Corrected Formulas for Computing Sample Sizes Under Indirect Range Restriction

Sackett and Wade's (1983) study added significantly to our understanding of the interaction of statistical power, reliability, predictor intercorrelation, and selection ratio in determining the sample sizes ( N s) required in validation studies where indirect range restriction is assumed. Although Sackett and Wade's conclusions were correct, their results were marred by computational problems. This note presents correct formulas for determining N under indirect range restriction. The N s obtained with the corrected procedure are larger than those reported by Sackett and Wade, but they are usually substantially smaller than those given by Schmidt, Hunter, and Urry (1976) for direct range restriction.