For a representative sample of adolescents, this article estimates the odds of various HIV risk behaviors for two groups: The General Risk Group reports behaviors indirectly related to HIV infections, and the HIV Risk Group engages in a variety of activities directly related to HIV infection. Multinomial logistic regression predicting risk group membership on the basis of demographic variables indicates heterogeneity in the combinations of risky behaviors relevant to HIV infection and the characteristics of adolescents engaging in these behaviors. The results highlight the weaknesses of behavioral risk factor studies and call into question the meaningsfulness of risk groups in public health assessment of social conditions. Finally, the article identifies the analytical process of decontextualization that operates to reduce the theoretical understanding of adolescent risk taking while also diminishing the practical ability to model and control the social causes of HIV infection.
Abstract