In-Person Voting in Elections

The link between the in-person voting experience and confidence in elections is well-documented. Voters generalize from their own experience at the polls to form judgments about how well elections are being run. Long wait times, confusing equipment, poor poll worker interactions, and inaccessible facilities all reduce confidence, not only in the specific election but in the overall integrity of election administration.  Other major factors influencing voter confidence are the “winner-loser effect” and statements by trusted officials.  For a more complete discussion about voter experience and voter confidence, see the Building Trust page on this Resource Hub. 

Most studies comparing voting modes across the United States find that in-person voters express greater confidence that their votes were accurately counted than vote-by-mail voters. This trend holds even after accounting for partisanship and other factors. The mechanism is intuitive because in-person voters can observe the process directly, hand their ballot to a poll worker or feed it into the scanner (or tabulation machine), and receive immediate confirmation. By contrast, vote-by-mail voters must rely on unobserved third parties to ensure their ballots arrive and are counted. Tools such as ballot tracking, curing processes, and livestreamed counting can partially increase transparency.

The Presidential Commission on Election Administration set the maximum acceptable voter wait time at 30 minutes. Most voters have wait times well below this threshold, although these times have varied over the years. .Election Day comparisons across years: 11% of Election Day voters waited over 30 minutes in 2024, compared with 14% in 2020; for early voters, it was 15% in 2024, down from 21% in 2020. 

Historically, wait times were generally longer in minority and higher-density precincts. In 2024, that gap disappeared, with Black, Hispanic, and white voters experiencing wait times over 30 minutes at similar rates (about 13% each).

• Insufficient resources. The most common cause of long lines is an imbalance between the number of check-in stations, poll workers, and voting booths and the expected number of voters. This is largely a planning and resource allocation challenge, not an unavoidable feature of elections.  Polling place consolidation can create new lines if the new polling place is inadequately resourced. 
• Voter arrival patterns. Voter arrival patterns are among the most difficult elements to predict in in-person elections. Arrival patterns depend on many variables, such as the type of election, location, voting system, and in-person polling place assignment. The most common voter arrival pattern is bimodal, with lines longest during the morning surge and the evening rush. Tools now exist, such as simulations and calculator(s), to model these surges and help officials plan accordingly based on their location and voting system.
• Polling place layout and voter flow. How a polling place is physically arranged (e.g., where check-in stations, voting booths, and ballot scanners sit, and how voters move among them) shapes how quickly people flow through and navigate the room. Simulation research finds that the layout method and one-way voter flow can reduce the distance and time each voter spends inside. 
• Voting equipment and check-in technology. Electronic poll books are intended to speed up check-in compared with paper rosters, though evidence that they actually reduce wait times is mixed. Yet when equipment fails (e.g., when an electronic poll book cannot sync or a scanner is jammed), significant bottlenecks can form, and lines can build within minutes, especially when no paper backup is available on the premises.
• Ballot design and complexity. Long or confusing ballots slow voters down and increase errors, extending the time each voter spends at a booth, regardless of how well the rest of the polling place runs. 

Where a polling place is located shapes who shows up and how. Understanding that election officials are often limited by considerations such as availability, accessibility, and cost when looking for polling locations, the following research suggests several factors that may impact turnout.

A large body of research finds that even small increases in the distance between a voter’s home and their assigned polling place measurably lower turnout. The most rigorous test, comparing near-identical neighbors assigned to different polling places across precinct boundaries, found that a roughly quarter-mile increase in distance cut ballots cast by 2 to 5 percent, roughly on the order of 1 to 3 percentage points of turnout, with the strongest effect among voters who start out closest to the polls. Newer work finds that results fall within the same range and that when mail voting is available, some voters facing longer trips switch to vote-by-mail rather than not voting at all. Small per-voter effects like these aggregate into meaningful vote totals across a precinct or jurisdiction.

These effects fall unevenly. Lower-income and minority communities are more likely to be served by lower-quality polling places, and the distance penalty is about three times as severe in high-minority areas during lower-turnout elections.

The research points to a few takeaways for recommended practices, including minimizing distances and disruption, utilizing vote centers, and optimizing site location for both access and equity.

Voting in person can take place at a voter-assigned polling place (i.e., traditional precinct model) or a jurisdiction-assigned polling place (i.e., vote center). Vote centers are polling places that allow any registered voter in a jurisdiction to cast a ballot, regardless of their home precinct. This represents a significant operational shift from the traditional model, enabling resource consolidation and reducing costs and burdens, although it does make it more difficult to determine voter arrival patterns and demand at any single location. Twenty-one states and D.C. now allow the use of vote centers.

This particular type of election reform is frequently justified to increase access and flexibility in in-person elections. Among these justifications are a modest increase in participation, particularly among infrequent voters, and lower administrative costs.

Implementing vote centers comes with risks.  In at least one county study, vote center voters actually reported a more negative experience than precinct voters, driven mainly by longer lines and less helpful poll workers.  Vote centers also increase check-in complexity and require electronic poll books that access the full voter list in real time, since they must serve any voter in the jurisdiction.  Finally, large vote centers concentrate risk if something major goes wrong at the polling location.

Accessibility for voters with disabilities is both a federal legal requirement and an area of active research. What we know:

• Barriers at the polls still depress turnout. Research links the ease of physical accessibility of the polling place to whether people with disabilities turn out at all. Yet, barriers to voters with disabilities are concrete and widespread. When the U.S. Government Accountability Office physically examined polling places during the 2016 general election, it found that roughly two-thirds had at least one potential impediment (e.g., steep ramps, poor paths to the entrance, missing accessible parking) and about a third had a voting station that did not afford a fully private and independent vote. 
• The minimum of one accessible device might be too low. The Help America Vote Act (HAVA § 301(a)(3)(B)) requires at least one accessible voting system at each polling place so voters with disabilities can vote privately and independently. However, simulation modeling of how many accessible machines a polling location actually needs suggests that a flat 1-per-location rule may be insufficient to avoid delays in jurisdictions with a higher percentage of voters with disabilities. The U.S. Election Assistance Commission (US EAC) has developed the Accessible Voting Machines Calculator, a planning tool that recommends deploying additional machines as needed. This approach is supported by advocates who contend that surpassing the single-device minimum fosters a more inclusive polling environment. 
• Quality poll worker training is crucial to developing the requisite skill set. How poll workers set up accessible equipment and assist voters, while preserving privacy and independence, shapes the entire experience. Training guides and protocols exist, including a poll worker curriculum from the Center for Civic Design, published with the National Institute of Standards and Technology (NIST).
• Accessible-equipment coverage has improved, but unevenly. Since HAVA, polling place accessibility has significantly improved, with nearly all locations equipped with at least one accessible device. However, quality gaps remain; these devices are often used by only a few voters, making them more likely to be left unplugged, uncalibrated, or unfamiliar to staff. Even as many states move to hand-marked paper ballots, accessible voting technology remains essential for enabling voters with disabilities to cast ballots privately and independently. When only a handful of voters use a dedicated accessible device, privacy itself can erode. For example, a ballot-marking device that produces differently sized or formatted ballots can help identify which ballots were cast by voters with disabilities. Meeting federal standards does not ensure that a device is genuinely usable for the full spectrum of disabilities.

A major study by the Government Accountability Office finds that possession rates of acceptable voter IDs vary by race and income, with Black, Latino, low-income, and elderly citizens less likely to hold a current government-issued photo ID. Native American voters can face yet other barriers when reservation residents do not have residential street addresses.  Whether these gaps actually lower turnout remains an unsettled question in election research.  Early studies found no negative turnout effects for any racial or demographic groups.  (See here and here.)  Other studies have found small but durable drops in turnout.

Although research into the turnout effects of photo ID requirements is mixed, there are more consistent findings regarding how photo ID laws. Research in New Mexico and Boston has revealed that ID laws can be applied unevenly, even following high-quality poll-worker training.

The operational implementation of photo identification introduces its own equity challenges. A multi-county study of the 2016 election found that scanning a driver’s license either expedited or slowed check-in at polling places, depending on race and ethnicity, because fewer voters in certain areas had scannable identification.

Ballot design is a consequential feature of in-person voting. When a ballot is confusing (e.g., a cluttered layout, unclear instructions, awkward formatting), voters make more mistakes. A landmark multidisciplinary study of voting systems showed that whether voters actually record the choices they intend can depend as much on how the ballot and machine are designed as on the voter. framed the core question—whether voters actually record the choices they intend. Research showed that the answer depends as much on how the ballot and the machine are designed as on the voter. 

The Brennan Center estimated that poor ballot design and instructions have cost tens of thousands, or even hundreds of thousands, of votes per election cycle, with the heaviest burden on older, lower-income, and first-time voters. Most of these errors are silent, meaning that neither the voter nor the poll worker notices them at the time of ballot casting. Therefore, a recommended practice is for election officials to design and test ballots well before Election Day.