Saturday, July 06, 2019

The Evenwel Gambit

Joseph Fishkin

I’m as surprised as everyone else that over the past 24 hours the Administration decided to flipflop, in such wildly Trumpian fashion, on whether to keep fighting to add a citizenship question to the Census. Our government’s clown show approach raises the usual questions of whether it is better or worse to temper malevolence with incompetence. (The Census fight seems to be one of those instances where the incompetence helpfully undermines the malevolence, rather than compounding it; other examples point the other way.)

Now that the administration is deciding to take more time to huddle together and settle on a new cover story, it seems to me that their most likely path is also one of the most dangerous: claim that in fact we need to add the citizenship question in order to redistrict in 2021 on the basis of citizen voting age population or CVAP, rather than on the basis of total population.

To review: Many of us argued from the start of this Census question fiasco that behind the flagrantly pretextual “we want to enforce the Voting Rights Act” cover story, there seemed to be two related but distinct reasons the Trump Administration was trying to add this question to this Census. The first—call it Real Reason 1—was to cause a targeted drop in the Census response rate among some large groups of people who don’t vote Republican often enough (mainly respondents with immigrants in their households or communities). The second—call it Real Reason 2—was to create block-level data on citizenship, in an effort to build a better foundation for Republican redistricters in 2021 to switch the fundamental basis of representation within states or other jurisdictions, abandoning the total population approach that is not only the near-universal American practice, but also the one specifically required by the Fourteenth Amendment, after considerable debate, in the federal context of the apportionment of congressional seats among states (that’s a topic for another post).  In place of total population, these Republican line-drawers would attempt to equalize the citizen voting age population or CVAP, which is supposed to stand in as a rough proxy for eligible voters (there is no good data for eligible voters). That way, communities that are older, and have fewer non-citizens, would get more representation, and communities that are younger and have more non-citizens would get less representation. Both of these two moves pretty obviously—in the immortal, material words of Republican redistricting mastermind Thomas Hoffeler—“would be advantageous to Republicans and Non-Hispanic Whites.”

Now that the Administration’s initial cover story has effectively been laughed out of the Supreme Court, any new effort to add the question faces obstacles. Imagine that you are the beleaguered government lawyer tasked with defending this indefensible thing in court. (Real advice: resign. Whistleblower aid is available. But let’s leave that aside.) One option you have is to try to concoct a completely new fake cover story. But that is probably a bad option. You don’t want to walk into court with a post hoc justification, concocted recently by lawyers, that was no part of the original decision-making process. That’s never a good look. Even if notionally the Administration claims to be starting completely over with a “new” decision-making process, leading to a new executive order, the haste of said process, and the involvement of high-level officials such as Secretary Ross in both processes, will make it difficult to plausibly sever the new order entirely from the original, years-long administration effort.   So what other options are there?  A second option might be to try to go with a very thin justification, of the form “we are adding this question because we just think it’s important to know how many citizens there are in this country.” But that approach is going to collapse under further questioning that my four-year old could manage (“why?”), and doesn’t seem well-calibrated to persuade a federal district judge. That leaves the third option which is this: just go with one or both of Real Reasons 1 and 2. Those reasons offer one great advantage: they’re honest.  Evidence in the record suggests that they were indeed central to the Administration’s decision-making process—evidence that, frankly, the plaintiffs have already carefully uncovered and pointed out in federal court. The only downside is that these reasons are, to their core, some combination of partisan and racial in their intent.

But will a court—ultimately, will Chief Justice John Roberts—agree with that? If the administration went with Real Reason 1, I am reasonably certain he would, and the government would surely lose again.  Having refused to rubber-stamp the VRA cover story under the bright lights on the last day of the Term, I see no good reason to believe that he would turn around over the summer and accept an official government reason that effectively said, “we are deliberately trying to selectively drive down response rates on the Census.”  Even though Roberts has recently held partisan gerrymandering claims to be nonjusticiable, it is hard to believe he would countenance such an obvious and deliberate undermining of the fundamental constitutional duty to conduct the decennial census, in the name of partisan and/or racial advantage.  

So that leaves what I’d call the Evenwel Gambit: fess up, selectively, to Real Reason 2. Claim that you want to add the citizenship question to the Census in order to enable the CVAP-based drawing of district lines in 2021. In the last round of fighting over switching from total population to CVAP—the Evenwel case from 2016, one significant issue was that the data we have about citizenship, although fairly accurate, is not that granular—it doesn't measure small enough units, like census blocks—that one would ideally want in order to comply with the exacting jurisprudence of one-person-one-vote.  (The data we have comes from the American Communities Survey (ACS), which has one interesting advantage over the Census: it uses statistical sampling, which the Census does not, and which can be essential for curing problems of nonresponse.)

I predict that the government may well conclude that the Evenwel Gambit is the best they can do: it has the great virtue of being true enough that it cannot be dismissed as post hoc pretext, yet it is not nearly so blatant in its disregard for basic norms of impartiality in governance as Real Reason 1. The case they would make is simple: CVAP and total population are two rival approaches to drawing district lines.  Each has its virtues, they would argue, with total population providing equal representation for equal numbers of people, while CVAP provides more of an equally weighted vote (an argument that collapses under close scrutiny, but that might be good enough for this purpose).  They would ask the Court to allow the Census question in order to give jurisdictions the option of choosing equally weighted votes over equal representation.

A competent district court ought to see through this gambit.  Explaining why is my real reason for writing this blog post.  There are two quite different things the government might be doing in attempting to enable jurisdictions to district by CVAP.  The first is actually drawing accurate CVAP-based maps, where voters’ votes will notionally enjoy equal weight, at the expense of equal representation.  The other is the naked pursuit of partisan and racial advantage, without regard to the accuracy of maps and without regard to the question of whether voters’ votes have an equal weight.  This is CVAPificiation as a form of partisan-and-racial constitutional hardball, with the straightforward aim of producing maps “advantageous to Republicans and Non-Hispanic Whites.”

Interestingly, there is a straightforward way to tell whether the government is pursuing the first of these or the second.  It’s actually very simple.  The test is whether the government is trying to add the citizenship question to the Census.

The point here is simple and depends only on the work of the Census Bureau itself that is already in the record in this litigation.  It is this: Estimates offered by the Bureau show that adding a citizenship question would actually undermine the overall accuracy of CVAP data available to line-drawers.  How is that possible? Because of Real Reason 1.  Asking the citizenship question of every Census respondent (as opposed to asking it on the broader ACS, where nonresponse can be accommodated through sampling) has the result of skewing the Census in a way that then predictably undermines the accuracy of your citizenship data.  If adding the citizenship question increases nonresponse rates at anything like the rate the Bureau itself predicts, the end result will be CVAP data that looks more granular but is less accurate, not more. As the Bureau told Commerce Secretary Wilbur Ross when he pressed for the citizenship question in the first place, if you want accurate CVAP data, instead of adding the question to the Census you should rely on other data sources not subject to this problem.

All this may sound a little head-spinning: if the Bureau knew all this from the beginning (and they did), then why was Real Reason 2 any kind of reason for adding a citizenship question to the Census in the first place? The answer is that the goal was never to produce accurate CVAP-based maps. The goal was to produce granular enough citizenship data to make it possible for CVAP-based maps to appear fully compliant with one person one vote, even if the underlying data was garbage.  The point was never to advance some abstract principle of equally weighted votes, but instead to further skew the partisan and racial balance of political power in this country, helping to further entrench the control of the Republican Party over more state governments, and generally to forestall the slow but inexorable rise of the political strength of groups other than “Non-Hispanic Whites.”

Judge Furman did a tremendous job with the opinion that led even Chief Justice Roberts to conclude the Court should not embarrass itself by accepting the flagrantly pretextual arguments the administration offered for adding this question to the Census. It is one of those opinions that really illustrates why the fact-finding role of district judges matters, even in the hottest cases where the Justices of the Supreme Court likely begin with the most strongly-held views.  If the Administration now attempts the Evenwel Gambit, it will be up to Judge Furman and the other judges overseeing the various venues of this sprawling litigation to enable the parties to develop a clear factual record, one that forces any higher court to confront the partisan and racial purposes that drive this sad and cynical effort to undermine the equal representation of millions of Americans in the legislatures whose laws bind us all.

Older Posts
Newer Posts