λ μY μX AUCtrue   MW* R* P*

0

0

0.25

0.570

 Bias(×103)
95% Cl		 −13
(-17, -9)		 0
(-4,4)		 0
(-4,5)
        MSE(×104)
95% Cl		 18
(16,20)		 19
(18,22)		 25
(22,28)
        Coverage (%)
 95% Cl		 93.7
(91.8,95.6)		 95.1
(93.4, 96.9)		 94.9
(93.0, 96.8)

0

0

0.5

0.638

 Bias(×103)
95% Cl		 -25
(-28, -21)		 0
(-4,4)		 0
(-4,5)
        MSE(×104)
95% Cl		 22
(19,24)		 18
(16,20)		 23
(20,26)
        Coverage(%)
95% Cl		 90.3
(87.6, 93.0)		 95.2
(93.5, 96.8)		 94.8
(92.9, 96.7)

0

0

1

0.760

 Bias(×103)
95% Cl		 -42
(-45, -39)		 0
(-4,3)		 0
(-3,4)
        MSE(×104)
95% Cl		 31
(27,34)		 14
(12,16)		 17
(15,20)
        Coverage (%)
95% Cl		 76.6
(73.0, 80.2)		 95.2
(93.3, 97.0)		 94.4
(92.4, 96.5)
*MW: Mann-Whitney estimate (i.e., image); R: Reiser’s (2000) method; P: Probit method.
**Simulation I was run 100 times. Each time, we generated 1000 simulated data sets. Each data set consists of 100 cases and 100 controls. Each subject provides two replicate biomaker scores. Both true values and random errors are assumed to come from normal distributions with image, image.
Table 1: Bias, mean square error (MSE), and coverage for image from simulation I ** .