«Urban Problems and sPatial methods VolUme 17, nUmber 1 • 2015 U.S. Department of Housing and Urban Development | Office of Policy Development and ...»
Discussion (that is, step 10, communicate meaningful information) As the RTM demonstrates, one or more features of the physical environment can elevate the risk of crime. Comparing RRVs across model factors is useful for prioritizing risky features so that mitigation efforts can be implemented appropriately. For instance, foreclosed properties may be the direct targets of burglary; however, other properties within close proximity to foreclosures may also be at high risk because of the absence of invested caretakers who would otherwise serve as eyes and ears within the area. After risk factors are identified, stakeholders can explore the (likely) mechanisms through which risks are presented and then initiate mitigation efforts, such as improved community surveillance and new homeowner investment campaigns. In Chicago, for example, the CPD developed strategies to work with other city agencies, including the Chicago Housing Authority, to target problem buildings using city ordinances to improve conditions conducive to crime. The city agencies are also working with private lenders to address the broader scope of the foreclosure crisis.
Using environmental factors for crime forecasting has many benefits, such as enabling intervention activities to focus on places—not just people located at certain places—that could jeopardize public perceptions and community relations. Another benefit is that RTM is a sustainable technique because past crime data are not needed to continue to make valid forecasts. Police use RTM to be problem oriented and proactive in their effort to prevent new crimes without having to be concerned that a high success rate (and no new crime data) will hamper their ability to make new forecasts. In fact, the researcher-practitioner collaborations forged through the aforementioned NIJ projects have led to new approaches to police productivity that go beyond a heavy reliance on traditional law enforcement actions, such as stops, arrests, or citations. The police are now able to measure their effects on mitigating the spatial influences of risky features—with the goal of reducing one or more risk factor weights in postintervention RTMs or, better yet, suppressing their attractive qualities completely and removing them from the post model altogether.
All places may pose risk of burglary but, because of the spatial influence of certain features of the landscape (not simply past crime locations), some places are riskier than others. As demonstrated here, RTM helps to explain why spatial patterns of crime exist in a jurisdiction and what can be done to mitigate risks, not just chase the hotspots. With such spatial intelligence (Kennedy and Caplan, 2012), key stakeholders can identify the most vulnerable areas in a jurisdiction, enabling them to predict, with a certain level of confidence, the most likely places where crimes will emerge in the future—even if they have not occurred there already.
Conclusion Giving high regard to place-based risk assessments makes theoretical and intuitive sense: offenders know they take risks and that these risks increase in certain locations, and police are often deployed to certain geographies to combat crime and manage other real or perceived public safety and security threats (Caplan, Kennedy, and Miller, 2011; Kennedy and Van Brunschot, 2009). In 14 Urban Problems and Spatial Methods Risk Terrain Modeling for Spatial Risk Assessment future work, additional research is needed to assess the temporal dynamics of burglary incidents, as well as the social and situational factors. In addition, RTM can be applied to a variety of other topics, including injury prevention, public health, traffic accidents, and urban development.
Acknowledgments The authors thank the Chicago Police Department for providing professional insights and valuable data for this project. This research was supported, in part, by funding from the Rutgers University Center on Public Security and a grant provided by the National Institute of Justice (Award #2012-IJ-CX-0038).
Authors Joel M. Caplan is an associate professor in the School of Criminal Justice at Rutgers University.
Leslie W. Kennedy is a University Professor in the School of Criminal Justice at Rutgers University.
Jeremy D. Barnum is a doctoral student in the School of Criminal Justice at Rutgers University.
Eric L. Piza is an assistant professor in the Department of Law and Police Science at the John Jay College of Criminal Justice.
References Arlot, Sylvain, and Alain Celisse. 2010. “A Survey of Cross-Validation Procedures for Model Selection,” Statistics Surveys 4: 40–79.
Brantingham, Patricia, and Paul Brantingham. 1995. “Criminality of Place: Crime Generators and Crime Attractors,” European Journal on Criminal Policy and Research 3: 1–26.
Caplan, Joel M. 2011. “Mapping the Spatial Influence of Crime Correlates: A Comparison of Operationalization Schemes and Implications for Crime Analysis and Criminal Justice Practice,” Cityscape 13 (3): 57–83.
Caplan, Joel M., and Leslie W. Kennedy. 2013. Risk Terrain Modeling Diagnostics Utility (Version 1.0). Newark, NJ: Rutgers Center on Public Security.
———. 2010. Risk Terrain Modeling Manual: Theoretical Framework and Technical Steps of Spatial Risk Assessment. Newark, NJ: Rutgers Center on Public Security.
Caplan, Joel M., Leslie W. Kennedy, and Joel Miller. 2011. “Risk Terrain Modeling: Brokering Criminological Theory and GIS Methods for Crime Forecasting,” Justice Quarterly 28 (2): 360–381.
Caplan, Joel M., Leslie W. Kennedy, and Eric L. Piza. 2014. “Risk Terrain Modeling for Public Safety.” http://www.rutgerscps.org/docs/RTM_SafetyDatapalooza2014_CaplanKennedyPiza.pdf.
———. 2013a. “Joint Utility of Event-Dependent and Environmental Crime Analysis Techniques for Violent Crime Forecasting,” Crime and Delinquency 59 (2): 243–270.
———. 2013b. Risk Terrain Modeling Diagnostics Utility User Manual (Version 1.0). Newark, NJ:
Rutgers Center on Public Security.
———. 2012. “Risky Places and the Spatial Influences of Crime Correlates.” http://www.rutgerscps.
Drawve, Grant. 2014. “A Metric Comparison of Predictive Hot Spot Techniques and RTM,” Justice Quarterly. Advance online publication. DOI: 10.1080/07418825.2014.904393.
Dugato, Marco. 2013. “Assessing the Validity of Risk Terrain Modeling in a European City: Preventing Robberies in Milan,” Crime Mapping 5 (1): 63–89.
Heffner, Jeremy. 2013. “Statistics of the RTMDx Utility.” In Risk Terrain Modeling Diagnostics Utility User Manual (Version 1.0), edited by Joel M. Caplan, Leslie W. Kennedy, and Eric L. Piza. Newark, NJ: Rutgers Center on Public Security: 35–39.
Hesseling, Rene B.P. 1992. “Using Data on Offender Mobility in Ecological Research,” Journal of Quantitative Criminology 8 (1): 95–112.
Kennedy, Leslie W., and Joel M. Caplan. 2012. “A Theory of Risky Places.” http://www.rutgerscps.
Kennedy, Leslie W., Joel M. Caplan, Eric Piza. 2011. “Risk Clusters, Hotspots, and Spatial Intelligence: Risk Terrain Modeling As an Algorithm for Police Resource Allocation Strategies,” Journal of Quantitative Criminology 27 (3): 339–362.
Kennedy, Leslie W., and Erin G. Van Brunschot. 2009. The Risk in Crime. New York: Rowman & Littlefield.
Taylor, Ralph B. 1997. “Social Order and Disorder of Street-Blocks and Neighborhood: Ecology, Microecology and the Systemic Model of Social Disorganization,” Journal of Research in Crime and Delinquency 24: 113–155.
Taylor, Ralph B., and Adele V. Harrell. 1996. Physical Environment and Crime. Washington, DC: U.S.
Department of Justice, Office of Justice Programs, National Institute of Justice. https://www.ncjrs.
Tomlin, C. Dana. 1994. “Map Algebra: One Perspective,” Landscape and Urban Planning 30 (1): 3–12.
Weisburd, David, Nancy A. Morris, and Elizabeth R. Groff. 2009. “Hot Spots of Juvenile Crime: A Longitudinal Study of Arrest Incidents at Street Segments in Seattle, Washington,” Journal of Quantitative Criminology 25: 443–467.
16 Urban Problems and Spatial Methods Linking Public Health, Social Capital, and Environmental Stress to Crime Using a Spatially Dependent Model Greg Rybarczyk Alex Maguffee University of Michigan-Flint Daniel Kruger University of Michigan Abstract This article reports the findings from a localized spatial modeling approach and visual assessment of crime determinants in Flint, Michigan. Factors pertaining to socioeconomic condition, public health, social capital, environmental stress, and neighborhood context were analyzed spatially and statistically using exploratory data analysis, exploratory spatial data analysis (ESDA), ordinary least squares regression (OLS), and geographically weighted regression (GWR). The ESDA indicated that elevated crime densities clustered in legacy residential areas, suggesting the need for a spatially explicit model. The OLS model was able to explain 46 percent of the variation in the model, although the GWR model proved superior, explaining approximately 56 percent. The GWR results largely supported the OLS results, while providing additional insights into the directionality, magnitude, and spatial variation of localized predictors of crime. The factors that contributed positively to crime rates may provide policymakers and law enforcement officials with nuanced information needed for targeted crime-reduction/prevention strategies.
Introduction Environmental criminology examines how contextual conditions influence criminal behavior throughout space. This strategy allows law enforcement officials to observe the spatial copatterning of criminal events and possible correlates, providing an avenue for more efficient crime-reduction strategies (Phillips and Lee, 2011). The field has surged in popularity because of the availability of
robust spatial datasets, advancements in Geographic Information Systems (GISs), and user-friendly spatial analysis tools, allowing for comprehensive spatial analysis of criminal activity (Anselin et al., 2000). Furthermore, environmental criminology and spatial analysis tools have taken root in large part because of the need to account for spatial autocorrelation. Evidence has shown that spatial correlation effects can undermine confidence intervals and significance tests in global regression models such as the ordinary least squares (OLS) model (Matthews et al., 2010). To counter these issues, spatial modeling has evolved to the extent that spatial effects can be minimized, providing a greater understanding of underlying criminal processes (Anselin et al., 2000).
As of the date of this writing, a minimal amount of research has used spatially explicit models to detail significant relationships among crime and its correlates. A few notable studies do exist, however. For instance, a study by Cahill and Mulligan (2007) used a geographically weighted regression (GWR) model to measure invariant local crime correlates throughout Portland, Oregon, and found that local variation of crime predictors was significant. Malczewski and Poetz (2005) used the same method to examine geographical variations of residential burglaries in London, Ontario, Canada.
The authors found significant statistical and spatial variations in burglary risk factors such as dwelling value and multifamily housing density. Another important study found that with the use of GWR, crime correlates were spatially and significantly variant throughout an urban area (Graif and Sampson, 2009).
Displaying and interpreting GWR outputs cartographically has historically been complicated at best.
Typical outputs from GWR is a choropleth map displaying the t-values for each parameter estimated (Mennis, 2006). For example, Graif and Sampson (2009) used classified surface maps of t-values to display the statistical significance of each explanatory variable. Using this approach, however, hides the magnitude, directionality, and statistical significance of each estimate on crime. Displaying the parameter estimates in conjunction with their statistical significance can provide more meaningful GWR results, while reducing the volume of maps to interpret (Mennis, 2006). With this concept in mind, one purpose for this research is to build on past studies that have efficiently produced robust GWR visualizations. The specific goals of this research are to (1) demonstrate the advantages of a using a local spatial modeling strategy for crime modeling; (2) institute a progressive cartographic technique that distinguishes among magnitude, directionality, and statistical significance of explanatory factors; and (3) provide a better understanding about the relationships among socioeconomic conditions, public health, social capital, environmental stressors, neighborhood context, and crime.
The organization of the article is as follows: the next section, which provides an explanation of the data and preprocessing steps, is followed by a section that presents the modeling approach, and the article concludes with a research summary.
Data and Preprocessing Reported crime incidence data (26,961 records) from the year 2010 were obtained from the City of Flint Police Department and served as the dependent variable. In this study the crimes selected were aggravated assault, homicide, robbery, and all forms of criminal sexual conduct. The crime incidence data were aggregated and normalized for each U.S. Census Bureau-defined census block group (CBG) because that was the unit level used for analysis.
The explanatory variables were grouped into five categories: socioeconomic condition, public health, social capital, environmental stress, and neighborhood context (exhibit 1). Several socioeconomic status (SES) factors were considered because of their long-standing links to crime (Sampson, 1995).
The SES variables were obtained from the U.S. Census Bureau for 2010 at the CBG level.