Research Article

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 6

Comparison of the performance of imputation methods based on the proportion of the variation attributable to the missing data (λ) of regression coefficients over 1000 simulations.

VariableMethodMARMCAR
10%20%30%40%50%10%20%30%40%50%

(BMIreal)MCR0.0690.1210.1780.2320.2670.0670.1320.1710.2220.264
MI0.1060.2060.2990.3980.4890.1090.2130.3080.3950.485
MRF0.1160.2060.2700.3300.3640.1190.1990.2650.3170.373
MS0.0100.0380.0800.1230.1730.0100.0380.0760.1210.171
(Betotal)MCR0.0760.1390.1920.2330.2720.0780.1430.1960.2320.271
MI0.1110.2050.2960.3940.4860.1130.2130.3090.3940.484
MRF0.1250.2130.2900.3450.3920.1330.2170.2790.3320.369
MS0.0120.0400.0790.1250.1730.0110.0410.0790.1210.169
(PACS4)MCR0.0930.1560.2050.2330.2790.0800.1430.2000.2300.279
MI0.1270.2430.3210.4260.5100.1120.2100.3080.4010.472
MRF0.1490.2380.2990.3510.3820.1280.2190.2840.3410.381
MS0.0140.0460.0840.1230.1750.0120.0380.0790.1290.172
(PSPS)MCR0.0910.1570.2090.2430.2730.0800.1430.1910.2310.271
MI0.1300.2390.3290.4290.5080.1130.2160.3070.3950.486
MRF0.1520.2360.3040.3510.3840.1300.2180.2740.3270.380
MS0.0140.0460.0840.1290.1740.0120.0410.0790.1240.173
(Gender)MCR0.0720.1290.1800.2280.2640.0720.1320.1750.2230.258
MI0.1050.2010.3040.3840.4720.1120.2170.3060.4000.482
MRF0.1160.1990.2610.3230.3490.1140.2080.2590.3190.358
MS0.0100.0380.0760.1190.1710.0110.0380.0760.1240.170
(BMIreal ∗Gender)MCR0.0950.1590.2190.2740.3090.0950.1710.2200.2630.301
MI0.1110.2130.3070.3980.4830.1070.2140.3000.3910.491
MRF0.1600.2500.3110.3680.4040.1590.2490.3110.3560.408
MS0.0150.0480.0890.1350.1870.0150.0470.0920.1340.185