The 2020 Census is the first U.S. census to use administrative records (ARs) to enumerate some households. The innovation used ARs to classify some addresses as Occupied, Vacant, or Nonresidential and to create a roster of the residents with their characteristics for some of the addresses classified as Occupied. The more recent U.S. censuses also have used ARs in some aspects of census-taking, but not for the enumeration of households. Other statistical programs at the U.S. Census Bureau have used and continue to use ARs to support their data collection operations. In this document, the term “administrative records” refers to data collected by governmental or nongovernmental agencies while administering a program or service.

Advances in computer technology and software have enabled the expanded use of ARs. In addition, the Office of Management and Budget (OMB) has encouraged federal agencies to leverage the ARs collected by operations and services in their statistical programs (U.S. Office of Management and Budget 2014). The advances and support of OMB also have led to the development of agreements between the Census Bureau and other government agencies that permit the use of their ARs for enumeration of households under specified conditions. In the nongovernmental sector, new technology and software have enabled the construction of commercially available databases that contain information that may or may not overlap with data available to the Census Bureau in ARs housed in government agencies.

The Census Bureau began a research program in 2012 to pursue its goal of identifying a methodology for using administrative records to reduce the cost of the 2020 Census Nonresponse Followup (NRFU) operation while preserving data quality. NRFU is the census field operation where enumerators visit addresses that did not submit a self-response and attempt to obtain an interview. Not surprisingly, NRFU was one of the most expensive operations in the 2010 Census. Improving the efficiency of the NRFU operation was viewed as critical to the goal of a cost-effective census in 2020 (U.S. Census Bureau (2019a, 2019b).

This document discusses the use of ARs in the enumeration of households at some addresses in NRFU. A brief overview of AR enumeration may be found in the blog post by Mule (2021). In contrast, this document provides a high-level discussion that includes more details regarding the evolution of the methodology for AR enumeration and its implementation in the 2020 Census. Not covered is the use of records maintained by group quarters administrators who use their records to enumerate the people residing in their facilities. For information about the enumeration of group quarters facilities, see Stempowski and Christy (2021), U.S. Census Bureau (2017).

In particular, the discussion in this document focuses on the research that led to the methodology for using ARs to enumerate the households at some addresses in NRFU and the implementation in the 2020 Census. These topics include the following:

• Section 1 is the introduction to the document.
• Section 2 has a brief overview of administrative records at the Census Bureau that includes historical and recent uses, data quality, and data protection.
• Section 3 contains a description of the research and development that occurred from 2012 through 2018 to prepare for using administrative records in census enumeration and includes the investigation of the feasibility, identification of a strategy, census tests, and lessons learned.
• Section 4 has a description of the models used in the administrative records enumeration process.
• Section 5 contains the original plan for using administrative records in 2020 Census enumeration.
• Section 6 discusses the unexpected delay in the 2020 Census data collection and processing because of the COVID-19 pandemic and the adjustments to the schedule.
• Section 7 focuses on the modifications and adaptations required to cope with the unforeseen disruptions in the implementation of the 2020 Census because of the COVID-19 pandemic.
• Section 8 contains a summary. • An Appendix contains two tables, one with the planned timeframe for the administrative records processing and another with the revised timeframe.

Throughout this document, the descriptions of the research and development include the rationale behind the resulting decisions. By focusing on data quality during each phase of the research, development, and implementation, the chosen methodology was able to employ ARs to improve the quality of the census enumeration. This investment proved fortuitous when the pandemic and a series of natural disasters disrupted traditional NRFU operations. Without the strategic use of administrative records, the imputation rate for the 2020 Census would have been higher.