Drug education is conceptualized as a socialization process, particularly as it is related to assuming the adult role. The effectiveness of two processes of social influence, expert power and referent (modeling) power, were measured in three different age groups-early adolescence, middle adolescence, and adulthood-in a drug education program. Learning of factual information and attitude change served as dependent variables. Results indicated strong age--specific effects. Expertise was most effective with adults, modeling most effective with early adolescents. Relationships between effectiveness and age level were monotonic. Implications are drawn for the study of adolescence, the study of attitude change, and the design of drug education programs. (Author)