Doctoral Dissertations

Date of Award

8-2004

Degree Type

Dissertation

Degree Name

Doctor of Philosophy

Major

Industrial and Organizational Psychology

Major Professor

Robert T. Ladd

Committee Members

Lawrence James, Michael Lane Morris, Chanaka Edirisinghe

Abstract

Adverse (or disparate) impact has probably represented one of the most persistent and pervasive problems in employee selection. Innumerable approaches to eliminating its presence have been attempted, but most have been met with limited success. To date, this success has been measured in only slight reductions in adverse impact unless substantial losses in validity are accepted. While a number of reasons for these results have been advanced, this research asserted that part of the problem originated in the narrow perspective with which employee selection is often defined. This narrow perspective has resulted in a singular focus on validity with insufficient attention allocated to multiple criteria. The purpose of the present research was to expand upon an earlier study (Henderson & Ladd, 2001) that introduced a methodology (constrained estimation) that incorporated multiple objectives into the decision-making process associated with employee selection. Specifically, the goals of the methodology included reducing adverse impact while maintaining validity. In order to test the efficacy of this methodology, constrained estimation was applied to both Monte Carlo data as well as archival data obtained from an assessment project conducted from 1992 to 1993. It was also compared to two commonly used predictor weighting methodologies – Ordinary Least Squares regression and Unit Weighting. Results suggested that constrained estimation was moderately successful in reducing, but not eliminating, adverse impact while maintaining validity. Implications, limitations, and suggestions for future research are discussed.

Download
Files over 3MB may be slow to open. For best results, right-click and select "save as..."

Share

COinS