How confidently can we infer health beliefs from questionnaire responses?

Recent theoretical and empirical research has challenged the assumption of health psychologists employing social cognitive models that beliefs can confidently be inferred on the basis of questionnaire responses. It is charged that responses to questionnaires are a function of the intelligibility of the items, constructs and theoretical framework to survey participants. To investigate this contention, cognitions specified by Protection Motivation Theory (PMT) were measured with respect to intentions to use a condom and intentions to use dental floss. Three alternative formats of a questionnaire which varied the order of presentation of PMT items were distributed to student samples. Analyses showed that while questionnaire format did not affect the reliability of cognition measures, there was some variation in the significance and strength of correlations between PMT cognitions across formats. Individual differences in social desirability and salience of the target behaviour had small and predictable effects upon reliabilities and intercorrelations. Implications for the measurement of health beliefs and the status of questionnaire responses are outlined and suggestions are made for controlling bias associated with respondents' capacity to "read" a questionnaire.