Esther Mertens

150 | Chapter 7 than their interactions and social perceptions (i.e., the interpersonal domain). Thus, an intervention’s effectiveness may depend on students’ vulnerability, number of people involved in the intervention, and the addressed domain. These results accentuate that there is no simple answer to the question “for whom is the intervention effective?”. The findings of the present study show the advantage of studying trajectories of change and highlight the complexity of evaluating “what works under what circumstances for whom”. Modeling trajectories of change rather than mean differences at one moment in time creates the opportunity to analyze intervention effects during an intervention (Greenberg, & Abenavoli, 2017). Not only might these trajectories reveal critical points of change, they might also provide insights for opportunities to optimize interventions. Therefore, I stress the relevance of assessing participants during an intervention, as well as before and after an intervention. Such interim assessments can be conducted with a relatively low frequency, as in the present dissertation (2 measurement points), but also with a high frequency, for instance after each session. Furthermore, to add to the complexity of differential effectiveness, intervention effects were not only dependent on context, but also on the combination of context and characteristics of participants. In the Light condition (i.e., narrow ecological focus) students benefitted from R&W regardless of their level of Extraversion. In contrast, in the Standard and Plus conditions (i.e., a broader ecological focus) only Extraverted students were able to benefit from the intervention. Notwithstanding this complexity, it is essential to examine under what circumstances and for whom an intervention is effective in order to determine the generalizability of intervention effects to different circumstances and populations. Aim 3: Working Mechanisms in Interventions The first way in which I studied working mechanisms in interventions was through examining mechanisms of change (i.e., mediators). Classmates’ modeling and reinforcement did not mediate the effect of R&Won the peer context in the classroom. Note that this conclusion has to be interpreted with caution as the mediators were only examined in a subsample and only in the first year of the intervention. Although the results could be due to research practices (e.g., conceptualization, coding), the finding that R&W was unrelated to modeling and reinforcement could also be an indication of action theory failure as the intervention is seemingly not able to successfully address the proposed mechanisms of change (O’Rourke & MacKinnon, 2018). In addition, modeling and reinforcement were generally unrelated to the peer context in the classroom, suggesting potential conceptual theory failure as well since the mechanisms of change are apparently not related to the outcomes (O’Rourke & MacKinnon, 2018). The exception is prosocial modeling which was related to less victimization in the class, especially when dyadic mutuality between classmates was high. Hence, prosocial modeling in the classroom could be a mechanism of change