Associating with substance using peers is generally considered as one of the most important predictors of adolescent substance use. However, peer association does not affect all adolescents in the same way. To better understand when and under what conditions peer association is most linked with adolescent substance use (SU), this review focuses on the factors that may operate as moderators of this association. The review highlighted several potential moderators reflecting adolescents' individual characteristics (e.g., pubertal status, genes and personality), peer and parental factors (e.g., nature of relationships and parental monitoring), and contextual factors (e.g., peer, school and neighborhood context). As peer association is a broad concept, important methodological aspects were also addressed in order to illustrate how they can potentially bias interpretation. Taking these into account, we suggest that, while the effects of some moderators are clear (e.g., parental monitoring and sensation seeking), others are less straightforward (e.g., neighborhood) and need to be further examined. This review also provides recommendations for addressing different methodological concerns in the study of moderators, including: the use of longitudinal and experimental studies and the use of mediated moderation. These will be key for developing theory and effective prevention.

Results support previous literature suggesting that peer victimization predates alcohol use and extends it by showing clear directionality between suicidal ideation and alcohol use over 1 year, supporting the self-medication model. Clarifying the empirical basis of these underlying models could allow for earlier prevention strategies, by targeting the risk factor that appears the earliest in the model.

Penile plethysmography (PPG) is a measure of sexual interests that relies heavily on the stimuli it uses to generate valid results. Ethical considerations surrounding the use of real images in PPG have further limited the content admissible for these stimuli. To palliate this limitation, the current study aimed to combine audio and visual stimuli by incorporating computer-generated characters to create new stimuli capable of accurately classifying sex offenders with child victims, while also increasing the number of valid profiles. Three modalities (audio, visual, and audiovisual) were compared using two groups (15 sex offenders with child victims and 15 non-offenders). Both the new visual and audiovisual stimuli resulted in a 13% increase in the number of valid profiles at 2.5 mm, when compared to the standard audio stimuli. Furthermore, the new audiovisual stimuli generated a 34% increase in penile responses. All three modalities were able to discriminate between the two groups by their responses to the adult and child stimuli. Lastly, sexual interest indices for all three modalities could accurately classify participants in their appropriate groups, as demonstrated by ROC curve analysis (i.e., audio AUC = .81, 95% CI [.60, 1.00]; visual AUC = .84, 95% CI [.66, 1.00], and audiovisual AUC = .83, 95% CI [.63, 1.00]). Results suggest that computer-generated characters allow accurate discrimination of sex offenders with child victims and can be added to already validated stimuli to increase the number of valid profiles. The implications of audiovisual stimuli using computer-generated characters and their possible use in PPG evaluations are also discussed.

No abstract