Formal Privacy Methods for the 2020 Census

In preparation for the 2020 decennial census, the Census Bureau asked JASON to examine the scientific validity of the vulnerability that the Census Bureau discovered in its traditional approach to Disclosure Avoidance, the methods used to protect the confidentiality of respondent data. To address this vulnerability, the Census Bureau will employ differential privacy, a mathematically rigorous formal approach to managing disclosure risk. JASON judges that the analysis of the vulnerability performed by Census is scientifically valid. The use of Differential Privacy in protecting respondent data leads to the need to balance statistical accuracy with privacy loss. JASON discusses this trade-off and provides suggestions for its management.

Related Information

Page Last Revised - October 28, 2021