British police are doing a pretty good job
Both sides in British politics have an interest in overstating the level of crime. Neither narrative is correct.
Written by Noah Carl.
Both sides in British politics have an interest in overstating the level of crime.
The left wants to blame out-of-control crime on years of Tory austerity that cut police budgets to the bone. Or they want to blame it on mistrust of the police that’s rooted in “systemic racism”. Meanwhile, the right wants to blame out-of-control crime on mass immigration or woke policing. “They spend so much time policing social media they haven’t got time to solve burglaries,” it’s often claimed.
Neither narrative is correct. Judging on the best available data, the British police are doing a pretty good job. Not a perfect job, of course, but a pretty good one. There’s no evidence that crime is spiralling out-of-control due to Tory austerity, systemic racism, mass immigration or woke policing.
As an aside, I’m completely against the policing of social media and the recording of “non-crime hate incidents” (except in obvious cases like incitement to violence). To the extent the police are engaged in those activities, they’re not doing a good job. Having said that, it’s not entirely clear who’s responsible – the police could be acting on guidance from the government. The College of Policing, which was taken to court for interfering with freedom of expression, is described as an “arms-length body” of the Home Office.
Naturally, I’m also against the pride flags on police cars, the dancing at pride parades, the kneeling for Black Lives Matter, and other completely inappropriate behaviour. Like many institutions, the police have become politicised during the Great Awokening, and this is surely detracting from their effectiveness. Nonetheless, my claim is that by-and-large they’re doing a good job at preventing and deterring serious crime.
The chart below shows two different measures of the total number of crimes in England and Wales from 1981 to 2016. The first measure, given by the blue line, is based on the Crime Survey of England and Wales – an annual survey of a representative sample of the population. The second measure, given by the blue, green and purple bars, is based on police recorded crime.
If you only looked at the bars at the bottom, you’d assume that crime in 2016 was barely lower than in the mid 1990s and was in fact trending upward. Yet looking at the line at the top, we can see that crime has fallen by more than half since the mid 1990s – and that’s before taking account of population growth. (The crime rate has fallen by an even larger amount.) Data for 2022 confirm that the total number of crimes has continued to fall.
But why should we trust the line and the not the bars? As the ONS repeatedly points out, police recorded crime data are not not suitable for making long-term comparisons. This is for the obvious reason that such data exclude crimes that are not reported or recorded by the police. If the number of unreported/unrecorded crimes changes over time (as indeed it does) the level of police recorded crime will not track the true level of crime.
Data from the Crime Survey of England and Wales, by contrast, are suitable for making long-term comparisons. These are based on an annual, nationally representative, face-to-face survey of more than 30,000 households – which has a high response rate of around 64%. Respondents are asked questions about their experience of criminal victimisation in the past year, which allows statisticians to derive estimates of the total number of incidents in the population. (The CSEW is equivalent to the NCVS in the US.)
For example, if 500 violent incidents are reported in the survey, then statisticians can infer that there were about 1,00,000 such incidents in the population. (It’s slightly more complicated than this, but you get the idea.) Due to the sample being large and approximately random, estimates derived from the survey tend to be quite precise (95% confidence intervals are small relative to the year-to-year variation). This means we can be confident about the direction and scale of trends.
As the ONS notes, the one area where police recorded crime data are superior to data from the CSEW is “lower-volume but higher-harm crimes”, notably homicide. Because homicides are rare in England and Wales (only 710 were recorded in 2022) you can’t obtain reliable estimates from a survey – even one with a sample size exceeding 30,000. If you asked each respondent, “Was anyone in your household the victim of a homicide this year”, the estimated number of homicides would fluctuate wildly from year-to-year. Hence it’s better to rely on police recorded homicide data.
Fortunately, homicides are almost invariably recorded by the police – for the simple reason that it’s hard to ignore a dead body. So we don’t need to worry about changes in recording practices skewing the figures up or down.
Okay then, the total number of crimes might be lower now than in the past, but what about the categories we really care about like violent crime, burglary and indeed homicide? They’re all lower too, as the chart below indicates. (Note that in this case I’ve calculated crime rates by dividing the figures by the population in the relevant year.)
The homicide rate in England and Wales peaked in 2003 and has since fallen by 35%. (It reached its nadir in 2015, before climbing back up over the next few years.) The rates of violence and burglary peaked in the mid 1990s and have since fallen by around 80%. Both crimes are now rarer than at any point since 1980.
There are two major exceptions to the trend of falling crime: knife crime and sexual offences. Assaults and robberies committed with a knife are now more common than they were in the mid 2010s. Likewise, rape and other sexual offences are more common now than they were in the early 2010s.
However, both of these categories are based on police recorded crime data, so are subject to the caveat above. As the ONS states, “High levels of non-reporting combined with changes in reporting trends can have a significant impact on sexual offences recorded by the police”. And in the case of knife crime, this may partly reflect substitution away from the use of firearms, which is down substantially since the mid 2000s. After all, what we really care about is violent offences, not violent offences with a particular type of weapon.
I have to admit that I was somewhat surprised when I looked into these figures. I’d assumed that crime was on the rise in England and Wales. Perhaps I spend too much time on Twitter, where people frequently post videos of gang members fighting with machetes – incidents that may not be particularly representative. (Before the smartphone era, few of us had a front-row seat to gang turf wars.)
It’s not impossible that the apparent fall in crime is due to some artefact of the data collection process. Perhaps survey respondents are more likely to lie about their experience of criminal victimisation than they were in the past? Though I can’t think of an obvious reason why. What’s more, the police recorded homicide data is consistent with a big drop in crime since the early 2000s. The data presented here are the best we’ve got and they show that crime is down.
Noah Carl is an Editor at Aporia Magazine.
Please consider supporting our work with a paid subscription and follow us on Twitter.
Imagine two countries: Syldavia and Borduria. In Syldavia, if your home lacks private security, it has a 1 in 3000 chance of being burgled; in Borduria it has a 1 in 3 chance. In Syldavia, if you wander the streets alone at night, you have a 1 in 3000 chance of being mugged; in Borduria you have a 1 in 3 chance. These statistics make behaviour very different in the two countries. In Syldavia, almost no one has private security for their home, and almost no one travels at night exclusively by registered taxi. In Borduria almost everyone does both. As a consequence, the actual crime rate is about the same in the two countries. In some years, Syldavia even has a higher crime rate than Borduria, because some Syldavians become too blasé about how wonderfully crime-free their society is.
What’s the point of my example? Even if the two countries have near-identical crime statistics, it’s obvious Borduria is the more fearful place to live. We’ve all heard the refrain that the fear of crime is worse than the likelihood of experiencing it. But this misses the point that, where people have a reasonable fear of crime, they modulate their behaviour to minimise risk. Syldavians learn to behave like Syldavians, and Bordurians like Bordurians. If a Bordurian behaves like a Syldavian in Borduria, he’ll likely be burgled and mugged. After that learning experience, he’ll change.
Detection rate? Something around 5% isn't it? Conviction rate for things like burglary? Somewhere approaching zero% isn't it? British justice - if defined as being on the side of the law-abiding and a fearsome deterrent to the criminally minded - is doing an appallingly bad job. It's difficult to imagine how it could be doing worse. https://grahamcunningham.substack.com/