COMPONENTS OF ARTIFACTUAL VARIANCE IN METAANALYTIC RESEARCH

Except for a few rough estimates, the literature contains no precise d ata on the relative size of each correctable artifact in a meta-analys is. The purpose of the present study was to determine these proportion s and see if artifacts were related to various sample and subpopulatio n parameters. The study confirmed the notion that sampling error accou nts for the bulk of the total artifactual variance (i.e., more than 90 % for small or medium samples and more than 70% for large samples, e.g ., N = 500). Nevertheless, variation in the percentages of each artifa ct was found to be primarily a function of sample size and mean correl ation. Total artifactual variance was found to be a function of subpop ulation correlation differences and sample size. Practical implication s of the results are presented.