. Author manuscript; available in PMC: 2018 Mar 28.

Published in final edited form as: Ann Appl Stat. 2013 Apr 9;7(1):154–176. doi: 10.1214/12-AOAS600

Table 1.

Comparisons between our areally referenced Gaussian process model and the three alternatives. p_D is a measure of model complexity, as it represents the effective number of parameters. Smaller values of DIC and Dawid–Sebastiani (D–S) scores indicate a better trade-off between in-sample model fit and model complexity

	P_D	DIC^*	D–S^*
Simple linear regression	79	9894	16,166
Random intercept and slope	165	4347	10,403
CAR model	117	7302	13,436
Areally referenced Gaussian process	5256	0	0

Both DIC and D–S shown are standardized relative to our areally referenced Gaussian Process model.