The Cross-Context Threshold Test: Detecting Discrimination Under Environmental Shifts

Jun Yuan, Xinyue Ye
Proceedings of The 29th International Conference on Artificial Intelligence and Statistics, PMLR 300:4330-4338, 2026.

Abstract

We study how threshold tests for detecting discrimination under environmental shifts, focusing on the Veil-of-Darkness (VoD) setting where visibility changes between daylight and darkness. We show that standard threshold tests, when applied separately to daylight and darkness data, violate key assumptions: risk distributions drift across contexts and thresholds fluctuate arbitrarily. We propose a cross-context threshold test that enforces distributional invariance and monotonic threshold decay. Using New York City stop-and-frisk data and synthetic experiments, we demonstrate that this model yields more reliable thresholds, improves bias detection, and aligns with the counterfactual logic of the VoD test. Our framework generalizes to fairness auditing whenever environmental context influences decisions.

Cite this Paper

BibTeX
@InProceedings{pmlr-v300-yuan26b, title = { The Cross-Context Threshold Test: Detecting Discrimination Under Environmental Shifts }, author = {Yuan, Jun and Ye, Xinyue}, booktitle = {Proceedings of The 29th International Conference on Artificial Intelligence and Statistics}, pages = {4330--4338}, year = {2026}, editor = {Khan, Emtiyaz and Li, Yingzhen and Solin, Arno and Ramdas, Aaditya}, volume = {300}, series = {Proceedings of Machine Learning Research}, month = {02--05 May}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v300/main/assets/yuan26b/yuan26b.pdf}, url = {https://proceedings.mlr.press/v300/yuan26b.html}, abstract = { We study how threshold tests for detecting discrimination under environmental shifts, focusing on the Veil-of-Darkness (VoD) setting where visibility changes between daylight and darkness. We show that standard threshold tests, when applied separately to daylight and darkness data, violate key assumptions: risk distributions drift across contexts and thresholds fluctuate arbitrarily. We propose a cross-context threshold test that enforces distributional invariance and monotonic threshold decay. Using New York City stop-and-frisk data and synthetic experiments, we demonstrate that this model yields more reliable thresholds, improves bias detection, and aligns with the counterfactual logic of the VoD test. Our framework generalizes to fairness auditing whenever environmental context influences decisions. } }
Endnote
%0 Conference Paper %T The Cross-Context Threshold Test: Detecting Discrimination Under Environmental Shifts %A Jun Yuan %A Xinyue Ye %B Proceedings of The 29th International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2026 %E Emtiyaz Khan %E Yingzhen Li %E Arno Solin %E Aaditya Ramdas %F pmlr-v300-yuan26b %I PMLR %P 4330--4338 %U https://proceedings.mlr.press/v300/yuan26b.html %V 300 %X We study how threshold tests for detecting discrimination under environmental shifts, focusing on the Veil-of-Darkness (VoD) setting where visibility changes between daylight and darkness. We show that standard threshold tests, when applied separately to daylight and darkness data, violate key assumptions: risk distributions drift across contexts and thresholds fluctuate arbitrarily. We propose a cross-context threshold test that enforces distributional invariance and monotonic threshold decay. Using New York City stop-and-frisk data and synthetic experiments, we demonstrate that this model yields more reliable thresholds, improves bias detection, and aligns with the counterfactual logic of the VoD test. Our framework generalizes to fairness auditing whenever environmental context influences decisions.
APA
Yuan, J. & Ye, X.. (2026). The Cross-Context Threshold Test: Detecting Discrimination Under Environmental Shifts . Proceedings of The 29th International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 300:4330-4338 Available from https://proceedings.mlr.press/v300/yuan26b.html.

Related Material