Esther Mertens

| 19 General Introduction dissertation, I investigated working mechanisms through examining mechanisms of change as well as intervention components. By analyzing mechanisms of change (i.e., mediators) I examine changes within the participants (Longabaugh & Magill, 2011), whereas by analyzing components of interventions I examine aspects of interventions related to intervention effects (O’Rourke & MacKinnon, 2018). First, I studied mechanisms of change (i.e., mediators). Mechanisms of change indicate behaviors or processes within the participants that have changed as a result of an intervention and can be related to subsequent changes in the ultimate outcome (Longabaugh & Magill, 2011). Mediation analyses provide insights of practical as well as theoretical importance and can be used to inform directions for future research. For instance, the intervention might have failed in successfully addressing the mediator (i.e., action theory failure), the mediator might not be related to the outcome (i.e., conceptual theory failure), or both (O’Rourke & MacKinnon, 2018). I examined whether the effect of R&W on competencies and problems in the interpersonal domain was mediated by changes in classmates’ deviant and prosocial modeling and reinforcement. As stated in the social learning theory (Bandura, 1977), peers influence each other through the mechanisms of modeling (i.e., learning new behaviors and tendencies by observing peers) and reinforcement (i.e., learning new behaviors and tendencies based on positive peer feedback). R&W explicitly addresses these two mechanisms in exercises and reflections aiming to subsequently improve the peer context in the classroom, represented by interpersonal relations in the class and victimization by bullies (aspects of the interpersonal domain). When classmates’ modeling and reinforcement of deviant behaviors decrease and those of prosocial behaviors increase, the overall peer context in the classroommight become more positive (e.g., Dishion & Tipsord, 2011; Telzer, Van Hoorn, Rogers, & Do, 2018). In addition, I examined whether classmates’ influences on the peer context was dependent on classmates’ dyadic mutuality (i.e., the level of responsiveness, reciprocity, and shared understanding). Dyadic mutuality can represent the relationship quality in various relations (Piehler & Dishion, 2007), including relations between classmates. Taking relationship quality into account is eminent when examining peer influences as research suggests that students are more strongly influenced by peers with whom they have high quality relationships (Barry & Wentzel, 2006; Berndt, 2002; Piehler & Dishion, 2007). Previous research studying peers’ influences often focused on deviant or prosocial behavior within a specific type of relationship (e.g., Dishion, Spracklen, Andrews, & Patterson, 1996) and mostly relied on questionnaires (e.g., Hofmann & Müller, 2018). I examined both deviant and prosocial influences enabling a comparison between the two types of influences. Furthermore, modeling and reinforcement were assessed through video-observations of randomly selected dyads of classmates in a subsample in the “Standard” (i.e., the entire teaching staff is involved in the intervention) and Control (i.e., Care As Usual; CAU) conditions in the first year of the intervention. The randomly selected dyads represented the broad range of types of relations that exists between classmates, minimizing selection effects and enabling studying classmates’ 1