A TEST OF 2 REFINEMENTS IN PROCEDURES FOR METAANALYSIS

This study used Monte Carlo simulation to examine the increase in accu racy resulting from 2 statistical refinements of the interactive Schmi dt-Hunter procedures for meta-analysis: the use of the mean correlatio n instead of individual correlations in the estimation of sampling err or variance, and a procedure that takes into account the nonlinear nat ure of the range-restriction correction. In all of the cases examined, these refinements increased the accuracy of the interactive procedure in estimating the variance of population correlations and resulted in more accuracy than other procedures examined. The use of the mean cor relation in the sampling error variance formula also increased the acc uracy of variance estimates for the multiplicative and Taylor Series p rocedures.