Times are displayed in (UTC-04:00) Eastern Time (US & Canada)Change
Noisy Geographies: The Influence of Differential Privacy and TopDown Algorithms on the 2020 Census Data
Topics: Political Geography
, Population Geography
, Medical and Health Geography
Keywords: Census, Demographics, Gerrymandering Session Type: Virtual Paper Abstract Day: Monday Session Start / End Time: 2/28/2022 11:20 AM (Eastern Time (US & Canada)) - 2/28/2022 12:40 PM (Eastern Time (US & Canada)) Room: Virtual 23
Authors:
John Hessler, Johns Hopkins University / Library of Congress
,
,
,
,
,
,
,
,
,
Abstract
The 2020 Census is one of most critical sets of demographic geospatial data used for the planning of Congressional districts, the allocation of funds for health care and other government programs, public policy, and geographic research. The 2020 Census was the first to use at scale differential privacy (DP) protections, in the form of noise injected into the data to protect respondents from identity reconstruction attacks. TopDown is the name given to the collection of DP algorithms used to generate confidentiality-preserving microdata that influences the released demographic information about the United States population. The complexity of the algorithm however makes it extremely difficult to study analytically. In this paper we will review the classic database reconstruction theorems and investigate the structure of the TopDown algorithm and show how it will influence geographic research in everything from the allocation and planning of Congressional district maps, to the distortion of age and race specific mortality rates from COVID-19 infections.
Noisy Geographies: The Influence of Differential Privacy and TopDown Algorithms on the 2020 Census Data