Beyond Internal Data: Bounding and Estimating Fairness from Incomplete Data

2025-08-18

AI Summary

Evaluating AI model fairness is challenging when complete data are unavailable—e.g., only internal institutional samples and external demographic marginals are accessible. Method: This paper proposes a fairness bounding framework for disjoint data settings. It constructs the set of feasible joint distributions consistent with both internal and external marginal constraints, then derives rigorous upper and lower bounds for fairness metrics—including equal opportunity and predictive parity—via distributionally robust optimization and marginal statistical inference. The approach requires no assumptions about the underlying data-generating process or access to raw shared data. Contribution/Results: The method enables compliant bias auditing under privacy-sensitive and data-siloed conditions. Experiments on synthetic, real-world healthcare, and credit datasets demonstrate that it yields tight, reliable, and interpretable fairness bounds—substantially outperforming existing proxy-based evaluation and imputation baselines.

Technology Category

• Philosophy and Ethics of AI: Bias, Fairness & Equity
• Computer Vision: Bias, Fairness & Privacy
• Machine Learning: Ethics, Bias, and Fairness

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Economics, Online Markets and Human Computation: Fairness and ethical considerations in crowd work and in human-in-the-loop AI systems
• Security and Privacy: Data transparency and provenance

Abstract

Ensuring fairness in AI systems is critical, especially in high-stakes domains such as lending, hiring, and healthcare. This urgency is reflected in emerging global regulations that mandate fairness assessments and independent bias audits. However, procuring the necessary complete data for fairness testing remains a significant challenge. In industry settings, legal and privacy concerns restrict the collection of demographic data required to assess group disparities, and auditors face practical and cultural challenges in gaining access to data. In practice, data relevant for fairness testing is often split across separate sources: internal datasets held by institutions with predictive attributes, and external public datasets such as census data containing protected attributes, each providing only partial, marginal information. Our work seeks to leverage such available separate data to estimate model fairness when complete data is inaccessible. We propose utilising the available separate data to estimate a set of feasible joint distributions and then compute the set plausible fairness metrics. Through simulation and real experiments, we demonstrate that we can derive meaningful bounds on fairness metrics and obtain reliable estimates of the true metric. Our results demonstrate that this approach can serve as a practical and effective solution for fairness testing in real-world settings where access to complete data is restricted.

Problem

• Estimating AI fairness without complete demographic data
• Leveraging separate internal and external datasets for fairness assessment
• Bounding fairness metrics when protected attributes are inaccessible

Innovation

• Leverage separate incomplete data sources
• Estimate feasible joint distributions
• Compute plausible fairness metrics bounds