Nonsensus: Pretext and the Decennial Enumeration

Justin Levitt Associate Dean for Research and Professor of Law, Loyola Law School, Los Angeles

It is impossible to overstate the importance of the Census. The obligation to conduct a decennial enumeration of the population appears in the sixth sentence of the Constitution, as the very first duty given to the new federal government: before the enumeration of legislative power, before the power to declare and wage war, before the resolution of federal judicial cases.[1] The Census enjoys this primacy because it is, logically, antecedent to the construction of a federal government that ostensibly obtains its power from the people. From the Founding, legislative representation in the federal government has always depended on knowing how many people live where.

Today, the Census continues to drive the allotment of congressional representation to various states, but it also does so very much more.[2] After the reapportionment revolution of the 1960s, our decennial population count determines the allocation of representation for federal, state, and local offices across the country.[3] The Constitution demands that districts foster equal representation. Whenever we take the Census to learn how many people are where, we redraw districts to recalibrate the representation we receive. Elections are the way Americans build the world we want to live in, together. The Census is the way we know who we are.

And the Census is also more than just the foundation of all government. It has also always been connected to funding: As early as 1798, it drove the apportionment of a new federal tax to backstop the federal government’s over-reliance on tariffs in the event of a coming war with France.[4] In the modern era, it is used more for giving than receiving: Hundreds of billions of federal dollars are distributed based on counts in the Census.[5] And it provides our national informational infrastructure: The reason we know that a survey or poll is representative, the way that local governments target policy interventions, one of the ways businesses decide to locate or relocate to a community with particular economic habits or characteristics—it all has the Census at its core.

I. The Controversy

In March of 2018, Secretary of Commerce Wilbur Ross determined that, for the first time in the country’s history,[6] the Census Bureau would ask every household about the citizenship of its residents.[7] To be sure, the Census Bureau has asked questions about citizenship before, of greater or lesser slices of the public. Right now, the public is asked about their citizenship on the American Community Survey (ACS), a survey designed to reach an average of 0.2% of households each month.[8] The question sits in a twenty-eight page instrument including seventy-one inquiries (some with subparts) and designed to take about forty minutes.[9]

Adding a question about citizenship to the ten basic questions asked of every household in the country dramatically elevates the question’s prominence. The last time we asked a significant portion of the American population about their citizenship in the basic enumeration conducted door-to-door was 1950.[10] At the time, about seventy-three percent of the public trusted the federal government to do what is right.[11] As of December 2017, that number was eighteen percent.[12]

Secretary Ross’s decision was immediately and enormously controversial. Career officials at the Census Bureau (and a succession of former Bureau directors) had long warned that asking questions about citizenship in the decennial enumeration itself would compromise the enumeration’s accuracy.[13] On the one hand, there were serious concerns that the prospect of a question on citizenship in a house-to-house enumeration would cause people to decline to respond to the Census at all, or to omit people in the household. On the other, there were concerns that, administered in this fashion, the question would cause noncitizens to falsely claim citizenship status, distorting the information collected.[14]

Those concerns were only magnified in recent years, given the increasingly toxic political climate around immigration and the consequent spread of fear in minority communities.[15] As early as 2015, Census officials noted decreased ACS response rates, disproportionately large in tracts with substantial concentrations of noncitizens, and an additional growing refusal to answer citizenship questions even when responding to the remainder of the survey.[16] An internal qualitative survey reported “[f]indings across languages [and] regions of the country, from both pretesting respondents and field staff[, that] point to an unprecedented ground swell in confidentiality and data sharing concerns, particularly among immigrants or those who live with immigrants.”[17] It also found that unprecedented groundswell represented the reaction to a citizenship question in a less prominent context than the door-to-door enumeration.

The precise magnitude of the new question’s impact was difficult to assess. Normally, one would expect studies by the Census Bureau itself to provide a sense of the consequences for a change of this nature: If this were a movie, the Census Bureau would be the team of lab-coated, white-bootied scientists titrating the antidote or prepping the space probe. They undertake even the most modest adjustments with exceptional care and ample testing and retesting. But the Census Bureau was offered no opportunity to evaluate the potential impact through quantitative tests of its own, in real-world context, before Secretary Ross made his decision.[18] Indeed, Secretary Ross cited the absence of “definitive, empirical” evidence of the potential impact as additional support for his decision.[19]

In the absence of meaningful testing in a real-world context, the Census Bureau estimated the extent of damage the question would cause as precisely as it could. It compared the drop-off in response rates from the (short) enumeration to the (long) survey in households without a noncitizen and in demographically similar households with a noncitizen; in 2010, that comparative incremental drop-off rate was identified as approximately 5.1%.[20] The chief scientist of the Census Bureau stressed that this was a “lower bound” and a “conservative estimate,” and that the real rate of nonresponse to the decennial enumeration “could be much greater” than 5.1% of households with noncitizens, itself a sizable number.[21] For that reason, among others, the career officials at the Census Bureau suggested collecting the information through administrative records, rather than through a question on the enumeration that could damage the enumeration itself.[22] Secretary Ross rejected the suggestion.[23]

Indeed, there is plentiful reason to believe that the real rate of damage would be much greater. The conservative estimate assumes that nonresponse rates for a citizenship question on the house-to-house enumeration would be no larger than nonresponse rates for the relatively below-the-radar survey, despite the increased prominence of the decennial enumeration; that nonresponse driven by the question should be expected only of households containing at least one noncitizen, despite fear within the broader community; and that nonresponse for a 2020 Census would be no larger than nonresponse rates in the 2010 Census, despite the substantial difference in overall political climate. There are substantial reasons to doubt each assumption.[24] A private study conducted after the decision was announced, and credited in litigation, predicted a reduced response rate—not just among households with noncitizens, but total—of 6.3% to 8.0% nationally and 10.5% to 14.1% in California.[25]

II. The Ostensible Rationale

At the time, Secretary Ross attempted to justify his decision to add a question on citizenship to the decennial enumeration by claiming a desire to assist the Department of Justice (DOJ) with its enforcement of the Voting Rights Act (VRA).[26] The DOJ had in late 2017 purported to need the extra information for extra enforcement: Data from the ACS, it claimed, were insufficient for the purpose.[27] Yet simply on its face, there was substantial reason to doubt the legitimacy of that need.

In theory, there are three primary instances in which better citizenship data might assist VRA enforcement, all in the redistricting arena. Among the threshold elements of one type of VRA redistricting case is the need to prove that a minority group (or groups) is sufficiently large enough to have its voting power diluted by the absence of responsive districts.[28] This element has been interpreted to require a showing that the minority community could comprise at least half of the electorate of a district-sized population.[29] Another threshold element requires a showing that the electorate is polarized based on race; these assessments hinge on nuanced evaluations of electoral patterns calculated from data including, inter alia, the demographic composition of voters within each precinct.[30] And when liability is found, remedial measures must allow equitable opportunities for minorities to elect candidates of their choice, which also requires an assessment of local electoral power, similarly drawn from calculations using data including (but not limited to) the same demographic information above.[31] Each of these elements relies on information about the electorate—or, as a proxy, the citizen voting-age population.[32]

To date, each and every VRA case using information about the electorate has drawn its data from the existing decennial enumeration or—when citizenship has been implicated—from surveys, like the data in the ACS. Since the VRA’s passage in 1965, no case had ever been brought using data on citizenship from the decennial enumeration, because in fifty years, the decennial enumeration had not asked for that data. To my knowledge, before December 2017, no DOJ official had requested citizenship data from the decennial enumeration to enforce the VRA.[33] To my knowledge, no proponent or opponent of asking the question has identified even one concrete circumstance in which a potential VRA case failed or was not brought for want of citizenship data from the decennial enumeration.[34]

Indeed, there is only one case that has emerged even potentially implicating a role for citizenship data on the decennial enumeration—and that case demonstrates why such data are unnecessary. In 2010, Latino plaintiffs claimed that the at-large election system in Farmers Branch, Texas, unlawfully diluted their right to vote.[35] They presented evidence, from the ACS, showing that Latino citizens could meet the VRA’s litigation thresholds.[36] But ACS data have several limitations in small jurisdictions with growing minority populations.[37] So plaintiffs also utilized data directly from the voter files, tallying voters with surnames highly likely to be Latino; this is a method not only used by the DOJ in its own litigation,[38] but also validated in the legislative history of the VRA itself.[39] The court found that the plaintiffs had met their burden under the VRA.[40] Without the need for citizenship data collected in the decennial enumeration.

It is true that better data might hypothetically yield better enforcement. Though no case has been identified thus far, circumstances can be imagined in which better data would make the difference. But the key relevant insight is that in this climate, collecting citizenship data on the decennial enumeration would not yield better data. It would yield data that is ostensibly more precise, but actually less accurate, in a direction disfavoring enforcement. More precise data could help only in marginal cases, where the population is actually sufficiently large or robust to create liability, but statistical uncertainty precludes the predicate proof. But collecting data on citizenship in a door-to-door enumeration will only depress participation by exactly the minority populations in need of protection, leaving an inevitable undercount.

In other words, better data would help populations just on the margin of enforcement. But door-to-door enumeration data would not be better: It would drive down response rates precisely in those populations, turning uncertainty about reaching litigation thresholds into ostensible certainty of falling short of litigation thresholds. And that ostensible certainty would be false.

Perhaps this is why the DOJ had never before requested citizenship data from the decennial enumeration. Perhaps this is why every single one of the nonprofit groups engaged in vigorous private enforcement of the VRA met the Census Bureau’s decision with condemnation and not applause.[41]

If the proffered justification of VRA enforcement appeared odd at the time, on the face of the request, the administrative record revealed in litigation firmly exposed it as a sham. It emerged that as early as February 2, 2017, officials in the Department of Commerce were “very interested” in the topics for the decennial Census,[42] and in May, months before any request from the DOJ, Secretary Ross became agitated about his “months old request that we include the citizenship question.”[43] Earl Comstock, his director of the Office of Policy and Strategic Planning at Commerce, responded that “we will get that in place. . . . We need to work with Justice to get them to request that citizenship be added back as a census question . . . . I will arrange a meeting with DOJ staff this week to discuss.”[44] Commerce shopped the desire for a question to DOJ, which declined any interest, then to DHS, which also declined interest, and then back to DOJ, before a discussion between Secretary Ross and Attorney General Sessions spurred DOJ to make its VRA request in December of 2017.[45] The process looked emphatically like a decision in search of a reason.

We may never know the real source of Secretary Ross’s strong interest in placing a citizenship question on the decennial enumeration as early as early 2017, well before any request from DOJ. It is possible that the likely statistical impact was its own reward: Depressed participation among noncitizens would (inaccurately) show fewer noncitizens in the country, which would support the administration’s appeal among both nativists and proponents of expansive immigration enforcement.[46] It is possible that the real motive lay in hoping that reduced participation would yield comparative geographic, racial, or partisan gains, when (inaccurate) Census data were used to distribute funding and political power in familiar ways.[47] It is possible that the real motive lay in creating the data environment for restructuring political power in unfamiliar ways, building a legally questionable redistricting base predicated on the exclusionary representation only of citizens (or voting-age citizens) rather than representation of the population as a whole.[48] It is possible there was another motive entirely.

III. The Litigation

Everybody sued. The State of California filed the first complaint on the day that Secretary Ross announced his decision,[49] and five other primary cases followed shortly thereafter, joined or consolidated in three venues: the Southern District of New York, the Northern District of California, and the District of Maryland.

One of the New York cases would become the lead case in the set, with two principal claims.[50] First, a set of state and local government plaintiffs alleged that the inclusion of the citizenship question violated the Constitution’s “Enumerations Clause”—the provision establishing the Census’s only constitutional duty, the decennial requirement to conduct an “actual Enumeration” of the “whole number of persons” in the country.[51] According to the plaintiffs, the presence of the citizenship question would cause an undercount, predictably impeding the enumeration itself.[52] Second, plaintiffs alleged violations of the Administrative Procedure Act (APA), including that the decision defied longstanding agency data-quality standards, that it was made without sufficiently complete or reasoned explanation, and that the rationale offered was pretextual.[53]

A second case, consolidated with the first, was brought on behalf of nonprofit groups. In addition to echoing the claims of the governmental plaintiffs above, these plaintiffs added equal protection claims based on intentional discrimination against immigrant communities of color and APA claims predicated on violations of specific additional statutory provisions governing Census Bureau procedures.[54]

Other sets of cases were lodged in California and Maryland, mostly with claims overlapping those brought in New York.[55] One, brought in California by the City of San Jose and the Black Alliance for Just Immigration, added a claim that the inevitable undercount would violate the Fourteenth Amendment’s clause requiring the apportionment of congressional seats according to the enumeration of population.[56] Another, brought in Maryland by a group of nonprofits, individuals, and caucuses of minority legislators, added a claim based on a conspiracy to interfere with civil rights—a claim originally codified in the Ku Klux Klan Act of 1871.[57]

In addition to the litigation above, two additional sets of plaintiffs lodged related suits. In Washington, D.C., the Electronic Privacy Information Center (EPIC) alleged that the Department of Commerce unlawfully failed to conduct full and complete Privacy Impact Assessments prior to announcing its decision to add the citizenship question to the enumeration.[58] And in New York, the NAACP Legal Defense and Educational Fund (LDF) initiated a Freedom of Information Act (FOIA) suit against the DOJ for records relating to the DOJ’s review of the citizenship question before issuing its December 2017 request.[59]

Litigation proceeded most quickly in New York, up against a June 30, 2019, deadline for determining the content of the Census forms to be printed.[60] In July 2018, the court dismissed claims brought under the Enumeration Clause, reasoning that the Clause permitted the government to ask demographic questions beyond a spare enumeration and that the Clause did not itself restrict the secretary’s discretion regarding which questions to ask.[61] But the court declined to dismiss the APA and equal protection claims.[62]

Over the government’s strenuous objection, the court granted discovery beyond the administrative record, including a September 2018 order permitting the deposition of Secretary Ross.[63] In part, the APA claims turned on whether Ross’s presented rationale fit the mustered evidence, was beyond the authority conveyed by Congress, or was pretextual. The equal protection claims turned, in part, on whether the rationale was discriminatory.[64] And though depositions of high-ranking government officials are permissible only in extraordinary circumstances, the court found that the unusual path to the decision and Secretary Ross’s increasingly questionable testimony, under oath to Congress, regarding the origin of the decision amounted to extraordinary circumstances warranting direct inquiry.[65] The Supreme Court first stayed the deposition of Secretary Ross and then agreed to take up the question of extra-record discovery more generally, though events would soon overtake this particular branch of the controversy.[66]

Meanwhile—despite eleven failed governmental applications for stays from various tribunals in the hierarchical chain[67]—a trial was held. On January 15, 2019, the court issued a 277-page opinion, vacating the decision to place the citizenship question on the enumeration.[68]

The court rejected the equal protection claim, finding insufficient evidence of intentional discrimination (even while acknowledging that the plaintiffs had not been able to depose the primary decisionmaker).[69] But it found violations of several technical statutory provisions governing the conduct of the Census.[70] And more sweepingly, the court found that Secretary Ross’s decision was unlawfully “arbitrary and capricious” under the APA—the “explanations for his decision were unsupported by, or even counter to, the evidence before the agency,”[71] the decision departed from established statistical quality standards without justification,[72] and “[f]inally, and perhaps most egregiously, the evidence is clear that Secretary Ross's rationale was pretextual—that is, that the real reason for his decision was something other than the sole reason he put forward in his Memorandum, namely enhancement of DOJ's VRA enforcement efforts.”[73]

Representing that the content of the Census was a matter of national importance, and that time was of the essence given the need to “finalize the decennial census questionnaire for printing by the end of June 2019,” the government sought Supreme Court review before Second Circuit proceedings.[74] The Court granted review and heard argument on the case in April.[75]

While Supreme Court review of the New York case was pending, litigation proceeded in California and Maryland. On March 6, 2019, the California federal court enjoined the secretary from adding a citizenship question to the Census; the court found statutory violations similar to those found in New York, but also granted relief on the Enumeration Clause claim, on the basis that including the question “will materially harm the accuracy of the census without advancing any legitimate governmental interest.”[76] On April 5, the Maryland court did likewise.[77] And between these two decisions, given the California judgment, the Supreme Court asked its New York litigants to brief the Enumeration Clause question as well.[78]

IV. The Supreme Court

On June 27, 2019, the last day of the 2018 Term, the Court handed down a split opinion with shifting majorities, affirming in part and reversing in part.[79] First, the Court addressed an important dispute over standing. The government contended that any harm to the plaintiffs occasioned by the citizenship question depended on the actions of individuals unlawfully refusing to respond to the Census, and that those independent legal violations broke any chain of causation tying Secretary Ross’s decision to the plaintiffs’ alleged harms.[80] The Court unanimously rejected that contention. It credited the trial court’s finding that individual nonresponse, while unlawful, was also predictable rather than speculative and that the injury required for Article III standing demanded no more.[81]

On the merits, the Court’s more conservative justices rejected the Enumeration Clause claim, agreeing with the trial court that the government had the constitutional authority to ask demographic questions—including a question on citizenship, if it chose—along with a headcount.[82] Unlike the trial court, however, the same bloc of justices also rejected the APA claim that Secretary Ross’s decision was not supported by the evidence before him.[83] Taking the secretary at his word that more precise data on citizenship were desirable for enforcing the VRA, the Court found that the secretary evaluated various means to get more precise data and selected a final approach consonant with, if not required by, the available evidence of likely harm and the merits and drawbacks of alternatives.[84] That is, the Court determined that the administrative record showed enough homework to justify the decision, even if it meant considering and rejecting the recommendations of career officials the secretary supervised.

A different majority of the Court, however, refused to take the secretary at his word as to the real reason for the decision. Chief Justice Roberts, joining the four more progressive justices, found that the evidence—the full administrative record and the extra-record discovery justified by the irregularities in the full administrative record—revealed that the secretary’s proffered explanation was a pretext.[85] Per the Court:

That evidence showed that the Secretary was determined to reinstate a citizenship question from the time he entered office; instructed his staff to make it happen; waited while Commerce officials explored whether another agency would request census-based citizenship data; subsequently contacted the Attorney General himself to ask if DOJ would make the request; and adopted the Voting Rights Act rationale late in the process.[86]

That is, the Court found that the secretary provided an explanation for his action “incongruent with what the record reveals about the agency’s priorities and decisionmaking process.”[87] And the absence of a genuine justification for the decision violated the APA.[88]

V.  The Aftermath

The Supreme Court’s June 27 opinion affirmed the vacatur of Secretary Ross’s decision and returned the matter to the trial court for remand back to the agency under the APA. Separately, the Maryland court had enjoined adding a citizenship question to the enumeration—independent of the cure for any APA violation—based on its interpretation of the Enumeration Clause.[89] And the government had repeatedly insisted in the courts, including in the filing convincing the Supreme Court to take up the case directly, that June 30 was the practical deadline for any decision on the content of the enumeration.[90] Though the secretary of Commerce had the legal authority to place a citizenship question on the enumeration given a plausible rationale and sufficient homework, the government was simply out of time to redo that process for 2020.

And so it was not terribly surprising when government attorneys said on July 2 that “the decision has been made to print the 2020 Decennial Census questionnaire without a citizenship question, and that the printer has been instructed to begin the printing process.”[91] The same day, Secretary Ross confirmed the start of printing without a citizenship question.[92]

The president, however, was not on the same page. On July 3, the president tweeted that news reports about dropping the question were “FAKE!” and declared that the Department of Commerce would be moving forward.[93] Later that day, government attorneys reported that they had been asked to reevaluate all available options to include the citizenship question in the enumeration questionnaire.[94] Four days later, DOJ announced that it would be replacing all of the attorneys on the case, swapping out an expert team specialized in defending federal agencies with a hodgepodge of attorneys from elsewhere in the Department. Such a move is highly unusual at best and strongly suggests attorneys’ desire to avoid potential allegations of misconduct.[95] When DOJ attempted to effectuate the change, courts in New York and Maryland declined to execute the swap without a statement of sworn reasons for the wholesale removal and assurances that the change would not prejudice the continuing litigation schedule.[96]

That particular chaos lasted only a few days. On July 11, 2019, the president issued Executive Order 13,880, declaring that he acknowledged that there was “no practical mechanism for including the question on the 2020 decennial census” and instead directing federal government agencies to supply the Census Bureau with administrative records concerning citizenship and immigration status.[97] As with other demographic characteristics, the Census Bureau already collects some administrative data with respect to citizenship; the efficacy of incremental efforts described in the Executive Order in improving the coverage or accuracy of existing records for various purposes is as yet unknown. Both the New York and Maryland courts have entered permanent injunctions reflecting the Executive Order’s representation that there is no practical mechanism for including the question on the enumeration in 2020.[98]

VI. The Upshot

The scramble described immediately above reveals one takeaway from the combat: We survived a near miss of a real constitutional crisis.[99] We are perhaps too quick to brand a dispute a constitutional crisis these days, when what we are really describing is a social crisis or policy crisis or yawning chasm where political leadership should be; invaluable norms and operative systems that we used to take for granted and that we may never recover are fraying and tearing,[100] but the institutional structures to rein in the wrongs exist, if only enough of the public cared to make it happen. When we the people are divided, our representatives reflect that divide. And many of the outrages of the day reflect broken policy and broken politics—or even a broken underlying order[101]—more than the broken execution of the governmental order the current Constitution bequeaths us. It is not that those other problems are less serious. They can be deadly. It is that they represent pilot error or design error rather than a gremlin pulling apart the wing of the plane.[102]

Direct attacks on the rule of law are different: They are threats to the system that live outside of problems with the political order and outside of problems of constitutional design. Two weeks after the Court handed down its opinion, the attorney general of the United States declared that he “agree[d] with [the president] that the Supreme Court decision was wrong,” and seemed to be searching for a new pretext to force a do-over, rather than demonstrating compliance with the principle the Court established.[103] The administration appeared to be seeking to place a question on the Census despite a voidable but still valid injunction from a federal court.[104] And the DOJ seemed to be asking government attorneys to violate their oaths as civil servants and as officers of the court in pursuit of their goal. It is possible that all of these appearances were misleading. But it looked very scary for a moment.

And then the moment was gone, which should come as a real relief. The president blinked. And no direct confrontation between the executive and the judiciary provided a twenty-first-century reboot of Andrew Jackson’s apocryphal aphorism.[105]

Another takeaway from the combat is that there may yet be an opportunity to save the accuracy of the 2020 Census. We are in no way out of the woods. The immigration environment remains toxic,[106] large segments of the public remain both afraid and confused about the content of the Census,[107] and serious logistical and budgetary hurdles remain.[108] But the fact that the citizenship question will not be included in the enumeration allows advocates for a complete count to reassure a nervous public that the instrument itself is safe, in a way that would not have been possible had Secretary Ross prevailed. There are statistical methods to compensate for errors and gaps in survey responses, but few permissible means to correct the enumeration that anchors them all. We have one opportunity to get it right, or as right as possible, every ten years.[109] The Court’s decision makes it more likely that we will be successful.

A third takeaway from the decision concerns administrative law and our expectations of the administrative state—here, the mood should be less joyful. It is to be expected that new administrations arrive with distinct preferences and priorities, and it is to be expected that agency approaches to the same substantive statutes will shift with changes in leadership. In partial dissent from the Court’s opinion, Justice Thomas said that “there is nothing even unusual about a new cabinet secretary coming to office inclined to favor a different policy direction, soliciting support from other agencies to bolster his views, disagreeing with staff, or cutting through red tape.”[110] On that we agree. Elections have consequences. And in a republic premised on at least a modicum of responsiveness to the electorate, that’s a good thing.

But administrative agencies may not create or change policy based on pure ideological preference divorced from evidence. The pursuit of pure ideology implicates a deep separation-of-powers concern: Pure policy preference is acceptable in government, but it is the province of the legislature. As directly representative bodies, we allow legislatures in the normal course to act with flawed, incomplete, or contradictory logic,[111] on flawed, incomplete, or contradictory evidence. Absent some other constitutional restriction, courts will invent plausible hypothetical support for legislative action, testing only for imagined potential rationality.[112] Congress may choose option X over option Y simply because it prefers X, as long as it is possible to conjure a legitimate goal that X might further. Because legislators are the elected mechanism for channeling public will, we normally accept what emerges from the black box of legislative process as legitimate, even without double-checking that the reasoning was sound.

In the exercise of its constitutional prerogative to legislate, Congress has often given the executive branch broad, but not unending, discretion. In particular, Congress has not wholly delegated its sole authority to act purely on policy preferences. The fundamental predicate for the administrative state is that agencies are properly granted a modicum of policy discretion because they possess valuable experience and expertise. The APA stands as the leading congressional requirement that agencies actually use that experience and expertise in the course of exercising the discretion they are given.

That is, Congress has demanded that agencies give reasons for their action and do at least some degree of public homework showing that the extant evidence indicates that taking the action will help accomplish the proffered reason.[113] Even the most sweeping substantive delegation arrives with this procedural limitation, tying agency action to at least a thread of experience and expertise over and above pure preference.

Congress may constitutionally act based on nothing more than “because I said so,” as long as the outcome is capable of being explained by some post hoc logic. Agencies may not. And in preserving pure policy preference as a legislative prerogative, the APA helps to maintain the constitutional separation of powers.[114]

Even without accounting for real motive, Secretary Ross’s decision on the citizenship question came awfully close to pure policy preference. It is unremarkable for cabinet members to have instincts for policy change within the bounds of discretion granted by Congress, test those instincts, and arrive at conclusions ultimately in line with those instincts. It is unremarkable for evidence to be equivocal or uncertain and for decisionmakers to evaluate the evidence and arrive at a contested but reasoned outcome based on the weight or importance assigned to particular pieces of the puzzle that others might prioritize differently. It is a different matter entirely to have the answer first and attempt to build a record to support the preconceived answer, ignoring evidence that contradicts the predetermined solution.

Justice Breyer’s dissent in the recent litigation catalogs the evidence available to Secretary Ross and convincingly demonstrates that the evidence that was not inexplicably short-circuited by the secretary’s decision actually pointed overwhelmingly against it.[115] He concluded that the agency decided to add the question to the decennial enumeration “on the ground that it will improve the accuracy of citizenship data, when in fact the evidence indicates that adding the question will harm the accuracy of citizenship data.”[116] The majority on this point accused the dissent of substituting its judgment for that of the secretary.[117] But the dissent was not improperly re-weighing controverted evidence; it was merely noting that there was no meaningful evidence of benefit to controvert the evidence of harm.[118]

The majority’s review of the administrative record is considerably more forgiving. It does not represent the absolute maximum of deference to administrative discretion: Justice Alito, for example, would have found the decision wholly insulated from APA review.[119] And it is admittedly difficult in the abstract to determine the precise point at which a political appointee swimming against the facts amounts to a departure from the proper administrative role rather than a permissible judgment call. But the majority’s decision here nevertheless blesses as sufficient some exceedingly shoddy homework. For members of the Court particularly concerned about administrative self-aggrandizement,[120] the thin review put forth by the majority is a lost opportunity.

Yet the final takeaway from the Census opinion—the dispositive takeaway in the legality of the secretary’s decision—is important, and it shows that this Court is not wholly prepared to abdicate its role as a check on the executive branch. Tests of fit between means and ends like heightened scrutiny, or even loose tests like arbitrary and capricious review, represent both a direct review of process and an indirect review of motive. A sufficiently poor fit between articulated ends and chosen means may indicate that the decisionmaker’s expressed intent was not the real reason for action.[121] Here, what the Court refused to do indirectly, it did directly. Though the Court blessed Secretary Ross’s evidentiary record as sufficient, it also found that he lied about his reason for acting.

It was hardly a foregone conclusion that administrative motive would matter. Three justices would have upheld the secretary’s action as consistent with the APA because the agency had “articulate[d] a satisfactory explanation for its action,” and would apparently have accepted that explanation even if it were acknowledged to be false.[122] Moreover, recent controversies in other contexts gave some support to the notion that the Court might prove all too willing to duck any serious examination of executive motive. Just a few weeks earlier, the Court seriously limited the availability of First Amendment claims for retaliatory arrests.[123] And in the final week of last year’s Term, the Court shut down meaningful inquiries into government purpose in the context of the Trump administration’s travel ban.[124]

But these decisions did not stand for a global retreat from consideration of motive. The Court’s retaliatory arrest decision was steeped in the practicalities of policing, and even then it did not shut the door entirely: Claims may proceed even in the presence of probable cause to arrest when a litigant can present “objective evidence that he was arrested when otherwise similarly situated individuals not engaged in the same sort of protected speech had not been,”[125] or when he can “prove the existence and enforcement of an official policy motivated by retaliation.”[126] And the strikingly constrained review in Trump v. Hawaii was expressly predicated on a level of extreme deference due executive action affecting the entry of foreign nationals into the country.[127] These cases seem to represent exceptions from the judicial capacity to examine the reasons for executive action, rather than a new normal.[128]

Serious second-guessing of governmental motive should be rare and subject to a high standard of proof. There are legitimate concerns about the intrusive nature of fact-finding coerced in the name of the judicial process when determining why government officials acted as they did, with opportunities for potential abuse by partisan opponents.[129] And there are legitimate concerns that judges overly eager to strike policies they view as unwise may turn to skepticism about motive as convenient cover.

But both of these concerns are inherent in the judicial review of governmental action. They do not imply that it is necessary to throw the baby out with the bathwater, disarming judicial response when irregularities are abundantly apparent. Discovery can (and should) be managed, and evidentiary thresholds can (and should) be adjusted to guard against renegade invalidation. What we gain in exchange is a measure of judicial protection against the ultra vires application of government power.

Agencies owe Congress, and the public, an explanation for their actions. To ensure adequate accountability, the explanation provided must be the real explanation.[130] Courts are not particularly well equipped to assess the full veracity of the explanation provided—to determine whether the explanation offered by an agency represents the truth, the whole truth, and nothing but the truth. But they can evaluate whether the explanation provided is, in context, reasonably plausible. And by refusing to accept explanations that are not, courts provide a valuable service in policing the extreme boundaries of agency accountability. It is not necessary for courts to be able to detect every half-truth in order to provide a service in calling out obvious lies.

That is precisely what happened in the Census case. The notion that Secretary Ross decided to include a question about citizenship in the decennial enumeration in order to allow the DOJ to better enforce the Voting Rights Act simply could not hold water as the real reason for his action. As the majority explained, “Our review is deferential, but we are ‘not required to exhibit a naiveté from which ordinary citizens are free.’ United States v. Stanchich, 550 F.2d 1294, 1300 (CA2 1977) (Friendly, J.).”[131]

Had the secretary offered his real reason or reasons, Congress would have been able to determine whether the agency’s experience and expertise were used appropriately, in the service of the authority the legislature delegated or at odds with that delegation. The pretext deprived Congress of that opportunity. And so the Court found, correctly, that the resulting action could not stand. The outcome of the case is, let us hope, unusual. But the legal principle is hearteningly sound.[132]

*Levitt is Associate Dean for Research, Professor of Law, and Gerald T. McLaughlin Fellow, Loyola Law School, Los Angeles. Some of this piece draws on my experience as a Deputy Assistant Attorney General in the Civil Rights Division of the Department of Justice, but nothing herein should be understood to reflect any official position of the DOJ. I am grateful to Sara Rohani and Thomas Tai for their research assistance. All errors, of course, are my own.

[1] See U.S. Const. art. I, § 2, cl. 3.

[2] U.S. Const. amend. XIV, § 2; id. art. II, § 1, cl. 2; 2 U.S.C. § 2a(a) (2012).

[3] In the 1960s, the Court ruled that constitutional claims challenging legislative districts that were unequal in population were justiciable. Baker v. Carr, 369 U.S. 186 (1962). The Court went on to demand rough equality of representation, based on roughly equal headcount, from legislative districts at every level of government. See Wesberry v. Sanders, 376 U.S. 1, 17–18 (1964) (applying the equal representation principle to congressional districts); Reynolds v. Sims, 377 U.S. 533, 577 (1964) (applying a similar principle to state legislative districts); Avery v. Midland Cty., 390 U.S. 474, 484–86 (1968) (applying a similar principle to local government districts).

[4] See U.S. Const. art. I, § 2, cl. 3; Act of July 14, 1798, ch. 75, 1 Stat. 597; Bruce Ackerman, Taxation and the Constitution, 99 Colum. L. Rev. 1, 24 & n.86 (1999).

[5] See Marisa Hotchkiss & Jessica Phelan, U.S. Census Bureau, Uses of Census Bureau Data in Federal Funds Distribution 3–10 (2017); Andrew Reamer, The George Washington Inst. of Pub. Policy, Counting for Dollars 2020: The Role of the Decennial Census in the Geographic Distribution of Federal Funds 5 tbl.2 (2017).

[6] See Thomas P. Wolf & Brianna Cea, A Critical History of the United States Census and Citizenship Questions, 108 Geo. L.J. Online 1 (2019).

[7] Letter from Wilbur Ross, Sec’y of Commerce, U.S. Dep’t of Commerce, to Karen Dunn Kelley, Under Sec’y for Econ. Affairs, U.S. Dep’t of Commerce (Mar. 26, 2018) [hereinafter Ross, Decision Letter].

[8] U.S. Census Bureau, U.S. Dep’t of Commerce, American Community Survey: Information Guide 6 (2017).

[9] U.S. Census Bureau, U.S. Dep’t of Commerce, The American Community Survey 28 (2018).

[10] Bureau of the Census, U.S. Dep’t of Commerce, 1950 Census of Population and Housing (1950).

[11] Public Trust in Government: 1958–2019, Pew Research Ctr. (Apr. 11, 2019). The earliest records date back to 1958, the first year that the National Election Study began reporting “trust in government” figures for the federal government in a manner comparable to the way the question is asked today. Id.

[12] Id. Similarly, 74.5% of Americans in 2017 reported being “afraid” or “very afraid” of corrupt government officials, making it the single largest source of fear among the population. See America’s Top Fears 2017, Chapman Univ. (Oct. 11, 2017).

[13] Letter from Vincent P. Barabba et al., Former Dirs., U.S. Census Bureau, to Wilbur L. Ross, Sec’y of Commerce, U.S. Dep’t of Commerce 1–2 (Jan. 26, 2018); Fed’n for Am. Immigration Reform v. Klutznick, 486 F. Supp. 564, 568 (D.D.C. 1980).

[14] In assessing the likely quality of the data to be gained from asking about citizenship in the enumeration, the Census Bureau compared answers given by individual respondents on the ACS against administrative data, and found that “nearly one third of [individuals listed as] noncitizens in the administrative data respond to the questionnaire indicating they are citizens.” Summary Analysis of the Key Differences Between Alternative C and Alternative D, Joint Appendix 145, at 147, Dep’t of Commerce v. New York, 139 S. Ct. 2551 (2019) (No.18-966), 2019 WL 1114907. That is, about one-third of individuals listed as noncitizens in bureaucratic administrative data said in response to the ACS—an instrument on which the question is less salient than the enumeration—that they were citizens. Id.; see also J. David Brown et al., Understanding the Quality of Alternative Citizenship Data Sources for the 2020 Census 21, 23 (Ctr. for Econ. Studies, Working Paper No. CES 18-38, 2018) (finding a 37.6% discrepancy). It is possible that the administrative data are outdated and do not capture more recent naturalization, and it is possible that the administrative data are incorrect (or incorrectly linked to census responses). But it seems at least as likely that some of the respondents were falsely presenting themselves as citizens in their responses to the questionnaire.

[15] See Justin Levitt, Citizenship and the Census, 119 Colum. L. Rev. 1355, 1363-67 (2019).

[16] See, e.g., Brown et al., supra note 14, at 8–12; William P. O’Hare, Ctr. On Poverty and Inequality, Citizenship Question Nonresponse 11–12 (2018). The rates at which individuals refused to respond to the American Community Survey were higher in 2015 and 2016 than ever before in the survey’s history. See American Community Survey: Response Rates, U.S. Census Bureau (last visited July 23, 2019). This effect was broadly distributed: In twenty-nine different states, the rate at which individuals refused to respond to the ACS was higher in 2016 than ever before; in forty-four different states, the rate at which individuals refused to respond to the ACS was higher in either 2015 or 2016 than ever before. Id.

[17] Mikelyn Meyers, U.S. Census Bureau, Respondent Confidentiality Concerns and Possible Effects on Response Rates and Data Quality for the 2020 Census 15 (2017).

[18] See Levitt, supra note 15, at 1369-70.

[19] See Ross, Decision Letter, supra note 7, at 4.

[20] Memorandum from John M. Abowd, Chief Scientist and Assoc. Dir. for Research and Methodology, U.S. Census Bureau, to Wilbur L. Ross, Jr., Sec’y of Commerce (Jan. 19, 2018) [hereinafter Abowd Memorandum, Joint Appendix], Joint Appendix 104, at 111, Dep’t of Commerce v. New York, 139 S. Ct. 2551 (2019) (No.18-966), 2019 WL 1114907.

[21] Id. at 115-16.

[22] Questions on the Jan 19 Draft Census Memo on the DOJ Citizenship Question Reinstatement Request no. 21, Joint Appendix 123, at 137, Dep’t of Commerce v. New York, 139 S. Ct. 2551 (2019) (No.18-966), 2019 WL 1114907.

[23] Ross, Decision Letter, supra note 7, at 4.

[24] See Levitt, supra note 15, at 1367-68.

[25] See California v. Ross, 358 F. Supp. 3d 965, 984-85 (N.D. Cal. 2019).

[26] Ross, Decision Letter, supra note 7, at 2.

[27] Letter from Arthur E. Gary, Gen. Counsel, Justice Mgmt. Div., U.S. Dep’t of Justice, to Dr. Ron Jarmin (Dec. 12, 2017) [hereinafter DOJ Letter].

[28] Thornburg v. Gingles, 478 U.S. 30, 50 (1986).

[29] Bartlett v. Strickland, 556 U.S. 1, 14–15, 18–20 (2009) (plurality opinion).

[30] Gingles, 478 U.S. at 51-53.

[31] See, e.g., United States v. Brown, 561 F.3d 420, 435 (5th Cir. 2009); Bone Shirt v. Hazeltine, 461 F.3d 1011, 1022 (8th Cir. 2006).

[32] Demographic information is necessary, but in no way sufficient, for the assessment of electoral power. See generally Justin Levitt, Quick and Dirty: The New Misreading of the Voting Rights Act, 43 Fla. St. U. L. Rev. 573 (2016).

[33] See Levitt, supra note 15, at 1375.

[34] Id. at 1380-82.

[35] Fabela v. City of Farmers Branch, No. 3:10-cv-01425, 2012 WL 3135545 (N.D. Tex. Aug. 2, 2012).

[36] Id. at *4-5.

[37] These limitations include the margin of error in any survey, and the fact that ACS data are most accurate when collected over a five-year span, which necessarily underestimates the size of populations that are swiftly growing. See Levitt, supra note 15, at 1378-80.

[38] See, e.g., United States v. Alamosa County, 306 F. Supp. 2d 1016, 1022 (D. Colo. 2004); see also United States’ Brief in Opposition to Defendants’ Motion to Exclude Bayesian Improved Surname Geocoding (BISG) Evidence, United States v. City of Eastpointe, 378 F. Supp. 3d 589 (E.D. Mich. 2019) (No. 4:17-cv-10079), 2018 WL 4144225.

[39] Congress explained, when it expanded VRA coverage to include language minorities, that “persons . . . of Spanish heritage” under the statute, 52 U.S.C. § 10310(c)(3), included “‘persons of Spanish language’ as well as ‘persons of Spanish surname’ in Arizona, California, Colorado, New Mexico, and Texas.” S. Rep. No. 94-295, at 24 n.14 (1975).

[40] Fabela, 2012 WL 3135545, at *8.

[41] 140+ Civil Rights Groups Call on Commerce Department to Strike Unnecessary Citizenship Question, The Leadership Conference on Civil & Human Rights (Aug. 1, 2018); Nick Brown, U.S. Census Citizenship Question Panned by Scientists, Civil Rights Groups, Reuters (Aug. 9, 2018).

[42] Email from Ellen Herbst, CFO & Ass’t Sec’y for Admin., U.S. Dep’t of Commerce, to David Langdon, Senior Policy Advisor, U.S. Dep’t of Commerce (Feb. 2, 2017).

[43] Email from Earl Comstock to Wilbur Ross (May 2, 2017, 02:19 PM), Joint Appendix 276, at 276, Dep’t of Commerce v. New York, 139 S. Ct. 2551 (2019) (No.18-966), 2019 WL 1470266.

[44] Id. (emphasis added).

[45] See Dep’t of Commerce v. New York, 139 S. Ct. 2551, 2575 (2019).

[46] See Levitt, supra note 15, at 1388.

[47] See id. at 1388-90.

[48] See id. at 1390-97. Evidence discovered quite dramatically in mid-2019 lends incremental support to this latter possibility. Longtime Republican redistricting consultant Thomas Hofeller passed away in 2018. There were contested proceedings concerning his surviving spouse and the estate; and in the course of a conversation with Common Cause about a referral for those proceedings, Hofeller’s estranged daughter mentioned that she’d found hard drives among his effects. Charles Bethea, A Father, A Daughter, and the Attempt to Change the Census, New Yorker (July 12, 2019); Michael Wines, Deceased G.O.P. Strategist’s Hard Drives Reveal New Details on the Census Citizenship Question, N.Y. Times (May 30, 2019). Common Cause subpoenaed the hard drives for North Carolina redistricting litigation and found, inter alia, a 2015 Hofeller study concluding that drawing districts based on citizen voting-age population—a “radical departure” from the status quo—“would be advantageous to Republicans and Non-Hispanic Whites,” but that it would be necessary to ask a citizenship question on the enumeration in order to get the data to make that change. NYIC Plaintiffs’ Motion for Sanctions exh. 6 at 8-9, New York v. U.S. Dep’t of Commerce, No. 1:18-cv-02921 (S.D.N.Y. July 16, 2019). Hofeller had spoken about adding a citizenship question to the enumeration with Secretary Ross’s advisor Mark Neuman, who had in turn delivered a draft letter concerning the need for data to the DOJ official actually drafting the request for citizenship data; a paragraph in Hofeller’s files concerning the Voting Rights Act justification for asking the question was incorporated verbatim in the draft letter Neuman gave to DOJ. See Kravitz v. U.S. Dep’t of Commerce, 382 F. Supp. 3d 393, 398-99 (D. Md. 2019). The variance between the Commerce Department’s origin story of the question, including in litigation, and the story suggested by these documents has become the subject of a motion for sanctions. NYIC Plaintiffs’ Motion for Sanctions, supra.

[49] Complaint, California v. Ross, No. 3:18-cv-01865 (N.D. Cal. Mar. 26, 2018).

[50] See Second Amended Complaint, New York v. U.S. Dep’t of Commerce, No. 1:18-cv-02921 (S.D.N.Y. July 25, 2018).

[51] Id. at 54; U.S. Const. art. I, § 2, cl. 3; id. amend. XIV, § 2.

[52] Second Amended Complaint at 54-55, New York v. U.S. Dep’t of Commerce, No. 1:18-cv-02921.

[53] Id. at 56-57.

[54] Complaint at 61-66, N.Y. Immigration Coal. v. U.S. Dep’t of Commerce, No. 1:18-cv-05025 (S.D.N.Y. June 6, 2018). Because the claims ran against the federal government, the equal protection claims were Fifth Amendment claims predicated on reverse incorporation. See Bolling v. Sharpe, 347 U.S. 497, 500 (1954).

[55] See, e.g., First Amended Complaint at 13-15, California v. Ross, No. 3:18-cv-01865 (N.D. Cal. May 4, 2018); Third Amended Complaint at 52-59, Kravitz v. U.S. Dep’t of Commerce, No. 8:18-cv-01041 (D. Md. Dec. 28, 2018); Second Amended Complaint at 37-39, NAACP v. Bureau of the Census, No. 8:18-cv-00891 (D. Md. Apr. 1, 2019).

[56] Complaint at 29-30, City of San Jose v. Ross, No. 5:18-cv-02279 (N.D. Cal. Apr. 17, 2018).

[57] First Amended Complaint at 106-09, La Unión del Pueblo Entero, No. 8:18-cv-01570 (D. Md. July 9, 2018); see 42 U.S.C. § 1985(3); Enforcement Act of 1871, Pub. L. No. 42-22, § 2, 17 Stat. 13 (1871).

[58] Complaint at 24-26, Elect. Privacy Info. Ctr. v. U.S. Dep’t of Commerce, No. 1:18-cv-02711 (D.D.C. Nov. 20, 2018). The suit was dismissed in June 2019 for lack of standing, Elec. Privacy Info. Ctr. v. U.S. Dep’t of Commerce, 928 F.3d 95 (D.C. Cir. 2019).

[59] Complaint at 13-16, NAACP Legal Def. & Educ. Fund, Inc., v. U.S. Dep’t of Justice, No. 1:18-cv-09363 (S.D.N.Y. Oct. 12, 2018). As of publication, this litigation continues. Separately, Alabama has sued based not on what the Department of Commerce did do, but what it has not done. Alabama sued in order to force the Department to exclude from the enumeration undocumented individuals, as allegedly not “persons in each State.” U.S. Const. amend. IV, § 2; Complaint, Alabama v. U.S. Dep’t of Commerce, No. 2:18-cv-00772 (N.D. Ala. May 21, 2018).

[60] See Plaintiffs’ Memo. of Law in Support of their Motion to Amend Judgment on Remand Pursuant to Rule 59(E), or for Injunctive Relief Pursuant to the All Writs Act at 11-15, New York v. U.S. Dep’t of Commerce, No. 1:18-cv-02921 (S.D.N.Y. July 5, 2019) [hereinafter Motion to Amend Judgment].

[61] New York v. U.S. Dep’t of Commerce, 315 F. Supp. 3d 766, 774-75 (S.D.N.Y. 2018).

[62] Id. at 775.

[63] New York v. U.S. Dep’t of Commerce, 333 F. Supp. 3d 282 (S.D.N.Y. 2018), vacated as moot by 351 F. Supp. 3d 502, 680 (S.D.N.Y. 2019).

[64] Id. at 286.

[65] Id. at 286-90.

[66] See In re Dep’t of Commerce, 139 S. Ct. 16 (2018); In re Dep’t of Commerce, 139 S. Ct. 566 (2018). The Court never substantively addressed the propriety of the Ross deposition: the deposition order was vacated after the trial court issued its substantive decision, New York v. U.S. Dep’t of Commerce, 351 F. Supp. 3d at 680, and the Supreme Court vacated the remaining underlying disputes after issuing its substantive decision. See Dep’t of Commerce v. U.S. Dist. Ct. S.D.N.Y., No. 18-557, 2019 WL 2649788 (U.S. June 28, 2019).

[67] See New York v. U.S. Dep’t of Commerce, No. 18-CV-2921, 2018 WL 6060304 (S.D.N.Y. Nov. 20, 2018).

[68] New York v. U.S. Dep’t of Commerce, 351 F. Supp. 3d at 679. Given the pending Supreme Court review of the propriety of extra-record evidence, the court carefully justified its opinion on the administrative record alone—but then also noted the consonance of the accumulated extra-record evidence. Id. at 635-36.

[69] Id. at 669-71.

[70] These included the requirement for the Census to seek information from sources other than direct inquiries where possible, like the administrative records career Census staff had recommended, see supra text accompanying note 22, and the requirement to present topics to Congress at least two years before the Census date itself. See New York v. U.S. Dep’t of Commerce, 351 F. Supp. 3d at 636-47.

[71] Id. at 647-54.

[72] Id. at 654-60.

[73] Id. at 660; see also id. at 660-64.

[74] Pet. for Writ of Certiorari before Judgment at *13-*14, U.S. Dep’t of Commerce v. New York, No. 18-966 (U.S. Jan. 25, 2019), 2019 WL 338906.

[75] Dep’t of Commerce v. New York, 139 S. Ct. 953 (2019).

[76] California v. Ross, 358 F. Supp. 3d 965, 1048-49 (N.D. Cal. 2019). As in New York, the government petitioned for certiorari before judgment, and the Court ultimately vacated the trial court’s decision in light of its final decision in the New York litigation. Ross v. California, No.18-1214, 2019 WL 1243674 (U.S. June 28, 2019).

[77] Kravitz v. U.S. Dep’t of Commerce, 366 F. Supp. 3d 681 (D. Md. 2019). As in New York, the Maryland court rejected the equal protection (and conspiracy) claim, finding insufficient evidence of the intent to discriminate. Id. at 752-54.

[78] Dep’t of Commerce v. New York, 139 S. Ct. 1316 (2019).

[79] Dep’t of Commerce v. New York, 139 S. Ct. 2551 (2019).

[80] Id. at 2565-66; see also 13 U.S.C. § 221.

[81] Dep’t of Commerce v. New York, 139 S. Ct. at 2566.

[82] Id. at 2566-67.

[83] Id. at 2569-71. The same justices also reversed the trial court’s holdings with respect to the other claimed statutory violations. Id. at 2571-73; see supra note 70.

[84] Dep’t of Commerce v. New York, 139 S. Ct. at 2569-71. Curiously, the Court misstated the evidence of harm even in validating the secretary’s reliance on it. Following the misrepresentation of the Solicitor General, Transcript of Oral Argument at 89, Dep’t of Commerce v. New York, No. 18-966 (U.S. Apr. 23, 2019), 2019 WL 1778161, the Court stated that “[t]he Bureau predicted a 5.1% decline in response rates among noncitizen households if the citizenship question were reinstated.” Dep’t of Commerce v. New York, 139 S. Ct. at 2570. This is untrue. As mentioned above, the Bureau found that it was likely that at least 5.1% of households with noncitizens would decline to respond, but that this was a “lower bound” and a “conservative estimate,” and that the real rate “could be much greater.” See supra text accompanying notes 20-21; Brief for Gov’t Respondents 10, Dep’t of Commerce v. New York, 18-966 (U.S. Apr. 1, 2019), 2019 WL 1468270.

[85] Dep’t of Commerce v. New York, 139 S. Ct. at 2574-76.

[86] Id. at 2574.

[87] Id. at 2575.

[88] Id. at 2576.

[89] Kravitz v. Dep’t of Commerce, 366 F. Supp. 3d 681, 755-56 (D. Md. 2019). Though the Supreme Court’s decision on the Enumeration Clause, Dep’t of Commerce v. New York, 139 S. Ct. at 2566-67, likely rendered that injunction voidable, no filing had requested that relief before the end of June.

[90] See Motion to Amend Judgment, supra note 60, at 11-15.

[91] Id. at 8.

[92] Id.

[93] Donald Trump (@realdonaldtrump), Twitter (July 3, 2019, 11:06 AM),

[94] Motion to Amend Judgment, supra note 60, at 9-10.

[95] See Katie Benner, Barr Says Legal Path to Census Citizenship Question Exists, but He Gives No Details, N.Y. Times (July 8, 2019); Michael Wines et al., Justice Dept. to Replace Lawyers in Census Citizenship Question Case, N.Y. Times (July 7, 2019); Marty Lederman (@marty_lederman), Twitter (July 8, 2019, 3:31 PM),

[96] Memorandum Opinion and Order, New York v. U.S. Dep’t of Commerce, No. 1:18-cv-02921, 2019 WL 2949908 (S.D.N.Y. July 9, 2019); Memorandum Opinion and Order, Kravitz v. U.S. Dep’t of Commerce, No. 8:18-cv-01041 (D. Md. July 10, 2019). It appears that the effort to swap out the attorneys was abandoned; though new attorneys entered appearances in the cases, the government did not pursue withdrawal orders for those who had been representing the Department of Commerce since the inception of the litigation.

[97] Exec. Order 13,880, 84 Fed. Reg. 33,821 (July 16, 2019). The Executive Order describes several reasons to pursue these additional administrative records beyond the Voting Rights Act rationale found to be pretextual; none of these reasons appears to warrant the collection of citizenship data with sufficient precision to justify the inclusion of the question in the decennial enumeration. Justin Levitt, Trump’s Executive Order on the Census, Election L. Blog (July 12, 2019).

[98] Order, New York v. U.S. Dep’t of Commerce, No. 1:18-cv-02921 (S.D.N.Y. July 16, 2019); Order, Kravitz v. U.S. Dep’t of Commerce, No. 8:18-cv-01041 (D. Md. July 16, 2019).

[99] See, e.g., Linda Greenhouse, A ‘Train Wreck’ Was Averted at the Supreme Court, but for How Long?, N.Y. Times (July 18, 2019).

[100] See generally Neil S. Siegel, Political Norms, Constitutional Conventions, and President Donald Trump, 93 Ind. L.J. 177 (2018); Daphna Renan, Presidential Norms and Article II, 131 Harv. L. Rev. 2187 (2018).

[101] See generally Sanford Levinson, Our Undemocratic Constitution: Where the Constitution Goes Wrong (And How We the People can Correct It) (2008).

[102] Cf. The Twilight Zone: Nightmare at 20,000 Feet (CBS television broadcast Oct. 11, 1963).

[103] Mike Balsamo, Barr Sees A Way for Census to Legally Ask About Citizenship, AP News (July 8, 2019).

[104] See supra note 89.

[105] In response to Worcester v. Georgia, 31 U.S. 515 (1832), regarding tribal sovereignty, President Andrew Jackson is often quoted as saying, “Marshall has made his ruling, now let him enforce it!” The remark is likely apocryphal. See, e.g., Matt Ford, When the President Defies the Supreme Court, New Republic (Apr. 24, 2018).

[106] See, e.g., Gwendolyn Wu et al., Few ICE Raids, but Much-Hyped Plans Stoke Fears in Immigrant Communities, S.F. Chron. (July 14, 2019); Lindsay M. Harris, Trump’s New Asylum Rule Will Guarantee More Separated Families, Wash. Post (July 17, 2019); Ted Hesson, Trump Officials Pressing to Slash Refugee Admissions to Zero Next Year, Politico (July 18, 2019).

[107] See, e.g., Alexis Dominguez, Scam or Not: Census Test Mailed Out May Seem Fake, but It’s Legitimate, KRDO (July 12, 2019) (describing a test of the decennial enumeration containing the citizenship question, mailed out while the Supreme Court case was pending and still in the field after the Court’s decision).

[108] See, e.g., Caitlin Emma, Budget Gridlock Imperils 2020 Census, Politico (July 15, 2019); Jeffrey Mervis, Census Citizenship Question is Dropped, but Challenges Linger, 365 Science 211 (2019); U.S. Gov’t Accountability Office, 2020 Census: Actions Needed to Address Key Risks to A Successful Enumeration GAO-19-588T (2019).

[109] See, e.g., David Russell, ‘We Get One Chance Every 10 Years, Queens Chron. (July 18, 2019).

[110] Dep’t of Commerce v. New York, 139 S. Ct. 2551, 2580 (2019) (Thomas, J., concurring in part and dissenting in part) (internal quotation marks omitted).

[111] Cf. Lawrence v. Texas, 539 U.S. 558, 604 (2003) (Scalia, J., dissenting) (noting that legislatures “need not carry things to their logical conclusion”).

[112] See, e.g., City of New Orleans v. Dukes, 427 U.S. 297, 303-05 (1976).

[113] Hence the requirement, made clear in the Senate Report to the APA, that “[t]he agency must analyze and consider all relevant matter presented,” and that “[t]he required statement of the basis and purpose of rules issued should not only relate to the data so presented but with reasonable fullness explain the actual basis and objectives of the rule.” S. Rep. No. 752, at 201 (1945).

[114] Cf., e.g., Kisor v. Wilkie, 139 S. Ct. 2400, 2432 (Gorsuch, J., concurring in the judgment) (“Some have even described [the APA] as a kind of constitution for our ‘administrative state.’”); Todd David Peterson, Procedural Checks: How the Constitution (and Congress) Control the Power of the Three Branches, 13 Duke J. Const. L. & Pub. Pol’y 209, 251-59 (2017) (describing the APA as a means of enforcing the separation of powers, albeit without focus on the function of the APA in requiring deployment of experience and expertise).

[115] Dep’t of Commerce v. New York, 139 S. Ct. 2551, 2587-93 (2019) (Breyer, J., concurring in part and dissenting in part).

[116] Id. at 2592.

[117] Id. at 2571 (majority opinion).

[118] Id. at 2592-93 (Breyer, J., concurring in part and dissenting in part).

[119] Id. at 2596-2606 (Alito, J., concurring in part and dissenting in part).

[120] See, e.g., Gundy v. United States, 139 S. Ct. 2116, 2133-35 (2019) (Gorsuch, J., dissenting); Kisor v. Wilkie, 139 S. Ct. 2400, 2425-42 (Gorsuch, J., concurring in the judgment).

[121] See, e.g., Elena Kagan, Private Speech, Public Purpose: The Role of Governmental Motive in First Amendment Doctrine, 63 U. Chi. L. Rev. 413, 453-55 (1996).

[122] Dep’t of Commerce v. New York, 139 S. Ct. at 2577-79 (Thomas, J., concurring in part and dissenting in part) (determining that “ordinary arbitrary-and-capricious review,” on the face of the materials presented, should end the inquiry).

[123] See Nieves v. Bartlett, 139 S. Ct. 1715 (2019).

[124] See Trump v. Hawaii, 138 S. Ct. 2392 (2018); Abbott v. Perez, 138 S. Ct. 2305 (2018).

[125] Nieves, 139 S. Ct. at 1727.

[126] Lozman v. City of Riviera Beach, 138 S. Ct. 1945, 1954 (2018).

[127] See Hawaii, 138 S. Ct. at 2418-20.

[128] Though government action enjoys the presumption of regularity, see, e.g., Dep’t of Commerce v. New York, 139 S. Ct. 2551, 2577 (2019) (Thomas, J., concurring in part and dissenting in part), many legal doctrines turn on the assessment of an executive actor’s impetus for action, and the Court will set aside that presumption of regularity where it feels the evidence so warrants. See, e.g., Flowers v. Mississippi, 139 S. Ct. 2228 (2019) (racial discrimination); J.E.B. v. Alabama, 511 U.S. 127 (1994) (gender discrimination); Masterpiece Cakeshop v. Colo. Civil Rights Comm’n, 138 S. Ct. 1719 (2018) (religious discrimination); Crawford-El v. Britton, 523 U.S. 574 (1998) (First Amendment); Florida v. Jardines, 133 S. Ct. 1409 (2013) (Fourth Amendment); Missouri v. Siebert, 542 U.S. 600 (2004) (Fifth Amendment); Michigan v. Bryant, 562 U.S. 344 (2011) (Sixth Amendment); Hudson v. McMillian, 503 U.S. 1 (1992) (Eighth Amendment).

[129] See, e.g., Dep’t of Commerce v. New York, 139 S. Ct. at 2583 (Thomas, J., concurring in part and dissenting in part) (warning that “the Court’s decision enables partisans to use the courts to harangue executive officers through depositions, discovery, delay, and distraction”); cf. Clinton v. Jones, 520 U.S. 681, 719-24 (1997) (Breyer, J., concurring in the judgment) (describing the potential for legal proceedings to distract from the time and energy of the president).

[130] The Senate Report for the APA makes plain what should otherwise be obvious: “The required statement of the basis and purpose of rules issued should not only relate to the data so presented but with reasonable fullness explain the actual basis and objectives of the rule.” S. Rep. No. 752, at 201 (1945) (emphasis added); see also H.R. Rep. No. 1980, at 259 (1946); Peterson, supra note 114, at 254; cf. Kenneth F. Warren, Administrative Law in the Political System 184 (3d ed. 1995) (highlighting the APA as a source of accountability).

[131] Dep’t of Commerce v. New York, 139 S. Ct. at 2575. The quotation—and, more vital, the sentiment—was perhaps particularly notable to Chief Justice Roberts, who specifically left the reference as denoted in the text. Stanchich was a unanimous appellate panel opinion, and its author would normally go unmentioned in a citation—but Roberts clerked for Judge Friendly two years after Stanchich was issued. See Todd S. Purdum et al., Court Nominee’s Life is Rooted in Faith and Respect for Law, N.Y. Times (July 21, 2005). The quote also echoes the wisdom of an earlier court, almost 100 years before: “Besides, we cannot shut our eyes to matters of public notoriety and general cognizance. When we take our seats on the bench we are not struck with blindness, and forbidden to know as judges what we see as men . . . .” Ho Ah Kow v. Nunan, 12 F. Cas. 252, 255 (C.C.D. Cal. 1879).

[132] There are cases ahead that will further develop the limits on pretextual explanations for executive action. Just one day after its decision in the Census case, the Court granted certiorari in Kelly v. United States, concerning the Bridgegate scandal in New Jersey, and criminal fraud prosecutions for the pretextual execution of a “traffic study” that was actually conducted purely for raw political payback. See, e.g., Kelly v. United States, No. 18-1059, 2019 WL 588845 (U.S. June 28, 2019); Respondent William Baroni’s Brief in Support of the Petition for a Writ of Certiorari, Kelly v. United States, No. 18-1059 (U.S. Mar. 15, 2019), 2019 WL 1240050. It will be interesting to see whether the Court expands, maintains, or constricts its vision of proper inquiry into the rationale for executive conduct in this substantially different context.