Evaluation of Four Multiple Imputation Methods for Handling Missing Binary Outcome Data in the Presence of an Interaction between a Dummy and a Continuous Variable

Table 3

Comparison of the performance of imputation methods based on the CRs of regression coefficients over 1000 simulations.

Variable	Method	MAR					MCAR
		10%	20%	30%	40%	50%	10%	20%	30%	40%	50%
(BMIreal)	MCR	1.000	1.000	0.981	0.964	0.885	1.000	1.000	0.982	0.955	0.924
	MI	1.000	1.000	0.998	0.997	0.984	1.000	1.000	0.998	0.992	0.985
	MRF	1.000	1.000	1.000	0.998	0.986	1.000	1.000	1.000	0.998	0.994
	MS	0.996	0.967	0.826	0.702	0.556	0.997	0.953	0.850	0.674	0.563
(Betotal)	MCR	1.000	0.999	0.982	0.944	0.906	1.000	0.997	0.987	0.955	0.901
	MI	1.000	0.999	0.998	0.994	0.983	1.000	1.000	0.998	0.995	0.986
	MRF	1.000	0.999	0.987	0.942	0.903	1.000	1.000	0.991	0.963	0.875
	MS	0.983	0.740	0.415	0.205	0.092	0.973	0.727	0.402	0.208	0.078
(PACS4)	MCR	1.000	0.996	0.965	0.893	0.834	1.000	0.997	0.987	0.961	0.907
	MI	1.000	1.000	0.998	0.991	0.987	1.000	0.999	0.999	0.998	0.983
	MRF	1.000	0.980	0.919	0.798	0.678	1.000	0.997	0.964	0.926	0.850
	MS	0.896	0.500	0.210	0.093	0.049	0.973	0.666	0.301	0.140	0.046
(PSPS)	MCR	1.000	0.994	0.955	0.907	0.824	1.000	1.000	0.983	0.946	0.899
	MI	1.000	1.000	0.997	0.995	0.985	1.000	1.000	1.000	0.998	0.989
	MRF	1.000	0.980	0.920	0.789	0.676	1.000	0.999	0.970	0.904	0.830
	MS	0.893	0.489	0.205	0.088	0.042	0.978	0.648	0.295	0.110	0.047
X5 (Gender)	MCR	1.000	0.999	0.995	0.981	0.959	1.000	0.999	0.990	0.979	0.957
	MI	1.000	1.000	0.998	0.997	0.989	1.000	1.000	0.999	0.996	0.988
	MRF	1.000	1.000	0.998	0.999	0.994	1.000	1.000	0.999	0.997	0.992
	MS	1.000	0.991	0.974	0.951	0.962	0.999	0.991	0.974	0.958	0.924
(BMIreal ∗Gender)	MCR	1.000	0.997	0.929	0.822	0.712	1.000	0.996	0.946	0.834	0.677
	MI	1.000	1.000	0.999	0.998	0.990	1.000	0.999	0.999	0.987	0.983
	MRF	1.000	0.988	0.867	0.714	0.558	1.000	0.984	0.908	0.734	0.545
	MS	0.910	0.455	0.173	0.072	0.015	0.916	0.469	0.166	0.037	0.014