Skip to main content

& Technology
in Policing

Tools for the Trade

Case study

Recruiting and building a high calibre, highly engaged police force is a priority across the country. To meet ambitious recruitment targets, policing needs to scale up the way that recruitment and vetting happens. That is why the service has invested in tools to help predict who might be at risk of committing serious misconduct, not only for serving officers but for those at entry point.

Thames Valley Police (TVP) has used prediction modelling to help develop a tool that they believe could identify risky behaviour. The success of this project will be judged on its ability to reduce misconduct committed across the country by officers and staff within the police service, and to vet the sort of personnel they’d like to attract.

They (TVP) repurposed an established method of analytical prediction from the Australian Institute of Criminology, working with Australian experts to roll out a similar programme across their force. In order to be effective, they had to ensure that the learning was compatible with UK policing. It had two approaches: to validate the historical data against known offenders and then to apply the same process to those as yet unknown or at risk in order to identify early intervention opportunities.

Detective Inspector Peter Semczyszyn, of the Thames Valley Police Counter Corruption Unit, said:

Similar to the Australian model, using the analytics work to predict officers and staff who could be at risk of offending was helpful in offering interventions and preventative measures. It was also a useful tool for highlighting potential issues for those undergoing vetting renewals and reviews.

The methodology included a survey in which 55 questions were applied to 100 individuals who had been through a gross misconduct procedure, and 100 comparable individuals who had not. The data resulting from this helped to guide and develop the process for prediction and early identification modelling. Questions were asked, such as: are prior instances of misconduct useful in predicting serious misconduct? They also explored when officers are most likely to be at risk of serious misconduct, and whether there are any demographic features that are notably predictive of serious misconduct.

The Australian Institute of Criminology analysed their prediction service and saw its potential for vetting future candidates for the force. Likewise, Thames Valley Police identified a system that they believe could be implemented nationally and reduce the cost of internal investigations.

Technology can serve not only to help cultivate an exceptional workforce, as demonstrated, but also to tackle similar challenges faced by the service. One example is of Cambridgeshire Constabulary having developed a Policing Pathways app to use as part of its selection process. Its objective was to reduce the attrition rate during recruitment. Historically the complexity of a gruelling 7 stage selection process has seen almost a 90% fall out rate due to the rigour of each recruitment round.

Working with developers, Cambridgeshire Constabulary designed a system that sends notifications to the applicant at key stages, and regular intervals. Push notifications were sent to candidates to reduce drop-out through 1) higher pass rates at selection stages, 2) greater candidate engagement and motivation, 3) higher levels of job-related fitness, and 4) reduced successful candidate drop-out by communicating work expectations.

In return, the service has been able to collect data from the user showing where there has been a dip in engagement or a withdrawal from the process. This information alerted the service to contact directly those candidates who appeared to have disengaged from the recruitment process, potentially keeping them on board.

Visit for the full article on serious misconduct prediction.