Population Estimates Sign-on Letter to the Census Bureau

View a PDF of the letter here.

Dear Associate Director Velkoff,

We write in appreciation of the progress already made and the plans in place to increase the accuracy of the Population Estimates in the wake of the 2020 Census. Following the challenges in 2020, and the increased differential undercounts of people of color, rural residents, young children and renters, we are pleased and gratified to see the whole of government consensus supporting the Census Bureau’s plans to modernize this critical dataset.

Although the reasons for developing the blended base initially were due to the unavailability of certain data sources from the 2020 Census, the impact of this work has been fantastic and is a credit to the agile, innovative staff of the bureau. Mitigating for the undercount of young children and correcting for the overcount of other groups are very positive steps. With increased support and dedication to innovate, much more can be done to increase the accuracy of the base on which the annual estimates are built.

We are encouraged that the Population Estimates Program (PEP) is focused on the use of additional administrative data sources to improve the base. We recommend that research focus on mitigating the differential undercounts, which are the most consequential issues in the decennial census data included in the current blended base. While continuing work on the undercount of young children is necessary, changes in the base should be expanded to fix more known problems. Specifically, the Base Evaluation and Research Team (BERT) should have a strong focus on the undercount of people of color, rural residents and renters.

Analyses of the 2020 Census confirm that undercounts of renters, Black and Hispanic populations were particularly bad in 2020. Although the differential undercount gap between rural and urban communities was not measured for 2020, rural areas have been consistently undercounted since at least 1960.

Mitigating these issues will require innovative research, which the Estimates Program and BERT are well positioned to do, and also the willingness to make data-driven decisions. Specifically, to mitigate the undercounts of children, people of color, renters, and rural residents, the blended base will need updated 2020 total resident estimates for both counties and states. We understand how big of a decision this is for the Census Bureau, and it is imperative in order to increase the accuracy of this essential data which determines the flow of over $2 Trillion in federal funding each year.

The Estimates Program should continue to research all viable options for making these improvements, and continue to engage experts throughout the scientific and advocacy communities. Here we provide several recommendations for lines of research.

First, we are glad to learn that you plan to use data sources even if they are not available for the whole population. We also urge the use of data sources that are not available for all of the country – either because they don’t cover the whole country, or because they are not shared by all jurisdictions.

One example here is the school enrollment data. Some states are known for the completeness of their student rosters, with solid coverage of even the home schooled and private school populations. Other states have much less even coverage. Working with the states with known complete data to start, and augmenting other states with additional data sets would be a great order of operations and provide an incentive for states to improve their school enrollment data coverage. Further, most states not only collect these data as counts, but also capture all the demographic characteristics needed for the Population Estimates. These data also have the potential to cover most if not all of the population of interest, including both traditionally easy to enumerate populations and populations that are persistently undercounted.

Another example is the use of data from Medicaid, SNAP and WIC. While these data sources will never cover the whole population, their coverage in some areas is highly concentrated. Finding statistical methods to estimate undercoverage is a viable option for estimating inflation factors in certain areas using these data sources. Fortunately, Medicaid and SNAP data for 2020 is likely to include expanded coverage of the population, as benefits were expanded in both programs in response to the COVID-19 pandemic.

Second, we are encouraged by how the great work of the Continuous Count Study and the Real Time Census Simulation could be used to inform the research that the BERT team is doing. We are sure this is already something that is on the radar of the PEP team, expanding upon these research projects with an eye towards if and how they can be molded to fit the needs of the postcensal estimates products. The sheer number of administrative data sources, matching processes, and analysis that has already been undertaken cannot and should not be left to languish.

For instance, during the recent CSAC meeting, the presentation on the Continuous Count Study the tremendous progress that has been made on a program that could be a great counterpart for PEP in testing and helping to productionalize the use of further administrative records, as well as modeling approaches, for the use in the blended base. The Continuous Count Study’s reliance on the FRAMES project and the MAF also lends the opportunity to adjust the housing base for subcounty estimates and the population derived therefrom.

Overall, the Census Bureau has a whole of government consensus behind its existing plans to improve the accuracy of the annual Population Estimates, with a significant investment and direction from the Administration and Congress to proceed. This presents an historic opportunity. We and the community of stakeholders and statistical experts here not only applaud this important work, but stand ready to support it in any way possible.


Allison Plyer
Co-Chair of the CQR Task Force
Chief Demographer, The Data Center

Cara Brumfield
Co-Chair of the CQR Task Force
Director of Income and Work Supports,
The Center for Law and Social Policy

The Leadership Conference on Civil and Human Rights
Asian Americans Advancing Justice | AAJC
Georgetown Center on Poverty and Inequality
Coalition on Human Needs
Partnership for America’s Children
Deborah Stein, Senior Consultant to Coalition on Human Needs for Count All Kids
Bill O’Hare, President, O’Hare Data and Demographic Services LLC