Tory crime statistics row deepens
Following revelations on my blog last week, shadow home secretary Chris Grayling has defended the Conservative Party's use of figures suggesting "significant" rises in violent crime - even though the official source statistics warn they should not be used in that way.
In fact, the party has now used the data in a release sent to every constituency in England and Wales - statistics which one police force has described as "extremely misleading" and which appear to have been labelled "inaccurate" by some senior Conservatives.
Mr Grayling argues he has done nothing wrong and has accused the government of using statistics in exactly the same way to suit their purposes.
Official statisticians say it is not possible to compare violent crime stats before and after April 2002 because of a big change in the way such incidents were recorded.
Before 2002 the decision as to whether an incident was a violent crime had been taken by police. After 2002, officers were obliged to record all incidents as violent crimes if the alleged victim said that is what it was. The aim was to stop police fiddling the figures and to get a better picture of violence. The obvious consequence was to send the raw numbers shooting up.
A warning printed on recorded crime stats after April 2002 says that "figures before and after that date are not directly comparable".
Now former Tory leader Iain Duncan Smith has written to me to say he agrees that such comparisons are profoundly misleading.
Mr Duncan Smith founded the policy think-tank "Centre for Social Justice" (CSJ) which includes on its board party heavyweights William Hague, Oliver Letwin and David Willetts.
"(T)he CSJ has long understood the inaccuracy of directly comparing present crime levels with those published before the National Crime Recording Standard (NCRS) was introduced - which as you note changed recording methods significantly and has rendered direct numerical comparisons with pre 2002/03 levels inaccurate."
Despite such anxieties, Conservative HQ has done exactly that and produced a list of "direct numerical comparisons" which have been punted to every candidate in England and Wales. The full release goes on for pages, but this extract gives you an idea of how they compare 2008/09 figures with stats from 1999/00.
Some Tory constituencies have started using the dodgy data - with predictable results.
In Milton Keynes, local MP Mark Lancaster's office put out a statement last week claiming that there were 6,015 "violent attacks" in the town last year, reflecting a 236% increase over the past decade.
The figures were given to Mr Lancaster by Conservative Central Office:
Milton Keynes sounds like Dodge City. That's a "violent attack" every 90 minutes.
However, according to the Milton Keynes Citizen newspaper, the town's police says that the statistics are "extremely misleading".
Local commander Nikki Ross told the local paper that the figure includes "everything from public order offences, to harassment, to allowing a dog to be out of control in a public place".
"The actual number of people who were victims of serious violence was 81," she said.
The point here is that the phrase "violent attacks" does not equate directly with the crime category "violence against the person". For instance, if someone swears at you in the street and you complain to the police about it, that incident goes down as an act of "violence against the person".
There was a similar reaction from police in Colchester. The Daily Gazette in Essex asked whether "shock north Essex crime figures add up". The story reports how:
"research, released by the Conservative Party, compares the number of recorded violent crimes against the person for 1999-2000 with 2008-09. They reveal the figure has risen dramatically, from 789 a decade ago, to 2,578 last year, an increase of 227 per cent."
However, the report goes on, "the research has been disputed by the police, as they compare periods in which crime data was recorded differently".
The Association of Chief Police Officers (Acpo) explained to me how it was they who pushed through the 2002 change in recording violence "to better reflect demands made of the police". Acpo knew that "the move would result in more crime being recorded, regardless of underlying trends".
That is why the warning on not making comparisons is so important.
"This caveat is stressed within official crime statistics to ensure that the public have the chance to reach balanced and informed conclusions on matters which are important to individuals and communities."
The architect of the Conservative strategy on crime is the shadow home secretary Chris Grayling, so I asked him about the use of comparative data. Did he think it was reasonable to employ them in this way?
"What else do we use?" Mr Grayling asked me, arguing that the figures were still being used by the Home Office and, therefore, it was legitimate for him to make comparisons across a decade.
He sent me three examples where, he argues, the government itself has compared violence statistics from before and after 2002.
Robberies peaked in 2001/02 at 121,359. The figure for 2007/08 is 30% lower than this peak.
Example 2: The Home Office crime statistics document for 2006/7 includes the line [1.78MB PDF]: "The number of most serious violence against the person offences increased by 21 per cent between 1998/99 to 2006/07, a much smaller rise than that of 111 per cent in other VAP (violence against the person) over the same period".
However, Mr Grayling did not send me the preceding paragraph which includes the warning: "Recorded VAP has more than doubled in the eight years between 1998/99 and 2006/07.
"Nearly one third of this increase occurred between 2001/02 and 2002/03, and much of this can be attributed to increases in recording of violent crime following the introduction of the National Crime Recording Standard (NCRS) in April 2002."
Example 3: A Home Office crime fact sheet suggests that recorded robbery in 2006/7 was "16% below the 2001/02 peak in robbery" [841KB PDF] although, again, Mr Grayling did not refer to the preceding paragraph which warns that "(t)rends in police figures are affected by changes in recording practices, coverage, public reporting and police activity".
At first sight, it would appear that the Home Office have made comparisons across the change in recording systems when it has suited them to do so - albeit with warnings posted elsewhere in their documents.
They would also argue that figures for the most serious violence are far less affected by the recording change than minor incidents.
There is another argument that Mr Grayling makes. He echoes the view of Mr Duncan Smith who, in his email to me, points out that people in the most deprived neighbourhoods are more reluctant to report violence to pollsters or police.
"Due to a lack of confidence in our largely dysfunctional justice system, there is a lower probability of people reporting crime in very deprived areas, and that within these areas people are also less likely to respond to victimisation surveys - yet we know from research that these communities are most impacted by crime and its contribution to social breakdown."
However, this problem is not new and is, therefore, not going to have a significant effect on the trend data. The only way it can be used to argue that violent crime is rising when BCS data suggests it is falling is if one believes that violence is increasing massively in poor areas but victims in those areas are revealing it less.
That there is more crime out there than either the British Crime Survey or police records identify is to state the obvious.
There are roughly 4.6 million crimes recorded by police in England and Wales each year. The British Crime Survey identifies about 10 million crimes. The Home Office recently suggested the annual number of offences was closer to 60 million crimes. And research for Downing Street in 2000 estimated the number of individual crimes at around 130 million a year.
Lots of violence goes unreported or unidentified - domestic violence and child abuse are common-place but won't necessarily show up in police records or survey data. The question, though, is whether violence is getting better or worse. Many people feel it is getting worse, but the statistical evidence suggests it is getting better.
There are, perhaps, three reasons to regret that the Conservatives are publishing the violent crime comparisons in this way:
• The comparisons are inaccurate
• The misleading figures will unnecessarily alarm people
• It makes it more difficult to identify solutions
Crime figures are notoriously misused by politicians of all flavours. Readers may recall my revelations on the shocking use of knife-crime figures by the government in a so-called "fact sheet" in December 2008 (numbers misrepresented to try and convince people that Home Office action on knives had been effective) despite the National Statistician pleading with ministers not to release the inaccurate data.
The selective use and spinning of crime statistics by politicians over decades helps explain why many people have lost complete faith in the data. Trust will not be restored by an election campaign in which official numbers are routinely abused.
Update 0900, 3 February 2010: Chris Grayling was on the Today programme defending his party's use of violent crime figures.