Prediction models for clustered data with informative priors for the random effects: a simulation study

Table 3 Results from the Bayesian models with informative priors including different percentages of discrepant expert opinion

	FREQ	BAYES.WI	BAYES.LI			BAYES.MI			BAYES.HI
Percentage wrong expert opinion	–	–	10%	30%	50%	10%	30%	50%	10%	30%	50%
Overall Brier score	.191	.192	.180	.192	.201	.174	.179	.182	.170	.173	.174
Overall C-index/AUC	.782	.781	.806	.781	.764	.818	.808	.801	.826	.821	.818
Overall calibration slope	.911	.907	.946	.874	.824	.982	.964	.950	.989	.988	.987
Within cluster C-index/AUC^a	.805 [.037]	.805 [.037]	.805 [.037]	.805 [.037]	.805 [.037]	.805 [.037]	.805 [.037]	.805 [.037]	.805 [.037]	.805 [.037]	.805 [.037]
Within cluster calibration slope^a	.914 [.102]	.914 [.102]	.946 [.091]	.939 [.100]	.935 [.100]	.953 [.077]	.939 [.084]	.935 [.085]	.962 [.059]	.953 [.068]	.951 [.070]

ISSN: 1471-2288