Fig. 3From: Multiple imputation by predictive mean matching in cluster-randomized trialsRelative percent error in the MI standard error of β2 after MI with a correctly specified imputation model. Missing data were generated completely at random (MCAR) at a rate of 15% or 40% (π=0.85 in red or π=0.60 in green, respectively). From top to bottom, results are presented in increasing order of cluster sizes m within an increasing order of number of clusters k. Filled circles indicate PMM methods while open circles indicate parametric (NORM) methodsBack to article page