Chief Justice John Roberts is a complex and subtly innovative jurist. The Supreme Court’s census decision revealed his skills, particularly seen against the backdrop of the gerrymandering case decided the same day. His ruling against the executive may actually strengthen deference to executive discretion when wielded by a future normal presidency, but in his decisive opinion (joined in full by no other justice), Roberts also refined a neglected counterdeference tool—pretext analysis. He did so in order to defeat one exceptionally consequential maneuver by this recurrently dishonest administration.
Some critics have scoffed at Roberts’s census ruling, hoping along with Justice Clarence Thomas in his dissent that it is “a ticket good for this day and this train only.” They are likely to be disappointed. Roberts will insist on a solid basis before deploying pretext analysis. But his opinion, by gingerly expanding the scope for skepticism, approves elevated judicial concern about governmental honesty. In performing this vital service to our polity during the Trump years, he has probably assigned himself a tightrope walker’s role in administrative law for the foreseeable future.
In this essay, I unpack the factors that contributed to Roberts’s unexpected result in the census ruling, and I venture some thoughts about how that outcome became a needed counterweight to the gerrymandering decision. I analyze what litigation approaches may succeed in drawing the closer scrutiny Roberts would support. Roberts does not suggest that his analysis applies only to domestic matters, though full pretext scrutiny in national security cases may be a tougher sell. Nonetheless, a lingering national security controversy—over President Trump’s travel ban—may provide an early occasion to test whether my analysis is on course.
These paired decisions on the final day of the term embody important characteristics of Roberts’s approach: traits that coexist in tension. He favors robust scope for executive authority. He also strongly wants to minimize the risk of boundless litigation. But he is genuinely concerned about official integrity, and he knows that this president violates without compunction the usual norms, customs and habits (such as truthfulness) that ordinarily keep broad executive authority within tolerable limits. Further, he rightly sees risks to legitimacy if judicial decisions skew strongly and persistently toward one side. This legitimacy concern probably accounts for his Obamacare decision in 2012, which gave conservatives a big doctrinal win on narrowing the reach of Congress’s commerce power but then rescued the individual insurance mandate based on the tax power. The census and gerrymandering cases are the jurisprudential heirs to that technique.
The chief justice doesn’t trust lower court judges to handle challenges to partisan redistricting, as he showed in his gerrymandering opinion, Rucho v. Common Cause. For him, that’s too close to judges picking winners in electoral contests, or to mandating a system of proportional representation—all without workable standards to guide decisions. At the same time, one can perhaps detect in the census decision’s bottom line a twinge of guilt about those Supreme Court decisions of the past decade, capped off by Rucho, that have helped move American politics away from contests over ideas. Instead those cases have empowered temporary majorities that seek to entrench their wins through brute manipulation of voting rules and maps.
Because Republicans had a banner year at the state level in the 2010 elections, the Republican Party has benefited the most from this reality. Take a look at the effects of its sophisticated redistricting initiatives, sometimes known as Redmap, which held Republican majorities in several legislatures and congressional delegations in 2018—even though Democrats won decidedly more votes in those states in that swing election. Or consider the speedy adoption (once the Supreme Court disabled the Voting Rights Act’s requirement for Justice Department preclearance) of cumbersome voter ID laws, strict rules for purging inactive voter lists, or the reduction of voting hours and sites.
This mix of conflicting tendencies accounts for Roberts’s unique resolution of the census case, which he dealt with in ways that have been misunderstood both by the White House and by critics of the decision.
First, the opinion clearly signaled that the chief justice wouldn’t go along with a newly minted rationale for including a question about citizenship in the 2020 census. Under Roberts’s logic, any new Commerce Department explanation hammered together in July 2019 would inevitably appear pretextual when offered to explain a March 2018 census design decision. After a few days of studying the Supreme Court ruling, Justice Department attorneys understood; they told the district court that the department was dropping further proceedings. One day later, President Trump belligerently reversed course in a tweet. Apparently transfixed by the fact that the court’s final order was technically a remand to the agency, he assumed the justices had offered him a do-over. It took eight more days for his staff to coax another U-turn. Attorney General William Barr explained the president’s final back-down as forced by timing constraints. But the savvy attorney general must have been aware of an immediate tactical advantage as well. Continuing the fight would immediately have subjected Commerce Secretary Wilbur Ross, an inept interviewee, to damaging discovery by the plaintiffs.
Prominent critics also misunderstood Roberts’s ruling. In a Washington Post column, for example, Hugh Hewitt wrote that “the census belonged to the president” until Roberts’s decision—and Trump needed to get it back. Otherwise the separation of powers would be gravely endangered, he argued, and “the checks don’t balance.”
On its face, this argument is nonsense. The Constitution mandates a census, an “actual enumeration,” every 10 years, “in such Manner as [Congress] shall by law direct.” It is true that Congress has delegated substantial discretion to the executive branch to shape the census, and Hewitt wants to preserve that discretion. But plainly the Constitution does not give the census to the president. And the exclusion of a single question for a single cycle hardly negates the separation of powers.
More importantly, Hewitt completely missed the careful game plan Roberts used to reach his result—which is generally favorable to Hewitt’s position over the long run. The central section of the census opinion, Part IV, is a virtual paean to executive branch discretion in implementing broad delegations from Congress. The lower court judges who found the census addition by Secretary Ross to be arbitrary under normal administrative review principles drew a firm scolding from Roberts: “[T]he choice between reasonable policy alternatives in the face of uncertainty was the Secretary’s to make. He considered the relevant factors, weighed risks and benefits, and articulated a satisfactory explanation for his decision. In overriding that reasonable exercise of discretion, the court improperly substituted its judgment for that of the agency.”
That part of the decision is what will be cited by the Justice Department in defending future exercises of executive discretion—and it should be. What Justice Robert Jackson aptly called “the actual art of governing under our Constitution” generally benefits from broad judicial deference to executive agencies as they labor to turn Congress’s enactments into operational reality.
These are hard words for me to write during the Trump presidency. But someday we will return to office a president who doesn’t act on daily whim, who respects the rule of law, reads briefing papers, values consistency and holds to the truth. Those who have served in the executive branch under serious presidents recognize the value of reasonable deference as officials work to serve their agency’s mission, confronting complexities that the judicial branch often undervalues.
But deference is not carte blanche. The courts face a perpetual dilemma of how to provide a judicial check against truly improper action without hamstringing or unduly delaying workable government. That is a tension often on Roberts’s mind. In Part V of the census opinion, the chief justice found a way to provide a unique check here. That section addresses the more tightly focused claim that Ross’s stated rationale was a pretext.
Ordinarily, Roberts emphasizes, judges should not inquire into the mental processes of administrators. But there is a narrow exception when the plaintiffs have made “a strong showing of bad faith or improper behavior.” The chief justice finds from the record of the case, taken as a whole, such a “disconnect between the decision made and the explanation given” that the lower court’s order had to be affirmed, thus leaving in place the injunction against the citizenship question. Deference, he says, channeling an insight by the respected jurist Henry Friendly, does not require judicial naiveté. “[I]f judicial review is to be more than an empty ritual,” Roberts summarizes, “it must demand something better than the explanation offered for the action taken in this case.”
Roberts does not commit to paper any hint about his view of the real explanation. But a chief justice who abjures naiveté has to be recognizing the obvious goal of Ross’s initiative: to skew electoral redistricting in favor of the incumbent party. And sure enough, eight days after the decision, when he temporarily ordered his lawyers to fight on, Trump himself gifted evidence to the plaintiffs about this very objective. When asked why he was changing course on the citizenship question, he said, “Number one, you need it for Congress—you need it for Congress for districting.” Trump was clearly expecting census undercounts in blue districts with a high number of noncitizen residents. And he may be thinking about someday pushing another dubious idea on the right-wing wish list—allocating seats in legislatures based on the citizen voting-age population, not total population numbers.
But why would the chief justice consider such an objective troublesome in the census case, when his Rucho decision upheld blatant partisan gerrymandering, done by drafters who boasted on the record about entrenching the incumbent party?
The two decisions are not really inconsistent. Indeed, the Court’s hands-off directive to district judges in Rucho may have helped fuel Roberts’s quest for a counterbalance. The Rucho decision, in Roberts’s own words, “does not condone excessive partisan gerrymandering.” It simply sees any judicial cure as unmanageable, because, in Roberts’s view, no workable standards exist to help courts distinguish appropriate partisan awareness from excessive partisan abuses and then to oversee the drafting of a replacement map. Green-lighting a judicial remedy also holds the prospect of subjecting federal, state and local redistricting officials—and also federal district judges—to hundreds of new lawsuits every reapportionment cycle. And, in the Rucho majority’s view, other remedies through Congress, state courts or legislatures, or citizen referenda, are possible.
The census case stands in sharp contrast. The decisional question is distinct, with a yes or no outcome, rather than a cascade of cartographic choices: Simply add the citizenship question to the census or not. No flood of nationwide lawsuits threatens. Most importantly, Ross’s citizenship question was likely to skew the redistricting process systematically—not just for one election or for one state, but for the whole nation, lasting a decade—by distorting the very foundational database from which all U.S. redistricting proceeds.
Roberts evidently came to see the census question as a far deeper, higher-stakes and more pervasive distortion than gerrymandering. Ross’s action was also a manipulation that lent itself to judicial remedy and could be definitively dispatched with one decision. At the same time, Roberts plainly left room for a full political debate over the following decade, involving both Congress and the executive branch, to decide whether a citizenship question should appear in the 2030 census.
What does Roberts’s ruling in the census case mean for future challenges to executive discretion, including in the national security realm? The dissenters predict that the chief will bring about a vast cascade of new administrative law cases alleging executive pretext and winning the right to burdensome discovery. Roberts is not likely to let this develop, however. Part IV of his census opinion has pride of place; it is so richly supportive of discretion that he will surely keep the pretext discovery threshold quite high.
But how high, and might it vary depending on the character of the incumbent administration? The formal standard for authorizing a serious pretext inquiry comes from the landmark Overton Park decision. Plaintiffs must make a “strong showing of bad faith or improper behavior,” and must do so before being granted discovery directed at the mental processes of decision-makers. Roberts expressly finds this threshold met in the census case. But his account of the evidence that led to that conclusion is so Delphic as to afford little future guidance. He provides colorful descriptors of what to look for—“disconnect” between decision and explanation, “significant mismatch,” “contrived reason,” explanation “incongruent” with the record—but not much more.
Future plaintiffs will have to do their best to satisfy the formal requirements despite this thin guidance. It may be more productive in actual practice to show enough departures by the administrator from normal procedures or normal substantive criteria that a judge will feel exposed as a judicial naïf if he or she fails to permit at least initial discovery. (Departures of this type are listed as key evidentiary factors in the leading legislative motive case, Village of Arlington Heights.) Reserving this kind of pretext argument for large-scale, widely applicable administrative decisions might also be prudent for plaintiffs.
If the Supreme Court has a new openness to pretext arguments, even in the national security arena—or at least a potential new framework for considering them—that drama may well play out in continuing litigation over the Trump administration’s travel ban. The central dispute there was whether the government’s asserted national security reasons for banning travel from seven specified countries (with various exceptions and qualifications) were genuine or were mere window-dressing for decisions based on animus against Muslims, in violation of equal protection or of the First Amendment’s religious establishment clause.
Of course, the Supreme Court already upheld, a year ago, the third iteration of the travel ban. Roberts’s majority opinion in Trump v. Hawaii applied deferential rational-basis review in expressly rejecting a pretext argument based on the record as it then stood. The court vacated preliminary injunctions and remanded. The plaintiffs did not pack up their tents, however. Instead they have sought discovery, and their cases have survived government motions to dismiss. They could be heading back to the Supreme Court, but only after months of discovery and perhaps a trial.
One element of Roberts’s reasoning on the pretext claim in Trump v. Hawaii provides a foothold for plaintiffs to achieve a different result after creating a more complete record (as Justice Stephen Breyer’s Hawaii dissent telegraphed). Roberts found that three “additional features” of the travel ban policy supported the finding that it was based on “a legitimate national security interest,” among them the existence of “a waiver program open to all covered foreign nationals seeking entry as immigrants or nonimmigrants.” The opinion goes on to quote government guidance documents giving examples of those who could qualify: a person who “seeks to reside with a close family member, obtain urgent medical care, or pursue significant business obligations.” The plaintiffs (whose ranks include people matching those examples) now have statistics and evidence on the actual implementation of the waiver program. Their preliminary evidence shows high rates of rejection in waiver cases (98 percent in the first year, a cumulative 94 percent after two years). The evidence also includes a declaration from a former consular officer indicating that officers had authority to deny waivers but could not approve them, instead having to send any prospective grant to headquarters.
Perhaps the plaintiffs can also find further support in the president’s continuing public statements and tweets showing animus toward Muslims, including four-Pinocchio falsehoods in recent weeks regarding Rep. Ilhan Omar. At some point, one would think, this president’s relentless dishonesty may ease the path toward finding “disconnects” or explanations “incongruent” with the record. But to win Justice Roberts over to such a result probably also requires a disciplined effort to sketch manageable standards for a workable injunction. The plaintiff’s odds remain steep, but they look better now than before the census ruling.
Chief Justice Roberts has added to the judicial toolbox for dealing with a wayward executive branch. He will make ample room for executive branch discretion, but he may help the polity survive the Trump era by finding occasions to patrol the guardrails that protect government integrity and legitimacy.