Advisor(s) - Committee Chair
Reagan Brown (Director), Elizabeth L. Shoenfelt, and Katrina A. Burch
Department of Psychological Sciences
Master of Science
Employee selection is an important process for organizations. Organizations seek to select the best employees for their available positions. Testing is key to many selection efforts. The results of studies assessing the criterion-related validity of a selection test are affected by a number of statistical artifacts, one of which is range restriction. Range restriction has the effect of attenuating the correlation coefficient. Statistical equations exist to correct for the effects of range restriction, and they enable researchers to obtain a more accurate estimate of the validity coefficient. Thorndike (1949) developed the best known and most frequently used of these correction equations. In the present study, Monte Carlo analyses were used to compare the accuracy of two indirect range restriction correction equations. The only difference between the two equations is the nature of the predictor intercorrelation employed; one equation uses the restricted predictor intercorrelation, whereas the other uses the restricted value. The distinction between these values is important as both forms of the correlation are likely available in a predictive design, and the magnitude of each can be quite different depending on the extent of range restriction. Given these differences between the two forms of the equation, I hypothesized that the equation utilizing an unrestricted predictor intercorrelation would be more accurate. Results indicated that the equation that made use of the unrestricted correlation was generally more accurate, particularly when the selection ratio was low, and the predictors were not highly correlated.
Human Resources Management | Industrial and Organizational Psychology | Psychology
Pelayo, Michael Thomas, "A Monte Carlo Analysis of Thorndike's Indirect Range Restriction Correction Equations" (2020). Masters Theses & Specialist Projects. Paper 3196.