Seven reasons dropping census questions would leave the US data-blind

The Trump administration's proposal for the 2030 census would stop asking Americans about their race and ethnicity, questions the census has asked in some form since 1790. The Heritage Foundation's Project 2025 agenda had urged a review of the race questions, citing "concerns among conservatives that the data under Biden Administration proposals could be skewed to bolster progressive political agendas." Dropping the questions altogether carries a different cost, a kind of data blindness: once a question leaves the census, the country loses the ability to see what it measured, and getting it back takes years.

Here is what goes dark when a question is cut, using the U.S. Census Bureau's own explanations of why each question is asked.

1. The laws that depend on the answers don't go away

Several federal laws rely on census race data. The bureau's page on why it asks about race lists the uses: monitoring compliance with the Voting Rights Act and enforcing bilingual election requirements, enforcing equal employment opportunity under the Civil Rights Act of 1964, identifying groups that may not be getting needed medical services under the Public Health Service Act, and allocating money to school districts for bilingual services.

Dropping the question doesn't repeal any of those laws. It removes the evidence used to apply them. A court weighing a Voting Rights Act claim, an agency checking hiring patterns, or a school district applying for bilingual funding would be working without the numbers it uses today.

2. Every district map starts with these counts

After each census, the bureau sends states the redistricting files known as the P.L. 94-171 summary files, which include counts by race and Hispanic origin down to the block level. States draw congressional and legislative districts from them, and the same counts are used in voting rights cases over whether districts dilute minority votes. Without the question, the 2030 files would arrive with the population counts but not the detail that districting fights turn on.

3. It breaks a 240-year record

The census has recorded race in every count since the first one, though the categories have changed many times, as we traced in how the census has counted race since 1790. A single missing decade would break that series for good. Researchers measuring whether gaps in income, homeownership, or health are closing would have no 2030 starting point to compare against.

The gap would spread beyond the census itself. The bureau's annual population estimates by race and Hispanic origin, published each June, start from the most recent census count, so a census without the question would remove the base those estimates are built on.

4. Money follows the data

A bureau working paper found that at least 353 federal assistance programs used census data, in whole or in part, to distribute more than $2.8 trillion in fiscal year 2021, for health care, nutrition, highways, housing, school lunches, and child care, among other things. Not all of that depends on race. But every program that targets help by group, or checks whether its money reaches the people it is meant for, needs the group data to do it.

5. Other countries have tried it

Canada offers the clearest warning about what happens when a government weakens its census. In 2011 it replaced its mandatory long-form census with a voluntary survey. The national response rate fell to 68.6 percent, and almost 12 percent of communities fell below the 50 percent response level considered reliable, CBC News reported in 2012. Canada restored the mandatory long form for 2016, and early counts put that year's response rate at 98 percent, The Globe and Mail reported. The 2011 gap stayed in the record.

6. It could depress response overall

Census advocates have warned that the proposal itself, whether or not it takes effect, could lower 2030 participation among Latino residents, a group the bureau has historically undercounted, as covered in our report on the 2030 proposal. An undercount does damage beyond any single question, since it feeds into House seats and federal funding formulas.

7. It stacks on top of other data losses

The race question is not the only loss on the table. A Commerce Department order has already banned the privacy method the bureau used to publish detailed tables, and bureau officials have said entire datasets will disappear and that small places and small population groups will be hit hardest, as covered in census officials confirming the data loss.

A lot of what we publish rests on these questions. The diversity rankings, the segregation scores, and the profiles of more than 100 groups on the origins and ancestry hub are all built from the race, Hispanic origin, and ancestry answers people give the bureau.

Congress, not the president, has final say over the census. The administration must send Congress its proposed 2030 question topics by April 1.

Sources

Federal uses of race data are from the U.S. Census Bureau's Why We Ask: Race page. Federal funding figures are from the bureau's working paper Uses of Decennial Census Programs Data in Federal Funds Distribution: Fiscal Year 2021 (June 2023). Redistricting file contents are from the bureau's P.L. 94-171 summary file page. Canada's 2011 response figures are from CBC News and The Canadian Press, "Lower response rates threaten census data in some places" (September 25, 2012), and the 2016 figure from Tavia Grant, "Census response rate is 98 per cent, early calculations show," The Globe and Mail (July 21, 2016). Details of the 2030 proposal and the Project 2025 quotation are from NPR's reporting and the Heritage Foundation's Project 2025 agenda, as covered in our earlier posts.