Fig. 1From: Estimating relative risks in multicenter studies with a small number of centers — which methods to use? A simulation studyBias of the estimates of β1 (calculated as βestimate − βtrue) for different scenarios under a multicenter randomized controlled trial study design (a–c) and an observational study design (d–f) based on 1000 simulations for each scenario. All scenarios used a β1 of log(1.5), a control outcome rate of 15%, and an intracluster correlation coefficient of 0.08. GEE Generalized estimating equation, GLM-Bin Log-binomial regression model, GLMM Generalized linear mixed modelBack to article page