Knife-crime assault data to help forecast fatal stabbings

The researchers suggest that with the right technological approach, police could receive automated daily updates of homicide risk based on the latest data – helping to guide patrol allocations. The study is published in the Cambridge Journal of Evidence-Based Policing.

“If assault data forecasts that a neighborhood is more likely to experience knife homicide, police commanders might consider everything from closer monitoring of school exclusions to localized use of stop-and-search,” said study co-author Prof. Lawrence Sherman from the University of Cambridge.

“Better data is needed to fight knife homicide. The current definition of knife crime is too broad to be useful, and lumps together knife-enabled injuries with knife threats or even arrests for carrying knives.”

“Police IT is in urgent need of refinement. Instead of just keeping case records for legal uses, the systems should be designed to detect crime patterns for prioritizing targets. We need to transform IT from electronic filing cabinets into a daily crime forecasting tool,” he said.  

However, Sherman and colleagues caution that solely focusing on assault hotspots is not a “panacea”. The 41 top hotspots in the study contained only 6 percent of the following year’s total knife deaths.

The new study was co-authored by DCI Massey as part of his Master’s thesis research at Cambridge’s Institute of Criminology, where he worked with Sherman and his colleague Dr. R. Timothy Coupe.

“These findings indicate that officers can be deployed in a smaller number of areas in the knowledge that they will have the best chances there to prevent knife-enabled homicides,” Massey said.

No single area in the 2017-18 financial year had more than one fatal stabbing. However, 69 percent of the knife homicides occurred in census areas where at least one non-fatal knife assault had taken place the year before.

The study’s authors say the last decade of deadly knife crime has been a “moving target”. The research suggests little repetition of homicide location. In the ten years up to 2018, there were 590 knife homicides across London spread over 523 different census areas.

The researchers write that geo-coding annual knife assaults to a census area provides a reliable – if far from perfect – basis for forecasting knife homicide.

Added Sherman: “When combined with intelligence-gathering on the streets, this form of data analysis could enhance the effectiveness of scarce resources to create a new and more powerful preventative toolkit. Our study is just the first step.”

— Read more in John Massey et al., “Forecasting Knife Homicide Risk from Prior Knife Assaults in 4835 Local Areas of London, 2016-18,” Cambridge Journal of Evidence-Based Policing (14 April 2019) (DOI: 10.1007/s41887-019-00034-y)