Emphasizing Interactions to Advance Violence Research.

Research on interpersonal violence has identified numerous biological, developmental, psychological, social, and situational risk factors, yet further progress requires greater attention to how these influences interact. This article argues that violence research should move beyond the study of isolated predictors toward multilevel, developmental models that explain how combinations of risk factors produce heterogeneous pathways to aggression. Two forms of interaction are emphasized: cross-domain interactions among influences operating at different levels and within-domain interactions among multiple risk factors that may yield similar behavioral outcomes through different causal pathways. Moffitt's developmental taxonomy of antisocial behavior and the Confluence Model of Sexual Aggression illustrate the value of this approach. Research on depression further demonstrates how multilevel models can facilitate the identification of meaningful subgroups and support more mechanism-focused interventions. An important next step for violence research is to integrate developmental risk models with intervention science by identifying individuals' dominant risk mechanisms and matching prevention strategies to those mechanisms. Such an approach may improve the field's ability to identify who is at greatest risk, understand the processes driving that risk, and determine which interventions are most likely to reduce perpetration.