Bayesian Analysis for Inference of an Emerging Epidemic: Citrus Canker in Urban Landscapes

Neri, Franco M.; Cook, Alex R.; Gibson, Gavin J.; Gottwald, Tim R.; Gilligan, Christopher A.

We assess the hypothesis that parameters vary spatially between sites as follows. The model is fitted to pairs of sites J and K independently (J, K = B1, B2, D1, D2), yielding a sample from the marginal distribution, e.g., for b J and b K (and similarly for the other parameters) for each of the sites respectively. Under the prior assumption of independence of parameters amongst sites, we can then build a joint posterior distribution for b J and b K , and empirically evaluate the probability p JK b ð Þ~Pr b J wb K Dcensored ð data for sites J and KÞ. Should p JK b ð Þ be near 1 or 0, there is evidence that there is a difference in parameter values between sites; if intermediate, the joint posterior straddles the line of equality and we cannot conclude in which location the parameter is greater. Further details are given in [42] .