What Do Structured Selection Interviews Really Measure? The Construct Validity of Behavior Description Interviews

Abstract

Structured selection interviews have been shown to predict performance at work, although the reasons for this success are not well understood. This study examined the construct validity of 2 behavior description interviews used to evaluate customer service manager candidates. A multitrait-multimethod analysis of the interview ratings yielded discriminant validities that were larger than the convergent validities. In addition, interview items designed to assess aspects of the Big Five, integrity, and customer service orientation were largely unrelated to established paper-and-pencil measures of these constructs. Confirmatory factor analysis suggested that interview ratings were related more to interviewee and interviewer factors than to the constructs the interviews were designed to measure. These results suggest that structured interviews may suffer from the same "validity paradox" commonly found in assessment center research.