The data analyzed were derived from the first two waves of the Buffalo Longitudinal Survey of Young Men (BLSYM). Data collection for the BLSYM began in 1992, and the study was designed to examine multiple causes of adolescent substance use and delinquency. A sample of 625 males ages 16 to 19 was obtained from the Buffalo area. Although the sample contained a higher proportion of young men of lower socioeconomic status than in the city as a whole, the entire socioeconomic spectrum was represented. The variables measured were drinking, drug use, delinquency, and psychopathy. A measurement model with a single first-order factor of problem behaviors was compared to a measurement model with three separate first-order factors of drinking, drug use, and delinquency for Wave 1 and 2 data. The significant improvement in fit for each wave indicated that the measurement model with three separate first-order factors was more statistically supported than the model with a single first-order factor. These finding support the hypothesis that drinking, drug use, and delinquency are different forms of problem behaviors and are represented by separate latent constructs. In addition, the data revealed high correlations among these latent constructs, indicating that these different forms of problem behavior shared much common variance. The model that contained the second-order common factors for each wave indicated that the variance of each form of problem behavior (drinking, drug use, and delinquency) was largely explained by the second-order common factor. These findings suggest that it is logical to conduct a second-order factor analysis to assess general deviance theories that posit that a single common factor can account for different forms of problem behaviors. In estimating a substantive model with psychopathic state as the underlying common factor, the study found that this construct was a powerful common factor that significantly and simultaneously predicted different forms of adolescent problem behavior. The data indicated that the problem behaviors of the same type were likely to remain stable across time, which accounted for the reduction of the initial power of psychopathic state to predict different forms of problem behavior. Analyses of the substantive model also revealed moderately high stability of psychopathic state across time. This finding provides some support for the argument that deviant personality traits do not change much across time, at least not in the stage of life covered by this study. 5 tables and 36 references