whether school context characteristics (e.g., urbanicity, grade span) moderate the relationship between security measures and student outcomes.
Products: Products include preliminary evidence of potentially promising school security measures and peer reviewed publications.
Setting: The project will analyze cross-sectional data from two large national surveys—the School Crime Supplement (SCS) to the National Crime Victimization Survey and the School Survey on Crime & Safety (SSOCS).
Sample: The SCS is a cross-sectional survey of 12–18 year old students in the United States. The analytic sample will include over 40,000 student-reported responses collected at five time-points (1999, 2001, 2003, 2005, 2007). The SSOCS is a cross-sectional survey of principals and administrators of schools in the United States. The analytic sample will include over 6,000 principal- and administrator-reported responses collected at four time points (1999/00, 2003/04, 2005/06, 2007/08).
Intervention: This exploratory study will help identify those school security approaches worth further refinement and empirical investigation thereby suggesting future use of school security measures that may be most beneficial to students' academic and behavioral outcomes.
Research Design and Methods: The analyses will occur in two stages. The first stage of analysis will involve an in-depth exploration of patterns of visible security measure utilization in schools including the identification of different patterns of security measure use that may typify most U.S. middle- and high- schools. Latent class analysis will be used to identify and describe the most common patterns of visible security measure utilization in schools. Because schools will not be randomly assigned to groups with different security use patterns, propensity scores will be used to create equivalent groups given other observable characteristics. The second stage of analysis will involve use of generalized linear models to explore the ways that security use patterns are associated with student outcomes, and the moderating effects of school context characteristics on those relationships.
Control Condition: Due to the nature of this research design, there is no control condition.
Key Measures: The School Crime Supplement includes student-reported items on perceptions of school safety, academic achievement (attendance, course grades, and academic expectations [do you expect to attend school after high school?]), behavior (aggression/violence victimization, property crime, weapon presence, and drug presence), and use of visible security measures in the school. The School Survey on Crime & Safety includes administrator-reported items on student academic achievement (attendance, percent of students below 15th percentile on standardized tests, and percent of students likely to go to college), student behavior (aggression/violence, property crime, weapon presence, and drug presence), and use of visible security measures in the school.
Data Analytic Strategy: Data from the two samples will be analyzed separately, but results will be compared and presented in parallel to determine the consistency and generalizability of findings. Latent class analysis (LCA) models will be used to identify different groups of schools that use similar patterns of visible security measures. The next phase of the analysis will compare student outcomes for the LCA groups using generalized propensity score methods to equate the multiple nonrandomized groups. Generalized linear models will be used to analyze the associations between security use patterns and student outcomes, with regression models estimated separately for each of the key outcomes of interest. Separate regression models will also be run to assess the potential moderating effects of school characteristics (e.g., urbanicity) on those relationships.