We are grateful for the kind comments of Underhill et al and Dr Tevendale regarding our article1 reporting significant effects of a theory-based abstinence-only intervention in delaying sexual initiation among African American adolescents. We randomly allocated participants to 5 interventions; therefore, pairwise comparisons involved only 40% of participants. This provided adequate statistical power to detect differences between each of the 4 human immunodeficiency virus (HIV)/sexually transmitted disease (STD) risk reduction interventions and the control group, but much less power for pairwise comparisons among the 4 HIV/STD interventions because the anticipated effect size was much smaller. Our article benefited from the rigorous review process that helps improve the quality of published trials and used an improved analytic approach compared with our presentation.2 We regarded the 4 comparisons between the HIV/STD interventions and control as primary and additional pairwise comparisons as post hoc to reduce type 1 errors that increase with multiple comparisons.
Jemmott JB, Jemmott LS, Stevens R. Integrating Past Research on Related Abstinence and Safer-Sex Interventions—Reply. Arch Pediatr Adolesc Med. 2010;164(7):679-681. doi:10.1001/archpediatrics.2010.121