Federal Data Infrastructure — Evidence Act, Chief Data Officers & data.gov

The Foundations for Evidence-Based Policymaking Act of 2018 (Evidence Act) is the most consequential federal data governance law since the Paperwork Reduction Act of 1980 — and most Americans, including most federal employees, have never heard of it. It created an entirely new data architecture across the federal government: Chief Data Officers (CDOs) at every major agency, Evaluation Officers responsible for independent program evaluation, Statistical Officials as liaisons to OMB data standards, and a framework for linking administrative data across agencies to answer policy questions that no single agency's data can answer alone. The law also authorized the National Secure Data Service — a Census-piloted system for privacy-preserving research data linkage — and strengthened the confidentiality protections in CIPSEA that became the legal battleground when DOGE sought access to IRS, SSA, and BLS microdata in 2025. Five years into implementation, most CFO Act agencies have CDOs and Learning Agendas on paper; the harder problem — actually using administrative data to improve policy — remains largely aspirational.

Legal Authority

• P.L. 115-435 (Foundations for Evidence-Based Policymaking Act of 2018) — The Evidence Act: requires CFO Act agencies to appoint Chief Data Officers, Evaluation Officers, and Statistical Officials; mandates four-year Learning Agendas; authorizes the National Secure Data Service for privacy-preserving research data linkage
• 44 U.S.C. §§ 3520, 3520A — CDO mandate and CDO Council: requires each agency to designate a Chief Data Officer and establishes the interagency Chief Data Officers Council at OMB
• 44 U.S.C. §§ 3561–3583 — CIPSEA (Confidential Information Protection and Statistical Efficiency Act): protects statistical data collected from individuals and businesses from non-statistical use; became a legal battleground in 2025 when DOGE sought access to IRS and SSA microdata
• 44 U.S.C. § 3511 — OMB statistical and science policy authority: empowers OMB to set government-wide data standards and coordinate the federal statistical system

Key Mechanics

The Evidence Act created a new governance layer across all 24 CFO Act agencies: each must have a Chief Data Officer (CDO) to manage data assets, an Evaluation Officer to plan and conduct program evaluations, and a Statistical Official to coordinate with OMB's data standards. Every four years, agencies must publish a Learning Agenda — a public plan identifying the key policy questions the agency will try to answer and what data and evaluations it will use. These Learning Agendas are the mechanism intended to connect data infrastructure to actual policy improvement. The CDO Council (chaired by OMB's federal CDO) coordinates across agencies on data standards, data sharing agreements, and interoperability. CIPSEA's confidentiality protections, which restrict how statistical data collected for surveys can be used for administrative or law enforcement purposes, became directly relevant in 2025 when DOGE requested access to IRS microdata for cross-agency matching — raising questions about whether the statutory protections barred the access regardless of executive directive.

How It Works

The Evidence Act Framework

The Foundations for Evidence-Based Policymaking Act was the product of a bipartisan commission chaired by economist Katharine Abraham and management scholar Ron Haskins, whose 2017 report documented that the federal government was making trillion-dollar policy decisions with little systematic evidence about what worked. The commission found that data existed within the government to answer most major policy questions — but agencies couldn't link it, couldn't share it, and didn't have the institutional capacity to use it for evaluation.

The Act responded with four structural requirements at every CFO Act agency (the 24 largest cabinet departments and major agencies):

Chief Data Officer (CDO): Mandated by 44 U.S.C. § 3520, the CDO is responsible for the agency's data governance framework, data quality standards, data catalog and inventory, data-sharing agreements, and coordination with other agency data users. CDOs report to agency leadership (typically the Deputy Secretary or CIO) and are members of the government-wide CDO Council coordinated by OMB. The CDO role was new to the federal government; most CDOs were hired from 2019–2021 as agencies stood up the function. CDO authority varies enormously across agencies — at some, the CDO directs significant resources; at others, it is a coordination role with no direct program staff.

Evaluation Officer: Each CFO Act agency must designate an Evaluation Officer — a senior official responsible for building the agency's capacity to conduct and use rigorous program evaluation. The Evaluation Officer must have relevant expertise and is responsible for developing the agency's Learning Agenda, coordinating with OMB, and ensuring that program offices have access to evaluation capacity.

Statistical Official: Designated as the agency's liaison to OMB's Office of the Statistical and Science Policy (OSSP). The Statistical Official ensures the agency complies with OMB statistical standards (including SPD 1 on independence) and coordinates the collection and dissemination of statistical data.

Learning Agenda (Evidence-Building Plan): Each CFO Act agency must develop a 4-year Learning Agenda — a systematic plan for what policy questions the agency will try to answer with evidence, what data and evaluation methods it will use, and what resources it will commit. Learning Agendas must be published, updated annually, and aligned with the agency's strategic plan under 5 U.S.C. § 306. The first generation of Learning Agendas (2022–2026) was widely criticized as aspirational rather than operational — listing broad questions without funded evaluation plans to answer them.

Federal Data Strategy

The Federal Data Strategy (2020–2030) is a government-wide framework published by OMB and the CDO Council, organized around 40 practices and 20 action steps across four principles: ethical governance, conscious design, learning culture, and data valuation. Annual action plans specify which agencies are responsible for which actions and the timeline for implementation. The Federal Data Strategy is a coordination framework, not a law — agencies are expected to incorporate it into their data governance plans.

data.gov

data.gov is the federal government's interagency open data catalog, managed by GSA's Technology Transformation Services (TTS). It indexes approximately 300,000 datasets from 80+ federal agencies — from EPA air quality monitoring data to USDA crop statistics to HUD housing data. Each dataset entry includes metadata (agency, format, update frequency, license, geographic coverage) and links to the original data source or download.

data.gov itself is primarily a discovery and metadata layer — it links to datasets hosted on agency websites rather than hosting them directly. The underlying open data mandate derives from OMB Memorandum M-13-13 (Open Data Policy, 2013), which directed agencies to publish data in machine-readable formats by default. As of 2025, data.gov coverage is uneven: some agencies publish extensive, well-documented datasets; others have outdated or sparse catalogs.

National Secure Data Service (NSDS)

The National Secure Data Service — authorized by the Evidence Act and piloted at the Census Bureau — is the federal government's attempt to enable privacy-preserving administrative data linkage for research and evaluation purposes. The concept: researchers could submit approved queries to the NSDS, which would link records across agency databases (tax, benefits, health, education) and return statistical results without ever exposing individual-level data to the researcher.

The NSDS addresses a longstanding problem: the best evidence about policy effectiveness often requires linking data across agencies (e.g., linking workforce training program records to IRS earnings data to measure employment outcomes), but each agency's CIPSEA or Privacy Act obligations prohibit sharing individual-level data. The NSDS would perform the linkage within a secure environment, returning only aggregate or de-identified research outputs.

As of 2025, the NSDS pilot has demonstrated technical feasibility but remains limited in scope and has not achieved the scale envisioned in the Evidence Act commission's report.

DOGE Data Access and the Evidence Act

The 2025 DOGE initiative's attempts to access federal administrative data — IRS taxpayer records, SSA wage histories, HHS claims data, BLS survey microdata, Education Department student records — put the Evidence Act's CIPSEA protections at the center of legal and political conflict.

DOGE's stated rationale was fraud detection and program integrity — identifying duplicate benefits, unauthorized recipients, and eligibility mismatches across programs. Opponents argued that:

1. CIPSEA's non-statistical use prohibition bars statistical agency data collected under confidentiality pledges from being used for any non-statistical purpose, including fraud detection (which is a regulatory/enforcement function, not a statistical one).
2. The Privacy Act (5 U.S.C. § 552a) prohibits federal agencies from sharing individually identifiable records for purposes incompatible with the original collection purpose without individual consent or statutory authority.
3. 26 U.S.C. § 6103 provides especially strong protections for federal tax return information, limiting disclosure to specifically authorized purposes and persons.

Federal courts issued temporary restraining orders in early 2025 limiting DOGE access to SSA records and IRS data pending further litigation. The cases established that DOGE personnel were not clearly authorized under existing law to access records protected by these statutes.

Federal Statistical Research Data Centers

The Federal Statistical Research Data Centers (FSRDCs) are a network of ~30 secure computing facilities at universities and research institutions where approved researchers can access restricted-use statistical microdata — including confidential survey responses, linked administrative records, and other data protected under CIPSEA — without the data leaving the secure federal computing environment. Researchers apply for access, have their proposals reviewed by the relevant statistical agency, and work on-site at FSRDC terminals under strict output-review protocols.

The FSRDC network enables the kind of research linking that the NSDS would eventually automate — but through a human-mediated, application-by-application process. Leading economics and public health research has been produced through FSRDCs, including studies linking Census demographic data to IRS earnings records, Medicaid claims to birth records, and Medicare data to environmental exposure information.

How It Affects You

If you are a citizen or consumer: The Evidence Act creates the legal framework for the federal government to learn whether programs are working — whether job training programs lead to higher wages, whether early education investments reduce crime, whether housing assistance improves health outcomes. Without systematic evaluation, federal programs can continue for decades regardless of effectiveness. The open data mandate means the raw material for this kind of civic accountability analysis is increasingly publicly available.

If you are a business, researcher, or analyst: data.gov is a starting point for federal data discovery; the underlying agency APIs and data portals contain the actual datasets. For research requiring restricted-use data, the FSRDC network provides access through an application process — each major statistical agency's website lists FSRDC access procedures. The CDO Council publishes a government-wide data asset inventory annually that catalogs all significant federal data holdings.

If you work at a federal agency: The Evidence Act's CDO, Evaluation Officer, and Statistical Official requirements are statutory — not discretionary. OMB guidance (M-19-18, M-20-12) specifies implementation expectations. Learning Agendas are required to be published on agency websites and submitted to OMB. Data sharing between agencies requires formal data use agreements that must address Privacy Act and CIPSEA compliance — the CDO is typically responsible for negotiating and overseeing these agreements.

If you are a journalist or policy analyst: OMB publishes an annual report on Evidence Act implementation that is the best overview of which agencies have made progress and which lag. data.gov's agency pages show which departments publish robust open data. The DOGE data access litigation documents (filed in federal district courts in 2025) are the most detailed public exposition of the CIPSEA legal framework and its limits. For FSRDC research, the Census Bureau's Center for Economic Studies publishes a bibliography of research produced through the FSRDC network.

Recent Developments

• 2025 — DOGE data access requests triggered federal court TROs citing CIPSEA, Privacy Act, and 26 U.S.C. § 6103; litigation defined the legal limits of cross-agency administrative data access for non-statistical purposes.
• 2024 — First full Learning Agenda cycle (2022–2026) hit the midpoint; OMB evaluation found most agencies had published Learning Agendas but fewer than half had funded evaluation plans with measurable milestones.
• 2023 — CDO Council published the Federal Data Strategy 2023 Action Plan; NSDS pilot at Census expanded to include linked IRS-Census earnings data for approved external researchers.
• 2022 — FedRAMP Authorization Act codified cloud security standards applicable to data infrastructure hosting.
• 2021 — First cohort of CFO Act agency Chief Data Officers onboarded; CDO Council established at OMB; data.gov underwent major infrastructure modernization.
• 2018 — Foundations for Evidence-Based Policymaking Act enacted (P.L. 115-435); created CDO mandate, Evaluation Officer, Statistical Official, Learning Agenda requirements, and CIPSEA II.