An empirical Monte Carlo test of the accuracy of the correlation, covariance, and regression slope models for assessing validity generalization.

An empirical Monte Carlo study with a large data base (N?=?84,808) was conducted to determine the accuracy of the correlation, covariance, and regression slope models for assessing validity generalization (VG). This study has resulted in the following three major conclusions: (1) The correlation between sampling errors and population parameters appears to be close to zero; (2) when the sampling error is the only artifact, all three models do well in estimating the relevant parameters across three sample sizes and 10 different numbers of validity studies per VG study; and (3) when the predictor reliability, criterion reliability, and range restriction are also included as artifacts, the accuracy of estimation depends on how closely the hypothetical and true distributions of artifacts match. All three models performed inadequately when the match between the two sets of distributions was poor. (PsycINFO Database Record (c) 2010 APA, all rights reserved)