Risk Terrain Modeling (RTM) was developed to assess risk geographically through the creation of a composite risk map. RTM builds off of hotspot mapping, which is the practice of identifying areas with high concentrations of crime to predict the location of future crimes. The hotspot perspective suggests that crime is not evenly dispersed, but instead concentrated in small areas or hotspots. RTM goes beyond just identifying the locations of crime clusters. It identifies future locations that are conducive for a particular crime based on the changing environmental characteristics. This allows for a more proactive response to the crime. RTM builds off ideas from environmental criminology, problem-oriented policing, and hotspot mapping.
RTM has been applied to forecast risk for a few crimes with varying success including residential burglary, aggravated assault, shootings, and street robbery. The predictive validity of the RTM for street robbery suggested for each unit increase in risk, there was a 127% increase in the likelihood of street robbery. The focus of the current project was to improve on the existing risk terrain model for street robbery, using data from Portland. The nature of street robbery, along with the prevalence of the problem in Portland, made it a good choice for risk terrain modeling.
PRIMARY RESEARCH QUESTION
This project aimed to assess three questions through development of a risk terrain model for street robbery:
- Could a risk terrain model for 2011 street robbery in Portland be created with good predictive validity?
- Did the risk terrain model improve on the predictive validity of using retrospective analysis of street robbery alone?
- Could the model be updated with 2011 data to accurately forecast risk for 2012 street robbery?
The methods for this project involved the creation of a risk terrain model for 2011 street robbery using 2010 risk data, statistical validation of the model, and finally cross-validation of the RTM for street robbery using 2011 risk data to forecast 2012 street robbery risk. The author of the study used the Risk Terrain Manual and its ten step process for creating a final risk terrain map. After the risk terrain model for 2011 street robbery was created, the statistical validity was checked. First, using a logistic regression analysis, the predictive power of the model was identified. Second, to assess whether the model was an improvement over simply identifying prior hotspots for street robbery, the final model was compared to a risk model created using only 2010 street robbery locations.
This study used the risk factors that were previously identified by the literature; there may be other risk factors that could improve the predictability of the model. Future risk terrain models could assess if a different means of operationalizing the risk factors will lead to better results.
Research and analysis identified seven risk factors as significant predictors of street robbery locations: prior street robbery, illegal drug activity, gang related activity, vandalism, recent street robbery offender residences, alcoholic establishments, and mass transit stops. Responses to the primary research questions:
- The statistical validity tests affirmed that a risk terrain model for 2011 street robbery in Portland with good predictive validity could be created.
- The areas identified as at risk for future street robberies by the risk terrain model captured all but three (or less than 1%) of the 2011 street robberies while using past street robberies alone missed 110 (or 30%) of the 2011 incidents. This suggests that the risk terrain model as a tool for problem-oriented policing is an improvement over using analysis of prior incidents alone,
- After updating the model with 2011 data, the model accurately forecasted risk for 2012 street robbery. The locations of 99.1% of the 2012 street robberies (that occurred between 1/1/12 and 4/15/12) fell within areas forecasted by the risk terrain model as being at risk for a future street robbery.
The results of this project indicate risk terrain modeling is a promising crime analysis tool with potential to help improve targeting of law enforcement responses to this crime. One potential police response could be directed police patrols in areas identified as high risk as an attempt to reduce street robbery occurrences. Another potential police response could be taking actions to reduce the opportunity for street robbery. Police actions to reduce other crimes associated with risk for street robbery, such as illegal drug activity or vandalism, could potentially reduce the opportunity for street robberies.