Comparison of Prior Setting Methods for Multilevel Model Effect Estimation Based on Small Sample Imbalanced Nested Data in Bayesian Framework

Table 1

Estimated accuracy results of treatment effects and nesting effects for the SSIND under all operating conditions (c = 8).

The number of subjects in each group	ICC	Prior setting	Convergence rate		Cover		Bias		RMSE
			Treatment	Nesting	Treatment	Nesting	Treatment	Nesting	Treatment	Nesting
			Effect	Effect	Effect	Effect	Effect	Effect	Effect	Effect
5	0.05	U	96.1	98.8	98.02	94.94	−245	1338	113	134
		G	97.6	97.9	93.75	97.55	−171	−186	114	27
	0.10	U	97	99.6	98.56	94.38	−272	1656	118	166
		G	97.8	98.4	93.76	97.76	−156	−96	121	27
	0.15	U	96.7	99.4	98.24	91.65	−301	1979	123	198
		G	97.1	97.6	92.07	96.72	−169	−3	125	29
10	0.05	U	96	99.6	98.44	95.78	−177	582	88	65
		G	97.3	98.7	95.17	95.85	−114	−114	88	32
	0.10	U	96.9	99.5	98.14	95.18	−193	982	97	102
		G	97.8	98.3	93.76	96.64	−103	−120	98	29
	0.15	U	96.7	99.5	98.35	89.15	−194	1433	104	146
		G	98	98.1	92.86	96.94	−105	33	106	32
15	0.05	U	96.3	99.9	99.17	96.4	−121	309	73	45
		G	97.7	97.7	94.68	90.79	−86	−303	73	33
	0.10	U	97.2	100	98.87	96.2	−145	767	84	84
		G	97	98.4	94.43	94.11	−108	−119	84	27
	0.15	U	96.3	99.8	98.44	89.18	−190	1242	93	128
		G	96.3	98.8	94.08	94.43	−100	59	93	30
20	0.05	U	96.9	99.6	97.94	96.89	−84	138	68	37
		G	98	98.2	94.18	84.52	−40	−334	68	36
	0.10	U	96.6	99.8	97.52	95.29	−88	597	79	70
		G	98.1	99.1	94.09	92.03	−55	−142	79	28
	0.15	U	96.5	99.6	97.31	89.76	−120	1061	88	112
		G	97.8	98.4	92.02	94.11	−68	44	88	30