| The proliferation of studies around key substantive questions in many fields has necessitated the development of procedures called meta-analysis for systematically integrating the results of multiple studies. In the present Monte Carlo study the well established Psychometric Meta-analysis (PMA) (Hunter, Schmidt, & Jackson, 1982) is compared to the Empirical Bayes meta-analysis (EBMA) of Raudenbush & Bryk (1986) under a variety of sample size, number of studies, and true population variance conditions. Comparisons of the Mean Squared Error due to the two procedures revealed that the EBMA was as efficient as the PMA in estimating the mean true effect size, and, under certain realistic conditions involving small N and large numbers of studies, the EBMA was more efficient than the PMA. Type I error rate and power were roughly comparable, with a slight tendency toward liberality in the PMA. |
