Assessing Statistical Disclosure Risk for Differentially Private, Hierarchical Count Data, with Application to the 2020 U.S. Decennial Census

TL;DR

This paper introduces Bayesian methods to evaluate the disclosure risk of hierarchical count data released under differential privacy, with applications to the 2020 U.S. Census and simulations based on historical census data.

Contribution

It develops novel Bayesian risk assessment techniques tailored for hierarchical, categorical data under zero-concentrated differential privacy, addressing a gap in privacy risk evaluation.

Findings

Risk increases with more hierarchical information used by intruders.

Privacy parameter choices significantly impact disclosure risk.

Method effectively quantifies risk in census data releases.

Abstract

We propose Bayesian methods to assess the statistical disclosure risk of data released under zero-concentrated differential privacy, focusing on settings with a strong hierarchical structure and categorical variables with many levels. Risk assessment is performed by hypothesizing Bayesian intruders with various amounts of prior information and examining the distance between their posteriors and priors. We discuss applications of these risk assessment methods to differentially private data releases from the 2020 decennial census and perform simulation studies using public individual-level data from the 1940 decennial census. Among these studies, we examine how the data holder's choice of privacy parameter affects the disclosure risk and quantify the increase in risk when a hypothetical intruder incorporates substantial amounts of hierarchical information.