Ballot access: the practical steps between being eligible and casting a vote that counts

In plain words

Ballot access, as used on this page, means the practical ability of an eligible person to register, cast a ballot, and have it counted. It covers registration, identification rules, polling place lines, mail ballot rules, and provisional ballots, and it is measured in the United States through official sources such as the Census Bureau’s voting supplement and the Election Assistance Commission’s Election Administration and Voting Survey.

Think of voting as a path with several gates. First you must be on the list. Then you must show you are who you say you are, in the way your state requires. Then you must get to a polling place during open hours, or get a mail ballot back on time with the right signature. Each gate exists for a reason. Each gate also stops some people who are eligible. The question is how many, which people, and whether the gate is worth it.

The problem, stated precisely

The problem is the gap between eligibility and a counted ballot, for people who are legally entitled to vote and want to. It has four measurable parts:

1. Registration gaps. Eligible citizens who are not registered, or whose registration is out of date.
2. Casting costs. Time, travel, and documents needed to vote, including wait times that differ across neighborhoods.
3. Counting losses. Ballots that are cast but not counted, such as rejected mail ballots or provisional ballots that are not accepted.
4. Uneven distribution. Whether these frictions fall harder on some groups than others.

A note on terms. In election law, “ballot access” also refers to the rules that decide which candidates and parties appear on a ballot. That candidate-side meaning is related to primary elections and is not the subject here. Disability and language barriers have their own pages: disability access and language access. The history of deliberate exclusion, from literacy tests to poll taxes, is on access and exclusion.

Defined terms

• Citizen voting-age population (CVAP). U.S. citizens aged 18 or older. Turnout rates are often stated as a share of this group.
• Mail ballot rejection rate. Mail ballots returned by voters but not counted, divided by mail ballots returned.
• Ballot curing. A process that lets a voter fix a problem with a mail ballot, such as a missing signature, so it can be counted.
• Provisional ballot. A ballot cast when a voter’s eligibility cannot be confirmed at the polling place; officials check eligibility afterward and count it in full, in part, or not at all.
• Strict photo ID. A rule requiring a voter without acceptable photo ID to cast a provisional ballot and take further steps after election day for it to count.
• Automatic voter registration (AVR). Registration triggered by a transaction at a government agency, such as a motor vehicle office, with the option to decline.⁠¹

Evidence

Registration and turnout

The Census Bureau reported in April 2025 that in the November 2024 election, 73.6 percent of the citizen voting-age population was registered, about 174 million people, and 65.3 percent voted, about 154 million.⁠² These are survey self-reports from the Current Population Survey, which tend to run higher than official counts. The Election Assistance Commission’s official administrative count put 2024 turnout at 64.7 percent of CVAP, with more than 158 million counted ballots, and reported more than 211 million active registered voters, which it called an all-time high.⁠³ The two numbers measure different things: one is what people said, the other what officials counted, and registration lists include people who have moved or died until they are removed.

How people voted, and what was not counted

According to the EAC’s 2024 Election Administration and Voting Survey, 35.2 percent of voters cast ballots in person before Election Day, 37.4 percent in person on Election Day, and 30.3 percent by mail.⁠³ States reported transmitting 66,867,671 mail ballots, of which 47,957,093 were returned. Of the returned ballots, 47,629,437 were counted and 584,463 were rejected, a rejection rate of 1.2 percent.⁠³ The most common reason for rejection was a signature that did not match the one on file or was incomplete, about two-fifths of rejections.⁠³ More than 585,000 mail ballots entered a cure process, and more than half were cured and counted.⁠³

For provisional ballots, the most common reason for rejection was that the person was not registered in the state, 52.2 percent of rejected provisional ballots.⁠³

A rejection rate of 1.2 percent sounds small. Applied to 47,957,093 returned ballots, it is more than half a million ballots from people who tried to vote. Whether those were mostly eligible voters who made fixable errors, or ballots correctly rejected, cannot be read from the totals.

Wait times

Wait times are an access cost that falls unevenly. Using smartphone location data from the 2016 presidential election, Chen, Haggag, Pope, and Rohla found that voters in entirely Black neighborhoods waited 29 percent longer than those in entirely white neighborhoods and were 74 percent more likely to spend more than 30 minutes at their polling place, and that the gap persisted within the same states and counties.⁠⁴

Voter identification

Thirty-six states request or require identification at the polls, ten have strict photo ID rules, and fourteen states plus the District of Columbia have no document requirement, using other checks such as signatures, according to the National Conference of State Legislatures as of September 2026.⁠⁵

The evidence on effects is mixed. The U.S. Government Accountability Office estimated in 2014 that turnout in Kansas and Tennessee fell more than in comparison states from 2008 to 2012 after new ID rules, by 1.9 to 2.2 percentage points in Kansas and 2.2 to 3.2 points in Tennessee, with larger reductions among young voters, new registrants, and African-American registrants.⁠⁶ The GAO also reviewed ten earlier studies: five found no significant effect, four found decreases, and one found an increase.⁠⁶ A later study by Enrico Cantoni and Vincent Pons, using about 1.6 billion observations from 2008 to 2018, found strict ID laws had no significant negative effect on registration or turnout, overall or for any group, and no effect on actual or perceived fraud.⁠⁷ The authors noted that campaign contact with non-white voters rose 4.7 points, which may have offset effects.⁠⁷

In Crawford v. Marion County Election Board (2008), the Supreme Court upheld Indiana’s photo ID law. The lead opinion accepted interests in deterring fraud and protecting public confidence, while noting that the record contained no evidence of in-person impersonation fraud in Indiana itself.⁠⁸

Causes and competing explanations

• Administrative design. Most registration and casting rules are set by states and run by thousands of local offices, so access varies by address. See election administration funding.
• Resource allocation. Unequal wait times may reflect fewer machines, poll workers, or polling places per voter in some areas.⁠⁴ Whether that results from budget limits, forecasting errors, or intent varies by place and is disputed.
• Mobility and paperwork. People who move often must re-register, and new registrants were among those GAO found most affected by ID changes.⁠⁶
• Integrity tradeoffs. Signature checks, ID rules, and registration deadlines exist to confirm eligibility and prevent double voting. Each also rejects some eligible voters. The disagreement is about the ratio.
• Competing explanation: motivation, not barriers. Some argue that most nonvoting reflects lack of interest rather than obstacles, and that measured effects of rule changes are small precisely because motivated voters adapt.⁠⁷ See unequal participation.

Who is harmed

The groups most exposed to access frictions in the evidence above are young voters and new registrants,⁠⁶ voters in predominantly Black neighborhoods in the wait-time data,⁠⁴ voters whose signatures have changed since they registered, and people who move often. Military and overseas voters face deadlines; the EAC reported that fewer than 4 percent of returned overseas and military ballots were rejected in 2024, most often for arriving after the deadline.⁠³ Every voter is also harmed if rules are so loose that people doubt results, which is the concern behind integrity measures; see trust in elections.

What has been tried elsewhere

Automatic voter registration. NCSL lists 24 states and the District of Columbia as having enacted or implemented AVR, with Oregon and California the earliest enactments in 2015.⁠¹ Results: AVR adds registrants through routine transactions. Limits: list accuracy depends on the agency data, and critics worry about ineligible people being registered by error; most designs let people decline.

Compulsory enrollment. In Australia, “Enrolment and voting is compulsory for every Australian citizen aged 18 years or older,” and the Australian Electoral Commission estimated national enrollment at 97.8 percent, 18,470,795 people, as at September 30, 2026.⁠⁹ Results: near-universal enrollment. Limits: compulsion is a different relationship between citizen and state that many Americans across the spectrum reject.

Ballot curing. States that allow curing let voters fix signature and similar errors. More than half of the 585,000-plus ballots that entered cure processes in 2024 were counted.⁠³ Limits: curing depends on timely notice and voter response.

Expanded early and mail voting. Mail voting rose to 43 percent in 2020 and settled at 30.3 percent in 2024, while early in-person voting grew.⁠³ Results: more options. Limits: mail voting introduces rejection risk and chain-of-custody questions that in-person voting does not.

Voter ID requirements. Ten states use strict photo ID,⁠⁵ and the Supreme Court upheld Indiana’s law in Crawford.⁠⁸ Results: the measured turnout effect may be near zero nationally.⁠⁷ or modest in specific states and years⁠⁶ Limits: whether ID rules raise confidence enough to justify any cost is a value judgment the data cannot settle.

Competing interpretations

The access-first view holds that every eligible voter has a right to a counted ballot, that frictions fall hardest on groups already underrepresented,⁠^4,6 and that the burden should be on the state to make voting easy, through tools like AVR and curing.

The integrity-first view holds that confirming eligibility is part of the right to vote, because an ineligible or duplicate ballot dilutes eligible ones. Supporters point to the Crawford Court’s acceptance of fraud deterrence and public confidence as legitimate interests,⁠⁸ and note that the best large study found strict ID laws did not reduce turnout.⁠⁷ On this view, reasonable verification rules are not barriers.

The evidence-skeptic view notes that both sides overstate. Cantoni and Pons found no effect on fraud as well as no effect on turnout,⁠⁷ which cuts against claims that ID laws either prevent much fraud or suppress many votes. The GAO’s mixed literature review points the same way.⁠⁶

The administration view holds that the largest measurable losses come from ordinary administration, such as signature mismatches and registration errors,⁠³ not from the rules that dominate debate, and that better forms, notice, and curing would recover more ballots than any change to ID law.

Application to this framework

My Vote Forward treats official eligibility and registration as belonging to the official election system, not to a participation website. The identity and registration chapter separates account authentication, identity proofing, and electoral eligibility. The private voting chapter explains why a future official ballot would have to coordinate with existing channels so no one is counted twice. The measurement chapter explains why counting who participated in a voluntary system says little about who was excluded from it.

Unresolved questions

• How many of the 584,463 rejected mail ballots in 2024 came from eligible voters who could have been cured with better notice⁠³?
• Why do the GAO and Cantoni and Pons findings differ: different states, years, methods, or the offsetting effect of campaign mobilization⁠^6,7?
• What is the right public measure of wait times, and who should collect it each election?
• Can list accuracy and ease of registration both improve at once, or is there a real tradeoff?