Abstract
Hedges’ Q-test is frequently used in meta-analyses to evaluate the homogeneity of effect sizes, but for several kinds of effect size measures it does not always appropriately control the Type 1 error probability. Therefore we propose a parametric bootstrap version, which shows Type 1 error control under a broad set of circumstances. This is confirmed in a small simulation study.
DOI
10.22237/jmasm/1051747620
Included in
Applied Statistics Commons, Social and Behavioral Sciences Commons, Statistical Theory Commons
