Methods: This study is a secondary data analysis of data from a 5 year CDC funded randomized control trial to test the effectiveness of Green Dot, a bystander-focused sexual and teen dating violence prevention program, in 26 high schools in a Southeastern state. Two matched schools from each region were randomized to the intervention (13) or control (13) condition. All students from each school were surveyed each Spring. 89,707 students completed paper pencil surveys which tracked victimization and perpetration of sexual violence, psychological and physical dating violence, stalking, and sexual harassment. In schools that received the intervention, fidelity was tracked using audio recordings of all bystander trainings delivered in implementation schools during the study period. Each recording was assessed for fidelity by two trained fidelity raters. Two separate analyses using linear mixed models highlight the differences in rates of school level violence of “as designed” (Coker et. al, 2015) controlling for implementation differences and “as implemented” (Coker et al., 2017). When implemented “as designed” school experienced a 50% reduction in school-level mean rates of violence compared to “as implemented” where a schools experienced a 17-21% reduction in mean rates of violence dependent on amount of training conducted. This presentation will examine differences in effectiveness in key variables associated with delivering the Green Dot program “as implemented”.
Results: While the Green Dot bystander prevention program has been shown to be effective in reducing both victimization and perpetration of multiple forms of interpersonal violence, rates of effectiveness for reducing sexual violence, dating violence, sexual harassment and vary by three variables associated with implementation fidelity. These include when student are recruited using the theoretically derived recruitment strategy, when curriculum is delivered as designed (with fidelity to program elements), and when a sufficient number of students were trained as specified by the program model. Findings will be presented by type of violence and type of implementation.
Conclusion and Implications: Bystander-focused violence prevention programs are primarily curriculum-based requiring intensive training of students. There is little research to address differences in program effectiveness when fidelity or implementation is compromised. By comparing program effectiveness data “as implemented” versus “as designed”, we find that the importance of delivering bystander-focused prevention programs without compromising key program elements and with fidelity to the curriculum is critical. Implications for increasing program effectiveness research that includes fidelity assessment measures, training social workers and violence prevention educators on the importance of delivery of programs as designed, working with schools to allow prevention programming to be delivered as designed, and training organizations on strategies to increase fidelity of curriculum-based programs will be discussed.