Estimating Predictive Validity When Range Restriction Due to Selection and Attrition Is Present
Jules I. Borack · Military Psychology · 1994
A percentage of students academically attrite from Navy training schools and receive no final grade. Exclusion of these individuals from predictive validity and other statistics yields estimates based on a nonrandom sample of selectees. A technique was developed for imputing a final grade for selectees who do not complete a course so that they can be included in validation analyses. The methodology was based on a mathematical model in which the grades of course cornpleters were distributed as the sum of truncated normal and normal variables. Another methodology, based on Guilford Case 111 and Guilford Case I corrections (Guilford, 1965), was also considered. The accuracy of these tech- niques was estimated using data from a variety of simulated school scenarios.