Simulated cervix cancer risk map and the results of the Bayesian and geostatistical analysis. The number of cases for each county was simulated by random sampling of a Poisson distribution that is defined by the white female population map of Figure 2 and a "true" risk map (A) generated using sequential Gaussian simulation. (C, E) Risk maps estimated from simulated rates using a Bayesian approach or ATA Poisson kriging. The class boundaries correspond to the deciles of the histogram of original rates. (D, F) Prediction variance associated with the Bayesian and geostatistical risk estimates.