| Abstract: | The correlations between each of five predictors and supervisory ratings of job effectiveness were not significantly different from zero. On the basis of the job analysis that preceded the validation study, the author concluded that the low correlations between each of the five predictors and the criterion were not due to problems inherent in either the tests or the performance appraisal instrument. Rather, it was hypothesized that the low correlations were a result of rating errors made by supervisors. The results of an eight hour training program (Latham, Wexley, and Pursell, 1975) designed to minimize rating errors supported this hypothesis. Four of the five predictors correlated significantly with the performance ratings that were conducted after the supervisors had received the training. |
