The primary goals of the meta-analysis was to inform school violence research and evidence-based policies by: (1) organizing the existing empirical knowledge base regarding school violence; and (2) identifying the key individual-, school-, and community-level factors that should be targeted for change and/or intervention to reduce school violence, aggression, and delinquency. Specifically, this research was guided by the following three objectives:
- Determine which individual-level factors are most salient in predicting whether aggression/delinquency and victimization will occur among students at school.
- Determine which school-level factors influence the likelihood that a student will either perpetrate aggression/delinquency or will be victimized at school.
- Determine which community-level factors are most strongly associated with youth aggression/delinquency or victimization at school.
Based on 693 studies, 31 different predictors of school violence at the individual, institutional, and community levels were determined.
The mean effect size estimates across all predictor domains were analyzed for different forms of victimization at school, and predictors were ranked by mean effect size to determine the strongest and most consistent predictors of school victimization, aggression, and violence.
The sample of studies was assembled in three phases: an extensive literature search through electronic databases for empirical studies published up to January of 2016, a search through electronic holdings of 107 different journals for any remaining studies not picked up during prior searches, and a final search of prior literature reviews, reports, and smaller meta-analyses. All studies
printed in English that presented a codeable statistical relationship between any of the identified individual-, school-, or community-level predictors and any form of violence, aggression, or delinquency at school were eligible for inclusion in the meta-analysis.
693 studies comprised the final sample, which contributed a total of 8,551 effect size estimates. These effect sizes were drawn from 545 independent data sets and 68 different nations. The majority of effect size estimates (56.22%) were based on U.S. samples.
Empirical research on school violence published up to January 2016.
Effect size of predictors of school violence
The data file Meta-Analysis-Data-for-NACJD.dta (n=8,551; 9 variables) contains variables related to results of statistical tests such as effect size estimates, and victimization and aggression/delinquency as dependent variables, and predictor.