Abstract
School-based aggression prevention programs may not be equally effective for all students and classes, depending on student and class characteristics. This study investigated moderators of a cluster randomized controlled socio-ecological aggression prevention program's effectiveness (change from pretest to posttest, sample: 2,042 preadolescents, mean age = 11.7 years, SD = 0.09, 47.6% girls) and sustainability (change from posttest to follow-up test, sample: 659 preadolescents, mean age = 12.7 years, SD = 0.08, 47.9% girls). The program worked better in multicultural classes, as greater ethnic diversity strengthened the program's effectiveness and sustainability. Moderating effects of a positive social class climate and higher baseline levels of aggressive behavior and victimization were also found. These results advance socio-ecological theorizing and can help develop more contextualized interventions.