Effects of residual smoothing on the posterior of the fixed effects in disease-mapping models.

摘要: Disease-mapping models for areal data often have fixed effects to measure the effect of spatially varying covariates and random with a conditionally autoregressive (CAR) prior account spatial clustering. In such regressions, objective may be estimate while accounting correlation. But adding CAR can cause large changes in posterior mean variance compared nonspatial regression model. This article explores impact on estimates variance. Diagnostics are proposed inflation from collinearity between each region's influence change effect's by effects. A new model that alleviates is developed extensions these methods point-referenced discussed.