Pretext and Remedy in the Census Case and Beyond

by Michael C. Dorf (cross-posted on Take Care)

Here is (a slightly cleaned up version of) what I tweeted in the minutes after I quickly read the Supreme Court's census opinion on Thursday of last week:
Chief Justice Roberts acknowledges the very strong evidence that enforcement of the Voting Rights Act was not the real reason for Trump/Ross adding the citizenship question to the 2020 census. So the Court approves the remand to the agency. That's good. But it is not clear what happens next. When a court remands a case to an agency because the agency did not comply with proper procedures, the agency is permitted to reach the same conclusion by dotting the i's and crossing the t's. 
Pretext is different, however. The Roberts opinion already says that the agency COULD have added the citizenship question for legitimate reasons. And we can be sure that the Trump administration will claim to be doing just that. But EVERYONE KNOWS that the political impact will still be the real reason. 
Based on the Travel Ban litigation, there is reason to fear that the SCOTUS will uphold the citizenship question after the administration "lawyers it up" better. The main difference is that in the Travel Ban case, the Court never actually found pretext. Here it has. And while an otherwise legitimate decision found to be pretextual should not be forever barred, where only a few months (at most) will pass, and the motives remain the same, a new determination to include the citizenship question should be viewed with extreme skepticism.
After a couple of disclaimers about timing that are specific to this particular case, I want to say a few words about a general remedial problem posed by any rule of law that says that an action that could be lawfully taken for some reasons cannot be taken for some other reasons.

The case poses a number of timing issues, because the government has heretofore insisted that it would have to send census forms to the printer by roughly today. Meanwhile, the plaintiffs had been saying the real deadline is early October. We can now expect the parties to reverse positions on the timing question. The plaintiffs will say that the government should be estopped from delaying further; with insufficient time for the Census Bureau to jump through whatever hoops it would need to jump through in order to tell a more persuasive lie about why it is adding the citizenship question, the forms should be printed without it. Meanwhile, the government will confess error and say that the plaintiffs were right all along; printing can wait until October.

Further complicating the timing question is the president's tweet suggesting that the census should be delayed as long as necessary to get the citizenship question back on in a way that satisfies the Supreme Court, because, you know, Donald Trump is all about effective enforcement of the Voting Rights Act. The hardest hard deadline is presumably the end of 2020, in order to comply with the constitutional obligation to conduct a decennial census. However, the real deadline for conducting the census is April 1, 2020, because that is set by a statute that the Democratic-controlled House certainly will not extend in order to help Trump under-count Democratic-leaning constituencies.

Presumably, the administration was working with the April 1, 2020 deadline when it told the Supreme Court that it needed to start printing on July 1, 2019. As noted above, it's possible that there's actually some wiggle room in the schedule, but there is not indefinite wiggle room. Meanwhile, Trump has no power to extend the April 1, 2020 deadline, so his tweet has one and only one impact: It serves as further evidence that the administration is determined to add the citizenship question, regardless of what the experts in the agency conclude. In other words, viewed rationally, Trump's tweet ought to undercut his legal case.

Whether it actually will have that impact depends on just what standards the courts use to evaluate the output of the Census Bureau, should it jump through the proper hoops and conclude that yes indeed, a citizenship question should be included for reasons having nothing to do with suppressing the count and thus the representation of Democratic-leaning constituencies. Yet it remains unclear exactly what kind of process or showing would suffice.

The Court allowed that the evidence fit a narrow exception to the general proposition that the existence of mixed motives does not invalidate an otherwise permissible agency action. That exception applies to such bad faith that the agency cannot be said to have disclosed the actual basis for its challenged action. And that is what the district court and ultimately the Supreme Court found. CJ Roberts wrote for the Court that "viewing the evidence as a whole, we share the District Court’s conviction that the decision to reinstate a citizenship question cannot be adequately explained in terms of DOJ’s request for improved citizenship data to better enforce the VRA."

What happens now? Here's the crucial passage of the SCOTUS opinion:
the District Court was warranted in remanding to the agency, and we affirm that disposition.  ... We do not hold that the agency decision here was substantively invalid. But agencies must pursue their goals reasonably. Reasoned decisionmaking under the Administrative Procedure Act calls for an explanation for agency action. What was provided here was more of a distraction.
Is there anything that the agency can do that should satisfy the Court? The answer seems like a clear no. If the agency discloses what everyone knows to be the real reason for the addition of the citizenship question -- to help Republicans by undercounting the Latinx and nonwhite persons -- then the government will have solved one problem at the cost of creating another: It will have given the real reason but in so doing disclosed an equal protection violation.

But now we have a puzzle. Why did the Supreme Court (and the district court for that matter) remand to the agency at all, rather than simply forbid the use of the citizenship question? I will suggest two possibilities.

(1) Perhaps the Court is giving the Census Bureau the opportunity to purge the taint of Secretary Ross's illicit intent. How to purge the taint of an illicit motive is an under-developed question in the Court's jurisprudence, even though taint questions are increasingly prominent in the Court's work (as I discussed here, here, and here last Term in connection with the Travel Ban and other cases). But logic suggests a few propositions, which I'll set forth and then apply:

(a) A decision that could have been reached on legitimate grounds but in fact was reached on illegitimate grounds cannot simply be reinstated by the same person or body reaching the same decision and claiming that the second (or successive) version was taken for the right reason. Absent more, a court should view the same decision with skepticism.

(b) The passage of time can be the kind of "more" that purges taint. Justice Alito's decision in the Bladensburg Cross case implicitly applies this principle. Even if a monument is originally erected for religious purposes (which he does not concede in the particular case), many years later we cannot assume that the people who maintain the monument have the same motives. How much time must pass to activate this principle? Any line will be arbitrary, but any defensible line will be far longer than the period in which the administration will need to act in order to get the census case back to the SCOTUS in time to conduct the census.

(c) A change in administration could also be the kind of "more" that purges taint. But the kind of change needed here is beyond unlikely: Say, the impeachment and removal of Trump and Pence, leaving Nancy Pelosi as President and the realization of Census Bureau statisticians that, actually, the citizenship question will not substantially reduce the response rate. We can effectively dismiss this possibility.

(d) In rare circumstances, we can imagine that the same administration in a very short period of time should be permitted an action despite the fact that it previously attempted it for an illicit reason. Indeed, in the scenario I'm imagining, a court should approve even the original tainted action.

Suppose that a malevolent actor in South America raises a zombie army that begins a slow, shuffling march northward, killing all in their path. Military experts inform Congress and the President that a wall at the southern border will sufficiently slow the zombie's march to enable air power to finish them off. The President is not interested in zombies, but he signs the appropriation along with the following statement: "This beautiful wall will protect our country from dark-skinned people from shithole countries trying to steal our jobs. MAGA."

Trump's racist motive taints the wall as a violation of equal protection. Under the Arlington Heights test, a court would ask whether it would have been built even without the illicit motive. Here the answer is no: the illicit motive really is a but-for cause of the challenged action. Nonetheless, if the zombies are truly an existential threat (as I assume for purposes of my example), then the government can build it. Why? Because the illicit purpose is the equivalent of an express racial classification, but here there is a compelling interest and, if (as I also assume) the military experts are correct, building the wall is necessary to serve that compelling interest.

Returning from the realm of fantasy (the zombies, not the racist President, alas), this kind of an argument could have worked in the Travel Ban case if there were a sufficient showing of national security necessity. As the case actually was decided, however, the Court (erroneously in my view) simply deferred to the claim of national security rather than strictly scrutinizing it.

In the census case, there is no argument based on national security, so even if one thinks deference was appropriate in the Travel Ban case, it is not appropriate here. The taint raises the level of scrutiny; national security is not invoked to lower it again; the inclusion of the citizenship question must satisfy strict scrutiny; it cannot, because even if better enforcement of the Voting Rights Act is a compelling interest, given the evidence, the addition of the citizenship question (with its attendant results) is not narrowly tailored to achieve it.

Accordingly, none of the mechanisms that one might ordinarily think capable of purging the taint of an illicit motive appears to be in play in the census case. So we must consider an entirely different sort of explanation for the remand.

(2) Nowhere does the Chief Justice actually say that Ross's motive for including the census question was illicit. What he says is that the explanation--to enforce the VRA better--was not the actual reason. The remand is therefore for the purpose of allowing the administration to come up with a better explanation. That is why more than one commentator has characterized the remand as an opportunity for the administration to do a better job of lying.

Put differently, the administration might now try to create a record that shows that Ross was indeed interested in adding a citizenship question all along and not for enforcing the VRA but for some other legitimate purpose. But what legitimate purposes fits the evidence? There are really only two possibilities.

(a) More effective lying would invoke some vague notion of the importance of citizenship, compliance with UN standards (as Justice Alito discusses in his separate opinion), or some other reason that was not the real reason. But it is a genuine mystery how the Census Bureau could generate a record that would show that the real reason for the decision was anything other than Ross's and Trump's partisan aims.

(b) That brings me to the final, perhaps most dispiriting possibility: that CJ Roberts would be prepared to uphold the inclusion of the citizenship question if the administration simply admits that it was included for partisan reasons. In various places, the Roberts opinion says there's nothing wrong with political considerations influencing agency action, and he takes issue with Breyer's view that the agency should be guided by expertise, not politics.

To be sure, there is a difference between "political" in the sense of policy-based and in the sense of "partisan." The sources Roberts cites for the proposition that an agency can consider political matters use "political" in the former sense, but it is possible to imagine the administration offering a policy-sounding reason. After all, it is true that citizenship is important to this administration. Thus, it is possible that the administration could abandon the VRA charade and build a record for the proposition that it was interested in adding a citizenship question all along simply because it cares about how many citizens there are; and although that would be a lie, it is possible to imagine five justices then approving the question.

I hope that's not what's in store, but there really is nothing the administration can now do that ought to lead to approval of the citizenship question, so one can only worry that the Court has remanded for the purpose of allowing the Trump administration to get away with something. Again, I hope that worry proves unfounded, but in the last few years I find my that hopes frequently go unfulfilled.