Race, Race, Go Away

The Trump administration is trying to delete race from national statistics, and it may very well succeed. The courts may not stand in the way, and Democrats have not even raised the issue this election year, probably because they want to avoid getting sucked into battles over race and identity politics.

On September 9, the administration announced a proposal to remove race and ethnicity questions from the decennial census. That announcement followed a vote last July by the Equal Employment Opportunity Commission to rescind requirements for employers to report the racial and gender makeup of their workforce. Just as important as these actions are the administration’s justifications for them, which suggest that the government should never collect or publish data based on race.

When government stops publicly reporting data for which no private substitute exists, the result is a “statistical blackout.” The decennial census is the only enumeration of the entire U.S. population; the EEOC data covers about 50 million employees. No other demographic surveys come close. If the Trump administration prevails on its announced proposals and pursues the same logic in other areas, the result would be a far-reaching blackout about racial change in America, including how well or poorly different groups are doing.

Yet while the Trump administration seeks to stop collecting data by race, it is demanding extensive racial data from colleges and universities. These contradictory actions are driven by one aim: preventing minorities from obtaining what the administration sees as unfair racial preferences. But in pursuit of that aim, it has not observed even a minimal level of intellectual and legal consistency.

In its official statements about the proposed census and EEOC changes, the Trump administration lays out a sweeping case against racial data. It points to “substantial doubts” in both “national discourse” and law about “the meaning, significance, and impact” of “attempting to categorize” people by race. It says that racial categories are “arbitrary” and “promote stereotyping.”

The notices justifying the census and EEOC changes invoke the Supreme Court’s decision to overturn Harvard’s race-conscious admissions policies. The administration cites Justice Clarence Thomas’s opinion that race is based on “ephemeral, socially constructed categories” and Justice Neil Gorsuch’s opinion that the categories were established by politically motivated bureaucrats and “have become only more incoherent with time.”

In the EEOC statement, the administration argues that the agency’s employer survey creates an “impermissible focus on ‘minorities’ and women.” Merely collecting the data invites discrimination because it may lead employers to take “sex and race-based actions to correct statistical imbalances.” The administration even floats the argument that federal requests for racial data in surveys should be deemed presumptively unconstitutional on the grounds that they violate the Equal Protection Clause of the 14th Amendment.

But the administration doesn’t take its own arguments seriously. At the president’s express direction, the Department of Education is requiring colleges to report race and sex data, including test scores and grades, for their applicant pool, admitted students, and enrollees. Apparently, it is “impermissible” for the government to demand race and sex data from employers but necessary and proper to make that demand of colleges. But if racial categories are arbitrary and incoherent stereotypes, why is any institution required to use them?

The administration’s moves to stop collecting racial data have been overshadowed by its proposals to add a census question about citizenship and to exclude both illegal immigrants and legal residents without green cards from the official count for congressional reapportionment. Those proposals have set off an intense reaction because they would remove millions of people from the census, distort population counts, and shift political representation and the distribution of federal funds toward areas with fewer immigrants and more Republicans.

Opponents of the proposal to narrow the count for congressional apportionment have a good chance to stop that idea in the courts because it violates the clear language of the 14th Amendment, which requires apportionment to be based on “the whole number of persons.” The administration, however, is in a much stronger legal position on adding a census question about citizenship and removing the question about race.

The Census Bureau is part of the Department of Commerce, and the secretary of commerce has the authority to decide which questions the census asks. President Trump also tried during his first term to add a citizenship question to the census, but he did it hastily, violated procedural requirements, and lost in the Supreme Court. The administration is not repeating its earlier procedural errors. Last spring, the Supreme Court also cleared away what may have been the only legal obstacle to removing race from the decennial census, when in Louisiana v. Callais the Court banned the use of racial data in redistricting.

Liberal democracies vary in whether they count people by race—France does not. Liberals and progressives in this country should also at least have misgivings about the practice. Collecting official racial data risks perpetuating the race myth, the idea that human beings are divided into discrete racial groups.

But race is a myth of a special kind: It is a myth turned social reality after centuries of being the basis of rights, dignity, wealth, and power. If making racial data disappear could make racial inequalities disappear, we should all support a statistical blackout on race. The liberal bet is that we are more likely to make progress toward a society where the race myth has reduced consequences if we know the facts. Suppressing data about race will not suppress racism itself. Without the discipline of facts—without science, without professional journalism, and without data—there is nothing to hold back the forces of racial denial and deceit.

Liberals have reconciled themselves to the need for racial data in part for another reason. Instead of “attempting to categorize” individuals by race, the government asks us to identify ourselves. When the Census Bureau introduced mail-in questionnaires in 1960, it began relying on respondents’ own racial self-identification, instead of racial appraisals by census interviewers. A general shift toward self-identification in official surveys and records took the federal government largely out of the business of defining an individual’s race. In 1977, when the government first adopted consistent standards for reporting race and ethnicity, it declared that the categories “should not be interpreted as being scientific or anthropological in nature.” Race became an identity, an expression of how people see themselves and a satisfactory measure of how others see and treat them.

Self-identification is the basis on which liberals and conservatives have long been able to agree on the collection and publication of racial data. The 1977 standards included a list of “minimum designations” for reporting: American Indian or Alaskan Native, Asian or Pacific Islander, Black, and White (for race), and Hispanic Origin and Not of Hispanic Origin (for ethnicity). The standards were revised in 1997 to enable people to check off two or more categories on the race question; in 2024, the Biden administration added Middle Eastern and North African and combined race and ethnicity into one question.

Whether the 2024 standards will ever go into effect is uncertain. After the 2024 election, some conservatives urged Trump to rescind the standards and abolish the use of racial categories entirely as part of an agenda aimed at driving “a spike through the heart of woke.” Trump has pursued most other elements of that agenda. He has not yet tried to purge race entirely from federal data, though the census and EEOC proposals are major steps in that direction.

A blackout of racial data in just the decennial census would have wide ramifications. Most other governmental and private surveys rely on the decennial enumeration to make representative estimates. Take away the baseline provided by the only count of the U.S. population, and other surveys lose their north star.

We depend on the federal government for reliable indicators of the state of the economy and American life, including the unemployment rate, birth and death rates, and crime victimization rates. You can’t compute a rate without knowing the denominator, which is what the census provides. Changes in those rates sometimes show progress and sometimes show problems. How those rates vary across racial lines is information that as citizens and voters Americans ought to have.

The administration’s proposals to remove race from national data came during a midterm election campaign when Democratic candidates wanted to downplay race and gender issues and keep attention focused on the economy. So they didn’t raise concerns about a racial blackout, and hardly anyone is aware that it might happen. Next year, however, the new Democratic Congress could try to reverse all or most of Trump’s changes to the census and other national data by undoing them in legislation that the president would agree to sign because of other provisions that he wants.

Would this issue be important enough to get that kind of legislative priority? I’m not sure that it would. Some Democrats who want to move away from identity politics might see the removal of race from the census as a positive step. I understand that urge; it’s not a good thing when race dominates politics.

But it’s a mistake to equate the collection of racial data with an obsessive identity politics. You don’t have to be woke to be concerned about racial disparities in employment, housing, and health. The collection of data by race doesn’t require that public problems be understood primarily in racial terms.

In fact, the data sometimes shows that race isn’t the issue in the way people might assume. For example, the findings by the economists Anne Case and Angus Deaton about rising “deaths of despair”—deaths from drug overdoses, alcoholism, and suicide—have had a wide public impact in part because they showed rising death rates among whites, specifically whites without a college education. If Case and Deaton had been unable to identify death rates by race, many people would have assumed those findings applied primarily to minorities and misunderstood the scope of the problems. Case and Deaton’s work helped clarify the economic basis of declining health in America.

Keeping public controversy grounded in facts is an unending struggle. We need some agreed-upon facts even to be able to disagree productively. Conservatives ought to recognize that on race they too need the illumination of facts, not a statistical blackout.