Spatial Scan Statistics Adjusted for Multiple Clusters

Table 2

The estimated type I error probabilities for the second most likely cluster when there is only one true cluster in Grand Isle county, with 95% confidence intervals. Different relative risks were used, two different bounds on the maximum window size as well as two different buffer sizes defining the number of neighboring counties that were removed together with the counties in the most likely cluster detected. The circle size is the maximum geographic size of the scanning window defined in terms of a percentage of the total populations.

Relative risk	Buffer size	Circle size	$\alpha =0.01$	$\alpha =0.05$
100	0	5	0.010 (0.008,0.012)	0.043 (0.039,0.047)
100	0	50	0.013 (0.011,0.015)	0.045 (0.041,0.049)
100	50	5	0.010 (0.008,0.012)	0.047 (0.043,0.051)
100	50	50	0.008 (0.006,0.010)	0.051 (0.047,0.055)
192.89	0	5	0.014 (0.012,0.016)	0.056 (0.051,0.061)
192.89	0	50	0.014 (0.012,0.016)	0.057 (0.052,0.062)
192.89	50	5	0.013 (0.011,0.015)	0.054 (0.050,0.058)
192.89	50	50	0.010 (0.008,0.012)	0.058 (0.053,0.063)
200	0	5	0.012 (0.010,0.014)	0.057 (0.052,0.062)
200	0	50	0.013 (0.011,0.015)	0.057 (0.052,0.062)
200	50	5	0.013 (0.011,0.015)	0.052 (0.048,0.056)
200	50	50	0.010 (0.008,0.012)	0.056 (0.051,0.061)
400	0	5	0.011 (0.009,0.013)	0.055 (0.051,0.059)
400	0	50	0.010 (0.008,0.012)	0.051 (0.047,0.055)
400	50	5	0.010 (0.008,0.012)	0.052 (0.048,0.056)
400	50	50	0.010 (0.008,0.012)	0.056 (0.051,0.061)