A Meta-Analysis of School-Based Violence Prevention Program Randomized Treatment-Control Studies

Background and Purpose: Providing violence prevention programming in schools can produce positive outcomes, however, there is very little research that has been replicated to establish a strong evidence base for specific school violence prevention interventions. As school violence prevention continues to be an area where programming dollars are spent, the question of what works stands out. By utilizing a meta-analysis of randomized treatment-control school-based violence prevention programs this study aims to draw the evidence base together and identify interventions/programs that have positive outcomes for students, schools and communities. Methods: For this meta-analysis, an exhaustive search of the literature was conducted utilizing 10 databases which looked solely at published articles in peer-reviewed journals. The search terms focused on school based violence prevention, intervention and physical aggression/violence outcomes. The initial search of the literature was conducted in 2007 and the search was recently repeated and updated using PsychInfo to include research published through 2010. The inclusion criteria for a study were established by 3 researchers. Two researchers read each abstract returned from the search. If the abstract met at least one of the inclusion criteria the article was retrieved and scored based on the inclusion criteria. Scoring disputes were reviewed and resolved with the third researcher. Results: This exhaustive literature search provided 1609 unduplicated articles and 102 studies were judged to meet criteria for inclusion. In these 102 studies, 1080 measures were reported of which 243 specifically measured violence or aggression according to the researcher established inclusion criteria. Of note, 67 studies utilized a screening procedure to select for study inclusion and 60 studies focused on elementary school children. Analyzable data (Ns, means and SD's at pre and post) were reported by 54 studies which represented 67 samples (n=54). Sample sizes ranged from 16 to 6,090 (median: 85, mean: 332) and totaled 24,270. Two effect size (ES) types were computed Post only and Pre-Post (pre-post correlation set at .40). A sample ES was also computed by averaging over the violence/aggression measures. Overall, the mean effect size for the Post only evaluation was -.041 with a confidence interval of -.067 to -.015; the Q statistic was 347.16 with df=66 and p<.001. For the Pre-Post, mean weighted effect size was -.067 with a confidence interval of -.096 to -.038; the Q statistic was 187.97 with df=66 and p<.001.

Conclusions and Implications: On average, school violence prevention interventions had a beneficial effect on outcomes, although these results varied widely. The findings of this meta-analysis indicate that randomized-control trials need to be replicated to show stability of effects across populations. To further the value of this research standardized measurement and definition of terms need to be developed. Basic summary statistics need to be reported in all published research studies. Nearly half of the studies that qualified for this study could not be included in the final analysis due to the inability to obtain these statistics from either published reports or directly from the researchers. To date, published school violence prevention research appears as a burgeoning field.