The Reality of Full Hand Counts: A Guide for Election Officials

Election officials in some jurisdictions are being pressed to consider replacing their automated vote-counting systems with full hand counts.

Full hand counts are different from limited hand counts, which are common and play a valuable role in elections. Limited hand counts are often used during the audit process, for example, to make sure automated counting systems produced accurate results.

In full hand counts, human teams manually tabulate every ballot and every contest, and their results are used as the official results. As this resource will show, this kind of effort to replace a modern voting system with a full hand count practically guarantees complexity, higher costs, delays, and mistakes.

The best testimony explaining these problems comes from local election officials across the United States. They have studied and tested full hand counts. And, as the chief operating officer for the Georgia secretary of state put it, their experience shows that a full hand count “is more expensive, less accurate (by a lot), delays results, puts outcomes in doubt.”

By comparison, election officials have determined that modern voting systems are cost-effective, reliable, and fast. As Clerk Nicci Kammerich of Osage County, Missouri, wrote, voting systems “are faster, accurate and more efficient to get the job done”—and they allow election officials to “ensure there is no fraud and everything is accurate in a timely manner for our citizens.”

In this resource, we’ll explain how full hand counts typically work. We’ll break down the challenges they present. And we’ll provide examples from across the country.

Finally, we’ll offer ways for election officials to build and maintain public trust in their existing voting systems.

Throughout the resource, we’ll hear from election officials who analyzed what a full hand count would mean in their jurisdictions. Their focus was on serving the voters and delivering election results with speed and accuracy. What they found should resonate with anyone who has an interest in accurate election results, timely election resolution, and functioning election departments.

1. Full hand counts cost more money, require more complex logistics, and take much more physical space than automated tallies. By contrast, modern voting systems are cost-effective and efficient. The increased burdens associated with full hand counts also may not be accompanied by additional funding necessary to meet the needs.
2. Full hand counts take far longer than automatic tabulation, especially in high-turnout elections or those with multiple races on the ballot. Voting systems tabulate quickly and facilitate timely results that enhance voter confidence. Full hand counts, on the other hand, may delay result reporting.
3. Full hand counts are likelier to lead to inaccurate results than automated tabulation. Voting systems are accurate and reliable. Jurisdictions performing full hand counts must build in additional time to audit—and sometimes redo—hand-tallied results to catch human error.
4. Public education on voting systems is key. Calls for full hand counts often spring from distrust in voting systems. So election officials should reinforce public trust in their systems—and do so early and often, especially during highly visible election cycles.

In Shasta County, the Board of Supervisors voted to move to a full hand count beginning with the November 2023 election. As her team prepared for that election and looked ahead to the presidential primary and general elections in 2024, County Clerk and Registrar of Voters Cathy Darling Allen developed a manual tally plan spanning more than 80 pages and conducted two mock elections.

In a detailed analysis prepared for the county board, Darling Allen described substantial costs, challenges in recruiting and training enough qualified workers, and potential delays in election results. She also warned of unforeseen complications—the “multiple points of potential failures” that could further hamper the count.

Implications of a Full Hand Count

Darling Allen’s plan for a full hand count included quality controls, carefully designed processes, and training guidelines, all of which she said would guard against failures “to the extent possible.” Despite her team’s best efforts, however, she warned that unforeseen complications could imperil the county’s ability to finish the count on time.

“We know that workers will make mistakes,” Darling Allen wrote to the board, “not out of ill will or lack of diligence but because these tasks will be repetitive and require significant attention to detail. Correcting those errors is often exceptionally time consuming, requiring portions of the count to be redone. As a result, if errors occur at a higher-than-anticipated rate, they, too, may threaten our ability to certify results by our legally mandated deadline.”

When a group of citizens proposed that Osage County try a full hand count for its municipal election in April 2023, the county clerk’s office was open to the idea. Even if it didn’t work out, Clerk Nicci Kammerich figured, it would be a learning experience.

The county normally uses a secure optical-scan system to count its ballots. After every election, the machine count is checked against a limited hand count. Under that system, the county has had no problems with accuracy since at least 2015, when Kammerich took office as county clerk.

The full hand count in April 2023 was a very different experience. Kammerich described it in a letter to a local newspaper. Her account reflects a broad range of challenges that election officials may expect if they move to a full hand count.

Based on her experience, and “[a]fter considering all factors of this election and comparing it to other elections that are similar,” Kammerich expressed her “fear that if we were to continue hand counting it would cost us more in time, money, losing volunteers, and accuracy of votes.”

She also noted that there was no issue with the county’s current voting system: “Our tabulation machines that the county uses for elections are faster, accurate and more efficient to get the job done. With the process of the recount when using tabulation machines, we can ensure there is no fraud and everything is accurate in a timely manner for our citizens.”

As a result, she concluded that “[o]ur office intends to move forward with our tabulation machines for upcoming elections.”

To understand why a full hand count takes so long, requires so many people, and can lead to so many mistakes, it helps to understand best practices for the process, step by step.

Tasks that are essentially effortless in an automated count require many additional steps during a full hand count. For example, voting systems automatically group ballots by ballot type or precinct. In a manual tally, a team of people must do it. In an automated tally, ballots do not normally pass through many hands, and results can be traced back to individual ballot images within the system. By contrast, in a manual tally, many different people handle ballots, and a high volume of paper must be physically tracked and processed simultaneously, all while guarding against inadvertent or intentional errors. Especially in the current environment, the number of people handling ballots in a manual tally can also give rise to concerns about tampering. As a result, manual tallies require many more tracking and quality control processes.

Note that some of these steps may not be necessary if the jurisdiction retains some voting system equipment, for example to scan ballots for tracking and audit purposes or to add up results. In addition, some of these steps can be simplified if state law permits the use of electromechanical systems in a hand count.

As you read the steps necessary for a full hand count, keep in mind that these steps represent best practices—and cover only the steps after the polls close and counting begins. Not shown here are the many additional steps required before the counting and canvass. Those include recruiting and training the people required to count the ballots, finding sufficient and secure space to do the counting, extensive preparation, and setup.

Also keep in mind that a failure at any step described below can add confusion, stop the process, and require redoing parts of the count.

The Manual Tally Process

Once counting begins:

Moving to a full hand count means election officials have to plan for many additional steps. To name just a few responsibilities, an election official overseeing a hand count must:

• Design a counting process, including the forms and training materials necessary to tally.
• Create ballot-tracking and storage systems that ensure adequate chain of custody and accurate aggregation of results.
• Design a ballot reconciliation process to make sure the number of ballots hand-tallied equals the number of voters who cast a ballot.
• Design an audit process that will reliably catch errors at each step.
• Design a tabulation and reporting mechanism for reporting results to state authorities and the public.
• Recruit, manage, and retain temporary staff, making sure those workers can carry out the process with consistency and accuracy.

Election officials have to do all this while complying with state law, protecting ballot integrity, ensuring accuracy and timely results, and explaining to voters how the election will be administered.

Because of reporting and certification deadlines, election officials already have little room for error in their administration timelines for large elections. The complexity of a hand count only tightens this window. Resources and procedures, including contingency procedures, must be planned and executed well before the election.

Designing a Counting System

There are many considerations involved in designing a counting system. Election officials then must design tally procedures that comply with state and federal law and ensure secure, fair, and accurate results. For example, election officials must make sure:

• Ballot integrity is maintained through the counting process. Ballots pass through many hands to get from the ballot bag or envelope to the tally team, through audit, and back to storage. Procedures to maintain ballot integrity, including robust accounting and reconciliation, are even more important if ballots are not scanned to preserve their images before they are handled.
• The secrecy of results is maintained before the close of the polls. A voting system can keep results secret until the polls close; it is more difficult to maintain secrecy in a hand count if counting is allowed to start prior to the close of polls.
• Batches of ballots are manageable for tally teams. Unlike voting systems, which can handle large numbers of ballots simultaneously, human teams can count only a limited number of ballots at a time—usually about 25.
• The process of calling and recording votes will catch errors. These considerations are especially important because counting can be monotonous, and workers can get tired and make mistakes.
• Reports of results accurately aggregate batches and ballots. Voting system tabulators can quickly add up the varying combinations of ballots that determine the results of each contest in each jurisdiction, but it’s much more challenging for humans.
• Audit processes reliably uncover problems in the manual tally.

Recruiting and Retaining Staff

Full manual tallies often require the hiring of a significant number of temporary workers. This has proved difficult in many jurisdictions. Election departments often turn to temporary workers, workers from other departments, or volunteers. Temporary workers lack experience, and the nature of the work can lead to a disruptive cycle of attrition and training new workers mid-count.

Some of the staffing challenges to consider:

• Election departments need several tally teams, each typically staffed with four to seven people. Large elections may require dozens of teams.
• The monotonous and physically taxing nature of the work requires giving these workers more frequent breaks, resulting in the need for more workers and additional shifts.
• Departments need staff to monitor tally teams and audit their results.
• Still more staff is required to serialize, sort, batch, track, and transport ballots.
• All these workers must be properly trained and supervised in the procedures developed by the election department.

Space Considerations

Many jurisdictions will need to lease, rent, or purchase additional space for large elections. Full hand counts require more infrastructure, including space for:

• Serializing, sorting, and batching ballots.
• Accommodating several multi-member tally teams.
• Separating teams to avoid confusion between teams during the tally, because the person recording the vote must be able to clearly hear the person calling the vote.
• Auditing tally results.
• Aggregating tally results.
• Securely and accessibly storing ballots for tabulation and audit.

Contingency Planning

Because hand counting is time-consuming and complex, a breakdown in any part of the process can threaten compliance with certification deadlines. For example, temporary workers can quit during the process because it is tedious or difficult, or results can fail audits because of recording or math errors. These failures can cause significant and time-consuming disruption. As a result, officials should have robust contingency plans to ensure that results can be obtained and certified in the face of the many challenges that can arise during a full manual tally. This may include scanning batches using voting equipment.

As we discussed above, hand counts require many more resources than automatic tallies. The additional staff and space requirements add considerable cost, especially in large elections, which may require new, larger facilities and hundreds of additional staff.

Many jurisdictions have found that these costs far exceed those associated with a certified voting system. This is especially true because many jurisdictions have found that they must retain at least a portion of their voting systems to comply with federal law protecting voters with disabilities, to design ballots, to audit the results of their hand counts, and to have a contingency plan if their hand count fails.

Timely and accurate election results are pivotal to Americans’ faith in elections. Candidates and voters demand them, and they reduce the time for misinformation and disinformation to spread as voters wait for an outcome. Timely counting is also critical to making sure election officials can meet their deadlines for canvassing and certifying election results and certifying their presidential electors. But election officials’ experience, as well as multiple studies, confirm that hand counting is much slower than automated tabulation. The counting process itself is slower, and identifying and correcting errors is slower still.

Close races can be decided by a handful of votes. Yet full hand counts in the United States are highly likely to lead to inaccurate results. Here are some of the reasons:

• U.S. ballots are exceptionally long, especially compared with countries like France, because they frequently include candidate races for federal, state, and local office, as well as state and local propositions and bond measures.
• A single consolidated election can require many ballot styles—even dozens. Different ballot styles are sometimes required to accommodate district, party, and language. Compared with other countries, where just one race might appear on a national ballot, manual tally teams in U.S. jurisdictions frequently must examine and accurately tally a variety of ballot styles and numerous races per ballot.
• Research shows that people lose focus when doing repetitive, monotonous tasks, like counting hundreds or thousands of votes. This can result in inaccurate tallies and requires multiple layers of checks and audits. Re-tallies are required when discrepancies are discovered, adding significant time to an already lengthy process.
• Manual tallies require accurately tracking a huge number of forms and ballots to make sure no ballots or tally sheets are misplaced or mixed. Failures in tracking can result in voters being disenfranchised or votes being double-counted.
• In a manual tally, many people touch ballots as part of the processes of sorting, tallying, and auditing. In addition, there are many points in the process where data is transferred from tally sheets, result logs, aggregation reports, and final results reports. This introduces a number of additional points where human error can enter the process and cause inaccurate results.

What the Research Says

Studies confirm machine counts are more accurate than full hand counts:

• Ansolabehere et al. (2018): This study examined statewide recounts for two elections in Wisconsin—a 2011 Supreme Court race and the 2016 presidential race. It found that election night hand counting produced error rates of approximately 1.4 to 1.8 times that of optical scanning.
• Ansolabehere and Reeves (2004): This study examined election night count and recount data from New Hampshire going back to the 1940s. It found that, in recent elections, election night hand counting produced error rates about two times that of election night machine scans, even after controlling for factors like the type of race and number of ballots cast.
• Goggin et al. (2012) and Goggin and Byrne (2007): These papers examined hand counts from small lab experiments where the true ballot counts were known. Each paper found an error rate of around 1.4 percent, larger than automatic tabulator error rates in other studies.

Calls for full hand counts often arise when voters are concerned with or lose confidence in voting systems for a variety of reasons. Modern voting systems are complex and use many pieces of equipment. And neighboring counties or other states featured in the news may use different systems, adding to potential voter confusion or questions. Election officials should communicate with voters clearly and regularly about these systems, building trust ahead of time rather than waiting until voters are confused. Public education and transparency are key to shoring up a lasting confidence in our democratic processes. In your jurisdiction, you can:

• Create and publicize simple, engaging voting system explainers so the public can understand your voting system and how it works.
  • Making these explainers available on your website and/or social media channels can help people find trustworthy content.
  • Putting your text directly on the website and/or social media channels, rather than featuring a linked PDF, can make it easier to find in searches.
• In these materials, explain key security and accuracy features in simple terms, without jargon. Key features include:
  • Certification of voting systems.
  • Lack of connection to the internet.
  • Use of paper ballots or voter verified audit trail.
  • Physical security measures and chain of custody procedures.
  • Audit processes that ensure the voting system worked properly.
• Promote opportunities in a variety of mediums—including on social media—to view logic and accuracy testing and other security processes.
• Amplify opportunities for bipartisan poll observers to observe the election.
• Similarly, promote opportunities to view audits and other post-election processes that guarantee accuracy.
• Enlist trusted bipartisan community messengers to attend these events and spread the word about them—and publicize their participation.
• Post videos and explainers of the processes and procedures that guarantee the voting system’s accuracy to your website and social media channels.
• Encourage local media to cover the process and procedures your team has in place to ensure voting systems are accurate and secure in spreading the word about your voting system and the procedures and features that guarantee its accuracy.
• Be transparent, simple, and complete about any issues that may arise and demonstrate proactive steps to address any problems. Similarly, make clear when a problem elsewhere does not affect the voting system used in your jurisdiction.
• Generally speaking, do not repeat or raise the profile of misinformation or disinformation by sharing or re-posting the misleading claims.
  • Instead, amplify and provide easy-to-understand information about why your voting system can be trusted, and how your office’s practices ensure a safe, secure election.

For more comprehensive communications guidance for election officials, please check out The Elections Group’s Telling Our Story: An Elections Communication Guide.

The experience of election officials across the country has made clear that full hand counts have critical disadvantages compared with automatic tallies, especially in elections with complex ballots and high turnout. They cost more money. They require more staff and more physical space. They take far longer than automated counts, putting jurisdictions at risk of missing reporting and certification deadlines. And they present an enormous logistical challenge: Election officials must plan many additional steps, each of which can fail—and add even more time and complexity to the canvass. After all this, full hand counts risk significantly higher error rates. Partial hand counts can provide a benefit when used in the common practice of tabulation audits. But a full hand count is an entirely different undertaking, and it carries significant risks and costs. By contrast, modern voting systems are cost- effective, faster, and more accurate.

Complexity & Cost

Many jurisdictions have provided real world examples of the resources required for a full manual tally—and the associated challenges.

• Mohave County, Arizona: $1 million in costs. After its hand count test, Mohave’s Elections Department projected that performing hand counts in its 2024 statewide elections would cost more than $1 million. This figure was largely based on the cost of staff time to count about 195,000 ballots, but the Elections Department also found that there was only one facility in the county with enough space to accommodate counting, and it had to be equipped with security cameras and guards, as well as a network with adequate bandwidth for live-streaming and recording.
• Osage County, Missouri: New costs and challenges recruiting counters. In Osage County, Clerk Nicci Kammerich initially “thought our county would be saving a great deal of money,” by implementing full hand counts, but “[u]nfortunately, that was not the case.” She ultimately found that costs “were actually higher” for the hand count. She also noted that a majority of election judges raised concerns about hand counts “and asked to please not do hand count again or they will have to quit being an election judge as it is very overwhelming and time consuming.”
• Kerr County, Texas: Election workers object to hand counting. In Kerr, Republican Party chairman and election judge Paul Zohlen noted that several past election workers had already indicated they would be unwilling to return if the county shifted to hand counting—and that he believed it would be difficult for the county to recruit replacements.
• Shasta County, California: Significant staffing needs and new building. In Shasta County, Clerk Cathy Darling Allen found that conducting a full hand count required coordination of eight teams composed of potentially hundreds of staff—375 extra help staff to staff counting boards for the presidential primary alone. She forecast that, in that election, a full hand count would cost $658,925. She estimated that the general election would likely cost roughly twice that much, with heavier turnout and more contests on the ballot. Finally, she estimated that it would cost the county $80,000 a year to house the necessary personnel and operations.
• Georgia: High costs in recount of 2020 presidential race. A state-mandated recount of just one race on the ballot cost some Georgia counties hundreds of thousands of dollars.

Delay

Real world examples show that full hand counts are considerably slower.

• Cobb County, Georgia: 100 days to hand count every race on every ballot. Following the state- mandated recount of the 2020 presidential race, a Cobb County election worker estimated that it would have taken the elections department 100 days to count every race on each ballot using the same procedure employed for just the recount of the presidential race.
• Mohave County, Arizona: 219 days to hand count. Mohave County, Arizona, conducted a test run in 2023 consisting of 850 ballots with approximately 36 races per ballot as it explored a full hand count. In the county’s study, teams of seven counters took up to three minutes to count marked ballots from the 2022 general election. For a seven person counting team working around the clock, that works out to as long as 219 days. To successfully count the 2024 primary and general elections in Mohave County by hand, Elections Director Allen Tempert calculated, the county would need to hire at least 245 tally workers, plus dozens more to serve on write-in and recount boards.
• Shasta County, California: More than 75 minutes per batch of 25. In a second mock election, Clerk Darling Allen found that it took her team of mostly experienced workers 75 minutes to tally a batch of 25 ballots. She also performed time trials for other portions of the process, which added more time. Extracting and serializing a batch of 100 ballots took 55 minutes, and auditing took roughly 9 additional minutes per batch of 25. Finally, she noted that she was unable, in a mock election, to estimate the time that compiling and reporting results would add to the process, but she anticipated that it would be significant, especially if she is barred from using electromechanical devices.
• Esmeralda County, Nevada: More than seven hours to count 317 ballots. After the county commission approved a hand count process for the 2022 primary, county commissioners and election workers spent more than seven hours hand counting a mere 317 ballots. Esmeralda County, the least populous county in Nevada, was the last to certify its results in this primary, less than two hours before the midnight certification deadline.

Inaccuracy

Real world examples show that full hand counts are less accurate than machine tabulation.

• Nye County, Nevada: 25 percent error rate. The Nye County clerk estimated a 25 percent error rate on the first day of the attempted 2022 hand count.
• Mohave County, Arizona: High error rates. In a 2023 hand count test run, a test deck of 850 ballots had roughly 36 races per ballot. The hand count produced errors in 46 cases. The clerk noted that, in a real election, rectifying those errors would require election workers to re-tally those races. The clerk attributed errors to disengaged and tired staffers who stopped paying attention to the process; sloppy handwriting that made it hard to discern the tally; workers speaking too quickly to accurately record results; and workers who heard or said the wrong candidate’s name.
• Tripp County, South Dakota: Voting system identifies human error after hand count. After Tripp County became the first South Dakota county in nearly 20 years to conduct a hand count, it discovered that 75 ballots went “missing.” The error, and its source, were discovered during a voting system audit of the hand count. The “missing” ballots had actually been misfiled. Staff had placed ballots in the wrong precinct’s ballot box. As Tripp County Auditor Barb Desersa put it, the sequence of events “shows the machine is more accurate” because it knew which precinct the ballots belonged to, and “read it as ‘those don’t belong there.’” “The machine caught it,” she said. “To me, that shows that the machine is more accurate than humans.”