King's College London

One of the key features of a state is the monopoly on the legitimate use of force, most often exercised through the police. The ability of the police to do their job effectively depends on this ability to legitimately use force, as well as the public’s consent to be policed.

Events of the last few years have served to undermine the legitimacy of the police. The Black Lives Matter movement, and in particular its rise to prominence following the murder of George Floyd in the United States in 2020, served to draw stark attention to racial disparities in policing, while in the UK, the murder of Sarah Everard by a serving police officer, followed by the heavy handed policing of a vigil in her memory, raised the issue of gender inequity and misogyny within the Metropolitan police. Baroness Louise Casey’s review offers a sobering assessment, stating: "Trust in the police is at an all-time low, and restoring it requires fundamental change". All of these events prompt a further question: what can be done about it?

Our new paper, published this week in the Justice Evaluation Journal, offers one tool that police forces can make use of. The paper describes the evaluation of a new form of training, delivered over the course of a year to all officers serving in the Avon and Somerset Police. The new form of training is entirely scenario-based and builds on conflict management skills officers were taught during their initial recruit training, such as de-escalation, risk management, physical restraint and use of personal protective equipment.

An expert reference group reviewed national data to identify the most common circumstances when the police and public came into conflict, and where force may have to be used. These points of conflict were then broadly defined into plausible scenarios for the training. The scenarios include booking in at custody, attending a domestic incident, attending a fight in the street, performing a stop and search and dealing with a vulnerable person. Each scenario has multiple levels, and officers repeat the scenarios multiple times over the two-day training course.

Officers are presented with a simple brief of the scenario before they start, and then must deal with the situation that unfolds in front of them. For example, they may receive the following briefing:

“You are called to a residential property following a report from a concerned neighbour about the couple that live there. The neighbour has heard a loud voice shouting, followed by a loud bang and then silence.”

As they respond to a mock-up of a house in the training environment, responding officers may find a verbally bickering couple that are responsive to police requests or a highly aggressive individual armed with a prop knife who reports they have just assaulted their partner, or anything in between. Officers are assessed on how they manage the role-players in front of them, what they do to de-escalate and what justification they have for using force if they choose to do so.

Because the training could not be delivered to all officers in the force at once, it was rolled out gradually, in a pragmatic stepped wedge trial, in which a different group of officers are trained every week over the course of a year, until all officers have been trained. This type of roll out allows us to compare use of force by those officers who have already been trained in any given week, with the use of force by those who have yet to be trained. Historical data also allows us to control for officers’ underlying tendency to use force, and any seasonal variations.

The results are incredibly encouraging, showing a 10.9% reduction in the likelihood that officers used force in any given week – with the biggest reductions seen for "hands-on" uses of force. We don’t detect any differences in effects across different races and genders, suggesting that the impact is similar across the board, although the relative lack of diversity of Avon and Somerset’s population might be a factor. The likelihood of a member of the public being injured during a use-of-force incident by a third, while there was no change to the likelihood that a police officer was injured in a use-of-force incident.

This research shows that we can pragmatically build in evaluation of police training to its rollout in a way that allows us to learn about its effectiveness without compromising operational delivery. Most importantly, it shows that this training is effective in reducing force, without increasing the likelihood that an officer themselves is harmed. We are therefore pleased to see the training being rolled out across forces in England and Wales, the training model provides a practical solution to address one specific aspect of the urgent need for change in policing.