Federal Data Users

 View Only
  • 1.  Big news on disclosure avoidance

    Posted 9 days ago

    The Commerce Department just issued an order banning "noise infusion"-the practice of adding small random values to published statistics to protect the confidentiality of census respondents. Going forward, the Census Bureau must use "coarsening" (rounding, aggregating, suppressing data) instead.

    Lots of unanswered questions-including why the Commerce Department made this call, whether coarsening alone can meet the Bureau's statutory confidentiality obligations under Title 13, and what it means for small-area data users. Worth noting: the Census Bureau's own research found that older methods like data swapping  weren't designed to defend against modern database reconstruction and reidentification attacks, which have grown more sophisticated since 2020.

    📰 Hansi Lo Wang has the story: https://hansilowang.com/ 📄 Full order: DAO 216-26 📚 New to this topic? PRB has a good explainer: Disclosure Avoidance in the 2020 Census 📘 Why the Census Bureau chose differential privacy: Census Bureau brief



    ------------------------------
    Mark Mather
    Associate VP
    PRB
    ------------------------------


  • 2.  RE: Big news on disclosure avoidance

    Posted 2 days ago

    NPR posted a helpful explainer over the weekend

    https://www.npr.org/2026/06/12/nx-s1-5855734/census-bureau-data-differential-privacy

    (Full transparency: one of the people quoted may be familiar. But the story provides helpful context even without my minor contribution.)



    ------------------------------
    Beth Jarosz
    ------------------------------



  • 3.  RE: Big news on disclosure avoidance

    Posted 22 hours ago

    Spread the word!  Also, does anyone know if the Joint Statement below is online anywhere?  The message from APDU is too long to fit on some social media platforms and I really want to refer people to the full message.

    APC, APDU, COPAFS, ICPSR, and PAA Statement on Commerce's "Disclosure Avoidance for Statistical Products"

    On June 4, 2026 the Department of Commerce issued a new order "Disclosure Avoidance for Statistical Products" (DAO 216-26) that limits the types of privacy protection methods that the Census Bureau and Bureau of Economic Analysis (BEA) can use for their data products.

    This order subverts processes developed over decades to foster transparency and public trust and creates a scenario in which there will either be less privacy for our personal information, or less usable data, or both

    This order raises several concerns:

    Lack of Transparency Undermines Integrity

    Federal law and policy provide guardrails by which statistical agencies work on behalf of the American people. As a result, agencies use established methods and processes to produce accurate information that respects individual privacy and protects confidentiality.

    Specifically, there were robust processes in place to update and refine privacy methods, based on federal expertise, with oversight from external watchdogs, and bolstered by multiple opportunities for public review and input. This order circumvents both the expertise and transparency that the public deserves, and undermines the integrity of well-established processes.

    Moreover, this order is in conflict with the Executive Order on Restoring Gold Standard Science, which requires federal government scientific endeavors to be, among other things, transparent, collaborative and interdisciplinary, subject to unbiased peer review, and without conflicts of interest.

    Privacy, Confidentiality, and Data Quality Are at Risk

    Limiting the Census Bureau and BEA to two of the oldest methods of privacy protection (rather than the full arsenal of other privacy-enhancing technologies that account for the rapidly evolving threats to privacy posed by modern computing and AI) will make it more difficult for these agencies to safeguard the privacy and confidentiality of the sensitive data they hold on each and every person, household, and business in the U.S.

    By limiting the Census Bureau and BEA to only two methods of privacy protection, the order also curtails the agencies' ability to produce detailed, decision-level information. The rule limits the agencies to "coarsening" (rounding, aggregating, and reporting in ranges) or suppression (not reporting). In other words, the now-mandated way to protect privacy and confidentiality is to reduce the amount of information that's published. This is particularly concerning for information we've come to rely on for neighborhoods and rural communities which are at most risk of being aggregated or suppressed.

    Moreover, while the order is clear that standard methods are banned, parts of the order are vague and are causing widespread confusion. We know, for example, that widely-used products like Longitudinal Employer-Household Dynamics are affected, which means that crucial information for workforce development, veteran outreach, and transportation planning is at risk. However, the text of the order is so vague that the full scope of products affected is unclear. And we do not know how the order will be carried out. Will county data products now only be released at the state level? Will products be suspended completely? Will detail continue to be published, but put our privacy at risk?

    A Ban Is a Bad Idea

    What we do know is that the result of this ill-conceived order will be either less privacy for our personal information, or less usable data, or both.

    Narrowing the scope of the order so that it affects just a few products also does not solve the problem. No decisions about our privacy should be made behind closed doors, without expert input, and without public input.

    Moreover, this order attempts to "solve" a problem that already had a solution. The Office of Management and Budget's 2025 Open Data Guidance already provides specifics on how agencies can and should balance the tensions between privacy and accuracy in data products. In an age of rapidly advancing technologies, constraining agencies' abilities to react in kind to such developments represents a hamstringing of capabilities resulting in an unnecessary restriction on the types of data available for evidence-based decision-making.

    The American people deserve accountability, transparency, and decisions made based on expertise not politics.

    What can you do?

    Contact your rep.

    We urge you to contact your representatives in Congress, regardless of their party affiliation, and let them know your concerns. Calling is most effective and takes just two minutes of your time.

    You can find contact information here:

    https://www.congress.gov/members/find-your-member

    Please share.

    Please also share this statement and call to action with others who may be impacted, particularly those who may not yet understand the magnitude of what has occurred.



    ------------------------------
    Michele Hayslett
    UNC Libraries
    ------------------------------



  • 4.  RE: Big news on disclosure avoidance

    Posted 20 hours ago
    Edited by Mark Mather 20 hours ago

    Hi Michele, Yes the joint statement is posted here:
    https://copafs.org/apc-apdu-copafs-icpsr-and-paa-statement-on-commerces-disclosure-avoidance-for-statistical-products/



    ------------------------------
    Mark Mather
    Associate VP
    PRB
    ------------------------------



  • 5.  RE: Big news on disclosure avoidance

    Posted 2 hours ago
    Edited by Beth Jarosz 2 hours ago

    Thank you, Michele! It's also posted on APDU's page: https://apdu.org/joint-statement-on-commerces-disclosure-avoidance-for-statistical-products/



    ------------------------------
    Beth Jarosz
    ------------------------------