Personally identifiable information collected for the decennial census is not released by the government for 72 years after it was collected. After 72 years, the records are released to the public by the National Archives and Records Administration. [source] In 2012, individual-level records from the 1940 census were released; 1950 census records are due to be released in 2022.
For scholarly research purposes, this means that 1940 and earlier census records can be used to search for information about individuals. For example, a genealogist can search for records by name and see the names, ages, occupations, and other information about specific individuals. For 1950 and later decennial census records, researchers must use aggregate data (with the exception of Public Use Microdata, a special kind of anonymized individual-level data set.)
After 72 years, census records become the domain of the National Archives, which makes census schedules all the way back to the first decennial census in 1790 available on microfilm, and increasingly as digital images online. [source]
U.S. Census Bureau – "The 72-Year Rule"
An official overview detailing the origins and implications of the 72-Year Rule, which restricts access to personally identifiable census data for 72 years.
National Archives – "Census Records: The 72-Year Rule"
An article discussing the historical context and significance of the 72-Year Rule in preserving individual privacy.
Pew Research Center – "The 72-Year Rule Governs Release of Census Records"
A brief explanation of the rule and its impact on genealogical research and public access to historical census data.
Public Use Microdata (PUMS) provides individual-level responses from a sample of people and households who participated in the American Community Survey (ACS), with identifying information removed to protect privacy. Unlike pre-tabulated summary data, PUMS lets researchers create custom analyses using raw data on variables like income, education, housing, and more. These datasets are available through the U.S. Census Bureau and via IPUMS USA, which offers a more user-friendly interface and harmonized variables for longitudinal analysis. Because PUMS data is complex and large in scale, it's best suited for advanced users or researchers with experience in statistical software like R, Stata, or SPSS. To ensure confidentiality, data is only available for large geographic areas called Public Use Microdata Areas (PUMAs), each containing at least 100,000 people.
IPUMS USA – Public Use Microdata
Access harmonized microdata samples from the U.S. Census and ACS for detailed research.
U.S. Census Bureau – Introduction to PUMS
Official guide to using the Public Use Microdata Sample from the ACS.
Journal of Privacy and Confidentiality – Public-Use vs. Restricted-Use
A comparison of public and restricted census microdata and their use in research.
Differential Privacy in the 2020 Census
To safeguard individual confidentiality in the 2020 Census, the U.S. Census Bureau implemented a modern statistical technique known as differential privacy. This method introduces controlled random noise into published data, ensuring that no individual's information can be discerned from aggregate statistics. By doing so, it addresses the heightened risks of re-identification posed by advanced computing and the proliferation of external data sources.
Differential privacy operates on a principle of balancing data utility with privacy protection. The Census Bureau's Disclosure Avoidance System (DAS), which incorporates differential privacy, was applied to most 2020 Census data products, excluding apportionment counts. This system ensures that while individual data remains confidential, the overall statistical insights remain valuable for analysis.
However, the application of differential privacy has led to discussions about data accuracy, especially in small or rural communities. For instance, some tribal nations have expressed concerns over reduced data granularity, which can impact resource allocation and policy decisions. The Census Bureau continues to engage with stakeholders to refine its methods, striving to maintain both data privacy and utility.
For a deeper understanding of differential privacy and its implementation in the 2020 Census, consider exploring the following resources:
These materials provide comprehensive insights into the rationale, methodology, and implications of differential privacy in the context of the 2020 Census.
Census Bureau – Disclosure Avoidance for the 2020 Census
Overview of how differential privacy protects personal data in modern census releases.
AEA Papers & Proceedings – Differential Privacy and Census Data
An analysis of how differential privacy impacts economic and social science research.
Harvard Data Science Review – Implementing Differential Privacy
Lessons learned from applying differential privacy to the 2020 Census.