Bias in Observed Validity Estimates When Using Multiple Valid Predictors

Simulated data, validity reports and a firefighter predictive validation study are used to examine bias created by three common selection problems-range restriction, applicant incumbent attrition, nonlinearity compression of high test scores. Top 20% samples drawn from an pool with known coefficients demonstrate that the sample estimates predictors differentially biased in both magnitude direction, depending on strategy used. Concurrent designs generally favor novel predictors. Corrections for direct range restriction across situations were mostly ineffectual. With proper scaling, corrections indirect accurate, but cross-variable biasing effects can occur when score distributions individual differ. Many biases found simulation results demonstrated where variations Pearson-Thorndike corrected validities full information maximum likelihood (FIML), approaches all compared as assessments. normalized predictors, Pearson FIML methods show general mental ability physically demanding job tasks predicted performance throughout 30-year study, no evidence interactions or leveling at