Table 7.

Estimated misclassification matrices in PRS based on normal subject- and rater-specific random effects. In the body of the table, we report P̂(Y = k|D = d) with their bootstrap standard errors based on 1000 bootstrapped samples in parentheses for each value of the covariate (resident and regional expert group) respectively.

	d = 0	d = 1	d = 2	d = 3	d = 4
Residents
k = 0	0.693(0.05)	0.196(0.04)	0.213(0.07)	0.060(0.02)	0.118(0.07)
k = 1	0.203(0.04)	0.622(0.05)	0.611(0.07)	0.307(0.07)	0.090(0.06)
k = 2	0.052(0.03)	0.166(0.03)	0.151(0.05)	0.348(0.07)	0.185(0.06)
k = 3	0.051(0.04)	0.015(0.02)	0.013(0.01)	0.224(0.05)	0.327(0.07)
k = 4	0.001(0.00)	0.001(0.00)	0.013(0.01)	0.061(0.04)	0.280(0.09)
Regional Experts
k = 0	0.748(0.05)	0.171(0.04)	0.256(0.07)	0.062(0.02)	0.141(0.07)
k = 1	0.170(0.04)	0.562(0.04)	0.344(0.06)	0.244(0.06)	0.087(0.05)
k = 2	0.041(0.02)	0.189(0.04)	0.342(0.07)	0.298(0.06)	0.089(0.05)
k = 3	0.034(0.02)	0.049(0.02)	0.047(0.02)	0.218(0.05)	0.430(0.11)
k = 4	0.006(0.01)	0.028(0.02)	0.012(0.01)	0.177(0.06)	0.253(0.10)