Statistics, Media, and Sentencing Policy.

A short while ago at MK Skeptics in the Pub we had a very interesting talk by Professor Kevin McConway on Statistics in the Media. To show that I was listening and learned something, here is a short reflection on a headline in of today's papers. I think it was the Mail, but it might have been another paper that majors on exploiting ignorance and prejudice.

The headline was "20,000 spared jail reoffend" prompting the knee-jerk response "terrible, far too many, lock more people up and don't give them a chance to reoffend."

The headline, of course, tells you nothing. How many offenders were given non-custodial sentences? What is the proportion reoffending? What if 200,000 offenders got non-custodial sentences and 20,000 reoffended? Well, in that case 90% did not reoffend, which surely would qualify as a success. What if all 200,000 had been locked up at great expense to the community? And what if the 20,000 offences were trivial? Where is the benefit in that? The headline provokes questions. It does not, by itself, give an answer or an indication of whether non-custodial sentences are successful or not. And above all, it does not tell you that the reoffending rate among prisoners is as high as 74% in one UK prison and that the comparative rate for non-custodial sentences compared to short terms of imprisonment (therefore hopefully for similar original offences) is lower, not higher, no matter whether 20,000 seems to be a large number or not. (Figures from a 2010 article in The Guardian, http://www.guardian.co.uk/uk/2010/nov/04/jail-less-effective-community-service)