Monday, 29 November 2010

Probation by Numbers

Just as the Ministry of Justice launches a consultation on making criminal justice statistics more transparent, it triggered several thoughts in me about numbers generally. Basically old-style probation officers don't like them, not least because of the difficulty of making people fit neatly into measuring boxes, like OASys and OGRS. Readers will be aware of my contempt for the former, but as regards the latter, I have to say I've never really understood the argument for the Offender Group Reconviction Scale.

Although refreshingly short in comparison to 'War and Peace' OASys, I've nevertheless found the process tedious and pointless when having to use it, especially at court when preparing FDR's. It's supposed to predict re-offending, but only measures static factors such as age, gender and criminal history, so does it really tell us anything we didn't know already? Along with dreaded OASys, it supposedly helps to categorise people neatly into one of four risk tiers, but we all know management manipulate that process in order to help massage the workload figures. 

I'm pretty sure magistrates and judges find the predictive risk section of PSR's confusing, irritating and dare I say, useless. I remember reading a report written a couple of years ago by a very well known probation officer on an elderly recidivist. Helpfully the risk of reconviction was stated to be something like 96%, a figure that the author felt in his professional opinion "was almost certainly an under-estimate."    

Of course figures and especially all this predictive stuff are the domain of the academic and I was interested to read an article on this topic written by Leigh Brauman and concerning Prof Lawrence Sherman in the latest issue of the Cambridge Alumni Magazine. Now to avoid any speculation I must hasten to say I have no connection with such an august institution, but happily blogging seems to attract information from all sorts of quarters. Anyway, this extremely eminent person who holds the prestigeous Wolfson Chair of Criminology, according to the article:- 

"........ is developing a general theory of crime, harm and criminal justice - a theory of how to use the criminal justice system to reduce total harm to society. Employing a 'crime-harm index' (CHI), the theory seeks to focus justice not on 'how to punish criminals' but on producing less harm to society at large."

He is about to embark on research with 10 police forces to test his hypothesis that society should focus on harm reduction, rather than punishment. Apparently one group of offenders will be 'prosecuted' fully, whilst the second will be processed using this predictive tool. Now I'm no lawyer, but I can already see some scope for an appeal here and wonder just how ethical this research is? The method seems to rely on measuring 49 variables and a computer with a brain the size of a planet in order to tell judges and magistrates which people should go to prison and for how long. The professor appears to be scathing of current sentencing practice using mere 'guidelines' and acknwledges that:-

"Social justice reformers don't like this approach because they are ideology-driven, and one of their ideologies is that numbers are bad. But that's not my problem."

Well I'm going to stick my neck out and say that maybe the professor has spoken a little prematurely and he has possibly under-estimated firstly the ability of finding an adequate IT system capable of performing these tasks (OASys regularly crashes) and secondly that anyone, especially sentencers, intend to take a blind bit of notice. People have that irritating habit of just not being willing to fit neatly into boxes, no matter how hard scientific endeavour tries to force it upon them. Thank goodness. 


1 comment: