Evidence for response set effects in structured research interviews

The Addiction Severity Index and NIMH Diagnostic Interview Schedule data of 20 methadone-maintained subjects with ''fake bad'' invalid profiles on the Personality Assessment Inventory, 15 methadone-maintained subjects with ''fake good'' invalid profiles, and 158 methadone-maintained subjects with valid profiles were compared. The findings revealed a number of significant group differences on both measures with the highest scores for the fake bad subjects and lowest scores for the fake good subjects. These findings suggest that the response sets exhibited in response to the Personality Assessment Inventory questionnaire extended to performance during the two semistructured interviews. There was no indication that interviewers were aware of misrepresentation. The limitations of the findings and alternative interpretations of the data are considered.