Session: Innovative Methods for Assessing the Etiology of Sexual Behaviors Among Adolescents and Young Adults (Society for Prevention Research 21st Annual Meeting)

Although sexual behavior among adolescents and young adults is normative, risky sexual behavior has been associated with an increased risk of negative health outcomes.  Understanding the heterogeneity of sexual behaviors, how those behaviors vary over time, and which early factors may determine engagement in risky behaviors is of continued public health significance.  Innovative methods have the potential to shed light on each of these aspects of sexual behavior and ultimately inform the design of effective interventions that promote sexual health while reduce negative health outcomes.  Each talk in this symposium applies a cutting-edge methodological approach to advance understanding of the etiology, determinants, and consequences of sexual behaviors during adolescence and young adulthood.  

The first talk introduces latent class analysis to describe complex patterns of sexual behavior identified in a national sample of adolescents.  The classes are then used in a novel model-based approach to assess whether or not sexual behavior class membership predicts mental, social, and physical health outcomes in young adulthood.  The second talk applies a new time-varying effect model to capture dynamic processes involved in sexual behavior over time in a sample of young adult women.  In particular, these models assess the evolving effect of predictors of sexual behaviors, including attitudes about sex and peer norms, throughout sexual partnerships.  The final talk employs modern causal inference techniques to determine whether depression in adolescence is causally related to lifetime number of sexual partners and having an STI by adulthood.  Inverse propensity score weighting is used to adjust for many confounding variables, providing a stronger causal inference of the effect.

Collectively, the talks present three innovative analytic approaches that answer important research questions that can be difficult to address using traditional approaches.  In particular, the talks in this symposium provide demonstrations of how latent class analysis with a distal outcome, time-varying effect models, and causal inference methods can provide new insight into the complexities of sexual behaviors, both within individuals and over time.  The discussant, a well-recognized expert in the prevention of adolescent sexual risk behaviors and HIV, will reflect upon the talks and moderate a discussion of how these new methods can guide the development of building more effective interventions.