- Conference date: 18–22 March 2012
- Location: Amparo‐SP, Brazil
Bayesian spatial models for areal data assume a fixed neighborhood structure. The stochastic components are comprised by the observed data at each area and the parameters of their distributions, together with the hyperparameters. In this paper, we allow the spatial structure to be part of the parameter space. We introduce a hierarchically increasing neighborhood structure that retain the Markov property of the usual Bayesian spatial models. We illustrate the use of the model with the analysis of the spatial evolution of Sudden Infant Death rates in North Carolina from 1974 to 2006.
MOST READ THIS MONTH
MOST CITED THIS MONTH
Article metrics loading...