New evidence on the relationship between substance use and adolescent sexual behavior.

Policymakers and the public have a clear interest in encouraging teens to delay becoming sexually active or, if they are sexually active, to use birth control. Many researchers have argued that reducing substance use among teens will accomplish both of these goals, yet work in this area has failed to control for unobservables that are potentially correlated with substance use and sexual behavior. Using a variety of estimation techniques and data from National Longitudinal Study of Adolescent Health, we estimate the effects of marijuana and alcohol use on two outcomes: the probability of being sexually active, and the probability of having sex without contraception. Our results highlight the importance of controlling for unobservables and indicate that the link between substance use and sexual behavior is much weaker than previously suggested.