Skip to main content

Table 5 Performance comparison of alternative estimators: mean square standardized residual. Results obtained on average over 100 realizations generated under two different population size scenarios and 3 types of risk map (1 = observed, 2 = smooth, 3 = random). Poisson kriging was conducted with the semivariogram estimated from the underlying risk values (true γR(h)) or the simulated mortality rates. Bold numbers refer to best performances outside the ideal case where the true semivariogram of risk is known.

From: Geostatistical analysis of disease data: estimation of cancer mortality risk from empirical frequencies using Poisson kriging

Estimators WF population BF population
BREAST CANCER Scenario 1 Scenario 2 Scenario 3 Scenario 1 Scenario 2 Scenario 3
Observed rates 0.913 0.948 0.983 0.891 0.948 0.980
Population-weighted average 0.817 0.735 1.076 0.473 0.260 0.624
Global Empirical Bayes 1.524 1.974 0.973 0.494 0.284 0.315
Local Empirical Bayes 0.823 0.799 1.011 0.519 0.318 0.667
Poisson kriging (true γR(h)) 0.929 0.986 1.151 2.324 1.705 1.447
Poisson kriging 0.901 1.436 1.197 2.365 0.947 2.066
CERVIX CANCER       
Observed rates 1.224 1.188 1.050 1.299 1.247 1.096
Population-weighted average 1.229 0.542 1.156 0.441 0.257 0.540
Global Empirical Bayes 1.506 1.476 1.028 0.354 0.248 0.289
Local Empirical Bayes 1.154 0.679 1.101 0.638 0.407 0.663
Poisson kriging (true γR(h)) 1.135 1.053 1.032 1.184 0.754 1.039
Poisson kriging 1.071 0.760 1.252 0.880 0.600 1.179