Stats watchdog barks back
The head of the UK Statistics Authority Sir Michael Scholar has again written to the shadow home secretary Chris Grayling. This time the data watchdog tells him that his "they are all it" defence for using misleading crime stats does not stand up.
Mr Grayling replied to the UKSA's rebuke in an unapologetic letter (see below) which simply argued that "a wide range of organisations" have used crime statistics "in the same way as the Conservative Party".
You may recall how this blog revealed that the Conservatives were claiming big rises in violent crime since Labour came to power by comparing two years in which the figures had been counted in very different ways. A warning on the statistics advises that the data are "not directly comparable".
Mr Grayling's letter attaches examples of what he claims are organisations including the Home Office, the Liberal Democrats and the BBC making exactly the same kind of comparisons.
However, Sir Michael simply says that "(w)ith regard to the examples cited in the attachment to your letter, these largely pre-date the creation of the Statistics Authority and/or they sufficiently, in our view, qualify the figures compared."
Mr Grayling's first example is from a Home Office document entitled British Crime Survey: Measuring crime for 25 years, 2007 [2.75MB PDF].
This seems a curious example since the third sentence on the page reads:
Earlier, the document explains why Mr Grayling should look to the British Crime Survey for crime trends, rather than non-comparable recorded stats.
Mr Grayling's second example is from the Home Office crime statistics in 2006/7 and concerns robbery [844KB PDF]:
A few pages earlier in this document there is, once again, a very clear warning as to why it is unwise to use police recorded figures to reveal trends in violent crime.
Mr Grayling's third example is from a Home Office booklet published a few years ago [1.78MB PDF].
The shadow home secretary draws Sir Michael's attention to the second bullet point which states that recorded violence fell 1% the previous year - the first such fall for eight years (ie since before the recording change).
However, Mr Grayling does not mention the warning on the previous page:
He also claims the BBC has compared violence figures before and after the recording change in 2002.
However, the type below the graph seems to contradict Mr Grayling's assertion, spelling out that the "statistics appear to indicate a year-on-year increase" but adding that "this coincides with a comprehensive overhaul of the way the figures are counted".
Finally, Mr Grayling quotes senior Liberal Democrat MPs in the House of Commons from 2007 and early 2008 claiming that violent crime has doubled since Labour came to power.
He may have a point that the claim is not backed up by the data, but the Liberal Democrats were not suggesting - as he does - that the claim of a "significant rise" in violent crime is supported by statistical evidence.
Mr Grayling has told the media that he will "take account" of the UKSA's concerns over the 2002/3 recording change but will continue to make the comparisons - with a caveat where necessary.
The question is what this caveat might say. Since Sir Michael's point is that the comparisons the Conservatives have been using should not be made because they are potentially misleading, it will be interesting to see how the party words the warning.
I'm 
Comments
Sign in or register to comment.