Saturday, February 06, 2010

Public Trust In Official Statistics

All over the place: reporting methods have changed twice since 1997

As you will know, shadow Home Secretary Chris Grayling has been rapped over the knuckles for misusing official statistics by the head of the UK Statistics Authority, Sir Michael Scholar.

Sir Michael wrote to Grayling:
"I must take issue with what you said yesterday about violent crime statistics, which seems to me likely to damage public trust in official statistics."
Grayling's offence was to claim that official stats show violent crime has been increasing under Labour. In reality they don't show that - or indeed anything - because of changes in recording methodology since Labour came to power.

Now, as we've blogged before, we have the greatest respect for Sir Michael. He's one of our few remaining mandarins of the old school: those Oxbridge men of myth and legend who believe duty to country always trumps duty to some here-today-gone-tomorrow politico. So if he raps someone, we must accept it's a fair cop.

Which is why the Prog Con is so jubilant. Alongside Rev Easton's sixpennyworth, the Independent whoops:

"The humiliating slapdown adds to doubts about how credible the Conservatives are as a potential government, when they appear to be only three months away from taking office."
Strange we can't recall them making any comparable point when Sir Michael read the riot act over Jacqui's shocking abuse of knife crime stats (see this blog).

But whatever the Prog Con says, the Big Truth here is that for judging long-term crime trends the UK's crime stats are next to useless. We've blogged this many times, but the key points are:

  1. Recorded crime statistics - supposedly the hard measure of crime that is serious enough to report to the police - have been through two changes of counting/recording methodology since 1997; Labour's criminal record is therefore conveniently masked (the latest Home Office report is here).
  2. Recorded crime stats come from the police who are heavily conflicted. Under Labour's tractor production regime, police performance is judged against targets based on these very stats, so naturally they are heavily gamed and distorted (eg see this blog, including Inspector Gadget's account of how "threatening behaviour" - a recorded crime of violence - gets downgraded to "drunk and disorderly" which does not show up in government stats; and see this blog for Her Majesty’s Chief Inspector of Constabulary confirming that a third of violent crimes don't get recorded, probably in order to meet government targets).
  3. The government's own preferred measure of crime - the British Crime Survey - is no more than an opinion poll on people's perceptions and experience of crime (how you feeling now love?). It is entirely voluntary, and it suffers from all the usual flakiness of such polls. When we looked at it, we found it claiming to have a 75% overall response rate, but much lower in high crime urban areas (62% claimed). It reckons to have a +/-3.5% statistical confidence interval around its estimate of total crime (an uncertainty margin that rarely gets mentioned by ministers). But that is not the sum total of the uncertainty, because we have no idea whether people with high experience of crime are more or less likely to respond. In Tyler's humble opinion, as a measure of actual crime it ain't worth the paper it's expensively printed on. 
  4. None of these stats are produced by the independent Office for National Statistics, but outrageously are still under the control of the Home Office - ie the very people who are most interested in persuading us that crime is falling. Sure, they now say they adhere to the National Statistics Code of Practice, but come on.
The fact is that the official crime statistics do not need Chris Grayling to damage public trust in them - they are doing a great job all by themselves.

All of which leaves us with a problem. Because if the official measures of crime are useless, how do we tell if Labour's record is good or bad?

Yup, we have to conduct our own crime counts. And here's Tyler's.

Since Labour came to power, Tyler Towers has suffered two assaults from burglars - including this one - against none previously. The houses on both sides have suffered as well. In the next road there has been an apparent contract killing. Prior to 1997 there were no such events.

One of Tyler's sons has been decked by a thuggish Balkanese (?) minicab driver, against no such deckings pre-1997.

Tyler's brother and his family have been subjected to an armed robbery in their own home, against no such robberies pre-1997.

From where we sit, the record looks all too clear.

No comments:

Post a Comment