Parent and peer influences reconsidered: the convoy of social support model of adolescent substance use

Longitudinal data were used to test hypotheses concerning the changes in parental and peer influence over a 6-year period in a panel of rural adolescents (mean age of 14 years at first assessment). Drawing on the assumptions of the social convoy model of social support (Antonucci, 1985; Kahn & Antonucci, 1980), a multivariate latent growth curve model tested the relative contributions of parental and peer influences on adolescent substance use. Evidence of a common pattern of alcohol and cigarette use was found, which was distinguished from the use of marijuana. In order to make use of all available information, a model-based approach was used to justify the imputation of missing data, using the expectation-maximization (EM) algorithm. Univariate growth curve models indicated that although adolescents reported an increase in parental support, the level of friend support did not change between the ages of 14 and 16. Similarly, separate univariate growth models showed that the reported level of parental influence on adolescents\u27 substance use decisions decreased across the six-year time, while the level of friend influence remained the same. The results of the multivariate model provided evidence that the relations between the growth parameters of social support and adolescent substance use were mediated by parents\u27 and peers\u27 influence on adolescents\u27 substance use-related decisions. Comparison of paths in the multivariate model suggested that the relative influence of parents on adolescents\u27 substance use was significantly higher than that of friends. While parents provided a protective buffer against early and escalating use, friends served to increase the risk of initial substance use. Thus, parental influence was found to be stronger than peer influence not only at early ages, but throughout the adolescent period