Anna Maria Barry-Jester, Ben Casselman and Dana Goldstein, have written a thought-provoking piece for FiveThirtyEight on sentencing considerations in Should Prison Sentences Be Based On Crimes That Haven’t Been Committed Yet? The authors question whether selected assessment factors reliably and fairly predict an offender’s future behavior. Pressures on prison systems and social reform as a whole, are moving states like Pennsylvania to incorporate risk assessment results as part of the information available to, but not binding on, judges in sentencing dispositions. Critics point out that a system that uses questionnaires as part of its risk-assessment process is prone to subjective interpretations and, further, that some of the factors considered and evaluated to predict future behavior are riddled with racial and socio-economic bias.
While a number of study outcomes on the integration of risk assessment tools are positive, some have met criticisms and are, therefore, inconclusive. In addition, in certain cases at least, the “technology” employed to amass such critical bits of life-impacting information seems surprisingly archaic (e.g, questionnaire-based assessments). In a world that more and more consistently relies on “smart” algorithms, the idea of less advanced measurement tools seems to strengthen critics’ arguments. As noted by a University of Pennsylvania statistician, who happens to have developed a more up-to-date method, “the most widely used tools are a generation behind a lot of the developments that are going on in computer science and statistics.” However, irrespective of the technology, when it comes to statistical generalizations stemming from risk assessments, Barry-Jester et al., in conclusion, ask one of the bigger questions, whether “it is fair to look at the behavior of a group when deciding the fate of an individual?”