ABSTRACT
The geocodes, or spatial coordinates, of sites where cancer patients live or work may constitute useful information for developing hypotheses about the etiology of the disease and for testing these hypotheses via statistical analyses. Several statistical methods for the analysis of geocoded cancer data were summarized in chapter 9. These included methods for detecting the existence and identifying the locations of spatial clusters of cases and regression methods for relating cancer incidence to spatially varying risk factors. The hallmark of statistical methods such as these is that conclusions drawn from the analysis can be made with quantifiable uncertainty, subject to the assumptions of the underlying probability model on which the analysis is based. An example of such a conclusion is the statement: The probability that we would observe a cancer cluster of this magnitude, if cancer cases occurred totally at random within the at-risk population, is less than .001.
