Five Situations to Watch to Ensure Trusted, Quality Federal Statistics

Coauthored with Connie Citro on behalf of our project team.

Federal statistics are essential US infrastructure supporting our economy, governance, and society. To fulfill their critical role, they must be relevant, timely, credible, accurate, and objective, not only in practice but also as required by the Foundations for Evidence-Based Policymaking Act of 2018 (Evidence Act). Moreover, trust in federal statistical agencies is paramount; any perception that government statistics have been influenced by an administration may undermine their credibility.

With the rapid transformation of the federal government underway, federal statistical agencies are likely to be impacted, either as collateral damage or as targets, and either flagrantly or more subtly. The impacts may be temporary, long-term, or permanent. The American Statistical Association-George Mason University project on the health of the principal federal statistical agencies is monitoring five types of actions that could undermine the quality or trustworthiness of federal statistics: (1) cuts in statistical programs (e.g., the Common Core of Data [CCD]); (2) delays, reduction of detail, or cancellations of data products (e.g. the, Digest of Education Statistics); (3) decreases in budget or staffing; (4) politicizing leadership and undermining staff security; and (5) accessing and using statistical data for nonstatistical purposes. We are watching other indicators of the health of the federal statistical agencies but are giving emphasis to the five described below. To see what we’ve observed so far, or to learn how to contribute, visit our monitoring page. See also our inaugural report, The Nation’s Data at Risk, issued in July 2024

?

Cuts in statistical programs

Statistical programs (e.g., ongoing surveys or administrative records programs) may outlive their usefulness, but abrupt termination of major programs without considered review and consultations with data users, OMB, and Congress is a disservice to policy makers and the public. Abrupt termination flouts Congressional authority, which has appropriated funds for the programs and which has explicitly authorized many of them. In addition, many statistical programs are needed to support other statistical programs (e.g., the Census Bureau’s annual population estimates provide denominators for birth and death rates and many other important socioeconomic indicators). Consequently, cutting a program in one statistical agency may damage important programs in other agencies.

?

Delays, reduced detail, or cancellations of data products

Federal statistical agencies are required to post well in advance and adhere to schedules for release of their data products (e.g., indicators, reports) to foster trust and objectivity. Statistical Policy Directive No. 3 covers schedules and release procedures for Principal Federal Economic Indicators (PFEIs), such as the unemployment and inflation rates, which move markets as well as serving many other important purposes. Statistical Policy Directive No. 4 covers schedules for other data products (e.g., income and poverty reports, criminal victimization reports). Agencies must give notice if there are delays. Delays are concerning whether they are ordered by an administration (e.g., delaying a report until after an election) or whether budget and/or staffing reductions have hampered an agency’s ability to be timely.?

Reduction in detail of a data product is concerning when the reduction is due to undue interference by an administration (e.g., to eliminate data for specific population groups) or to budget and/or staffing reductions that make it not possible to provide detail (e.g., an agency may lack funds to maintain survey response rates and therefore not be able to publish detail due to smaller effective sample sizes).?

Elimination of data products is concerning for the same reasons, particularly if there is no opportunity for public comment and consideration of trade-offs (e.g., cutting a less-used product in order to keep or expand a more relevant product). The removal of taxpayer-funded data along with resources to help data users access and understand the data contradicts the need for federal statistics to be accessible and transparent to data users.

?

Resources: Decreases in budgets or staffing

A decline in agency staffing can be reason for concern because the work of federal statistical agencies is labor-intensive. Based on our report in the summer of 2024, we know that at least a few agencies were understaffed to an extent that affected their ability to maintain their programs, not to mention their ability to innovate. Moreover, 11 of the 13 principal statistical agencies have lost purchasing power in the last 15 years, and a majority of the agencies have lost more than 14%. In contrast, federal discretionary, nondefense spending—after accounting for inflation—increased 16%. The National Center for Education Statistics and the Bureau of Economic Analysis have had to cut products in recent years, and, as of January, the Bureau of Labor Statistics had to cut product detail.?

It’s important to note that program cuts in a constrained budget environment are generally exercised after other deferments have already been executed. IT upgrades are often the first to be deferred followed by program updates and redesign. Next, data purchases are skipped followed by sample-size cuts. The last steps to be taken are generally program terminations and staff furloughs. Our point is that program cuts like those made at BEA and BLS mean the agency’s efforts to innovate in order to improve efficiency and data quality, and reduce respondent burden, were already sacrificed.?

?

Undermining leadership and staff security

Professional leadership and staff who are able to design and provide objective, trustworthy, high-quality statistics are essential to enable federal statistical agencies to carry out their mission to serve the public good. Personnel actions that can undermine this mission include:

• Abrupt, wholesale reductions in staff. Such reductions, which send a signal to staff that they are not valued, are concerning given that data collection, analysis, and dissemination are labor- and expertise-intensive activities and the need to maintain institutional knowledge.?
• Replacing the head of an agency with a person who does not meet the high professional and statistical qualifications.. Such a step undermines trustworthiness and gives the impression that an administration wishes to improperly influence official government statistics.?
• Classifying some or all of an agency’s senior staff as Schedule Policy/Career (formerly known as Schedule F), who serve at the pleasure of an administration. Such an action undermines trustworthiness and puts the Policy/Career-designated staff in a difficult position of possibly having to comply with an executive order that undercuts their agency’s mission to be objective.?
• Adding political appointees to a statistical agency’s staff beyond what is traditional (e.g., heads of BJS, BLS, Census Bureau, EIA, and NCES, and, for Census, the communication director). Such a step undermines objectivity and gives the impression that an administration wishes to improperly influence official government statistics unless the need for and role of the political employee and that he or she will not interfere with an agency staff’s professional judgment are clearly stated.

Accessing and using statistical data for nonstatistical purposes*

Statistical agencies collect data under strict statutory protections, violations of which are felonies punishable by steep fines and prison terms that were established in the Privacy Act of 1974 and the Confidential Information Protection and Statistical Efficiency Act of 2002, and reaffirmed in the Evidence Act that President Trump signed into law in January 2019. The purpose of such protections is, as stated by the Evidence Act, to fulfill the requirement that statistical agencies “protect the trust of information providers by ensuring the confidentiality and exclusive statistical use of their responses” (in contrast to such nonstatistical uses as conferral of benefits or sanctions on individuals). The law also requires the head of the host agency for the statistical agency (e.g., the Department of Labor for the Bureau of Labor Statistics) to “enable, support, and facilitate statistical agencies or units in carrying out” the confidentiality and other responsibilities of the statistical agencies. ? Because the data collected by statistical agencies include sensitive personal and business information, the agencies have strict protocols and sophisticated protections that have proven, time-tested effectiveness to prevent any breaches of sensitive information. ? Access to these data by individuals outside of the federal statistical agency who have not been approved to use the data for statistical purposes in a secure environment violates federal law. More importantly, the American people’s trust in statistical agencies would be compromised by outside, unauthorized access to these data. Without this trust that has been earned and nurtured over decades, statistical agencies will not be able to collect the data needed to produce trusted, quality statistics. Because of survey-response, sample-size, and other challenges for almost all statistical agencies,?many statistical products, including BLS’s monthly employment situation releases, would be put at risk by outside, unauthorized access to statistical-agency data.

*This entry was adapted from a February 10 LinkedIn post by ASA Executive Director Ron Wasserstein.