Direct and indirect assessment of explanatory models of illness.

Patients' self-reports of explanatory models (EMs) are sensitive to distortions, particularly as a result of social desirability, uncertainty towards one's own beliefs, and ethnic disparities with the interviewer. In contrast, reaction-time-based indirect measures are thought to be less sensitive to such factors. This article reports on two studies that applied direct (interview) and indirect (reaction-time-based association task) measures of EMs. Study 1 found evidence for the convergent validity of the direct and indirect measures, indicating that the two measures were essentially related. Furthermore, social desirability and uncertainty towards one's beliefs affected the association between the measures on two categories of EMs. Study 2 showed that, unlike the self-reports of EMs, indirect measures were less sensitive to the ethnicity of the interviewer. The nature of the indirect measure, and the construct that it measures, are discussed.