A Tale of Two Swing States

Pennsylvania should prepare itself for the kind of “voting wars” litigation that Ohio has experienced in recent years, including a claim that the state’s current opportunities to cast a ballot impose a disproportionate burden on the state’s African-American voters.

Since 2004, Ohio has been at the epicenter of the so-called “voting wars” because of the pivotal role the state played in George W. Bush’s reelection victory over John Kerry. Given the possibility that a presidential election might again hinge upon which candidate prevails in the state, over the last dozen years both Democrats and Republicans have pursued litigation in an effort to adjust Ohio’s voting rules in ways that the parties perceive to be favorable to their candidates. On the topic of provisional voting alone, there has been at least one major lawsuit actively litigated in every federal election year since it became clear that a close presidential election—like Bush-Kerry itself—might turn on which of Ohio’s disputable provisional ballots end up being counted.

Pennsylvania, by contrast, has largely escaped the intensity of electoral litigation to which Ohio has been subjected. To be sure, when Pennsylvania adopted a new voter identification law in 2012, it provoked lawsuits and, indeed, was invalidated by the state’s judiciary. Also, fears about the reliability of Pennsylvania’s voting machines in 2008 prompted the NAACP to seek, and a federal court to grant, an emergency injunction requiring the state to supply precincts with backup paper ballots. Still, for the most part, Pennsylvania’s voting procedures have not been subjected to nearly the same degree of judicial scrutiny as have Ohio’s during the past decade.

Now we have a report from Dave Wasserman at the 538 website that Pennsylvania is most likely to be the state that tips the Electoral College for the winning candidate. I’m sure Ohio would be happy to share the spotlight that comes from having this “swingiest” of swing-state status. But is Pennsylvania ready for it? One advantage of all the litigation that Ohio has experienced over the last decade is that its electoral system has been tested. Sure, there are new problems that could emerge in Ohio this November, but many of the ambiguities concerning how to count certain categories of ballots in particular circumstances have been clarified through judicial resolution. Because Pennsylvania has not been put through the same type of clarifying process, there potentially will be more available grounds for disputation if the presidential election there is close enough and, as 538 warns, the White House hangs in the balance.

Are Currently Available Voting Procedures Discriminatory?

One issue that is presently under litigation in Ohio, and which Pennsylvania has not yet faced, is whether the state’s current rules concerning the opportunities available for registering to vote and casting a ballot either discriminate on the basis of race or otherwise deny citizens the equal protection of the laws. In Ohio, the current law is that voters must register 30 days in advance of Election Day and, having done so, then may cast a ballot in one of three ways: first, under a regime of “no excuse” absentee voting, all registered voters may choose to cast an absentee ballot and deliver it by mail (or, if they prefer, drop it off at their local board of elections); second, under a regime of early in-person voting that begins 27 days before Election Day (on October 12 in 2016), all registered voters may choose to go to a designated early voting center in their county on any of the 23 days in which this early voting is available; and third, under a regime of traditional neighborhood-based precinct voting, all registered voters may choose to cast their ballot on Election Day itself at the polling place for their local precinct. In Pennsylvania, current law also requires that voters register 30 days in advance, and all registered voters also may cast their ballots on Election Day at a traditional neighborhood polling place for their particular precinct, but Pennsylvania currently does not offer any in-person early voting, and Pennsylvania also restricts absentee voting to those voters who satisfy a limited list of justified reasons for being unable to vote at their local precinct on Election Day (like military service, disability, and business travel).

Ohio’s current law is being challenged as racially discriminatory and a violation of equal protection because it is less expansive in its provisions than a previous Ohio law, under which early voting started 35 days before Election Day and, as a result, provided a five-day period—dubbed “Golden Week”—in which it was possible to register to vote and cast an early in-person ballot at the same place at the same time.   In a recent ruling after a trial on the merits of these claims, however, the federal court found that Ohio’s change in its law was not motivated by racial discrimination. [See opinion at 115.] Therefore, under longstanding constitutional doctrine, which requires a racially discriminatory intent to prevail, it is impossible to claim that Ohio’s law is unconstitutional race discrimination. [See Washington v. Davis, 426 U.S. 229 (1976).] Laws that are racially discriminatory in effect, but not intent, may violate federal statutory laws, and we will consider whether Ohio’s voting rules have a racially discriminatory effect that violates the federal Voting Rights Act. In the absence of a racially discriminatory intent, however, the presence of a racially discriminatory effect does not violate either the Fourteenth or Fifteenth Amendment. [See Rogers v. Lodge, 458 U.S. 613 (1982) (applying the rule of Washington v. Davis to voting cases).]

Equal Protection Analysis

             Before turning to the Voting Rights Act, we should consider a separate equal protection challenge to Ohio’s current voting, one not based on a claim of race discrimination. Under jurisprudence dating back to the 1960s, the differential treatment of citizens with respect to the opportunity to vote calls for special judicial scrutiny given the fundamental importance of voting in a democracy. Thus, as the Supreme Court held in the canonical case of Harper v. Virginia Board of Elections, 383 U.S. 663 (1966), the imposition of a poll tax violates the Constitution’s equal protection clause insofar as it conditions the right to vote on having the financial resources to pay the tax. The Supreme Court applied this line of equal protection cases most recently in Crawford v. Marion County Election Board, 553 U.S. 181 (2008), which involved Indiana’s voter identification law.

The plaintiffs in the pending Ohio case, which include the Ohio Democratic Party in the lead, have invoked this line of cases—known to lawyers as Anderson-Burdick after two intermediate cases in the line—to argue that Ohio’s current rules for registration and voting violate the Constitution’s equal protection clause. But it is difficult to see how this could be so. Ohio’s current rules requiring registration 30 days in advance of Election Day and providing three alternative methods of casting a ballot (again, no-excuse absentee voting, 23 available days of in-person early voting starting on October 12, and traditional precinct-based Election Day voting) apply equally to all citizens.

The plaintiffs say that Ohio’s elimination of Golden Week imposes a burden on those citizens who in the future would like to take advantage of the opportunity to register and vote at the same place at the same time. Whether or not it is appropriate to characterize the elimination of this convenience as a burden, it is hard to understand it as differentiating among voters and thus denying equal protection to any. No new voters get to take advantage of Golden Week any more. All are required to register 30 days in advance and then separately decide which of the three ways they would like to cast their ballots.

It is the most rudimentary requirement of equal protection jurisprudence that in order for the government to be found to be violating equal protection the government must be treating some persons differently than others.   This was true in the poll tax case: those who paid the poll tax were permitted by Virginia to vote; those who did not pay the poll tax were disenfranchised. Analytically, a similar differentiation among citizens existed in the Indiana voter ID case: those with proper ID were treated by the law differently—more favorably—than those without the proper ID. But no such differentiation among citizens appears to exist in the pending litigation over Ohio’s current rules for casting a ballot. Again, the same three ways of casting a ballot apply equally to all. An equal protection challenge to these current rules would seem to fail in this most basic respect: in terms of its availability of early voting, Ohio is not presently engaging in any differential treatment at all among its would-be voters.

To be sure, there may be some potential new Ohio voters who would find Golden Week to be an attractive option.   Of course, any potential new voter would prefer having that option added to other available options, without the loss of those other alternatives. Who would not prefer more choices to fewer?

Yes, but some would-be new voters really would like to take advantage of the extra convenience of simultaneous registration and voting that Golden Week offered. They have an especially strong preference for that form of convenience. Doesn’t the absence of Golden Week deny them equal protection? No. In providing its rules concerning voter registration and the casting of ballots, Ohio law is treating them just the same as other would-be voters. It is providing equal opportunity to all (although obviously not equally as generous as if the options also included Golden Week).

Consider this analogy. Suppose each summer the government offers Shakespeare in the Park. If the government charged the state-subsidized price of $5 per ticket, that would be a government rule that differentiated among individuals in terms of access to this government benefit. That differential treatment would trigger judicial scrutiny under equal protection analysis, although it likely would easily pass muster insofar as access to Shakespeare (while desirable) is not fundamental to democracy. Likewise, if the government required college students to show a valid student ID in order to take advantage of free admission to Shakespeare in the Park for all college students. This ID requirement would differentiate among those having and those not having the ID necessary to obtain free admission.

Suppose that for the summer of 2016 the government decides to offer four weeks of Shakespeare in the Park, starting Monday, August 1. Suppose, however, that there are some individuals who really wish the government would offer an additional week—the last week in July. Maybe these individuals are out of town on vacation the entire month of August and therefore cannot take advantage of Shakespeare in the Park if it is only in August, without an extra week in July. Even so, the government’s decision to limit Shakespeare in the Park to just four weeks in August would not be a denial of equal protection to these individuals. Rather, everyone would be entitled to attend Shakespeare in the Park on equal terms during the four weeks in August that the government provides it.

This equal treatment would be the same even if in previous summers the government operated five weeks of Shakespeare in the Park, with the last week of July included along with four weeks in August. And it would also be true even if in those previous summers one reason why some individuals really liked to attend Shakespeare in the Park during that last week in July, rather than in August, was because it so happened that during that week ArtsFest was also taking place in the same park. Thus, previously arts lovers could take advantage of the opportunity to view the ArtsFest exhibits on the same trip to the park that they watched Shakespeare. Limiting Shakespeare in the Park to four weeks in August, as planned for the upcoming summer, makes this “doubleheader” no longer possible. Even so, limiting Shakespeare in the Park to four weeks in August is not a denial of equal protection to anyone. Everyone still is entitled to attend Shakespeare in the Park during the four weeks of August, and everyone now must make two separate trips to the park if they want to see both ArtsFest and Shakespeare. In limiting Shakespeare in the Park to four weeks rather than five for upcoming summers, the government has not engaged in any differential treatment of individuals that would trigger even minimal “rational basis” judicial review for purposes of the Constitution’s equal protection analysis.

Voting is not Shakespeare in the Park, but the same analytic point applies. At a minimum, there needs to be differential treatment of individuals in order to trigger judicial review under equal protection analysis.   But Ohio’s decision to limit in-person early voting to four weeks rather than five, thereby eliminating simultaneous registration and voting, is not a differential treatment of anyone.

In the pending lawsuit, the federal district judge failed to consider this basic point. Instead, the judge jumped right into judicial scrutiny under so-called Anderson-Burdick balancing. But Anderson-Burdick balancing is part of equal protection analysis, and thus it requires differential treatment of individuals in the first place. (As Rick Hasen already has observed, the federal judge in the Ohio case also oddly conflated Anderson-Burdick balancing with the question of whether Ohio’s currently law has the effect, without the intent, of disproportionately disadvantaging African-American voters; in doing so, the judge apparently contravened the aforementioned longstanding doctrine that racially discriminatory effects in the absence of racially discriminatory intent are not unconstitutional under either the Fourteenth or Fifteenth Amendments.)

To my knowledge, consistent with this basic equal protection point, the U.S. Supreme Court has never applied the Anderson-Burdick inquiry in any circumstance that did not involve differential treatment of individuals with respect to the operation of the electoral process. As already indicated, both Harper (poll tax) and Crawford (voter ID)—the first and latest word respectively from the Court in this line of cases—involved the requisite differential treatment of individuals with respect to the entitlement to cast a ballot. Anderson involved a filing deadline for candidates to get on the ballot, thereby differentiating among candidates who did and did not meet the deadline. Burdick involved a prohibition against write-in candidates, and thus could be viewed either as a differentiation among candidates or a differentiation among the voters who wanted to cast ballots for them. In limiting a voter’s choice among candidates, the state law in Burdick also can be viewed as implicating distinctive Free Speech interests that are not at play in Ohio’s rule concerning the number of weeks available for in-person early voting, or in the current lack of overlap between early voting and the registration deadline.

In two cases from the 1970s, before both Anderson and Burdick, the Supreme Court considered the constitutionality of laws that imposed early deadlines on the right to vote.   In Dunn v. Blumstein, 405 U.S. 330 (1972), the Court invalidated a Tennessee law that, although providing for a 30-day registration deadline (like Ohio), required anyone attempting to register to have lived in Tennessee for a year. Then, a year later, in Rockefeller v. Rosario, 410 U.S. 752 (1973), the Court sustained a New York statute that required voters to affiliate with a political party 30 days in advance of a general election in order to be able to vote in the first party primary following that general election. (The current version of this New York law drew a lot of criticism this year insofar as it required independents to declare a party affiliation many months before New York’s hotly contested presidential primary between Hillary Clinton and Bernie Sanders.) Both of these laws can been seen as differentiating between those individuals who satisfy the applicable deadline and those who do not, and thus both of these Supreme Court decisions are consistent with basic Equal Protection analysis. In Dunn, the Court explicitly made this point:

Durational residence laws . . . divide residents into two classes, old residents and new residents, and discriminate against the latter to the extent of totally denying them the opportunity to vote. The constitutional question presented is whether the Equal Protection Clause of the Fourteenth Amendment permits a State to discriminate in this way among its citizens.

405 U.S. at 334-35. In Rockefeller, moreover, the Court there observed that New York’s party-affiliation deadline

did not absolutely disenfranchise the class to which the petitioners belong—newly registered voters who were eligible to enroll in a party before the previous general election. Rather, the statute merely imposed a time deadline on their enrollment, which they had to meet in order to participate in the next primary.

410 U.S. at 757.

In this respect, the 30-day registration deadline in Ohio’s current law—considered by itself—can also be viewed as differentiating between those who meet and do not meet this deadline. But there is little doubt that this 30-day registration deadline would easily pass constitutional muster, given the Court’s precedents. See Marston v. Lewis, 410 U.S. 679 (1973) (rejecting an equal protection challenge to a 50-day registration deadline); Burson v. Fortson, 410 U.S. 686 (1973) (same).  In any event, it is not Ohio’s 30-day registration deadline about which plaintiffs complain in the current litigation; instead, it is Ohio’s current provision of early voting for a period of time that is not long enough to overlap with the time in which the registration window is still open. Yet the challenged aspect of Ohio’s current law, unlike the registration deadline itself, does not differentiate between anyone: it simply makes early voting available for the specified period of 23 days. One might try to say that this aspect of Ohio’s law differentiates between those who do and do not take advantage of this rather extensive period of early voting, but that would be like saying that the government’s decision to provide four weeks of Shakespeare in the Park differentiates between those who do and do not take advantage of this rather extensive theater opportunity.

To say that the government treats persons differently just by offering everyone an ample amount of Shakespeare would be to undermine a basic analytic precept of equal protection jurisprudence. Indeed, to provide something to everyone who shows up to receive it during the time when it is available would seem to satisfy the essence of equal treatment. (If one needs another example to confirm this fundamental point, imagine a municipal swimming pool at which anyone may swim during the time in which it is open between Memorial Day weekend and Labor Day weekend.   Is the city that invites everyone to swim during these available times really denying equal treatment to someone who would prefer to swim at the municipal pool before it opens on Memorial Day weekend?) Insofar as the pending Ohio case involves an equal protection challenge to the availability of early voting in the state, the same point would seem true. Thus, by jumping immediately to Anderson-Burdick, the federal district court in the pending Ohio case seems vulnerable to reversal on the basic ground that Anderson-Burdick does not apply since the case does not involve a differential treatment of would-be voters.

Voting Rights Act Analysis

To my mind, the Voting Rights Act issue in the pending Ohio case is more difficult to think through. That is because section two of the VRA, as revised by Congress in 1982, does not require the existence of differential treatment among voters on the basis of race, but instead imposes liability on a state whose uniform treatment of voters nevertheless results in black voters facing greater obstacles to casting a ballot than white voters. The exact text of section two, as amended, is as follows: “No voting qualification or prerequisite to voting or standard, practice, or procedure shall be imposed or applied by any State or political subdivision in a manner which results in a denial or abridgement of the right of any citizen of the United States to vote on account of race or color.” (Emphasis added.) It might have been possible to interpret the phrase “denial or abridgement” narrowly to encompass only outright governmental impediments to the act of casting a ballot. But that possibility has been superseded by decades of Supreme Court jurisprudence, subsequently accepted by Congress in its revisions of the VRA, which apply this “results in . . . abridgement” language to the dilution of African-American voting power as a consequence of the particular way in which legislative districts are drawn during decennial reapportionments. Those district lines are not impediments to casting a ballot; nonetheless they can cause African-American voters to suffer inferior electoral opportunities compared to white voters.

Think, too, of this example: a particular town is residentially segregated, and the town’s only polling place is on the white side of town very far from the neighborhood where black voters live. There is no doubt that the government’s location of this sole polling place close to white voters and far from black voters, even if not motivated by intentional race discrimination, would be a “practice” that “results in . . . abridgement” of the right to vote based on race—or at least a court could so conclude after considering “the totality of circumstances,” as required by additional explicit language in the statute. Indeed, even Justices Thomas and Scalia, the two members of the Supreme Court who adopted the most restrictive interpretation of this language in the VRA, explicitly acknowledged that it would apply to this example: “the section thus covers all manner of registration requirements, the practices surrounding registration (including the selection of times and places where registration takes place and the selection of registrars), the locations of polling places, the times polls are open, the use of paper ballots as opposed to voting machines, and other similar aspects of the voting process that might be manipulated to deny any citizen the right to cast a ballot and have it properly counted.” Holder v. Hall, 512 U.S. 874, 922 (1994) (Thomas, with Scalia, concurring in the judgment).

Note, too, that for the purpose of section two of the VRA, it matters not whether the location of the town’s sole polling place in a white neighborhood (and far removed from where black voters live) is a newly adopted “practice” or instead a longstanding one.   Either way, this practice violates VRA’s section two if the totality of the circumstances demonstrates that the polling place’s location “results in” black voters having more difficulty in casting a ballot than white voters (as would clearly seem to be the case from the rudimentary facts of this example). In this respect, section two of the VRA differs from the Act’s section five, which focuses on whether a change in a state’s voting practices causes a deterioration in the electoral opportunities of racial minorities compared to what those opportunities were prior to the change. Section five has been rendered inoperable by the Supreme Court’s decision in Shelby County v. Holder (2013), and in any event section five would not have applied to Ohio or Pennsylvania because neither state was among the “covered jurisdictions” governed by that particular provision of the VRA. Therefore, in considering whether the current voting practices in Ohio or Pennsylvania violate the VRA, only section two matters, and under section two what is relevant is not the state’s current practices compared to its previous ones, but instead whether the state’s current practices cause minority voters to face greater difficulty in casting a ballot than white voters. Section two’s focus on the present and future, rather than the past, however, means that a practice is not immunized from section two liability just because the practice is longstanding. “It’s been that way forever,” in other words, would be no defense in a section two lawsuit brought against a town with a single polling place in a white neighborhood far from where the town’s black voters live.

But the applicability of section two to “the location of polling places, the times polls are open, . . . and similar aspects of the voting process,” as Justices Thomas and Scalia put it, inevitably raises difficult line-drawing questions. Some states keep the polls open for twelve hours on Election Day, usually from 7 am to 7 pm. But other states are willing to keep the polls open for fourteen or even fifteen hours, from 6am to either 8 or 9 pm. Suppose it could be shown that in a state with only twelve hours of voting on Election Day black voters disproportionately face greater difficulties in getting to the polls than white voters—the principal reason being that black voters disproportionately have work or family constraints that prevent them from having the flexibility to get to the polls within the twelve-hour window—while having the extra two or three hours that other states provide would be an especially great benefit to minority voters. On the basis of this evidence, would the legal conclusion follow that a state with only twelve hours of voting on Election Day violates the VRA? Or would twelve hours be deemed enough, even if it were shown that African-American voters would especially benefit from having fourteen or fifteen hours? Congress presumably did not think it was requiring states to provide more than twelve hours of voting on Election Day when it amended the VRA in 1982 to incorporate the “results” test in section two, but it may be that regardless of congressional intent the 1982 amendment has that legal consequence.

What about the fact that Election Day is on Tuesday, not Saturday or Sunday? Suppose it also could be shown that holding Election Day on Tuesday, rather than Saturday or Sunday, disproportionately burdens black voters more than whites—because, again, blacks on the whole have less flexibility than whites in their weekly schedules. Would that mean that holding Election Day on Tuesday, rather than Saturday or Sunday, is a “practice” that “results in” abridging voting rights “on account of race,” as provided in VRA’s section two? It is hard to believe that Congress meant to make holding elections on Tuesdays vulnerable to a VRA challenge under the results test of section two, especially as Congress itself has set Tuesday as Election Day for federal elections. But it is also hard to identify a principled basis for protecting Tuesday-as-Election-Day from a finding of liability under the results test in VRA’s section two, at least if the evidence shows that holding elections on Tuesdays in fact has a disproportionately adverse impact on African-American voters.

These line-drawing questions seem very pertinent to evaluating the claim that Ohio’s current rules regarding voter registration and the casting of ballots violate VRA’s section two. The federal district court said that they do, because Ohio’s current rules entail an elimination of Golden Week, a practice that African-Americans disproportionately took advantage of when it existed. But recognizing the need to consider Ohio’s rules under section two rather than section five, the federal court also said that it was not applying the “retrogression” analysis applicable to section five claims. [See opinion at 97.] Yet it is difficult to square these two statements from the district court. If the court truly was not employing section five’s “retrogression” inquiry, then it should not matter that Ohio law previously contained Golden Week. What matters, instead, under section two is whether the current set of rules and practices “results in” African-American voters having less opportunity to cast a ballot than white voters.

Citing a since-vacated appellate opinion, the federal district judge (op. at 98) in the pending Ohio case asserted that the elimination of Golden Week was relevant to determining whether currently blacks have equal opportunities to vote as whites. But this assertion seems to be an analytic error. To see this point, consider again the analogy to the government’s provision of Shakespeare in the Park during four weeks in the summer. Suppose the question is whether blacks have equal opportunities to see Shakespeare during those four weeks as whites. For this purpose, the timing of this theater is the only relevant issue (not cost of the tickets, or the location of the particular park relative to where blacks and whites live, etc.) Suppose we have reason to think blacks on the whole are equally able to attend theater during the month of August as whites. In other words, there is no reason why blacks would face greater difficulty attending Shakespeare in the Park during these four weeks in August than whites would.  If this were the case, then we would say the government’s timing of its Shakespeare in the Park does not “result in” blacks having less opportunity to attend this theater than whites.

This conclusion would not change, it is important to recognize, just because blacks might have preferred more than whites would have for the government to offer its Shakespeare in the Park during the last week of July. And on this point it matters not whether the government did nor did not offer Shakespeare in the Park in July previously. (Suppose Ohio had not eliminated Golden Week, but blacks disproportionately would have utilized an additional second week of Golden Week; the fact that Ohio never provided that extra desirable week of simultaneous registration and voting would not cause the provision of one week of Golden Week to result in blacks having greater barriers to voting than whites during the single week of Golden Week.) Even if in previous summers the government provided a fifth week of Shakespeare in the Park during the last week of July, and even if blacks more than whites attended Shakespeare in the Park during this last week of July—and even if they did so because blacks more than whites enjoyed the possibility of seeing Shakespeare on the same trip to the park as viewing Artsfest—it does not follow that blacks have less opportunity than whites to attend Shakespeare in the Park in the upcoming summer when the government is now offering it only during four weeks in August. If it is still true that it is no more difficult for blacks than for whites to go to the park during these four weeks in August, then it does not become more difficult just because blacks more than whites found it preferable to see Shakespeare at the same time as Artsfest. Only on a retrogression analysis would the provision of four weeks of Shakespeare in the Park, rather than the previously available five weeks, become a discriminatory burden on black theater-goers. For the same reason, because VRA’s section 2 does not involve the retrogression analysis that belongs to section 5, the relevant inquiry is whether Ohio’s current provision of 23 days of in-person early voting makes it more difficult for blacks to vote than for whites to vote—not whether blacks more than whites would take advantage of an in-person early voting period that started even earlier.

With this analytical point now clarified, how does Ohio’s current law fare under section 2 analysis? According to US Census data, in 2014 voter turnout among Ohio’s black citizens (40.7%) was actually a little bit higher than among the state’s white citizens (40.2%, for “non-Hispanic whites” as the relevant comparison of majority-race voters). The 2014 election was held after Golden Week had been eliminated, and thus this data suggests that under Ohio’s current rules and practices for voter registration and the casting of ballots the state’s African-American citizens do not suffer disproportionate obstacles in the ability to vote compared to the state’s white citizens. In the previous sentence, the most important word is “suggests” because, as political scientists have observed, there are significant limitations to the US Census methodology that generates this data. Nonetheless, unless there is better data to rebut what the US Census numbers at least superficially suggest (and it is important to remember that plaintiffs bear the burden of proving wrongful “results” in a section two lawsuit), it presumptively appears that blacks citizens in Ohio have an equally effective ability to register and, using one of the state’s three available methods, to cast a vote as white citizens in Ohio. To be sure, African-American voters as a group might disfavor one of Ohio’s available methods of voting. (The district court found that blacks in Ohio tend to distrust absentee voting.) Likewise, African-American voters as a group might prefer a longer period of early voting, or might like to have the convenience of being able to simultaneously register and vote. But under the regime as it currently exists, Ohio’s black citizens do not appear to confront disproportionate obstacles to the casting of ballots that are not confronted by Ohio’s white citizens.

Yet if Ohio’s current regime violates section two of the VRA, as the federal district court ruled, then what of Pennsylvania’s current regime? The federal judge said that in evaluating Ohio’s regime under section two it is irrelevant what the voting rules and practices of other states may be. Not quite. Whatever legal principle leads to a conclusion that Ohio violates section two must also apply to all other states. After all, unlike section five of the VRA, section two governs all fifty states. Thus, if Ohio law violates section two, then we can ask whether according to the same interpretation of section two Pennsylvania law also violates section two?

Based on the conclusion of the federal judge in the Ohio case, Pennsylvania’s existing regime would seem to be extremely vulnerable. As we have seen, Pennsylvania provides much less opportunity to cast a ballot than Ohio: Pennsylvania lacks “no excuse” absentee voting and also lacks any in-person early voting. In other words, Pennsylvania voters without a valid reason for voting absentee are confined to casting a ballot on Election Day (and must have registered 30 days in advance to do so). The fact that this severely limited set of voting options is longstanding “practice” in Pennsylvania would be no defense to a section two challenge to that practice. Rather, the question would be whether this practice of severely limited options “results in” Pennsylvania’s African-American voters facing disproportionately greater difficulties in casting a ballot than the state’s white voters. In this regard, it is worth observing that the same US Census data for the 2014 election shows voter turnout for Pennsylvania’s black citizens to be only 37.1%, significantly lower than the 42.1% turnout rate for Pennsylvania’s white citizens.

Thus, it would seem that Pennsylvania should be even more vulnerable to a section two lawsuit than Ohio. If one resists this conclusion on the ground that Pennsylvania provides enough voting opportunities for both its black and white citizens—it keeps its polls open on Election Day for thirteen hours, from 7 am to 8 pm—then it becomes difficult to see how Ohio could be violating the same section two by providing so much more voting opportunities than Pennsylvania. If one then points out that Ohio used to provide even more voting opportunities than it currently does, including offering the much-desired Golden Week—then it seems like the analysis has converted to a “retrogression” inquiry despite protestations to the contrary. If blacks have no greater difficulty voting than whites do under current Ohio law, but blacks utilized the even more convenient aspects of Ohio’s previous regime at higher rates than whites did, a finding of liability because of this change in Ohio’s law would fit the theory of section five, not the jurisprudence of section two.

The Value of a Retrogression Inquiry

To say that the distinctive retrogression analysis properly belonging to section five should not be smuggled into section two “results” jurisprudence—despite the obvious temptation to do so after the neutering of section five in Shelby County—is not to deny the importance of an anti-backsliding principle in voting rights law. Indeed, one can see the argument of extending section five’s retrogression analysis nationwide, so that it covers northern states like Ohio as well as southern ones like North Carolina.   But as frustrating as it is to wait for Congress to fix the coverage formula applicable to section five in the aftermath of Shelby County, it is seems inappropriate as an exercise of statutory interpretation to convert section two, which was always intended to be distinct from section five, into the functional equivalent of a nationally applicable anti-backsliding rule. (In a current research project, I am pursuing whether the Due Process Clause of the Fourteenth Amendment, which in a wide variety of non-voting contexts is understood to prevent the government from improperly undermining legitimate settled expectations and reliance interests, might be interpreted as encompassing an anti-backsliding principle in electoral contexts and thus could do some of the work of the incapacitated section five.)

But as important as an anti-backsliding principle is to voting rights, it is also necessary to recognize that it comes with costs. If it becomes too difficult for a state to undo an experiment in electoral reform, states will be reluctant to experiment in the first place. States without any early voting will be advised by their attorneys that if they choose to adopt some early voting they may be stuck with whatever amount they initially provide, so perhaps they better think twice before adopting it.

There are legitimate reasons why a state might not want to offer more than two weeks of early voting. One of the presidential debates this year is scheduled for October 19. A state might think that it is much better if voters cast their ballots after they have seen all the debates, and that voters should not cast their ballots “too early” based on incomplete or outdated information compared to what is known on Election Day itself. A state that expanded its available days of early voting from zero to two weeks, but no more, would be seen as embracing a reasonable balance between making voting more convenient and having the election capture the electorate’s collective opinion at a particular moment in time.

(The American Law Institute, at its annual meeting on May 16 this year, approved a set of principles for early and absentee voting that recognized, among other relevant considerations, the “information gap” among voters that potentially increases as the period of early voting lengthens. This “information gap” was a factor in the ALI’s deliberations on what to recommend as a minimum amount of in-person early voting for those states choosing to establish an early voting regime. The ALI settled upon a recommendation that early voting begin no later than ten days before Election Day, rather than adopting a recommendation that early voting necessarily should begin even earlier in October than that, in recognition that some states legitimately might wish to avoid the “information gap” associated with a longer period of early voting. Full disclosure: together with my Moritz colleague Steve Huefner, I serve as reporter for this ALI project.)

If a state experimented with a month of early voting, but subsequently decided that two weeks was a better balance of the competing considerations, a state should not be unduly hindered from making this kind of adjustment to its election laws. To be sure, as section five required before Shelby County, if a state has a history of racial discrimination in voting and the state decreases its early voting period from four weeks to two, the state should be required to justify this cutback in nondiscriminatory terms, and the federal judiciary should make sure that any nondiscriminatory reason that the state asserts is not mere pretext. But in principle a state’s decision to hold two rather than four weeks of early voting can be understood in racially neutral terms, and no state that experiments with a longer period of early voting should be forever barred from deciding that a two-week period would be better.

In any event, that kind of point would be relevant to the application of a retrogression analysis. But for the reasons elaborated, retrogression analysis is not germane to the evaluation of Ohio’s change in its early voting rules. What matters, instead, is a proper application of the section two “results” analysis.

For an interesting comparison on how section five and section two apply differently to the same reduction in early voting, one can examine the two decisions concerning Florida’s changes in early voting in advance of the 2012 election. As a consequence of new legislation, Florida’s early-voting period shrank from a maximum of twelve days to a maximum of eight, although the legislation also provided other changes that newly guaranteed three weekend days of early voting, including one Sunday of early voting, and potentially retained the same number of total hours of early voting compressed into a fewer number of days. In Florida v. United States, a three-judge district court (which included Supreme Court nominee Judge Merrick Garland) ruled that under section five Florida had failed to demonstrate that its changes to early voting were not retrogressive. By contrast, in a separate lawsuit presenting a section two challenge to the same legislative changes in early voting, Brown v. Detzner, a federal district judge in Florida ruled that plaintiffs had failed to establish that the changes resulted in the state’s black voters having less opportunity to cast a ballot than the state’s white voters. It is hard to square the federal court decision in Brown v. Detzner, the Florida section two case, with the recent ruling in Ohio.

What Next?

As the Ohio case moves from the district to the circuit court, this much seems clear: the status of section two analysis is extremely fluid given the line-drawing difficulties involved. On appeal, the Ohio case could go either way. But if the district court is affirmed, meaning that current Ohio law does violate section two, then Pennsylvania should beware. For the Ohio ruling on appeal to be principled, it necessarily would mean that Pennsylvania is at significant risk of also violating section two. With Pennsylvania coming into the spotlight as the most likely Electoral College “tipping state,” litigation over Pennsylvania’s voting procedures presumably would follow quickly.