We’ve heard from a number of data users lately who have questions about the status of our plans to strengthen confidentiality safeguards for our data products, including the American Community Survey (ACS), the nation’s premier source for detailed population and housing information. We wanted to provide an update as we continue to evolve these safeguards in the face of challenges posed by the digital world we live in.

Our current assessment is that the science does not yet exist to comprehensively implement a formally private solution for the ACS. We expect a multiyear development period, including data user review and feedback, that will extend beyond 2025.

As you might recall, in 2018, we shared that the U.S. Census Bureau was modernizing disclosure avoidance safeguards for 2020 Census data products by moving to an approach known as differential privacy, a new, advanced, and far more powerful confidentiality protection system uniquely suited for a large-scale data product like a national census. Differential privacy is considered a formally private methodology because it provides an objective framework for quantifying global disclosure risk. We indicated at the time that we would eventually adapt this approach to the ACS and other products.

In 2019, we shared that adaptation to the ACS wouldn’t occur prior to 2025, given the complexity of implementing formal privacy for a complex multistage probability survey like the ACS. We also shared that our priority focus was to first develop the disclosure avoidance system for the 2020 Census data products. 

Strengthening confidentiality protections for the ACS is a continuous process, and we are taking the time to carefully research options and engage with the data user community. Our ongoing research has made it clear that the science for a formally private solution for the ACS does not yet exist. It’s also not clear that differential privacy would ultimately be the best option. Other formally private disclosure avoidance approaches may end up being a better fit for the ACS.

Science and research will lead the way in deciding the best long-term strategy for achieving formal (globally quantifiable) privacy protections for the suite of ACS data products, and this will be a multiyear process. In the meantime, as we identify vulnerabilities in ACS data products, we will strengthen the disclosure avoidance methods used for each one. This may involve using solutions that are not formally private, such as expanding reliance on synthetic data for the ACS Public Use Microdata Sample (PUMS) product. 

To address some of the known vulnerabilities in the ACS PUMS data product, we are researching the feasibility of creating a high-quality, fully synthetic public-use microdata file and accompanying validation service. This would allow users to verify the validity of analyses run on the microdata file while meeting ACS data users’ needs.

These products have several potential benefits to data users, such as including finer geographic detail for both the synthetic data and the validations and allowing users to account for privacy protection in their measures of uncertainty. Once developed, and after user review and feedback, these data products could replace the public ACS PUMS.

For the time being, the ACS PUMS data product will still be protected using traditional disclosure avoidance methods. Ultimately, we will develop a comprehensive disclosure avoidance solution that can assess and mitigate the cumulative disclosure risk posed by the separate ACS data products.

To help data users stay informed of developments in our ongoing disclosure avoidance research and decision-making, we recently launched a dedicated ACS disclosure avoidance webpage to serve as a central repository for up-to-date information. We also have a dedicated email for this topic at ACSprivacy@census.gov should you have questions.

In closing, the Census Bureau takes its data stewardship responsibilities very seriously. We also recognize the importance of keeping our data users informed and engaged as we make data stewardship decisions that could impact how our data are used, both internally and externally. We will make these decisions based on the best science available. We will continue to engage with our stakeholder community while we research and test the ACS synthetic PUMS file and formal privacy.