in their domains and they must have a clear mission (see Practice 1). A major advantage of the decentralized federal statistical system in the United States is that separate federal statistical agencies are located in appropriate departments and are closer to the policymakers and programs in those areas. However, this can also present a challenge when different agencies produce different statistical estimates of the same or similar phenomena. It is essential that statistical agencies coordinate and collaborate with one another (see Practice 7) to ensure that coherent and consistent statistical information is provided on major policy issues, such as the federal collection and dissemination of race and ethnicity data (see Appendix A).

To ensure that they are providing relevant information, statistical agencies need to reach out to a wide range of their data users, including staff in their own departments and other federal departments who use their data, members of Congress and congressional staff, state and local government agencies, academic researchers, businesses and other organizations, organized constituent groups, and the media.

Agencies may need to expend considerable energy to open avenues of communication more broadly with current and potential users (see Practice 9). The recently introduced requirement that federal agencies create learning agendas could assist statistical agencies in prioritizing their outreach activities (Commission on Evidence-Based Policymaking, 2017; Foundations for Evidence-Based Policymaking Act of 2018, 2019; Office of Management and Budget [OMB], 2019b).

Statistical agencies have used a variety of approaches to engage with users. Advisory committees, such as the Census Bureau’s National Advisory Committee, are one tool to obtain the views of users outside a statistical agency (National Research Council [NRC], 1993b, 2007a).¹ Many agencies obtain advice from committees that are chartered under the Federal Advisory Committee Act.² Some agencies obtain advice from committees and working groups that are organized by an independent association, such as the American Statistical Association’s Committee on Energy Statistics for the Energy Information Administration. Regardless of the exact organization of these advisory committees, statistical agencies should examine the

___________________

¹ Some statistical agencies had advisory committees that were later disbanded by their departments. This has occurred for the Bureau of Transportation Statistics and the National Center for Education Statistics.

² Examples include: the Board of Scientific Counselors for the National Center for Health Statistics; the Data Users Advisory Committee and the Technical Advisory Committee for the Bureau of Labor Statistics; the Census Scientific Advisory Committee and the National Advisory Committee for the Census Bureau; and the Federal Economic Statistics Advisory Committee, which provides substantive and technical advice cutting across the major economic statistics programs of three agencies: the Bureau of Economic Analysis, the Bureau of Labor Statistics, and the Census Bureau. See https://apps.bea.gov/fesac/. See also PL-92-463.

institutional affiliation and occupation of advisory committee members to ensure equitable coverage in representation.

Additionally, holding workshops and conferences for data users or engaging with them at professional conferences are valuable activities for facilitating interchange among users and agency staff (NRC, 2013c). Online mechanisms, such as blogs and web surveys, may assist a statistical agency in obtaining input from users (see Practice 9). Similarly, agencies can use web analytics to better understand their user base and to assess the accessibility and usability of their website and data products. Offering positions to data users as fellows or temporary employees can also help a statistical agency gain a richer perspective on user interests and concerns.

Statistical agencies should periodically review their data collection programs and products to make sure they remain relevant (see Practice 6). Relevance should be assessed, not only for particular programs or closely related sets of programs but also for an agency’s complete portfolio, to assist each agency in making the best choices among program priorities given available resources.

To increase data quality and relevance, an agency’s own staff should actively analyze its data (Martin, 1981; Norwood, 1975; Triplett, 1991). Such analyses may examine correlates of key social or economic phenomena or study the statistical error properties of the data. Carrying out such work can lead to improvements in the quality of the statistics, to the identification of new needs for information and data products, to a reordering of priorities, and to a deeper understanding of data users’ needs (see Practice 5).

The substantive analyses that statistical agencies produce as a regular part of their dissemination and research activities will likely be helpful to policy analysis units in their departments, as well as other data users. These analyses typically describe relevant conditions and trends over time and across geographic areas and population groups (e.g., high school completion rates by race, poverty rates for each year, or state variation in employment rates). A statistical agency may expand upon such initial analyses at the request of a policy analysis unit or other data user, for example by examining trends for particular population groups, while being mindful of the agency’s responsibility to remain independent from undue influence (Principle 4). Further, statistical agencies have an obligation to provide useful statistical information to Evaluation Officers to conduct their work.

However, statistical agencies should be careful not to become involved with policy development or implementation (beyond policies directly affecting their operations), because these activities could affect their ability (or the perception of their ability) to conduct impartial and objective statistical activities. Examples of policies that are appropriate for statistical agencies to help develop and implement include statistical system policies, federal data quality standards, research access, privacy protections, and

information technology security protocols—all as they relate to statistical operations (see Practices 2, 3, 4 and 8). Beyond such exceptions, a statistical agency should neither make policy recommendations nor conduct substantive analyses of policies, although it may advise on the availability and strengths and limitations of relevant information in a policy-neutral manner. The distinction between analysis consistent with the mission of a statistical agency and policy analysis is not always clear, and a statistical agency must carefully consider the extent of policy-related activities that are appropriate for it to undertake to maintain its primary mission of providing impartial and objective statistical information for public use (see Practices 1 and 2). Principle 1 is summarized in Box 3-1, below.

PRINCIPLE 2: CREDIBILITY AMONG DATA USERS AND STAKEHOLDERS

The value of a statistical agency rests fundamentally on the accuracy and credibility of its data products. Because few data users have the resources to verify the accuracy of statistical information, users rely on an agency’s reputation to disseminate high-quality, objective, and useful statistics in an impartial manner. Only if its products are viewed as credible can an agency be regarded as working in the national interest, not beholden to a particular set of users (Ryten, 1990; see Practice 2).

Credibility therefore stems from the respect and trust of users and stakeholders in the statistical agency. Agencies build this respect and trust, not only by producing accurate and objective data and meeting all of their deadlines for the release of their statistics, but also by adhering to the

other principles for federal statistical agencies and by following some key practices. When different agencies produce similar or related estimates, it is essential that statistical agencies coordinate and collaborate with one another (see Practice 7) to ensure that users understand the differences and can determine which data are most useful for their needs.

Agencies build and maintain respect and trust through clear public commitments to professional practice and transparency in all that they do. For example, statistical agencies should actively engage with users in determining priorities for data collection and analysis; make their data widely available on an equal basis to all users, formally and informally (see Practice 9); conduct research to improve efficiency and data quality (see Practices 3 and 5); and fully inform users of the strengths and limitations of the data (see Practice 10). Such activities demonstrate an agency’s respect toward, and openness with, its users and stakeholders.

A statistical agency’s website is a key vehicle for conveying not only its statistical data, but also key information about its data. Providing clear and easy access for users to locate, work with, and understand the strengths and limitations of the agency’s data is a vital part of an agency’s mission (see Practice 1) and requires ongoing efforts to continue to meet users’ and stakeholders’ evolving needs. An agency’s website can enhance its credibility by providing information about its policies for data access (e.g., explaining which tables and microdata files are publicly available and which data require approval to access in secure sites to protect confidentiality; see Practice 8); scientific integrity policies; standards for data quality and for documenting sources of error in data collections and estimation models (see Practice 3); procedures and schedules for the release of new and continuing data series; procedures for timely notice of errors and corrections to previously released data; procedures and schedules for archiving historical data; and documentation of ongoing research efforts to provide accurate statistics that meet users’ needs (see Practice 5). To support its credibility, it is essential that each statistical agency website be clearly distinguishable from its associated parent agency, particularly if that parent agency is a regulatory, direct service, or policy development agency.³

To keep up with an ever-changing society and technology and thereby maintain the trust of their users and stakeholders, statistical agencies need to recruit, develop, and retain high-quality professional staff who are dedicated to providing high-quality products and upholding high ethical standards (see Practice 4). Statistical agencies also need to regularly review and evaluate

___________________

³ Note that the Trust Regulation authorizes each recognized statistical agency or unit to maintain a distinctive, outward-facing website with its own domain name and with adequate control over the website content and management to uphold the fundamental responsibilities. See final § 1321.4(e)(3) and Appendix A.

their programs and share the results of these evaluations with their users and stakeholders (see Practice 6). Principle 2 is summarized in Box 3-2, below.

PRINCIPLE 3: TRUST AMONG THE PUBLIC AND DATA SUBJECTS

Nearly every day of the year, individuals, household members, businesses, state and local governments, and other organizations provide information about themselves when requested by federal statistical agencies. Without the cooperation of these data subjects, federal statistical agencies could produce very little useful statistical information.

Some information provided is required by law or regulation for government tax and transfer programs, such as reports of employers’ wages to state employment security agencies or payments to program beneficiaries. A small number of federal statistical surveys are so important that participation is mandatory. But most of the data come from the voluntary cooperation of respondents. In all cases, the willing cooperation of data subjects reduces costs and promotes accuracy.

Because virtually every person, household, business, state or local government, and organization is the subject of some federal statistics, public trust is essential for the continued effectiveness of federal statistical agencies. Individuals and entities providing data directly or indirectly to federal statistical agencies must trust that the agencies will appropriately

handle and protect their information. Implicitly and explicitly, they expect the following:

1. The agency’s computer systems will not be hacked and their information taken;
2. Their information will not be accessed and used by others for nonstatistical purposes;
3. The statistical agency has a lawful, legitimate need for each piece of information it is requesting;
4. The statistical agency will use their information to create useful statistics;
5. The statistical agency will not release or publish their information in identifiable form;
6. The statistical agency will control who is allowed to see and access their information; and
7. The statistical agency will evaluate and update its systems for using and protecting data subjects’ information to take account of new information needs, new threats to confidentiality, and new technological developments that offer improved protection.

Federal statistical agencies not only have legal and ethical obligations that require them to fulfill these expectations; they also have the obligation to effectively communicate how they fulfill them. Consistent ethical conduct on the part of a statistical agency is critical for obtaining the trust of the general public and of data subjects, whether those subjects are individuals, organizational entities, or custodians of administrative records. Statistical agencies should coordinate and collaborate with each other to ensure that their communications and internal practices are clear and consistent, as this will strengthen the trust of data subjects who may interact with more than one agency (see Practice 7).

Data subjects must trust that the information the agency seeks is important for the government to collect and is being collected in a competent manner, for the good of the larger society, and only for the purposes that the agency has described (see Practice 2). To engender trust, a statistical agency should also respect the privacy of data subjects in other ways and ensure that each individual’s consent to respond to a survey is given knowingly and with full information. Agencies should describe the intended and likely future uses of the data being collected, the data’s relevance for important public purposes, and the extent of confidentiality protection that will be provided. Agencies should minimize the intrusiveness of questions and the effort needed to respond to them, and they should seek administrative or other non-survey sources to fulfill needs consistent with each agency’s requirements for information (see Practices 5 and 8). Trust among data

subjects also requires that an agency treat respondents with courtesy in appreciation for their time (National Academies of Sciences, Engineering, and Medicine [NASEM], 2016a).

The mission of federal statistical agencies is to produce statistical information by aggregating the data provided by individuals, businesses, or other entities. These agencies pledge to use the information they collect only for statistical purposes, not to provide individual records for any administrative, regulatory, or judicial use, and to make every effort to protect the confidentiality of individual information in the data they publish. This pledge is supported by many statistical agencies’ individual statutes as well as the Confidential Information Protection and Statistical Efficiency Act of 2018 (Title III of Foundations for Evidence-Based Policymaking Act of 2018, 2019). (See Appendix A.)

Data subjects must be able to trust that a statistical agency will scrupulously honor its pledge of confidentiality and will fulfill the expectations noted above. Earning this trust, however, goes beyond what the agency is simply required to do by law, and recognizes that there are many potential threats, some outside the control of the statistical agency, that the agency must anticipate and guard against. In the world of “big data” and the “dark web,” agencies must guard against the use of external data to re-identify information provided by individuals. Agencies must consistently innovate in privacy-protecting technologies to protect—to the extent possible—against re-identification of individual records in statistical data products. (See NASEM, 2017a,c, 2024c; and Practice 8.)

When data are obtained from the administrative records of other federal, state, or local government agencies or other third-party providers, the same requirements of trust building apply to justify their cooperation.⁴ Data-providing organizations need to trust that their records are important and legitimate for a statistical agency to obtain, that their own restrictions on data access will be honored, and that the statistical agency will make every effort to minimize their burden in responding.

An effective statistical agency has policies and practices to instill the highest possible commitment to professional ethics among its staff and build a culture of confidentiality. When issues arise or guidance is unclear, it must be able to rely on its staff to keep this culture (see Practices 3 and 4).

We end by acknowledging that statistical agencies’ mandate to establish and maintain the trust of data subjects, holders, and users is made more difficult when overall trust in government data and government agencies

___________________

⁴ See (OMB, 2014a) in Appendix A, which asserts the legitimacy and benefits of use of administrative data from other federal agencies for statistical agency purposes. It also provides guidance for best practices and procedures to engender mutual respect and trust and facilitate such data sharing.

is waning (American Statistical Association, 2024; NASEM, 2023b). Some of this diminished trust is due to disparagement of particular statistical releases or agencies by public figures. An unfavorable climate raises the need for statistical agencies to focus intently on practices that help to build and maintain trust. Principle 3 is summarized in Box 3-3, below.

PRINCIPLE 4: INDEPENDENCE FROM POLITICAL AND OTHER UNDUE EXTERNAL INFLUENCE

A statistical agency must be impartial and execute its mission without being subject to pressures to advance any political or personal agenda. It must avoid even the appearance that its collection, analysis, or reporting processes might be manipulated for political or other purposes. Only in this way can a statistical agency serve as a trustworthy source of objective, relevant, accurate, and timely information (Bohman, 2024; Bowen, 2023; Citro et al., 2023; Cohen, 2023; Habermann & Louis, 2020; Habermann et al., 2023).⁵ Their independence and their high-quality products also allow statistical agencies to perform another increasingly important service for the country: they assist and improve citizens’ understanding of statistical products amid an avalanche of information of varied provenance and quality (American Statistical Association, 2024; Behzad et al., 2023).

___________________

⁵ The importance of independence from undue external influence is reflected in national statute and regulation, as well as international guidance. See (Foundations for Evidence-Based Policymaking Act of 2018, 2019; OMB, 2014b) in Appendix A and (European Statistical System Committee, 2017; United Nations Statistical Commission, 2014) in Appendix C.

___________________

⁶ The heads of the Bureau of Justice Statistics (BJS), the Bureau of Transportation Statistics (BTS), and the National Center for Education Statistics (NCES) had been appointed by the President and confirmed by the Senate in the past; these positions were changed, two to become Presidential appointments without Senate confirmation (BJS and NCES) and one to become a career civil servant (BTS).

the leader’s term not coincide with the presidential term to better ensure that professional criteria, rather than political ones, guide the appointment process. Appointment by the President with Senate confirmation for a term that is at the pleasure of the President, as is the case for the head of the Energy Information Administration (EIA), provides less assurance of independence (however, it is worth noting that EIA does have other strong legislative protection for the authority of its administrator). Appointment of a qualified career civil servant as the head of an agency is another method considered helpful for maintaining the independence of a statistical agency.⁷

Having its agency head report directly to the secretary of the department can also be helpful for a statistical agency to maintain a position of independence from political or other undue external influence. Such access allows the head to inform new secretaries about the role of their statistical agency and to directly present the case for new or changed statistical initiatives. Such direct access currently is provided by legislation for BLS and EIA. Other statistical agencies have one or more layers of departmental management between the statistical agency head and the secretary (see Figure B-2 in Appendix B). Over time there has been an increase in the “layering of statistical agencies,” that is, positioning them lower in their department’s administrative structure, a trend that the National Research Council (NRC, 2009c, p. 226) has identified as “a subtle, but increasingly common” threat to independence because it increases the number of political appointees and career staff who could seek to exercise control over the agency without transparency to external stakeholders or users (Habermann et al., 2023; Hartman et al., 2014).

Another means to protect against political and other undue external interference is for the statistical agency to have its own funding appropriation from Congress separate from that for other departmental agencies or programs.⁸ (See Practice 2.) This provides greater visibility and accountability to Congress, both by the agency and by its department, something

___________________

⁷ Agencies headed by career civil servants, many of whom hold their positions for long periods of time, include the Bureau of Economic Analysis; the Bureau of Transportation Statistics; the Economic Research Service in the U.S. Department of Agriculture; the National Agricultural Statistics Service; the National Center for Health Statistics; the National Center for Science and Engineering Statistics (NCSES); the Office of Research, Evaluation, and Statistics (ORES) in the Social Security Administration; and the Statistics of Income (SOI) Division in the Internal Revenue Service.

⁸ Most, but not all, federal statistical agencies receive appropriations directly. Three federal statistical agencies—National Center for Science and Engineering Statistics, Statistics of Income, and Office of Research, Evaluation, and Statistics—receive funding through appropriations to their parent agencies. In addition, some federal statistical agencies, such as NCES, are not involved in the annual budget request and communication process.

that is reinforced when the statistical agency participates in appropriations briefings. Other funding arrangements, such as the statistical agency being completely dependent on allocations from the budget of its parent department or agency, risk giving the department a great deal of leverage over the statistical agency without transparency to external stakeholders and users, and potentially compromising its ability to fulfill its mission.

A key aspect of a federal statistical agency’s mission is its ability to release its statistical products without review or approval by policy officials outside the statistical agency (see Practice 2). Some agencies have this authority spelled out in statute, while others have departmental policies that support it. OMB provides governmentwide protocols and assurances on the release of key federal statistics and publishes in advance a release calendar for the entire year for Principal Federal Economic Indicators.⁹ A strong internal and external evaluation program (see Practice 6) can also help ensure that all agency statistical programs are adhering to standard procedures and are not manipulated.

In the long run, the effectiveness of an agency depends on its reputation for impartiality: thus, an agency must be continually alert to possible infringements on its credibility and be prepared to strenuously resist such infringements (Bohman, 2024). OMB has stated that it is also the responsibility of the statistical agency’s department to “enable, support, and facilitate federal statistical agencies and recognized statistical units” as they implement their responsibilities to produce objective data.¹⁰ A federal statistical agency that has the respect and trust of its stakeholders and users, who can help publicly defend the agency, is better equipped to ward off or resist attempts by others to exert political or other undue external influence on the agency (see Principle 3). Within an agency, the professional staff’s adherence to the mission of the agency and its quality standards and ethical principles (see Practices 3 and 4) is also key to the agency’s preservation of its independence from political or other undue external interference.

Over the past decade, the federal government has issued a series of policies and frameworks to reaffirm and further support scientific integrity, including statistical production. These policies aim to protect against undue external influence in scientific processes by specifically rejecting the suppression or alteration of scientific findings by political officials; requiring transparency and accessibility of scientific findings; and requiring the selection of policy staff in scientific roles to be based on scientific and technological knowledge, credentials, experience, and integrity. (See Appendix A.) Principle 4 is summarized in Box 3-4, below.

___________________

⁹ See OMB’s Statistical Policy Directives No. 3 (OMB, 1985) and No. 4 (OMB, 2008) in Appendix A.

¹⁰ See OMB’s Statistical Policy Directive No. 1 (OMB, 2014b) in Appendix A.

___________________

¹¹ See for example https://www.statspolicy.gov/assets/docs/ICSP-The%20Use%20of%20Private%20Datasets%20by%20Federal%20Statistical%20Programs-1-6-2023.pdf.