Criminal justice agencies around the world collect and organise data related to the assessment of risk, increasingly relying on big-data analysis using predictive algorithms. In this chapter, we argue that this new system of knowledge production challenges comparative criminology, particularly at a moment when a comparative perspective is crucial. The transition from data defined by social science theory to a stream of independent variables under algorithmic analysis undermines the ability to compare units of analysis. By comparing the goals of modern comparative criminology (characterised by the commensurating approach to comparison) with the epistemic change of algorithmic knowledge, we argue that there are three main points of friction: commonalities are replaced with idiosyncrasies; the loss of generalisability and policy reform as possible goals; and defiance of parsimonious comparative models. This transition from positivist knowledge to prediction viability requires recalibrating how we make inferences from comparison.
|Title of host publication||Research Handbook of Comparative Criminal Justice|
|Editors||David Nelken, Francis Pakes, Claire Hamilton|
|Publisher||Edward Elgar Publishing|
|State||Published - 2022|