HOW INTERCHANGEABLE ARE DIFFERENT ESTIMATORS OF EFFECT SIZE

The computation of effect sizes is a key feature of meta-analysis. In treatment outcome meta-analyses, the standardized mean difference stat istic on posttest scores (d) is usually the effect size statistic used . Hoc-ever, when primary studies do not report the statistics needed t o compute d, many methods for estimating d from other data have been d eveloped. Little is known about the accuracy of these estimates, yet m eta-analysts frequently use them on the assumption that they are estim ating the same population parameter as d. This study investigates that assumption empirically. On a sample of 140 psychosocial treatment or prevention studies from a variety of areas, the present study shows th at these estimates yield results that are often not equivalent to din either mean or variance. The frequent mixing of d and other estimates of din past meta-analyses. therefore, mag have led to biased effect si ze estimates and inaccurate significance tests.