If your neighbor’s house is burglarized, yours might be next in line. That’s because some crimes, such as burglary and gang violence, tend to happen close to each other in both time and space. In recent years, police departments have used a method called crime-mapping to document clusters and deploy surveillance to the latest “hot spots.”
But George Mohler, Martin Short, P. Jeffrey Brantingham, Frederic Schoenberg, and George Tita (Journal of the American Statistical Association, forthcoming) have found a new way to anticipate crimes: seismological models that predict aftershocks from earthquakes.
In an analysis of burglaries in one Los Angeles neighborhood during 2004 and 2005, Mohler and his colleagues observed that initial break-ins led to spikes in nearby burglaries (“aftershocks”) within a few hundred meters and a few days of the first event. In other words, areas around the first burglary are at heightened risk for further break-ins, but this risk fades over time and distance.
The researchers compared their predictive method against retrospective crime-mapping (the traditional method) to see which best anticipated future crime. The seismological models did a better job than “hot spots” maps in all cases. This enhanced ability to predict crime might not prevent the first “quake,” but it could help police departments better anticipate the tremors that follow.
