Abstract
Behavioral theory informs HIV prevention programs by identifying important predictors of the relevant behavior. Changes in those predictors brought about by program messages should theoretically translate into behavior change. Program evaluation, however, may examine only whether message exposure is associated with behavior change, assuming that if people changed their behavior, the intervention must have effectively changed the determinants. This practice obscures how intervention programs work in the field. We use data from the AIDS Community Demonstration Projects (ACDP) to illustrate this important idea. Earlier research found that the ACDP moved at-risk communities to more consistent condom use with both main and non-main partners. This study demonstrates that these behavioral effects are explained by the ACDP's ability to improve intention and self-efficacy for condom use with both main and non-main partners. We conclude that to understand how HIV prevention programs work, program evaluations need to assess how the intervention affected what it was designed to change: one or more critical behavioral determinants.
Acknowledgements
We thank Wayne Johnson for his valuable data-analytic advice.