TABLE 5.

Simulation Study: Bias, MSE, and Nominal 95% NCP under High Separation Logistic Model

	β0			β1			β2			α1			α2
	Bias	MSE	NCP (%)	Bias	MSE	NCP (%)	Bias	MSE	NCP (%)	Bias	MSE	NCP (%)	Bias	MSE	NCP (%)
n = 500
Two-stage	−0.010	0.027	5.0	−0.002	0.003	4.0	−0.031	0.037	4.5	0.071	0.060	6.0	0.021	0.045	5.5
Imputation	−0.018	0.027	5.0	−0.002	0.003	4.5	−0.032	0.037	4.0	0.094	0.064	6.5	−0.029	0.045	4.5
EM	0.014	0.029	3.5	−0.007	0.003	4.5	−0.020	0.038	5.0	−0.036	0.068	3.5	−0.003	0.048	5.0
Bayesian	0.029	0.023	3.0	−0.005	0.003	4.5	−0.059	0.037	6.0	0.014	0.055	5.0	−0.010	0.042	5.0
n = 200
Two-stage	−0.028	0.079	4.5	−0.006	0.008	3.0	−0.004	0.097	4.0	0.099	0.12	4.0	0.042	0.16	5.0
Imputation	−0.034	0.079	6.0	−0.006	0.008	3.5	−0.004	0.096	4.5	0.11	0.12	4.0	0.049	0.17	4.5
EM	0.014	0.061	4.5	−0.007	0.009	7.5	−0.001	0.082	4.0	−0.021	0.061	1.0	0.015	0.074	1.0
Bayesian	0.059	0.052	4.0	−0.003	0.009	7.5	−0.106	0.084	4.5	0.018	0.078	1.0	0.029	0.091	1.5
n = 100
Two-stage	0.001	0.16	7.0	−0.003	0.016	4.5	−0.029	0.211	4.0	0.045	0.29	4.5	−0.042	0.26	2.5
Imputation	−0.005	0.16	6.5	−0.002	0.016	4.5	−0.031	0.21	4.0	0.055	0.29	5.0	−0.038	0.26	2.5
EM	−0.011	0.16	7.5	−0.018	0.014	3.0	0.010	0.22	4.5	−0.044	0.086	0.5	−0.004	0.061	0.5
Bayesian	0.094	0.11	3.5	−0.017	0.017	7.0	−0.15	0.18	7.5	−0.031	0.12	0.5	−0.002	0.11	0.5

MSE indicates mean square error; NCP, noncoverage probability.