Brian Snyder / Reuters

The outcome in Department of Commerce v. New York, which the Supreme Court will hear on April 23, will affect virtually every aspect of our national life, from the right to vote to the balance of power in Congress and the Electoral College to the scope of federal educational, health, and welfare programs. At issue is whether next year’s federal census form can include a question asking whether each member of a household is a U.S. citizen. Although the case has constitutional dimensions—the census itself is created by the Constitution— the result is likely to turn on whether the Department of Commerce, which administers the census, properly read Title 13 of the United States Code.

Every American knows that we have “separation of powers” in our system. But what does that mean? In essence, it means that Congress can create, change, or repeal statutes but can’t enforce them, and the executive branch can enforce statutes but can’t create, change, or repeal them.

Courts are called in as referees when someone alleges that an agency’s enforcement actions don’t match the authority Congress has given it—that, in essence, the president or those who work for him or her have changed federal statutes rather than enforced them.

If that question sounds simple, the answer is, “Have you ever met a statute?” Written by multiple committees, often in haste, important statutes can be impenetrable, tail-swallowing word salads that daunt even judges. “[W]hat happened to the Eighth Amendment?” the late Justice Antonin Scalia once snapped at a lawyer who suggested the Court read the entire Affordable Care Act. “You really want us to go through these 2,700 pages?”

But lawyers and courts must read entire statutes.  That’s because agencies are “creatures of statute”; no agency has any power not specifically given to it by statute. Administrators can’t just wake up one morning and say, “Hey, let’s do this!” They must follow set procedures when they make their own internal rules, when they make regulations that govern the public, and when they penalize people or companies for violations of federal statutes and rules.

Back to Department of Commerce v. New York: The census form hasn’t asked whether each person listed is a U.S. citizen since 1950.

Whatever you think of the idea as a general matter, it presents complex issues in the 2020 census context. The first is whether the department followed proper procedures in changing the question; the second is whether a citizenship question will make the 2020 census count more accurate and useful.

To understand the stakes, recall that the U.S. census is one of the few in the world that is required by its country’s constitution. The Framers in 1787 were well aware that in England, seats in the House of Commons were handed out in arbitrary fashion: Some huge towns had only one member—or were lumped in with rural areas—while other “rotten boroughs” had a full seat despite having as few as half a dozen voters. (Old Sarum, Wiltshire, for example, comprised three houses and seven voters, but had its own member of Parliament.)

The framers wanted the U.S. House of Representatives to be the “people’s house,” and to do that, distribution of seats had to be fair. For this reason, they provided in Article I § 2 cl. 3 that representatives would be awarded to states “according to their respective numbers,” and that those numbers would be determined by a new count (an “actual enumeration”) every 10 years. Conducting the census is not simply a power granted to Congress; it is a positive duty as well.  And since the demise of the infamous three-fifths clause (which counted each slave as only 60 percent of a “free person”), courts have read “respective numbers” to mean everybody, not just citizens or voters.

Congress delegated that job to the Census Bureau in the Census Act. As amended over the years, it specifies both the aims of the census and the procedures to be used in creating and compiling it. The act in particular restricts what information can be gathered, how it can be gathered, and how it can be used.

The census questionnaire is just one of the ways the bureau gathers information; the others involve consulting records of other state and federal agencies and summarizing or extrapolating figures from them. Because the questionnaire is so intrusive, the act provides that the bureau should prefer gathering data from these other sources, and should gather only “necessary” information from the questionnaire. The historian Margo J. Anderson, in The American Census: A Social History, noted that leaders from George Washington on have agreed that “any changes in census questions or calls for additional data would be evaluated in terms of the potential loss in accuracy and timeliness of the basic apportionment data.”

This takes us to March 2018, when Secretary of Commerce Wilbur Ross directed the bureau to add a citizenship question to the census questionnaire.

According to a sweeping 277-page opinion by U.S. District Judge Jesse Furman of the Southern District of New York, that wasn’t legal. After a three-week-long trial, Furman in January concluded, first, that the question was rushed onto the questionnaire in a hasty power play by Trump appointees, who did not properly notify Congress of their intention and then provided the court with a misleading explanation of the procedures used to develop the question; and, second, that statistical authorities, most particularly the Census Bureau’s experts, agree that the question will make the census less reliable. That’s because noncitizens filling out the form might become worried that the government will retain the information and use it for immigration enforcement, and thus will throw away the form rather than returning it. Both of these defects violate the Census Act’s instructions, Furman said; he enjoined the Commerce Department from adding the question to next year’s questionnaire.

Ross had apparently decided immediately upon taking office in 2017 that a citizenship question should be added. But for months afterward, he and his staff shopped around the federal bureaucracy trying to convince some other federal agency to ask for the question. Homeland Security rebuffed them; so, at first, did the Department of Justice—until Ross had a series of meetings with top officials, including then–Attorney General Jeff Sessions. At that point, DOJ submitted a letter stating that the question would help it enforce the Voting Rights Act. In the half century since the act became law, DOJ had always relied on other survey research for citizenship information, and had never asked for a change in the questionnaire before.  

The initial record filed with the court misstated Ross’s role in creating the question. And it did not disclose that he had discussed the issue with Steve Bannon, the notoriously anti-immigrant Trump strategist then in the White House, and then–Kansas Secretary of State Kris Kobach, known for his repeated false claims that noncitizens are voting in large numbers.

The record did disclose that the bureau’s professional statisticians and career census officials were unanimous: Adding a citizenship question would make the census less accurate. When the bureau asked for public comment on the new question, moreover, almost no one came forward to support it.

Thus, Furman concluded, the department used improper procedures to reach the decision—and the decision itself, far from being “necessary,” will make the census less valuable and accurate.

Furman has been careful not to base the opinion on the political aspects of Ross’s conduct. There is an obvious reason why President Donald Trump and his appointees might want the census to be less accurate: Undercounting immigrants—who are concentrated in urban areas that trend Democratic—will mean increased representation for areas with fewer immigrants, which are predominantly Republican. In a nation so closely divided politically, the divergence could make the difference between Republican and Democratic control of both the presidency—due to the Electoral College—and the House.

Are there good reasons to include a citizenship question on the questionnaire? Maybe, but the administration hasn’t made a serious effort to cite them. Instead, it has, from the beginning, simply lied. As we have seen, the claim that the Justice Department wanted the question was a lie. Beyond that, when the Commerce Department sent its draft questionnaire to Congress, it stated that the citizenship question had been on the questionnaire from “1890 to present”—not only a lie but a clumsy one.

The administration has reacted to this challenge with barely concealed contempt. It tried to shut down the trial by demanding the Supreme Court block the whole thing, then asked the Court to prevent the plaintiffs from questioning Ross during discovery. The Court expressed some concern about the district court’s discovery orders, but permitted the trial to go forward. In his opinion, Furman points out that his conclusions about the bureau’s misconduct are supported by discovery evidence that the government did not attempt to bar.

In its appeal briefs, the government now argues that the state of New York and other challengers lack standing, because they can’t prove that noncitizens won’t answer the questionnaire. It’s true there is no direct proof. That’s because the citizenship questionnaire was never tested to determine its impact, as required by bureau procedures. Every statistical authority in the bureau, and virtually all of those outside it, agree that the question will result in fewer responses and less reliable data. The government responds by changing the subject from statistical evidence—which indicates a virtual certainty of fewer responses—to talk-radio moralizing: There wouldn’t be any falloff in accuracy, it argues, if the wretched refuse would just “completely and truthfully answer the census questionnaire, as required by federal law.” It’s not the government’s fault if immigrants break laws, is it?

Beyond that, the government argues that what Ross said about his motives and procedures in his public statements, even though contradicted by the testimony developed later, should still settle the issue. This was a winning argument in the “travel ban” case: Since Trump officially gave a “neutral” explanation for barring entry from certain countries, the Supreme Court said in Hawaii v. Trump, it should ignore his other statements explaining that the real reason was religious bigotry. Similarly, Ross provided a “neutral” explanation; it was a lie, but so what?

The New York case is not the only challenge to the citizenship question. By taking steps that will clearly reduce the accuracy of the count, a federal judge in California reasoned, Ross and the Census Bureau violated Article I of the Constitution. After granting review in the New York case, the Supreme Court later expanded the questions presented to include the constitutional issue as well; it has also provided for expanded oral argument, agreeing to hear from the government, the state of New York, an immigrants’ rights group, and a representative of the Democratically controlled House of Representatives.

No one misunderstands the scope of the case. American politics in 2019 is a death duel between a side that, buoyed by popular support and demographic trends, favors broad voting rights, nonpartisan apportionment, and an accurate population count; and an opposing side that, facing a demographically diminishing political base, is moving to restrict the franchise and skew political power toward its strongholds.

That restrictionist side wants a census that will strip power from cities and states with racially diverse and immigrant-heavy populations. That side has also moved aggressively to exert control over the federal bench in general and the Supreme Court in particular, and now oozes confidence that its handpicked justices will dance to Wilbur Ross’s tune.

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.