Watchdog: Grayling 'likely to damage' trust in statistics
I can reveal that shadow Home Secretary Chris Grayling has just been sent a sharp letter from telling him that his use of figures in violent crime is "likely to damage public trust in official statistics".
Chairman of the UK Statistics Authority Sir Michael Scholar says he does "not wish to become involved in political controversy" but writes that he "must take issue" with what Mr Grayling "said yesterday about violent crime statistics", referring to .
Here is the letter in full:
The authority's concern follows this blog's observations (here and here) that the Conservative Party has been using non-comparable stats to suggest big rises in violent crime.
In notes attached to the letter from Sir Michael (see below), the authority says: "we regard a comparison, without qualification, of police recorded statistics between the late 1990s and 2008/09 as likely to mislead the public".
Instead, the UKSA advises the shadow Home Secretary that: "The British Crime Survey (BCS) provides a more reliable measure of the national trend in violent crime. This is because the BCS is not affected by changes in reporting, police recording and local policing activity, and has been measuring crime in a consistent way since the survey began in 1981."
Mr Grayling , however, that he regarded the BCS as "fundamentally flawed":
"There are certainly changes in the recording methods, but the point is that they are the only comparators available. They are published by the Home Office... as an opposition party, we don't make the statistics. We can only use what the Home Office publishes."
The BCS suggests there has been a 49% fall in violent crime since 1995 - an unhelpful conclusion for any politician keen to demonstrate that violence has "risen significantly" under Labour.
A spokesperson in Mr Grayling's office did not appear to know about the letter, but I will report the response as soon as I have it.
Update 1242: In the meantime, below are some links to related posts at this blog from the past couple of years.
The shadow home secretary was, in this case, defending his decision to issue violent crime figures to every Conservative candidate in England and Wales which were based on a comparison between figures for 1999-00 and 2008-09. The source statistics include a warning that a change in recording practice in April 2002 means that "figures before and after that date are not directly comparable".
The UK Statistics Authority has, in the past, issued criticism of comparable uses of statistics by the government. Sir Michael Scholar wrote to Downing Street in December 2008 to say that the use of potentially "misleading and inaccurate" statistics on knife crime was "corrosive of public trust in official statistics and incompatible with the high standards which we are all seeking to establish".
In June last year, Sir Michael wrote to Equalities Minister Harriet Harman saying that her use of statistics "may undermine public trust in official statistics" and "risks giving a misleading quantification of the gender pay gap".
The UKSA code on the use of national statistics is legally binding on Whitehall departments, but the authority is keen to emphasise its role in policing the use of official data by those operating outside government.
Update 1322: A Conservative "rebuttal note" issued by Mr Grayling's office this afternoon argues that "(n)ot only is it commonplace and totally acceptable to make these comparisons on the basis of published Home Office recorded crime figures, to do anything else would be irresponsible".
The note dismisses the British Crime Survey (BCS) as "essentially an opinion survey" and "highly flawed", arguing that "recorded crime statistics... demonstrate trends in violent crime because these are real crimes reported to real police stations".
Mr Grayling regards the recorded crime figures as "the best available measure of the amount of crime that occurs in society", although statisticians might point out that far more violent incidents are identified in the BCS. The survey reflects victim accounts of more than two million crimes of violence, as compared with around 900,000 incidents being recorded by police.
I have written before about the weaknesses within the British Crime Survey. As Mr Grayling points out, it did not until recently interview under-16-year-olds about crimes they may have suffered; nor does it include crimes such as murder and manslaughter. Its methodology means it does not cover the population living in group residences like care homes, and it does not include crime against commercial or public sector bodies.
But while it may be far from a complete picture of crime, it is a fuller picture than the recorded statistics offer. And the point about the survey is that it is designed to provide information about crime trends. With violent crime, the trend has been downwards or flat for 15 years. Only if you believe that there has been an equivalent - or rather, a greater - increase in crime suffered by people within the categories not covered by the survey could it be argued that the trend is the reverse.
The difficulty for Mr Grayling at this point is that his election strategy on crime is based on the use of figures which the independent watchdog on official statistics says he should not use.
Update 1556: Chris Grayling has now responded to the UKSA letter, saying that he "will take account of the request by the Statistics Authority, particularly with regard to the changes to recording practices made in 2002-3".
However, the shadow Home Secretary stresses that he will "continue to use recorded crime statistics, because they reflect an important reality; that the number of violent crimes reported to police stations, and particularly serious violent crimes, has increased substantially over the past decade, even taking into account any changes to data collection".
I have asked for further explanation of what this means.
Does it mean he will withdraw the figures he sent out to constituencies which the UKSA believes are "likely to damage public trust in official statistics"?
Does he still regard it as "irresponsible" to use anything other than the recorded statistics to look at trends?
And, given the UKSA's note saying the British Crime Survey is "a more reliable measure of the national trend in violent crime", does he still dismiss the BCS as "essentially an opinion survey" and "highly flawed"?
In response, his office simply said: "you have got the line".
I'm 
Comments
or to comment.