Table 3.

Results for the Beam et al mammography study [2] for an ordinal GLMM with several characteristics including patient’s age (young = 0, old = 1), rater experience (experienced = 0, inexperienced = 1), rater’s annual volume of reading mammograms (<2500 mammograms = 0, ≥2500 mammograms = 1) and rater gender (1= male, 2 = female).

Parameter	Symbol	Estimate	S.E.	Z-value
Ordinal GLMM parameters:
Thresholds: ( ${\alpha }_0=-\infty ,{\alpha }_5=+\infty$)
Between categories 1 and 2	${\alpha }_1$	−0.621	0.170	−3.657
Between categories 2 and 3	${\alpha }_2$	0.079	0.170	0.467
Between categories 3 and 4	${\alpha }_3$	1.037	0.170	6.103
Between categories 4 and 5	${\alpha }_4$	2.816	0.171	16.425
Fixed Coefficients:
Subject’s age (Older)	${\beta }_{11}$	0.549	0.258	2.130
Rater Inexperience (Inexperienced=1)	${\beta }_{21}$	−0.063	0.099	−0.635
Rater Volume (Higher)	${\beta }_{22}$	0.134	0.079	1.700
Rater Gender (Female)	${\beta }_{23}$	0.008	0.120	0.063
Random Effect Variance Components:
Subject intercept	${\sigma }_{u0}^2$	2.746	0.324
Subject’s age slope	${\sigma }_{u1}^2$	0.719	0.084
Subject correlation coefficient	${\rho }_{u_0{u}_1}$	−0.505	0.062
Rater intercept	${\sigma }_{v0}^2$	0.154	0.022
Rater’s inexperience slope	${\sigma }_{v1}^2$	0.142	0.019
Rater’s volume slope	${\sigma }_{v2}^2$	0.089	0.012
Rater’s gender slope	${\sigma }_{v3}^2$	0.009	0.001
Rater correlation coefficient	${\rho }_{v_0{v}_1}$	−0.126	0.097
Agreement Measures:
Experienced male radiologists with a high volume rating younger patients:
- GLMM Observed Agreement	p0	0.470
- Model-based Kappa	κm	0.306	0.016
Inexperienced male radiologists with a low volume rating older patients:
- GLMM Observed Agreement	p0	0.462
- Model-based Kappa	κm	0.254	0.019