Ratios comparing results obtained from adequately powered studies only with results obtained from all studies, in subset A of 1,107 meta-analyses: results shown are percentiles of the distribution of such ratios across meta-analyses.

1 in the all-studies meta-analysis in 256/1107 meta-analyses. In 199/256 (78%), also in the meta-analysis including adequately powered studies only. In 57/256 (22%), increased, but trivially, when underpowered studies were removed.