CALL FOR INPUT
Committee on National Statistics
Workshop on 2020 Census Data Products: Data Needs and Privacy Considerations
December 11 - 12, 2019
Auditorium, National Academy of Sciences Building
2101 Constitution Ave. NW, Washington, DC 20418
REGISTER HERE (note: this event will be Webcast live)
The planning committee for the Committee on National Statistics (CNSTAT) Workshop on 2020 Census Data Products is pleased to issue this public call for input, with the intent of performing the broadest possible sweep of users (and use cases) of decennial census data. We want and need to hear from prospective users of 2020 Census data products, to inform the Census Bureau's final decision making about the 2020 products.
On October 29, the Census Bureau released a set of “2010 Demonstration Products” — tables comparable to those released as official data products of the 2010 Census, but generated by applying the Census Bureau’s new-for-2020 protocol for disclosure avoidance (based on the concept of differential privacy) to raw 2010 census data. A crosswalk file describes the tables currently planned for release in 2020 and the subset of tables included in the 2010 Demonstration Products. In brief, the Demonstration Products comprise:
· * A complete version of the P.L. 94-171 Redistricting Data Summary File, and
· * A subset of tables from the planned 2020 Demographic and Housing Characteristic (DHC) File, which in turn is meant to cover most of the content of the 2010 Summary File 1. Notable exclusions from the Demonstration Products include tabulations beyond the most top-level race/Hispanic origin codes (including detailed American Indian and Alaska Native affiliation) and tabulations based on within-household composition, for which improved disclosure avoidance protocols have not yet been identified by the Census Bureau.
Again, we want to hear from census data product users, those whose applications of the products are directly covered in the 2010 Demonstration Products as well as those whose are not.
We recognize that time until the December 11–12 workshop is extremely short and regret any inconvenience — but we also know this to be a critical time, to inform the setting of final parameters for the 2020 Census products before major field operations begin and consume the time and energy of Census Bureau leadership. Hence, we ask for and welcome your input along several possible lines:
· * If your use of decennial census tabulations is covered in the Demonstration Products: How does the accuracy and utility of the new data compare with the original 2010 census data releases? Are there new challenges in interpretation, and is the level of geographical detail planned for 2020 adequate for your use? Are there dimensions (geographic unit type, basic demographic characteristics, age groups) for which it is essential to your use of the data that the tabulations afford more accuracy (in dividing up the overall privacy loss budget)?
· * If your use of decennial census tabulations is not covered in the demonstration products: How critical are decennial census data products to your work or satisfying legal/regulatory needs? What census data tabulations are absolutely necessary for your use, and at what degree of geographic and population group resolution? Are there alternatives (other data resources) that might be brought to bear?
· * Can you produce a “case study” of the use of privacy-protected census data in your application area by comparing the 2010 Demonstration Products with the applicable 2010 products as originally released, or, alternatively, by describing in detail how your application used the original 2010 products and why they were essential? We welcome case studies in all varieties of application area, from redistricting to provision/targeting of government services to deriving the denominators for calculation of policy-important rates and percentages.
· * Can you address the essential tension between privacy and accuracy of 2020 census data products, with reference to specific data? Looking at the demonstration products holistically, do they afford enough privacy protection to safeguard individuals while still being usable for essential needs?
What we need at this point are brief responses, signals of interest — indications of whether you plan to scrutinize the 2010 Demonstration Products and whether you have a case study or an application that you think would be good to showcase at the workshop. Even short, fragmentary expressions will be helpful in shaping the workshop agenda. We may be back in touch with you about presenting/discussing at the workshop; we may ask you to flesh out use case examples into “one-pagers” that could be distributed at the workshop or presented in a virtual “poster session” at the workshop. But, more generally, we want to hear from and reflect in this workshop as broad a swath of the census data user community as possible.
To answer this Call for Input, please email Daniel Cork (firstname.lastname@example.org). In addition to your brief commentary on your data product use, please include your name and affiliation; whether you anticipate using the 2020 P.L. 94-171 Redistricting Summary File or the 2020 Demographic and Housing Characteristic File (akin to 2010’s Summary File 1), and which geographic summary levels or characteristics/census variables are most critical to you. Please also indicate whether you plan to analyze the 2010 Demonstration Products to evaluate the effects of the new disclosure avoidance system, and whether any results might be available on or before the December 11–12 workshop.
The Workshop on 2020 Census Data Products is organized by the Committee on National Statistics (CNSTAT) of the National Academies of Sciences, Engineering, and Medicine, at the request of the U.S. Census Bureau. The workshop is open to the public and will be webcast for remote viewing; for planning purposes, please RSVP HERE. A proceedings will be issued based on the workshop presentations and discussions, and may include additional materials submitted in response to this Call for Input.