The study is the first to examine the effects of children’s and therapists’ in-session behaviors on later therapeutic alliance (TA; i.e., relational bond, task collaboration) as rated by children and therapists in an intervention for children with aggressive behavior. One hundred eighty children (ages 9.3–11.8; 69% male; 78% Black), screened as having aggressive behavior by teacher and parent ratings, received a 32-session group-based cognitive-behavioral intervention (Coping Power) at their schools. TA ratings were collected from children and therapists at mid- and end-of-intervention using the Therapeutic Alliance Scale for Children. Children’s and therapists’ behaviors during the first 16 sessions were coded by independent observers. Children’s negative in-session behaviors predicted lower child- and therapist-rated TA (averaged across mid- and end-of-intervention). Children’s in-session positive behaviors, at both the individual and group-wide level, predicted higher later TA. Therapists’ efforts to manage deviant behavior predicted stronger child-reported ratings of the relational bond and of child- and therapist-rated task collaboration. Exploratory analyses indicate that the effect of children’s in-session behaviors on later TA is moderated by therapists’ skills in managing the group and in managing deviant talk and behavior in sessions. Clinical and research implications of the findings are discussed.

This study originated in collaboration with Thomas Dishion because of concerns that a group format for aggressive children might dampen the effects of cognitive-behavioral intervention. Three hundred sixty aggressive preadolescent children were screened through teacher and parent ratings. Schools were randomized to receive either an individual or a group format of the child component of the same evidence-based program. The results indicate that there is variability in how group-based cognitive-behavioral intervention can affect aggressive children through a long 4-year follow-up after the end of the intervention. Aggressive children who have higher skin conductance reactivity (potentially an indicator of poorer emotion regulation) and who have a variant of the oxytocin receptor gene that may be associated with being hyperinvolved in social bonding have better outcomes in their teacher-rated externalizing behavior outcomes over time if they were seen individually rather than in groups. Analyses also indicated that higher levels of the group leaders’ clinical skills predicted reduced externalizing behavior problems. Implications for group versus individual format of cognitive-behavioral interventions for aggressive children, and for intensive training for group therapists, informed by these results, are discussed.