Persuading the doubleminded: An examination of expectancy-values and ambivalence related to condom use to prevent HIV infection

This study examines expectancy-values (beliefs about the consequences of a behavior) and ambivalence (having both positive and negative attitudes toward a single attitude object) in the context of condom use to prevent HIV. A modified expectancy-value model is proposed and tested. The relationship of ambivalence to attitude certainty, attitude stability and attitude-behavior consistency is examined.;Elicitation and pretesting procedures were used to develop a questionnaire. Principal components analysis during the questionnaire development phase of the project revealed underlying dimensions to the utilitarian, normative, emotional and control expectancy constructs. The main study consisted of two surveys administered to college students at a six-week interval.;Several utilitarian and emotional expectancy-values were found to be significant predictors of global attitude. Findings for normative beliefs and emotional associations as predictors of global attitudes and behavior were mixed. Past behavior did not predict current behavior in this study.;Ambivalence toward using condoms was found to be positively related to attitude certainty as expected. However, ambivalent attitudes were found to be more stable than unambivalent attitudes, an unexpected finding. Attitude-behavior consistency was found to be lower for respondents who were highly ambivalent than for those with little ambivalence. Ambivalence was a marginally significant predictor of inconsistency in behavior.;Implications for HIV education, including message strategy and campaign evaluation, are discussed. Future research is recommended to examine multiple influences on behavior, particularly normative beliefs and emotion, and the relationship of expectancy-values to ambivalent attitude structures. Research is also recommended to examine the effectiveness of two-sided messages for persuading the ambivalent.;Like most research, this study suffers from some limitations. The study does not examine several variables which may be relevant to sexual behavior. The ethnically homogeneous student sample used in the study limits generalizability of the findings. The limitations of the measures employed, particularly self-reported behavior and pencil-and-paper measures of emotion, are discussed. Despite these limitations, the findings reported here have important implications for future research and for social marketing applications.