Invariants Set for 2020 Census Data Products
On November 24th, the Census Bureau’s Data Stewardship Executive Policy Committee (DSEP) finalized the list of “invariants” for the first set of 2020 Census data products. Invariants are statistics that are published without noise infusion.
The first set of data products includes the P.L. 94-171 file (redistricting data), the Demographic Profiles, the Demographic and Housing Characteristics File, and the Congressional District Summary File. These products will be protected using the main TopDown Algorithm (TDA) central to the Disclosure Avoidance System (DAS). We will employ other formal privacy methods to handle the remaining more detailed data tables separately to preserve the greatest degree of accuracy, as they pose difficult and unique privacy challenges.
Per the decision, the following statistics for this first set of products will be invariant at these levels of geography and higher:
- Total population (at the state and state-equivalents level (the District of Columbia and Puerto Rico)
- Total housing units (at the census block level)
- Number of group quarters facilities by type (at the census block level)
Although state-level data for American Indian and Alaska Native (AIAN) tribal areas will not be invariant, the Census Bureau will ensure that the statistics will be equally as accurate, if not more so, as those released in the most recent Privacy-Protected Microdata Files (PPMF) (v. 2020-11-16), and as reflected in its associated Detailed Summary Metrics.
Our recent research verifies that this level of accuracy is achievable by allocating a separate, larger, dedicated privacy-loss budget for AIAN tribal area data and placing the data on the geographic hierarchy spine.
This treatment has other advantages over a state-level invariant. First, it ensures improved accuracy for AIAN tribal areas comparable to an invariant without undermining the privacy guarantee or jeopardizing system stability. Just as important, it better protects the confidentiality of tribal area residents in those states with very small tribal area populations, bringing those populations under the umbrella of guaranteed, quantifiable confidentiality protection.
The DSEP will consider and determine privacy-loss budgets for 2020 Census products at a separate meeting closer to the ultimate data release date.
This DSEP decision will be formally documented and posted on the 2020 Census Memorandum Series page.
Was this forwarded to you?
Sign up to receive your own copy!
Useful Links:
Have Suggestions?
Do you have specific questions you'd like us to answer in this newsletter or topics you'd like discussed? Send us an email at 2020DAS@census.gov and let us know!
|
