Multidimensional Predictors of Sexual Behaviors and Program Effectiveness: A person-centered approach

• In Event: 3-194 - Multilevel Risk Factors and Sexual Behavior in Diverse Adolescents

Integrative Statement

Sexual behaviors are influenced by risk and protective factors at multiple levels (DiClemente et al., 2005). Most research on correlates of adolescent sexual behavior has used a variable-centered approach, which examines individual risk factors in isolation. However, a person-centered approach can be used to identify how particular patterns of risk and protective factors are associated with sexual behaviors. In this study, we applied latent class analysis (LCA) to data from a teen pregnancy prevention program to identify subgroups of youth characterized by multidimensional risk and protective factors and examine how these are associated with a) initiation of vaginal and oral sex and b) program effectiveness.
Methods: LCA was applied to data from the baseline (7th grade) sample of 1,693 sexually inexperienced students (48.3% Hispanic/Latino, 34.4% black) participating in a randomized controlled trial of It’s Your Game…Keep It Real. We used LCA indicators that had been identified as predictors of sexual behavior in prior studies, which included family structure, having a parent in jail, having a mother who was a teen parent, living in multiple homes, household language other than English, and religious service attendance. Multilevel analysis was then used to examine rates of sexual initiation (vaginal, oral) and program effectiveness in reducing initiation by latent class.
Results: LCA identified three classes of adolescents: Family Disruption; Other Language Household; and Frequent Religious Attendance. In the control condition, individuals in the Family Disruption class were most likely to engage in both vaginal (26%) and oral (11%) sex, compared to the Other Language Household (18% and 7%) and Frequent religious Attendance (4% and 2%) classes. Within the Family Disruption class, being in the treatment condition was association with 30% lesser odds of initiating vaginal sex. There was no effect for the other classes, or for initiation of oral sex.
Conclusions: Findings suggest the utility of a person-centered approach to examine predictors of sexual behavior and to evaluate the effects of prevention programs. Individuals who experienced multiple types of family disruption were more likely to initiate oral and vaginal sex; however, the program was most effective for these adolescents, and these multidimensional profiles better predicted subgroups for whom the program was effective than individual demographic factors typically studied (Peskin, 2017). This approach can help us better understand the multidimensionality of influences on sexual behavior, as well as how programs work and how they can be more effective.

Authors

• Sara A Vasilenko, Syracuse University
  Presenting Author

• Jill R. Glassman, ETR Associates
  Non-Presenting Author

• Kari C. Kugler, Pennsylvania State University
  Non-Presenting Author

• Melissa F. Peskin, University of Texas Health Science Center
  Non-Presenting Author

• Ross Shegog, University of Texas Health Science Center
  Non-Presenting Author

• Christine Markham, University of Texas Health Science Center
  Non-Presenting Author

• Susan Tortolero Emery, University of Texas Health Science Center
  Non-Presenting Author

• Karin Coyle, ETR Associates
  Non-Presenting Author