There is a common refrain that, in an ideal election, it is easy to vote and hard to cheat. Applying that framework to the modern American voting experience indicates that a perfect remote identity verification policy meets two key goals: 1) accuracy, ensuring that the verification method filters out all ineligible ballots without disqualifying any valid votes; and 2) accessibility, guaranteeing that the verification process doesn’t establish requirements that restrict access to the ballot box. While this is the ideal, in practice these goals often conflict with one-another. For instance, some mail voting verification measures—like requiring a photocopy of one’s ID card—can be immensely restrictive.
U.S elections take place in the context of a stratified society, wherein race, class, and ability can all impact an individual’s access to voting and likelihood of having their vote accepted and tabulated. Verification policies need to actively counter contextual disparities, rather than passively perpetuate or exacerbate them. Additionally, given that under-resourcing for state and local elections infrastructure is a chronic, national issue, verification measures must be feasible for elections jurisdictions to implement and administer.
Signature matching—the most widely-used identity verification process for absentee ballots— continues to be the “gold standard” of remote voter identity verification. As mail voting ramped up to unprecedented levels during the COVID-19 pandemic, the accuracy and integrity of signature matching came under public review. While some of the distrust around signature verification has been blown out of proportion and rates of voter fraud are extremely low, signature verification does have a tendency to over-correct. Signature verification tends to result in higher rates of ballot rejection for young, lower-income, and voters of color. While signature matching remains the best policy alternative to requiring photo IDs when voting by mail, these concerns present an opportunity to revisit the utility of signature verification procedures in the modern context, and perhaps to reimagine how elections administrators can and should go about verifying voters’ identities during mail voting.
Looking beyond the elections policy space, technological innovation has enabled actors and institutions in a variety of industries to accurately and efficiently discern individuals’ identities. Additionally, supplemental policies like extended pre-processing and ballot curing procedures may effectively bolster signature matching and ensure that the voting experience is secure and accessible.
The 2020 election was the most secure in our nation’s history. Yet in response to false claims of rampant voter fraud, some state legislatures have introduced strict identification requirements for mail voting. These identification requirements too often hamper voter access without addresssing the most pressing security threats facing our elections system: sustained underfunding and outdated cybersecurity. Our desire to explore identification measures in this paper is born out of a desire to expand voters’ access to the ballot while protecting against ever-evolving future threats. An election system which is free, fair, and accessible must not be static; an election system that truly meets voters’ needs must be flexible and responsive. This paper is just the beginning of scoping out what a modernized remote voter identity verification system might look like.
As the nation trends towards a wider reliance on convenience voting methods, this report explores how states and localities can ensure that their voter verification policies achieve the nexus of accuracy, accessibility, equity, and practical feasibility. This paper is not intended to provide specific recommendations about how election officials should be conducting identity verification. Rather, it provides a survey of the major benefits and drawbacks of the policy alternatives in use today, as well as the methods that might gain traction down the road.
The United States electoral system is highly decentralized, with great discretion for election procedures at the state and local levels. This section provides a high-level, inexhaustive overview of the most common remote voter verification policies in use today. We primarily focus on the verification methods used to verify voted mail ballots, but many of the conclusions also apply to identity verification for absentee ballot applications.
Signature verification is the most commonly used method of remote voter identity verification. All states require that voters provide a signature when voting by mail, and at least 36 states rely on signature verification to confirm the voter’s identity. Signature verification, also referred to as signature matching, is the process by which a signature on a ballot envelope is compared against a signature on file with the elections office to confirm a voter’s identity.
Signature verification is typically conducted in multiple stages. Most jurisdictions rely on two-person teams to manually compare signatures. For some larger jurisdictions, the first stage instead involves using Automatic Signature Verification software (ASV) to identify discrepancies between the signature on the ballot envelope and a signature on file. Ballots that do not meet a certain match threshold are then fed into a second stage of review, which generally involves manual review by trained inspectors. Signatures rejected during manual inspection (stage one for jurisdictions without ASV and stage two for jurisdictions with ASV) usually then undergo a final level of review, in which the signature on the ballot envelope is compared against older signatures from the voter’s record.
The amount of resources expended by the election office increases with each stage; while using automatic verification software (when available) has a low per ballot cost, this cost rises as election workers’ time is spent conducting manual review. Election workers must be trained in signature comparison best practices, thus increasing the cost and difficulty associated with this policy. Furthermore, the chronic underfunding of elections and recent uptick in mail voting has resulted in many offices using untrained staff for signature matching, risking an increased likelihood of erroneously rejected signatures.
Disproportionate rates of ballot rejection among marginalized communities is one of the primary drawbacks of signature verification. Recent studies have sparked concerns that signature verification erroneously screens out some valid ballots, and that ballots that are cast by young voters and voters of color are rejected at higher rates. Litigation in California revealed that “[i]n the 2016 general election, the statewide average rejection rates were 0.88% for Latino voters and 0.61% for Asian-American voters, versus the 0.45% statewide rejection rate for non-Latino, non-Asian votes.” Analysis of Florida’s 2020 presidential preference primary conducted by the Stanford-MIT Healthy Elections Project revealed similar findings. Their study found that young and first-time voters, as well as Hispanic, Black, and other minority voters, were about twice as likely to have their mail ballots not counted. Their study was not able to differentiate the cause of the ballot rejection, and these rates include ballots rejected for late arrival and other non-signature related defects.
Additionally, there is a disparity between large and small jurisdictions regarding the baseline signature used to verify the return envelope. Smaller jurisdictions generally have only one signature to compare to, usually from the voter’s driver’s license. Larger jurisdictions often have automated equipment that captures (and stores in the voter registration system) the images of signatures on every accepted ballot. As a signature evolves over time, a large county will have images of that evolution, whereas smaller jurisdictions might only have one static image until the changes get so significant that the voter has a signature rejected and must cure the signature.
One benefit of signatures as an identification tool is that they are less widely available than personal identification numbers. Signatures, unlike ID numbers, are not easily sourced online. Even if found, a nefarious actor would have to forge the signature, increasing the likelihood that the ballot would be rejected.
Signature matching is widely regarded as the best way to balance security and accessibility in mail voting, but only when it is paired with robust curing measures to protect voters from undue ballot rejections. The BPC Task Force on Elections recommends that states both notify voters of rejected ballots and give them sufficient time to cure eligibility deficiencies, even if that period extends beyond Election Day. To shorten the turnaround time involved in ballot curing, the task force encourages election officials to collect, but not make publicly available, multiple points of contact for voters. Additional best practices for signature verification include regularly updating signature databases and publishing signature verification standards to increase transparency, as well as importing signature files from driver’s licenses and state IDs.
Signature curing can be costly for election offices. In total, a jurisdiction can expect to spend as much as $26.47 per ballot in resources and staff time on both signature verification and curing. While there is inconsistent data on how many ballots must undergo ballot curing in each election, anecdotal evidence from BPC’s Task Force on Elections suggests that as the quality of handwritten signatures is decreasing with technological advancements, more signatures are being rejected and thus more curing is needed.
The table below provides cost estimates for signature verification with a ballot cure. This table was compiled by Paddy McGuire, Auditor of Mason County, Washington, and accounts for the salary and compensation of the election worker performing the task at each stage, the time it takes to conduct each task, and the cost of any supplies associated with curing the signature.
Personal Identification Number Requirements
In addition to providing valid signatures, some states require voters to submit their personal identification numbers for identity verification when voting by mail. Voters’ identification numbers—including driver’s license numbers, identification card numbers, and social security numbers—are compared against the numbers on file to approve or reject a ballot.
This policy option gained notoriety with the passage of Georgia’s 2021 election bill, which replaced reliance on signature matching with the use of personal identification numbers. According to the National Conference of State Legislatures, Minnesota, Ohio, and Kansas are among the states that use personal identification numbers to verify absentee ballots as of April 2020. Unlike Georgia, each of these three states conduct both signature and identification number verification (Minnesota only conducts signature verification if the identification number does not match that which is on file and also requires a witness signature). Other states, such as Pennsylvania, require that personal identification numbers be provided in the absentee ballot application, but not on the ballot return envelope.
Requiring that voters supply their personal identification number on their outer ballot envelope has both security and privacy concerns. Despite some state and federal data protection efforts, it is fairly easy for nefarious actors to obtain personal identification numbers online. Election fraud is incredibly rare, but in theory, nefarious actors could use this information to submit false ballots. Furthermore, even if we assume that identification numbers are private, the act of supplying that information on a ballot envelope risks their disclosure. Voters are typically required to verify their ballot on the ballot’s outer envelope (not on the ballot itself to preserve the privacy of one’s vote choices). However, by requiring voters to supply their personal identification numbers on the ballot’s outer envelope, in order to vote by mail the voter has to risk leaking that information to whoever processes or otherwise interacts with that piece of mail. Requiring the last four digits of a voter’s social security number might be more secure than requiring their driver’s license or identification card number, but the stakes of accidentally leaking even a partial social is even higher. Some ballot envelope designs provide a privacy flap that protects voters’ personal information. These envelopes should be utilized whenever election officials request personal identification numbers on a ballot envelope.
Despite the security and privacy concerns, this policy option scores high on accessibility, especially when compared against requiring voters to provide a copy of their identification documents. Even voters who do not have a photo ID should have access to their social security number, and enabling voters to provide the number without needing a copy removes an additional barrier to voting that traditionally restricts lower-income groups.
Copy of Identification Documents and Notary or Witness Requirements
Some states require voters to submit a copy of their photo identification card, or other accepted identification document when voting or applying to vote absentee. Arkansas is one of the few states that currently requires a copy of a voter’s personal identification with their voted ballot, but this policy option has gained popularity as state legislatures revise their election laws in the wake of 2020. Requiring that copies of identification documents be submitted with an absentee ballot or absentee ballot application hinders voter access without proportionate improvements to election security.
This policy creates burdensome barriers to absentee voting that disproportionately exclude lower income populations. Not only are certain forms of photo identification less widespread among lower income communities, young people, and communities of color, requiring a printed copy of identification documents further restricts access. Many households lack the tools necessary to make copies at home, and getting copies made commercially incurs costs onto the voter. Absentee voting was originally conceived to extend voting access to individuals unable to vote in-person. Given that for some voters procuring photocopies involves paying for copies in a public venue with varying levels of access for individuals with disabilities, this policy option fails to uphold the level of accessibility absentee voting aspires to achieve.
The accessibility limitations of document copies extend to witness and notary requirements as well. This policy requires voters to provide a valid witness or notary signature on their ballot envelope or absentee ballot application. Witness and notary requirements are not proven to increase security and are especially challenging for lower-income voters and voters with disabilities who may not have the time, resources, or ability to attain a witness or notary signature, which usually costs about $10 per signature.
Witness and notary requirements are often paired with other verification measures, such as personal identification numbers, document copies, and signature matching. To apply for an absentee ballot in Minnesota, for instance, a voter must provide their personal identification number in the application. When voting, the voter must have a valid witness (registered voter or official notary) sign the ballot envelope to verify that the ballot was delivered blank and voted in private. South Dakota, in contrast, allows voters to submit either a copy of their photo identification or a notary signature when applying to vote absentee. Submitted ballots then undergo signature verification but do not require any additional proof of identity.
While less resource-intensive for election officials than signature matching, there is still significant staff time and resources involved in reviewing submitted documentation and curing any shortcomings or discrepancies.
BPC spoke with representatives from four leading identity verification companies to discover which identity verification practices used in the private sector might be applicable to mail voting. Three practices emerged as having the most utility for remote voter identity verification: data verification, identity document verification, and two-factor authentication. Additionally, each company recommended the use of a “step-up” system that combines two or more practices to reach as many voters as possible.
This section provides an overview of how these approaches might be implemented in the elections space.
Third Party Identity Verification
It is common practice for private businesses to outsource identity verification to third party companies. Identity verification companies have expansive data stores that allow them to confirm consumers’ identity more accurately and efficiently than businesses can on their own. Private businesses are not alone, and a variety of government bodies also outsource their identity verification efforts.
Identity verification companies employ a variety of methods to confirm identity, from data verification to biometric analysis. Data verification, identity document verification, and two-factor verification are the three of the most promising approaches for election administration.
- Data Verification is the simplest form of identity verification, in which a user’s personally-identifiable information—such as their name, date of birth, and last four digits of their social security number—is compared against information on file. If the information matches, their identity is confirmed. Additional metrics can be added to data verification to enhance security—such as questions about addresses or persons associated with the user—but given the extremely low prevalence of voter fraud, it is unlikely this level of security is necessary for elections. Election jurisdictions who collect voters’ personal identification numbers to verify identity are conducting data verification. For the purpose of this paper, we distinguish data verification managed by a third party company from the collection of personal ID numbers by election officials. While the process is broadly the same, third party companies leverage multiple data sources and provide a streamlined, outsourced approach to data verification that differentiates it from that which election officials perform.
- Identity Document Verification requires a user to scan or otherwise provide a copy of their identity documents, such as a utility bill, passport, state, or other government-issued ID. This scanned document is then analyzed for validity and, if valid, is used to confirm the users’ identity.
- Two-Factor Authentication relies on a trusted device or means of contact to verify a voter. Most commonly, two-factor authentication involves a code being sent to a phone number or email address associated with an individual. This code is then used to verify the identity of the person in question. Two-factor authentication could be used in conjunction with voter registration and ballot tracking systems. For example, when a voter signs their mail ballot envelope, they could be prompted to go to the trusted voter registration or ballot tracking website to retrieve a one-time code to put on their ballot envelope.
Third party identity verification has the potential to improve the accuracy of voter verification. While there is an extremely low rate of voter fraud, processes like signature verification do result in rejected ballots that must be dealt with manually. Curing deficiencies in rejected ballots takes significant time and resource investments from election workers, exacerbating the ecosystem-level costs of signature verification as an identity verification tool. Third party identity verification companies are highly specialized and may be able to perform remote voter identity verification in a more efficient, secure, and user-friendly manner than election officials are able to do on their own.
Third party identity verification is common practice in the public and private spheres, and most voters likely have some prior experience using these tools. Signature verification can sometimes be murky and difficult to explain, but if a cure is not required the process itself is simple – voters must simply sign their ballot envelope. Third party identity verification has the benefit of being a trusted and secure alternative, but it would likely add steps to the identity verification process for voters. The added degree of security could improve voters’ confidence and trust in the electoral system, but this increased security must not compromise the ease of voting.
Additionally, data verification alone—the simplest and most elections-friendly option—does not meet the needs of all voters. While some voters might find data verification more trustworthy than signature matching, others lack consistent internet access or are likely to be uncomfortable sharing their data with government entities. Furthermore, third party identity verification often relies on a users’ financial history, which proves difficult for young or underbanked voters without an extensive banking trail. It is critical that such voters be given an equitable alternative they are comfortable using to preserve voter trust and engagement.
Data verification covers about 80% of the U.S. population, and thus requires that it be paired with a secondary verification system for individuals who cannot be verified the first time around. Identity verification companies typically funnel persons not verified through data verification into a secondary identity verification procedure, often identity document verification. Yet due to the equity concerns of voter-facing photo ID and photo ID photocopy requirements, identity document verification is not a feasible policy option for all election jurisdictions. The next section on Multi-Tiered or “Step-Up” Verification Systems explains how a mixed methods approach could address these concerns and better meet the dynamic needs of a diverse population.
Mixed Methods: “Step-Up” Verification Systems
The core principle of election administration is that all eligible voters should be able to cast a ballot freely and fairly. Data verification, when conducted through a trusted third party, is effective for about 80% of the American public. Data verification alone cannot replace signature verification, but a mixed-methods approach holds promise for both a more secure and a more efficient remote voter verification system.
Election officials hoping to modernize their identity verification systems might consider extending third party verification to voters whose signatures could not be verified. For example, a voter whose signature is not verified could be prompted to verify their identity through data verification or two-factor authentication. When a signature is rejected, election officials could notify the voter and provide them with a link (or other chosen mechanism) to confirm their identity in an online portal.
This sort of “Step-Up” approach (in which voters who do not pass one type of verification are passed onto another) could be further embedded within a jurisdiction’s ballot tracking system. Ballot tracking systems provide a trusted mode of communication with voters and provide a way for election officials to seamlessly introduce modernized security measures. Furthermore, many identity verification companies provide clients with verification widgets that can easily be embedded into existing websites. These widgets come enhanced with extensive security protections, and thus do not run the same risk of hacking or interference that many under-resourced elections websites face. They also have the benefit of being embedded into a trusted .gov website with the election office’s branding.
Tiered identity vericiation processes are common practice in the private industry and, while seemingly burdensome, is something the public is growing increasingly familiar with and stands to mitigate the high resource cost of signature curing.
Despite increased access to technology and growing rates of tech-literacy among the American public, a remote voter verification system must be able to accommodate voters without broadband access. A mixed-methods approach could help ensure these populations are reached without any additional barriers. While, depending on implementation, data verification can take place without an online interface, election officials opting for more tech-heavy solutions must provide an alternative to voters without a mobile phone or internet access. For example, if data verification or two-factor authentication were used in place of signature curing for most voters, traditional signature curing could still be used for voters without broadband access. This mixed-methods setup would reduce the resource demand on election officials by reducing the quantity of voters needing signature curing, while still ensuring that no voter is excluded from the process.
While the idea of working with a third party company may sound daunting to jurisdictions with a long history of signature verification, working with an outside group could actually lower costs and reduce the administrative overhead associated with signature curing. The total costs associated with curing a ballot signature that does not pass the first round of review can be as high as $26.47 per ballot. In contrast, assuming a voter had to go through two levels of review to confirm their identity (data verification and two-factor authentication), the total cost when working with a third party company would be roughly $1.80. While true costs will vary by jurisdiction, voter, and company, this example shows how outsourcing identity verification to companies specialized in the field could mitigate the high resource strain of signature curing on election offices.
It is worth noting, however, that while third party verification may be more cost effective for ballots that require a cure, the average cost of signature verification for accepted signatures is only about $0.06 per ballot,* significantly more cost effective than the third party alternatives. While at this point there is insufficient data on how many ballots require a cure to make a cost-benefit argument for or against third party verification, at this early stage pairing signature verification with third party verification for rejected signatures appears to be the most cost effective alternative.
The status quo of voting in America is shifting away from predominately in-person voting on Election Day to a near even balance of early, mail, and Election Day voting. In this period of flux, it is critical that our voting system not backtrack on ease and accessibility, and that the remote verification policies implemented by policymakers promote both security and access.
After a contentious 2020 election season marked by rampant disinformation, state legislatures are implementing reactionary remote voter identity verification reforms. Strict identification requirements, such as requiring an ID copy to vote absentee, are an outsized response to equally outsized claims of fraud.
Rather than provide a single solution, the intention of this paper is to broaden the conversation happening around remote voter verification alternatives. By framing out the risks, benefits, and costs of each of the primary modes of identity verification we hope that policymakers will look beyond a purely reactionary stance and instead vouch for the policy alternatives that best promote a free and fair election system.