We use a randomized experiment to test whether and what information changes teenagers' sexual behavior in Kenya. Providing information on the relative risk of HIV infection by partner's age led to a 2
(a)OR = Adjusted Odds Ratio (matched by age and admission time).Final conditional logistic regression model: HIV risk factors among MSM, Project Horizonte, 1994–2010.
*exposure prevalence in cases equals the number of HIV-positive individuals with exposure to the risk factor over the total number of HIV-positive individuals; † exposure prevalence in controls equals