Table 2.

Estimated risk for each parameter set (conditional intercepts, marginal mean, covariance matrix) by estimation model under each of the four data generating mechanisms.

Loss Function	Estimation Model
Mean: Dependence: GARP:	mvn-mar equal sparse	pattern equal sparse	shrink equal sparse	shrink shrink sparse	pattern pattern non-sparse
Data Generating Model (A)
Zeta	33.9	68.6	8.2	34.9	56.7
Mu	53.5	130.6	9.2	54.4	120.7
Sigma	22.8	33.6	16.7	49.8	54.2
Data Generating Model (B)
Zeta	29.8	65.5	16.8	46.9	56.9
Mu	47.3	120.3	32.1	57.3	110.4
Sigma	17.6	30.8	15.0	49.8	53.2
Data Generating Model (C)
Zeta	310.8	67.9	61.9	140.9	95.1
Mu	191.7	111.8	90.8	97.2	123.2
Sigma	30.3	32.8	20.6	51.1	52.9
Data Generating Model (D)
Zeta	243.6	245.7	190.5	31.8	85.1
Mu	250.7	120.5	106.9	92.3	119.2
Sigma	388.4	377.7	367.6	29.9	45.6