Testing a Geospatial Predictive Policing Strategy: Application of ArcGIS 3D Analyst Tools for Forecasting Commission of Residential Burglaries

Abstract

Classical placed-based crime prevention theories suggest existence of a relationship between certain characteristics of spatial design and configuration and crime occurrence. This study explored the relationship between natural surveillance - one of the least studied and understood principles of crime prevention through environmental design (CPTED) - and burglary commissions in three-dimensions.Natural surveillance has been claimed to differ when seen by neighbors, pedestrian passersby or individuals in vehicles, and to be influenced by viewing distance. Thus, the notion of natural surveillance was quantified to three categories of occupant, road and pedestrian surveillability. In addition, length of sightlines were restricted by the distance at which human eye is considered effective to eyewitness and interpret events.Employing a mixed methods research design, qualitative data (sketches made from oblique aerial imagery, field observations of architectural and landscape features, burglary crime reports and field observations of crime sites) were embedded and provided a supportive role for the quantitative data (georeferenced spatial and crime data) and quantitative analysis (univariate and multivariate statistical analysis). Firstly, ArcGIS geospatial tools were utilized for processing spatial and crime data in three-dimensions. Then, ESRI ModelBuilder was employed for automating the procedure of enumerating natural surveillance intensity. Spearman's rank correlation, Mann-Whitney U and binary logistic regression were employed to investigate the univariate and multivariate association between natural surveillance and burglary commissions or burglary occurrence. The results at the building opening level revealed that the log of the odds of burglary commission was negatively related to occupant surveillability and positively related to road surveillability. Findings at the building level showed that the log of the odds of residential burglary occurrence was positively related to road surveillability.This research shed light on the importance of the notion of "eyes upon the street" (Jacobs, 1961) even in a low socioeconomic-high criminogenic area. It has implications for developing proactive design and planning policies to help design crime out at the early stages of planning and development. It also demonstrates how law enforcement can further leverage societal investments in geospatial data to benefit public safety more effectively.