. Author manuscript; available in PMC: 2015 Jul 1.

Published in final edited form as: Comput Stat Data Anal. 2014 Jul;75:53–65. doi: 10.1016/j.csda.2014.02.009

Table 3.

Simulation results with n = 20; B1, B2 and B3 are top three selected models.

	Parameter	True Value	B1	B2	B3	Estimate	95% CI	Pr(Inclusion)
n_i = 10	β ₁	0.8	0.446	0.426	0.642	0.430	(0.000, 0.739)	0.937
p_l₀ = 0.2	β ₂	0.4	0.310	0.298	0.353	0.292	(0.018, 0.453)	0.952
	β ₃	0.4	0.393	0.387	0.429	0.394	(0.258, 0.555)	1.000
	β ₄	0		−0.025		−0.004	(−0.061, 0.015)	0.191
	Σ ₀₀	0.4 (0.33)	0.637	0.636	0.908	0.702	(0.229, 1.554)	1.000
	Σ ₁₁	0.2 (0.21)	0.269	0.275		0.267	(0.000, 0.683)	0.887
	Σ ₂₂	0				0.011	(0.000, 0.090)	0.144
	Pr(Selection)		0.562	0.142	0.078

n_i = 10	β ₁	0.8	0.507	0.479	0.519	0.503	(0.338, 0.693)	1.000
p_l₀ = 0.8	β ₂	0.4	0.287	0.269	0.251	0.281	(0.144, 0.409)	0.993
	β ₃	0.4	0.358	0.352	0.366	0.357	(0.206, 0.522)	1.000
	β ₄	0		−0.011		−0.002	(−0.044, 0.034)	0.160
	Σ ₀₀	0.4 (0.33)	0.561	0.546	0.392	0.558	(0.194, 1.151)	1.000
	Σ ₁₁	0.2 (0.21)			0.040	0.000	(0.000, 0.000)	0.008
	Σ ₂₂	0				0.000	(0.000, 0.000)	0.002
	Pr(Selection)		0.825	0.159	0.007

n_i = 20	β ₁	0.8	0.738	0.715	0.696	0.733	(0.494, 0.976)	1.000
p_l₀ = 0.2	β ₂	0.4	0.360	0.388	0.376	0.366	(0.268, 0.457)	1.000
	β ₃	0.4	0.327	0.329	0.330	0.327	(0.243, 0.421)	1.000
	β ₄	0		0.057		0.011	(0.000, 0.096)	0.201
	Σ ₀₀	0.4 (0.32)	0.495	0.461	0.462	0.488	(0.205, 0.928)	1.000
	Σ ₁₁	0.2 (0.28)	0.376	0.352	0.377	0.371	(0.199, 0.627)	1.000
	Σ ₂₂	0			0.002	0.000	(0.000, 0.000)	0.008
	Pr(Selection)		0.793	0.200	0.006

n_i = 20	β ₁	0.8	0.696	0.791		0.705	(0.586, 0.837)	1.000
p_l₀ = 0.8	β ₂	0.4	0.395	0.464		0.401	(0.335, 0.475)	1.000
	β ₃	0.4	0.334	0.328		0.333	(0.297, 0.361)	1.000
	β ₄	0		0.035		0.003	(0.000, 0.037)	0.090
	Σ ₀₀	0.4 (0.32)	0.379	0.386		0.380	(0.215, 0.634)	1.000
	Σ ₁₁	0.2 (0.28)	0.191	0.200		0.192	(0.117, 0.308)	1.000
	Σ ₂₂	0				0.000	(0.000, 0.000)	0.000
	Pr(Selection)		0.910	0.090