The Census Bureau is planning an extensive revision of procedures for ensuring the confidentiality of public use data, based on a method known as "differential privacy." The new disclosure avoidance system will rely on injecting noise into the data, and the new procedures may reduce or even eliminate the usability of public use data for many common research and policy applications. For more information, see our recent working paper (
https://z.umn.edu/working-paper-2018-6) and forthcoming article (
https://z.umn.edu/diffprivacy).
The American Community Survey (ACS) is the single most important statistical source in the social sciences. It generates 9,000 articles a year, and a new paper using these data appears about every hour. If these data become unusable, unreliable or available only through restricted-access environments because of the new confidentiality measures, that would have profound consequences across many areas of demographic, economic, and policy research.
Acknowledging the importance of the ACS data, we are seeking clarification and engagement with the research community from the Census Bureau. Our open letter (
z.umn.edu/lettertoCensusBureau) requests:
1.
A public timeline for changes to the ACS summary files and microdata.
2.
Engagement with the research community in advance of any changes.
3.
Public versions of new noise-infused ACS summary files and microdata for 2005-2017 so that researchers can assess their usefulness by replicating existing studies.
If you support these requests, please sign our letter by completing the form below.