Saturday, May 27, 2006

Beware the undeclared remainder

The current AADirections magazine has an interesting article on crime and crime reduction policies, however one quote from Rick McKee (strategic Advisor on Crime reduction with the Policing Development Group) caught my eye:

"About 10% of offenders are responsible for half of all crime, so if a road patrol turns over a known offender in a hot location, then 'alarm bells' go off. Even if we can't legally search the car, we have on record that this person was in this location at this time and, if it turns out later that crimes have been committed in the area we know who to start looking at"

While I applaud the efforts gone to for this type of approach, and also believe the logic behind it, what can be dangerous is the use of the above statistic to justify it. Mainly because the statistic probably is not for "all crime" as Rick probably believes, but more likely "all solved crimes". Which means that it can become both a self-fulfilling statistic, and an impediment to catching new offenders. While initially this will lead to more crimes being solved; it will lead to more crimes being solved where the offender is already known (and flagged as one of the "high risk" 10/20%) - meaning the 10% responsible for 50% of crime will trend upwards. If the statistic is watched and continued to be a justification in the use of these techniques then the mind set can increasingly move towards only concentrating on known offenders - which then means the statistic continues to trend upwards as less crimes with new offenders are caught... My main point here is actually nothing to do with crime - it is a warning against taking any statistic at face value.

The above stat has probably had no spin of any kind applied to it, and the person delivering it believes in what he is saying. However in interpreting a stat we must always ask what is the undeclared remainder of the original set - in this case the unsolved crimes - and how can that affect the interpretation and movement of the statistic. What makes this more difficult is that it needs to be realised that the undeclared remainder is frequently not known by the reporter of a statistic, and is as likely to be through a genuine simplification - rather than deliberate obfuscation.
Tags: , , ,

Ads by

No comments: