The Impact of Geocoding Accuracy on Near Repeat Analysis

• In Event: New Directions in the Analysis of Repeat/Near Repeat Crime

Abstract

Criminologists have examined many technical aspects of the near repeat phenomenon, including its generalizability across crime types and its susceptibility to changes in analyst-selected parameters like spatial and temporal bandwidth. One fundamental question that remains, however, is how sensitive near repeat patterns are to variations in geocoding accuracy and precision for the source data, usually representing crime incidents. This study features residential burglary data from three cities, geocoded alternatively to the address point, parcel centroid, street centerline, and block interval. Results indicate general stability in the observed empirical near repeat patterns across geocoding techniques, although evidence for patterning appeared to decay somewhat for the block-level geocoded data at progressively higher spatial and temporal bands, especially when using narrower (7 day) bandwidths. Implications for research and policy are discussed.

Authors

• Jonathan A Grubb, Georgia Southern University

• Matthew Nobles, University of Central Florida