How differential privacy will affect our understanding of health disparities in the United States

Abstract

<jats:p>The application of a currently proposed differential privacy algorithm to the 2020 United States Census data and additional data products may affect the usefulness of these data, the accuracy of estimates and rates derived from them, and critical knowledge about social phenomena such as health disparities. We test the ramifications of applying differential privacy to released data by studying estimates of US mortality rates for the overall population and three major racial/ethnic groups. We ask how changes in the denominators of these vital rates due to the implementation of differential privacy can lead to biased estimates. We situate where these changes are most likely to matter by disaggregating biases by population size, degree of urbanization, and adjacency to a metropolitan area. Our results suggest that differential privacy will more strongly affect mortality rate estimates for non-Hispanic blacks and Hispanics than estimates for non-Hispanic whites. We also find significant changes in estimated mortality rates for less populous areas, with more pronounced changes when stratified by race/ethnicity. We find larger changes in estimated mortality rates for areas with lower levels of urbanization or adjacency to metropolitan areas, with these changes being greater for non-Hispanic blacks and Hispanics. These findings highlight the consequences of implementing differential privacy, as proposed, for research examining population composition, particularly mortality disparities across racial/ethnic groups and along the urban/rural continuum. Overall, they demonstrate the challenges in using the data products derived from the proposed disclosure avoidance methods, while highlighting critical instances where scientific understandings may be negatively impacted.</jats:p>

The 2020 U.S. Census Differential Privacy Method Introduces Disproportionate Discrepancies for Rural and Non-White Populations

“…The Census Bureau has provided researchers with test data for evaluation, and multiple papers have found significant issues with each successive vintage of experimental data. In general, demographers find there is little issue with overall population counts, but serious issues for population sub-groups and small geographic areas (Hauer & Santos-Lozada, 2021;Kenny et al, 2021;Santos-Lozada, Howard, & Verdery, 2020). As foreshadowed by Figure 1 below, our results do not suggest the Census Bureau has resolved this issue in their April 2021 data release.…”

The Role of Chance in the Census Bureau Database Reconstruction Experiment

“…The results of the database reconstruction experiment do not justify the substantial degradation of the nation's statistical infrastructure resulting from the implementation of differential privacy. The intentional errors introduced by Census Bureau's new disclosure control methods will compromise the utility of the data for demographic analysis, policy research, and planning (e.g., Hauer & Santos-Lozada, 2021;Santos-Lozada et al, 2020;Winkler et al, 2021). Weighing the high cost of the new disclosure protocol against negligible benefit for respondent confidentiality, it is apparent that differential privacy for census data is an unfortunate mistake.…”

“…Rigorous evaluation of the Census Bureau's database reconstruction experiment is important because differential privacy will add error to every statistic the agency produces for geographic units below the state level, and this error will significantly reduce the usability of census data for social, economic, and health research, and will compromise basic demographic measures (Hauer & Santos-Lozada, 2021;Ruggles et al, 2018a, 2018bSantos-Lozada et al, 2020Winkler et al, 2021).…”

The sociospatial factors of death: Analyzing effects of geospatially-distributed variables in a Bayesian mortality model for Hong Kong

“…Others have investigated a wide range of socioeconomic, psychological, and behavioral factors of fatality risks [10]. We often examine the notion of disparity in health and mortality risks using population-scale inputs and sensitive individual variables such as age, race, and gender respecting the privacy concerns that emerge from such applications [11]. Ou et al [12] infer socioeconomic status by type of housing, education, and occupation.…”