Sequential Sensitivity Analysis for Multiple Assumptions: A Framework for Understanding Racial Disparity in Police Use of Force

2026-05-20

AI Summary

Inferring racial discrimination in police use of force requires simultaneously satisfying two assumptions: non-discriminatory stops and unbiased post-stop encounters. Existing methods can only test one assumption at a time. This study proposes the first sequential sensitivity analysis framework that jointly perturbs both assumptions, revealing how their interaction affects estimates of racial disparities and overcoming the limitations of conventional single-assumption tests. Leveraging causal inference techniques calibrated with census data, the framework is applied to New York City’s “stop, question, and frisk” records from 2003 to 2013. Results indicate significant racial disparities under plausible levels of stop-level discrimination, yet these findings are highly sensitive to encounter-level bias, exhibiting notable fragility within demographically feasible ranges.

Abstract

Inferring racial discrimination in police use of force -- the average causal effect of civilian race on use of force -- requires two assumptions about policing prior to potential use of force: that officers do not discriminate in whom they would stop (no discrimination in stops) and that, conditional on patrol context, the probability that an encounter is with a minority rather than a white civilian does not vary across encounters (no bias in encounters). As Knox et al. (2020) show, violations of the first can mask racial disparity in force. Whether it reflects discrimination in force also depends on the second. Existing sensitivity analyses address one assumption at a time. We develop a framework that varies both sequentially and apply it to NYPD Stop, Question, and Frisk data (2003--2013). Under plausible levels of discrimination in stops, we find substantial racial disparity in force. However, the conclusion that this disparity reflects discrimination is fragile to modest departures from no bias in encounters that census-based calibration suggests are demographically feasible. By jointly addressing both confounding channels, the framework reveals how they interact in ways that separate analyses cannot, contributing to understanding what generates racial disparities and how they might be addressed.

Problem

• racial disparity
• police use of force
• sensitivity analysis
• discrimination in stops
• bias in encounters

Innovation

• sequential sensitivity analysis
• racial disparity
• police use of force
• confounding bias
• causal inference