The results for boys are presented in Table 3. When family environment variables were added (Model 2), there was a reduction in the OR for age (approximately 20%), indicating that the growth in alcohol use over adolescence was partially mediated by lagged changes in the family environment. The results were different from the pattern for girls, with the mediating effect of age primarily attributable to decreases in parental disapproval of alcohol use (OR: 0.59, p < .001) rather than mother closeness. Family conflict and emotional closeness (to either parent) were nonsignificant for boys in these models. These findings are unaltered by the addition of peer alcohol use (Model 3). As for girls, the odds of alcohol use increased nonlinearly with the number of alcohol-using peers. There was no meaningful change in the effect for parent disapproval of alcohol use when peer alcohol use was added, indicating that these variables were independent of each other. When controls were entered (Model 4) there was no meaningful changes in the ORs for family/peer variables, though sensation seeking was significant. SES was unrelated to alcohol use for boys.
