Date of Award


Degree Type


Degree Name

Doctor of Philosophy (PhD)



First Advisor

Robert Hornik


It is difficult to determine whether influence or selection drive the association between an adolescent's behavior and his or her friend’s behavior. To understand what role influence plays on adolescent risk activities, this research analyzed the longitudinal network sample of the AddHealth dataset to examine whether any random friend, a best friend, or a peer group shapes an adolescent’s risk behavior. The project conducted cross-behavior analyses of five activities — cigarette smoking, alcohol consumption, tobacco chewing, marijuana use, and sexual debut — among a sample of 1,969 adolescents aged 12-17 years at time one. The data contained real behavior measures for the adolescents and their nominated friends. The analyses used logistic regressions to predict the respondents’ time two behaviors and to determine whether demographic variables, self-esteem, or parental factors modified peer influence. The results from this project contain four important findings. First, there is a main effect for peer influence and it is equivalent across risk behaviors. On average adolescents were twice as likely to engage in a risk behavior if their friend participated in the activity at time one. Second, peer influence may be both harmful and protective. For cigarette and marijuana use, there was only influence to initiate a risk behavior. In comparison, for alcohol consumption there was equal influence to conform to friends who drank and friends who did not drink. In contrast, for chewing tobacco use there was significantly more influence to stop chewing than to begin. This suggests that friends offer teens protection from risk activities. Third, the group analysis found that a linear measure of peer influence, which accounts for each group member’ s behavior, provides significantly more detail about the peer influence process when compared to a dichotomous measure of group influence, which does not detail how many peers engage in a risk activity. Finally, results found that best friends are not more influential than other close peers, suggesting that adolescents have multiple friends who exert equal levels of influence. In summary, this project found that peer influence is a real phenomenon that takes on varying roles across adolescent risk behaviors.