Predictive Policing, Bias, and Community Legitimacy: A Social Justice Evaluation Framework

Abstract

Predictive policing has been gaining ground as a data driven process for crime prevention, resource allocation and public safety management. But concerns persist about algorithmic inequity, lack of uniform oversight, and the racialization of risk stratification and diminishing confidence in police institutions. The study assesses predictive policing in a social justice lens, where the technical aspects of predictive policing are analyzed and linked to notions of fairness, accountability, transparency and community legitimacy. The analysis is the potential for predictive policing systems to reinforce policing disparities from the past when developed based on biased crime data, arrest data, or policing locations. The results indicate that the use of predictive policing can be effective in making operational decisions and in the identification of hotspots, but with low levels of fairness safeguards, community oversight and explainability mechanisms. The findings also suggest that communities where algorithms are used for high levels of surveillance tend to experience lower levels of trust, procedural justice, and police cooperation. The framework for evaluation is proposed, revealing that bias auditing, participatory governance, transparent model reporting, and ongoing testing and monitoring of social outcomes must be added to predictive accuracy to facilitate ethical deployment of policing technologies. This paper contends that predictive policing should be assessed as a justice-sensitive governance practice, in addition to a technological innovation. A socially legitimate predictive policing system should help to minimise harm, ensure the safety of vulnerable communities, and enhance democratic accountability whilst achieving public safety goals.