Latent Racial Bias -- Evaluating Racism in Police Stop-and-Searches

TL;DR

This paper introduces a new metric and method to evaluate latent racial bias in police stop-and-search events, using probabilistic models on UK data to quantify and predict racial biases.

Contribution

It proposes a novel probabilistic graphical model and inference techniques to measure and analyze racial bias in police stops, including bias prediction and application examples.

Findings

Identified racial bias patterns in UK stop-and-search data

Developed probabilistic models to quantify bias levels

Predicted biases in police responses based on racial groups

Abstract

In this paper, we introduce the latent racial bias, a metric and method to evaluate the racial bias within specific events. For the purpose of this paper we explore the British Home Office dataset of stop-and-search incidents. We explore the racial bias in the choice of targets, using a number of statistical models such as graphical probabilistic and TrueSkill Ranking. Firstly, we propose a probabilistic graphical models for modelling racial bias within stop-and-searches and explore varying priors. Secondly using our inference methods, we produce a set of probability distributions for different racial/ethnic groups based on said model and data. Finally, we produce a set of examples of applications of this model, predicting biases not only for stops but also in the reactive response by law officers.