The Net Undercount of Children Under 5 Years of Age in the Decennial Census: An Art of the Possible Use Case

The undercount of young children, ages 0-4, is a persistent problem in the decennial census and surveys. It is the largest net undercount of any age group. This Art of the Possible Use Case identifies and proposes an approach to fill this data gap. Extending this to a demonstration use case would help to identify and build Curated Data Enterprise (CDE) capabilities to address this gap. The CDE approach goes beyond statistical products derived from single surveys, employing data re-use and integration to allow for accurate, timelier, and granular statistical products that help to fill data gaps and address pressing societal needs.
Since 1980, the deficit of young children has increased relative to adults, reaching an all-time high of 5.4 percent in 2020. Despite the best efforts of the Census Bureau and its local partners, this problem persists and is intractable using current methods. Finding ways to address this coverage error is important, given that funding for early childhood programs is linked to an accurate count of young children. Also, the undercount of young children affects the accuracy and timeliness of the decennial census overall. This paper proposes a new source of data to enhance the counts of children 0-4 years of age as part of the demographic frame being constructed by the Census Bureau. State-based Early Childhood Integrated Data Systems (ECIDS) collect, store, integrate, and maintain administrative data from early childhood programs across multiple agencies within a state. Analysis of these data informs policies, practices, and programs. ECIDS have become more prevalent as computational and data science skills facilitate the creation and use of these data systems. ECIDS offer a potential source of data for identifying young children who are missed in the census and surveys.

Salvo J, Lancaster V, Shipp S. The Net Undercount of Children Under 5 Years of Age in the Decennial Census: An Art of the Possible Use Case. Proceedings of the Biocomplexity Institute, Technical Report. TR# BI-2023-259. University of Virginia; 2023 October. DOI: https://doi.org/10.18130/nzyj-m621.

Funding: This research was sponsored by the U.S. Census Bureau Agreement No. 01-21-MOU-06 and Alfred P. Sloan Foundation grant, Grant No. # G-2022-19536

Acknowledgments: The authors would like to acknowledge the contribution of Ford Heilizer, who provided invaluable assistance with analyzing state administrative data on young children. Ford conducted this work while an intern at the Biocomplexity Institute and is now studying Computer Science at the University of Southern California. In addition, University of Virginia undergraduate fellows, Stephanie Zhang and Olivia Davis, provided valuable assistance with the manuscript.

We would also like to thank our colleagues: John Thompson, former Director of the U.S. Census Bureau; William O’Hare of O’Hare Data and Demographic Services; David Dolson, former Director of Statistics Canada; Matthew Snipp, Stanford University; Joan Naymark, Minnesotans for the ACS and 2030 Census (MACS) and Zhengyuan Zhu, University of Iowa, for their comments on earlier drafts of this paper. We would also like to thank our UC Census Bureau colleagues: Jennifer Ortman, Principal Demographer; Eric Jensen, Senior Advisor for Population Estimates and Coverage Measurement; and Sallie Keller, Chief Scientist and Associate Director Research and Methodology Division. (The views expressed in the report are those of the authors and not the Census Bureau.)