Space-Time Clustering and Prospective Hot-Spotting of Canadian Crime

Abstract

Previous research has consistently shown that repeat crime victimization is common. More recently, research has shown that near repeat victimization is also common, whereby targets located in close proximity to previously victimized dwellings/people/vehicles (depending on the crime) are at an increased risk of also being victimized. However, this elevated risk is only temporary and subsides over time. This near repeat space-time clustering has been found across various crime types (e.g., burglary, theft from motor vehicle (TFMV), gun crime, etc.) as well as across jurisdictions. However, the precise space-time patterning of crimes is location-specific. To date, no published research exists that has examined near repeat victimization using Canadian data; the current study fills this gap. This dissertation consisted of 4 phases of analyses. Phase 1 determined the exact space-time clustering of three crime types (burglary, TFMV, common assault) across three Canadian cities (Edmonton, AB, Moose Jaw, MB, Saint John, NB). Phase 1 results found significant near repeat space-time clustering for Edmonton burglary, Edmonton TFMV, and Saint John TFMV, with the exact near repeat space-time pattern varying from one data file to the next. Phase 2 analyses used the time and distance over which crime clusters (as found in Phase 1) to generate prospective risk surfaces. Risk surfaces were also generated using two traditional hot-spotting methods. Overall, the various hotspot mapping techniques examined were found to be comparable in their accuracy at predicting future crime. Phase 3 examined whether it was possible to improve the accuracy of prospective hot-spotting by considering three different strategies. Although Phase 3 results suggested that the three strategies examined were not effective at improving predictive accuracy of the maps, some interesting trends did emerge, which may have practical implications. Finally, Phase 4 investigated whether near repeat burglaries in one Canadian city (Edmonton, AB) were more likely to be committed by the same offender than more distant burglaries. Phase 4 results suggested that serial offending by the same offender offers a viable explanation for near repeat crime. The theoretical and practical implications of these results, as well as some limitations and directions for future research, are also discussed.