This is the third of three chapters on hot-spot analysis from the user manual for CrimeStat IV, a spatial statistics package that can analyze crime incident location data.
This chapter discusses methods for identifying hot spots with zonal data. It focuses on four techniques for analyzing hot spots with zonal data or with individual-level data that have attributes (count or interval variables that measure a characteristic associated with the X and Y coordinates). The four techniques are Anselin's Local Moran, the Getis-Ord Local "G", the Zonal Nearest Neighbor Hierarchical Clustering algorithm, and the Risk-adjusted Zonal Nearest Neighbor Hierarchical Clustering algorithm. These techniques are needed because sometimes it is not possible to analyze data at the individual level. The user may need to aggregate individual data points to spatial areas (zones) in order to compare the events to data that are only obtained for zones, such as census data, or to model environmental correlates of the data points. Also, at times individual data are not available (e.g., when a police department releases information by police beats but not individual streets). Zonal data can include crime counts by zone, socio-economic information, or some other data that are aggregated to the small areas. Since the zones are not events, they must be spatially analyzed by assuming that all the data reside at a single point within the zone. This is usually the centroid (the geographical center of the zone). Before describing in detail the techniques for using zonal data in crime analysis, the chapter discusses the "local indicator of spatial association." An attachment to this chapter discusses the use of Local Moran's "I" to detect spatial outliers in soil organic carbon concentrations in Ireland. 7 references and extensive figures that include computer screens and maps to illustrate examples