Who Counts in America?

The census’ new citizenship question could entrench Republicans in power for a decade.

An attendee holds her flag and her naturalization papers as she is sworn in during a U.S. citizenship ceremony in Los Angeles on July 18.
An attendee holds her flag and her naturalization papers as she is sworn in during a U.S. citizenship ceremony in Los Angeles on July 18. Mike Blake/Reuters

The Trump administration’s decision to add a citizenship question to the 2020 census could mark the beginning of a radical shift in American democracy. If the question survives a court challenge, it will fundamentally alter the composition of government, exacerbating overrepresentation in predominantly white regions while diminishing representation of immigrants and minorities. The question will tilt political power to the right in both Congress and state legislatures while depriving cities of vital funds for public health, schooling, and other social services. And it opens the door to a new kind of gerrymandering that will further suppress minority votes. The addition of this question may be Donald Trump’s most lasting move as president, one with vast ramifications that will harm communities disfavored by the Republican Party through at least 2030.

How did we get here? The census is meant to be a politically neutral tool designed to ensure that Congress is truly representative of the country it serves. Every 10 years, the federal government must conduct a census to determine the “actual enumeration of the people,” and then mandates congressional redistricting based on its findings. From the start, the Constitution required the counting of people rather than just citizens—though it notoriously counted individual slaves as three-fifths of a person. The 14th Amendment remedied this shameful exception by apportioning representatives “according to their respective numbers, counting the whole number of persons in each state.”

The language of the Constitution couldn’t be clearer: Congress represents people, not citizens or voters or legal residents. Indeed, the framers of the 14th Amendment considered and rejected the notion of counting citizens alone. Ohio Rep. John Bingham, the principal author of the amendment, explained that it would be foolish to “strike from the basis of representation the entire immigrant population not naturalized.” Why? Because “under the Constitution as it now is and as it always has been, the entire immigrant population of this country is included in the basis of representation.” The census counts all people, immigrants included, and all people receive representation in Congress.

In 1950, the federal government stopped asking about citizenship on the decennial census. But in recent years, Republicans have crafted various theories designed to justify restoring the citizenship question. Some have argued that including undocumented immigrants in the census is illegal, an absurd claim that contravenes the original understanding of the 14th Amendment. Others assert that Congress must represent voters, not people—a theory the Supreme Court rejected in 2016’s Evenwel v. Abbott, holding that states may count the overall population to satisfy the Equal Protection Clause’s “one person, one vote” principle.

Secretary of Commerce Wilbur Ross added the citizenship question to the census at the behest of the Department of Justice. In December, the DOJ told the Census Bureau, which is part of the Department of Commerce, that it needs a “reliable calculation of citizen voting-age population” to enforce the Voting Rights Act, which bars the dilution of minority votes.

The DOJ is full of it. There is zero chance that Attorney General Jeff Sessions legitimately cares about VRA compliance given that he has abandoned the law in court. And the DOJ’s theory doesn’t even make sense. As the attorneys general of 19 states and the District of Columbia pointed out in a letter, the Supreme Court has never measured vote dilution by simply asking whether a majority of citizens in a district are minorities. To be clear, plenty of conservative lawmakers want the court to do so. Texas, for instance, created multiple districts that contain a majority of Hispanic citizens but relatively few reliable Hispanic voters. In 2011, the DOJ took the position that this method did not satisfy the VRA.

Sasha Samberg-Champion, a former attorney with the DOJ’s Civil Rights Division who now practices at the firm Relman, Dane & Colfax, told me the DOJ’s justification for a citizenship question may indicate the agency will soon flip positions and side with Texas. The Justice Department looks poised to embrace the theory that the state can satisfy the VRA by packing a majority of Hispanic citizens into a district even if they do not consistently vote. If courts accepted this position, states like Texas could effectively sidestep the VRA’s goal of empowering minority voters.

There are two equally ominous reasons why the Trump administration would ask for citizenship data. First, in Evenwel, the Supreme Court did not rule out the possibility of drawing districts based on the population of citizens or voters rather than overall population. Citizen-based apportionment would deprive immigrants of representation, shifting power toward white rural voters at the expense of minorities and city-dwellers.

A Brennan Center analysis found that counting citizens instead of people would disproportionately affect “booming metro areas” like Dallas, Houston, and Los Angeles given their large populations of immigrants. It would also dilute the votes of Latinos and make it more difficult to draw VRA-compliant districts. Harvard political scientist Carl E. Klarner found that apportionment by citizen population would substantially reduce Latino representation in state legislatures and the House of Representatives. An analysis by Andrew A. Beveridge, a professor of sociology at Queens College, reached the same conclusion.

In Evenwel, the Supreme Court didn’t address the legality of such a scheme because it was, at the time, theoretical. By soliciting data on the nation’s immigrants, the Trump administration seems to be laying the groundwork for Texas and other states to use citizen population in the next cycle of redistricting. The upshot would be fewer Democratic representatives in Texas; Beveridge estimated that Republicans would gain at least seven seats in the state House (which has 150 members) and one in the state Senate (which has 31). He also found that Republicans would win one to two more congressional districts in the state.

Second, adding a citizenship question to the census will almost certainly suppress responses from both legal and undocumented immigrants. These individuals may fear that the government will use their information to target them for deportation or investigate their family’s citizenship status.

How large might this effect be? Unfortunately, the Census Bureau did not model the impact of a citizenship question. But after the 2010 census, the bureau noted that Hispanic communities were already difficult to count, in part because immigrants are understandably wary of government officials who knock on their doors asking questions. In September 2017, the bureau’s Center for Survey Measurement found that Hispanics and immigrants were mistrustful of census-takers. It found that fears of the census, “particularly among immigrant respondents, have increased markedly this year,” with immigrants expressing concerns about the “Muslim ban” and the dissolution of DACA. Some immigrants were uncomfortable “registering” members of their household; others were told by “community leaders” not to open the door without a warrant signed by a judge.

That was before the census featured a citizenship question. Civil rights advocates believe that the addition of such a question will compound distrust in the census among immigrants, lead undocumented individuals to refuse to participate altogether, and push those with legal status to lie out of fear that they will inadvertently report undocumented family members. During the Evenwel litigation, several former directors of the census told the Supreme Court that a citizenship question would “invariably lead to a lower response rate,” particularly among noncitizens.

A significant majority of immigrants live in blue states; if they don’t respond to the census, these states’ populations will be undercounted. As a result, the states could lose seats in the House of Representatives as well as funding for Medicaid, food stamps, transportation, and education. The problem would stretch well beyond the Trump era; the numbers collected in 2020 will be in use until the 2030 census.

California Attorney General Xavier Becerra appears to recognize this threat; on Monday, he filed a lawsuit challenging the legality of the citizenship question. Becerra argues that the addition violates the Constitution’s Census Clause itself: By “diminish[ing] the response rates of non-citizens and their citizen relatives,” the question will interfere with the “actual enumeration of the people” required by the Constitution. The strength of this claim is difficult to gauge given the novelty of the question. It seems possible that the courts will find that a citizenship question does not infringe on “actual enumeration” because immigrants can still respond to the census if they want to, even if the question has a disparate impact on their response rate.

Becerra also argues that the new question violates the Administrative Procedure Act, a federal statute that bars any “arbitrary or capricious” agency action. This claim is quite strong. The citizenship question was added at a very late stage over the objection of career attorneys and with no field testing to estimate its effects. That marks a dramatic departure from the bureau’s standard procedures, implying a political motivation that may be unlawfully “arbitrary” under the APA.

There is thus a decent chance that the courts could block the question. But there is also a good possibility that it will remain in place, redrawing the country’s political map for at least 10 years, creating a self-perpetuating cycle of disenfranchisement. Trump will eventually leave office. But the dire consequences of this move could stay with us for a long time to come.