TABLE 4.

Results for Simulation 3 for sample sizes 500, 1000, and 5000. The true and false models for Designs A–C are described in Table 2.

			Estimators
			Simulation 3
			${\widehat{\mathrm{\Delta }}}_{{\text{IPW}}_1}^{\ast }$			${\widehat{\mathrm{\Delta }}}_{{\text{IPW}}_2}^{\ast }$			${\widehat{\mathrm{\Delta }}}_{\text{AIPW}}^{\ast }$
n	Models	Design	Bias	SD	MSE	Bias	SD	MSE	Bias	SD	MSE
500	True	A	−0.011	0.350	0.123	−0.003	0.328	0.108	−0.003	0.329	0.108
	False	A	−0.091	0.367	0.143	0.002	0.332	0.110	−0.001	0.329	0.108
	True	B	−0.050	1.133	1.284	−0.000	0.655	0.429	−0.019	0.697	0.486
	False	B	−1.388	2.402	7.691	0.045	0.778	0.607	0.004	0.856	0.731
	True	C	0.059	3.546	12.566	−0.018	1.260	1.587	−0.033	1.430	2.043
	False	C	−10.041	23.663	660.216	0.077	1.579	2.496	0.031	4.416	19.487
1000	True	A	0.010	0.250	0.063	0.012	0.239	0.057	0.012	0.239	0.057
	False	A	−0.078	0.257	0.072	0.015	0.242	0.059	0.010	0.239	0.057
	True	B	−0.016	0.703	0.493	−0.006	0.440	0.194	−0.008	0.445	0.198
	False	B	−1.379	1.473	4.068	0.031	0.542	0.294	−0.001	0.558	0.311
	True	C	−0.139	2.347	5.522	−0.000	0.952	0.906	−0.002	1.028	1.055
	False	C	−12.065	28.245	942.510	0.118	1.364	1.872	−0.159	6.912	47.752
5000	True	A	0.008	0.110	0.012	0.009	0.107	0.012	0.009	0.107	0.012
	False	A	−0.076	0.114	0.019	0.014	0.108	0.012	0.010	0.107	0.012
	True	B	0.001	0.291	0.085	0.005	0.206	0.042	0.005	0.206	0.043
	False	B	−1.356	0.598	2.197	0.042	0.260	0.069	0.006	0.255	0.065
	True	C	−0.033	0.988	0.976	0.008	0.449	0.202	0.007	0.462	0.213
	False	C	−12.955	12.913	334.401	0.166	0.933	0.898	0.010	3.233	10.442