Fig. 14From: The effect of high prevalence of missing data on estimation of the coefficients of a logistic regression model when using multiple imputationEmpirical coverage rates of 95% CIs (when smoking is subject to missingness)Back to article page