Skip to main content


Fig. 5 | BMC Medical Research Methodology

Fig. 5

From: Multiple imputation methods for handling missing values in a longitudinal categorical variable with restrictions on transitions over time: a simulation study

Fig. 5

Estimated mean difference in body mass index (BMI) for age z-scores and 95% confidence intervals for ex-smokers and current-smokers compared to never-smokers for the case study analysis obtained from a random intercept linear mixed effects model using different methodsa for handling missing data in maternal smoking. ACA, available case analysis; CCA, complete case analysis; continuous-calibration, imputation as a continuous variable using multivariate normal imputation with calibration; indicator-PDBR, indicator based imputation using multivariate normal imputation with projected distance-based rounding; PMMb, predictive mean matching. a Results are not shown for indicator-PDBR with restrictions, fully conditional specification with multinomial and ordinal logistic imputation, and two-fold fully conditional specification methods because the imputation models failed to converge. bMinimal differences were observed between the results of predictive mean matching with 5 and 10 nearest observations. Therefore, only the results for this method with 5 nearest observations are presented

Back to article page