Fig. 2From: The effect of high prevalence of missing data on estimation of the coefficients of a logistic regression model when using multiple imputationRelative bias in estimated regression coefficients (%) (when cholesterol is subject to missingness)Back to article page