Bridging the gap between PETs in books and PETs in action: privacy-enhancing techniques for criminological research

Abstract

The Big data turn and the increasing amount of information in criminological research and practice puts emerging privacy risks on the agenda. Especially in a field where sensitive data are not rare and data sharing a common practice, privacy is a major concern. However, increasing the level of privacy risks reducing the data's utility as little would be left to “learn.” Hence, the objective is to analyze the impact of applying privacy‐enhancing techniques (PETs) on the utility of data. To assess the utility level of privacy‐enhanced data, a sample of sickness absences across the entire Belgian police organization was applied in a pre‐experimental design. In the first phase, descriptive, predictive analyses, and machine learning models were conducted using the original data. In the second phase, the dataset was anonymized with state‐of‐the‐art PETs after which the same analyses were applied. We have found that PETs can be applied with a negligible cost on the data's research utility, indicating there is room for privacy gains in criminological science. Taking into account the size of the dataset, the research objectives, and the sensitivity of the data, PETs can be applied on a case‐to‐case basis and support quantitative research in criminology.