A study of the validity of interview judgments and other predictors in the selection of professional and scientific personnel for a chemical company

The management of the home office plant of a division of a large chemical manufacturer requested an empirical validation of the predictions made in their selection procedure for professional personnel. The predictors consisted of Judgments by both trained and untrained interviewers, test scores, and certain biographical data. Of particular interest was the hypothesis of differential interviewer validity. A review of the literature indicated that interview reliability and validity has been found to be low. Several of the possible sources of error conducing to the low validities were summarized. Standard procedural requirements were reviewed. Some tentative evidence in favor of the hypothesis of differential interviewer validity was found, justifying further research. After describing the recruiting and assessment methods employed, the design of the study was explained. The predictors consisted of: (a) seven ratings made by the trained comprehensive interviewers, (b) seven ratings by the panel interviewers, some of whom were trained, (c) written interview comments pertaining to technical background of the candidates, (d) two test scores, and (e) certain biographical data. The reliabilities and limitations of each of these predictors were described. The two criteria were (a) a merti index which reflected the average percentage pay increase per month of active service, and (b) current job performance ratings by supervisors on a forced distribution basis. The limitations and interrrelationships of the two criteria were enumerated. Results were as follows. Group comprehensive validities were significant on only one of the seven scales. While some evience for differential interviewer validities was found for four of the comprehensive interviewers, it was not statistically significant. A group of panel interviewers had validities which were slightly better than chance. With the performance rating criterion, the differential validities for ten of the panel interviewers were somewhat higher than chance would allow. The validities of the interview write-ins were not significant. The trained interviewers conducting panel interviews contributed considerably more to both the positive and negative individual validities than did the untrained panel interviewers. For the biographical data, a low level of significant relationship was found in some departments for age, marital status, source of recruitment, grades, and college degree. Some of the test validities and individual interviewer validities were higher than the validities of the groups of comprehensive and panel assessors. The test validities were approximately equivalent to the individual validities of four comprehensive and ten panel interviewers. A comparison of the results of this study with the results of other studies was made. It was suggested that a fruitful problem for further investigation was that of the casual factors of differential interviewer validity.