TABLE 3.

Simulation Study: Bias, MSE, and Nominal 95% NCP under High Separation Linear 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	−0.015	0.007	5.5	−0.001	0.001	4.5	−0.007	0.008	5.0	0.060	0.015	5.5	0.021	0.012	3.5
Imputation	−0.020	0.007	6.5	−0.001	0.001	3.5	−0.007	0.008	5.0	0.075	0.017	7.5	−0.025	0.012	4.0
EM	−0.002	0.006	7.5	−0.001	0.001	4.5	−0.007	0.008	5.0	0.014	0.013	6.5	0.007	0.012	4.5
Bayesian	−0.004	0.006	5.5	−0.001	0.001	3.0	−0.001	0.008	5.0	0.012	0.012	5.0	0.002	0.012	3.5
n = 200
Two-stage	−0.001	0.022	9.5	−0.002	0.002	6.5	−0.026	0.020	4.5	0.058	0.034	4.5	−0.007	0.030	4.0
Imputation	−0.008	0.022	9.5	−0.002	0.002	6.5	−0.026	0.020	4.5	0.073	0.035	7.0	0.001	0.030	5.0
EM	0.021	0.016	6.0	−0.002	0.002	6.0	−0.026	0.020	4.5	0.011	0.034	6.5	−0.002	0.031	6.0
Bayesian	0.016	0.017	6.0	−0.001	0.002	6.5	−0.015	0.022	9.0	0.005	0.035	5.0	−0.005	0.030	3.0
n = 100
Two-stage	−0.033	0.034	5.5	−0.001	0.004	5.0	−0.004	0.037	4.5	0.12	0.090	12	0.027	0.063	5.5
Imputation	−0.038	0.036	5.5	−0.001	0.004	5.0	−0.003	0.037	5.5	0.13	0.094	12	0.031	0.064	4.0
EM	−0.014	0.031	3.5	−0.001	0.004	4.5	−0.005	0.037	5.0	0.067	0.098	12	0.016	0.067	6.0
Bayesian	−0.001	0.027	6.0	0.001	0.003	8.5	−0.009	0.031	5.5	−0.011	0.070	5.5	0.024	0.057	5.5

MSE indicates mean square error; NCP, noncoverage probability.