Voter ID

Accuracy About Voting Needed on Both Sides of Debate

by

The Golden Rule fully applies: speak truthfully about voting as you would have others also speak truthfully about voting.

This recent tweet from Professor Larry Tribe caught my eye: “Call it what you like, but the # of voters turned away for not having required forms of ID exceeded margin of T’s victory in MI, Pa & Wis”

As soon as I read it, I said to myself, “That can’t be right.”

First of all, no voter ever should be “turned away” for lack of ID.  Instead, the Help America Vote Act of 2002 (HAVA) requires that voters lacking required ID receive a provisional ballot. To be sure, some poll workers may fail to enforce the mandates of HAVA, but in volumes exceeding Trump’s margins of victory in Michigan (about 11,000), Wisconsin (about 23,000), and Pennsylvania (about 44,000)??? If there had been a massive failure of election administration on that scale, which could have accounted for the outcome of the presidential election, presumably we would have heard news reports of it by now.

Just because voters cast provisional ballots does not mean, of course, that those provisional ballots will be counted. In some states, a voter who casts a provisional ballot because the voter lacked a required form of ID is not permitted simply to sign an affidavit to get the ballot counted, but instead within a limited period of time must find a way to get the required ID and show it to local election officials. It is accurate to say that provisional voters who are unable to procure the required ID within the available amount of time are disenfranchised because of the ID requirement. But in Michigan, Wisconsin, and Pennsylvania, are there more voters who fall in this category than Trump’s margin of victory in each of these states? If so, then the number of rejected provisional ballots in each state would exceed Trump’s margin of victory there. I have seen no reports of provisional ballots rejected in such volumes in any of these three states—much less rejected for the particular reason of lacking the required ID, rather than for some other reason (like not being a registered voter, or casting a ballot at an improper polling place given the voter’s home address).

Instead I saw this report about provisional ballots in Wisconsin: “618 were issued because the voter didn’t have an acceptable photo ID,” of which 502 were not counted. It is unfortunate for any eligible citizen who took the trouble to go to the polls to be disenfranchised, but 502 is a far cry from Trump’s 23,000-or-so margin of victory.

It is possible, also, that the number of voters deterred from going to the polls to cast a ballot, because of a required ID law, is greater than a winning candidate’s margin of victory.  If so, then one credibly could argue that the suppressive effect of the ID law was responsible for that particular electoral outcome. Did something like this occur in Michigan and Pennsylvania and Wisconsin? I know of no evidence attempting to make this case for any of these three states. Of course, conclusively proving the deterrent effect of an ID law might be a difficult task (how would you show that the reason that a citizen didn’t go to the polls was because of the ID law?); even so, I’m not aware of any effort to make this connection.

Moreover, there are sound reasons to doubt that any such effort could be successful for any of these three states, much less all of them. According to the National Conference of State Legislature’s very useful website, Pennsylvania’s supreme court invalidated that state’s voter ID law, and thus there is no such enforceable obstacle in place there. Michigan, moreover, is one of the states that permits a voter who lacks the required ID to sign an affidavit instead. And with respect to Wisconsin’s strict ID law, for this year’s election the Seventh Circuit required the state to provide a temporary form of ID to any eligible citizen who attempted to get the required ID but was unable to do so. While the Seventh Circuit’s temporary remedy is not as voter-friendly as Michigan’s affidavit option, it does indicate that many voters who really wanted to vote this year—and thus were not easily deterred—had an available mechanism to do so and thereby could avoid disenfranchisement as a consequence of the ID law. As a result, even in Wisconsin, do we have sound reason to believe that about 20,000 would-be voters were actually thwarted from casting a ballot because of the state’s ID law?

Finally, yes, it is possible that the number of otherwise eligible citizens lacking required ID might exceed a winning candidate’s margin of victory.  But that’s a far cry from saying that this number of citizens was deterred from voting because of the ID requirement.  They might have been eligible citizens who, for whatever reason, decided not to vote in the election.  For an ID rule to cause disenfranchisement, it must be an obstacle that prevents, or at least deters, a citizen from voting.  It’s not enough to say more potential voters lack the required ID than Trump’s margin of victory.  That’s like saying all of Jill Stein’s voters, if she had not been on the ballot, would have voted for Hillary Clinton, as opposed to exercising the option of just not voting for president at all.

In sum, I think Tribe’s tweet is wildly irresponsible—indeed demonstrably false. The assertion that “the # of voters turned away for not having required forms of ID exceeded margin of T’s victory in MI, Pa & Wis” simply does not square with HAVA, the applicable ID rules governing these three states, or available data.

Ordinarily, I would ignore such misinformation, except that it comes from a “big name” source—a prominent Harvard Law professor, who (among other well-known accomplishments) served as one of Al Gore’s leading lawyers in the 2000 recount litigation.

As celebrated as Tribe may be, he certainly is not as well known as Donald Trump. Nor is Tribe’s tweet as egregiously false as Trump’s claim that he would have won the national popular vote were it not for “the millions of people who voted illegally.” Still, if one is to decry Trump’s blatant falsehoods about the electoral process, one must also call to account leading figures on the other side of the political aisle when they also propagate outright untruths about elections. In this context, as so many others, the Golden Rule fully applies: speak truthfully about voting as you would have others also speak truthfully about voting.

In this regard, I am also deeply disappointed about the League of Women Voters, whose website prominently displays the declaration of the League and its president, Chris Carson, that the “the 2016 presidential election WAS rigged”—a declaration subsequently reported by the Washington Post.

The League of Women Voters should know better. It is supposed to be a responsible organization about the nature of the voting process. It has a venerated history that it should not squander by sullying itself with Trump-like falsehoods about the nation’s voting process.

What is the League’s evidence that this year’s election “WAS rigged”? Among the assertions that the League makes to support its outlandish allegation is the fact that several states, including Ohio and Wisconsin, cut back the amount of early voting that they previously offered—not eliminating early voting completely, but not offering it as generously as they had before. While one certainly can disagree with this rollback of early voting as a matter of policy, it is impossible to say truthfully that this rollback disenfranchised any voter or rigged any election.

I, for one, refuse to accept the proposition that we live in a “post-truth” world, at least when it comes to talking about voting. Democracy depends on the ability of reasonable people to accept a certain set of shared premises about their electoral procedures, even if they sincerely disagree about matters of social policy, like taxes and economic regulations. Some of these essential shared premises are empirical facts about how elections actually work.

Another element of these essential shared premises is a willingness to use language carefully and precisely when talking about voting procedures. Don’t say “disenfranchisement” when you are talking about voting laws that may make voting less convenient than previously, but which do not actually prevent anyone from participating in an election. Don’t say that voters were “turned away” from the polls “for not having the required forms of ID” when, in fact, that is not what actually happened.

The overheated rhetoric of the “voting wars” has transgressed the realm of exaggeration and reached the danger zone of willful dishonesty. This must stop.

Whoever utters demonstrable falsehoods about the voting process should be condemned for doing so, whether these falsehoods come from one side or the other.

When Should a Voter’s “Clerical Error” Invalidate a Ballot?

by

Not when the state already has enough information to verify the ballot’s validity.

Roland Gilbert accidently wrote the current date, instead of his birthdate, when filling out the form on the envelope for submitting his absentee ballot in Ohio’s 2014 general election (which included a gubernatorial race). It’s a mistake that all, or at least most of us, have made at one time or another in our lives when filling out forms. Is it a mistake that should disqualify Roland Gilbert’s absentee ballot from being counted?

As a policy matter, I certainly think not. Moreover, this policy position recently has been adopted by the American Law Institute, a prominent nonpartisan organization most famous for its Model Penal Code, Uniform Commercial Code, Restatements of Law covering a wide variety of fields (like contracts, torts, and property law), and other law-improvement projects. In its new Principles of Law project concerning Election Administration, the ALI takes the position that an absentee ballot should not be invalidated if the identity of the absentee voter can be verified and the voter is registered and eligible to cast the ballot. (Full disclosure: together with my Election Law @ Moritz colleague Steve Huefner, I serve as Reporter to the ALI project that developed this and related principles.)

With respect to Roland Gilbert, there was no dispute that he was a registered and eligible voter. Nor was there any dispute that the absentee ballot in question had been cast by him. As required by a separate provision of Ohio law, he had supplied identification information for his absentee ballot (driver’s license number, or SSN, or the like), and this information sufficed to verify his identity—and his ballot’s authenticity. Yet his ballot was rejected—and thus Roland Gilbert disenfranchised—apparently for the sole reason that he mistakenly wrote the current date instead of his birthdate on his absentee ballot envelope.

This disenfranchisement seems wrong and undemocratic, disturbingly so. But is it unconstitutional? That legal issue is at the heart of the pending case, NEOCH v. Husted, in which federal district court Judge Algenon Marbley recently ruled that this disenfranchisement does violate the equal protection clause of the federal Constitution. Ohio’s Secretary of State Jon Husted has since appealed the case to the Sixth Circuit federal appellate court.

The Applicable Constitutional Analysis

To determine whether a state violates equal protection in the administration of its voting laws, the U.S. Supreme Court has devised what has come to be called the “Anderson-Burdick” balancing test. As I discussed in a recent essay on another pending federal court case involving Ohio’s voting laws, the one involving the state’s repeal of so-called “Golden Week” (when eligible citizens could both register to vote and cast their ballots during the same trip to an early voting location), the first task under the Anderson-Burdick balancing test is to identify the allegedly objectionable way in which the state is differentiating between two groups of voters. In the Golden Week case itself, this seemed impossible to do, since even after the repeal of Golden Week, Ohio was offering the same amount of voting opportunities to all eligible voters: 23 days of early voting, no-excuse absentee voting, and traditional Election Day voting. As I explained using an analogy to government-provided Shakespeare in the Park, the fact that in previous years the government used to provide five weeks of summer theater whereas now the government provides only four weeks does not cause a differentiation among would-be theater-goers that would raise an equal protection question. As long as the currently available four weeks are equally available to any person who would like to attend, there can be no equal protection objection to the shorter length of this government-provided benefit.

By contrast, in the NEOCH case, there is no doubt that Ohio is differentiating between two groups of voters: (a) those who correctly complete their birthdate on their absentee ballot envelopes and (b) those who do not. Thus, it is entirely appropriate to scrutinize this differential treatment of voters under the Anderson-Burdick balancing test. Although the justices of the U.S. Supreme Court have not been entirely clear—or in agreement among themselves—on exactly how Anderson-Burdick balancing is supposed to work, it essentially involves a comparative weighing of (i) the burdens imposed on the voters who suffer the differentially adverse treatment and (ii) the state’s justifications for imposing this adverse treatment on the negatively affected group of voters.

The most directly relevant precedent from the U.S. Supreme Court is Crawford v. Marion County Election Board, where the justices applied Anderson-Burdick balancing to Indiana’s voter ID law. There, the nine justices divided into three groups, of three justices each. The three most liberal justices on the Court at the time (Souter, Ginsburg, and Breyer) would have invalidated the Indiana law in its entirety as being unduly burdensome in relation to its potential benefits. The three most conservative justices (Scalia, Thomas, and Alito) would have upheld the law in its entirety as being minimally burdensome on the state’s voters considered as a whole. The outcome of the case was controlled by the three justices in the middle (Stevens, Kennedy, and Chief Justice Roberts), who refused to invalidate Indiana’s ID requirement completely, but explicitly left open the possibility that the ID requirement would be invalid as applied to those voters for whom it actually operated as a barrier to participation in an election.

The problem in the case, as these three centrist justices carefully explained, was that the plaintiffs were attempting to obtain an across-the-board injunction, which would have prevented Indiana from enforcing its ID rule even with respect to voters who already possessed the required ID or easily could obtain one (and thus for whom the rule imposed no consequential burden). Moreover, the plaintiffs had failed to identify any individual Indiana voter who actually would be disenfranchised as a consequence of the ID rule’s enforcement.   Thus, the centrists concluded: “Finally we note that petitioners have not demonstrated that the proper remedy—even assuming an unjustified burden on some voters—would be to invalidate the entire statute.”

The reasoning of the centrists in Crawford is instructive for the pending NEOCH case. It is plainly acceptable for Ohio to ask absentee voters to supply their birthdate, as Judge Marbley’s opinion acknowledges. Having this information can help confirm a voter’s eligibility and identity, thereby increasing the government’s confidence concerning a ballot’s validity, and asking a voter to supply it imposes only a trivial burden. (To be sure, as Judge Marbley also observed, there are low-literacy or otherwise impaired voters who cannot fill out an absentee ballot on their own, but under Ohio and federal law they are entitled to assistance in completing their absentee ballot envelope, and thus asking them to include their birthdate is not unduly burdensome even as to these voters.)

But invalidating a ballot, and thus disenfranchising the voter, because of an inadvertent error—like the one Roland Gilbert made, accidently putting the current date, instead of his birthdate—is an entirely other matter. Here, in contrast to Crawford, there is proof of a specific individual voter who suffers a severe burden: disenfranchisement itself, the loss of the right to participate on equal terms with other eligible voters in the election. And what is the state’s justification for this disenfranchisement? In Roland Gilbert’s case, and others like it, there isn’t any. Indeed, it is the existence and enforceability of a voter ID requirement that makes this so.

Ohio is able to ascertain Roland Gilbert’s identity and eligibility because of the ID information that is also required on the absentee ballot envelope. There is no dispute about the accuracy or validity of this information. In this circumstance, confirmation of Ronald Gilbert’s birthdate on his absentee ballot envelope is superfluous, and voiding his ballot because of an inadvertent error concerning this unnecessary extra information serves no valid government purpose. Only in a Kafkaesque bureaucracy, where government functionaries take pleasure in declaring “Gotcha” when they trip up citizens in the enforcement of their administrative regulations, would invalidation of a ballot solely for this kind of innocent mistake seem reasonable.

Thus, under Anderson-Burdick balancing it seems entirely appropriate to order—as Judge Marbley did—that Ohio not disqualify any absentee ballot because of the absentee voter’s failure to provide the voter’s correct birthdate when other available identification information provided on the absentee ballot envelope suffices to establish the voter’s eligibility and the ballot’s authenticity. Were I the judge in the case, I might have tried to frame the injunction against the state even more explicitly than Judge Marbley did to make clear that its scope was limited to those voters for whom an accurate birthdate was indeed superfluous. It is possible to imagine circumstances in which an accurate birthdate would be necessary to authenticate a ballot. Suppose, for example, that two individuals named “John Smith” reside at the same address, perhaps a father and son. They might have indistinguishable signatures, and the identification information provided with the returned absentee ballot—a utility bill, for example—might be insufficient to determine which John Smith was the one who cast the returned ballot. (Maybe one John Smith at that address already has cast another ballot, and it is important not to permit any individual to cast a second ballot in the same election.) In this instance, an accurate birthdate on the returned absentee ballot envelope might enable the government to determine that it was the father, not the son, who cast the ballot (and if it had been the son, not the father, who was the one who already cast the previous ballot, then this ballot now identified as from the father would be valid and eligible to be counted).

As I read Judge Marbley’s order in light of the rest of his opinion, it seems clear enough that it would not bar disqualification of an absentee ballot for lack of an accurate birthdate in those instances, like this hypothetical example, where the birthdate is indeed necessary to verify the ballot’s validity. Instead, Judge Marbley’s order essentially is limited to those other instances, like Roland Gilbert’s, where an accurate birthdate is unnecessary to validate the ballot. So limited, the order permits Ohio both to collect birthdate information in the first place and to rely upon this information in those situations when it is necessary to verify a voter’s identity.

In this respect, Judge Marbley’s order is consistent with post-Crawford developments concerning the constitutionality of voter ID laws. For example, in litigation over Wisconsin’s version of a voter ID requirement, the U.S. Court of Appeals for Seventh Circuit (like the Supreme Court in Crawford itself) refused to invalidate the ID requirement in its entirety. But in a subsequent opinion by Judge Easterbrook, the Seventh Circuit signaled that Wisconsin’s ID law would be invalid as to individual voters who, despite good faith efforts, were unable to obtain the required ID and therefore would be disenfranchised:

Instead of saying that inconvenience for some voters means that no one needs photo ID, plaintiffs contend that high hurdles for some persons eligible to vote entitle those particular persons to relief. Plaintiffs’ approach is potentially sound if even a single person eligible to vote is unable to get acceptable photo ID with reasonable effort. The right to vote is personal and is not defeated by the fact that 99% of other people can secure the necessary credentials easily. Plaintiffs now accept the propriety of requiring photo ID from persons who already have or can get it with reasonable effort, while endeavoring to protect the voting rights of those who encounter high hurdles. [Page 4; emphasis added.]

By the same logic, Judge Marbley’s order protects Roland Gilbert and other individual voters just like him, for whom Ohio’s birthdate requirement imposed an inappropriately disenfranchising burden, even as it imposes no such burden on most other voters.

Judge Marbley’s order is also consistent with a previous appellate decision in an earlier round of the same NEOCH litigation (696 F.3d 580). Back in 2012, the U.S. Court of Appeals for the Sixth Circuit considered the requirement in Ohio law that voters, when casting a provisional ballot, accurately supply on their provisional ballot envelope their name, their signature, and one of several acceptable pieces of identification information (driver’s license number, SSN, and so forth).   In the context of a preliminary injunction proceeding, when consideration of the relevant constitutional claim was tentative and not definitive, the Sixth Circuit surmised that demanding that provisional voters accurately supply these three pieces of information would not be unduly burdensome under Anderson-Burdick balancing. But that pronouncement is easily distinguishable: those three pieces of information are in most instances necessary to verify a provisional voter’s eligibility. Without the voter’s name, election officials simply do not know who the voter is. Without some sort of identification information (like a driver’s license or Social Security number), election officials ordinarily cannot confirm the authenticity of the voter’s identify. And unless a provisional voter has a religious objection to signing the provisional ballot envelope (in which case the signature requirement is waived upon an alternative demonstration’s of the voter’s authenticity), the absence of a signature is an indication of the voter’s unwillingness to attest to the voter’s own eligibility. Thus, none of these three pieces of information are superfluous in the way that an accurate birthdate often can be. Invalidating a provisional ballot for failure to accurately supply any of those three pieces of information would be justifiable in a way that invalidating an absentee (or provisional) ballot for failure to accurately supply a superfluous birthdate amounts to nothing more than the gratuitous disenfranchisement of an eligible voter who has cast a ballot in good faith.

Provisional ballots and address information

There is more to the pending NEOCH case than just the disqualification of absentee ballots because of inadvertent mistakes in supplying one’s birthdate. Like the earlier round of litigation in NEOCH, this round also involves provisional ballots as well as absentee ballots. Under current Ohio law, provisional ballots along with absentee ballots are being disqualified when a provisional voter inadvertently writes the current date instead of the voter’s birthdate on the provisional ballot envelope. According to the same Anderson-Burdick balancing, the same conclusion should apply: the federal judiciary should prohibit Ohio from disqualifying that provisional ballot when the provisional voter accurately supplied other identification information sufficient to verify the voter’s eligibility.

The pending NEOCH case also involves the disqualification of ballots, both absentee and provisional, for inaccurate address information. For example, when filling out an absentee or provisional ballot envelope, a voter mistakenly might transpose digits of the voter’s zip code, writing 42309 accidently instead of 43209. Obviously, a voter’s address is important for determining a voter’s eligibility. In our residence-based system of elections, where which legislative representative one votes for (as well as which municipal offices) depends on where a voter lives, election officials must be able to determine where a voter lives in order to verify the eligibility of a voter’s ballot. Even so, it does not follow that the government should be entitled to disqualify a voter’s absentee or provisional ballot automatically just because the voter made a clerical error concerning the voter’s address when completing the envelope in which the ballot is submitted.

In this context, it is important to remember that the pending litigation concerns voters who already have registered at least 30 days in advance of Election Day, as they are required to do under Ohio law, and when registering have supplied their address. Neither an absentee nor a provisional ballot can count if the voter is not previously registered. Invalidating an absentee or provisional ballot because the voter was not registered, or properly registered, is entirely appropriate and constitutional. Thus, invalidating an absentee or provisional ballot just because of an inadvertent clerical mistake on the absentee or provisional ballot envelope itself is, by definition, disenfranchising a voter whose proper registration status is not in dispute.

For all properly registered voters, the government already has on file the voter’s address. Thus, when examining an absentee or provisional ballot envelope, the only relevant question is whether the government has adequate information to link the ballot to a properly registered voter who purportedly cast it. If the envelope contains the name and signature of a properly registered voter, as well as a key piece of identification information (like a driver’s license or Social Security number), and if those three pieces of information match what is on file in the government’s voter registration database, then the government has established a positive link between the ballot and properly registered voter who cast it. If there is no reason to believe that an inadvertent error concerning completing the voter’s address on the envelope undermines that positive link, then there is no reason to disqualify the properly registered voter’s ballot just because of that inadvertent error.

This would be the case, for example, when a voter has made a simple transposition of digits in the voter’s zip code or made some other minor error, like writing “123 Park Avenue” instead of “123 Park Place”. If the rest of the address is correct—and the voter’s name, signature, and identification information all match what’s in the government’s registration database for that voter—then election officials have no reason to doubt the validity of the voter or the authenticity of the ballot. In this situation, it is obvious that a routine and innocent clerical error has occurred. Under Anderson-Burdick balancing, as explained by both the Crawford centrists and Judge Easterbrook, this particular voter should not be disenfranchised just because of the clerical error. When the government knows it has an authentic ballot from a valid voter, the government should not be disqualifying the ballot according to some “Gotcha” theory of clerical perfection. The right to vote on equal terms with other eligible citizens is too important for that kind of bureaucratic mentality.

To be sure, there may be instances in which inaccurate information concerning a voter’s address on an absentee or provisional ballot envelope might undermine the government’s confidence that it has an authentic ballot from a valid voter.  (The government legitimately asks voters to include their address on their absentee or provisional ballot envelope, even though the government already has a registered voter’s address on file, as part of its overall efforts to increase the accuracy of its ballot authentication processes, just as it does by asking voters to include their birthdates. Even if this extra information is not always necessary, sometimes it is useful, and there is nothing wrong in asking for it. Ohio also uses addresses on provisional ballot envelopes to create new valid registrations for provisional voters whose previous registration status the government cannot verify, thereby enabling these voters to cast countable ballots in future elections.)

If the address on the envelope looks nothing like the address on file in the voter registration database, then the government might have reason to doubt the ballot’s eligibility. In that situation, the government might be justified in calling upon the voter to clarify the discrepancy before the ballot is definitively entitled to be counted. But in those situations in which it is readily apparent that a discrepancy concerning the properly registered voter’s address is nothing more than a routine, innocent clerical error—which in no way undermines the government’s confidence that the ballot in question was cast by a voter entitled under state law to cast that very ballot—then the government is not justified to require anything further from the voter before proceeding to count that valid vote. To demand anything more from the voter in this instance would be an inherently unjustified burden, and thus one that necessarily flunks Anderson-Burdick balancing.

Simply put, in this situation, the government already has enough information to validate the ballot. To make the voter do anything more is overkill. For this reason, the voter should not be required to correct the obvious clerical error in order for the ballot to count. Accordingly, for this set of circumstances, it is appropriate for the federal courts to require (as Judge Marbley had done) that the state count the properly registered voter’s authentic ballot.

(Under Ohio law, a voter’s change of address may sometimes present a confusing situation for both the voter and the poll worker. If a previously registered voter has moved residences within the same county, but has failed to update the voter’s registration with the new address—as the voter is now able to do online—then the voter is supposed to cast a provisional ballot at the precinct appropriate for the voter’s new address, and that provisional ballot should count once the voter’s new address is verified. If the voter accidently puts the voter’s old address on the provisional ballot envelope, instead of the new address, there will be a mismatch between what the voter should have supplied and what the voter did supply.   If election officials can figure out what happened without additional information, and thereby determine that the voter actually cast a ballot in the correct precinct for the voter’s new address, then the officials should go ahead and count the ballot as properly cast. But this may be one of those instances in which election officials need additional information from a voter in order to corroborate the change of address, in which case the government would be entitled to disqualify the ballot if the additional corroborating information was not forthcoming.)

Other issues in the NEOCH case

When an absentee or provisional voter is legitimately required to supply additional information in order to validate the voter’s ballot, there is the question of how much time the voter should have to supply this information. In Ohio, voters previously were permitted ten days for this purpose. Now they have only seven. One of the issues in the pending NEOCH case is whether this shortening of the so-called “cure” period violates either equal protection or the Voting Rights Act.

This particular issue, it seems to me, is more like the curtailment of early voting that caused the elimination of Golden Week. Whether it is good or bad policy to cut the cure period from ten to seven days, it is not obvious to me that this cutback triggers Anderson-Burdick balancing. All voters now receive seven days. Where is the differential treatment among different groups of voters? That some voters would like, or would especially benefit from having, three extra days means that the shorter period imposes a “disparate impact” but under equal protection law that does not mean that the law subjects these voters to differential treatment—and it is only differential treatment, not disparate impact, that triggers equal protection scrutiny (whether under Anderson-Burdick, rational basis review, or any other level of judicial review). To invoke another basic analogy, suppose the government used to provide free transit for anybody on municipal buses from July 1 to July 10, but now provides this free transit only from July 1 to July 7. Yes, it is a cutback, but as long as the current free service is available to anyone who wants to ride the bus, it does not seem to be a differential treatment among would-be riders, even if there are some citizens who would prefer to ride the free bus during the extra three days no longer available.

But however the cutback of the cure period should be analyzed under Anderson-Burdick balancing, and however some other ancillary issues in the pending NEOCH case should be handled, it should be clear from the discussion above that that there are serious constitutional questions at stake. At issue is the potential disenfranchisement of properly registered voters who have cast their ballots, in circumstances where the government has no reason to doubt these voters or their ballots.   As Judge Easterbrook has observed, even if the voters and ballots that fall into this particular category are a tiny faction of all the voters and ballots in an election, the right to vote belongs to each eligible individual, who is entitled to protection from improper disenfranchisement. The federal courts are capable of tailoring injunctions carefully, so that a state law may be enforced in those many circumstances—indeed most of them—where it causes no improper disenfranchising effect, while at the same time the federal court’s injunctive relief safeguards those relatively few voters who need protection from disenfranchisement caused by the unnecessarily excessive enforcement of an otherwise unobjectionable state law.

When all is said and done, the upshot of the pending NEOCH case should be clear: Roland Gilbert and other voters just like him should be protected from being disenfranchised solely because they made a simple and understandable error when filling out their birthdate or address on their absentee or provisional ballot envelope.

Judge Easterbrook on the Voting Rights Act: Asking Good Questions, Making Bad Law

by

By Christopher Elmendorf

Professor of Law, UC Davis School of Law

Judge Easterbrook’s opinion upholding Wisconsin’s voter ID requirement asks the right questions.

Earlier this week Rick Hasen blasted Judge Easterbrook’s opinion upholding Wisconsin’s voter ID requirement as cavalier with the facts and “heartless and dismissive” in tone. But in one respect the opinion is extremely helpful: it asks the right questions.

Three questions foregrounded by Easterbrook are particularly important to the future of the Voting Rights Act:

  • What limiting principle keeps the Section 2 “results test” from obligating every state to tinker with its election machinery until rates of voter participation by race have been equalized?
  • Insofar as Section 2 conditions state obligations on past or present societal discrimination, how does this square with the 14th and 15th Amendments, which by their terms reach only state action?
  • To what extent are the “fact” questions in Section 2 cases questions that district judges should try to answer on the basis of expert testimony, as opposed to questions of belief, faith, or policy that ought to be settled by appellate courts as a matter of law?

As this post will explain, Judge Easterbrook’s answers to these questions are not convincing. But unless proponents of robust voting rights protections come forth with better answers—answers that a conservative judge can appreciate—Easterbrook’s opinion is likely to prove a harbinger of things to come at the Supreme Court.

  1. Does the “results test” of Section 2 obligate states to adjust their election rules so as to equalize rates of voter participation across racial groups?

For Judge Easterbrook this question is rhetorical. An affirmative answer would be “implausible,” he says. Perhaps more to the point, but unstated: an affirmative answer would make Section 2 a powerful voting wars weapon for the Democratic Party. This much conservative jurists are likely to resist.

But what limiting principle would permit a court to strike down Wisconsin’s voter ID law, or Ohio’s or North Carolina’s recent rollbacks of early voting and same-day registration, without also compelling states to adopt every feasible election administration reform that would tend to equalize political participation across racial groups?

Ruling in the Ohio and North Carolina cases, liberal panels of the Sixth and Fourth Circuits suggested that the reach of Section 2 depends on “social and historical conditions.” The argument, in a nutshell, is that blacks and Latinos faced substantial disparate-treatment discrimination in the past; that some societal discrimination continues in the present day; that the lower socioeconomic status of blacks and Latinos is due in part to such discrimination; and that Section 2 therefore obligates the states to remove any barrier to voting which in practice limits participation by low-SES voters relative to high-SES voters. This argument, though plausible as a matter of statutory construction, is less a limiting principle than a principled basis for no limits. There is a history of discrimination against racial minorities almost everywhere.

Judge Easterbrook offered a different and severely limiting principle: Racial inequalities in rates of voter participation violate the results test only if the state has made registration and voting very difficult for a class of citizens, defined by race or color, that is under-represented in the voter turnout statistics.

Easterbrook’s rule is dubious as a matter of statutory construction and ironic to boot. His rule would make the Section 2 redundant with the equal protection clause in election administration cases. Yet when Congress enacted the Section 2 results test, Congress was responding to a Supreme Court decision that read equal protection too narrowly. The whole point of the results test was to sweep more broadly, to provide a cause of action where minorities have formally equal opportunities to participate but too little “voting strength” in practice. It is also clear that the enacting Congress saw racial disparities in voter participation as a Section 2 problem, at least where there’s a pattern of official and societal discrimination.

So if Easterbrook’s limit is bad, and that of the Fourth and Sixth Circuits illusory (though principled), what else might do the job? One possibility is to treat Section 2 as reaching even minor voting barriers with a racially disparate impact, insofar as plaintiffs show that it is at least “significantly likely”—a relaxed evidentiary standard—that the voting requirements at issue (1) were enacted to diminish the number of votes cast by the racial minority, or (2) enable or encourage disparate-treatment discrimination by front-line election administrators, such a poll workers. Section 2 could also reach voting requirements that give electoral effect—via impacts on voter participation—to (3) disparate-treatment discrimination by state actors outside of the electoral realm (e.g. school or prison administrators); or even (4) discrimination in the larger society. However, if the significant likelihood test is to serve as a limit in practice, plaintiffs whose claims are predicated on societal discrimination should probably be expected to differentiate conditions in the defendant jurisdiction from typical conditions elsewhere.

I am not sure how the Wisconsin voter ID case would come out under this approach, but North Carolina’s rollback of same-day registration would clearly be vulnerable. In the latter case, the district court indicated that the question of discriminatory intent under the equal protection clause was close. A factual showing that almost establishes discriminatory intent under the equal protection clause should suffice to establish a “significant likelihood” of subjective discrimination for purposes of Section 2.

  1. Whither state action? 

Objecting to “social and historical conditions” arguments, Judge Easterbrook wrote that Section 2 cannot be treated as a remedy for societal discrimination because societal discrimination does not violate the Constitution. That societal discrimination is beyond the Constitution’s ken (for want of “state action”) is very well established as a general matter. But might voting be different?

Certainly there are some hints that it is. As Ellen Katz has shown, the Waite Court, which rolled back the First Reconstruction in the late 19th century, pointedly sidestepped the question of whether Congress could use its enforcement power under the 15th Amendment to remedy societal discrimination that hinders minority political participation.

In the 1950s, the Supreme Court found state action in racially exclusionary straw poll conducted by a private political club. In the 1970s, the Supreme Court emphasized societal discrimination as it developed racial vote-dilution doctrine under the equal protection clause.

The Court later abandoned the racial vote dilution jurisprudence of the 1970s, but Congress in the 1982 amendments to the VRA provided a statutory remedy for the same injury. And, importantly, conservative judges in subsequent vote dilution cases have argued that there is no constitutional problem with the 1982 amendments so long as plaintiffs must show that white voters discriminate against minority candidates on the basis of the candidates’ race. Seemingly private discrimination by voters stands in for state action. Elsewhere I have argued that this apparent exception to the state-action requirement is no exception at all, once one recognizes that to put in office persons who wield the coercive power of the state is to perform a “public function.”

“Social and historical conditions” may also be relevant to a Section 2 claim not because societal discrimination can violate the 14th or 15th Amendments when it touches the right to vote, but because societal discrimination can create political incentives for elected officials to restrict voting by members of a racial group. If this argument is correct, it answers a question that has perplexed Easterbrook and other judges: Is racially polarized voting germane to “vote denial” claims under Section 2? Many courts and commenters have deemed it irrelevant. But if societal conditions matter for Section 2 because of the incentives they create for legislators and their agents, then racially polarized voting is centrally important to vote denial as well as vote dilution claims.

  1. Distinguishing “adjudicative” from “legislative” facts?

A few years ago, the Supreme Court upheld Indiana’s photo-ID requirement for voting on the theory that it promotes “public confidence” in the electoral process. Plaintiffs in the Wisconsin voter-ID case said this argument is now a loser because subsequent empirical research found no relationship between voter ID laws and citizens’ confidence that their ballots will be counted. Judge Easterbrook held that the new findings are irrelevant because the relationship between voter ID and public confidence is a “legislative fact,” rather than a fact to be determined on the record in a particular case.

The idea here is that some putative facts are essentially just matters of belief. These “facts” must be settled as a matter of law, or else the law itself will consist of nothing more than district judges applying their personal and quite divergent prior beliefs.

I think Easterbrook was wrong that voter ID / public confidence is such a fact. Questions about causation can in principle be answered with statistics when the question concerns some discrete intervention in the world, and there is a way to estimate counterfactual outcomes—e.g., voter confidence in the absence of the ID requirement.

However, many facts of central importance to Section 2 are quintessentially legislative. Here’s an example: to what extent are current socioeconomic disparities among racial groups due to disparate-treatment discrimination? The Democratic district judge in the Wisconsin voter ID case credited an expert who said race discrimination is mostly to blame. In another case, a Republican judge might rely on conservative scholars who maintain that socio-economic disparities are largely due to differences in human capital, which in turn (some conservatives say) are substantially unrelated to race discrimination.

If we’re going to be honest, we must admit that the accounting-for-socioeconomic-disparities question is one that statistics—and hence experts—probably cannot answer. I see no way to estimate the relevant counterfactual, i.e., the pattern of socioeconomic outcomes that would be observed today if the United States had had no history of race discrimination. Furthermore, judicial answers to the accounting-for-socioeconomic-disparities question implicitly depend on normative judgments about the relevance of discrimination that took places generations ago. Justice Roberts’s famous pronouncement, “Things have changed in the South,” is as much as normative as factual.

This is not to say that all questions about race discrimination and its consequences are unanswerable. Experts can make across-state and over-time comparisons of racial attitudes, for example, and of the “treatment effect” of race in experimental settings.

One could also argue that Section 2 delegates authority to district courts to find legislative facts, rather than reserving these questions to the courts of appeal. But however one comes down on this question, there can be little doubt that voting rights law would be much improved if judges paid a little more attention what can and cannot be learned from statistics. By asking courts to distinguish legislative from adjudicative facts, Judge Easterbrook has taken a small but potentially constructive step in this direction—even if the line he drew was misplaced.