With cross-classification used to probe for possible male-female and child-adolescent variation, recently published evidence on the mental and behavioral disturbances of young people seems to confirm dimensional models of anxiety and depression (Wadsworth et al., 2001) as well as attention problems (Hudziak et al., 1999). In this new R01 project, the main goal is to investigate youth drug use as it relates to latent structure of mental and behavioral disturbances across age strata in adolescence (12-13,14-15,16-17 years), with due consideration for different drug types (inhalants; club drugs like ecstasy; marijuana; tobacco; alcohol) and levels of drug involvement. Once the underlying latent class structures have been elucidated, and class memberships have been assigned based on the best-supported class structures, the applicant will turn to regression models to probe research questions pertinent to (a) the co-occurrence of adolescent drug involvement and disturbances of the mental life, behavior, and social adaptation, (b) probing for variation in the latent structure across subgroups defined by race-ethnicity to perhaps aid in understanding health disparities that affect minority subgroups, and (c) the clustering of the class members within local areas of residence within the United States, with a specific focus upon analyses that will probe how drug use might sustain delinquent peer groups. The data are from 13,831 youths (age 12-17) recruited after nationally representative sampling for the National Household Survey on Drug Abuse (NHSDA) in 1994-1996. For the NHSDA, the youths completed standardized modules on drug involvement, as well as a survey adaptation of Achenbach's Youth Self Report (YSR). The YSR is perhaps, the most. widely used instrument for measuring youth disturbances in the mental life, behavior, and socxal adaptation. In sum, the plan of research is one that exploits an untapped research opportunity to learn more about adolescent drug taking as it relates to disturbances of the mental life, behavior, and social adaptation. The plan involves the use of two relatively recent biostatistical developments in the form of latent class regression and alternating logistic regression models, which are innovative approaches to the study of research questions that previously have not been asked in large-sample epidemiological studies of national scope.