Fig. 4From: Mind the gap: covariate constrained randomisation can protect against substantial power loss in parallel cluster randomised trialsPower under simple randomisation with covariate adjustment, for a data generation model with four, eight or 12 covariates, adjusted for A of the covariates in the analysis, with covariate coefficients of (1) 0.25, (2) 0.5 or (3) 1.0. All with 18 clusters and an intra-cluster correlation coefficient of 0.05. (Total variance changes across scenarios)Back to article page