On November 2, 2026, the National Academy of Education (NAEd) urged the U.S. Census Bureau to withdraw a proposed rule that would depart from every census since 1790, both in whom it counts and in what it asks. The rule would remove the race and ethnicity questions from the decennial census, prohibit any question about sexual orientation, and exclude from the apportionment count every foreign citizen who is not a lawful permanent resident. The Bureau has also raised the prospect of adding an untested question about legal status, although its own research shows that such questions drive down response. The new residence criteria and a legal-status question would make the census less accurate, and children, whom the census already counts least well, would bear much of the harm. As the NAEd comment explains, “The nation cannot determine whether it is treating its children equally if it chooses not to count them accurately.”

Federal, state, and local governments rely on census data to distribute federal education funds, to enforce civil rights protections, and to plan for the students they serve. Researchers rely on these data to determine whether those public investments reach children of every background and whether every child has a fair opportunity to learn and to thrive. Drawing on decades of research, much of it by the Census Bureau’s own researchers, the NAEd comment shows that the proposed rule would harm children and the schools that serve them on multiple fronts:

• Lost evidence on race, ethnicity, and educational opportunity. The proposed rule bans any questions about race or ethnicity. Census race and ethnicity data allowed researchers to show that, even among children whose parents had the same income, Black boys have lower incomes in adulthood than White boys in 99 percent of census tracts. The census also provides the most complete count of the American Indian and Alaska Native children whom the federal government has a trust responsibility to educate. No other source can replace that count. Administrative records match the self-reported race of American Indians and Alaska Natives only about three-quarters of the time, and tribal leaders have told the Bureau that enrollment records “cannot capture the full Native population, including descendants, citizens of multiple Tribal Nations, and Native peoples otherwise not formally enrolled.” Federal agencies also rely on census race and ethnicity data to enforce civil rights laws in education.
• No substitute for census race and ethnicity data. The Bureau suggests that administrative records could supply race and ethnicity data, but Census Bureau researchers found that these records held no race information for about 20 percent of individuals, roughly 69 million people. Children ages 0 to 17 were nearly 17 times more likely than adults ages 18 to 44 to lack race information. The American Community Survey cannot fill the gap either, because the Bureau calibrates its race and ethnicity estimates to the decennial census.
• Residents left out of the count. The census should count everyone who lives in the United States, regardless of citizenship or immigration status. Census data informs community’s needs such as schools, roads, and hospitals, so a count that leaves out any group misstates the needs of the places where its members live. Every census since 1790 has sought, in the words of former Census Director Robert Groves, “a complete enumeration of all persons, residents, in the country.”
• More children not counted. The 2020 Census already missed about one million children ages 0 to 4, an undercount of 5.46 percent. The rule’s legal-status standard would likely signal to immigrant families that answering the census carries risk. The Bureau does not explain how the standard would apply to the more than 6.3 million children, over five million of them U.S. citizens, who live with at least one parent who lacks legal status. Census Bureau research shows that legal-status inquiries depress response with Bureau researchers predicting that a citizenship question alone would reduce self-response by 8.0 percentage points more in households that may include noncitizens than in other households.
• Misdirected education funds. In fiscal year 2021, more than $2.8 trillion in federal funds flowed through at least 353 programs that use census-derived data, including Title I, special education grants under the Individuals with Disabilities Education Act, the National School Lunch Program, and Head Start. Children whom the 2030 Census misses would remain missing from these funding formulas for the next decade.

The NAEd also explains how the rule fails to reflect the reasoned decision-making that the Administrative Procedure Act requires. It ignores the many uses of census data and the states, school districts, tribal nations, and researchers who rely on them. It cites response burden as a reason to remove the race and ethnicity questions, yet the studies it cites do not evaluate those questions, and it considers adding a legal-status question that the Bureau’s own research indicates would reduce response far more. The Bureau’s concerns about disclosure risk call for better disclosure avoidance methods, not less data. The rule would permanently ban a question about sexual orientation that has never appeared on the census questionnaire and that the Bureau has never tested for that purpose. And it disregards Congress’s direction in Public Law 94-311 to collect and publish statistics on Americans of Spanish origin or descent. The NAEd urges the Census Bureau to withdraw the proposed rule in its entirety and offers its assistance in strengthening the 2030 Census.

Read the Full Comment

For more information, contact Amy I. Berman, Deputy Director, National Academy of Education, at aberman@naeducation.org.