A comparison of the mediational properties of four adolescent smoking expectancy measures

This prospective study compared the ability of 4 smoking expectancy measures to mediate the influence of peer, parent, and current smoking on adolescents ’ cigarette use 3 months later. No evidence for mediation was found when expectancies were operationalized as unidimensional subjective expected utility (SEU), multidimensional SEU, or unidimensional SEU decomposed into probability and desir-ability main effects and their interaction. Evidence for partial mediation was found for the decomposed multidimensional SEU measure. The results suggest that (a) peer and current cigarette use may influence future smoking indirectly through adolescents ’ probability estimates that smoking will control negative emotions and (b) the relationship between current and future smoking also may be mediated by adolescents ’ beliefs about the desirability of weight control. An important assumption underlying many models of adoles-cent problem behavior is that expectancies are not only an impor-tant proximal cause of smoking, drinking, and drug use, but they also mediate more distal effects, such as temperament, community factors, family dynamics, and peer influence (Petraitis, Flay, & Miller, 1995). Although the expectancy mediation hypothesis ap-pears to be widely accepted in the literature, direct empirical tests have been rare. In the present study we investigated whether smoking expectancies mediate the effects of three factors previ-ously shown to predict adolescent smoking intentions: (a) peer smoking, (b) parent smoking, and (c) respondents ’ current smok