Heinz Holling
Westfälische Wilhelms-Universität Münster, Germany
Abstract
In this study, we will first give an overview of utility analysis of personnel selection. The development of the utility models used most often will be briefly outlined. Furthermore, we will summarize empirical studies reported in literature. These studies show considerable financial benefits due to an introduction or improvement of personnel selection programs. But, nearly all studies suffer from the often discussed problems of estimating the parameters, especially the standard deviation of job performance. Consequently, the distributional assumptions and the requirements for regression analysis cannot be proved. Our own study is based on sales figures representing an objective job performance measure. These sales figures allow a simple and more valid computation of the standard deviation of performance, the Achilles' heel of utility analysis. Also the validity, the correlation of predictor scores with sales figures, is estimated more precisely. Furthermore, we are able to prove distributional prerequisites. Presupposing a normal distribution for our performance data would lead to severe violations of the necessary assumptions for regressing performance on assessment center scores. As a consequence, the utility would be strongly overestimated. The results of our study, based on more reliable assumptions, show a more realistic financial benefit compared to previous studies.