Thursday, 28 January 2010

Duking the stats, redux

A while back I had a bit of a go at Chris Grayling's claim that "under Labour violent crime had risen 70%", on the grounds that it was total nonsense. But I wasn't able to find out what the shaky basis for his claim was.

Happily Mark Easton has rather more clout with Conservative Central Office than I do* and so when he made the same challenge to David Cameron recently, they very kindly provided him with the figures they used.

They're appalling. Transparent horseshit. Do read the full analysis, but the headline version of how you turn a consistent fall in violent crime into a 70% rise is:
  • Compare figures that you have been explicitly told are not comparable
  • Cherry-pick your comparison years so that downward trends become upward leaps
  • Have the unmitigated gall to pretend this is a meaningful contribution to political debate
It's not even sophisticated statistical bullshit. It's out and out manipulation of the figures to argue that fewer stabbings are really more stabbings, fewer shootings are actually more shootings, less crime is more crime.

I'm sure Grayling and Cameron weren't the architects of this. But you have to wonder if, when they were told that the violent crime stats were exactly what they wanted them to be, they even asked themselves whether they ought to check that their researchers hadn't been a little too helpful. Or whether they really cared - it's not very likely they're about to change their policies or rhetoric to match the evidence.

*A low bar there, to be sure.

No comments: