Several metanalytic studies of group variability use variance ratios as measures of effect size. Problems with this approach are discussed, including limitations of using means and medians of ratios. Mean logarithms and the geometric mean are not adversely affected by the arbitrary choice of numerator. (SLD)