Skip to content

2025 Data Integrity Summit Report

  • by

Quick Summary of the Report

The federal statistical system has been struggling with budget constraints for several years. In the first six months of 2025 threats to the integrity of the system increased substantially, including cuts to staffing and contracts, data removals, revisions to data collections, and delays and missed deadlines in the publication of key information. Day 1 of the Data Integrity Summit brought together data experts and users across the U.S. to strategize about recovering from the current disruptions and rebuilding an efficient federal data infrastructure for the years ahead. Day 2 took a deeper dive into issues of transparency and public trust.

Recurring Themes

Across the two days, four key themes emerged.

Transparency and public trust are paramount

Across all breakout sessions, transparency and public trust were identified as critical considerations for any data system changes. Clear communication (from planning through publication), broad public access to both data and documentation, and robust protections against misuse are essential to maintain public confidence.

Data quality is key

A recurring theme was the challenge of tracking and ensuring data quality in new or consolidated systems. This involves defining new metrics, conducting robust testing (e.g., parallel runs for surveys, comparing administrative data to traditional methods), providing error metrics, and publishing clear documentation on data methods and limitations.

Education is needed

While participants were all deeply knowledgeable about threats to the federal statistical system, many data users are a step (or more) removed from knowledge about data infrastructure and the current threats to data access and quality. More education is needed at several levels including among elected officials, membership organizations, and data users.

Privacy protection is critical

Given the increasing use of linked survey and administrative data, as well as changes in the way the government is using data, many participants called for enhanced privacy laws, strict data use agreements, and robust oversight mechanisms. This includes ethical/privacy review boards, community-informed oversight, and increased penalties for breaches to protect against misuse by government agencies or third parties.

Overall Action Items and Next Steps

There are specific actions identified for each discussion above, with themes that recur throughout. APDU’s key action items fall into categories that mirror the key themes above.

Transparency and Public Trust

Short-term Actions

  • Identify and collaborate with business ambassador(s) to help advocate for data for businesses and for workforce development.
  • Develop a “roadshow” or “plug-and-play” presentation that anyone could give that can be tailored to local governments, philanthropic organizations, industry groups / associations, and others.
  • Define transparency in a way that data producers, data consumers, and the general public can understand.
  • Consider using the advisory committee model as a mechanism for keeping up pressure for transparency in decision-making.
  • Track instances where regulations are not being followed and put pressure on OMB.
  • Facilitate introductions between recently-released federal staff and journalists by making introductions, hosting mixers, and/or providing media training to former fed staff.
  • Connect with Congresspeople who are engaged on transparency issues now.
  • Include transparency and public trust in education and outreach activities (described below).

Long-term Actions

  • Push for transparency at every step of the data lifecycle–from planning through publication.
  • Catalogue and start to work through the hard questions:
    • How do we define transparency?
    • What do we do when transparency goes bad?
    • What do we do when there is malicious compliance?
    • When is transparency not enough?
    • When is transparency not the goal?
    • How should we be crafting transparency strategies?
    • What should the standards be for transparency?
  • Propose a definition of and standards for data transparency, broadly. And also…
    • Propose standards for transparency around linked data, such as (but not limited to) minimum reporting standards.
    • Propose standards for transparency about the implementation of AI/ML tools. Most of these are “black box,” so there needs to be robust processes (like making training data available) to ensure transparency.
    • Under this umbrella, there is also a need to support additional research into the implications of ongoing changes such as the use of AI and ML, use of administrative data, consolidation across surveys, implementation of privacy-enhancing technologies, and the associated and implications of changes.
  • Support the “Stat Influencer” idea for Tik Tok to influence younger generation.

Data Quality

  • Ensure that any changes to data collection are based on a scientifically rigorous implementation strategy (and technical plan to permit benchmarking, quality assessment).
  • For AI/ML learn more about what has already been done within the federal statistical system and what are the lessons learned that could inform future ethical use.
  • Also identify the goal(s) of any changes–are the goals to reduce respondent burden? Reduce cost? Improve accuracy? Improve timeliness or granularity? Improve privacy? Be prepared to demand that those goals are clear at the outset and that there is evidence that the change(s) will support the desired outcome(s) before implementation.

Education and Outreach

Develop a strategic communications plan, that may include:

  • Segment the strategy by audience.
    • Key audiences include: businesses, foundations, state/local governments, and academics.
  • Include a risk assessment (will communications help or hinder progress).
  • Hold bilateral meetings with left, right and center think tanks to find common ground.
  • Conduct research into what messaging strategies are effective in this environment, and if we find nothing, hold focus groups to learn more. Subsequently, conduct training sessions for the media and data stakeholders on best practices.
  • Update secondary school and college curriculum standards to make data literacy a core requirement.

Specific actions may include:

  • Hold a convening and invite agency staff so we can learn what they need and want. Also to connect federal agency staff with data users so we can turn data users into data advocates.
  • Compile primary uses for key statistics/data elements so that we can reference these as we develop messaging; flesh out use cases for specific audiences. (Editor’s note: Work of this type is underway–APDU’s role may be to support and uplift those efforts.)
  • “Take APDU out to the streets,” not just a yearly DC conference.
  • Continue to host events where data users can learn about the latest state of affairs, share information, and strategize.
  • Conduct outreach to partner and peer organizations (like NASBO, NCSL, and others) to raise their awareness of ongoing changes in the federal statistical system.
  • Develop an open-access slide deck that anyone can use to share the latest on what’s going on with public data.
    • COMPLETE – You can access the slide deck here.

Privacy protection

  • Outline a plan for how enhanced privacy standards and oversight would function (work with Representative Trahan’s office).
  • Support efforts from community-led organizations to make sure all voices are heard.