Chapter 36 — Voting Rights: The Long Struggle and Current Battles

Why this chapter matters

Of all the chapters in this book, this is the one where the gap between rhetoric and evidence is widest. On voting rights, you can read two articles published the same morning, in serious outlets, that present mutually incompatible pictures of American democracy. One says voter suppression is the gravest threat to the republic since Jim Crow. The other says claims of suppression are partisan invention designed to delegitimize lawful election rules. Both sides will accuse the other of bad faith. Both sides will accuse the courts of being captured. Both sides will accuse the press of being in the tank.

The honest answer is harder than either polemic. The American franchise has expanded enormously over two and a half centuries, from a property-restricted franchise of perhaps six percent of the population to an inclusive franchise that, in 2024, registered the highest raw-vote total in U.S. history. Within that long arc, the period from 1965 to roughly 2010 was one of relatively bipartisan voting-rights consensus, anchored by the Voting Rights Act and its periodic reauthorizations. The period from 2010 to the present is one of bitter partisan division about election rules. The empirical evidence on whether recent rule changes have meaningfully suppressed turnout is genuinely contested — not because the science is broken, but because the effects are real but smaller than activist rhetoric on either side suggests, and because the targets and motives of the changes are themselves part of what the evidence has to assess.

This chapter walks through that history, presents the empirical evidence honestly (including where the evidence cuts against the position of the side closer to the institutional voice of academic political science), and refuses to flatten a complicated picture into a slogan. We also state plainly something the evidence requires: the 2020 presidential election was not stolen. That is not a contested claim within the political-science literature or the courts. It is a finding. The political-cultural movement that holds otherwise is empirically wrong. We will explain what happened, what the courts ruled, what Republican election officials in swing states actually did, and why the persistence of the contrary belief is a fact about American political culture that this chapter has to describe even when describing it is uncomfortable.

Section 1: The historical arc — from six percent to universal

The original franchise

When the Constitution was ratified in 1788, the right to vote was not federally guaranteed. Article I, Section 2 specified that members of the House of Representatives would be chosen "by the People of the several States, and the Electors in each State shall have the Qualifications requisite for Electors of the most numerous Branch of the State Legislature." In plain English: the states decided who voted, using whatever rules they used for their own legislative elections.

In practice, those rules excluded most adults. The franchise typically required ownership of real property, sometimes a minimum amount of personal wealth, and almost always being a free white male. Estimates of the size of the eligible electorate at the founding range from roughly six percent of the total population (the most common estimate for the 1790 census era) to perhaps eight percent if you count the most expansive state rules. The total population in 1790 was about 3.9 million; the voting electorate was on the order of 250,000 to 320,000. Slaves (about 700,000 in the 1790 census) could not vote. Free Black men could vote in some Northern states, but not most. Women could not vote anywhere in the new United States by 1807, when New Jersey closed the small loophole that had briefly allowed propertied women to vote in that state.

This was not an oversight. Most of the Founders openly believed that the franchise should be limited to those with a "stake in society," which they understood to mean property holders, who would be more deliberative and less susceptible to demagoguery. The view was held by Federalists and many Democratic-Republicans alike. Jefferson in some letters argued for broader yeoman-farmer suffrage, but yeoman farmers were still men, still white, and usually still owned land.

The expansion of white male suffrage, 1820s–1850s

Property requirements eroded across the early Republic, mostly at the state level. The new states admitted between 1815 and 1845 — Indiana, Illinois, Mississippi, Alabama, Maine, Missouri, Arkansas, Michigan, Florida, Texas, Iowa, Wisconsin — almost universally entered the Union with universal-white-manhood-suffrage constitutions. The older states followed in waves: New York eliminated its property requirement in 1821 (replacing it with a small taxpaying requirement for some races), Virginia held out until 1850, North Carolina until 1856.

By the eve of the Civil War, "universal white manhood suffrage" was the U.S. norm. Roughly 80% of adult white men could vote. The voting age was 21. Free Black men in most Northern states could not vote, even when the state had eliminated property requirements for whites — the Pennsylvania Constitution of 1838, for example, expanded the franchise for white men while explicitly disenfranchising Black men. The trajectory was simultaneously democratizing along a class axis and tightening along a racial axis.

The Reconstruction Amendments and their unraveling

The Civil War, the Thirteenth Amendment (abolishing slavery, 1865), the Fourteenth Amendment (citizenship and equal protection, 1868), and the Fifteenth Amendment (1870) transformed the constitutional structure of voting rights. The Fifteenth Amendment is short:

Section 1. The right of citizens of the United States to vote shall not be denied or abridged by the United States or by any State on account of race, color, or previous condition of servitude.

Section 2. The Congress shall have power to enforce this article by appropriate legislation.

In plain English: the federal government cannot deny you the vote because of race. State governments cannot either. And Congress can pass laws to enforce that.

The constitutional rule was clear. The on-the-ground reality, after the withdrawal of federal troops from the South in 1877 (the compromise that ended Reconstruction), was very different. By the 1890s, Southern states had constructed an elaborate legal architecture to disenfranchise Black voters without using race-explicit language: literacy tests administered with grossly unequal severity, poll taxes that exempted descendants of pre–Civil War voters (the "grandfather clause"), residency rules, and white primaries from which Black voters were procedurally excluded. The Mississippi Plan (1890), the Louisiana Constitution of 1898, and parallel measures across the former Confederacy reduced Black voter registration in many counties from majority levels to near zero within a decade. Mississippi's Black voter registration fell from roughly 70% of eligible Black men in 1867 to under 6% by 1892. Louisiana's Black registration fell from 130,344 in 1896 to 1,342 by 1904.

The Supreme Court partly enabled this. Williams v. Mississippi (1898) upheld the literacy-test architecture on the grounds that the law itself was facially race-neutral. Giles v. Harris (1903) declined to order Alabama to register Black voters even after a finding of systematic exclusion, on the now-notorious ground that no federal court could enforce such an order against a determined state. The result, by 1910, was a South in which the Fifteenth Amendment was a written promise honored almost nowhere on the ground.

Women's suffrage

The Nineteenth Amendment (1920) prohibited denial of the right to vote on account of sex. The seventy-two-year campaign that produced it began at the 1848 Seneca Falls Convention and ran through the long Suffragist organizing of the late nineteenth century, the militant tactics of the Women's Party in the 1910s, and state-level expansions (Wyoming had granted women the vote in 1869 as a territory; by 1919, fifteen states allowed full women's suffrage and many more allowed it in particular elections).

The Nineteenth Amendment was, in formal terms, an enormous expansion: it roughly doubled the legal electorate. In practice, registration and turnout grew more slowly than the legal expansion implied, partly because women, like men, had to navigate the registration architectures that varied by state, and partly because Black women in the South were excluded by the same Jim Crow apparatus that excluded Black men.

Native Americans

The Indian Citizenship Act of 1924 recognized U.S. citizenship for Native Americans born within the United States. Citizenship did not automatically mean the right to vote: states retained the power to set voting qualifications, and several states (notably Arizona and New Mexico) used various rationales — guardianship status, residence on reservations as ineligible "non-resident" status, literacy tests — to keep Native Americans from registering. Arizona's state constitution did not allow most Native Americans to vote until Harrison v. Laveen (1948); New Mexico did not fully allow Native voting until 1962. Some states' practical exclusion continued into the 1970s.

The Twenty-Fourth Amendment and the Civil Rights Era

The Twenty-Fourth Amendment (1964) prohibited poll taxes in federal elections. Harper v. Virginia Board of Elections (1966) extended the prohibition to state and local elections under equal-protection grounds. The combination eliminated one of the central Jim Crow disenfranchisement mechanisms.

The Civil Rights Act of 1964 included voting-rights provisions, particularly Title I, but the Act's voting protections were limited compared to what would come a year later. Federal officials still could not effectively oversee Southern elections.

The Voting Rights Act of 1965

After "Bloody Sunday" (March 7, 1965), when Alabama state troopers attacked civil-rights marchers crossing the Edmund Pettus Bridge in Selma, President Lyndon Johnson addressed a joint session of Congress and announced he would send what became the Voting Rights Act. Johnson's speech included the line that gave the movement its anthem: "And we shall overcome." Congress passed the bill in August 1965; Johnson signed it on August 6, with Martin Luther King Jr. and Rosa Parks present.

The VRA was a watershed because it did three things at once that the Fifteenth Amendment, on its own, had not been able to do for almost a century.

First, Section 2 prohibited any voting "qualification or prerequisite" or "standard, practice, or procedure" that resulted in denial or abridgment of the right to vote on account of race or color. The 1965 version had an "intent" test (you had to prove the rule was adopted to discriminate), but the 1982 amendments shifted to a results test: a rule that produced racially discriminatory effects, given the totality of circumstances, was unlawful regardless of whether intent could be proven. Section 2 applies nationwide and is still good law. The 2023 Allen v. Milligan case, which we will discuss, reaffirmed this.

Second, Section 5 required that certain "covered jurisdictions" — initially most of the Deep South — submit any change to election practices, however small, for preclearance by either the U.S. Department of Justice or the U.S. District Court for the District of Columbia. Until preclearance, the change could not take effect. The covered jurisdictions had the burden to show the change was not retrogressive — that is, did not make things worse for minority voters. This was an extraordinary federal supervision regime. It worked: between 1965 and 2013, DOJ objected to approximately 1,000 proposed changes, and the existence of the regime deterred many more.

Third, Section 4(b) specified the coverage formula that determined which jurisdictions were subject to Section 5 preclearance. The formula keyed on whether a jurisdiction had used a literacy test or similar device in 1964, and whether voter registration or turnout had been below 50% in the 1964 (later 1968 and 1972) presidential election. The formula initially captured Alabama, Georgia, Louisiana, Mississippi, South Carolina, Virginia, and most of North Carolina, with later updates pulling in additional jurisdictions in Texas, Arizona, and elsewhere.

The Twenty-Sixth Amendment (1971) added 18-year-olds to the electorate, ratified in record time after the Vietnam-era argument that a citizen old enough to be drafted should be old enough to vote. This produced the modern legal franchise: every U.S. citizen over 18 has a constitutionally protected right to vote that cannot be denied on account of race, sex, age (above 18), or failure to pay a tax.

A note on the Twenty-Fourth Amendment's specific language is worth pausing on, because it illustrates how the constitutional system handles disenfranchisement mechanisms. The amendment reads, in plain English, that the right to vote in federal elections "shall not be denied or abridged by the United States or any State by reason of failure to pay any poll tax or other tax." It does not prohibit poll taxes themselves; it prohibits using payment of a poll tax as a precondition for voting in federal elections. Texas, Virginia, Alabama, Mississippi, and Arkansas were still imposing poll taxes when the amendment was ratified. The Harper decision two years later, decided under the Fourteenth Amendment's Equal Protection Clause, finished the job by extending the same prohibition to state and local elections. The pairing — constitutional amendment plus subsequent Court decision — is a recurring pattern in voting-rights doctrine: the formal text reaches one element of the structure, and follow-on litigation closes the gap.

The same pattern applies to the literacy test, formally banned in covered jurisdictions by the Voting Rights Act in 1965 and nationwide by the 1970 amendments. Literacy tests, like poll taxes, were not unconstitutional on their face — South Carolina's literacy test had been challenged and upheld in Lassiter v. Northampton County (1959). What made them unlawful was not their form but their use: applied with grossly unequal severity to Black applicants, used in conjunction with grandfather clauses that exempted whites whose ancestors had voted before 1867, and embedded in a registration system designed to filter on race. The 1965 statute was Congress's response to the empirical reality, not an abstract reading of the Fifteenth Amendment.

The arc, summarized

In 1788, perhaps 6% of the U.S. population could vote. In 2024, the U.S. registered approximately 168 million voters, with about 245 million U.S. citizens age 18 and older eligible (turnout was about 64% of eligible voters, a near-record). The expansion of the franchise across two and a half centuries is one of the great democratic achievements in human history. It is also incomplete: roughly 5 million Americans currently cannot vote because of felony status, several hundred thousand more are functionally disenfranchised by registration friction, and Native Americans on some reservations still face practical access barriers. The arc bends, as the saying goes, but slowly, and in fits.

Section 2: The Voting Rights Act in detail

The VRA is the central voting-rights statute in modern American law. To understand current battles, you have to understand the Act's three core provisions and what happened to them.

Section 2: the nationwide rule

Section 2 (52 U.S.C. § 10301) prohibits any "voting qualification or prerequisite to voting or standard, practice, or procedure" that "results in a denial or abridgment of the right of any citizen of the United States to vote on account of race or color, or in contravention of the guarantees" for language minorities (added in 1975). The 1982 amendments specified that violations could be proven by showing discriminatory results based on the "totality of circumstances," even without proof of discriminatory intent.

The Section 2 framework, refined by the Supreme Court in Thornburg v. Gingles (1986), evaluates challenges using factors including:

• Whether minority groups are sufficiently large and geographically compact to constitute a majority in a single-member district.
• Whether the minority group is politically cohesive.
• Whether the white majority votes sufficiently as a bloc to defeat the minority group's preferred candidates.
• A "totality of circumstances" inquiry into history of discrimination, racial appeals in campaigns, and other factors.

Section 2 cases produce remedies. A successful Section 2 vote-dilution challenge can require redrawing district lines. A successful challenge to a voting practice (e.g., a particular voter ID rule, polling-place reduction, or registration cutoff) can require modification or elimination.

In Allen v. Milligan (2023), the Supreme Court — by a 5-4 vote with Chief Justice Roberts and Justice Kavanaugh joining the three liberal justices — affirmed that Section 2 still requires Alabama to redraw a congressional map that diluted Black voting strength. The decision surprised many observers who had expected the Roberts Court to weaken Section 2; instead, the Court applied the traditional Gingles framework. Section 2 remains active law. Most VRA litigation since the Shelby County decision (below) has occurred under Section 2.

Section 5: the preclearance regime

Section 5 required covered jurisdictions to submit proposed election changes for preclearance — federal sign-off, in plain English — before they could take effect. The regime applied to changes in voting qualifications, electoral procedures, district lines, polling places, and any other "voting" practice broadly defined. The burden was on the jurisdiction to prove the change was not retrogressive.

Between 1965 and 2013, Section 5 generated approximately 1,000 DOJ objections and a much larger number of changes that were modified or withdrawn under the threat of objection. Empirical studies (Davidson and Grofman 1994; Issacharoff, Karlan, and Pildes textbook; Persily 2007) generally found that Section 5 had substantial effects in deterring retrogressive changes and in expanding minority political opportunity.

Section 4(b): the coverage formula

Section 4(b) specified the coverage formula. The formula was based on whether a jurisdiction had used a "test or device" (literacy test, etc.) and on voter registration or turnout in particular years. Congress reauthorized the VRA in 1970, 1975, 1982, and 2006 — but did not significantly update the coverage formula in those reauthorizations. By 2006, the formula was using data from 1964, 1968, and 1972.

Shelby County v. Holder (2013)

In Shelby County v. Holder (2013), the Supreme Court — 5-4, Chief Justice Roberts writing for the majority — held that the Section 4(b) coverage formula was unconstitutional because it relied on data that was "decades old" and no longer reflected current conditions. The Court did not strike down Section 5; it held that the formula could not be used to determine which jurisdictions were subject to it. Without a coverage formula, Section 5 became a "shell," in Justice Ginsburg's dissenting phrase: a provision that still existed in law but had no jurisdictions to apply to.

The doctrinal logic was a federalism argument: requiring a subset of states to seek federal preclearance for ordinary election administration is a significant departure from the federalism baseline, and that departure has to be justified by evidence of current discrimination at least roughly proportional to the burden imposed. The 1964 voter-registration disparity that justified the coverage formula in 1965, the majority held, no longer existed in 2013. Justice Ginsburg's dissent took up the metaphor: "Throwing out preclearance when it has worked and is continuing to work to stop discriminatory changes is like throwing away your umbrella in a rainstorm because you are not getting wet."

What happened next is contested. Within hours of Shelby County, several states announced election changes that had previously been blocked by preclearance, including Texas's voter-ID law and North Carolina's omnibus voting bill (which became the McCrory litigation discussed below). Some scholars argue Shelby County unleashed a wave of restrictive changes; others argue Section 2 has continued to police those changes, that DOJ's enforcement priorities under both Trump and Biden administrations have continued to litigate, and that the on-the-ground turnout effects since 2013 have been smaller than predicted (turnout in covered jurisdictions has not declined relative to non-covered jurisdictions). The 2024 turnout data — high in many post-Shelby states — complicates the strong-suppression narrative. The data also shows continued litigation, continued findings of dilution, and continued state-level changes.

In Congress, the John Lewis Voting Rights Advancement Act has been proposed repeatedly to update the coverage formula, in light of Shelby County. As of 2026, it has not passed; the Senate filibuster has been the principal obstacle. The political composition of Congress has not produced sixty votes for a coverage-formula update, and most Republicans argue that Section 2 plus state-level legislation is sufficient.

It is worth marking why the Section 2 vs. Section 5 distinction matters in practice, because the chapter asks readers to take it seriously. Section 5 was a prophylactic rule: a covered jurisdiction had to seek federal approval before a change could take effect, and it had to prove the change was not retrogressive. The burden of proof was on the jurisdiction; the time horizon was prospective; and a wide range of changes — even minor ones, like moving a polling place a few blocks — required submission. Section 2, by contrast, is a post hoc litigation rule. A plaintiff sues after a change has been adopted, the burden of proof is on the plaintiff, and the litigation can take years. A plaintiff under Section 2 can win — Allen v. Milligan shows that — but the institutional structure is much slower and more expensive than preclearance was. A jurisdiction that adopts a change knows that a Section 2 lawsuit is possible, that it might lose, and that, if it loses, it might have to redraw a map or modify a rule for the next election cycle. A jurisdiction operating under Section 5 had to clear the change before the current election cycle. The deterrent effect is qualitatively different.

This is the structural reason why the empirical question — has Shelby County affected voting rights on the ground? — is genuinely difficult to answer. Some changes that would have been blocked under Section 5 may now be operating in covered jurisdictions; many of those changes are being challenged under Section 2; some of those challenges are succeeding; and the net effect on minority voting opportunity is the residue of all those processes plus the broader trends in turnout and registration that would have happened anyway. The Shelby County decision is thus not equivalent to "voting rights have been weakened" or "voting rights have been preserved." It is a change in the enforcement architecture, and the on-the-ground consequences are still being measured.

Brnovich v. DNC (2021)

A second major Section 2 case, Brnovich v. Democratic National Committee (2021), interpreted Section 2 as applied to vote-denial rules (not vote-dilution maps). The Court, 6-3, upheld two Arizona rules: a requirement that out-of-precinct ballots be discarded, and a prohibition on most third-party ballot collection ("ballot harvesting"). Justice Alito's majority opinion specified five "guideposts" for evaluating Section 2 vote-denial claims: the size of the burden imposed, the degree to which the rule departs from standard practice in 1982, the size of the disparity in the rule's impact, the availability of other voting opportunities, and the strength of the state interest served. Critics argued the guideposts make Section 2 vote-denial cases harder to win. Defenders argued the Court was clarifying that not every disparate impact violates Section 2. Brnovich and Allen v. Milligan, taken together, mark the current Section 2 landscape: the statute is alive, vote-dilution maps remain a viable cause of action, and vote-denial rules are evaluated under a multi-factor inquiry that weighs burdens against state interests.

Section 3: Felony disenfranchisement

Roughly 5 million Americans cannot vote because of a felony conviction. The state-by-state variation is enormous, and the moral arguments cut in genuinely different directions.

The state landscape. Maine, Vermont, and the District of Columbia allow voting from prison; Maine and Vermont have done so historically, and the District extended voting to incarcerated residents in 2020. Most states restrict voting during incarceration but restore rights upon release or after parole/probation. A smaller group of states permanently disenfranchise some categories of felons, sometimes requiring a discretionary restoration process through the governor or a parole board. Florida, Iowa, Kentucky, and Virginia have historically been among the most restrictive.

Florida Amendment 4 (2018). In November 2018, Florida voters approved a ballot amendment that automatically restored voting rights to most felons who had completed "all terms of their sentence including parole or probation." Supporters expected the amendment to restore voting rights to roughly 1.4 million Floridians. The Florida Legislature in 2019 enacted SB 7066, which interpreted "all terms of their sentence" to include payment of all fines, fees, and restitution. The 11th Circuit Court of Appeals in Jones v. DeSantis (2020) upheld the legislature's interpretation, finding it not unconstitutional. The practical effect: most Florida felons could not afford to fully pay their fines and fees, and a system to determine the actual amount owed proved administratively difficult. Estimates of the number of people whose rights were practically restored ranged from about 50,000 to several hundred thousand, far short of the 1.4 million envisioned by the amendment's drafters. The episode is a case study in how a ballot-measure win can be substantially limited by subsequent legislative implementation, and reasonable people can read the legislature's action as either faithful interpretation of the amendment or evasion of it.

The competing arguments. Supporters of broader felony enfranchisement argue: (1) civic reintegration is a precondition for successful re-entry; (2) excluding felons from democratic participation is a continuing punishment beyond what the criminal sentence required; (3) the racial disparities in felony rates produce racial disparities in disenfranchisement that, while not constitutionally prohibited per Richardson v. Ramirez (1974), are still morally significant; (4) other democracies generally do not permanently disenfranchise felons. Supporters of narrower enfranchisement argue: (1) the loss of voting rights is part of the social contract that incarcerated felons broke and is a legitimate consequence; (2) states have broad authority under the Constitution to set voting qualifications, and Richardson v. Ramirez explicitly permits felony disenfranchisement; (3) the question is properly settled at the state level, where it has been; (4) restoration after completion of sentence (including financial obligations) is a reasonable rule that many states already apply.

The federal Constitution, as currently interpreted, permits both approaches. The result is the patchwork that exists, with some states (Maine, Vermont, DC) at one end and others (in the historically more restrictive group) at the other, and most states in between.

A specific fact worth registering: of the approximately 5 million Americans currently disenfranchised by felony status, roughly 75% are not in prison. They are on probation, on parole, or — in states with persistent restrictions — have completed their sentences entirely. The Sentencing Project's regular surveys (most recently 2022) put the in-prison fraction at about 25% and the post-release fraction at 75%. This shifts how the policy debate looks: the question is not principally whether incarcerated felons should vote (a genuinely contested values question on which states differ), but whether ex-felons living in their communities, working, paying taxes, and raising children should regain the franchise — and on what schedule, with what conditions. Most state-level reform since 2000 has focused on this second category: New York's 2010 conditional restoration on parole, Virginia's gubernatorial restoration regime under several governors, California's 2020 Proposition 17 restoring rights on parole, and similar measures elsewhere. The trend across most of the country has been toward earlier restoration; Florida's experience shows the trend is not unidirectional and that ballot-measure outcomes can be substantially shaped by subsequent legislation.

Section 4: Recent voting-law changes

Since roughly 2010, state legislatures across the country have made significant changes to election laws — often in opposite directions, with red states tending to tighten and blue states tending to loosen, and with significant variation within both. The Brennan Center for Justice and the Heritage Foundation maintain regular catalogs of these changes; we sketch the categories here.

Voter identification. As of 2026, approximately 36 states require some form of identification at the polls, with about 10 of those requiring strict photo ID. The strictness varies: some states accept a wide range of IDs (utility bill, bank statement, student ID, signed affidavit); some require government-issued photo ID; some require a photo ID and offer free state ID cards as an accommodation; some require photo ID with limited exceptions and provisional ballots for those without. The federal Help America Vote Act (HAVA) of 2002 imposed an ID requirement for first-time voters who registered by mail without verifying their identity, but did not impose a general voter-ID requirement. The expansion of state voter-ID rules has been overwhelmingly a post-2010 phenomenon.

Registration changes. Online voter registration is now available in approximately 42 states and DC. Automatic voter registration (where eligible citizens are registered when they interact with a state agency, typically the DMV, unless they opt out) is in approximately 22 states and DC. Same-day registration (registering and voting on the same day, typically Election Day or during early voting) is in approximately 22 states and DC. There is significant overlap among these, and the geography of adoption tends to cluster: the West Coast, Northeast, and upper Midwest tend toward easier registration; the Southeast and parts of the South tend toward more conventional preregistration deadlines.

Voting hours and locations. State changes since 2010 have moved in opposite directions. Some states have expanded early voting (Colorado, Oregon, and Washington moved to all-mail voting; California significantly expanded early voting and ballot drop-off; Nevada expanded early voting; Michigan in 2022 expanded early voting). Other states have reduced early voting periods or polling places (Georgia's SB 202 in 2021 reduced some early voting in some counties while expanding it in others; Texas's SB 1 in 2021 limited drive-through and 24-hour voting that Harris County had used in 2020). Whether any particular state law is best characterized as "restriction" or "standardization" is itself a contested empirical question.

Mail and absentee voting. Mail voting expanded sharply during the 2020 COVID-19 pandemic; the share of votes cast by mail rose from roughly 25% in 2016 to about 46% in 2020. Some states have retained the expansion (Vermont moved to universal mail voting; California universalized mail-ballot mailing), some have partially rolled back (Texas tightened ID requirements for mail ballots; Georgia limited drop boxes). The 2022 and 2024 elections saw mixed mail use depending on state rules; total mail use settled at roughly 32-35% of votes nationally in 2024. Some empirical research finds that mail expansion in 2020 produced symmetric increases in Democratic and Republican turnout (Yoder et al. 2020; Thompson et al. 2020).

Drop boxes. Drop boxes for completed mail ballots expanded dramatically in 2020. Some states have restricted them post-2020 (Wisconsin Supreme Court ruled them unauthorized under state law in 2022, then 2024 ruled them allowed); some have expanded them (California). Drop-box availability varies enormously by state and even by county.

Provisional ballots. HAVA requires states to offer provisional ballots to voters whose eligibility is in question; the rules for when provisional ballots are counted vary widely.

Voter roll maintenance. The federal National Voter Registration Act (NVRA) of 1993 requires states to make reasonable efforts to maintain accurate voter rolls, including removing voters who have died or moved. The administration of these "purges" varies and has been the subject of significant litigation. Husted v. A. Philip Randolph Institute (2018) upheld Ohio's "supplemental" process, in which voters who failed to vote in two consecutive federal elections and did not respond to a notice were removed; critics argued it disproportionately removed minority voters; defenders argued the federal NVRA framework permits exactly this kind of maintenance.

The terminology in this area is itself politically charged. Defenders of roll maintenance call the practice "list hygiene" or "list maintenance"; critics call it "voter purging" or "voter roll purges." The underlying activity is the same: removing names of voters who have died, moved out of the jurisdiction, or — under some state rules — failed to vote and respond to notices for a defined period. The activity is required by federal law (the NVRA's Section 8 imposes affirmative maintenance obligations) and is also constrained by federal law (the NVRA prohibits maintenance solely on the basis of non-voting and imposes notice and waiting-period requirements). The honest framing: rolls do need to be maintained, the federal statute does require maintenance, and reasonable disagreement is possible about how aggressive the maintenance should be and how much benefit-of-the-doubt should go to voters who appear on the rolls but have not recently voted. Texas, Wisconsin, and Georgia have all been the subject of significant maintenance litigation since 2018; courts have ruled both ways on different facts.

ERIC and its political fragmentation. The Electronic Registration Information Center (ERIC) is a multistate consortium founded in 2012 with Pew Charitable Trusts support, intended to help member states cross-check voter rolls against motor-vehicle and other databases to identify outdated registrations. ERIC was bipartisan in design and at peak membership included thirty-three states across the partisan spectrum. Beginning in 2022 and accelerating in 2023-24, several Republican-led states (Florida, Louisiana, Missouri, West Virginia, Alabama, Ohio, Iowa, and others) withdrew from ERIC, citing concerns about a board structure that included a non-voting member from a left-leaning research organization, about ERIC's voter-outreach requirements, and about data-privacy questions. Some withdrawals were driven by claims circulating on conservative media that ERIC was a "voter registration scheme" funded by left-leaning donors; ERIC defenders responded that the consortium had been founded with broad bipartisan support and was the most effective tool available for cross-state list maintenance. As of 2026, ERIC has lost about a third of its membership and the practical effect on cross-state list-maintenance accuracy is being studied. The episode illustrates how technical election-administration tools have become part of partisan-political discourse in ways their designers did not anticipate.

Section 5: The empirical evidence on suppression effects

This is the section that most invites motivated reasoning. Here is the honest state of the literature.

Voter ID

Strict photo-ID laws produce small overall turnout effects. The most-cited meta-analyses (Highton 2017; Hopkins, Hanmer, Wolfinger 2017) find effect sizes typically below 1 percentage point on aggregate turnout. Some studies find no significant aggregate effect at all. Other studies (Hajnal, Lajevardi, Nielson 2017, with a controversial replication debate) find larger effects, especially for minority voters, though the magnitude is contested.

The point of agreement across most studies: even small aggregate effects can include larger effects on specific subgroups — older Black voters who lack driver's licenses; lower-income voters; some young voters; some Native American voters on reservations where state-issued ID is harder to obtain. An aggregate turnout effect of less than 1 percentage point can mask a 3-5 percentage-point effect on a particular subgroup.

The point of legitimate disagreement: whether the costs to particular subgroups exceed the benefits to election integrity. Defenders of voter ID argue (a) most voters already have ID; (b) ID is a routine requirement for many activities; (c) the existence of an ID requirement reinforces public confidence in elections, even if its anti-fraud effects are small (because in-person impersonation fraud is empirically rare, per the Brennan Center, the Government Accountability Office, and most academic studies); and (d) state-issued free ID programs mitigate access concerns. Critics argue (a) the access burden is real and falls disproportionately on certain groups; (b) the empirical fraud problem is small; (c) the legislative motivations have sometimes been clearly partisan; and (d) reinforcing public confidence by addressing a problem that does not significantly exist comes at the cost of voters who are excluded.

Polling-place reductions

Polling-place consolidations can have moderate local effects. The empirical work (Gronke et al. 2007; Burden et al. 2014; more recent work in Georgia) suggests polling-place reductions produce localized turnout effects that are sometimes larger than aggregate voter-ID effects, particularly when consolidation increases travel distances or wait times. The effects are real but are typically smaller than activist estimates suggest and often smaller than the broader effects of campaign mobilization or weather.

Same-day and automatic registration

Same-day and automatic registration produce measurable positive turnout effects. The empirical estimates typically range from 3 to 7 percentage points for same-day registration (Burden et al. 2014; Hanmer 2009; Hill 2003) and 2 to 5 percentage points for automatic registration depending on implementation. These are larger and more consistently estimated effects than the suppression effects.

Mail voting expansion

The 2020 mail-voting expansion produced large turnout increases, with estimates suggesting roughly symmetric effects on Democratic and Republican turnout (Yoder et al. 2020; Thompson, Wu, Yoder, Hall 2020 in PNAS). Mail voting did not, on the available evidence, produce a partisan turnout advantage.

The progressive critique

Progressive analysts argue that the wave of state-level changes since 2010 has been targeted, in legislative purpose if not always in formal language, at Democratic-leaning groups: minorities, urban voters, students, and lower-income voters. Several court findings support this characterization in particular cases. NAACP v. McCrory (4th Cir. 2016) found that the North Carolina omnibus voting law of 2013 (passed within months of Shelby County) "target[ed] African Americans with almost surgical precision," based on legislative records showing that the legislature had requested and used race-coded voting-pattern data when designing provisions to reduce same-day registration, early voting, and out-of-precinct provisional voting. Similar findings, with different language, have been made in Texas redistricting and voter-ID cases. The progressive view is that Shelby County enabled this wave and that the cumulative effect, even where individual provisions have small turnout effects, is a meaningful constriction of access. Carol Anderson's One Person, No Vote (2018) is a representative articulation.

The conservative response

Conservative analysts argue that election integrity has independent value; that the empirical aggregate turnout effects are small; that rules apply equally to all voters of whatever race or party; and that concerns about clean voter rolls, ID, and rule consistency are legitimate state interests. Hans von Spakovsky and John Fund's Who's Counting? (2012) and the Heritage Foundation's election-integrity work make this case. The conservative view emphasizes that turnout in 2020 and 2024 reached near-record highs even as ID and registration rules tightened in some states, that states with more restrictive rules sometimes have higher turnout than states with more permissive rules, and that "suppression" framing inflates relatively modest changes.

The honest middle

The honest middle: suppression effects are real but typically smaller than activist rhetoric on the left suggests. Election-integrity concerns are real but not justified by widespread evidence of fraud at a scale that would change election outcomes. Both criticisms can be true at once. The legislative motivations for some state-level changes have, in particular cases, been demonstrably partisan and demonstrably racially-aware (the McCrory finding). The aggregate turnout consequences of those changes have, on average, been modest. The cumulative effect over many provisions across many states is harder to estimate; it is almost certainly real and almost certainly smaller than rhetoric suggests.

This is not a comfortable position for activists on either side. It is what the evidence shows.

What the fraud evidence shows

A separate empirical question, often run together with the suppression debate but distinct from it, is whether voter fraud occurs at scale. The evidence here is unusually clear, which is why this section is shorter than the suppression section. The Brennan Center's compilation of fraud cases finds rates of in-person voter impersonation in the range of 0.0003% to 0.0025% of votes cast — that is, three to twenty-five cases per million votes, almost all of which are caught and prosecuted. The Government Accountability Office's 2014 study reached comparable conclusions. Lorraine Minnite's The Myth of Voter Fraud (2010, with a 2017 update) is the standard academic treatment. The Heritage Foundation maintains a database of prosecuted election-fraud cases that, as of 2026, contains roughly 1,500 cases nationwide across all years and all forms of fraud (including absentee-ballot fraud, registration fraud, and ballot-stuffing — not just impersonation). Heritage's database is used by some advocates to argue fraud is more common than the Brennan Center claims; in absolute terms, 1,500 cases across many years is a small number relative to the billions of votes cast in that period, and Heritage itself does not claim the cases approach a magnitude that would change federal-election outcomes.

The honest synthesis: voter fraud occurs, prosecutors prosecute it, and the cases are real. The rate is too low to plausibly determine the outcome of a federal election. This finding does not settle the policy question of whether voter ID and similar rules are warranted — reasonable people can argue, as Justice Stevens did in his Crawford v. Marion County (2008) opinion upholding Indiana's voter-ID law, that public confidence is itself a legitimate state interest even if the underlying fraud is rare. Reasonable people can also argue, as Stevens later said publicly that he had not adequately considered, that the burden on certain voters can outweigh that interest. The empirical fraud rate is a parameter; the policy conclusion is contested.

Section 6: The 2020 election and the truth about fraud

Let us state plainly what the evidence shows about the 2020 presidential election: it was not stolen. President Joe Biden defeated President Donald Trump by 7,059,526 votes nationally and 306-232 in the Electoral College. The result was certified by states, by Republican and Democratic election officials, and by Congress on January 6-7, 2021.

Sixty-plus court cases challenging the 2020 election results were filed by the Trump campaign and allied litigants. The Trump campaign and its allies lost those cases — overwhelmingly, in front of judges appointed by presidents of both parties, including judges appointed by Trump himself. Where evidence was offered, judges examined it and found it insufficient. Where evidence was not offered, judges noted its absence. The Supreme Court declined to take the cases. (For a comprehensive catalog, see Trump v. Wisconsin Elections Commission (7th Cir. 2020); Donald J. Trump for President v. Boockvar (3rd Cir. 2020); the Pennsylvania Supreme Court's In re Canvassing Observation opinion; and the Cybersecurity and Infrastructure Security Agency's joint statement that the 2020 election was "the most secure in American history.")

Recounts and audits were conducted in multiple states. Georgia conducted a full hand recount, then a second machine recount, then a separate audit of signature matching in Cobb County. The results confirmed Biden's victory. Arizona's "Cyber Ninjas" audit, paid for by partisan donors and conducted outside standard election-administration practice, ultimately found that Biden had in fact won by a slightly larger margin than the official count. Wisconsin's Republican-led legislative review, after a year of investigation, did not find evidence that would change the result.

Republican election officials in key states refused to overturn results. Georgia Secretary of State Brad Raffensperger, a Republican, certified Georgia's results despite a January 2, 2021 phone call from President Trump in which Trump asked him to "find 11,780 votes" — a recording that became central evidence in subsequent investigations. Arizona Republican Governor Doug Ducey signed Arizona's certification. Michigan Republican Senate Majority Leader Mike Shirkey refused to summon legislators to overturn Michigan's results. Pennsylvania's Republican legislative leaders did not move to substitute a Trump slate of electors. The system held because election officials, many of them Republicans, did their jobs.

The persistence of election denial is a fact about American political culture. Polling consistently shows that approximately one-third of Americans, and roughly two-thirds of self-identified Republicans, say the 2020 election was illegitimate or stolen. Pew Research, ANES, and other major surveys find this pattern from 2021 through 2025. This is a meaningful and durable belief structure, regardless of its empirical basis.

The chapter is honest: the empirical claim is wrong, the political-cultural fact is real, and the consequences for democratic legitimacy are significant. Subsequent chapters (especially Chapter 37, on January 6 and the erosion of democratic norms) examine the political consequences in detail. For the purposes of this chapter, the point is narrower: voting-rights debates after 2020 occur in a context where a substantial portion of one party's electorate believes elections are illegitimate. That context affects what reform proposals are politically feasible, what rule changes are framed as "election integrity" versus "voter suppression," and how election administrators do their jobs.

The 2024 election, as of the writing of this chapter, did not produce an analogous denial movement. Donald Trump won the 2024 election, and his supporters were not motivated to claim the election was stolen. The 2024 result was certified, electors were counted, and the peaceful transfer of power occurred on January 20, 2025. Among the politically significant facts about American democracy as of 2026 is that the country had a peaceful transfer of power in 2025, conducted under election rules that had been litigated and challenged but not overturned, and that 2020-style denial did not produce a 2024-equivalent movement. The institutional system that Republicans criticized as biased in 2020 produced a Republican victory in 2024; the system Democrats had defended in 2020 was the system that produced their loss in 2024. The system, on both occasions, worked.

Section 7: Election administration as a policy area

Behind the high politics of voting rights is the practical work of running elections. The United States has roughly 10,000 election administration officials nationwide, depending on how you count. Most of them work for counties or municipalities. Many are part-time, low-paid, and elected; others are appointed civil servants. The work is detailed, deadline-driven, and unforgiving: ballots have to be printed correctly, machines have to be tested, polling places have to be staffed, voter rolls have to be updated, mail ballots have to be processed, and the entire operation has to satisfy federal, state, and local rules that frequently change.

The post-2020 wave of resignations and threats against election officials is a genuine phenomenon. Surveys by Reed College and the Brennan Center find that roughly one-third of local election officials in 2022 reported being personally threatened or harassed since 2020. Veteran election officials have left the profession in numbers that worry institutional observers across the political spectrum. Maricopa County, Arizona; Fulton County, Georgia; Wayne County, Michigan; and other large jurisdictions have lost senior staff. The replacements are sometimes less experienced; in some jurisdictions, the replacements have been chosen from movements that question 2020 results, raising the institutional question of whether the 2024 election would be administered as professionally as past elections. (As it happened, 2024 was administered professionally and produced a result accepted by the losing party.)

Professional infrastructure for election administrators has expanded in response. The Election Center provides training and certification (Certified Election/Registration Administrator, or CERA, program). The National Association of Secretaries of State (NASS) and the National Association of State Election Directors (NASED) coordinate across states. The Brennan Center, the Heritage Foundation, R Street, the Bipartisan Policy Center, and other organizations across the ideological spectrum work on election-administration improvements that often command bipartisan support: clear deadlines, secure voting equipment, adequate poll workers, accessible polling places, and prompt counting.

This is a policy area where the politics is bitter at the top and surprisingly cooperative at the working level. Most election officials, of both parties, want elections that are free, fair, and not the subject of national controversy. The challenge is that national controversy now reliably attaches to local administration in close races.

A consequential concrete example: in 2024, Maricopa County, Arizona — one of the most-watched election jurisdictions in the country — was administered by Recorder Stephen Richer, a Republican who had publicly affirmed that the 2020 results in Maricopa were accurate, and Board of Supervisors Chairman Bill Gates, also a Republican. Both faced primary challenges from candidates who had questioned 2020 results. Both were re-elected or stepped down on their own terms. The 2024 election in Maricopa was administered without major operational failure, results were released on a published timeline, and Donald Trump won Arizona. The system that had been the subject of intense conspiracy theorizing in 2020-2021 produced, in 2024, an outcome accepted by all parties. The election officials who had been pilloried for accurate administration in 2020 conducted accurate administration again in 2024. This is not a triumphalist story — many of those officials left the profession after years of threats — but it is a story about institutional resilience that the chapter has to record alongside the genuine concerns about election-official attrition.

Section 8: Federal voting-rights legislation and the future

Several major federal voting-rights bills have been proposed since 2020.

The Freedom to Vote Act (2021–22) would have set federal minimum standards for early voting, mail voting, and voter registration; required disclosure of large political donations; and addressed gerrymandering. It passed the House but fell short of the 60 votes needed in the Senate. Senator Joe Manchin (D-WV) and Senator Kyrsten Sinema (then-D-AZ) declined to support eliminating or modifying the filibuster to pass the bill, and the bill failed.

The John Lewis Voting Rights Advancement Act would have updated the Section 4(b) coverage formula in light of Shelby County. It also fell short of 60 votes in the Senate.

The Electoral Count Reform Act of 2022 did pass, with bipartisan support. It reformed the 1887 Electoral Count Act in light of January 6: clarified that the Vice President's role in counting electoral votes is purely ceremonial; raised the threshold for congressional objections (now requires one-fifth of each chamber to object, up from one member); and tightened state procedures for designating electors. The ECRA is the major bipartisan voting-rights legislation of the post-2020 period, and its passage is genuinely significant: it closed several of the legal ambiguities that the Stop the Steal effort had attempted to exploit.

The ECRA is worth describing in more detail because its provisions are concrete and consequential. First, the Act specifies that the Vice President's role in the joint session of Congress that counts electoral votes is "ministerial only." This addresses the John Eastman memo theory advanced in late 2020, under which Vice President Mike Pence might have unilaterally rejected slates of electors or returned them to states. Pence rejected that theory at the time. The ECRA writes that rejection into statute: a Vice President in a future close election cannot claim, even arguably, that the role is anything other than ceremonial. Second, the Act raises the threshold for congressional objections from one member of each chamber to one-fifth of each chamber. The 2021 process saw several Republican senators and over 100 Republican House members vote to sustain objections to slates from Arizona and Pennsylvania; under the ECRA, organizing such objections requires a much larger coalition. Third, the Act establishes that each state has a single, identified executive responsible for certifying electors (typically the governor unless the state designates otherwise), and requires that any challenge to that certification proceed through expedited federal-court review with direct Supreme Court appeal. The intent is to ensure that disputes about who won a state are resolved through courts before the joint session, not at the joint session. Fourth, the Act eliminates the antiquated provision that allowed a state to "fail" to choose electors and then designate them through legislative action, which had been a subject of speculation in late 2020.

The ECRA is not a complete solution to the structural vulnerabilities that 2020 exposed. It does not address state-level election administration, does not federalize counting procedures, and does not preempt state-court election litigation. But it closed specific legal ambiguities — about the Vice President's role, about the threshold for congressional objections, about the path of disputes to the courts — that determined adversaries had attempted to exploit. The bipartisan vote (61-37 cloture in the Senate, with sixteen Republicans joining all Democrats) is the kind of cross-partisan institutional self-protection that the post-2020 period has otherwise been short on. The Act's history is a counterexample to the thesis that all voting-rights legislation is hopelessly polarized.

Federal voting-rights legislation under future congressional majorities will depend on the partisan composition of the Senate and the willingness of either party to alter the filibuster. Both the Freedom to Vote Act and the John Lewis Act have been reintroduced in subsequent Congresses; both remain short of 60 votes. The state-level battle over voting rules continues to be the principal arena.

Section 9: Comparative perspective

The United States has historically had lower turnout than most peer democracies. Turnout in U.S. presidential elections (66.6% of voting-eligible population in 2020, ~64% in 2024) is below the average of OECD democracies (~75%). The gap is partly driven by registration burden: most U.S. voters must affirmatively register, while most peer democracies register voters automatically as a function of citizenship records or population registries. Some democracies (Australia, Belgium, parts of Latin America) have compulsory voting with small penalties for non-voting; others (Germany, France, Sweden) have non-compulsory but very high-turnout systems anchored by automatic registration and weekend or holiday voting.

A textbook on American government should report this comparative fact without using it as an argument for any particular reform. Different turnout systems reflect different democratic traditions, different conceptions of the relationship between citizen and state, and different judgments about whether a higher or lower turnout serves democratic legitimacy. A 99% Australian turnout includes many voters compelled by law; a 64% U.S. turnout includes many voters who chose voluntarily not to register or not to vote. Both are democracies. The data is the data.

There is a related comparative fact about administration that bears on the U.S. case. Most peer democracies run elections through national or quasi-national bodies — the Australian Electoral Commission, Elections Canada, the UK's Electoral Commission, India's Election Commission. These bodies set procedures, standardize ballots, and certify results across the country. The United States runs its elections through approximately 10,000 local jurisdictions operating under fifty different state legal regimes, with federal floors set by HAVA and the NVRA but no federal central administration. The decentralization is a deliberate constitutional design choice — the Elections Clause (Article I, § 4) leaves "the times, places and manner of holding elections" to the states, subject to congressional override — and it has both costs and benefits. The cost is variability: a Pennsylvania voter and a Georgia voter and a California voter face quite different procedures, deadlines, and ballot designs, and disputes about election rules become disputes about state law in fifty separate venues. The benefit is institutional resilience: there is no single point of failure, no national election authority that can be captured or compromised in a way that affects every jurisdiction at once. Whether decentralization is on net good or bad for U.S. democracy is itself a contested policy question. Several recent reform proposals (most prominently the Freedom to Vote Act discussed earlier) would have set federal floors on registration and early voting; opponents argued these floors would intrude on state authority that the Constitution allocates to states. The comparative perspective is useful for understanding the choice; it does not, on its own, settle the question.

What this chapter has argued

The American voting franchise has expanded enormously over two and a half centuries, and that expansion is one of the durable achievements of the republic. The Voting Rights Act of 1965 was a watershed; Shelby County v. Holder (2013) substantially altered its enforcement architecture; Section 2 remains active. Felony disenfranchisement varies enormously by state and is a contested policy area where reasonable disagreement persists. Recent state-level voting-law changes have moved in opposite directions across the partisan landscape, with red states tending to tighten and blue states tending to loosen. The empirical evidence on suppression effects is more modest than activist rhetoric suggests on both sides, but it is real and uneven across subgroups. The 2020 election was not stolen; the political-cultural movement that holds otherwise is empirically wrong; the system worked in 2020, and it worked again in 2024. Federal voting-rights legislation has produced one major bipartisan reform (the Electoral Count Reform Act) and has fallen short on others.

What you should take from this chapter is a set of frameworks, not a set of conclusions. Voter ID, mail voting, drop boxes, polling-place hours, and registration rules are policy choices that have real but typically modest turnout effects. They are also choices about what kind of democracy you want — how much access you weight against how much administrative consistency, how much state autonomy you weight against how much federal floor. Reasonable people, looking at the same evidence, can reach different conclusions. The point of the chapter is to give you the evidence that a reasonable person should be looking at.

The next chapter takes up January 6 and the broader question of democratic erosion. It is the chapter that follows from this one because, when the empirical question of who won an election becomes contested at the level of a substantial fraction of the public, voting rights become inseparable from democratic norms more broadly. We turn to that there.

A final note on what this chapter has tried not to do. It has not characterized the Democratic Party as the party of voting rights and the Republican Party as the party of voter suppression; it has not characterized the Republican Party as the party of election integrity and the Democratic Party as the party of fraud denial. Both parties contain serious people working in good faith on these questions, and both contain people who are using the questions as partisan weapons. The veteran political scientist's job is to describe what the law says, what the data shows, and what the actors did, and to leave the moral evaluation to readers who are entitled to their own conclusions. Voting rights are central to the American constitutional project precisely because they are the mechanism through which the consent of the governed is registered. Disagreements about how that mechanism should operate are real, durable, and unlikely to be resolved by any single chapter or any single textbook. What you should leave with is the vocabulary, the legal architecture, and the empirical bearings to participate in the debate as a citizen who can read a study, follow a court case, and tell when an argument is being made on the strongest ground rather than the cheapest.