摘要：Mentoring is one of the most commonly‐used interventions to prevent, divert, and remediate youth engaged in, or thought to be at risk for delinquent behavior, school failure, aggression, or other antisocial behavior. We conducted a meta‐analytic review of selective and indicated mentoring interventions that have been evaluated for their effects on delinquency outcomes for youth (e.g., arrest or conviction as a delinquent, self‐reported involvement) and key associated outcomes (aggression, drug use, academic functioning). Of 112 identified studies reported published between 1970 and 2005, 39 met criteria for inclusion. Mean effects sizes were significant and positive for each outcome category. Effects were largest (still moderate by Cohen's differentiation) for delinquency and aggression. However, these categories also showed the most heterogeneity across studies. The obtained patterns of effects suggest mentoring may be valuable for those at‐risk or already involved in delinquency and for associated outcomes. Moderator analyses found stronger effects in randomised controlled trials compared to quasi‐experimental studies, for studies where emotional support was a key process involved in mentoring, and where professional development was a motivation for mentors. However, the collected set of studies are less informative than expected with quite limited detail in studies about what comprised mentoring activity and key implementation characteristics. This limitation encourages caution particularly in interpreting the moderated effects. These findings add to the longstanding calls for more careful design and testing of mentoring efforts to provide the needed specificity to guide effective practice of this popular approach.
