Table 1.

Bayesian hierarchical models for the effects of annual precipitation on the relative risk of human West Nile virus in Mississippi, the United States.

Model	DIC	ΔDIC	Weight	Mean and 95% CI of precip coefficient
c	326.77	76.73	0.00	NA
c + Precip	309.58	60.54	0.00	-0.014 (-0.020, -0.007)
c + Precip + UH	257.80	8.76	0.01	-0.008 (-0.020, 0.005)
c + Precip + UH + CH	249.58	0.54	0.43	-0.005 (-0.020, 0.008)
c + UH +CH	249.04	0.00	0.56	NA

Note: letter c denotes constant; Precip is annual precipitation of the previous year; UH is uncorrelated heterogeneity; CH is the correlated heterogeneity modeled by conditional autocorrelative distribution; DIC is deviance information criterion; ΔDIC is the difference between the DIC of a candidate model and the lowest DIC; weight is equivalent to Akaike weight; and CI is credible interval.