EMPAC Fellow’s geospatial research on crime harm

Crime volume and severity: a numerical and geospatial study

EMPAC Fellow John Tanner of Derbyshire Constabulary, has completed new research on Crime Volume and Severity: a numerical and geospatial study at Nottingham Trent University. John’s work, as part of a Post Graduate Certificate, was supervised by Dr James Hunter. This study represents the first of its type within the published literature that examines
the relationship between crime volume and crime severity at a geospatial scale smaller than that of entire police force areas. The discovery that crime volume and crime
severity are linearly scaled fundamentally calls into question the benefits of adopting a severity based policing model over a more traditional crime volume based approach.

The use of crime volume to determine the impact of crime upon the public has recently been called into question by several authors who note that all crime is not created
equally (Ignatans and Pease 2016; Sherman, Neyroud and Neyroud 2016). It follows therefore, the impact of individual crime types upon the very same public must also be
unequal. Despite this, crime volume is the primary metric utilised by the United Kingdom Home Office for measuring the performance of individual police forces within
England and Wales, even though significant efforts to develop alternative, more robust, approaches have been attempted over the years that focus upon an attempt
to measure the severity1 of crime (Pease, Billingham and Earnshaw 1977; Pease et al. 1977; Pease 1988).

Recent work by Sherman, Neyroud and Neyroud (2016) on the Cambridge Crime Harm Index (hereafter CHI) and Bangs (2016) on the Office for National Statistics
Severity Index (hereafter ONSI), indices that both purport to measure and reflect the severity of crimes, has resulted in significant interest in readdressing the shortcomings
of crime volume as a metric for the impact of crime on the public. In fact, the use of these severity indices to inform the allocation of decreasing police resources has
already begun by several police forces within the East Midlands Policing Collaboration Region (Travis and Dodd 2015; The Economist 2016) despite there not having being
any form of evaluation (either from the public or through academia) of the validity of the use of these indices for such a purpose. Recent work by Ashby (2017) has critically
examined and compared these severity indices, but the question still remains unanswered as to whether resource allocation informed by crime severity offers any
benefit over that based upon more traditional crime volumes.

The purpose of this study was to address the unanswered question as to how crime volume and crime severity (using the newly developed ONSI and CHI severity indices)
are related, and if a severity based policing model offers any benefits over that of one based upon a more traditional crime volume based approach. It is the first study of this
type in the published literature to address this question, despite the fact that a severity based approach is already being used to inform the allocation of ever decreasing
police resources by several police forces within the East Midlands Policing Collaboration Region.

Crime volume and crime severity have been shown to scale linearly, such that there will always be the same proportion of severity within an area (LSOA for this study)
irrespective of the volume of crime experienced within that same area. The bivariate linear regression models, developed using numerical and geospatial analysis
techniques, are so well specified that there is little room to doubt their validity. As such, the findings of this study fundamentally call into question the benefits of adopting a
severity based policing model over a more traditional crime volume based approach, and also challenge the assertion, originally made over two hundred years ago by
Cesare Beccaria, that crime should be differentiated on the basis of severity.

See John’s full research report here:-

Crime volume and severity full report

Comments