Table 2.

Simulation results for ${\beta }_1$ under scenario 2 where the model ${\pi }_1(\mathbf{W},\mathit{\gamma })$ was misspecified with $\eta =0.5$

n	$p_m$ (%)	Method	Bias	MCSD	ASE	CP	MSE	RE
200	27	Proposed MPPLE	0.006	0.424	0.419	0.955	0.180	1.000
		AIPW	0.004	0.427	0.442	0.957	0.182	1.014
		MI(5)	0.001	0.445	0.446	0.956	0.198	1.100
	46	Proposed MPPLE	0.015	0.471	0.458	0.954	0.222	1.000
		AIPW	0.013	0.484	0.500	0.955	0.235	1.059
		MI(5)	$-$ 0.007	0.487	0.495	0.948	0.237	1.071
	59	Proposed MPPLE	0.009	0.520	0.504	0.939	0.271	1.000
		AIPW	0.009	0.556	0.579	0.952	0.310	1.143
		MI(5)	$-$ 0.010	0.536	0.553	0.951	0.287	1.061
400	27	Proposed MPPLE	0.000	0.301	0.298	0.952	0.091	1.000
		AIPW	$-$ 0.002	0.306	0.305	0.946	0.094	1.034
		MI(5)	$-$ 0.004	0.312	0.306	0.943	0.097	1.070
	46	Proposed MPPLE	$-$ 0.001	0.332	0.326	0.948	0.110	1.000
		AIPW	$-$ 0.006	0.350	0.343	0.945	0.122	1.111
		MI(5)	$-$ 0.007	0.348	0.337	0.933	0.121	1.098
	59	Proposed MPPLE	$-$ 0.004	0.364	0.359	0.946	0.132	1.000
		AIPW	$-$ 0.012	0.399	0.390	0.941	0.159	1.203
		MI(5)	$-$ 0.004	0.381	0.372	0.940	0.145	1.100
2000	27	Proposed MPPLE	0.006	0.130	0.133	0.960	0.017	1.000
		AIPW	0.004	0.132	0.134	0.953	0.017	1.035
		MI(5)	0.003	0.132	0.134	0.957	0.018	1.044
	46	Proposed MPPLE	0.006	0.141	0.145	0.955	0.020	1.000
		AIPW	0.005	0.146	0.149	0.952	0.021	1.084
		MI(5)	0.002	0.150	0.147	0.950	0.023	1.141
	59	Proposed MPPLE	0.005	0.152	0.159	0.958	0.023	1.000
		AIPW	0.003	0.163	0.167	0.957	0.027	1.150
		MI(5)	$-$ 0.001	0.163	0.161	0.952	0.027	1.153

$p_m$, percent of missingness; MCSD, Monte Carlo standard deviation; ASE, average estimated standard error; CP, coverage probability; MSE, mean squared error; RE, variance of the estimator to variance of the proposed MPPLE (relative efficiency); MPPLE, maximum partial pseudolikelihood estimator; AIPW, augmented inverse probability weighting estimator; MI(5), Lu and Tsiatis type B multiple imputation based on 5 imputations