Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

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

William Sexton, Skye Berghel, Bayard Carlson, Sam Haney, Luke Hartman, Michael Hay, Ashwin Machanavajjhala, Gerome Miklau, Amritha Pai, Simran Rajpal, David Pujol, Ruchit Shrestha, Daniel Simmons-Marengo

TL;DR

SafeTab-H addresses the challenge of privately releasing Detailed DHC-B household type and tenure counts by iterating over population groups defined by geography and detailed race/ethnicity using a discrete Gaussian mechanism under zero-concentrated differential privacy. The approach selects table variants based on $T01001$ counts, computes basis vectors, adds calibrated noise with level-specific budgets, and applies postprocessing to ensure demographic reasonableness while preserving privacy. The paper provides formal zCDP guarantees, a bounded zCDP conversion, and a detailed account of implementation through Tumult Analytics, input validation, and postprocessing steps for marginals, coterminous geographies, and suppression. It also discusses parameter tuning via an analysis tool, highlighting the trade-offs between privacy loss, accuracy (MOE), and data availability, and demonstrates practical decisions (e.g., eight-race multiplicity) guided by policy and data users.

Abstract

This article describes SafeTab-H, a disclosure avoidance algorithm applied to the release of the U.S. Census Bureau's Detailed Demographic and Housing Characteristics File B (Detailed DHC-B) as part of the 2020 Census. The tabulations contain household statistics about household type and tenure iterated by the householder's detailed race, ethnicity, or American Indian and Alaska Native tribe and village at varying levels of geography. We describe the algorithmic strategy which is based on adding noise from a discrete Gaussian distribution and show that the algorithm satisfies a well-studied variant of differential privacy, called zero-concentrated differential privacy. We discuss how the implementation of the SafeTab-H codebase relies on the Tumult Analytics privacy library. We also describe the theoretical expected error properties of the algorithm and explore various aspects of its parameter tuning.

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

TL;DR

SafeTab-H addresses the challenge of privately releasing Detailed DHC-B household type and tenure counts by iterating over population groups defined by geography and detailed race/ethnicity using a discrete Gaussian mechanism under zero-concentrated differential privacy. The approach selects table variants based on counts, computes basis vectors, adds calibrated noise with level-specific budgets, and applies postprocessing to ensure demographic reasonableness while preserving privacy. The paper provides formal zCDP guarantees, a bounded zCDP conversion, and a detailed account of implementation through Tumult Analytics, input validation, and postprocessing steps for marginals, coterminous geographies, and suppression. It also discusses parameter tuning via an analysis tool, highlighting the trade-offs between privacy loss, accuracy (MOE), and data availability, and demonstrates practical decisions (e.g., eight-race multiplicity) guided by policy and data users.

Abstract

This article describes SafeTab-H, a disclosure avoidance algorithm applied to the release of the U.S. Census Bureau's Detailed Demographic and Housing Characteristics File B (Detailed DHC-B) as part of the 2020 Census. The tabulations contain household statistics about household type and tenure iterated by the householder's detailed race, ethnicity, or American Indian and Alaska Native tribe and village at varying levels of geography. We describe the algorithmic strategy which is based on adding noise from a discrete Gaussian distribution and show that the algorithm satisfies a well-studied variant of differential privacy, called zero-concentrated differential privacy. We discuss how the implementation of the SafeTab-H codebase relies on the Tumult Analytics privacy library. We also describe the theoretical expected error properties of the algorithm and explore various aspects of its parameter tuning.
Paper Structure (39 sections, 8 theorems, 4 equations, 9 tables, 3 algorithms)

This paper contains 39 sections, 8 theorems, 4 equations, 9 tables, 3 algorithms.

Table of Contents

  1. Introduction
  2. Problem Setup
  3. Households
  4. Geography
  5. Race and Ethnicity
  6. Population Groups
  7. Detailed Demographic and Housing Characteristics File B
  8. Private Release Problem
  9. SafeTab-H Algorithm
  10. Input Data Description
  11. Household Data
  12. Total Population Counts
  13. The Algorithm Description
  14. Privacy Preliminaries
  15. Privacy Definitions
  16. ...and 24 more sections

Key Result

Lemma 1

(Adaptive sequential composition of zCDP BunS16) Let $M_1: \mathcal{X} \rightarrow \mathcal{Y}$ and $M_2: \mathcal{X} \times \mathcal{Y} \rightarrow \mathcal{Z}$ be mechanisms satisfying $\rho_1$-zCDP and $\rho_2$-zCDP respectively. Let $M_3(x) = M_2(x, M_1(x))$. Then $M_3$ satisfies $(\rho_1 + \rho

Theorems & Definitions (19)

  • Definition 1: Neighboring Databases
  • Definition 2: Bounded-Neighboring Databases
  • Definition 3
  • Definition 4
  • Definition 5
  • Lemma 1
  • Lemma 2
  • Definition 6: L2 Sensitivity
  • Definition 7: Bounded L2 Sensitivity
  • Definition 8
  • ...and 9 more