In the Office of Homeland Security Statistics (OHSS), we follow all laws and regulations to protect the privacy of the public. We work with the Department of Homeland Security’s (DHS) Office of Privacy to review and clear how we receive and process data from other agencies. These agencies obtain data from members of the public. They generally get the data through applications for immigration benefits, sponsorship petitions, or enforcement-related documentation. More information on DHS privacy protection is available on the Office of Privacy website.
We ensure our public data do not pose an undue risk of identifying individuals whose data are in our products by:
- Striving to meet all the principles and practices outlined in the National Academies’ Principles and Practices for a Federal Statistical Agency. These principles include best practices for protecting the confidentiality of data providers’ information.
- Continually reviewing other federal statistical offices’ disclosure prevention methods.
- Participating in the Federal Committee on Statistical Methodology’s Disclosure Review Officer (DRO) Council.
- Keeping up on disclosure best practices outlined by the American Statistical Association's Committee on Privacy and Confidentiality.
Rounding Policy (2023 - Present)
Generally, for any data regarding individuals, we round the cells to the nearest ten. This includes the data cells, subtotals, and totals. This means cells do not always add up to the totals.
We round cells to the nearest ten in the common way. We round 0-4 to 0; 5-14 to 10; 15-24 to 20; … and 1,213,067 to 1,213,070.
We round to make it difficult to identify the original, exact counts within small cells. Small cells pose the risk of identifying individuals.
Cell Suppression Policy (Before 2023)
We adopted the current policy of rounding in 2023. We used a different data disclosure avoidance method for earlier products. In the previous system, we didn’t round. Zeroes remained zeroes (often shown as a “-“ dash). However, we replaced all other cells below a certain threshold (usually 3) with a “D” for “Data Disclosure Threshold Not Met.” We also suppressed other cells when the count in those cells could be used to calculate the exact value of small cells.
Reducing Small Cells
In some cases, we may construct tables to reduce the number of smaller cells. This might include only breaking out larger categories or rolling smaller categories into an “All Other” category. We base this on a data-specific threshold for each table or may present only a certain number of the largest categories (for example, the top 50).
