Hidden Under the ROC. A Critique of Dangerousness Assessment Models Evaluation

Abstract

Contemporary sociology of punishment has identified a shift in penal policies during the last quarter of the 20th century, part of which is a transformation of punishment practices from an “old penology” based on the rehabilitation ideal to a new technocratic and managerial penology challenging this ideal (Feeley and Simon, 1992). This paper hypothesizes that it would have been difficult for violence risk assessment to know its current success in the Western world had it not proceeded to an esoterization of empirical research. Esoterization of recidivism research is a process of epistemological complexification that obfuscated the two major difficulties faced by academics during the larger part of the 20th century. This process has apparently solved the two main types of critique faced by violence risk evaluation research: (1) the inconsistency critique – results concerning the “good predictors” of recidivism are too inconsistent to be ready for use by an administration; and (2) the undecidability critique – scientists cannot neutrally “decide” what is the best dangerousness assessment tool, as it is primarily an political and ethical choice between liberty and security. Drawing on a recent meta-analysis and on an original demonstration,I show that instead of answering the two fundamental critiques, contemporary methods have hid and displaced them. Overall, this paper intends to contribute to the understanding of the shifts within scientific research that allowed for a new penology to compete with the old ideals of rehabilitation, and concludes in trying to draw a parallel between the newfound strangeness of an ethical choice determined by “new” radar technology methods and the classical strangeness of “older” truth-producing methods such as ordeals, arithmetical proofs, and clinical psychiatric expertise.