THE USE OF FISHERS-Z IN SCHMIDT-HUNTER-TYPE METAANALYSES

Over the past few years, there has been some dispute as to whether Fis her's Z or the Pearson correlation (r) should be used in Schmidt-Hunte r-type meta-analyses. The two major reasons that Z has not been recomm ended are the possibly larger positive bins in estimating mean populat ion correlation (M(rho)) and the problem of estimating the standard de viation of population correlations (SDrho). In this study, two new met hods of estimating M(rho) and SDrho by using Z are suggested and teste d by Monte Carlo simulations. The results show no consistent advantage to using r instead of Z or vice versa in estimating both M(rho) and S Drho In fact the estimated M(rho) and SDrho from all three methods wer e highly similar and were almost identical when rounded off.