The U.S. Census Bureau is a federal statistical agency that is tasked with producing statistics on demographic and economic characteristics across the United States and, in some instances, its territories. The Census Bureau collects data using several methods, including surveying respondents directly and using administrative data (also referred to as administrative records). These data are then analyzed to produce federal statistics.
Administrative data are data that were not generally collected for the purposes of producing federal statistics. Instead, these data are collected and maintained by federal agencies in order to "administer (or run) programs and provide services to the public."
The Census Bureau has increasingly used administrative data to reduce respondent burden and data-collection costs, create new data products, and increase the research potential of census data.
Members of Congress may have an interest in how the Census Bureau uses administrative data as it pertains to cost savings, quality of data and its fitness for purpose, and confidentiality of the administrative data sources.
Administrative data are generally not collected through surveys and are produced for nonstatistical purposes. Rather, administrative data are collected to support administering some kind of program or service, such as mailing addresses linked to individuals for the purpose of mail delivery by the United States Postal Service (USPS).
Other examples of administrative data include but are not limited to individual data, such as social security data and income tax return data; business identifying information, such as descriptions of operating, financial, ownership, and other characteristics; and address data, such as latitude or longitude points, telephone numbers, and street addresses.
The Census Bureau is authorized under Title 13, Section 6, of the U.S. Code to acquire and use administrative data from other federal departments, agencies, or establishments as well as "private persons and agencies" when these data are consistent with the kind, timeliness, quality, and scope of statistics required.
Additionally, the Privacy Act of 1974 (5 U.S.C. ยง552a(b)(4)) permits federal agencies to share records concerning a U.S. citizen or lawful resident with the Census Bureau solely for statistical purposes.
In 2014, the Office of Management and Budget (OMB) established guidance for federal agencies to follow when working with administrative data for statistical purposes.
The latest release of the Census Bureau's statistical quality standards provides standards the agency follows when acquiring and using administrative data. These requirements are broadly that the Census Bureau
Additionally, the Census Bureau established some guiding principles to follow when considering using administrative data in the American Community Survey (ACS), one of the bureau's largest surveys, in a 2018 report. Below is a selection of those principles.
Table 1. Census Bureau Statistical Quality Standards for Using Administrative Data for Enhancing the ACS
|
Requirement |
Description |
|
Coverage |
How comprehensive is the coverage of the administrative data with respect to geographies and population subgroups? |
|
Quality |
What is sufficient data quality for the published estimates? Do the administrative data meet these quality requirements? |
|
Conceptual Alignment |
Do the administrative data correspond to the concept the ACS currently intends to measure? |
|
Temporal Alignment |
Do the administrative data correspond to the time period referenced in the ACS? |
|
Impacts on Estimates |
To what extent does the administrative data source diverge from survey response (direct impact)? Do the differences carry over to other variables, for example, through editing and imputation (indirect impact)? |
Source: U.S. Census Bureau, Realizing the Promise of Administrative Data for Enhancing the American Community Survey, November 2018.
The Census Bureau considers these guiding principles when evaluating administrative data sources for the ACS. Many of these administrative data contain some demographic information on the individuals being covered in surveys. Particularly, some relatively fixed demographic characteristics, such as sex and date of birth, will have high conceptual and temporal alignments. Meanwhile, there may be uses for administrative data with less-fixed characteristics, such as incomes.
Other demographic characteristics, such as race and ethnicity, are collected using different methods that vary from one source to another. Some sources collect race and ethnicity in a mutually inclusive manner (e.g., the individual is both Black Alone and Hispanic), whereas others treat these characteristics as mutually exclusive (e.g., Black Alone or Hispanic).
The availability of administrative data from certain programs is not necessarily consistent or complete across different geographies. With differential coverage across geographic areas, the Census Bureau cannot equally use administrative data within all areas and housing units, limiting the potential for its universal use in products.
During the 2020 census, administrative data were used to support activities prior to data collection, enumerate households, process data, and evaluate data quality.
The Census Bureau maintains a Master Address File (MAF), which is intended to contain an up-to-date inventory of all known living quarters in the United States and Puerto Rico. The MAF is updated using USPS administrative data.
During data collection for the decennial census, administrative data supported quality assurance by being matched against actual survey responses. Additionally, households that did not respond after multiple visits from enumerators were enumerated by using administrative data if the Census Bureau considered these data reliable. When necessary, administrative data also supported group quarters enumeration by utilizing preexisting information from the given facility, such as expected resident count or maximum capacity.
As a last resort after completing all other means of data collection, the Census Bureau will impute missing information from households. Imputation is a statistical technique where missing data are replaced with substituted values. First the Census Bureau conducts count imputation to resolve whether households were occupied and how many people occupied them, followed by characteristic imputation to fill in missing demographic characteristics. The imputation process uses administrative data from several sources, including the United States Postal Service (USPS), Internal Revenue Service (IRS), and Social Security Administration (SSA).
To evaluate the quality of the decennial census, the Census Bureau used administrative data, among other data sources, to conduct the Demographic Analysis (DA) program. The DA uses vital records, international migration data, and Medicare records to produce national estimates.
Using administrative data to support the production of federal statistics might be an effective supplement to in-person enumeration operations. Use of such data might also alleviate the ongoing decline in survey response rates. Members of Congress may have an interest in the Census Bureau's use of administrative data in its surveys as it pertains to cost, data quality, and access.
The Census Bureau found that fieldwork operations were the most expensive aspects of the 2020 decennial census. The Census Bureau estimated it could save $1.4 billion by using administrative data in the 2020 decennial census, as reported by the Government Accountability Office (GAO) in 2015. A 2021 JASON report found that administrative records accounted for a lower-than-expected amount of the nonresponse follow-up workload. Members of Congress may consider requesting analysis of the final cost savings from the use of the administrative data during the 2020 decennial census and the potential cost savings for the 2030 census, as well as other surveys that could utilize administrative data. Members might approve costs to effectively test and implement use of administrative data to maximize usefulness and overall cost savings.
The Census Bureau aims to use administrative data to improve existing data quality, expand upon information currently available in Census products, and reduce response burden. With these aims in mind, Members may wish to direct the bureau to report to them on any limitations associated with using administrative data.
Members may also consider directing reporting on the various sources of administrative data used, data quality of these sources across agencies, and what may be done to encourage consistent data quality. To encourage public support, Congress may also want to ensure sound confidentiality practices are followed by the Census Bureau when using these administrative data.