FactCheck: knife crime stats
Updated on 15 December 2008
Why has the statistics watchdog criticised the government for "misleading" and "selective" use of statistics?
The claim
"The number of teenagers admitted to hospital with knife wounds fell by 27 per cent in nine tackling knives action programme police force areas."
Tackling knives action programme - six months on, home office press release, 11 December 2008
The background
Selective use of statistics has been the preserve of politicians of all colours for far longer than FactCheck has been FactChecking - but last week saw an unusual criticism of government numbers on knife crime.
Fears about rising knife crime, particularly among young people, have dominated headlines this year.
Get the latest FactCheck news: sign up for the newsletter
In June, the government launched a £5mn anti-knife crime initiative in 10 police force areas including London, Merseyside and Greater Manchester.
The Tackling knives action programme included more searches, knife arches and wands, a crackdown on shops selling knives to teenagers, and more alternative activities for young people.
Last week, the home office pinged out a press release claiming the programme was already having a big impact, including fewer knife wound injuries, fewer young people carrying knives, and tougher sanctions for those caught carrying a knife.
The story was widely reported on the BBC on Thursday, and made many of the Friday's newspapers.
But on that Friday, the chairman of the official statistics watchdog hit out, writing to Downing Street to say the stats were premature, irregular and selective.
So are the claims worth the paper they're written on?
The analysis
Figures released as official "national statistics" are subject to a strict code of practice and should be of high integrity and quality. They're usually released to a regular schedule rather than timed to coincide with political events.
It's fairly common practice for national statistics to be released with an accompanying government press release highlighting selected aspects of the numbers.
But the full data set would also be presented so those who wish can go through and check the government's interpretation.
The knife crime figures the home office put out last week are not national statistics.
In this case, the press release was simply accompanied by a two-page "factsheet" containing a few bullet-pointed statistical claims based on data supplied to the government by the forces involved, rather than raw or detailed data.
Let's look at a few of these boasts.
The government claimed hospital admissions were 27 per cent lower for assault by a sharp object (including knives) for July to September 2008 in the nine TKAP areas in England compared to the same period in 2007.
It was this claim that particularly enraged Sir Michael Scholar. "These statistics were not due for publication for some time, and had not therefore been through the regular process of checking and quality assurance," he wrote to Downing Street.
It doesn't mean they're wrong - but it does mean that we can't say whether they are reliable.
The release also boasts of 17 per cent fewer serious knife crimes against young people in October 2008 than in June 2008 in the 10 areas.
This sounds good, but it's questionable how much this rather random comparison tells us. Statistics often do fluctuate month on month - this is why it's normal to look at a year-on-year comparison alongside any more recent analysis of data.
And the 17 per cent fall is really just a drop of 17 offences: from 98 offences to 81.
No serious knife offence is a trivial matter, least of all to those involved. But when dealing with such small numbers in a statistical sense, it's particularly important to be wary of fluctuations which don't necessarily show a wider trend.
The factsheet also claims there were 18 per cent fewer knife crime victims under 20 between April and September 2008, in comparison with the same period in 2007, and that youth violence in Halloween week was 30 per cent lower than the year before.
This sounds good - but how much does it have to do with the crackdown? The government's scheme only kicked off in June, halfway through the April-September period.
Damned lies and crime statistics?
Crime statistics are a notoriously slippery area.
The government often boasts about the increased number of police and the reduction in crime on its watch - but whether one causes the other is pretty tricky to prove.
Although the most recent figures showed a fairly hefty year-on-year fall in crime, the longer-term trend of reduction actually started back in 1992 or 1995, depending on how you want to count it. Good for the country in general, but not a victory for which Labour can claim all the credit.
So what do the official figures tell us about knife crime?
The government has only recently started collecting more detailed figures about crime involving knives across the country.
The much broader measure of violent crime across the country seems to have dropped fairly heftily, according to the most recent figures (which still pre-date the government's boasts).
The British Crime Survey - which measures people's experiences of crime (though not, significantly for this topic, under-16s) and is seen to be the most reliable way of assessing crime trends - showed a 12 per cent fall in violent crime between 2006-7 and 2007-8.
This is backed up by an 8 per cent fall in police-recorded violent crime during the same period.
London figures, which we dug into in detail before this spring's mayoral election, suggest knife-enabled crime has been falling since 2004, with a 13 per cent year on year decrease from 12,985 incidents in 2006 to 10,699 in 2007.
The verdict
It's not necessarily the case that the government's facts are wrong - more that they should be taken with a vast dose of caution.
If, as the government claims, more searches are being carried out each month but a smaller proportion are uncovering a weapon, that would generally seem to be a good thing.
But to claim, as the government does, that, for example, crime is down by 17 per cent based on the comparison of just one month with another doesn't take into account a wider picture of seasonal fluctuations.
And this is at a time when violent crime more broadly seems to be falling, anyway.
FactCheck rating: 4
How ratings work
Every time a FactCheck article is published we'll give it a rating from zero to five.
The lower end of the scale indicates that the claim in question largerly checks out, while the upper end of the scale suggests misrepresentation, exaggeration, a massaging of statistics and/or language.
In the unlikely event that we award a 5 out of 5, our factcheckers have concluded that the claim under examination has absolutely no basis in fact.
The sources
Home Office factsheet
Hospital admission statistics and knife crime: letter from Sir Michael Scholar
Sir Michael Scholar talks to Channel 4 News
Crime in England and Wales 2007/08
Your views
You've read the article, now have your say. We want to know your experiences and your views. We also want to know if there are any claims you want given the FactCheck treatment.
Email factcheck@channel4.com
FactCheck will correct significant errors in a timely manner. Readers should direct their enquiries to the editor at the email address above.
