Letter: Reconviction: scientific methodology vs empirical experience

Click to follow
The Independent Online
From Professor J. B. Copas

Sir: Your article "Home Office's calculated crime doesn't add up" (24 July), and the comment on your leader page, add up to a cynical and unwarranted distortion of what the Offender Group Reconviction Scale does and why it is useful.

The scale is a concise and accurate description of what has been observed in the past, nothing more and nothing less. It is no surprise to know that, other things being equal, older offenders tend to reoffend less often than young offenders, that men tend to re-offend more often than women, and that those convicted of burglary or car theft tend to re-offend more often than those convicted of fraud or forgery. But by how much, and how should these indicators of the likelihood of re-offending be combined? The score takes a small number of such indicators, chosen to be those most indicative of re-offending, and puts them together into a single formula. The formula is based on an extensive statistical analysis of the records of over 10,000 recent offenders. To dismiss the scale as nonsense is to say that the rate of re-offending observed in the past amongst like individuals is of no value in indicating what might happen in the future. And, in fact, there is nothing new in the idea of reconviction scales: a number of area probation services already have their own local scales.

The scale is a probability and does not say whether any particular individual will or will not re-offend. It is certainly not a prescription for sentencing. It does not replace judgement, but is just one of the items of information available to probation officers. It is neither a "magic formula" nor "a fiendishly complicated sum".

Our work (in collaboration with my research student Chris Stride) is a serious study addressing a serious social issue. To describe it as "pseudoscientific" is demeaning to those of your readers who do not share the level of numeracy all too evident in your leader writer. The fact that complex data on offending can be summarised in a simple mathematical formula is the strength and not the weakness of the scientific method: risk to the public is too important for scientific tools to be discarded. People are welcome to dismiss it if they will, but they should do so by rational argument and not by ridicule.


John Copas

Professor of Statistics

University of Warwick


26 July