Predicting Later Adolescent Risk Behaviors: Which Proximal Outcomes Matter?

Prevention programs are often delivered before unhealthy risk behaviors emerge, yet their effectiveness is frequently judged by impacts on those distal behaviors. This creates a need for stronger evidence on proximal outcomes that predict later adolescent risk behaviors. We analyzed longitudinal survey data from approximately 1,100 youth in grades 9-12 who had received a relationship education program in 27 schools across seven U.S. states. The sample was 60% White, 23% Hispanic, and 49% girls. Outcome-specific regression models estimated associations between 30 posttest proximal outcomes and nine later sexual behavior and substance use outcomes over six months. Because analyses were exploratory, findings were interpreted with attention to effect-size magnitude, directional consistency, and theoretical coherence. Twenty-two proximal outcomes had statistically significant associations with one or more later behavioral outcomes. Sexual behavior associations were generally coherent and moderate in magnitude, whereas substance use associations were more often smaller, isolated, or theoretically inconsistent. Risky social media behavior emerged as the clearest cross-domain risk factor. Self-efficacy to use relationship skills and self-awareness about personal strengths and weaknesses showed the clearest protective patterns for sexual behavior outcomes. Intentions to have sex also showed directionally consistent negative associations across domains, although this pattern was contrary to prior expectations and was interpreted cautiously. These findings strengthen the evidence base on longitudinal predictors of adolescent risk behaviors and identify proximal outcomes that may be useful when direct assessment of later behavioral outcomes is infeasible or uninformative.