Mark SayManaging EditorWednesday 22 February 2017

Police can struggle to convert data insights to action

HMIC official highlights potential of big data in understanding the demands on policing, but warns it is only a first step in change

Police forces can make use of big data to gain a better understanding of the context and place of incidents they have to attend, an official of HM Inspectorate of Constabulary (HMIC) has said.

Peter Langmead-JonesHowever, Dr Peter Langmead-Jones (pictured), the inspectorate’s head of better inspection, warned that there will often be difficulties in changing the way police handle a situation in response to the insights.

He said the lessons have emerged from two research strands running over recent years, one led by the HMIC and the other by Greater Manchester Police, at which he also holds the role of head of external relations and performance.

He was speaking at the Big Data in Government Conference staged by the Reform think tank in London yesterday.

Predictive effort

HMIC has been working on a predictive model on which areas are most challenging to police. The work involved taking the national census output areas, of which there are 181,000 in England and Wales, as a starting point and building in a profile using open data of around 2,500 variables.

This can take in details from sources such as surveys and records of property prices, along with that from police forces, and less obvious factors such as the number of ATM machines or movements of mobile phones.

It showed that 10% of the predicted demand for policing in England and Wales is in 1% of the output areas. Langmead-Jones said that these were areas with high densities of people and the geographic percentage would be even smaller.

“That piece of work is progressing and is being used to contextualise the assessments HMIC makes every year,” he said.

Work in Manchester has looked at indicators of vulnerability; for example, in what factors cause domestic violence. It showed it tends to go up when the temperature rises, especially in deprived areas.

An unexpected factor was when one of the major football teams unexpectedly lost a game: the force used records of bookmakers’ odds to establish the correlation.

Radio data

Another project has involved mapping more than 800 million data points from the locations of police officers’ radios – indicating where they were at specific times – against where the demand had been.

“It says some really interesting things about where police officers are,” he said. “And it’s as revealing in telling us where they are not.

“There is a potential for that be used in supervision and planning activities.”

However, he warned that obtaining insights from the data would be only a first step, and that those that follow are not easy.

“No matter how profound the insights, we should not underestimate how difficult it is to translate those into changes of practice. For example, I was to tell you with 95% confidence there would be an incident of domestic abuse in this area within eight hours, what are you going to do about it.?

“It revealed something about how our mindset works. A lot of policing is wired to be post-event, with someone cast as victim, someone as offender, and the framework for the interaction is clear to the police officer.

“If the officer turns up and it hasn’t yet happened we don’t know what the respective roles are and we’re unsure about the framework for the interaction.”

Langmead-Jones said the Manchester work has been taking place over four to five years with a focus on how the challenges in policing differ according to place; and the HMIC programme has been running for two and a half years based on how data can be used in the assessing performance against the PEEL Assessment factors of policy, effectiveness, efficiency and legitimacy.