Deviant Peers as a Mediator of Pubertal Timing-Substance Use Associations: The Moderating Role of Parental Knowledge

Abstract

Purpose: Early perceived pubertal timing and faster maturation have been linked to increased risk of adolescent substance use (SU), particularly for girls, but the mechanisms underlying this association are not well understood. We sought to replicate and extend findings from Westling et al. (2008) showing that peer deviance mediates the link between early puberty and SU with stronger pathways in the context of low parental knowledge of adolescents' whereabouts and activities.

Methods: Participants (n = 1,023; 52% female, 24% nonwhite, and 12% Hispanic) were recruited through middle schools. Pubertal timing and tempo were derived from repeated measures of perceived pubertal development. Specific sources of parental knowledge included child disclosure and parental solicitation. Two measures of peer deviance (problem behaviors and SU) were obtained. The use of any substances (alcohol, cigarettes, marijuana, and other illicit drugs) was coded from all assessments.

Results: For girls, earlier pubertal timing was associated with higher likelihood of SU but only in girls who disclosed less. For boys, slower tempo predicted greater SU, equally across parental knowledge groups. Pubertal timing and tempo were generally not associated with peer deviance; however, we detected a significant indirect effect such that peer problem behavior mediated the association between girls' early pubertal timing and SU. Parental knowledge did not moderate this effect.

Conclusions: Peer deviance was not strongly supported as a mechanism underlying atypical pubertal risk for SU (supported in one of the eight models). Parental knowledge appears to serve as a contextual amplifier of pubertal risk, independent of peer influences.

Keywords: Parental knowledge; Peer; Pubertal tempo; Pubertal timing; Puberty; Substance use.

Figure 1. Most Parsimonious Models for Boys

Unstandardized parameter estimates from the best-fitting, most parsimonious model, wherein all paths that could be constrained to equality across high and low knowledge groups were constrained, are presented (hence, the parameter estimates may differ somewhat from the results for the full model presented in Tables). Only significant paths are depicted. Results from all four models for boys are presented: models including child disclosure as a moderator are presented in the top panel (1a, 1b), and models including parental solicitation as a moderator are presented in the bottom panel (1c, 1d). Models including peer behavior problems as the measure of peer deviance are presented on the left (1a, 1c) and models including peer substance use as the measure of peer deviance are presented on the right (1b, 1d). When paths show significant moderation, the low disclosure group is on the left of the / and the high disclosure group is on the right. * p < .05. Paths with a single estimate were constrained across groups without a decrement in model fit. Timing and tempo were assessed from repeated measures across all waves (the mid-point of development occurring prior to W6). Peer deviance and parental knowledge measures were assessed at W6, and substance use was assessed at W7.

Figure 2. Most Parsimonious Models for Girls

Unstandardized parameter estimates from the best-fitting, most parsimonious model, wherein all paths that could be constrained to equality across high and low knowledge groups were constrained, are presented (hence, the parameter estimates may differ somewhat from the results for the full model presented in Tables). Only significant paths are depicted. Results from all four models for boys are presented: models including child disclosure as a moderator are presented in the top panel (2a, 2b), and models including parental solicitation as a moderator are presented in the bottom panel (2c, 2d). Models including peer behavior problems as the measure of peer deviance are presented on the left (2a, 2c) and models including peer substance use as the measure of peer deviance are presented on the right (2b, 2d). When paths show significant moderation, the low disclosure group is on the left of the / and the high disclosure group is on the right. * p < .05. Paths with a single estimate were constrained across groups without a decrement in model fit. Timing and tempo were assessed from repeated measures across all waves (the mid-point of development occurring prior to W6). Peer deviance and parental knowledge measures were assessed at W6, and substance use was assessed at W7.