Novelty and interdisciplinarity in criminology: how data-drivenness connects both

With the discipline of criminology as a case study, I explore how papers using what can broadly be considered “data-driven methods" leverage novelty to experience citation benefits. The study of criminal justice is notable because it is a discipline in which applied and also controversial methods such as predictive policing and automated bail allotment have emerged in recent years (Brayne and Christin in Social Problems 68:608, 2020; Brayne (Predict and surveil: Data, discretion, and the future of policing. Oxford University Press, 2020). Many of these technologies rely on the combination of traditional criminological concepts with novel computational methods. I generate different measures of novelty, including the scientific journal and article content of references, to demonstrate that articles which use data-driven methods experience a particular advantage in impact measured by citation. I create estimations of article content using Top2Vec and compare novelty as measured by atypical combinations of references (Uzzi et al. in Science, 342(6157):468–472, 2013) and novelty as measured through sub-field integration (Moody in The Nonproliferation Review 3(3):92–97, 1996). To conclude, I discuss the larger implications that research in the era of “Big Data" has on the production of scientific knowledge, as well as how the discipline of criminal justice may or may not be disrupted by emerging machine learning-based technologies.