Supreme Court observers have begun to pay more attention to the Court’s “shadow docket”: disputes the Court resolves summarily, without the usual briefing, argument, explanations, or even indications how each Justice voted. Still mostly overlooked, however, is that the Court doesn’t just have a shadow docket; it also has shadow doctrines, rules the Court applies only in its non-merits cases. In my own field of election law, the most prominent of these shadow doctrines is undoubtedly the Purcell principle, named for the 2006 (non-merits) decision of Purcell v. Gonzalez. In Purcell, the Court strongly disfavored judicial changes to election regulations close to election day. Such changes, according to the Court, “can themselves result in voter confusion and consequent incentive to remain away from the polls.” “As an election draws closer,” moreover, “that risk will increase.”
Since it was announced, the Purcell principle has reared its head every two years as elections have approached. But it has never been as important as this year for one simple reason: There has never been as much litigation in the leadup to an election. According to the COVID-Related Election Litigation Tracker, more than three hundred election law cases have been filed in 2020 in more than forty states. Among (many) other issues, these suits have addressed polling place locations and procedures, deadlines for requesting and returning absentee ballots, witness and notarization requirements for absentee ballots, and signature thresholds for qualifying for the ballot. The Supreme Court has already resolved half a dozen disputes on Purcell grounds—an all-time high—three of which included reasons for the Court’s actions or written dissents. And the Court will surely confront Purcell again before this election is over. Lower-court election litigation continues to rage, making it inevitable that more appeals will land on the Court’s doorstep.
Despite all this activity, the Purcell principle remains remarkably opaque. Precisely because it is a shadow doctrine, appearing only in the Court’s shadow docket, its contours have never been clarified. The above quote from Purcell itself was almost all the Court had to say about the rule against late-breaking judicial intervention when the Court first unveiled this policy. Since Purcell, the Court has added only a few more sentences about the doctrine’s operation. In April of this year, the Court praised “the wisdom of the Purcell principle,” and in August, the Court held that the principle carries less weight when “state election officials support the challenged decree.” True, occasional dissents from the Court’s Purcell jurisprudence have explored in somewhat more detail when courts should and shouldn’t change election regulations close to election day. But these have still been skimpy opinions that didn’t purport to offer a comprehensive analytical framework.
My aim in this piece, then, is to put some meat on Purcell’s bones: to consider more fully when judicial intervention near an election is inadvisable and when, conversely, it’s prudent or even urgent. I make two main points. First, the Purcell principle can’t be an ironclad rule. The Court’s own rationales for the doctrine indicate that judicial revision of election regulations can sometimes be appropriate despite the imminence of an election. Second, the circumstances under which judicial action is warranted, even though an election beckons, are reasonably foreseeable. They include (1) when a court’s remedy will cause little voter confusion; (2) when a court’s remedy will cause little administrator error; (3) when, if a court fails to intercede, significant disenfranchisement will ensue; (4) when plaintiffs have diligently pursued their claim; and (5) when an election is further rather than closer based on Congress’s judgments about election proximity.
Start with the argument that Purcell should be understood as a presumption against—not a prohibition of—judicial intervention near an election. This is the reading most consistent with Purcell’s actual language. In deciding whether to issue injunctions in election law cases, the Court held, lower courts are “required to weigh” the possibility that their orders will cause voter confusion and consequent disenfranchisement. Of course, to weigh a factor is to take it into account, to examine carefully all the evidence that bears on it. To weigh a factor is not to make it dispositive in all cases, to prioritize it over all competing values. So Purcell can’t fairly be construed as a categorical bar on courts amending election regulations close to election day. The decision is more like an admonition that courts considering such amendments take seriously the prospect that the changes will confuse or even disenfranchise certain voters.
With respect to this “voter confusion and consequent incentive to remain away from the polls,” furthermore, it’s plainly variable rather than fixed. As the Court observed in Purcell, injunctions near an election “can themselves result in” this harm. But such orders don’t necessarily do so. It all depends on what exactly is directed, what the law on the books previously said, how much time there is to implement the injunction, and so on. Put another way, courts shouldn’t assume that their interventions close to election day will be confusing or even disenfranchising. Instead, they should analyze whether this will be the case. Sometimes, upon examination, it will turn out that their proposed remedies will not perplex voters or deter them from voting. In this scenario, Purcell doesn’t require judicial abstention.
This discussion all fits within Purcell’s four corners. The point is that, under Purcell’s own logic, courts shouldn’t always refrain from stepping in when an election is nigh. But Purcell doesn’t exhaust the equitable factors that are relevant to shaping a proper remedy. In particular, it doesn’t take into account the potential illegality or impact of a challenged policy. Suppose a state enacts a blatantly unconstitutional law, on the eve of an election, that would disenfranchise many of its citizens—a poll tax, say. Also imagine that judicial nullification of this law, right after it was passed with much fanfare, would confuse some people. Would Purcell really compel a court to sit on its hands in this situation? Surely not. The confusion caused by judicial interference would be outweighed by the policy’s legal invalidity and democratic illegitimacy.
That Purcell isn’t always binding is demonstrated, too, by the Court’s record applying it. Several months prior to the 2018 election—well before Purcell’s concerns could have been triggered—a district court enjoined North Dakota’s requirement that voters show IDs that include their residential addresses. Then, in late September 2018 (and so quite close to the election), the Eighth Circuit stayed the district court’s order. Although “[t]he risk of voter confusion [was] severe here because the injunction against requiring residential-address identification was in force during the primary election,” as Justice Ginsburg argued in dissent, the Court upheld the Eighth Circuit’s stay. The Court thus gave short shrift to Purcell’s worry about voter confusion, instead prioritizing issues nowhere to be found in the decision (presumably, the Court’s view that ID requirements are lawful).
This sequence just repeated itself in litigation about the re-enfranchisement of ex-felons in Florida. In October 2019, a district court enjoined a Florida law requiring ex-felons to pay their outstanding fines in order to qualify to vote. This injunction remained in place until July 2020, when it was lifted by the Eleventh Circuit less than three weeks before the voter registration deadline. Again, the Court upheld the lifting of the injunction even though, as Justice Sotomayor wrote in her dissent, “the Eleventh Circuit . . . created the very ‘confusion’ and voter chill that Purcell counsels courts to avoid.” Justice Sotomayor then underscored the Court’s inconsistent application of Purcell. “[F]aced with an appellate court stay that disrupts a legal status quo and risks immense disfranchisement—a situation that Purcell sought to avoid—the Court balks.”
The Court’s own practice, then, confirms that Purcell isn’t an absolute. But if courts sometimes should and sometimes shouldn’t intervene near an election, how are they supposed to distinguish one setting from the other? Most of the factors that should guide courts in making this distinction have already come up. But I now want to describe these considerations in somewhat more detail.
The first relevant factor is the one stressed by Purcell itself: the probability that judicial changes to election regulations close to election day will confuse voters and dissuade some of them from voting. As explained earlier, courts shouldn’t assume that this probability is high; they should assess it based on the best available evidence. In this assessment, much will often hinge on the kind of policy that’s being challenged. Some policies, like a district plan or a jurisdiction’s underlying electoral rule (at-large voting, plurality voting, ranked-choice voting, and so on), affect every aspect of the political process. Candidates choose to run (or not) based on these policies, and the policies also drive candidate fundraising, media attention, voter interest, and the style of the campaign. Such basic building blocks of the election should almost never be upset when time is limited. Significant voter confusion is likely when the electoral environment is transformed shortly before votes are cast.
On the other hand, other policies are incapable of causing voter confusion (whether they shift or stay the same) because they apply only to administrators. Consider a signature-match requirement for absentee ballots, under which voters’ signatures on the ballots must resemble those on the registration rolls. This requirement is enforced exclusively by election officials, and if the rule is amended or waived, they’re the only ones who must modify their behavior. For exactly this reason, a North Dakota district court recently dismissed Purcell’s concerns when it enjoined the state’s signature-match law. “[T]here is no potential for voter confusion or dissuasion from voting because the process for submitting an absentee ballot will remain unchanged.”
Between these poles, unfortunately, lie most litigated policies. They’re not pillars of the electoral system, but they do apply directly to voters. With respect to these policies, courts should keep in mind that how they craft their remedies can influence how much voter confusion ensues. Take a requirement that voters procure two witnesses for absentee ballots and then have them notarized. If a court struck down the notarization rule but left the two-witness rule in place, then voters might be quite confused. They might be unsure what their new obligations were, and some might throw up their hands and forget about voting. But if a court invalidated both rules, then its ruling might be substantially more understandable. Voters wouldn’t have to interact with any third parties before returning their absentee ballots. That’s a simple message that’s easy for administrators to convey and voters to grasp.
This reference to administrators brings me to a second consideration: Courts should avoid changing election regulations near an election when, by doing so, they would likely cause election officials to make serious mistakes. Pause for a moment to sympathize for election officials. Even in the absence of judicial intervention, they have to comply with labyrinthine election codes, directives from the secretary of state, and local guidelines. When courts step in at the last minute, they can (though they need not) make it still harder to run an election. Court orders can disrupt administrators’ familiar routines, compel them to make determinations for which they lack training or experience, and extend how long each step in the process takes. As a result, the vote count can be slowed or even rendered inaccurate thanks to election officials’ missteps under the new court-imposed rules. Beyond the possibility of voter confusion, then, courts contemplating action close to election day should evaluate the risk of administrator error. When this risk is severe, discretion may be the better part of valor.
Administrator error, though, isn’t equivalent to administrator inconvenience. Almost any judicial revision of election regulations—near or far from an election—will lead to more work for election officials. This extra work is no reason for courts not to remedy legal violations unless it genuinely threatens to delay or distort the vote count. As the Eleventh Circuit recently stated in response to Alabama’s argument that it would be burdensome to abide by judicially amended absentee ballot procedures, “requir[ing] defendants to provide additional training to ballot workers” is “a feat hardly impossible in the allotted time.” Sharpening the point, the court added, “Purcell is not a magic wand that defendants can wave to make any unconstitutional election restriction disappear so long as an impending election exists.”
Third, courts should take into account the disenfranchisement that will follow if they decline to intercede. Fear of disenfranchisement is the driver of the previous two factors. When courts change election regulations close to election day, they sometimes cause voter confusion and administrator error. In turn, confused would-be voters sometimes choose not to vote, and harried election officials sometimes fail to count validly cast ballots. Crucially, however, judicial intervention isn’t the only step that can lead to disenfranchisement. Judicial abstention can, too, when it allows an unconstitutional policy that unjustifiably burdens voting to stay in effect. Courts thinking about action near an election should therefore balance the disenfranchisement if they do interfere (from voter confusion and/or administrator error) against the disenfranchisement if they don’t (from the unlawful status quo). When the latter is larger than the former, the judicial calculus should tilt in favor of enjoining or otherwise amending the illegal policy. In this case, Purcell’s own goal of minimizing voters’ “incentive to remain away from the polls” is advanced by courts entering the fray.
Justice Ginsburg twice made this argument in dissents from the Supreme Court’s Purcell rulings. In 2015, the Court stayed a district court injunction barring Texas from enforcing its photo ID requirement for voting. According to Justice Ginsburg, there was “little risk” that this injunction would “in fact disrupt Texas’s electoral processes.” In contrast, the Court’s stay posed “[t]he greatest threat to public confidence in elections” since it would lead to Texas “denying the right to vote to hundreds of thousands of eligible voters.” In April of this year, similarly, the Court reversed a district court order that would have permitted Wisconsin voters to postmark absentee ballots after election day. Justice Ginsburg again criticized the Court’s decision because many more votes would have been enabled by the district court’s remedy than deterred due to voter confusion. “The concerns advanced by the Court . . . pale in comparison to the risk that tens of thousands of voters will be disenfranchised.”
Fourth, it matters whether plaintiffs diligently developed their claim or, conversely, dallied when they should have hurried. Suppose litigants waited to attack an election regulation until an election was imminent, even though the law was on the books for years and its burdens were constant over time. Then a court might sensibly postpone any action until after the election. Pre-election intervention could cause the usual Purcell harms of voter confusion and administrator error. And these harms could have been entirely avoided had the lawsuit only been launched earlier. As the Sixth Circuit held in 2016, Purcell’s worries ring “especially true when a plaintiff has unreasonably delayed bringing his claim.”
Other challenges to election regulations, though, couldn’t have been mounted any sooner. Say a state passes an onerous election law close to election day. Then pre-election litigation is both inevitable and unattributable to any tardiness on plaintiffs’ part. Likewise, circumstances can change when an exogenous shock like a hurricane or a pandemic strikes near an election. Under these new conditions, election rules that are normally unproblematic can become far more burdensome. Here, too, plaintiffs couldn’t have sued any earlier because the event that gave rise to their claim hadn’t yet occurred. In the words of a Georgia district court, this is why “a hard-and-fast rule against modifying election regulations close to an election” is untenable. It’s “inherently incompatible with emergencies, which by definition arise unexpectedly and may jeopardize fundamental voting processes.”
Finally, it’s significant just how close the next election is. When it’s very near, it’s more likely that a court order will cause voter confusion and administrator error. “As an election draws closer,” to quote Purcell,“that risk will increase.” By the same token, when an election is less proximate, there’s more time for voters and election officials to adjust to judicially mandated changes to election procedures. Beyond this intuitive point, federal law can provide some guidance as to when Purcell’s concerns are more and less acute. The Bipartisan Campaign Reform Act (BCRA) applies special restrictions to campaign ads aired within 60 days of a general election. The Military and Overseas Voter Empowerment (MOVE) Act requires absentee ballots to be sent to certain voters at least 45 days before a federal election. These statutes suggest that 60 days and 45 days prior to an election are pertinent dates for courts considering granting relief. After the first, Congress thinks the campaign is proceeding in earnest, and after the second, the actual mechanics of voting are in motion.
It won’t be lost on readers that we’re now within both BCRA’s and the MOVE Act’s windows. This fact should weigh against judicial intervention at this point in the election calendar. But it also shouldn’t be decisive. The central theme of this discussion is that courts should examine a series of factors when contemplating action close to election day, all of which are relevant and none of which is dispositive. So, now or even nearer the election, courts shouldn’t hesitate to step in if their remedies won’t baffle voters, won’t lead administrators to make mistakes, will prevent disenfranchisement, and couldn’t feasibly have been imposed sooner. This is simply the nature of a multifactor standard. It can sometimes lead to a conclusion (judicial relief) even if a particular prong (proximity to the election) points in a different direction.
For some lawyers, of course, that’s precisely the problem: I’m interpreting Purcell as a standard, not a hard-edged rule. To reiterate, though, the Supreme Court doesn’t treat Purcell as a rule either. In at least two recent cases, the Court has allowed lower courts to disrupt the electoral status quo quite close to election day. Moreover, Purcell is just a part of the broader legal analysis that courts conduct when determining whether to grant or lift a stay. And that broader analysis is—you guessed it—a multifactor standard including four distinct elements. So Purcell is an odd hill for proponents of rigid rules to die on. It lies in territory, the fashioning of remedies, that has long been ceded to flexible standards.