The contribution of the interview to the prediction of job performance

A model of a naturally occurring selection process was presented and evaluated. The model included: (a) a cross-validated, empirically keyed biographical data measure, (b) applicants\u27 self ratings of skills and abilities relevant to the job, (c) interviewers\u27 ratings of applicant skills and abilities, (d) interviewers\u27 hiring decisions, and (e) objective measures of job performance for hired applicants.^ The present study improved upon previous interview validity studies by using a complex battery of predictors, which was more representative of the actual information used in hiring decisions than would be the case with a single predictor. Also, a methodology was used which estimated the criterion-related validity of interviewer ratings while controlling for interviewer effects, thereby overcoming the problem of nonindependence of interviewer judgments. The model used was the generalized group effect model (Kenny, 1985).^ Complete data were obtained for 102 Canadian life insurance sales agents who had been interviewed by one of 37 interviewers. Interviewers were managers who were normally involved in the hiring process in the agencies in which the applicants sought employment.^ Significant interviewer effects were found for interviewer ratings. Therefore, the applicant-within-interviewer unit of analysis was used within program LEVEL (Kenny & Stigler, 1983). Preinterview information such as biodata score and self ratings were found to have significant effects on interviewer ratings. Implications for including preinterview information when estimating interview validity were discussed. Using applicants as the unit of analysis overstated the predictivity of interviewer ratings.^ Interviewer ratings of applicant skills and abilities were not predictive of sales commissions. However, biodata score and seven of eight self ratings were predictive of commissions. Biodata score and self ratings of three of the self ratings were positively predictive of commissions, while four of the self ratings were negatively predictive of commissions. A focus of effort phenomenon was postulated to explain the pattern of negative and positive effects of self ratings. The nonsignificant predictivity of the interviewer ratings may be attributable to direct restriction in range due to selection based on the interview. Implications for future interview research were discussed.