Bridging Disciplinary Boundaries (January 11 - 14, 2007)

Purpose: This theory-based evaluation was conducted to examine the effects of the Making Choices (MC) program, a universal, preventive intervention designed to reduce aggression in elementary school-age children. The study tested whether social information-processing (SIP) mediators accounted for program effects on posttest overt aggression, which were identified in a recently published study. In addition, it explored moderating effects of gender. Examining mediating and moderating effects in the context of intervention can help us to achieve a more sophisticated understanding of “what works” and “for whom” – improving our capacity to prevent problem behavior in youth.

Methods: The study utilized a non-randomized cohort design with treatment withdrawal. The study sample consists of three cohorts of third graders from two rural elementary schools. Third graders in 2001-02 (n=156) received the MC program delivered by intervention specialists. Third graders in 2002-03 (n=193) received the MC program plus supplemental activities requiring greater teacher and parent involvement. Third graders in 2004-05 (n=131) received a routine health curriculum. Students came from low-to-middle SES backgrounds and were ethnically diverse.

Intervention students were hypothesized to have greater pretest-to-posttest decreases in overt aggression and hostile attribution than comparison students and greater increases in SIP skills. In addition, posttest SIP skills (i.e., encoding, hostile attribution, goal formulation, response selection, and emotion regulation) were hypothesized to partially mediate program effects on overt aggression at posttest. Informed by preliminary research evidence suggesting gender differences in SIP skills and in the association between social cognition and aggression, the study hypothesized that effects would vary by gender.

Hierarchical linear modeling (HLM) was employed to estimate program effects on six posttest scores – overt CBCL aggression, encoding, hostile attribution, goal formulation, response selection, and emotion regulation. Posttest scores from each measure were regressed on the following variables: pretest scores, race/ethnicity, and gender (Level 1-Student Level) and MC and MC+ (Level 2-Classroom Level). Multilevel mediational analyses were conducted to assess whether the targeted SIP skills explained program effects on overt aggression and whether gender moderated these effects.

Results: Both interventions positively affected study outcomes. Program effects on posttest overt aggression were larger than expected (ESMC = -.43 and ESMC+= -.51). Medium effect sizes (ESs) were also found for SIP skills. MC decreased hostile attribution (p<.001) and increased response selection at posttest (p<.05). Effects of MC+ resulted in increases in response selection (p<.001) and goal formulation (p<.001) and trends for emotion regulation and hostile attribution. Marginally significant indirect effects were found for emotion regulation and goal formulation, indicating that SIP skills may have mediated program effects on overt aggression. Moderating effects were limited to a cross-level program-by-gender interaction effect on overt aggression.

Implications: The findings of this study are promising and consistent with empirical findings in child development. Descriptive analyses of main effects and explanatory analyses of theory-based mediators can help to shed light on causal processes underlying treatment effects. Conducting these analyses requires that intervention researchers adequately address methodological challenges (e.g., obtaining valid and reliable measures of social cognition, employing rigorous tests of mediation) posed by these analyses.