Bias and Efficiency of Meta-Analytic Variance Estimators in the Random-Effects Model

作者：Wolfgang Viechtbauer · 发表于：Journal of Educational and Behavioral Statistics · 年份：2005 · DOI：10.3102/10769986030003261 · 被引用次数：1711 · 研究领域：Meta-analysis and systematic reviews、Diverse Approaches in Healthcare and Education Studies

The meta-analytic random effects model assumes that the variability in effect size estimates drawn from a set of studies can be decomposed into two parts: heterogeneity due to random population effects and sampling variance. In this context, the usual goal is to estimate the central tendency and the amount of heterogeneity in the population effect sizes. The amount of heterogeneity in a set of effect sizes has implications regarding the interpretation of the meta-analytic findings and often serves as an indicator for the presence of potential moderator variables. Five population heterogeneity estimators were compared in this article analytically and via Monte Carlo simulations with respect to their bias and efficiency.