Jan 31 -- Today, the Census Bureau released the Proof of Concept for the 2020 Census Detailed Demographic and Housing Characteristics File A (Detailed DHC-A) for public review and feedback. This Proof of Concept is based on 2010 Census data. The Census Bureau is also hosting a webinar to discuss the Proof of Concept; see below for webinar information.
The 30-day public feedback period on the Proof of Concept will conclude on March 2. The final Detailed DHC-A based on 2020 Census data is slated for release in August.
As reported previously, this release demonstrates how the product’s differentially private algorithm, called SafeTab-P, uses an adaptive design that determines the amount of data detailed racial and ethnic groups and American Indian and Alaska Native tribes and villages receive based on population size and geography level while ensuring sufficient confidentiality protections.
In addition to the description of the algorithm and guidance for data users, today’s release includes example tables and various performance metrics.
About the Proof of Concept Release
-- Different requirements, different algorithms. The requirements for the Detailed DHC-A are different than those for the 2020 Census Redistricting Data (Public Law 94-171) Summary File, Demographic and Housing Characteristics File (DHC), and Demographic Profile. Unlike the Detailed DHC-A, those products require the creation of a privacy-protected microdata file, consistency through aggregation, and postprocessing. They are protected using the TopDown Algorithm (TDA), which meets those requirements. The Detailed DHC-A uses an algorithm called SafeTab-P, which produces privacy-protected tabulations directly, rather than requiring the creation of a privacy-protected microdata file, consistency through aggregation, or postprocessing.
-- Not a full set of demonstration data. Rather, example data with margins of error. Unlike the data products protected with the TopDown Algorithm, the Detailed DHC-A is tuned to meet pre-set margins of error, within a 95% confidence interval. This allows for an easier analysis of accuracy rather than researching the margins of error across population groups, tables, and geographies.
-- Adaptive design maximizes data availability. The Detailed DHC-A uses an adaptive design that determines the amount of data racial and ethnic groups and American Indian and Alaska Native tribes and villages receive based on population size and geography level. By using minimum population counts (i.e., thresholds) to determine how much data are produced for each racial or ethnic group by geography level, we are able to produce accurate data while ensuring sufficient confidentiality protections. In 2010, we used a single population threshold of 100 for every table and every geographic area. The Detailed DHC-A uses dynamic population thresholds, all of which are less than 100 for the Proof of Concept product. (See the adaptive design fact sheet.) https://www.census.gov/library/fact-sheets/2023/dec/adaptive-design.html
-- Data for more groups than ever before. Improvements in questionnaire design, processing, and coding allow for the release of more detailed racial and ethnic data than ever before: approximately 370 detailed racial and ethnic groups and 1,200 detailed American Indian and Alaska Native (AIAN) tribal and village population groups. Note that the increase in the amount of detailed race and ethnicity data that will be available in the 2020 Census Detailed DHC-A is not reflected in the Proof of Concept because this product is based on 2010 Census data.
-- Population data. Racial and ethnic groups will receive total population data at the national and state levels. With adaptive design, groups that meet minimum population counts will also receive total population data for counties, tracts, places, and American Indian/Alaska Native/Hawaiian Home Land (AIANNH) areas.
-- Sex by age data. Population size for racial and ethnic groups at each level of geography will determine whether the group receives sex by age tables and the level of detail in those tables (four, nine, or 23 age categories).
-- Data for both detailed and regional groups. The Proof of Concept includes tables for both detailed and regional groups based on the adaptive population thresholds. Detailed groups include disaggregated groups such as Puerto Rican, Chinese, the Navajo Nation, Samoan, etc., while regional groups include groups such as Caribbean, East Asian, American Indian, Polynesian, etc.
The Detailed DHC-A data come with limitations that are important for data users to understand. As outlined in greater detail in the Proof of Concept, these limitations include:
-- We caution data users against aggregating data, such as geographic levels, age categories, and detailed racial or ethnicity groups, as the data generally will be less accurate. We recommend using data for specific tables and geographies as they are published in the data product.
-- Data are not always geographically consistent.The SafeTab-P algorithm generates a count for each geography individually. This means that values from lower-level geographies may not sum to values from higher-level geographies.
-- Data are not always consistent between data products. For reasons described above, Detailed DHC-A data will not always “match” comparable data from the Redistricting File or Demographic and Housing Characteristics File (DHC).
Your feedback will be instrumental in making final improvements to the Detailed DHC-A. Data users are invited to send feedback on the Detailed DHC-A Proof of Concept to 2020DAS@census.gov with the subject “2020 Census Detailed DHC-A.” To meet our 2020 Census Detailed DHC-A production release date of August 2023, all feedback must be received by March 2, 2023. If you do not want us to publish your feedback or if you want us to remove your identifying information, please indicate that in the email.
We would like to hear from you about how your work will be positively, negatively, or not impacted if we release the final product using the current specifications outlined above. Please provide the level of geography, description of the use case(s), and implications should the data be released as reflected in this Proof of Concept.
Proof of Concept: https://www2.census.gov/programs-surveys/decennial/2020/program-management/data-product-planning/2010-demonstration-data-products/03-Detailed_DHC-A/2023-01-31
News release: https://www.census.gov/programs-surveys/decennial-census/decade/2020/planning-management/process/disclosure-avoidance/newsletters/released-today-detailed-dhc-a-proof-of-concept.html