Debates about voting policy tend to center on the parts of the process voters interact with most, such as the number of days of early voting or availability of mail voting options. It is, however, the way states approach the period that begins as polls close that can shape whether voters have confidence in election results.
The canvassing and certification period is a black box for most Americans who rarely think about what goes into counting, certifying, and auditing ballots once they are cast. During an era of close contests and hyperpolarization, there is minimal room for error and ample opportunity for misinformation and misunderstanding. In 2021, the perceived need for further analysis of election results gave rise to a series of semi-private, unofficial audits.1 These audits cast doubt on the American voting process, despite being unable to find any evidence to support their claims of fraud.
Perhaps the greatest threat to American elections is not any particular of election administration itself, but the decoupling of reality from perception. As the gap between the reality and perception of elections grows, we see private audits haphazardly attempt to achieve what election officials already do in the aftermath of each election: confirm the accuracy of the results. Yet unlike the official certification processes already in place, these circus-like audits are intended to usurp rather than instill voter confidence. Furthermore, many of the audits failed to uphold security best practices and threatened the integrity of voting systems – forcing some jurisdictions to invest millions in new technology, an unnecessary waste of already limited election office resources.
While there is always room for progress in how elections are run and certification is performed, there is a clear need for public education about how existing policies uphold election integrity. Policymakers, elected officials, and public figures seeking security improvements must invest in strengthening these existing processes, and not throw money towards unproven, unofficial audits that only stand to threaten our democracy.
Our hope is that this paper will inform the debate surrounding election integrity by giving a behind-the-scenes look into the canvassing and certification process. As with any component of the American voting process, the post-election period is unique in each state. This paper is not meant to as a manual for certification, or as a comprehensive survey of existing policy. Rather, it is intended to highlight the three main components of the post-election period that largely apply across the country: unofficial results reporting, canvassing and certification, and audits.
For most Americans, the 2020 election was a crash-course in learning that the results they see on election night are not final. Rather, election night results are often a snapshot of votes cast in-person and any absentee or mail ballots processed before Election Day and counted by the close of polls. These unofficial results are thoroughly checked and vetted in the days and weeks following the election as officials process remaining provisional and absentee ballots and, ultimately, complete an official canvass in accordance with state law.
As soon as the polls close, poll workers, local election officials, and state election officials work together to begin the unofficial results reporting process. Typically, results are aggregated first at the individual precinct level, then funneled up to the local or county election authority, and ultimately shared with the state. At each step of the process, there are extensive protections in place to ensure an accurate count.
While each jurisdiction is unique, some of the most common security and integrity measures include:
- Chain of Custody. According to the U.S. Election Assistance Commission, chain of custody “refers to the processes, or paper trail, that documents the transfer of materials from one person (or place) to the next.” A chain of custody for election technology and ballots is established by applying seals to equipment and ballot boxes, and then recording the seal’s serial number for subsequent verification. For example, election workers might seal a ballot container and record that seal’s number in the poll book. The seal prohibits unauthorized or undocumented entry into the container because doing so would break the seal. Each time election officials open a ballot container, a new seal is attached and the serial number is recorded demonstrating who had access to the container. Chain of custody processes apply not only to ballot containers, but also to voting machines, electronic poll books, and other key election materials.
- Dual record of tabulations. Voting sites typically have both paper ballots and a ballot tabulator that counts the votes. Once the polls are closed, poll workers print several copies of tabulation results—often referred to as ‘tabulator tapes’—which report the number of ballots processed by the tabulator, the number of votes received for each candidate, the number of votes for each proposal, and the number of write-in votes for each item. Poll workers sign the tabulator tapes and attach them to an official statement of results that aggregates results from all the tabulators in use in that polling place. The results are also electronically stored on an external memory device that is secured inside the tabulator while the polls are open. This process results in a dual record of each precinct’s results – one paper and one electronic.
- Paper audit trail. Paper ballots are the gold standard of election security. Should any discrepancies arise in the tabulation of results, election officials can revert to the voter-verified paper ballots to confirm the true number of votes cast for each candidate or initiative. A small handful of jurisdictions remain that use touchscreen voting systems without a paper trail, but they are fewer in number each cycle and BPC’s Task Force has recommended that all jurisdictions transition to paper-based ballots.
- Poll books. The poll book is used to balance the number of voters who checked in to the polling place with the number of ballots tabulated. The poll book also might store comments from poll workers explaining any incidences or irregularities in ballots or materials, such as noting voters who left after being checked-in but before voting. These comments are often essential in explaining balancing issues—discrepancies between the number of voters checked-in and the number of ballots cast—identified during the canvass.
These represent just a few of the many steps poll workers and election officials take throughout the unofficial results reporting process to protect election integrity. Additional security measures might include reporting the number of unused ballots, the use of “air-gapped” technology for the storage of election results, and software protections that prohibit the tabulation of results before the close of polls. Each of these measures supports the operation of the official canvass and, ultimately, the certification of results.
All election night results are unofficial and subject to change. Yet while Americans expect expedient results following the close of polls, the nature of unofficial results reporting creates opportunities for voter confusion and misinformation. As with any process, unofficial results are prone to human and technical error and do occasionally reflect mistakes made during data processing on election night. Critically, errors pertaining to unofficial results reporting do not mean any ballots were compromised, but simply that somewhere in the aggregation of precincts some data was input or processed incorrectly. Thankfully, the process of unofficial results reporting followed by official certification is intentionally structured to catch errors and protect the integrity of the process. In the days after the election, local election officials rigorously double check results and make corrections to the unofficial totals that reflect the ballots as cast.
Canvassing refers to the collection and reconciliation of all ballot materials used during an election. It begins in earnest at the close of polls on Election Day, but includes voting materials that are completed beforehand, such as early and mail ballots. Canvassing leads to certification, or the final and official confirmation of election results by the relevant authoritative body in the state once the canvass is complete.
Specific processes vary by state, but generally there are two canvasses performed after each election – the local and the state canvass. The local canvass includes a thorough proofreading and review of documents provided by poll workers from each precinct, review of chain of custody materials, verification of mail ballots, and occasionally audits and recounts. The state canvass primarily involves the compilation of locally certified results to create statewide and multi-district results.
State law dictates how long election officials have to perform the local and state canvasses, usually beginning within a few days after an election and concluding up to 30 days after the election. Most of this time is reserved for the local canvass, with a few days at the end of the period reserved for the state. The surge in mail voting in 2020 stretched canvassing resources, validating calls for sufficient time to complete an accurate canvass. BPC’s Task Force on Elections believes that 14 days is the minimum amount of time needed to perform the canvass of a presidential election, with some states needing another week as the complexity of the canvass increases.
The Local Canvass
The local canvass is the process of moving results from unofficial to official status, often completed by a bipartisan canvassing board with support from election office staff. The canvass verifies that the standing election results accurately reflect vote totals from each precinct and counting board, that each precinct balances (meaning that the number of voters indicated as having voted matches the number of ballots cast), and that poll workers have complied with chain of custody and other documentation requirements.
The bulk of the canvass typically focuses on comparing the paper tabulator tape from each precinct to the electronic results spreadsheet. This proofreading confirms that the vote totals for each candidate and proposal printed by the tabulator immediately after the polls closed are the same results uploaded on election night.
The goal of the canvass is to run down every possible discrepancy in vote counts or turnout rates. When timelines are short, election officials focus primarily on the precincts and counting boards with the largest relative discrepancies. In some states, poll workers are even called back during the canvass to correct or clarify any identified errors. The analysis of discrepancies focuses on multiple variables, such as:
- Comments in the poll book that may disclose and resolve the problem.
- A review of mail ballot return envelopes for ineligible ballots or envelopes that arrived without a ballot enclosed.
- Mail ballots assigned to the wrong counting board.
- Voters who moved after being issued a mail ballot but were not removed from the list.
- Spoiled ballots that may not have been voided when a new one was issued.
The chain of custody described in the prior section is critical to the security of the ballots and the integrity of the ballots post-election recounts, audits or legal proceedings. During the canvass, chain of custody materials are reviewed to ensure that election workers properly completed all protocols and no voting materials were compromised. If any ballot containers are opened during the canvass, new seals are affixed and the numbers recorded.
Local election results become official at the conclusion of the canvass and are then transmitted to the state election official for aggregation.
The State Canvass
The primary role of the state canvass is to compile vote totals for statewide offices and ballot measures as well as for any contests comprising all or parts of more than one local jurisdiction. In many states, a bipartisan board or a committee of specific statewide elected officials conducts the state canvass, but the work of compiling the official local election results into a statewide official certification is done by staff of the state election official – either the Secretary of State, state board of elections, or comparable office. The state canvass follows the completion of the local canvass.
The state canvass is becoming more robust as states develop queries to test the accuracy of local or county results. Queries might include checks of turnout percentages, searches for duplicate results, or deep dives into select local returns to affirm accuracy.
Official audits are a central component of our elections system and have long been acknowledged as a trusted source for affirming election integrity. While the rise of partisan, unofficial audits seen this year threatens the reputation of audits in the public eye, official audits function to assess and ultimately demonstrate the integrity of the election to voters.
Audits are most often conducted after the canvass, once results are certified, though the specific type and timing of audits performed is dictated by state law. The three types of audits most widely used include performance audits, precinct results audits, and Risk Limiting Audits. Audits can be performed at the state or local level, and no audit type is mutually exclusive. Jurisdictions will often conduct multiple types of audits—when resources permit—to assess the integrity of various parts of the voting process. Audits are different from recounts, which might follow an audit that identifies discrepancies in results.
Election officials and poll workers must execute a variety of security and integrity protocols before, during, and after the election. A performance audit assesses these actions, and ultimately determines whether these critical duties were performed properly.
Pre-election duties might include poll worker training, publishing notices of registration and elections, conducting logic and accuracy tests on precinct and central counting board tabulators, using certified ballot containers, and encrypting electronic pollbook data. Election day procedures analyzed in a performance audit might include checking that all pollbook paperwork was properly completed, provisional ballots were properly issued and documented, ballot containers were properly sealed, and that any ballot duplication was properly done and documented.
Precinct Results Audits
This traditional, workhorse audit involves randomly selecting a number or percentage of precincts within a jurisdiction to have the results counted in full a second time. Usually, the audit of selected precinct results is conducted by a hand count of one specific contest on the ballot. Precinct results audits are a regular part of the canvassing and certification process in many states, becoming popular 15 – 20 years ago. These audits are not considered recounts as they do not change the results but may lead to further action if a substantial deviation is found.
Precinct results audits confirm the accuracy of results and build public confidence in the ballot tabulation equipment and performance of election officials. Yet even though the precincts may be randomly selected, this type of audit is not designed to project conclusions about the accuracy of any other precincts’ results (as is the case with Risk Limiting Audits). While major discrepancies are few and far between, the question each state must consider is what the next steps should be in the case of significant deviation. One option includes selecting more precincts to be hand-counted to identify the scope of the problem. When audits are conducted after the results are certified, there is limited ability to alter the results or outcomes.
Risk Limiting Audits
Risk Limiting Audits (RLAs) are a newer form of post-election audit currently gaining traction around the country. RLAs are akin to a random sample survey, not of voters’ opinions, but of their ballots cast in an election. Using accepted statistical sampling procedures, ballots across a state or local jurisdiction are selected and the result for a specific contest on each ballot is recorded. From this, a projected statewide or countywide result is determined. The projected result is then compared to the actual election result to determine the accuracy of the tabulation. If a pre-determined confidence level is not reached, the number of sampled ballots is increased. A very close contest could evolve into a full recount as the sample size grows to satisfy the set confidence level.
RLAs assess the accuracy of the results that extends to all ballots cast, rather than only the ballots cast in specific precincts as the precinct results audit provides. The primary drawbacks of RLAs draw from their sheer complexity. Not only are they resource intensive to implement, given their complex statistical nature they are also difficult to explain to the general public—thus minimizing their effectiveness as a tool for shoring up voter confidence.
In 2020, the timing of audits became a flash point as demand grew for audits to occur before election results were certified. There are clear benefits to conducting pre-certification audits, namely that any discrepancies—though rare—would be identified before the results are certified.
The biggest obstacle to pre-certification audits is time. Today, most audits are done after results are certified. Election officials are often pressed for time to conduct all necessary canvassing tasks in the weeks following the election, leaving little time with existing resources to conduct audits. This is especially true in states with less than 14 days to canvass, as a local canvass including audits would likely require 20 days or more to complete. Some vote by mail states conduct audits before certification, as they tend to have more generous canvassing periods and the process of centrally counting and storing ballots makes it easier to conduct audits in a timely and efficient manner.
There are ways to increase the ease of pre-election audits, including extending the canvass period or utilizing the technology already embedded in most tabulation equipment. Images of each ballot made during machine tabulation at the polls or centrally offer a means to sample the universe of ballots more quickly. In addition to time and resource limitations, some claim that there are integrity issues with breaking seals and working with ballots before post-election remedies such as recounts are conducted. That said, this concern can be largely alleviated through thorough chain of custody procedures.
Like audits, there is a split in states on whether to conduct recounts pre- or post-certification. Those using unofficial results before certification will begin and end earlier. However, the local canvass may find and correct any errors in the lead up to certification that negates the need for a recount in the first place.
Elections are a zero-sum game. Someone will win, and someone will lose. However, it has become too easy for any losing candidate to claim that an election had fatal “irregularities” without any evidence.
BPC believes that one way to disincentivize these often fact-free claims is to lift the veil on how ballots are verified at every step in the process and create a counting and certification process that Americans know and trust. The processes we have in place today are far more complex and secure than most voters realize, and most policymakers and candidates acknowledge.
Given these complexities, it’s critical that election administrators and policymakers be proactive in explaining the certification process to voters. Specifically, the following three areas warrant strengthened public education, as they are especially prone to mis- and dis-information:
- Election night. States need to weigh speed and accuracy in the tense and emotional period on Election Day. Wait too long to release unofficial results and there will be some who suspect foul play. Move too quickly and administrators are prone to making mistakes inputting results into computerized systems that will haunt them for the rest of the cycle. States must ensure that election officials have the resources they need to report results, as well as clearly communicate how unofficial results reporting works to voters.
- Path to certification. The long period after Election Day, when final results are unclear and uncertified, provides ripe ground for misinformation to spread. The tasks that local and state officials perform during this period with early, mail, Election Day, provisional, and military and overseas ballots are incredibly resource intensive (not to mention integral to voter access). States and localities must be vigilant in communicating key deadlines to voters in the lead up to the final certification of results.
- Audits and recounts. While intended to uphold election integrity, audits and recounts are prone to misinformation. How states and localities conduct audits, when audits occur, who is involved, and how audits are communicated can either add or detract from confidence in the outcome. Policymakers can improve the likelihood of acceptance of the results if they are clear about the auditing process and ensure that party representatives are present throughout.
There will always be a segment of the electorate that believes something must have gone wrong if their preferred candidate did not win. Policymakers looking to strengthen the system have real options for improving how canvassing and certification are administered, but at some point, no amount of technical improvements will convince all voters that the election was legitimate. Policymakers must pair meaningful improvements to certification with rigorous public education and civic engagement to better communicate to voters how our election system already works to fight fraud and protect integrity. This careful balance of investment in election administration with civic education is what is needed to move the needle on ending this vicious cycle of recriminations between parties.
1 These exercises are not audits (official activities conducted by election authorities) as they do not meet widely acknowledged professional standards of an audit or recount.