Wednesday, July 23, 2008
These Foolish Things
Picking up our disbelief about those "falling crime stats", Mr Fink tells us:
"Tougher prison sentences and tougher policing are working. Denying the facts only makes the Right look stupid...
...Why? Because the British Crime Survey is already, very precisely, a survey of what we “know” about crime. It is a statistically valid review of our experiences. That is the whole point of it. The rest is just anecdote. Are we seriously supposed to measure what is happening to crime by counting the column inches devoted to terrible stories about victims?
Crime is falling. Sorry, but it just is"
Interesting. Can he really think the BCS opinion poll is a "statistically valid review of our experiences"? Even though it's put together by Home Office officials who have an all too obvious axe to grind? And even though pretty well all high profile government figures are now routinely spun and distorted? And even though the government ignored the specific recommendation of the Statistics Commission that the BCS should be taken away from the Home Office and switched across to the ONS?
Fink seems a very nice man, but as George W is wont to say, "you can fool some of the people all the time, and those are the ones you want to concentrate on."
As it happens, there's another interesting article today on numbers and government targets. The excellent John Kay writing in the FT says:
"Performance in any but the simplest tasks has many dimensions. Focusing on a small number of these dimensions as targets directs attention on these at the expense of others of equal importance. The difficulty is compounded when there is a shared interest on the part of target setter and target recipient in asserting that the goal has been met.
The story of the Soviet factory that achieved a quota based on weight of output by manufacturing a single gigantic nail is no doubt apocryphal. But the story that ambulance crews that were set a target of meeting emergency calls within eight minutes met a large proportion of calls in seven minutes and 50 seconds is true. We do not know how far figures were doctored or the workload prioritised to enable the target to be met. All we know is that the reported statistics ceased to give useful information."
Now, that's the world we actually live in.
To spell it out, the reason we discount the BCS is that it's being used as a target scorecard, and it's compiled by the very same people whose performance is being targeted. It's a single gigantic nail that no longer impresses anyone - or at least, nobody I've been able to discover who's sampled the crime scene round our way.
PS Danny F quite rightly discounts the recorded crime stats. Not only have the counting rules changed twice in a decade, data collection is now seriously distorted by the police forces making the returns. For chapter and verse on the scams involved see Inspector Gadget. Yet another example of the gigantic nail syndrome.