Effects of alternative point pattern geocoding procedures on first and second order statistical measures

The objective of this research is to examine whether or not the same dataset, geocoded using different methods, produces significantly different results when applying various methods used to examine both first and second order effects of a point pattern dataset. Observations from Sydney, Nova Scotia, Canada, and vicinity were geocoded in two ways: using a civic address point file derived from 20cm resolution orthophotos; and postal code geocoding using two different products. The resulting three point patterns were analysed using a bivariate K‐function, a ratio of kernel estimates and Kuldorff's spatial scan statistic. Analysis indicates that conclusions can vary depending on geocoding, while of the methods tested the bivariate K‐function produces results that are least affected by the type of geocoding employed.