The research also examined the extent of the overlap in these behaviors according to age, gender, race, and family structure. The study used log-linear statistics to examine multiple contingency tables for all possible main effects and interactions while selecting the most parsimonious model that minimized the differences between observed cell counts and expected values while assuming no relationships (null hypothesis). Results revealed statistically significant relationships between heavy alcohol use, the use of marijuana, and sexual activity, but most teenagers did not engage in all four behaviors. In addition, all demographic factors entered into interactions with these behaviors. Findings indicated that the construct of the deviance syndrome has qualified utility and needs refinement with inclusion of demographic characteristics. Findings also indicated the need for further research that includes characteristics such as parental income and education, property offenses, and crimes against persons. Tables and 46 references (Author abstract modified)