Skip to main content
. Author manuscript; available in PMC: 2011 May 20.
Published in final edited form as: Stat Med. 2010 May 20;29(11):1206–1218. doi: 10.1002/sim.3862

Table 6.

The 95% confidence/credible interval length of disease prevalence based on simulation studies with 2000 replicates. The bolded cells represent the correctly chosen model.

n α (θ or ρ) Maximum Likelihood Methods Bayesian Models
⊥ model α model θ model ρ model ⊥ model α model θ model ρ model BMA*
⊥ model
5000 0.0706 0.1062 0.1415 0.0752 0.0726 0.1060 0.1394 0.0791 0.1204
25000 0.0309 0.0473 0.0633 0.0329 0.0312 0.0476 0.0629 0.0332 0.0534
α model
5000 1.25 0.0592 0.1018 0.1333 0.0686 0.0607 0.1018 0.1308 0.0724 0.1149
5000 3.00 0.0318 0.0920 0.1061 0.0500 0.0323 0.0923 0.1053 0.0527 0.1736
25000 1.25 0.0260 0.0454 0.0596 0.0300 0.0262 0.0456 0.0592 0.0304 0.0604
25000 3.00 0.0141 0.0412 0.0474 0.0220 0.0142 0.0414 0.0475 0.0223 0.1454
θ model
5000 1.25 0.0485 0.0888 0.1236 0.0569 0.0496 0.0890 0.1210 0.0597 0.1074
5000 3.00 0.0141 0.0183 0.0736 0.0150 0.0142 0.0184 0.0729 0.0150 0.1622
25000 1.25 0.0215 0.0397 0.0553 0.0251 0.0216 0.0400 0.0550 0.0253 0.0624
25000 3.00 0.0063 0.0081 0.0329 0.0067 0.0063 0.0082 0.0326 0.0067 0.1450
ρ model
5000 0.20 0.0369 0.1392 0.1028 0.1121 0.0377 0.1383 0.1024 0.1267 0.2384
5000 0.60 0.0163 0.2424 0.0727 0.2152 0.0163 0.2380 0.0721 0.2638 0.4371
25000 0.20 0.0163 0.0624 0.0459 0.0475 0.0164 0.0624 0.0461 0.0488 0.2034
25000 0.60 0.0073 0.1096 0.0325 0.0829 0.0073 0.1089 0.0324 0.0897 0.3685
*

BMA = Bayesian model average for the α, θ and ρ dependent models.