2020 Census Redistricting Data (P.L. 94-171) Noisy Measurement File
Year: 2023
Type: dataset
Abstract: The 2020 Census Redistricting Data (P.L. 94-171) Noisy Measurement File (NMF) is an intermediate output of the 2020 Census Disclosure Avoidance System (DAS) TopDown Algorithm (TDA) (as described in Abowd, J. et al [2022] https://doi.org/10.1162/99608f92.529e3cb9, and implemented in https://github.com/uscensusbureau/DAS_2020_Redistricting_Production_Code). The 2020 Redistricting NMF was an intermediate output of the DAS during the execution of the algorithm to produce the 2020 Census Redistricting Data (P.L. 94-171) Summary File. The NMFs are intermediate privacy-protected outputs of the DAS; they were generated using the Census Bureau's implementation of the Discrete Gaussian Mechanism (https://arxiv.org/abs/2004.00010), calibrated to satisfy zero-Concentrated Differential Privacy (https://arxiv.org/abs/1605.02065) with bounded neighbors (https://dl.acm.org/doi/10.1145/1989323.1989345). The NMF values, called "noisy measurements," are the output of applying the Discrete Gaussian Mechanism to counts from the 2020 Census Edited File (CEF). They are generally inconsistent with one another (for example, in a county composed of two tracts, the noisy measurement for the county's total population may not equal the sum of the noisy measurements of the two tracts' total population), and frequently negative (especially when the population being measured was small), but are integer-valued. The NMF was later post-processed as part of the DAS code to take the form of microdata and to satisfy various constraints. The NMF documented here contains both the noisy measurements themselves as well as the data needed to represent the DAS constraints; thus, the NMF could be used to reproduce the steps taken by the DAS code to produce microdata from the noisy measurements by applying the production code base.
Source: Harvard Dataverse
DOI: 10.7910/dvn/5lavkv
Authors: Abowd, John M.,, Ashmead, Robert, Cumings-Menon, Ryan, Garfinkel, Simson, Heineck, Micah, Heiss, Christine, Johns, Robert, Kifer, Daniel, Leclerc, Philip, Machanavajjhala, Ashwin, Moran, Brett, Sexton, William, Spence, Matthew, Zhuravlev, Pavel
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
Funder: U.S. Census Bureau
OpenAlex ID: W4398397510