This dissertation explores the program evaluation results from a domestic violence prevention initiative designed to reach African American adults with a dramatic radio campaign. The impact evaluation found associations between program exposure and outcomes, consistent with a claim of impact, however, low exposure levels and evidence of selectivity led evaluators to reject the hypothesis of impact. This paper addresses the question of explaining an association between exposure and outcomes if it is not due to program effects. Two prominent alternate explanations are explored: selectivity and response bias. Through two paired analyses I approach the data set in two different ways. In the first pair I seek evidence of variables that explain two exposure measures - program recall and false exposure claims. The first analysis corroborates the finding of selectivity, as beliefs and behaviors related to domestic violence prevention explain recall, as do racial identification and media use. Domestic violence-related measures are also positively related to false exposure claims. In addition, respondents who are male, listen to the radio more, and spoke with Black interviewers, are more likely to falsely claim exposure. vi In the next paired analysis I reexamine the selectivity hypothesis, testing whether other factors underlie the association between domestic violence measures and the two exposure measures. No other factors account for the association between domestic violence measures and recall, and I conclude that listeners respond differently to a persuasive message depending on their prior beliefs, intentions and behaviors. Most of the association (79%) between domestic violence measures and false exposure remains, while racial identification and media use each account for a small portion (11% and 5% respectively) of the associations. I conclude that issue involvement leads to central processing of the message and consequent recall. The program may thus serve to reinforce and strengthen prosocial norms. I also conclude that the tendency to falsely claim exposure does not reflect social desirability, but demand compliance in response to the interview situation. The study suggests that selective perception and response bias are distinct cognitive processes motivated by different factors.