. Author manuscript; available in PMC: 2014 Dec 26.

Published in final edited form as: Stat Methods Med Res. 2013 Jun 26;25(4):1596–1619. doi: 10.1177/0962280213492588

Table 5.

Trivariate model parameter estimates and standard errors

	Se (SE)	Sp (SE)	Disease prevalence (SE)	Test prevalence (SE)	σ_μ (SE)	σ_ν (SE)	σ_ε (SE)	ρ_μν (SE)	ρ_εμ (SE)	ρ_εμ (SE)	AIC	-2logL
model I	0.826 (0.041)	0.887 (0.021)	0.448 (0.075)	0.433 (0.058)	1.123 (0.236)	0.671 (0.203)	1.387 (0.260)	0	0	0	2653.8	2641.8
model II	0.827 (0.041)	0.888 (0.021)	0.448 (0.075)	0.433 (0.059)	1.130 (0.238)	0.673 (0.205)	1.387 (0.260)	−0.311 (0.328)	0	0	2655.0	2641
model III	0.826 (0.042)	0.877 (0.026)	0.448 (0.076)	0.438 (0.062)	1.138 (0.241)	0.715 (0.235)	1.396 (0.263)	0	0	−0.401 (0.471)	2654.5	2640.5
model IV	0.822 (0.042)	0.887 (0.021)	0.447 (0.075)	0.430 (0.060)	1.109 (0.237)	0.678 (0.206)	1.383 (0.260)	0	0.176 (0.256)	0	2654.5	2640.5
model V	0.823 (0.042)	0.879 (0.026)	0.447 (0.075)	0.435 (0.062)	1.117 (0.240)	0.702 (0.229)	1.381 (0.260)	0	0.122 (0.270)	−0.328 (0.547)	2656.2	2640.2
model VI	0.828 (0.042)	0.880 (0.025)	0.448 (0.076)	0.437 (0.062)	1.146 (0.244)	0.697 (0.223)	1.396 (0.263)	−0.247 (0.340)	0	−0.325 (0.516)	2655.9	2639.9
model VII	0.823 (0.042)	0.888 (0.021)	0.447 (0.075)	0.429 (0.061)	1.114 (0.238)	0.681 (0.208)	1.384 (0.259)	−0.284 (0.329)	0.165 (0.256)	0	2655.9	2639.9
model VIII	0.823 (0.042)	0.881 (0.025)	0.447 (0.075)	0.434 (0.064)	1.117 (0.239)	0.703 (0.226)	1.387 (0.261)	−0.298 (0.323)	0.175 (0.256)	−0.343 (0.502)	2657.5	2639.5

Model I–VIII is trivariate GLMM with all possible combinations of correlation parameters. Model I assumes all correlation coefficient ρ_μν,ρ_εμ and ρ_εν equal to 0. Model II, III, IV assume only one of the correlation coeffecients is not 0: ρ_μν ≠ 0, ρ_εν ≠ 0 and ρ_εμ ≠ 0, respectively. Model V–VII assumes two of the correlation coefficients are not 0: ρ_εμ ≠0 and ρ_εν ≠ 0, ρ_μν ≠0 and ρ_εν ≠ 0, ρ_μν ≠0 and ρ_εμ ≠ 0, respectively. Model VII assumes none of ρ_μν, ρ_εμ and ρ_εν are 0.