Back to School – Crime - Apollo Mapping
Posted on December 3rd, 2013

Back to School – Crime

In a study about forecasting crime patterns, researchers at Rutgers University set out to devise risk terrain maps for gang behavior and shootings to see if they could estimate the likelihood of future criminal activities. Using risk terrain modeling, they applied their model to two six-month periods for a comparison of hot-spots based on past crimes data. The authors argue that current systems of spatially-based policing may be outdated because they operate on the premise that new crimes will predominately occur where older crimes have already taken place. Through the analysis of shootings in urban New Jersey, the authors seek to consider crimes as a hazard that can be evaluated through risk assessment.

Risk terrain and shooting overlay.

Past research has suggested that variations in crime are explained by opportunities to commit crimes; however the ability to operationalize ‘opportunity’ has been an ongoing issue. Using risk terrain modeling (RTM), it functions similarly to the offender-base risk assessment model, in that it allows for offender characteristics to be incorporated into a scale of risk, such as re-arrest, absconcion while on bail or violation of parole. However, the RTM approach goes beyond this tactic to include professional experiences of police, practitioner knowledge such as parole officers and bondsmen, compiled literature on offender patterns and other empirical methods to feed their model.

Hot-spot mapping is the default for police agencies across the board, though the authors believe that this base data can be greatly improved upon with the use of multivariate methods and raster data. Working with the New Jersey state police, the researchers mapped out an area of 2.8 square miles in Irvington, NJ, an urban community with a population of 65,000 people. Murder rates for this area in 2007 were 38.7 per 100,000, as compared to the national average of 4.9 per 100,000 people. The town has seen an influx of drug organizations and has a high gang population.

Clustering of top 10% high-risk locations in risk terrains during both periods of evaluation.

Being that shootings are a relatively infrequent type of crime in general, yet frequent in this area, the authors operationalized three variables they felt would help predict them. They were: the dwellings of the known gang members (722); locations of retail business infrastructure (108); and location of drug arrests (496). While these were not deemed the only worthwhile variables to consider, based on existing data, they were the most effective data points for their model.

Through their creation of risk terrains, the authors felt that they had created a reliable predictor for future crime, but also stressed that the model does not indeed guarantee crime. Future follow-up studies based on the model and study area will be able to speak to the validity of their prognosis, but the model as a tool for law enforcement has great potential to use the limited resources in the most effective manner.

Justin Harmon
Staff Writer

This entry was posted in The Geospatial Times and tagged , , , , Bookmark the permalink.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

    The Geospatial Times Archive