Table 6.

ML and Bayesian Estimation Performances for Multilevel MIMIC Models

					Parameters vs. ML estimates		Parameters vs. Bayesian estimates
Factors	Items	Covariates	Sample size	Parameters	AAB	RMSE	AAB	RMSE
1	3	1	500	9	0.017	0.022	0.008	0.011
			1,000	9	0.015	0.020	0.005	0.006
			3,000	9	0.016	0.020	0.002	0.002
		2	500	10	0.015	0.019	0.007	0.010
			1,000	10	0.013	0.017	0.004	0.005
			3,000	10	0.011	0.015	0.002	0.003
	6	1	500	15	0.027	0.037	0.005	0.008
			1,000	15	0.025	0.034	0.003	0.004
			3,000	15	0.025	0.033	0.001	0.002
		2	500	16	0.017	0.023	0.004	0.005
			1,000	16	0.017	0.023	0.002	0.002
			3,000	16	0.017	0.022	0.001	0.001
2	3	1	500	18	0.028	0.035	0.007	0.009
			1,000	18	0.026	0.034	0.003	0.005
			3,000	18	0.025	0.033	0.001	0.002
		2	500	20	0.019	0.025	0.005	0.007
			1,000	20	0.018	0.024	0.002	0.003
			3,000	20	0.019	0.025	0.001	0.001
	6	1	500	30	0.050	0.064	0.007	0.010
			1,000	30	0.046	0.059	0.004	0.006
			3,000	30	0.042	0.054	0.001	0.002
		2	500	32	0.031	0.040	0.005	0.007
			1,000	32	0.031	0.039	0.003	0.004
			3,000	32	0.029	0.037	0.001	0.002

Note. AAB represents averaged absolute bias; RMSE represents root mean square error.