Esther Mertens

102 | Chapter 5 Table 1 Means and Standard Deviations of Students’ Perceived Classroom Peer Context, Modeling, Reinforcement, and Dyadic Mutuality per Condition and Group Comparison at Post-test R&W Control Post-test differences (ANCOVA) Baseline M ( SD ) Post-test M ( SD ) Baseline M ( SD ) Post-test M ( SD ) F p η 2 partial Outcomes Comfort 4.35 (.80) 4.19 (1.04) 4.57 (.56) 4.36 (.76) .23 .633 .002 Cohesion 4.27 (.72) 4.05 (.95) 4.36 (.65) 4.23 (.69) .04 .850 .000 Conflict 2.58 (1.15) 2.43 (1.24) 2.31 (1.00) 2.39 (1.02) .44 .508 .004 Victimization 1.29 (.67) 1.32 (.86) 1.18 (.60) 1.14 (.43) 1.32 .253 .011 Mediators Deviant modeling .27 (.17) .34 (.21) .25 (.14) .33 (.15) .11 .742 .001 Prosocial modeling .17 (.11) .15 (.11) .16 (.09) .14 (.08) .67 .417 .005 Deviant reinforcement .30 (.28) .38 (.33) .29 (.21) .43 (.27) .91 .343 .007 Prosocial reinforcement .46 (.34) .39 (.34) .45 (.28) .33 (.28) 1.79 .183 .014 Moderator Dyadic mutuality 4.27 (.57) 4.51 (.53) Mediation analyses Both changes in modeling and in reinforcement did not mediate the effect of the intervention on students’ perceived classroom peer context (see Table 2). More specifically, the intervention did not predict changes in modeling or reinforcement (Path a). Changes in modeling or reinforcement also did not predict the outcomes, except concerning victimization (Path b). An increase in prosocial modeling was related to a decrease in experienced victimization. Moderation analyses Dyadic mutuality moderated the relation between changes in prosocial modeling and experienced victimization (see Table 3). The negative relation between changes in prosocial modeling and experienced victimization was stronger for dyads with higher levels of mutuality ( B Low  = -0.79; B Average  = -1.70; B High  = -2.60; see Figure 1). No other moderations by dyadic mutuality were found on either modeling or reinforcement.