U.S. Census Data for Research
Data Ethics
As with all data sources, it is important to consider the Census through a critical ethical lens if you use it for your research project. Data ethics encompasses a variety of topics. The topics discussed in detail in this LibGuide are privacy concerns and the impacts of COVID-19. For more information on data ethics overall, see the following resources.
Privacy Concerns
By law, the Census Bureau is required to maintain the privacy and confidentiality of individuals who respond to the survey. Over the years, the Census Bureau has conducted research into the risks of private information exposure resulting from public data release, known as re-identification. The tenet of protecting the privacy of individual survey respondents is what the Census refers to as disclosure avoidance. The 2020 Census is the first Census to use a statistical technique known as Differential Privacy to balance the tradeoff between accurate data and privacy protection. The Census Bureau has a long history of researching and protecting privacy, which has come to a head with the introduction of differential privacy methods to protect the privacy of respondents beginning in 2020.
The impacts of differential privacy on research with the Census continue to be researched and understood. While the Census Bureau maintains that differential privacy does not compromise the quality of the 2020 Decennial Census greatly, the effects of differential privacy are not felt evenly for all data users. It is understood that the greatest impacts are for users of small geographies, more specifically users of data at the sub-county level.
- Gong, R., Groshen, E. L., & Vadhan, S. (2022)This special issue of the Harvard Data Science Review, titled, "Harnessing the Known Unknowns: Differential Privacy and the 2020 Census," has become an authoritative source on the range of issues involved in managing the confidentiality-utility tradeoff in the Census context.
- Hotz et al. (2022)Hotz et al. explore the nuances of the confidentiality-utility tradeoff in the Census context, arguing that the public benefit of Census data products is not properly weighed in cost-benefit frameworks, and provide recommendations on the future of disclosure avoidance.
- Christ et al. (2022)Christ et al. (2022) explore theoretically differential privacy methods, including those used in the 2020 Decennial Census, to examine the impact of differential privacy on disclosure avoidance across racial groups, finding that the privacy burden is unequally placed on racial minorities.
- Kenny et al. (2021)Kenny et al. (2021) studied the impact of differential privacy on the 2020 Decennial Census redistricting data, finding that the data may have been systemically biased against mixed-race and mixed-partisan precincts.
- Krieger et al. (2021)Krieger et al. (2021) provide an application of the impact of differential privacy in the American Community Survey to the field of health equity. They do not find evidence that differential privacy has adversely affected health equity estimates at the Census tract level.
Impacts of COVID-19
Because the 2020 Decennial Census, as well as the American Community Survey, were being administered during the COVID-19 pandemic, there were several changes to the normal administration, processing, and dissemination of the survey. The first impact was that the survey took longer to administer, and thus the data took longer to be released to the public. The Census Bureau maintains that COVID-19 did not significantly impact the quality of the 2020 Decennial Census. Research is still being conducted to understand why COVID-19 caused an undercount or overcount, in which areas, and if this miscount was or was not even across demographics. It is important for data users to understand if the Census data sample is biased to evaluate its appropriateness for their research.
A Census employee wearing PPE in 2020
U.S. Census Bureau, Public domain, via Census Photos
- National Academies (2022)The National Academies Interim Report titled, "Understanding the Quality of the 2020 Census" is a detailed report on the Academies' ongoing assessment of the latest decennial census.
- Shuster et al. (2020)Shuster et al. (2020) explore historic inaccuracies in the Decennial Census, emphasizing inequity in the count by race and ethnicity. Implications are explored for the 2020 Decennial Census and undercounting or overcounting populations.
- Warren (2020)Warren (2020) cautions users of the 2020 American Community Survey estimates for researching foreign-born individuals, finding evidence for an undercount of this population.