Esther Mertens

| 105 Classmate Influence in Intervention Discussion The present study extended previous research on peer influence by examining classmates’ influences through observations within a randomized controlled trial of a universal intervention study, examining influence in not self-selected dyads, and focusing on both deviant and prosocial modeling and reinforcement. Contrary to our expectations, changes in classmates’ modeling and reinforcement were not mechanisms of change in the universal school-based intervention R&W, that aims to improve positive peer relations. Overall, regardless of condition, we did find that increases in classmates’ prosocial modeling was related to less victimization in the class, especially when the dyadic mutuality between the classmates was high. The finding that classmates’ modeling and reinforcement were not mechanisms of change in R&W might indicate that modeling and reinforcement were not intensely enough addressed during the intervention. More intensive attention to setting negative consequences for deviant behaviors and reinforcing prosocial behaviors in the classroom has been related to more positive behavior of students (Phillips Smith, Dumas, & Prinz, 2006). During the R&W lessons the trainer sets negative consequences for deviant behavior, reinforces prosocial behavior, and serves as a role model, as described in the interventionmanual (Ykema, 2002). However, the interventionmanual lacks guidelines for applying these techniques outside the intervention lessons. Hence, this classroom management approach may not have been implemented during regular lessons or implemented differently among teachers. This decreases the intensity and consistency with which modeling and reinforcement are addressed, limiting the opportunities for students to learn fromprosocial models and refrain from deviant models (Phillips Smith et al., 2006). Another explanationmight be that the time frame of R&W is too short. Sijtsema and Lindenberg (2018) indicated in their review that studies showing a relation between peer influence and antisocial behavior used a time span from six months to one year, whereas the R&W intervention takes approximately four months. A longer time frame might be necessary to establish change in classmates’ modeling and reinforcement and to have an influence on the classroom peer context as group dynamics take time to change. The finding that an increase in prosocial modeling is related to a decrease in victimization is particular eminent for anti-bullying interventions and in line with the social learning theory (Bandura, 1977). Classmates showing prosocial modeling are more inclined to show affiliation and involve all classmates in classroom activities degrading the level of exclusion and rejection (Juvonen & Ho, 2008). Other students might imitate this prosocial behavior which results in less victimization. In contrast to Ando and colleagues (2005), we did not find a relation between deviant peer influences and victimization. Given that Ando and colleagues (2005) studied peer influences of friends and the present study influences of classmates, the difference in findings might 5