Table 3.

Results of treatment effect estimates from propensity matching and inverse probability weighting, when an unmeasured confounding exists

		Homogeneous treatment effect/unmeasured confounding
		Propensity matching			Propensity weighting
		Coefficient		MSE	Coefficient		MSE
		Mean	SD		Mean	SD
No missing	No adjustment	2.011	0.154	4.068	2.011	0.154	4.068
	After adjustment	0.377	0.111	0.154	0.328	0.168	0.136
MCAR	Complete case analysis	0.362	0.152	0.154	0.336	0.233	0.167
	Missing indicator	0.870	0.138	0.776	0.774	0.171	0.628
	Multiple imputation
	With Y	0.376	0.119	0.155	0.330	0.171	0.138
	Without Y	0.807	0.119	0.665	0.771	0.165	0.621
	Combined method
	With Y	0.375	0.119	0.155	0.330	0.171	0.138
	Without Y	0.808	0.119	0.667	0.770	0.165	0.620
MNAR	Complete case analysis	0.145	0.163	0.048	0.157	0.255	0.089
	Missing indicator	0.514	0.117	0.277	0.345	0.197	0.158
	Multiple imputation
	With Y	0.422	0.141	0.197	0.354	0.200	0.165
	Without Y	1.003	0.129	1.023	1.028	0.154	1.079
	Combined method
	With Y	0.240	0.114	0.071	0.169	0.191	0.065
	Without Y	0.469	0.105	0.231	0.386	0.175	0.180