Predictive Analytics and Criminal Justice
A corporate press release last week boasted that the Florida Department of Juvenile Justice is using IBM “predictive analytics” to determine which juvenile offenders in custody are likely to re-offend. An offender’s rehabilitation program may depend upon the results of this analysis, which is a form of data mining to discover from data sets correlations that would otherwise be hidden. The software is a product created by SPSS, a company recently acquired by IBM.
Evidently, the U.K. Ministry of Justice is already using SPSS’s predictive analytics for the same purpose in connection with the population of adult offenders in custody. As well, the Richmond Virginia Police Department makes use of the software to assist in better deploying its forces and in identifying potentially serious crimes or offenders.
The SPSS program was initially developed for marketing to businesses as an aid to increase their “business intelligence.”
Am I wrong to be made nervous about this use of statistics? What of human judgment? what of the ability of individuals to break free of their statistical ‘inclinations’? what of mercy? Do Charter values require more individual attention to people who are to be condemned to more prison time? The language of Simon’s text, particularly the lead-off item, suggest very little human intervention in the result produced by the machine.
Will this kind of analysis lend itself to people arranging their behaviour to ‘beat’ the system (or would that just mean behaving well, which is what the system is trying to achieve)? I am remined of the appendix to The Organization Man in the 1950’s, “how to cheat on a personality test”. Will people learn to cheat on predictive analytics?