Repository logo

Targeting Knife-Enabled Homicides for Preventive Policing: A Stratified Resource Allocation Model

Published version

Change log


Harinam, Vincent 
Sherman, Lawrence W 


jats:titleAbstract</jats:title>jats:secjats:titleResearch Question</jats:title>jats:pHow can police translate differing risk levels for knife homicide into a resource allocation model that follows the evidence?</jats:p></jats:sec>jats:secjats:titleData</jats:title>jats:pThe data for this publication are taken from the open access tables published in this journal by Massey et al.jats:italicCambridge Journal of Evidence-Based Policing</jats:italic>, 3:1-20, (2019). Those data show the linear relationship between the number of non-fatal knife assaults in a lower super output area (LSOA) in 1 year and the risk of a knife-enable homicide in the subsequent year, as well as how many of the 4835 LSOAs fell into each of five levels of increasing homicide risk.</jats:p></jats:sec>jats:secjats:titleMethods</jats:title>jats:pThe data from Massey’s research are re-calculated to show how a hypothetical number of 15-min police patrols could be allocated across all areas on the basis of ajats:italiccombination</jats:italic>of knife-enabled (KE) homicide risk level and the volume of LSOAs at each of the five levels of knife homicide risk. We display these results using both tables and multi-layered “wedding cake” images to show the size of different dimensions of each level, including proportion of total homicides and directed patrol frequency per LSOA at each of the five risk levels.</jats:p></jats:sec>jats:secjats:titleFindings</jats:title>jats:pBased on the hypothetical allocation of 10,000 patrol visits of 15 min in length, the highest risk group, with a forecasted 6% of all KE homicides, would receive 600 police patrols, divided by the 41 LSOAs at that risk level = 15 patrols across every 10 days. At the lowest level of risk, the 2787 LSOAs would share the 3000 patrols that a group of LSOAs would recieve for having 30% of homicides, which equals 1.1 patrols every 10 days. The hypothetical premise is that every LSOA gets some patrol, but the highest risk areas get 15 times more patrol to follow the evidence of risk. The formula is to (1) allocate resources by proportion of homicide at each risk level; (2) divide the allocated resources by the number of areas in each risk level group; and (3) allocate the resulting resources per day to each area in each of the 5 levels.</jats:p></jats:sec>jats:secjats:titleConclusion</jats:title>jats:pPolice face difficult tradeoffs between targeting more policing tojats:italicfewer</jats:italic>areas of higher risk (with more efficiency) or tojats:italicmore</jats:italic>areas of lower risk (with more effectiveness). The use of a formula combining risk and volume can help guide such decisions, illustrated by a layered “wedding cake” visualization for gaining clarity and legitimacy in communications.</jats:p></jats:sec>


Funder: University of Cambridge


48 Law and Legal Studies, 4805 Legal Systems, 4402 Criminology, 44 Human Society, 16 Peace, Justice and Strong Institutions

Journal Title

Cambridge Journal of Evidence-Based Policing

Conference Name

Journal ISSN


Volume Title



Springer Science and Business Media LLC