Rebalancing the Algorithm: Pathways for Mitigating Bias in Predictive Policing

By Lata Nautiyal, Preeti Malik, Varsha Mittal, and Baraa Zieni. Published in Service Oriented Computing and Applications, July 17, 2026.

Summary: An empirical study using the Chicago Crime Dataset finds that predictive policing models disproportionately flag minority and historically over-policed neighborhoods as high-risk. Applying data re-weighting, counterfactual analysis, and algorithm auditing, the authors show the models are learning patterns of policing activity rather than actual crime distribution.

Why this matters: Directly relevant to the Crime, Policing & Carceral Systems subdomain — this study shows a concrete mechanism by which “objective” risk-scoring tools launder pre-existing enforcement bias into future predictions, with clear implications for social workers engaging with justice-involved clients and communities.

Key findings: “Enforcement intensity rather than underlying crime incidence” drives biased predictions. The authors argue for transparency, fairness-aware modeling, and systematic auditing throughout development and deployment.

Read the original: https://link.springer.com/article/10.1007/s11761-026-00501-1

DOI: 10.1007/s11761-026-00501-1