University of Florida

Friday, April 25,1997

3:30 pm - refreshments

4:00 pm - talk

Room 327, Yost Hall

Abstract

We first provide a general overview of how hierarchical Bayes generalized linear models can be utilized for the analysis of spatial data, incorporating in particular the spatial correlation structure. The general results are then utilized to produqe estimates of leukemia incidence rates for 281 census tracts in an eight-county area of upstate New York. These estimates are used for mapping the said disease rates across the 281 census tracts. Such maps can be utilized for detecting clusters or aggregations of cancer cases. The hierarchibal Bayes methods are also used to find whether there is any clustering of the cases near prespecified putative sources of increased risk. In particular, we examine the effect of proximity of residence to eleven inactive hazardous waste sites containing the volatile organic compound trichioroethylene, a common contaminant of groundwater.