1 Making Voter Registration Harder

One of the post-VRA strategies by which many states reduce African American voter turnout is making the registration process as hard as possible.Footnote ¹³³ Responding, Congress in 1993 passed the National Voter Registration Act (the NVRA).Footnote ¹³⁴ Among other things, that Act requires every state to offer its citizens three ways to register to vote in federal elections: simultaneously with an application for a driver’s license; by mail; and in person at designated federal, state, or nongovernmental offices.Footnote ¹³⁵ (States are free to offer additional options as well). Although the NVRA governs only federal elections, states typically conduct federal and state elections at the same time. As a practical matter, therefore, a single registration usually qualifies the person to vote in elections for both federal and state offices.

A second important provision of the NVRA requires states to “accept and use” (for federal elections) the standardized voter registration application form prescribed by a federal agency called the Election Assistance Commission (the EAC).Footnote ¹³⁶ This application is commonly referred to as the “Federal Form.”Footnote ¹³⁷ Although the Federal Form contains uniform, nationwide content, it also includes any necessary state-specific content – most importantly, the address for mail-in registrations and the state’s voter eligibility and registration criteria.Footnote ¹³⁸ All state-specific information requires EAC approval.Footnote ¹³⁹ If the EAC denies a state’s request to add particular content to its registration form, the state has the right to challenge the EAC’s decision in court.Footnote ¹⁴⁰

On the issue of citizenship, all the Federal Form requires is that the applicant attest under penalty of perjury that he or she is a US citizen. Undeterred by the NVRA, the State of Arizona adopted, by popular initiative in 2004, a law that required something additional before one could register to vote – furnishing documentary proof of US citizenship. That is often more easily said than done. Birth certificates and passports – the most common ways in which those who were born in the US can prove their citizenship – are often harder for the poor to access. In Arizona v. Inter Tribal Council of Arizona [ITCA], the US Supreme Court held that states may not require the applicant “to submit information beyond that required by the [Federal Form].”Footnote ¹⁴¹ By insisting on further proof of citizenship in the form of documents, the state was therefore in violation of the NVRA.

At the same time, the Court reaffirmed the distinction between the “times, places, and manner” of federal elections – a decision on which the Constitution gives Congress the last word – and the substantive voter eligibility requirements – which the state has the sole power to decideFootnote ¹⁴² (subject to any specific constitutional constraints). Thus, a state may, and every state does, require US citizenship as a condition for voting in federal elections. But it may not insist on more proof of US citizenship than what the Federal Form regards as sufficient. Arizona’s only recourse, the Court said, was to request the EAC’s permission to add documentary proof of citizenship to the state-specific information on its version of the Federal Form and, if denied, to seek review in court.Footnote ¹⁴³ The Court’s decision is comforting, but as explained in Section B.4 below, there are reasons to fear successful future state efforts to condition voting on documentary proof of US citizenship.

Arizona, joined by Kansas, took up that suggestion. Both requested the EAC’s permission to add documentary proof of citizenship to the state-specific instructions to their respective versions of the Federal Form. The EAC denied both requests, and both states sought judicial review. The federal district court granted relief, holding that the EAC has a “non-discretionary” duty to approve those states’ requests, but in Kobach v. U.S. Election Assistance Commission,Footnote ¹⁴⁴ the US Court of Appeals for the Tenth Circuit reversed. The EAC decision was discretionary, the court held, and it found no basis for second-guessing the Commission’s call. The Tenth Circuit reached the same result in a 2020 case, adding that requiring documentary proof of citizenship also violated the Constitution’s equal protection clause. In the past nineteen years, the requirement had caused the cancellation or suspension of more than 30,000 voter registrations of US citizens in Kansas, compared to those of only thirty-nine non-US citizens who had managed to register to vote.Footnote ¹⁴⁵ The combination of a miniscule “citizenship fraud” problem and a massive disenfranchisement solution was held to violate the constitutional requirement of equal protection.

But there were (and are) still other ways to make voter registration harder. Earlier discussion highlighted the actions taken by North Carolina, starting one day after the Supreme Court had gutted the VRA in the 2013 Shelby County case.Footnote ¹⁴⁶ The package of five voter restriction laws passed by the North Carolina state legislature in the wake of Shelby County included two that specifically targeted the registration process. The US Court of Appeals for the Fourth Circuit, in N.C. State Conference of the NAACP v. McCrory,Footnote ¹⁴⁷ ultimately struck down all five restrictions on equal protection grounds. As noted earlier, the court found overwhelming evidence of a specific intent to discriminate against African American voters with “almost surgical precision.”Footnote ¹⁴⁸

One of the two registration-related laws was a repeal of preregistration. This was a process that had allowed sixteen- and seventeen-year-olds to preregister to vote when obtaining driver’s licenses or attending mandatory high school registration drives. The process had been highly successful, enabling elections officials to verify eligibility in advance and register eligible voters once they reached age eighteen. It had increased turnout among young voters.Footnote ¹⁴⁹

The other law relevant here was a repeal of same-day registration. That procedure had been especially valuable to several groups – those who had not been able to register earlier, those who had been shunted into the “incomplete registration queue” after previous unsuccessful attempts to register, those who moved frequently, and those who needed personal assistance from poll workers.Footnote ¹⁵⁰

These two laws had two common denominators. For multiple reasons laid out by the court, they had a disproportionate adverse impact on African American turnout. And both laws had been passed only after the legislature had sought and obtained data demonstrating those disproportionate adverse effects.Footnote ¹⁵¹ The evidence of discriminatory intent was unmistakable.

It shouldn’t have to be this hard. As Issacharoff et al. point out, one state (North Dakota) has decided that no voter registration at all is necessary. You just show up and vote. Twenty others allow same-day registration.Footnote ¹⁵² The United States, as the same authors observe, “is distinctive among western democracies in that the government takes virtually no affirmative responsibility for registering citizens [to vote].” In almost all other democracies, registration is either automatic when citizens reach voting ageFootnote ¹⁵³ or proactively assisted by the national government – typically without requiring identification or allowing non-photo identification. And some countries, including Argentina, Australia, and Brazil, make voting mandatory.Footnote ¹⁵⁴

2 Purging Voter Rolls

Getting your name onto the voter registration list is one thing. Making sure it stays there is another.

From time to time, states cull names from their voter rolls. They have legitimate reasons to do so. Voters die, move out of their districts, or become ineligible to vote because of felony convictions. Updating the list helps prevent people from either impersonating dead voters or voting in multiple districts in the same election, rare as voter fraud is in reality.Footnote ¹⁵⁵ It also aids states in keeping their registration lists at a manageable length. The NVRA in fact requires states to make a “reasonable” effort to remove the names of voters who have died or moved out of their voting districts.Footnote ¹⁵⁶

At the same time, administrative errors in maintaining voter registration lists are not uncommon; voters should not be disenfranchised because of those errors. Beyond that concern, an unbridled power of state election officials to purge voters on change of residence grounds would allow election officials of one particular political party to deliberately target population areas that tend to vote for the opposing party. Examples of precisely such practices, typically aimed at minority populations, are endemic and are highlighted below. Until passage of the NVRA in 1993, in fact, some states would purge people from the voting rolls without any notice and therefore without any opportunity to confirm their continued residence within the district.Footnote ¹⁵⁷

A word on terminology: As explained by Ballotpedia, “Voter caging [by either public or private actors] is the practice of sending mail to registered voters and challenging their eligibility to vote if the mail is returned as undeliverable.” The Ballotpedia article offers several modern examples of voter caging, all by either the national or state Republican Parties. The next step, “[v]oter purging, is the practice of removing names from the voter rolls …”Footnote ¹⁵⁸

As discussed earlier, the Supreme Court’s 2013 decision in Shelby County v. HolderFootnote ¹⁵⁹ cut out the heart of the 1965 VRA by effectively nullifying Section 5 of that Act. This was the provision that required states and counties with histories of racial discrimination in voting, as measured by specified metrics, to obtain advance Justice Department clearance for any proposed changes to their election laws. As the earlier discussion illustrated, once freed from that requirement, the states and counties that had been subject to it wasted no time in adopting or accelerating a range of strategies to suppress African American and other minority votes.

Voter purging is one of the many suppression strategies that has proliferated in the wake of Shelby County. A Brennan Center study found that, in the five years immediately following Shelby County (2013–18), four states actually carried out, and four others adopted rules that would permit, illegal purges. Those eight states together account for one-fourth of the nation’s registered voters.Footnote ¹⁶⁰

After Shelby County, actual voter purging escalated dramatically in the counties that the Court had freed from the preclearance requirement. The Brennan Center report doesn’t assess how many of those purges were improper, but the circumstantial evidence is strong. Using data compiled by the US Election Assistance Commission (the EAC), the Brennan Center spotted two empirical patterns. First, the purge rates of counties whose histories of racial discrimination had subjected them to the VRA’s preclearance requirement had been at roughly the national average while the preclearance requirement was keeping them in check. As soon as the Supreme Court lifted that requirement, their purge rates increased instantly and sharply. Second, their post-Shelby County purge rates were significantly higher than those for counties that had not been subject to preclearance.Footnote ¹⁶¹ So much higher, in fact, that if the former preclearance states had purged only at the lower rate prevailing in the other states, two million fewer voters would have been purged during the interval between the 2012 and 2016 elections.Footnote ¹⁶²

The numbers are significant. The same Brennan Center study compared the national voter purge figures for the 2006–2008 period (when preclearance was still in force) to the 2014–16 period (just after Shelby County). From the former period to the latter, the number of voters purged from the rolls increased by one-third – and by an even higher percentage in the former preclearance states and counties.Footnote ¹⁶³

Wrongful purges haven’t been confined to any one geographic region of the country. Thousands of eligible voters were erroneously purged in Virginia in 2013 and in both New York and Arkansas in 2016.Footnote ¹⁶⁴ Wisconsin experienced a near miss. In 2019, a Wisconsin state court judge ordered the purge of 200,000 voters, a majority of whom lived in Democratic Party strongholds. His order was reversed by both the state court of appeals and the state’s Supreme Court.Footnote ¹⁶⁵ And “Georgia purged twice as many voters – 1.5 million – between the 2012 and 2016 elections [i.e., mainly after Shelby County] as it did between 2008 and 2012 [i.e., before Shelby County].”Footnote ¹⁶⁶

But the mother of all voter purges – in terms of both numbers and audacity – has to be Georgia’s subsequent 2017 purge of 560,000 voters, just in time for the 2018 gubernatorial election. These individuals were purged for having failed to vote often enough in recent elections. What makes that purge particularly noteworthy is that it was ordered by Georgia’s then-secretary of state, Brian Kemp, himself a candidate for governor. A Republican, he was opposed in the general election by Democrat Stacey Abrams, an African American with strong support in the very places that the purge disproportionately singled out. The tactic was successful; Kemp won the election by fewer than 55,000 votes out of almost 4 million votes cast.Footnote ¹⁶⁷ An investigation revealed that approximately 107,000 of the purged voters would have been eligible to vote; 70,000 of them, in fact, reregistered to vote after the election.Footnote ¹⁶⁸

In theory, the NVRA should constrain manipulative voter purges. For purposes of federal elections, it prohibits states from removing a voter’s name “by reason of the person’s failure to vote.”Footnote ¹⁶⁹ But there are loopholes. The Help America Vote Act (HAVA), passed in 2002, contains a similar provision but with the addition of a keyword: It says only that the person’s name may not be removed “solely” for failure to vote.Footnote ¹⁷⁰ That language permits consideration of failure to vote when it is coupled with other factors. In addition, the NVRA provision expressly allows the state to remove a voter’s name for presumed change of residence in two (though only in two) circumstances. It may do so if the person confirms in writing that he or she has moved out of the district. And it may do so if the elections officials mail the person a preaddressed, postage-prepaid card requesting confirmation that the voter is still at the same address and the person does not reply – and then fails to vote in at least one of the next two general federal elections.Footnote ¹⁷¹ Moreover, as noted earlier, the NVRA affirmatively requires states to make a “reasonable” effort to remove the names of people who have either died or left their districts.Footnote ¹⁷²

But here’s the question: To whom may a state send these confirmation requests in the first place? Limitations are obviously essential. Unfettered, a state could send the confirmation notices selectively to voters or population centers that are known to support one particular political party, with the result that some percentage of the addressees will fail to return the cards and thereby disqualify themselves from voting. The NVRA allows states to send those requests to people who have submitted change-of-address information to the US Postal Service,Footnote ¹⁷³ but it says nothing about whether the cards may be sent under any other circumstances. In particular, may the state choose to send the notices only to those who have failed to vote recently enough or often enough?

This is where Ohio and the US Supreme Court come in. Under Ohio law, the state automatically sends confirmation requests to anyone who hasn’t voted (or registered to vote or signed a petition or updated a voting address) in the past two years. If the person doesn’t respond and doesn’t vote in the next four years (including two general federal elections), that person’s name is then removed from the registration list.

In Husted v. A. Philip Randolph Institute,Footnote ¹⁷⁴ the Supreme Court split along partisan lines in upholding the Ohio law. The five Republican Justices interpreted the NVRA to mean that the person’s failure to vote can still be a factor in its decision to remove the person’s name – just not the sole factor. Because Ohio doesn’t remove people for failure to vote unless they also failed to return their confirmation cards, the Court held that Ohio’s practice was lawful – even though the sole reason for sending the cards to those particular individuals in the first place had been their failure to vote in the past two years.

Justice Sotomayor, joining the principal dissent and adding her own, is explicit in highlighting which of these voters are most likely to be erroneously disenfranchised. Calling out the elephant in the room, she observes that “Congress enacted the NVRA against the backdrop of substantial efforts of States to disenfranchise low-income and minority voters, including programs that purged eligible voters from registration lists because they failed to vote in prior elections” [emphasis added]. She cited an additional study documenting various states’ registration practices that “sharply reduced turnout, particularly among blacks and immigrants.”Footnote ¹⁷⁵ The American Bar Association agrees, observing that “[f]ailure to vote regularly correlates with lower socioeconomic status and, at least in some places, with being a member of a racial minority.”Footnote ¹⁷⁶

Justice Breyer, writing for the four dissenters and citing the House committee report on the bill that would become the NVRA, similarly points out a history of “selective purges” in the late nineteenth and early twentieth centuries. These, the report emphasized, were designed “to keep certain groups of citizens from voting” and were among the very reasons Congress passed the NVRA.Footnote ¹⁷⁷

Because the negotiations and strategy sessions that precede the adoption of purging policies are ordinarily hidden from view, there is no way to know how much of the disparate impact on minority voters is intentional and how much is the product of either a “let the chips fall where they may” decision or innocent error. One Brennan Center study attributes some of the problems to a combination of faulty data and faulty methodologies.Footnote ¹⁷⁸ Another Brennan Center study pinpoints errors involving voters with the same names and dates of birth, especially in large states. That study points out that identical names are particularly common in minority communities.Footnote ¹⁷⁹ One organization capsulizes these multiple problems in one succinct sentence: Voter purging can be “shrouded in secrecy, prone to error and vulnerable to manipulation.”Footnote ¹⁸⁰

Apart from the disparate impact of these practices, failure to vote in recent elections is a highly inaccurate proxy for likelihood of death or change of residence. More than one-third of all eligible voters nationwide sit out even the high turnout presidential elections.Footnote ¹⁸¹ Far fewer than that number die or move out of their districts within those same narrow time periods. Singling out that population for the required return of confirmation notices thus does not make even statistical sense.

A use-it-or-lose-it policy also doesn’t make legal sense. In the United States, eligible voters are never obligated to vote. You have the right to sit out an election without risking the forfeiture of your right to vote in future elections.

If failure to vote in recent elections is a weak proxy for death or change of residence, failure to reply to a card received in the mail is, if anything, an even weaker proxy. As Justice Breyer’s dissent observes, Ohio sent cards to 20 percent of the state’s registered voters. About two-thirds of the recipients – that is, roughly 13 percent of the state’s registered voters – failed to return the cards. Yet, only about 4 percent of Americans move from their counties each year. And there is no reason to think Ohioans move across county lines three times as often as other Americans. The much more obvious explanation is “the human tendency not to send back cards received in the mail.” It is clear that Ohio removes far more eligible voters than ineligible voters.Footnote ¹⁸²

Finally, reliance on such poor proxies for voter ineligibility is entirely unnecessary. If the goal really were simply to cull the names of voters who have died or moved out of the district, or who otherwise had managed to double register, there are simpler and far more reliable methods. The American Bar Association explains:

ERIC began in 2012.Footnote ¹⁸⁴ At its peak, it had 32 member states.Footnote ¹⁸⁵ By any objective measure, it has been a stunning success. Jesse Wegman describes why:

He continues:

Republican and Democratic officials alike offered “glowing” reviews.Footnote ¹⁸⁷

Then something happened. In January 2022, the far-right website “Gateway Pundit” made the puzzling announcement that ERIC was “essentially a left-wing voter registration drive disguised as voter roll cleanup.” Former President Trump piled on two months later, claiming falsely that ERIC “pumps the rolls” for Democrats. Predictably, Republican-controlled states began bailing out. In the eighteen months following the Gateway Pundit posting, at least eight states, all Republican-controlled, left ERIC. More are expected to follow.Footnote ¹⁸⁸ Their departures will make it politically easier to substitute far less accurate purging methods that target unfavorable voting populations in the partisan ways previously discussed.

Provisional voting has mitigated some of the harmful effects of voter purging. The 2002 Help America Vote Act (HAVA), mentioned earlier, gives voters the right to cast provisional ballots. They may do so when an election official tells them their name doesn’t appear on the registration list and the person then declares in writing that he or she is a registered voter who is eligible to vote in the particular jurisdiction. If the election officials later verify that the person was indeed eligible to vote in that jurisdiction, the provisional ballot is counted.Footnote ¹⁸⁹

That is a welcome development. Surely, however, it would be better to address the problem of wrongful purging than to rely on after-the-fact, case by case, mitigation. Not every voter will be willing to go through the written declaration process. Voters who learn in advance that their names have been purged might also have less incentive to stand in long voting lines with the prospect of then being turned away. At any rate, the need to verify the provisional ballots adds further delay to a vote-counting process that too often is already needlessly prolonged. More effective would be to prohibit the practice of states requiring the return of confirmation notices solely for those who have not voted recently enough – the step the five Republican Supreme Court Justices were unwilling to take in Husted.

A final note: Apart from purging the names of voters, can a state constitutionally purge members of Congress? In 1973 Arkansas, by voter initiative, amended its constitution to prohibit anyone who had already served three terms in the US House or two terms in the US Senate from appearing on the ballot for another term.Footnote ¹⁹⁰ At the time, whether coincidentally or not, three of the state’s four US House representatives were Democrats (Reps. Alexander, Thornton, and Anthony), as were both of its US Senators (Pryor and Bumpers).Footnote ¹⁹¹ In Term Limits v. Thornton,Footnote ¹⁹² the Supreme Court struck down the amendment, holding that the Constitution lays out the exclusive requirements for serving in Congress and that the states’ power to decide the “manner” of congressional elections did not authorize them to superimpose additional qualifications.

3 Requiring Photo IDs

In 1950, South Carolina passed the nation’s first voter ID law. Since then, these laws have proliferated.Footnote ¹⁹³ By March 29, 2023, some 36 states were requiring specified types of identification documents for in-person voting. The remaining states and DC verify the voter’s identity in other ways, usually by comparing the voter’s signature against signatures that are already on file.Footnote ¹⁹⁴

One state’s experience bears special mention. As noted earlier, the North Carolina Republican-controlled legislature’s all-out assault on voting rights began literally 24 hours after the Supreme Court in Shelby County had cut out the heart of the preclearance requirement for states (including North Carolina) with histories of racial discrimination in voting. The legislature commissioned various studies of the impact that certain changes in the election laws would have on African American turnout. It then used those data to pass a series of measures to depress African American votes with what the US Court of Appeals for the Fourth Circuit in the McCrory case called “almost surgical precision.”Footnote ¹⁹⁵

One of those strategies was to require voters to present one of several specified photo ID documents (including driver’s licenses) that the legislature’s data showed to be disproportionately lacking among African American voters. It simultaneously excluded alternative documents that African Americans were disproportionately likely to possess.Footnote ¹⁹⁶ The Fourth Circuit in McCrory had little difficulty in striking down this entire bundle of restrictions.

In Crawford v. Marion County Election Bd.,Footnote ¹⁹⁷ the Supreme Court upheld Indiana’s voter ID law, mainly on the now familiar theory that it helps prevent voter fraud. But the only type of voter fraud that voter ID laws even ostensibly prevent is impersonation of eligible voters.Footnote ¹⁹⁸ ID documents establish, at most, that the people who present themselves at polling stations are who they say they are. They don’t prove US citizenship or other voter eligibility requirements such as noncriminal backgrounds. Some of the acceptable ID documents display a residential address, but even these don’t prove that the address is still current. Nor do IDs prevent double voting.

This limited purpose is worth keeping in mind, because, rare as voter fraud is in general,Footnote ¹⁹⁹ voter impersonation fraud is rarer still. In his Crawford dissent, Justice Souter says this:

The Brennan Center for Justice agrees, adding “Our research has established that impersonation fraud rarely occurs. Indeed, more Americans are struck by lightning each year.”Footnote ²⁰¹

Those results should not be surprising. An individual would rarely find voter impersonation to be worth the risk. There’s almost no chance that a single vote will alter the outcome of an election and, for anyone caught, the criminal penalties would be severe.Footnote ²⁰²

Against these minimal or nonexistent benefits, there are huge costs. They disproportionately hamper the poor, racial minorities, and the elderly.Footnote ²⁰³ That is both because those groups are less likely than the general population to have qualifying IDs and because the “free” ID cards that voter ID states provide for voters who need them are seldom actually cost-free. The two principal costs relate to travel and documentation.

As for travel, a Brennan Center study by Keesha Gaskins & Sundeep Iyar sums up the practicalities:

Apart from travel costs, applicants need certain primary documents in order to acquire the acceptable IDs. Those primary documents can themselves place travel, time, and financial burdens on poor citizens. Again, Gaskins and Iyar explain:

As of 2012, this combination of factors had left 10 percent of voting-age US citizens without unexpired government-issued IDs. For African Americans, Hispanics, and those over age sixty-five, the percentages in 2012 were higher still – 25%, 16%, and 18%, respectively.Footnote ²⁰⁶ Justice Souter’s powerful dissent in Crawford supplies further documentation of these same problems. He points out that, for thousands of the state’s residents, the travel required to obtain a qualifying ID would be burdensome, and the costs of both the travel itself and the procurement of the documents needed to obtain the required ID significant.Footnote ²⁰⁷ The studies he cites show how disproportionately the ID requirements burden the poor and African Americans especially.Footnote ²⁰⁸

The particulars of many of the voter ID laws – not all – specifically disadvantage college and university students as well. Those effects require special discussion and are considered below.

Given the flimsy character of voter ID laws’ claimed benefits and their immense toll on whole swaths of the voter-eligible population, it is fair to ask why these laws have become so popular. There is only one credible answer – crass partisanship by state legislatures and governors.

How do we know? Occasionally, Republican Party operatives have let their guard down either orally or in writing, saying the quiet part out loud. A staffer for a Wisconsin Republican state legislator quit his job and left the party before writing “I was in the closed Senate Republican Caucus when the final round of multiple Voter ID bills were being discussed. A handful of the GOP Senators were giddy about the ramifications and literally singled out the prospects of suppressing minority and college voters.” Both the former Republican Party Chair and the former Republican governor of Florida similarly admitted that their state’s “voter ID law was designed to suppress Democratic votes.”Footnote ²⁰⁹

Today, of course, most key operatives are savvy enough to whisper their partisan motives behind closed doors. But the circumstantial evidence is hard to dismiss. First, as the foregoing discussion demonstrates, the chasm between the minimal-to-nonexistent policy benefits claimed for voter ID laws and their conclusively proven harms belies any suggestion that the decisions to pass these laws were based on the merits. Second, there is a clear correlation between strict voting laws (including voter ID laws) and Republican control of the legislative process.Footnote ²¹⁰ Even the Supreme Court in Crawford acknowledged that, since every Republican legislator had voted in favor of the ID requirement and every Democrat had voted against it, “it is fair to infer that partisan considerations may have played a significant role in the decision to enact” the law.Footnote ²¹¹

Third, those groups that are disproportionately hampered by voter ID laws tend overwhelmingly to vote for Democrats. To be fair, this is not universally the case. Two of the groups especially likely to feel the burden of voter ID requirements are rural voters and elderly voters. Rural voters can be affected mainly because of difficulties in getting transportation to the government offices that issue free IDs. As noted earlier, these voters tend to skew Republican.Footnote ²¹² And the elderly can be affected for the same reason and for lack of current driver’s licenses, although any Republican losses attributable to the effects of voter ID laws on elderly voters are likely minimal. That is because, despite voter ID laws, seniors already have very high turnout rates. Moreover, their voting preferences fluctuate and, in times when they favor Republicans, they typically do so only by narrow margins.Footnote ²¹³

But three other groups – all of them heavily Democrat-leaning – bear by far the greatest brunt of the voter ID laws: the poor, racial minorities, and students. In Veasey v. Abbott, for example,Footnote ²¹⁴ a federal district judge had found that the Texas voter ID law was passed with the specific intent to discriminate against African American and Hispanic voters and that it would in fact produce discriminatory results, both in violation of the VRA. Even the ultraconservative US Court of Appeals for the Fifth Circuit agreed with the judge’s finding of discriminatory results. It also found ample evidence of discriminatory intent, though it ordered the judge to reconsider that finding because some of the evidence the judge had relied on was in its view not probative enough.

Perhaps the strongest evidence of ulterior partisan motives can be found in the lists of the specific documents that will satisfy particular states’ voter ID laws. As of March 2023, at least seven states prohibit the use of college and university IDs entirely.Footnote ²¹⁵ At least two others allow them only if the college or university is located within the particular state.Footnote ²¹⁶ And at least one other state (Georgia) allows only those IDs that are issued by public institutions, thus ruling out student IDs issued by the historically black private colleges and universities.Footnote ²¹⁷

Gun licenses are another story. Among the states that prohibit the use of student IDs or confine their acceptability to IDs from public educational institutions or to institutions located within the state, at least six take precisely the opposite approach with gun licenses. These states explicitly include them in the list of acceptable IDs, regardless of whether they were issued within the particular state.Footnote ²¹⁸

That contrast between accepting gun licenses and rejecting university IDs lends strong support to suspicions of partisan motives for voter ID laws. Young Americans – and especially college students – tend to vote overwhelmingly for Democratic candidates and progressive initiatives, and in recent years their turnout rates have increased rapidly.Footnote ²¹⁹ Republicans’ desires to minimize student turnout have been well publicized and manifest themselves in suppression strategies that go beyond voter ID laws.Footnote ²²⁰ In sharp contrast, as of December 2022, some 48% of Republicans reported that they owned at least one gun, compared to only 20% for Democrats. And 66% of Republicans reported living in a household in which there was at least one gun, compared to 31% for Democrats. Republicans, in other words, were more than twice as likely as Democrats to own at least one gun and more than twice as likely to live in a household where there was at least one gun.Footnote ²²¹

Despite the conclusive empirical refutations of claims of widespread voter fraud generally and impersonation of eligible voters in particular, despite voter ID laws’ serious and discriminatory impediments to voting, and despite strong evidence of inappropriate partisan motives, the Supreme Court in Crawford approved photo ID laws. Why?

The Court in Crawford identified three state interests that photo IDs are claimed to serve: verifying the person’s legal eligibility to vote; preventing voter fraud; and promoting citizens’ confidence in the integrity of the election process.Footnote ²²² All three interests are indisputably valid, but their connections to photo ID laws are embarrassingly thin. As noted earlier, and despite the Court’s statement to the contrary, photo IDs do not verify one’s eligibility to vote; at most, they verify the voter’s identity, a check that many other states find other secure ways to accomplish.Footnote ²²³ As for voter fraud, the Court acknowledges that the state failed to identify a single instance of it,Footnote ²²⁴ and as discussed earlier voter impersonation is truly rare. And while the Court extolls the state interest in citizen confidence in the integrity of the voting system, it never considers the effects on public confidence of the partisan manipulation of the voting laws that it candidly acknowledges.Footnote ²²⁵ Moreover, when there arises the very reasonable perception that laws making it as hard to vote as possible are deliberately designed to minimize turnout by citizens of color, one can safely assume that the threats to public confidence in the integrity of the electoral system become greater still.

The Court concludes nonetheless that the law is “nondiscriminatory” and “supported by valid neutral justifications.”Footnote ²²⁶ But in describing the law as “nondiscriminatory,” the Court is content to ask only whether the law is neutral on its face and whether it serves any theoretically plausible benefits. That is all it was willing to require.

Having stopped at that point, the Court never considers either the actual disproportionate impact on racial minorities or the state legislature’s motives in passing this law. After acknowledging the fair inference that partisan considerations might have played a significant role, the Court never asks the logical follow-up question: Why would the legislature’s partisan majority think photo ID’s serve the party’s interests? By failing to ask that question, the Court is able to avoid the answer that is obvious to everyone with a pulse: Because the Republicans know full well that people of color vote overwhelmingly for Democrats and that this is the voting bloc most adversely affected by restrictive voting requirements. (Republican politicians have protested that the reason photo ID laws benefit their party is that they help prevent voter fraud by Democrats; again, however, significant voter fraud has been shown to be nonexistent, and at any rate, there is no evidence to suggest that Democrats are more likely than Republicans to engage in it.)

Legislatures today are quite sophisticated in concealing impermissible motives. When the two opposing parties divide so sharply on this and practically every other voting issue, the Court needs to be willing to examine the magnitudes of both the benefits and the harms far more closely, and with its eyes more open to the realities on the ground, than it did in the Crawford case.

Justice Scalia’s concurring opinion, joined by Justices Thomas and Alito, goes even further. Those Justices find it “irrelevant” whether a voter ID law imposes a “special burden on some voters.”Footnote ²²⁷ As with so many of the other issues discussed in this book, state sovereignty was paramount. “That sort of detailed judicial supervision of the election process would flout the Constitution’s express commitment of the task to the States,” the three concurring Justices explained.Footnote ²²⁸ Once more, state sovereignty took precedence over political equality.

4 Requiring Documentary Proof of US Citizenship

Some states have attempted to go beyond voter ID requirements, additionally insisting on documentary proof that the voter is a US citizen. But in 2013 the Supreme Court in Arizona v. Inter Tribal Council of Arizona [ITCA]Footnote ²²⁹ substantially restricted (though it didn’t entirely foreclose) such efforts.

The ITCA case, discussed in Section B.1 above for its general role in the voter registration process, focuses on the effects of the “Federal Form” that the EAC requires for registering to vote in federal elections. That form reminds the voter, in distinctive red letters at the top of the form, that only US citizens may vote. To that end, the form requires the registrant to attest under penalty of perjury that he or she is a US citizen.Footnote ²³⁰ It does not require that the oath be supported by written documentation proving one’s citizenship; the declaration under oath is enough. Arizona, by statewide initiative in 2004, nonetheless passed a law that requires such documentary proof. The Supreme Court interpreted the relevant provision of the NVRA as establishing the exclusive requirements for voter registration; states could not insist on more documentation than the Federal Form requires. And the Constitution’s elections clause,Footnote ²³¹ the Court held, gives Congress the last word when it comes to the times, places, and manner of congressional elections.

But the Court left a potentially wide opening for proof of citizenship laws. As noted earlier, the “times, places, and manner” of congressional elections are not the same as voter qualifications, which are left to the states.Footnote ²³² Thus, Arizona can require US citizenship for voter registration; in fact, every state does, for all state and federal elections. And it can enforce that requirement. It’s just that if it wants to enforce it by requiring information beyond that required by the Federal Form, it has to follow the procedure specified by Congress. Under that procedure, the state must request the approval of the EAC. If the EAC denies the state’s request, the state may appeal the decision to the federal courts. So Arizona, joined by Kansas, requested EAC approval for adding documentary proof of US citizenship to their state-specific versions of the Federal Form. The EAC denied those requests, and in Kobach v. U.S. Election Assistance Commission,Footnote ²³³ the US Court of Appeals for the Tenth Circuit upheld the denials on the grounds that the proposed restrictions would violate both the NVRA and the Constitution’s equal protection clause.

In 2016, after ITCA and Kobach, three states – Alabama, Georgia, and (again) Kansas – all requested permission to amend the Federal Form to require documentary proof of citizenship for voting in their respective states. This time the EAC granted their requests, but in League of Women Voters v. Harrington,Footnote ²³⁴ the federal District Court for the District of Columbia struck down the EAC’s approvals. Under a provision of the Help America Vote Act (HAVA), the court noted, the Federal Form may contain only whatever information is “necessary” for election officials to assess voter eligibility or administer the election.Footnote ²³⁵ In this case, the EAC had approved the requests without even considering whether the states had made the required necessity showing. Indeed, one of the commissioners had inexplicably stated that proof of necessity was “irrelevant.”

This line of cases is a welcome development, but the danger posed by proof of citizenship laws has not passed. As the Supreme Court made clear, states may still request permission from the EAC to require documentary proof of citizenship. A federal Administration sympathetic to such a requirement, or to voting restrictions generally, might well grant such a request; as noted, the EAC granted three states’ requests, only to be reversed in the Harrington case. And even if the EAC were to deny the request, a sympathetic federal court could reverse the denial on appeal. Given the ease, described earlier, with which states can choose the specific federal forum – and in many cases even the specific judge – this is always a live possibility. Moreover, the NVRA applies only to federal elections. States always have the option of declining to use the Federal Form for elections to state legislative and other state offices, though their usual preference for conducting federal and state elections on the same days would make such a practice inefficient.

Indeed, despite the Supreme Court’s decision in ITCA and the subsequent lower court decisions in the Kobach and Harrington cases, Arizona in 2022 passed another law requiring documentary proof of US citizenship.Footnote ²³⁶ Its apparent hope is that the current Supreme Court will reverse course and approve the legislation. And as of May 15, 2023, bills to require documentary proof of citizenship were under active consideration in at least ten states.Footnote ²³⁷

In practice, laws that require documentary proof of US citizenship put voting beyond the reach of millions of eligible voters. A 2006 survey found that up to 7 percent of all US citizens lack ready access to documents that would prove their citizenship.Footnote ²³⁸ Between 2013 and 2016, the Kansas law that the court ultimately struck down had blocked approximately one out of every seven new voters (more than 14 percent) from registering or voting. Almost half of the disenfranchised voters were under age thirty,Footnote ²³⁹ a voting bloc that as noted earlier skews heavily Democratic.Footnote ²⁴⁰

But if you have to be a US citizen in order to vote, one might ask, what’s wrong with requiring you to prove it? And what is so hard about it, anyway?

For people born in the United States, by far the two most common documents for proving US citizenship are their birth certificates or, if they have them, passports. For obvious reasons, the poor are disproportionately unlikely to be able to afford either passports or the international travel that necessitates them. As of 2024, passports alone cost $165.Footnote ²⁴¹ They also take a lot of time. As of March 24, 2023, the average processing time was 10–13 weeks, plus several additional weeks for mailing the applications, having them accepted for processing, and mailing the passports.Footnote ²⁴² In addition, those who want passports in order to prove US citizenship are caught in a catch-22: You need proof of US citizenship to get the passport.

Birth certificates, fortunately, are a more widely available option. But they too are costly and typically take a long time to acquire, depending on one’s state of birth. The costs generally range between $20 and $55 and they require an average of 4–8 weeks to process.Footnote ²⁴³ Apart from the required expense, time, and energy involved, many Americans are unable to obtain birth certificates for the same reasons that they were born without them in the first place. Alfred Lubrano explains:

Married women who changed their surnames have an additional problem. For them, the original birth certificate will not be enough. Additional documentation to prove that the would-be voter is the same person described in the birth certificate will be required.

There is yet another effect of making documentary proof of US citizenship a requirement for voter registration. Organized voter registration drives are typically held in parks, shopping malls, athletic events, concerts, and political demonstrations sites, among other venues. People don’t ordinarily bring their passports or their birth certificates with them when they visit those places.

All these harms might at least be understandable, if not acceptable, were there important reasons for requiring that citizenship be proved through written documentation rather than by declarations under penalty of perjury. But the only justification offered by supporters of these laws tends to be the elimination of voter fraud. Previous discussions have documented the rarity of voter fraud generally and impersonation of eligible voters in particular.

Citizenship fraud in voting is rarer still,Footnote ²⁴⁵ and for obvious reasons. First, noncitizen voting in a federal election is (with limited exceptions) a federal crime punishable by fines and imprisonment for up to a year.Footnote ²⁴⁶ Also, since the Federal Form requires voters to attest under oath that they are US citizens, one who knowingly makes such a false claim is additionally guilty of perjury, a federal felony punishable by a fine and imprisonment of up to five years.Footnote ²⁴⁷ On top of that, both false claims of citizenshipFootnote ²⁴⁸ and unlawful votingFootnote ²⁴⁹ are independent grounds on which non-US citizens can be deported from the United States. Who would do this? The probability that one vote will change an electoral outcome is minute. For any noncitizen to take so huge a risk for so miniscule a potential gain would be beyond irrational.

5 Severely Curtailing Early and Mail Voting

Upon the founding of the American Republic, all voting in both federal and state elections took place in person on election day. Since then, the ways in which Americans vote have evolved steadily.Footnote ²⁵⁰ Today, two particular adaptations have taken root in a big way and are the subject of the present subsection – early voting and mail voting.

The terminology varies and can be confusing. This book uses the term “early voting” to mean any voting in which ballots are cast before election day. The voter might send the completed ballot by mail or deliver it in person at an officially designated polling place or drop box. As of March 23, 2023, the National Conference of State Legislatures reports that forty-six states (and DC, Puerto Rico, Guam, and the US Virgin Islands) offer early in-person voting to all eligible voters. Many of those states also allow early voting by mail, eight of them (and DC) conducting elections almost entirely by mail.Footnote ²⁵¹ The early voting periods vary, with an average starting date of twenty-seven days before Election Day and an average duration of twenty days.

“Mail voting,” as the term is used in this book, describes what many people think of as absentee voting. Elections officials mail the blank ballot to the voter, who then returns it either in person or by putting it in a mailbox or drop box. As of July 12, 2022, thirty-five states and DC allow any eligible voter to vote absentee. The other states offer that option only to those with specified excuses, and the list of qualifying excuses varies from state to state. In eight of the no-excuse-needed states, absentee ballots are automatically mailed to all registered voters; in the others, absentee ballots must be requested individually.Footnote ²⁵²

Regrettably, both compilations of state practices and statistical analyses of partisan voting patterns often lump these two voting methods together. But they are apples and oranges. What distinguishes early voting is simply the timing; what distinguishes absentee voting is the way the voter receives the blank ballot – by mail, as opposed to in person at the polling station.

The partisan effects of early voting and mail voting are not the same either. In recent elections, when early in-person votes and early mail votes are combined, early Democratic voters have consistently outnumbered early Republican voters. That was especially true for the 2018, 2020, and 2022 November elections.Footnote ²⁵³ In 2022, when early voting (driven mainly by COVID) accounted for approximately 45% of the total national vote,Footnote ²⁵⁴ the nonprofit organization TargetEarly estimated that among early voters Democrats outnumbered Republicans 51%–38%.Footnote ²⁵⁵

As elections expert Michael McDonald pointed out, it was not always that way. At one time, Republicans were more likely than Democrats to vote early; the pattern reversed only when President Trump warned his supporters not to trust early voting, urging them to cast their ballots on election day.Footnote ²⁵⁶

Again, however, these numbers fail to disaggregate early in-person voting from early mail voting. The partisan divide in mail voting is more extreme. As anyone who has stayed awake at night watching election returns is well aware, Democrats today are far more likely than Republicans to vote by mail. In 2020, when the pandemic prompted states to greatly expand opportunities for mail ballots, Biden supporters chose that option at twice the rate of Trump supporters. In some states, the ratios were even more dramatic – for example, 3-1 in North Carolina and almost 4-1 in Pennsylvania. Republicans, in contrast, are far more likely to vote in person, whether early or on election day.Footnote ²⁵⁷

To be clear, these percentages alone don’t necessarily prove that either generous early voting rules or generous mail voting rules favor Democrats. It might be that the number of early or mail-voting Democrats who would not have voted on Election Day if that had been their only option is no greater than the number of early or mail-voting Republicans for whom the same can be said. McDonald posits that early voting “activates people who are already high propensity voters. And so who are the high propensity voters? Well, by and large they tend to be more Republican than Democratic.”Footnote ²⁵⁸

Whether the first part of McDonald’s premise is correct is not clear. It might instead be the case that, all else equal, it is precisely the low-propensity voters for whom the ease of voting is more likely to be the decisive factor. Perhaps, in other words, the high-propensity voters would have been the more determined to find their way to the polls on Election Day if that had been their only option.

Regardless, as long as Republican strategists perceive that early and/or mail voting generally benefits Democrats, that belief can be a powerful incentive to pass laws that restrict those voting methods and other turnout-boosting measures, with the specific aim of making voting as hard as possible. In recent years, that perception has been evident. In 2013, for example, North Carolina’s Republican legislature requested and obtained data showing that in both 2008 and 2012 African American voters, who skew heavily Democratic, were far more prone to vote early than were white voters – especially during the first week of the early voting period. Immediately upon receiving those data, the legislature eliminated that first week, shortening the early voting period from seventeen days to seven.Footnote ²⁵⁹ In 2014, the Wisconsin legislature cut back the hours for early voting on weekdays and eliminated all early voting on weekends.Footnote ²⁶⁰ In 2023, Arkansas’s Republican legislature made it a criminal offense for election workers to send ballot applications – not even actual absentee ballots – to voters who had not requested them.Footnote ²⁶¹ And on the national level, the same Republican strategists who were campaigning to suppress voting by college students were strenuously advocating against early voting and mail voting.Footnote ²⁶²

One specific early voting issue – weekend voting – requires special mention. As of April 2023, twenty states offered Saturday voting, and seven others gave local election officials the discretion to do so. Eight states offered Sunday voting, and eight others gave local election officials the discretion to offer it.Footnote ²⁶³

Sunday voting is the real issue, as it holds both symbolic meaning and exceptional practical importance for African American communities. Part of a tradition that dates back to Reconstruction and Jim Crow, Sunday voting has continued through the modern civil rights era.Footnote ²⁶⁴ Known as “Souls to the Polls,” it remains a critical tool with which African Americans try to overcome voter suppression. In the words of Bishop Reginald Jackson of the African Methodist Episcopal Church, “[w]e gather in our churches on Sunday morning, you have morning worship and then after the service you get on the church buses, church vans, get in cars and people go to vote.”Footnote ²⁶⁵ Church official Christy Jackson adds that “[i]t became something that gave us comfort, to go in a group with our church group during those times to go and vote” and that “[c]hurches are especially important for mobilizing African Americans in rural counties,” because “[t]here are some polling locations, voting locations that are 10, 15, 20 miles from where a person lives. And so these central[ly] located churches, for example, then become what we call neighborhood hubs.”Footnote ²⁶⁶

Preparing for the 2012 elections, several Republican state legislatures and local elections officials restricted early voting periods, including targeting Sunday voting.Footnote ²⁶⁷ In 2013, North Carolina’s string of voter suppression measures included the elimination of one of the state’s two Sunday voting days.Footnote ²⁶⁸

Some of the more blatant actions took place in Georgia, again during the governorship of Brian Kemp. In February 2021, the Republican-controlled state House of Representatives passed a bill that would have eliminated Sunday voting entirely. Amidst a public outcry, the Republicans backed down, opting instead for a law that gives local elections officials the discretion whether to allow Sunday voting – but in no event more than two Sundays.

Even that action was just one element in a larger plan. “The GOP-led legislature also handed control of more election board appointments to conservative local judges or GOP-controlled county commissions in at least five counties.” This was a critical move, because in Georgia “local boards of elections oversee voter registration lists and certifications of elections and decide when voting can take place.” Exercising their newfound powers, election officials in at least seven Georgia counties eliminated Sunday voting in advance of the 2022 midterm elections. For further context, this occurred during the same election cycle in which Georgia’s Republican legislature “created new voter identification requirements for absentee ballots, empowered state officials to take over local elections boards, limited the use of ballot drop boxes and made it a crime to approach voters in line to give them food or water.”Footnote ²⁶⁹

It is hard to construct even a theoretical defense for these actions. Of course, any day in which staff must be available for early in-person voting entails some public expense, but the costs need not be any greater for Sunday voting than for weekday voting, except perhaps when there is a small add-on for overtime pay. Nor have opponents of Sunday voting even attempted to play the fraud card, as there is simply no reason to associate Sunday voting with fraud. About the most they have been able to come up with is that “Black church leaders influence their members on how to vote.”Footnote ²⁷⁰ But even assuming for the sake of argument that that is true,Footnote ²⁷¹ there is a name for that kind of influence: free speech. If church leaders or anyone else want to advocate for their beliefs, that is their right. And if congregants or anyone else want to hear those opinions, and choose to find them persuasive, that is their right.

The real purpose of laws and policies eliminating or restricting Sunday voting, often part of a broader package of voting restrictions, is surely more cynical. In the words of historian Rebecca Brenner Graham, “Bills aiming to eliminate Sunday voting have a transparent purpose: countering [Souls to the Polls] because African Americans vote overwhelmingly Democratic.”Footnote ²⁷²

One antidote – not a form of early voting, but a step that would at least mitigate the effects of restrictions on Sunday and other early voting – would be to make Election Day a national holiday.Footnote ²⁷³ As others have pointed out, this would enable people to vote without taking time off from work.Footnote ²⁷⁴ From time to time, bills have been introduced in Congress to do precisely that. One such bill, introduced on January 6, 2021 (Insurrection Day), would have made Election Day, in November of every even-numbered year, a national holiday for purpose of federal employment. The same bill would also have encouraged private employers to give their employees the day off.Footnote ²⁷⁵ But there were not enough Democratic votes to overcome staunch Republican opposition. Like its predecessors, the bill died.

6 Closing and Strategically Locating Election Day Polling Places

Several states have been closing polling places, in carefully selected locations, with lightning speed. Like so many of the other voter suppression strategies described in this section, the modern explosion of polling place closures traces back to the 2013 decision of the Supreme Court in Shelby County v. Holder.Footnote ²⁷⁶ As previously noted, that decision gutted the heart of the 1965 VRA, leaving the states and counties with histories of racial discrimination far freer to suppress the votes of racial minorities without the need for Justice Department preclearance. By far the most detailed, carefully documented, and comprehensive study of the link between Shelby County and polling place closures is a 2019 report by the Leadership Conference Education Fund (LCEF).Footnote ²⁷⁷ In this subsection, I draw liberally from that report.

The LCEF’s 2019 report, which updated and expanded on its analogous 2016 report,Footnote ²⁷⁸ studied polling place closures during the period 2014–18 by those states and counties that the Shelby County decision had freed from the VRA’s preclearance requirements.Footnote ²⁷⁹ In total, it found a decrease of 1,173 polling places in those counties during this period “despite a significant increase in voter turnout.”Footnote ²⁸⁰

In absolute numbers, the top three offenders were Texas (by far), Arizona, and Georgia. What the report described as “quieter efforts” to close polling places “without clear notice or justification” took place in Louisiana, Mississippi, Alabama, North Carolina, and Alaska (in that order). South Carolina, constrained by some specific state laws governing polling places, closed very few polling stations, relying instead on other post-Shelby County voter suppression strategies.Footnote ²⁸¹

Georgia once again deserves special mention. In terms of percentages of polling places closed, the top five counties in this study (having closed between 80% and 89% of their polling places), and seven of the top ten, were all in Georgia.Footnote ²⁸² It was there, you might recall, that the state official in charge of elections – Republican Secretary of State Brian Kemp – was running for governor and purging the voter rolls disproportionately in minority communities.Footnote ²⁸³ Kemp sent a memo to local elections officials encouraging them to “consolidate” polling stations and reminding them that in light of the Shelby County decision they no longer had to worry about Justice Department preclearance or even notification. And in the lead-up to his 2018 gubernatorial contest, Kemp’s recommended elections consultant, Mike Malone, “led an effort to close polling places in 10 counties with large Black populations. Malone told local boards of elections that Kemp had recommended polling place consolidation and sought [unsuccessfully] to close seven of nine polling places in Randolph County, which is 60 percent African American.”Footnote ²⁸⁴ In 200-square-mile Lanier County, which is 24 percent African American, only one of the four polling stations survived, even though its population had recently almost doubled.Footnote ²⁸⁵ The report adds: “In addition to five-hour lines, voters in communities of color faced countless obstacles on Election Day, including delayed polling place openings and broken voting machines.”Footnote ²⁸⁶

Similar examples, not always quite as blatant, occurred throughout the other former preclearance states. In absolute numbers, the data from Texas are particularly striking:

And then there is North Carolina. Elizabeth City is 52 percent African American but is situated in Pasquotank County, which is majority white. Citing cost concerns, the county closed four of the eight polling places in Elizabeth City – and none anywhere else in the county.Footnote ²⁸⁸

The effects of these closures have been stark. The LCEF report summarizes some of them:

As that summary explains, the mass closing of polling places does more than increase the time and cost of travel to the remaining stations. It also spells longer lines once you get there. Worse, those longer lines are not distributed equally. The data reveal a clear pattern of significantly longer waiting times for African American and Latino voters than for white voters. From a nationwide standpoint, part of that differential can be ascribed to the fact that African American population percentages happen to be especially high in the southeast, where the average wait times for all voters are generally higher than the national averages.Footnote ²⁹⁰ But even within the southeast, the wait time disparities can be huge, as the five-hour waits experienced by Georgia voters in communities of color illustrate. Closures in Arizona’s Maricopa County, where 31 percent of the population is Latino, have similarly resulted in five-hour lines.Footnote ²⁹¹

Admittedly, waits of that length are the exception, even in minority communities. Still, the empirical evidence now confirms that the average wait times correlate strongly with the percentage of nonwhite voters in a given precinct. Nationwide, in precincts where more than 90 percent of the voters are white, the average wait time in the 2018 elections was only 5.1 minutes. In contrast, in precincts where over 90 percent of the voters are nonwhite, that wait time increased to 32.4 minutes.Footnote ²⁹² The many additional studies cited by the Brennan Center for Justice have made similar findings.Footnote ²⁹³

Consistently with those findings, the Bipartisan Policy Center reports that in 2018 “[r]esidents of the most densely populated neighborhoods waited 25% longer than residents of the least densely populated neighborhoods.”Footnote ²⁹⁴ Recall that the most densely populated neighborhoods are those that tend to vote for Democrats by heavy margins.Footnote ²⁹⁵ Further, the average 2018 wait times in precincts in which the mean annual income was $40,000 or less was exactly twice the wait times in districts in which the mean annual income exceeded that amount.Footnote ²⁹⁶ That is not surprising, given the high positive correlation between racial minorities and average annual incomes.Footnote ²⁹⁷

All told, the Bipartisan Policy Center found, in 2018 “over 560,000 eligible voters failed to cast a ballot because of problems related to polling place management, including long lines.”Footnote ²⁹⁸ As noted, it is the poor, the urban, and racial minorities – particularly African Americans and Latinos – who are disproportionately burdened.

To be clear, there can be legitimate reasons to reduce the number of polling stations. Populations can shift. Election officials might have reasons to expect lower turnout rates for particular elections; midterm elections, for example, almost always generate lower turnout rates than those in presidential election years. An increase in the use of early and mail voting might be expected to reduce the total in-person Election Day turnout. And some states, especially Texas and Arizona, have made ambitious use of “vote centers,” where any eligible voters may cast their ballots regardless of which precincts they live in.

But there are far more closings, and far more selective closings, than these factors can explain. None of the examples noted earlier coincide with reductions in population. The lower turnout expectations for midterm elections are similarly a nonfactor. To the contrary, the LCEF Report, analyzing the data for the former preclearance states, found that 69 percent of the closures from 2012 to 2018 occurred only after the 2014 midterms.Footnote ²⁹⁹ Nor does the increased use of mail voting support the need to close polling places. As the Bipartisan Policy Center study shows, “Although the number of people voting by mail has been steadily increasing over the past two decades, 2018 also set a record for the number of votes cast in-person in a midterm election, 91.2 million. This was a 39 percent increase in the number of in-person ballots cast compared with the last midterm election in 2014.”Footnote ³⁰⁰

As for the vote centers in Texas and Arizona, the flexibility that they afford has been another stated justification for “massive reductions in polling places.”Footnote ³⁰¹ The LCEF reports:

Similarly, only 24 percent of Texas’s counties participate in the state’s vote center program but account for about two-thirds of the poll closings.Footnote ³⁰³

One might well ask whether voters would prefer voting at an assigned polling place in their neighborhoods or the “flexibility” of finding transportation to a distant vote center. Moreover, many of the Texas counties that don’t create vote centers have also closed the vast majority of their polling places. These include “Somervell (–80 percent), Loving (–75 percent), Stonewall (–75 percent), and Fisher (–60 percent) – all of which have large Latino populations.”Footnote ³⁰⁴

If none of these factors explains the massive number of polling station closures, then what does? There is ample room for cynicism here. As the foregoing data make clear, the closures were overwhelmingly in the states and counties that, by reason of their histories of racial discrimination in voting, had been subject to the VRA preclearance requirements. The numbers skyrocketed the moment the Supreme Court in Shelby County lifted that constraint. The cuts took place very disproportionately in heavily African American and Latino communities. And in several of those states – including Texas, Alabama, and North CarolinaFootnote ³⁰⁵ – the severe reductions in polling stations coincided with a range of other measures clearly aimed at depressing turnout in minority communities.

For all these reasons, there can be little doubt that at least a significant chunk of the racial disparities in polling station closures is the product of conscious racial discrimination. Moreover, whatever the explanation for the disparities, the very fact that racial minorities must systematically endure longer voting lines than whites is unacceptable in a democracy that purports to embrace political equality.

A related strategy is to prohibit otherwise eligible voters from voting, even provisionally, at other than their assigned precincts. Such a rule might initially seem reasonable from a management standpoint, but it has a sinister side. The US Court of Appeals for the Fourth Circuit, reviewing a package of voter restriction laws enacted by the North Carolina state legislature immediately after the Shelby County decision, had this to say:

But five years later, the Supreme Court in Brnovich v. Democratic National CommitteeFootnote ³⁰⁷ saw things differently. The Arizona legislature had passed a law requiring election officials to reject any Election Day votes not cast in the precinct in which the voter lived. The evidence at trial had demonstrated that Arizona changed its polling places with uncommon frequency, that minority voters on average lived further from their precinct polling places than other voters, and that, as a result of the confusion and the distances, the votes that were thrown out were disproportionately those of minority voters.

None of that mattered. Splitting 6-3, once again along strictly partisan lines, the Court upheld the restriction. It did so despite the explicit provision in Section 2 of the VRA that discriminatory “results,” occurring because members of a particular group lacked an equal “opportunity” to vote, would be a violation. In addition, the majority held, Arizona had a strong interest in preventing voter fraud. The fact that Arizona had not had any serious voter fraud problem also did not matter; for the majority Justices it was enough that, perhaps one day, such a problem might arise.

Polling place decisions have also targeted college students, another solidly Democratic voting bloc. In her speech at a Republican National Committee donor retreat in 2023, a top party strategist, Cleta Mitchell, called for Republicans to oppose locating polling stations on college campuses. She targeted campus voting sites in five states – Arizona, Georgia, Nevada, Virginia and Wisconsin. All have in common that they are swing states with huge public universities that enroll large in-state student populations.Footnote ³⁰⁸ And in Texas, a bill introduced in 2023 would prohibit all voting on college campuses.Footnote ³⁰⁹

Ironically, in the lead-up to the ratification of the Constitution, the antifederalists had expressed repeated fears that Congress would limit the number of polling states in areas unfavorable to the majority party.Footnote ³¹⁰ Once again, however, it is state legislatures – not the federal government – that are to blame.

7 Limiting Ballot Drop Boxes

Ballot drop boxes have become a popular alternative to both voting in person and mailing in a completed ballot. Voters may deposit their completed ballots in locked boxes, which are typically located in or outside city halls, public libraries, and/or various other state or local office buildings, including but not limited to elections board offices and polling places.Footnote ³¹¹

These ballot drop boxes bring a range of benefits. On the eve of the 2020 elections, when huge numbers of voters wanted to avoid in-person voting because of the COVID-19 pandemic, ballot drop boxes became a safe way to vote without health risks to oneself and without further spreading of the disease.Footnote ³¹² Otherwise, in the words of one observer, voters “could be forced to choose between their vote and their health.”Footnote ³¹³ For elderly and disabled voters, those considerations take on additional importance.

Beyond their public health benefits, ballot drop boxes make voting more convenient for many. As Democracy Docket’s Mac Brower points out, they “offer voters an accessible, 24/7 option for returning their ballot. Voting becomes as easy as filling out a ballot at home and depositing it in a secure box whenever the voter has time, a much more convenient process than going to a polling place or dropping it off at an election office during their limited hours of operation.”Footnote ³¹⁴

He adds: “Drop boxes are more convenient for election officials too, as they allow election offices to collect ballots directly from voters. They don’t need to worry about issues with the US Postal Service causing unforeseen delays in returning ballots.”Footnote ³¹⁵ Avoiding postal delays, it might be added, does more than help beleaguered elections officials; it reduces the burden on the US Postal Service as well.Footnote ³¹⁶ For similar reasons, drop boxes eliminate voters’ fears that the combination of strict absentee ballot deadlines and lengthy mail delivery times will prevent their ballots from arriving in time to be counted.Footnote ³¹⁷ Prompt arrival of ballots also means that the elections officials can complete their tabulations, and the American people can thus learn the results of the election, that much sooner.

With all these benefits, one would expect ballot drop boxes to be a no-brainer. Yet, throughout the nation, Republican legislatures, executive officials, and judges have made it their mission to ban or restrict them whenever they can. In Brower’s words, “What started as a Trump-led attack in 2020 has turned into an all-out Republican war on drop boxes, a cornerstone of their coordinated strategy to stop people from voting.”Footnote ³¹⁸

The campaign against drop boxes began in earnest with a 2020 tweet from then-President Trump, running for reelection. He wrote: “So now the Democrats are using Mail Drop Boxes, which are a voter security disaster. Among other things, they make it possible for a person to vote multiple times. Also, who controls them, are they placed in Republican or Democrat areas? They are not COVID sanitized. A big fraud!”Footnote ³¹⁹

Although Trump’s claims about drop boxes were false (as explained below) and in the pre-Elon Musk era Twitter flagged them as such, they remained on Twitter and became the gospel among many Republican voters and elected officials. In the run-up to the 2020 election, Ohio’s Republican Secretary of State issued an order prohibiting all off-site ballot drop boxes and limiting each county, regardless of population, to one on-site drop-off location. The federal district judge, a Clinton appointee, had enjoined the order, finding that the latter limit had a “disproportionate effect on people of color” in the more populous counties and thus significantly burdened their right to vote.Footnote ³²⁰ In Ohio’s Franklin County, where African Americans and Latinos together comprised 30% of the population, there was only one drop box for 1.3 million voters.Footnote ³²¹ “While it maybe said that the 7,903 registered voters in Noble County may find a single drop box location sufficient, the record demonstrates that the 858,041 registered voters in Cuyahoga County will likely not,” the district court wrote. But the all-Republican three-judge panel of the US Court of Appeals for the Sixth Circuit voted 2-1 to reverse the district court and uphold the restrictions.Footnote ³²²

During the same period, Texas’s Republican governor announced a similar plan to limit drop boxes to one per county, regardless of population differences. Harris County, with a population that was 64 percent African American and Latino, provided one drop box for its 4.7 million residents.Footnote ³²³ A three-judge panel of the US Court of Appeals for the Fifth Circuit – all three of them Trump appointees – upheld the governor’s order.Footnote ³²⁴

In some of the Republican-controlledFootnote ³²⁵ states, even one drop box per county was too much. At least three of those states (Mississippi, Tennessee, and Missouri) did not allow ballots to be dropped off in person at all, even at county offices. North Carolina, also under Republican control, allowed drop-offs at a county’s one elections office (which on average served 100,000 people) and was the only 2020 battleground state that provided no ballot drop boxes anywhere else. At least seven other Republican-controlled states similarly limited ballot drop-off locations to elections offices.Footnote ³²⁶

After the 2020 elections, these efforts metastasized. As of March 2023, at least eleven states have adopted complete bans on drop boxes.Footnote ³²⁷ The bans are ordinarily passed in the normal course by state legislatures, but in Wisconsin the state supreme court used its conservative majority to ban them via judicial fiat – extending the ban even to those drop boxes that are located inside the official offices of the county election clerks.Footnote ³²⁸

The stated justification for these bans is almost always that drop boxes are unacceptably vulnerable to fraud. President Trump’s 2020 tweet to that effect has already been discussed. That tweet and its progeny have since been echoed repeatedly in state legislative chambers that are under Republican Party control.Footnote ³²⁹

But the claims of widespread fraud in the use of drop boxes are just as empty as the claims of widespread fraud in the other voter suppression contexts. There are two points here: The security measures that attend the use of drop boxes make widespread fraud nearly impossible. And the evidence, as consistently reported by both nonpartisan elections officials and those with either Democratic or Republican Party affiliation, is that drop box fraud on a scale that remotely approaches what would be required to alter the outcome of an election simply has never occurred.

As to the security measures, ballot drop boxes are made of steel, bolted to the ground, often monitored by surveillance cameras or election workers, and often located in public buildings.Footnote ³³⁰ The Orange County [California] Registrar of Voters, described them this way: “They are very large, a thousand pounds, quarter-inch steel, sturdy boxes … They are designed with fire suppression systems as well as liquid damage protection. And so I’ll tell you what, they’re a lot more secure than a mailbox.”Footnote ³³¹ Moreover, the ballot signatures are verified in the same way as those on other ballots, and they are closely tracked as they work their way through the process.Footnote ³³²

The evidence confirms the absence of widespread voter fraud in the use of drop boxes. Brower describes their “proven track record in many states – including red ones.”Footnote ³³³ The Associated Press surveyed elections officials in every state and received responses from all but five states. Those responses “revealed no cases of fraud, vandalism or theft that could have affected the results.” … And “[n]one of the election offices in states that allowed the use of drop boxes in 2020 reported any instances in which the boxes were connected to voter fraud or stolen ballots. Likewise, none reported incidents in which the boxes or ballots were damaged to the extent that election results would have been affected [emphasis added].”Footnote ³³⁴ In a similar survey of all South Dakota county auditors, to which almost all replied, a majority reported using drop boxes in 2020 and again in 2022. According to the county auditors, there was not a single recorded case of voter fraud related to ballot drop boxes in either 2020 or 2022.Footnote ³³⁵

The real reason for banning or limiting ballot drop boxes, of course, is that Republican legislators and strategists believe – seemingly correctly – that Democratic voters are more likely than their Republican counterparts to use them. As the earlier discussion revealed, the impact of banning or restricting drop boxes falls particularly heavily on large African American, Latino, or other Democratic-leaning communities. In Texas, for example, where the one-box-per-county edict disproportionately harms voters in high-population counties, the burden falls especially heavily on voters in “Harris County (Houston), Travis County (Austin), [and] Fort Bend County (in suburban Houston), which are not only huge, they’re also … more diverse than the state … and in reality, far more blue than the rest of the state.”Footnote ³³⁶ The executive director of Common Cause Wisconsin explains the Republicans’ similar strategy for that state: “[M]ore Democrats vote by absentee ballot overall than Republicans, so they view it in their interests to be able to restrict how absentee ballots are returned. By getting rid of all the drop boxes, their calculation is that this will help them.”Footnote ³³⁷

Again, this is state government at work. And it is anathema to small-d democratic rule.

8 Restricting Third-Party Collection of Absentee Ballots

Collection and return of absentee ballots by individuals other than the voter has become another flash point in the running debate over how best to encourage voting without unduly sacrificing election integrity. Pejoratively called “ballot harvesting” by those who seek to cultivate an image of a practice riddled with fraud, third-party ballot collection is now the subject of sharp state-to-state variation.

As of March 2023, Ballotpedia reports that the states break down as follows: Twenty-four states and DC allow voters to entrust the delivery of their absentee ballots to anyone they choose. Fourteen states permit only specified individuals to return other voters’ ballots; typically, the eligible group includes some combination of family members, other household members, and caregivers. One state (Alabama) prohibits all third-party absentee ballot deliveries. The laws of the remaining eleven states do not specify which third parties, if any, may return ballots.Footnote ³³⁸

Some of the state laws restrict third-party absentee ballot delivery in other ways. In nine states, again as of March 2023, laws affirmatively prohibit certain specific individuals from returning other people’s ballots; the more common examples include employers, union agents, and campaign staff (including candidates). Twelve states restrict the number of ballots that any third-party may return. In seven states, only voters with disabilities, emergency circumstances, or other specified circumstances may entrust their ballot returns to third parties.Footnote ³³⁹

As with the other voter restriction measures discussed in this section, those who favor tight restrictions on third-party returns of absentee ballots invoke the specter of fraud and related misconduct. Unlike in those other contexts, however, these fears are not as easily dismissed. In the words of the Heritage Foundation:

To show that these worries are more than hypothetical, the Heritage Foundation cites two examples. Both involve mayors who were seeking reelection. Both mayors were criminally convicted of coercing local citizens to vote for them and, in one case, actually filling out the voters’ ballots.Footnote ³⁴¹ A third example, not cited by the Heritage Foundation, would be the earlier-discussed episode in which a Republican operative had generated fraudulent absentee ballots in a state legislative race, although in that case there was no evidence that the numbers involved could have tipped the election.Footnote ³⁴²

So the potential for abuse cannot be denied. Still, while the potential exists, examples of actual abuse remain rare.Footnote ³⁴³ Moreover, supporters argue, restricting ballot collection is ineffectual, because anyone intent on any of the feared violations – coercion, fraudulent tampering with the ballots, or discarding them – could avoid detection simply by depositing the ballots in a mailbox.Footnote ³⁴⁴

But the principal cost of these restrictions is, again, their disproportionate adverse impact on several minority voting populations. They particularly burden “communities that have traditionally relied on ballot collection, such as the elderly, differently-abled, Native American, and Latino communities, by preventing them from casting a ballot in the manner that is easiest for them.”Footnote ³⁴⁵ Increasingly, this method of voting is organized by black churches as well; African American churchgoers might either travel to post offices en masse to mail their ballots after Sunday services or allow church leaders to collect the ballots and mail them off in bulk.Footnote ³⁴⁶

The impact on Native American communities requires elaboration. As the ACLU points out,

Additional restrictions on third-party collection and delivery of absentee ballots might well be on the horizon, because in 2021 the Supreme Court gave at least one state the green light. In Brnovich v. Democratic National Committee,Footnote ³⁴⁸ the Court addressed two of Arizona’s voter restriction laws.Footnote ³⁴⁹ One of those laws makes it a criminal offense for anyone other than the voter or his or her family member, household member, or caregiver, or a postal worker or elections official, to collect or return the person’s ballot.

Splitting 6-3 along strictly partisan lines, the Court upheld these restrictions. The Arizona legislature had been well aware that the practice of ballot collection by community volunteers was an important way to make voting easier and that it was utilized disproportionately in minority communities. Additionally, the court of appeals had held that the record at trial should have required a finding of discriminatory intent. But the Supreme Court’s majority Justices disagreed. The majority also chose to give Section 2 of the VRA a narrow reading, despite the law’s explicit command that discriminatory “results,” occurring because members of a particular group lacked an equal “opportunity” to vote, would be a violation – whether or not there had been discriminatory intent. On its face the Arizona law applied equally to all voters, and for the Court’s majority that was enough. Besides, the majority held, Arizona had a strong interest in preventing voter fraud. The fact that Arizona had not had any serious voter fraud problem – let alone voter fraud relating specifically to third-party collection or delivery of absentee ballots – did not matter; again, it was enough that the problem might conceivably arise one day.

The correlation between Republican Party control of a state’s legislative process and the adoption of restrictions on third-party collection or delivery of absentee ballots is weaker than for many of the other voter restriction laws considered in this section. Some solid blue states – including California and Maryland – are also in the group that have restricted the practice.Footnote ³⁵⁰ Still, the vast majority of the restrictions are in solid red states,Footnote ³⁵¹ a reflection of the Republicans’ probably accurate perceptions that the minority populations most dependent on the assistance of third parties skew strongly Democratic. And again, in those states the mere possibility of a small number of fraudulent ballots has taken precedence over the far greater number of eligible minority voters effectively disenfranchised.

9 Disenfranchising Citizens Convicted of Crimes

Denying the vote to citizens who have been convicted of particular crimes is neither a new, nor a distinctively American, phenomenon. It was imported from England.Footnote ³⁵² In the US, several states adopted the practice right from the outset; over time, other states followed suit. The trend began to reverse in the late twentieth century, when a number of states loosened the restrictions, typically by reducing the length of the ineligibility. In part, as Alexander Keyssar observes, these relaxations were a reaction to the extraordinary, and growing, size of the US prison population: “Between 1972 and 2003, the number of persons in prison or jail [in the U.S.] increased tenfold, from 200,000 to more than two million.”Footnote ³⁵³ In combination with the number who had already been released, the result was that in the year 2000, some 4.7 million US citizens were rendered ineligible to vote because of criminal convictions.Footnote ³⁵⁴

Despite the liberalizing trend in several states, the prison population continued to grow. By 2008, the number of citizens disenfranchised because of criminal convictions had soared to a record 5.3 million.Footnote ³⁵⁵ Since then, the number has pretty much plateaued; as of 2020, it was estimated to be roughly 5.2 million.Footnote ³⁵⁶

Today, every state except Maine and Vermont disenfranchises citizens who have been convicted of specified categories of crimes.Footnote ³⁵⁷ The specifics vary widely. The two main variables are which crimes trigger disenfranchisement and when, if ever, the voting rights are restored.

As of June 2022, the main categories of qualifying crimes include the following: any felony; any felony or election-related crime; any felony or election-related crime or treason; any felony under the laws of the disenfranchising state (as opposed to convictions in federal court or other states’ courts); and any felony, or any misdemeanor that leads to any incarceration.

Depending on the state law, the right to vote might be restored upon release from incarceration (either automatically or upon the citizen’s application); upon release followed by completion of all probation or parole requirements; upon release followed by completion of all probation or parole requirements and payment of all required fines, fees, or victim restitution; only two years after both release and the completion of all probation or parole requirements; upon release followed by the earlier of five years or completion of all probation or parole requirements; or never (depending on the particular felony and sometimes subject to a gubernatorial power to dispense relief). In some states, the length of the disenfranchisement depends on such variables as the number of felonies, whether any of them were violent, and the dates and places of the convictions.Footnote ³⁵⁸

Four states – Florida, Tennessee, Alabama, and Arizona – deny convicted felons the right to vote, even after release from incarceration and completion of all probation or parole requirements, until they have also repaid all fines and fees. Tennessee additionally conditions the restoration of voting rights on payment of any legally required child support. Some have compared the practical effects of such financial conditions to those of now-prohibited poll taxes.Footnote ³⁵⁹

The impacts of these laws have been borne grossly disproportionately by African American men. Although African Americans comprised only 12.9 percent of the US population in 2000,Footnote ³⁶⁰ “[a]lmost half of the prison and parole populations were African American; … In some states more than 15% of adult African American men were disenfranchised.”Footnote ³⁶¹ By 2020, there were at least seven states (all Republican-controlled) in which more than one out of every seven African American adults was disenfranchised.Footnote ³⁶²

The disproportionate suppression of the African American vote has not always been a mere side effect of laws that add disenfranchisement to the penalty for commission of a crime; too often, the special impact on African Americans has been the very point. As Jeanne Chung has noted, “[i]n the post-Reconstruction period, several Southern states tailored their disenfranchisement laws in order to bar Black male voters, targeting those offenses believed to be committed most frequently by the Black population.”Footnote ³⁶³ Keyssar points out that in the late nineteenth century “[c]riminal exclusion laws … were altered to disenfranchise men convicted of minor offenses, such as vagrancy and bigamy. … The overarching aim of such restrictions, usually undisguised, was to keep poor and illiterate blacks – and in Texas, Mexican Americans – from the polls.”Footnote ³⁶⁴ That trend continued in the early twentieth century, when southern states passed additional restrictions that “were often more detailed and included lesser offenses, targeting minor violations of the law that could be invoked to disenfranchise African Americans.”Footnote ³⁶⁵

In this century, as with the other voter restrictions discussed in this section, crass partisanship continues to drive criminal disenfranchisement laws. In Keyssar’s words, “[b]oth Democrats and Republicans assumed, probably correctly, that most ex-felons – coming from poor or working-class backgrounds and being disproportionately African American – would vote Democratic. Many Republicans were consequently reluctant to support legislation that could hurt their own electoral fortunes, and they regarded Democratic support for liberalization as transparently partisan.”Footnote ³⁶⁶

Are racism and partisanship the only motives for criminal disenfranchisement, or are there meritorious arguments for it? Keyssar notes persuasively that the traditional arguments for criminal punishment lend no support to criminal disenfranchisement:

I would add that it obviously does not serve the additional criminal function of incapacitation – isolating the offender from society.

My view is that, if a legislature considers the criminal punishment allowed by law sufficient to achieve all the legitimate penal functions, then adding disenfranchisement is superfluous. And if it thinks the existing criminal penalties are insufficient for any of those purposes, it has a much more tailored remedy – increase the penalty for the particular crime. Either way, disenfranchisement serves none of the traditional goals of criminal justice.

Disenfranchisement, of course, is not the only civil disability that attaches to criminal convictions. People convicted of various crimes might be barred from carrying firearms, or excluded from particular occupations, or (if not US nationals) removed from the United States. The usual justification for civil disabilities is that the crime of which the person was convicted makes them unsuitable for the particular function. We don’t want people convicted of violent crimes to walk around with guns. We don’t want people convicted of drunk driving to pilot airplanes. We don’t want people convicted of assault with a deadly weapon to become police officers. But no otherwise eligible voter – not even one who has been convicted of a serious crime – endangers anyone by voting.

So what, exactly, is the harm that we fear their voting would cause? Historically, one argument offered by defenders of criminal disenfranchisement is that ex-felons might vote for prosecutors or other officials who were perceived as soft on crime, or even vote to repeal the criminal laws entirely.Footnote ³⁶⁸ Apart from the lack of any empirical evidence to support those assumptions, there is nothing inappropriate about a person choosing to vote his or her self-interest. In a democracy, you get to do that.

That leaves us with what appears to have been the favored justification offered for criminal disenfranchisement in the late nineteenth century – to “preserve the purity of the ballot box.” Letting ex-felons vote, the argument went, “would corrupt the electoral process.”Footnote ³⁶⁹

One might bristle at the brazenness of legislators who have no inhibitions about enacting a whole range of partisan-inspired voter suppression laws daring to invoke the purity of the ballot box in the first place. Regardless, a single wrongful act does not conclusively define a person’s moral character. Perhaps most importantly, in a democracy, the right to vote should be seen as a fundamental attribute of citizenship, not a privilege dependent on legislators’ speculative judgments about the intrinsic morality of their constituents.

1 Electing Congress

Article I, Section 4, Clause 1 of the US Constitution (commonly referred to as the “elections clause”) reads: “The Times, Places and Manner of holding Elections for Senators and Representatives, shall be prescribed in each State by the Legislature thereof; but the Congress may at any time by Law make or alter such Regulations, except as to the Places of chusing Senators …” [emphasis added]. This clause has given rise to a legal argument known as the “independent state legislature theory” (the ISLT). In its pure form, the ISLT posits that the phrase “the legislature” means “only the legislature.” Under that theory, the power that the elections clause confers on a state legislature is not constrained by the state’s constitution, the state’s courts, the state’s governor, the state’s voters, or any other organ of the state.

The ISLT has never gained a solid foothold in either federal or state courts. But it has become the Harold StassenFootnote ³⁷⁰ of the electoral distortion movement, popping up with astonishing regularity only to be shot down every time before being trotted out yet again. And despite an important 2023 Supreme Court decision that will be discussed presently,Footnote ³⁷¹ the ISLT remains a major threat to the integrity of congressional, as well as presidential, elections.

The ISLT made its first major appearance in 1916. The Ohio state constitution authorized statewide popular referenda approving or disapproving laws passed by the legislature. When the voters passed such a referendum disapproving their state legislature’s congressional districting map, aggrieved citizens asked the courts to disregard the referendum. They argued that the elections clause of the US Constitution gives the state legislature the sole authority to draw congressional districting maps. The Ohio Supreme Court disagreed, holding that the legislature’s power under the elections clause is subject to the state constitution and that the referendum results would therefore stand. The US Supreme Court affirmed unanimously.Footnote ³⁷²

The Court was similarly unanimous in 1932, when it decided Smiley v. Holm.Footnote ³⁷³ There, the Court upheld a gubernatorial veto of the legislature’s congressional districting map. The state constitution authorized the governor to veto legislation and the Court again held that the legislature’s exercise of its elections clause power was subject to the state constitution.

More recently, the Court held yet again that the state legislatures’ powers under the elections clause are subject to the limitations in their states’ constitutions. In Arizona State Legislature v. Arizona Independent Redistricting Commission,Footnote ³⁷⁴ the challenge was to a state constitutional provision that transferred the congressional redistricting function to an independent commission. But by this time the Court had become more conservative and far more partisan, with the result that the state constitutional provision survived only by a tenuous margin of 5-4. All four dissenters were Republican appointees.

Support for the ISLT has not been limited to gerrymandering issues. In Republican Party v. Degraffenreid,Footnote ³⁷⁵ a Pennsylvania state statute allowed mail-in ballots for the 2020 elections, but only if they arrived by 8:00 pm on Election Day. In light of the COVID-19 pandemic and the accompanying long postal delays that were beyond the voters’ control, the state supreme court understood that that time limit would result in significant numbers of even timely mailed ballots arriving too late to be counted. It thus interpreted the state constitution’s requirement of “free and equal” elections to require a more realistic receipt date. For that purpose, the court extended the deadline for the ballots’ arrival for three days. The Republican Party petitioned the US Supreme Court to review the decision, and the Court declined.

But three Justices dissented. Justice Thomas hinted at support for the ISLT;Footnote ³⁷⁶ Justice Alito, in an opinion joined by Justice Gorsuch, endorsed it more fully, though not by name. Justice Alito wrote

Justice Alito’s dismissive language, especially the term “simply by claiming,” minimizes what happened in this case. The state supreme court had carefully explained why, under the unusual conditions created by COVID, failing to count ballots that were likely delayed by postal logjams would violate the state constitution’s requirement of “free and equal elections.”

That brings us to the Supreme Court’s landmark 2023 decision in Moore v. Harper.Footnote ³⁷⁸ As was discussed in Section A, the North Carolina Supreme Court – with Democrats then holding a 4-3 majority – had struck down the legislature’s congressional districting map and had approved a new map, affirming the trial court’s finding of intentional, extreme, partisan gerrymandering. The Republican Speaker of the North Carolina House and others asked the US Supreme Court to invalidate that remedial map.Footnote ³⁷⁹ Their argument rested on the independent state legislature theory. The elections clause, they maintained, should be interpreted as insulating state legislatures’ congressional districting plans from all constraints otherwise imposed by state law, including state constitutions. Since the US Supreme Court in Rucho had barred federal court review of partisan gerrymandering claims just four years earlier,Footnote ³⁸⁰ embracing the state legislature’s argument would have meant that partisan gerrymandering could not be constrained by any courts at all.

The Supreme Court declined to overrule its previous decisions rejecting the ISLT. By a 6-3 vote, it rejected the theory once again – in its pure form. Quoting from the Smiley case, the Court held that “[a] state legislature may not ‘create congressional districts independently of [requirements imposed] by the state constitution with respect to the enactment of laws.’”Footnote ³⁸¹ Quoting historical sources, the Court relied heavily on the framers’ understanding that state legislatures “are the mere creatures of the State Constitutions. … [T]hey owe their existence to the Constitution: they derive their powers from the Constitution: It is their commission; and, therefore, all their acts must be conformable to it, or else they will be void.”Footnote ³⁸²

So far, so good. For now, the Supreme Court has contained the fire. And that is a big deal. But rather than extinguish the blaze completely, the Court has left the embers burning. It is one thing, the Court essentially says, to hold state legislatures to the limitations imposed by their state constitutions. But that, the Court explains, doesn’t mean the legislatures are limited by their state supreme courts’ interpretations of their state constitutions. In the Court’s words, “Although we conclude that the Elections Clause does not exempt state legislatures from the ordinary constraints imposed by state law, state courts do not have free rein.” It then cites Chief Justice Rehnquist’s concurring opinion in Bush v. Gore for the proposition that the Court may disapprove state supreme court interpretations of state law if the latter “distorted” the state law “beyond what a fair reading required.” Later, it adds: “We hold only that state courts may not transgress the ordinary bounds of judicial review such that they arrogate to themselves the power vested in state legislatures to regulate federal elections.” Finally: “[F]ederal courts must not abandon their own duty to exercise judicial review. In interpreting state law in this area, state courts may not so exceed the bounds of ordinary judicial review as to unconstitutionally intrude upon the role specifically reserved to state legislatures by [the Elections Clause].”Footnote ³⁸³

At least at this writing, no one knows what that means. What are the “ordinary” constraints of state law? What are the “ordinary bounds of judicial review?” If states don’t have “free” rein, how much rein do they have? What will it take for federal judges to decide that they not only disagree with a state supreme court’s interpretation of its state’s laws, but think the state court “distorted” its state’s law or that its reading of state law wasn’t “fair” or that the state court “arrogated” the state legislature’s power under the elections clause or that the state court “unconstitutionally intruded” on the legislature’s elections law powers?

Readers of the Court’s opinion waited in vain for answers to these questions. The Court declined to provide them: “We do not adopt these or any other test by which we can measure state court interpretations of state law in cases implicating the Elections Clause.” That is because “[t]he questions presented in this area are complex and context specific.”Footnote ³⁸⁴

Perhaps, upon reaching this stage of the Court’s opinion, some readers assumed that at least limited guidance would be provided once they got to the part where the Court tells us whether the North Carolina Supreme Court’s interpretation of its state’s law in the present case passes muster. But no: “We decline to address whether the North Carolina Supreme Court strayed beyond the limits derived from the Elections Clause. The legislative defendants did not meaningfully present the issue …”Footnote ³⁸⁵

Justice Kavanaugh joined the majority opinion and wrote a separate concurrence to consider what standards future federal courts might employ in deciding how far is too far. Ultimately, he expressed a preference for the Rehnquist standard described above.

When the Supreme Court tells us this many times in the same opinion that federal courts may still reject state supreme courts’ interpretations of their own states’ constitutions when reviewing their legislatures’ congressional districting maps, we have to assume the Court really means it. And Chief Justice Roberts, the author of the majority opinion, is anything but sloppy. He has proven quite adept at giving conspicuously with one hand while taking back subtly with the other. In this case, I find the theory of democracy expert Richard Pildes highly plausible. His take is “that a majority was cobbled together among conservative and liberal justices by agreeing to decide this part of the case in the narrowest terms.” The result, as Pildes points out, is simply “a weaker version” of the ISLT, one that he fears will infect the 2024 elections.Footnote ³⁸⁶ I will term this tamed-down version “ISLT-lite.”

The US Supreme Court unquestionably has the power, and arguably the responsibility, to interpret the elections clause of the US Constitution. It does so here, rejecting the argument that the elections clause insulates state legislatures from the constraints imposed by their state constitutions. Once the Court has decided that federal constitutional issue, there is only one remaining question: Does the North Carolina constitution empower the state supreme court to review claims that the legislature’s districting map reflects impermissible partisan gerrymandering? And that is purely a question of state law. I am at a loss to understand what business any federal court, including the Supreme Court, has in second-guessing a state supreme court’s interpretation of its state’s law. More perplexing still is that, as in Bush v. Gore, this same usurpation of a state supreme court’s authority is endorsed by three of the Justices who consistently trumpet their fidelity to federalism and states’ rights.

And yet, for at least two of the dissenting Justices, even ISLT-lite was not enough. Justice Thomas, in an opinion joined by Justice Gorsuch, dissented primarily because he felt the case was moot, but secondly because he disagreed on the merits. As to the latter, these two Justices adopted the extreme version of the ISLT, acknowledging only those state supreme court decisions that interpret the word “legislature.” The decisions in cases like Hildebrant and Smiley, discussed earlier, tied their hands on that score, since those decisions had recognized popular referenda and gubernatorial vetoes as part of the legislative process. But state constitutions, they maintained, cannot place any substantive limitations on the scope of state legislatures’ powers under the elections clause.Footnote ³⁸⁷

Interestingly, Justice Alito joined only the mootness section of the Thomas dissent, not the ISLT portion of that opinion. But there is no doubt where he stands. His embrace of a full-throated ISLT in dissenting from the denial of certiorari in Degraffenreid has already been noted. And when the Republican state legislator plaintiffs sought an emergency stay of the state court-approved remedial districting map at issue in this very case, Justice Alito invoked the substance of the ISLT once again, dissenting from the majority’s refusal to grant the requested stay.Footnote ³⁸⁸

How much damage can ISLT-lite do? Quite a bit, as it turns out. Earlier discussion asked what terms like “fair reading” and “exceed the bounds of ordinary judicial review” mean. But that was a rhetorical question, for every Supreme Court scholar knows the answer. It means whatever five members of the Court say it means. Given the propensities of the current Justices, one can hardly be confident that the Court will constrain state legislatures when they enact various measures testing the limits of the Supreme Court’s willingness to overturn state supreme court rulings on their legislatures’ exercise of their election clause powers. For that matter, there is no way to know the ultimate fate of even the North Carolina legislature’s extreme partisan gerrymander, much less similar efforts by other states, because the Court declined to say.

There are additional reasons to fear that we are not yet out of the woods. With the current more solidly conservative, more Republican, and more activist Supreme Court, Justices who embrace even the extreme version of the once-unthinkable independent state legislature theory can no longer be laughed off. The Court’s current Republican majority has not hesitated to overrule precedent when doing so furthers its politically conservative views or the electoral welfare of the Republican Party, as previous discussion has already demonstrated.Footnote ³⁸⁹ Three of the current Justices have already signed on to a potential absolutist version of the independent state legislature theory. Overruling Moore would require just two more votes. They could come from new appointments. Or they could come from either a change of mind or a specific application of the Court’s amorphous restrictions on state supreme court interpretation powers by the Chief Justice or Justices Kavanaugh or Barrett, or some combination of both.

Finally, not every case reaches the Supreme Court. And even those that do will not always reach the Supreme Court in time for its ruling to affect a pending election. Federal district and court of appeals judges whose mindsets are similar to those of the three dissenters in Moore – and as previous discussion showed, their ranks are plentiful – have many ways to circumvent the holding in Moore. They can interpret elastic terms like “ordinary” and “fair” creatively. Or they could slow the process enough to ensure that extreme gerrymandering or voter suppression laws are not struck down until after a pending election, knowing that undoing an election after the fact is exceptionally difficult. As the same previous discussion also showed, the easy availability of judicial forum shopping – and of dependable lower court judges willing to shape their legal analyses to fit their own ideological agendas – can give cynically motivated state legislatures the cover they need.

Leaving any state government officials free to disregard the constraints of their state constitutions and the legal interpretations of their courts would be dangerous enough. Here, though, an earlier point is worth repeating: Despite public perceptions, state legislatures are often profoundly counter-majoritarian to begin with – in the words of Miriam Seifter, “the least majoritarian branch” of state government.Footnote ³⁹⁰

2 Electing the President

Might the independent state legislature theory apply to presidential elections? Article II, Section 1 of the US Constitution (the “electors clause”) reads:

The differences between this provision and the analogous provision governing congressional elections are notable. As the preceding subsection points out, state legislatures decide not only the manner, but also the times and places, of congressional elections, but Congress may supersede all of those choices. When it comes to presidential elections, only the state legislatures decide the manner of selecting presidential electors, and only Congress decides the date on which those electors must be chosen.

In the early years of the republic, several state legislatures chose their presidential electors themselves, rather than leave the decision to the people.Footnote ³⁹¹ But for at least the past 150 years, the laws of every state provide for presidential electors to be chosen by popular vote.Footnote ³⁹²

Could state legislators today resume the historical practice of naming their own preferred presidential electors instead of letting their constituents decide? The thought is disconcerting, and the question is an open one.

The Supreme Court’s 1892 decision in McPherson v. Blacker takes a broad view of the state legislatures’ powers under the electors clause. Upholding the Michigan legislature’s decision to allocate its electoral votes via district-by-district elections, the Supreme Court described the legislature’s power in sweeping terms. The power to decide the manner in which the state’s presidential electors are chosen, the Court said,

Two points stand out. First, in contrast to the Court’s analogous congressional election decisions, the Court says that the legislature’s power to decide the manner of choosing presidential electors is not subject to the state’s constitution. Second, the Court declares, even when a state statute or state constitution provides for popular election of the presidential electors, the legislature has the right “to resume the power at any time” – meaning that the legislature can change the law and decide to name the presidential electors itself, regardless of what the state constitution says.

To be clear, both those statements are nonbinding dicta, for neither issue was before the Court in McPherson. There was no state constitutional provision allegedly prohibiting the legislature’s districting plan; the challenge had been grounded in provisions of the US Constitution. Nor had the legislature attempted to substitute its own electors for those chosen by the people.

Nonetheless, the language in McPherson has had an impact. The Congressional Research Service concluded that “state legislatures still retain the constitutional option of taking [the election of presidential electors] out of the voters’ hands, and selecting electors by some other, less democratic means.” That very option, in fact, was “discussed in Florida in 2000 during the postelection recounts, when some members of the legislature proposed to convene in special session and award the state’s electoral votes, regardless of who won the popular contest in the state.”Footnote ³⁹⁴

That radical step proved unnecessary. As discussed earlier, a 5-4 majority of the Supreme Court in Bush v. Gore awarded Florida’s electoral votes, and with them the presidency, to Republican nominee George W. Bush. The five Justices in the majority (all Republicans) ultimately decided that the differing standards employed in various Florida counties during the recount violated the equal protection clause of the US Constitution.

But in that case, Chief Justice Rehnquist wrote a separate concurring opinion, which was joined by two of the other majority Justices (Scalia and Thomas). In it, these three Justices offered an additional reason to overturn the decision of the Florida Supreme Court. Ordinarily, federal courts defer to state courts’ interpretations of their states’ laws. But because the US Constitution’s presidential election clause left the manner of choosing the electors to the state legislatures, these three Justices believed that the Florida Supreme Court’s interpretation of laws on that subject was not entitled to the usual deference. These Justices didn’t specifically say that the state legislature is immune from any constraints imposed by the state constitution. Still, these Justices’ explicit refusal to defer to the state supreme court – which is typically, also, the ultimate arbiter of what the state constitution means – comes perilously close to endorsing the ISLT and extending it to the presidential realm.Footnote ³⁹⁵

Moreover, the majority – not just the three concurring Justices – took pains to emphasize the teachings of McPherson on this subject. “The individual citizen has no federal constitutional right to vote for electors for the President of the United States unless and until the state legislature chooses a statewide election as the means to implement its power to appoint members of the Electoral College [citing article II, section 1].” Citing McPherson, the Court added that

Dissatisfied with the results of the 2020 presidential election, “the Trump campaign was seriously considering asking several of the Republican-controlled state legislatures to select Trump electors.”Footnote ³⁹⁷ Trump campaign lawyer John Eastman, in fact, discussed precisely that strategy with Pennsylvania Republicans.Footnote ³⁹⁸

There is at least one fairly clear legal limit on such a strategy – timing. It is true that the Supreme Court in both McPherson and Bush v. Gore opined, albeit only in nonbinding dictum, that the US Constitution would permit state legislatures to reassert the power to select the presidential electors rather than let the voters decide. But Congress retains exclusive authority over the time when the electors are chosen. And it has chosen the first Tuesday after the first Monday of November in even-numbered years. Thus, in the minds of some, legislatures could not constitutionally substitute their own electors for the ones that the voters had selected on Election Day. By then, it would be too late.Footnote ³⁹⁹

I’m not so sure. It does seem clear that state legislators could not take such action after Election Day. Nowadays, however, in the vast majority of states, the major television networks can project the winners of the electoral votes on election night. And even without a formal projection, state legislators might feel that the results are not looking good or simply that the outcome looks too close to leave to chance. In any of those scenarios, what legal barrier would there be to the legislature convening before midnight on election night, making any necessary rules changes to permit immediate action, amending its statute to allow legislative selection of its state’s presidential electors (this assumes the governor is on board), and then choosing those electors? With that strategy, the legislature might argue that it is complying with the requirement that it choose its electors on the date designated by Congress.

My hope is that if such a strategy were attempted, even the current Supreme Court would reject the ISLT in much the same way that it rejected its application to congressional elections in Harper. The Court would simply need to read the Constitution’s electors clause as assuming that the state legislatures, being creatures of the state constitution, are constrained by it. If the right of the people to select presidential electors is part of the state’s constitution, and not merely statutory, this would not be a heavy lift. There are judicial dicta to the contrary, but no binding holdings.

Still, the same uncertainties exist. Three members of the current court are already on record as endorsing the ISLT; forming a majority would take only two more votes, from either new appointments or changes of minds by existing members. When the presidency hangs in the balance, Bush v. Gore teaches us that Republican Supreme Court Justices’ allegiance to states’ rights can be quite flexible.

The greater constraint, many others have argued, would be strong political headwinds. One can imagine the outrage were a state legislature to reject the votes of the people on an issue so important and so emotionally absorbing.Footnote ⁴⁰⁰ Then again, partisan polarization being as acute as it now is, one hesitates to make even that assumption.