Figure 4.

Plots of power (%) versus heterogeneity for the random effects meta‐analytic estimator calculated from four combinations of analytical methods (all SAFV, all SACS, all ANCOVA and Rand. select.) with ${\sigma }_{Y_{int}}^2=1$. Rows represent small, moderate and large underlying intervention effects. SAFV, simple analysis of final values; ANCOVA, analysis of covariance; SACS, simple analysis of change scores; Rand. select., random selection