The reliability of the behavioral interview

The primary objective of the present study was to investigate the reliability of the behavioral interview. This objective was operationalized in three ways. First, a generalizability (G) study was performed to establish the generalizability across interviewers with respect to the number of areas identified as problems per client. Second, the agreement among the interviewers as to those specific areas which were identified as problems for a particular client and as to the specific problem-items within an identified area was determined. Third, the accuracy of interview data was measured by establishing the agreement between each human interviewer and a criterion (computer) interview. Four interviewers conducted comprehensive behavioral interviews with the same four clients. These interviews were audio-taped and transcribed in order to provide a verbatim account of the content. Each interviewer dictated a summary following each interview. In addition, each client completed a standardized computer interview. Transcriptions of interviews, dictations, and computer interview printouts were coded independently by two raters for areas and items identified as problems and areas and items questioned