Figure 1
From: Risk of bias: a simulation study of power to detect study-level moderator effects in meta-analysis
Power simulation, moderator distribution 50:50. The figure shows the power for each of the combination (number of studies in each meta-analysis ranging from 5 to 200 controlled trials; study sample size ranging from 20 to 500 participants; moderator effect ranging from 0 to 0.4; residual heterogeneity ranging from τ2 = 0 to 0.8; for a 50:50 distribution of the moderator (for example, 50% ‘high quality’, 50% ‘low quality’).