Thursday, 20 June 2013

Pareto and Policing: The need for UNcommonsense

Last week's New Scientist reminds me that crime is not random. Criminals, victims and the times and places of crimes are all much more likely to follow a Pareto distribution ('20% of the villains commit 80% of the crimes') than a random distribution.

This is hardly a new idea. Police officers and criminologists know about habitual offenders and the likelihood of violence outside pubs on Saturday nights. So far so obvious. Yet police forces have been reluctant to apply these insights systematically. And the public still like to see patrols on their streets - even if their streets see little crime. (Which is, in any case, not actually reduced by random patrols.)

Just as some kinds of people are more likely to become criminals than others (men rather than women, young men rather than older ones just for a start) so some people and properties are more likely to become victims. It makes good sense to at least ensure that those concerned know of the enhanced risk and understand what they could do about it. For civil liberty reasons there should be no criticism of those who do not adjust their behaviour but many people will be only to happy to fit better locks and keep out of dangerous areas.

In fact the evidence about crime if often counter-intuitive. Most people believe that crime is rising and that primitive societies are more peaceful than ours. Neither is true. In the UK, for instance, violence, criminal damage and burglary have all fallen since 2006. Theft, exceptionally, fell until 2009 but then rose.

As humanists we should ask for evidence whenever people say silly things about crime.

Hmm - sounds like a lifetime job!