Violence is a significant public health issue. Levels of violent crime have been increasing in England and Wales. As a result, 18 Violence Reduction Units (VRU) were established in England and Wales as part of an intercept model, where those who pose the greatest risk of violence recidivism are supported away from the criminal justice system to see if it is possible to reduce violent offending and harm in communities. Thames Valley’s VRU has been developing these intercepts to reduce violence in the area.This research looks to validate an empirically grounded and novel risk assessment tool, known as OxREC, which has been developed by The University of Oxford using high quality datasets in other jurisdictions within a local Thames Valley context. This is a validation study and will not seek to operationalise the tool at this stage. However the tool offers the opportunity to more accurately identify risk of violent recidivism at the point of release from prison or those who have recently been released, which may offer opportunities for improved support, diversion and risk management.
This research will look to examine the predictive validity of a statistical model called OxRec in the UK using data from historic anonymised administrative data available to Thames Valley Police. OxRec has been developed and validated in Sweden and the Netherlands, but there has been considerable interest in the UK for more transparent and evidence-based approaches to policing. More specifically, we will test how well the OxRec can forecast violent and any crime committed within 1 year and 2 years after conviction/release from UK prisons of individuals released to the Thames Valley geographic region. This collaboration with The University of Oxford, a world leader in crime science, is one step to improve the service to the public and keep people safe in the Thames Valley.Statistical models should be open and explainable and this model will be fully published for public review. These models should also only form part of professional decision making where professionals can make more informed decisions as a result of taking account of these statistical models as aids.