Dear Scan Statistics Modelers,
Have you ever tried to estimate overdispersion of a spatio-temporal clustering which use a Poisson distribution in Satscan software? Currently, I am using Kulldorff’s scan statistics in Satscan software to build a series of moving cylindrical windows in which the circular base and height correspond to a geographic area and time span, respectively. The statistic was configured to identify clusters with high rates and to include a maximum of 50 % of the at-risk population in a cluster using a space-time permutation model with 999 replicates in SatScan v 9.6 47. Statistically significant clusters were identified using a value of p < 0.0001 and standard Monte Carlo inference.
However, I need to estimate overdispersion in the model, any idea how to do it?
Best regards