SafeTab-P: Disclosure Avoidance for the 2020 Census Detailed Demographic and Housing Characteristics File A (Detailed DHC-A)

2025-05-02

AI Summary

Fine-grained race/ethnicity statistics in the 2020 U.S. Census Detailed Demographic and Housing Characteristics File A (DHC-A) pose significant privacy risks due to re-identification vulnerabilities. Method: We propose the first differentially private framework for releasing multi-level geographic and cross-tabulated group statistics, featuring an adaptive granularity control mechanism that dynamically adjusts the number of statistics and the resolution of geographic and categorical dimensions based on group size, coupled with discrete Gaussian noise injection under zero-concentrated differential privacy (zCDP). Contribution/Results: Implemented and deployed via Tumult Analytics within the official census release pipeline, our approach achieves strong formal privacy guarantees (ε = 0.48 zCDP) while substantially improving statistical utility over prior methods. Empirically tuned and budget-validated, it balances practical deployability with regulatory compliance, establishing a scalable paradigm for privacy-preserving large-scale official statistics.

Technology Category

• Machine Learning: Privacy
• Computer Vision: Bias, Fairness & Privacy
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

• Security and Privacy: Data transparency and provenance
• User Modeling, Personalization and Recommendation: User privacy protection in personalized systems
• Responsible Web: Data and user privacy-enhancing technologies for the Web

Abstract

This article describes the disclosure avoidance algorithm that the U.S. Census Bureau used to protect the Detailed Demographic and Housing Characteristics File A (Detailed DHC-A) of the 2020 Census. The tabulations contain statistics (counts) of demographic characteristics of the entire population of the United States, crossed with detailed races and ethnicities at varying levels of geography. The article describes the SafeTab-P algorithm, which is based on adding noise drawn to statistics of interest from a discrete Gaussian distribution. A key innovation in SafeTab-P is the ability to adaptively choose how many statistics and at what granularity to release them, depending on the size of a population group. We prove that the algorithm satisfies a well-studied variant of differential privacy, called zero-concentrated differential privacy (zCDP). We then describe how the algorithm was implemented on Tumult Analytics and briefly outline the parameterization and tuning of the algorithm.

Problem

• Develops SafeTab-P to protect 2020 Census demographic data
• Adaptively releases statistics based on population group size
• Ensures zero-concentrated differential privacy (zCDP) compliance

Innovation

• Uses discrete Gaussian noise for privacy
• Adaptively selects statistic granularity by group size
• Implements zero-concentrated differential privacy (zCDP)