Effects of Truncating a Reference Population on Correction of Validity Coefficients for Range Restriction
Published Date: June 1, 1985
Correlation coefficients based on samples from occupational specialties that differ in qualification standards cannot be compared. The sample coefficients need to be put on the same metric by correcting them to a common reference population. The purpose of this analysis is to evaluate the effects of truncating the reference population on the correlation coefficients and on the inter-correlation of performance measures. Population wide estimates were computed in the full population and in the truncated population with the bottom 10 percent deleted.
