Estimating and maximizing the utility of sequential selection decisions with a probationary period

Abstract

This paper provides a double extension of the two‐stage sequential model developed by Cronbach & Gleser (1965) to estimate the utility of a single‐cohort personnel selection decision. The first extension generalizes the formulation of the two‐stage procedure to that of a general multi‐stage model, whereas the second addresses the situation in which some of the newly hired employees are later judged to perform at an unacceptable level and are therefore dismissed at the end of the training and probationary period. Thus, the new model and the corresponding computational procedure can be used to estimate the utility of selecting sequentially either a fixed quota of selectees (FQS) or a fixed quota of successful selectees (FQSS). In addition, the paper presents an efficient procedure to derive predictor cutoff scores that maximise the utility of the overall selection decision.

Predicting adverse impact and mean criterion performance in multistage selection.

“…We also make available a computer program that implements the method. The present method extends previous related work on the analytical computation of multistage selection outcomes (Cronbach & Gleser, 1965; De Corte, 1998) to the case in which the applicant pool is not homogeneous but rather is a mixture of several applicant groups (both majority and minority groups) that differ in terms of their average performance on the predictors. The method is also related to the commonly used simulation approach (e.g., Doverspike, Winter, Healy, & Barrett, 1996; Hattrup & Rock, 2002; Sackett & Roth, 1996) in that it is based on the same assumptions and that its application is contingent on identical information with respect to the predictors and the criterion dimensions.…”

Combining predictors to achieve optimal trade-offs between selection quality and adverse impact.

“…The extension to multistage selections will not be easy, however, because for these situations an analytical method to link adverse impact and selection quality to the relevant selection parameters is not yet available. One could eventually resolve this issue by adapting the approach outlined by De Corte (1998) to the case in which the total applicant group is a mixture of majority and minority candidates.…”

Weighing job performance predictors to both maximize the quality of the selected workforce and control the level of adverse impact.

“…As a consequence, the results of the study are only indirectly relevant to obtaining stage-specific predictor composites that maximize the average quality of the selected applicants and, at the same time, eliminate adverse impact. To adequately solve the latter problem, the multistage selection utility framework of De Corte (1998) must be extended to express the condition that the total candidate group is a mixture of majority and minority applicants. Because the mixture condition is formally equivalent to the situation in which the quality of the applicants varies from one recruitment source to the other, the extension can proceed along the lines discussed by De Corte (1996).…”

“…In fact, such an approach is to be recommended in general because it also improves the overall quality of the selected employees (cf. Cronbach & Gleser, 1965; De Corte, 1998).…”