A few years ago, I taught a statistics course for undergraduate criminology students. My main priority was to develop their critical skills in reading statistics. We started out with a map with 8 districts. The students first determined each districts crime levels by rolling two dice. The higher the throw, the higher the crime rate.  Students were then asked to allocate their police resources to districts. Instinctively they targeted “high crime neighbourhoods” and often left low crime areas unpoliced. With their crime fighters deployed they threw the dice again to determine an updated crime rate. Almost by magic the police had done their work, districts with the highest concentration of police almost all saw a decline in their crime rates, whilst those unpoliced experienced, sometimes dramatic, increases in crime. It was clear that levels of policing effected crime levels. This was of course their ‘common sense’ view, so many just believed it. Others were unconvinced, so they repeated the experiment. They got the same result.

Obviously, the crime rates were determined by a throw of the dice. So how come it appeared they were influenced by police numbers? Eventually, some more reluctantly than others, they accepted that their deployment of police was not actually influencing the crime levels determined by the dice. What I was trying to teach them was the difference between correlation and causality. Clever statistical links didn’t prove anything.

I remembered these classes when I read the recent Institute for Fiscal Studies (IFS) report Police infrastructure, police performance, and crime: Evidence from austerity cuts. Its conclusion is that closing police stations is not cost effective. It, the report claims, reduces clearance rates, reduces reporting of non-violent offences, and (most scarily) lowers house prices. Methodically, it draws on micro-economic analysis and requires fairly advanced specialist academic skills to follow and understand. It draws on data sets that would require considerable time to replicate. Instinctively I am suspicious of claims based on correlation and their vulnerability to selective cherry picking in the data used (see my critique of Charles Murray’s claim that prison works). But I have neither the time nor the inclination to undertake the detailed work to test the detail of this report.

However, it is worth a few swift tests. The report identifies that in the decade 2008-2018 the number of operational police stations in London decreased from 160 to 45, with 80% of closures happening in 2013. Given the focus on the impact of station closures the report’s empirical analysis is focused on 2011- 2016. The reports conclusion’s opening sentence includes the claim ‘that reductions to police spending contribute to a significant rise in violent crimes’ (p. 29). Whilst I can’t decipher the complex mathematical formula that leads to this conclusion, it is clear its focus is on “murder and assaults” (rather than domestic or sexual violence, both of which, despite increasing during this period, are not mentioned in the body of the report) To test this I did a quick internet search looking for the violent crimes closest to this category – ‘murder’ and ‘knife crime’. 

The graphic below from the Guardian shows London homicides have been declining. The decline has slowed in the last decade, but murder rates are no higher today than they were in 2013 when the closures took place. There has certainly been no ‘significant rise’ in murder rates.

If we look at ‘knife crime’, statistics are distorted by variations in the number of possession cases that fluctuate dramatically as police priorities change. However, given that we are concerned with violence a much better measurement is the number of hospital admissions for knife or other sharp object wounds. The graphic below shows that since 2013 we have (happily) seen a small decline. Certainly not a significant increase.

What Police infrastructure, police performance, and crime: Evidence from austerity cuts is attempting to do, is convince us that closing police stations increased crime (and reduced house prices). It does this through impenetrable mathematical formula which prove some correlation between some data. The IFS as an organisation, and this report in particular, sees the world from a particular perspective which assumes that police are a ‘good thing’. If you start with that assumption, it is unsurprising that reduced funding is perceived to be a bad thing. It is not difficult to select some evidence to support your desired conclusion.

One of my students’ favourite research papers was on Stork theory.

This uses the clear evidence of a positive correlation (in some areas) of birth rates to stork population to “prove” (statistically) that storks deliver babies! Of course, whatever the correlation, [Spoiler alert] storks do not deliver babies. However, equally crude conclusions are drawn by the IFS report. For example, having discovered that public confidence in the police was declining at the same time as the closure of police stations the report confidently declares that the closures are responsible for the declining confidence. A conclusion that is helped by totally ignoring the recent Casey review which highlighted the institutional sexism, racism and homophobia of London’s police. 

No doubt this dubious report will be used to justify increased police expenditure as we move into an intensification of austerity (and thus increase the cuts elsewhere). It is best understood as an ideological intervention of very limited scientific merit. The case for defunding the police in the UK has not been diminished by this report