Open Letter to Census Bureau Leadership
Dr. Steven Dillingham, Director
Dr. Ron Jarmin, Deputy Director and Chief Operating Officer
Dr. Victoria A. Velkoff, Associate Director for Demographic Programs
Dr. John M. Abowd, Associate Director for Research and Methodology
We, the undersigned organizations and individuals, request that the Census Bureau (1) clarify the timeline for changes to the American Community Survey (ACS) public use data files, (2) engage the research community in advance of these changes, and (3) enable empirical assessment of the impact of any new confidentiality measures on research applications using ACS summary files and microdata. We believe that partnering with the scientific research community will produce a more thorough and valid understanding of real-world research and policy applications of these crucial data, and will help the Census Bureau optimize the trade-offs between scientific utility and confidentiality.
We know the Census Bureau is focused on the 2020 Census. We hope that decisions regarding the ACS are not made hastily, and we would be happy to assist in the evaluation of any new ACS data products before they are released.
Value of ACS Data
The ACS is by far the most intensively used data source in the social sciences. To date 54,000 papers have relied on the ACS, including 9,000 papers in the past year alone. These papers span an enormous diversity of research across many disciplines, including topics such as inequality, labor force participation, aging, population health, fertility, family structure, mortality, migration, population distribution, residential segregation, transportation and economic development. Harmonized versions of the data are disseminated through IPUMS, which has prepared over a million customized microdata extracts for over 100,000 investigators. Use of the ACS data are not confined to academic research. The data are widely used by government agencies at all levels and by nongovernmental organizations for policy-relevant research. Over 1,000 journalists have registered to use the IPUMS version of the data, representing organizations such as the New York Times, Washington Post, Boston Globe, Wall Street Journal, and National Public Radio.
The most feasible way to make a comprehensive assessment of the impact of the new procedures is to leverage the talents of the research community. Dozens of researchers with diverse research topics and methodologies are prepared to quickly replicate their peer-reviewed studies and research reports using data prepared with the new disclosure avoidance methodology, and to report their findings in a standard format. This will enable a transparent evaluation of the impact of the changes. It would also be feasible to evaluate multiple versions of the data to determine optimal noise-infusion strategies.
To execute this plan, we need a version of the 2005-2017 ACS summary files and microdata samples produced with the new generation of noise-infusion software. In the interest of full transparency, we therefore request that the Census Bureau produce such files as soon as possible. IPUMS has agreed to harmonize and disseminate the files and collate the results of the analyses.
Thank you for considering this request.