Background
One of the most important civil rights laws passed in our nation's history, the Americans with Disabilities Act (ADA) guarantees that the 61 million people with disabilities in the United States have equal access to public facilities, including voting. This requirement has been interpreted to provide a guarantee that all voters, regardless of ability, have the right to vote independently and privately, whether voting in person at a polling place or by absentee ballot. Current Census data estimates that there are approximately 38 million eligible voters with a disability, representing a nearly 20% increase since 2008, and meaning voters with disabilities now make up a larger share of the electorate than voters who are black (29.9 million) and Hispanic (31.3 million). And since nearly everyone will experience temporary or permanent disability at some point in their lives, this guarantee has the potential to benefit nearly every American.
Despite the federal guarantee in the ADA, voters with disabilities continue to face barriers to voting and are too often unable to exercise their right to vote. In the 2022 general election, for example, voters with disabilities voted at a 10% lower rate than voters without disabilities of the same age. Voters with disabilities were nearly four times as likely to report difficulties voting in person as voters without disabilities, and six times more likely to have difficulty voting by mail than voters without disabilities. Voters who are blind face particular barriers to voting and continue to be the most likely to face difficulties voting, with over 38% reporting difficulty voting by mail and over half reporting difficulty voting in person in 2022.
​
Mail voting can help address some accessibility barriers, but traditional mail voting is not accessible for all voters with disabilities, particularly voters who are blind or have a print disability and cannot independently handmark a paper ballot. Mobile voting would provide a fully accessible option that ensures any voter, regardless of ability, can complete the voting process entirely independently and privately, meeting the federal guarantees in the Americans with Disabilities Act.
Accessible Voting with VoteHub
The Free Democracy Foundation (FDF), with generous grant support from Tusk Philanthropies (TP), has supported the development of VoteHub, a new mobile voting system designed to not only provide a fully accessible absentee voting system for voters with disabilities, but to also mitigate the inherent security risks when casting ballots over the internet.
VoteHub is a native mobile voting application that is designed as a digital form of absentee voting. In this way, voters follow the same procedure used to cast a ballot by mail but on their mobile device. Here are the steps voters use to access and cast a ballot with VoteHub:
-
Voters begin by providing identifying information, including their state and county/municipality, name, year of birth, and one additional identifier - either the last four digits of their social security number, their driver's license or state ID number, or their house number and zip code.
-
VoteHub uses the identifying data to match a voter record in the county's eligible voter database. Voters must confirm their record is correct, including their address and email address.
-
Voters then select whether they want to use digital or physical return. For voters using digital ballot return, they are prompted to retrieve a one-time access code from the email in their voter record and enter it into the app. This step provides the authorization to encrypt their ballot for digital return.
-
Next, voters mark their ballot. They will be prompted with warnings if they skip a contest or do not vote for all available options. They also will be unable to vote for more options than the contest permits.
-
Voters must then sign an absentee affidavit affirming their eligibility and identity. The app gives voters options to sign the affidavit, including using a finger or stylus, or, if unable to physically sign, by typing or speaking their name, uploading a photo of their signature, or making a mark. Those options may also require the voter to add a photo of acceptable identification or have a witness sign.
-
Finally, the ballot and affidavit are separately encrypted and prepared for submission to a digital ballot box. Voters can submit their ballot immediately or first perform a check to ensure it is encrypted correctly before submitting.
-
Once their ballot is submitted, voters receive a tracking code in the app and by email, which they can use to verify their ballot was received.
The ballot check process is a key security feature of VoteHub. This step enables voters to verify that their ballot is correct and that nothing has secretly changed any of their votes. To perform the check, voters must follow these steps:
-
VoteHub displays the web URL for an external verification site. Voters are directed to open the site, ideally using a second device, such as a laptop or tablet.
-
After entering the code into the verification site, voters are then directed to tap a button in VoteHub labeled "Code Entered."
-
Both VoteHub and the verification site display a passkey. Voters are directed to indicate whether the passkeys match by selecting Yes or No in VoteHub.
-
Once the passkey is confirmed, the verification site displays the decrypted ballot in plain text for voters to review.
-
If everything looks correct, voters are directed to tap Yes in VoteHub, which then re-encrypts the ballot and prepares the package for submission again. Voters can choose to perform the check again - including as many times as they want - before submitting the ballot.
-
If voters see any issue with the ballot during the check, they are prompted to answer No in VoteHub, which then directs them to use physical return.
​​
The workflow diagram below displays the complete voting process for a voter using VoteHub:
Voter Workflow Diagram
Testing VoteHub with Blind and Low Vision Voters
To ensure VoteHub is fully accessible for blind and low vision users, FDF and TP collaborated with the National Federation of the Blind (NFB) to invite members of the NFB to participate in mock elections designed to mimic a real election pilot. Two mock elections were conducted, including one on August 1, 2023, and a second from December 5-11, 2023.
Users were invited to "register" to participate in the elections ahead of time. NFB contacted members by email in the weeks before with an invitation and a link to a Google form through which participants could register. Here is a sample of the email NFB sent to recruit participants:
Help Test Accessible Mobile Voting Technology
If you are looking forward to the day when you can vote using your smartphone, here is your opportunity to help make that desire a reality! Tusk Philanthropies is hosting a mock election on Tuesday, August 1st to test a new mobile voting app for iOS and Android devices designed to make accessible absentee voting more secure. Want to participate? Sign up at the link below by 5pm ET on Friday, July 28. Once you sign up, you'll receive instructions via email from mock election sponsor, Tusk Philanthropies, on how to install the app on your device and participate.
https://forms.gle/xwu3psq5EYZ7yJEt5
​
A total of 207 people signed up to participate in the August election and 165 signed up for the December election. Sixty-three people signed up to participate in both elections. The sign-up form asked users to provide their first and last name and email address. In the December election, participants also had the option to provide their mobile phone numbers if they wished to receive text alerts about the election. Of the 165 who signed up, 131 also provided phone numbers.
​
While the goal in both mock elections was to mimic as much as possible a real-world election, we opted to not use any personal data from participants. Instead, we provided users with "mock" registration information they could use to perform voter lookup and verify their registration. Each user was given the same data in instructions delivered by email on the day voting opened. This may have made the voter lookup process more complicated, particularly for users using voice over and other accessible technology, since it required voters to refer to an email to retrieve their assigned year of birth and social security number. In a real election, voters would know this information and likely not need to retrieve it from another source.
We used the same ballot for both elections, and all voters were presented with the same ballot content. The ballot contained three contests, including two candidate contests and one ballot measure. One of the candidate contests was a multi-candidate winner in which voters could vote for up to two candidates, while the other enabled voters to vote for one person. Both candidate contests included six candidates plus a write-in candidate option. See Appendix C for screen shots of the complete mock election ballot.
​
To install the VoteHub mobile app, users had different experiences between Android and iOS, and between the August and December elections. For iOS users, the mobile app was not available in the Apple app store in either election and could only be installed through TestFlight, Apple's service for beta testing applications. In August, users had to receive a separate email from TestFlight in order to install the app. In December, we used a public link to the TestFlight app, which made it much easier for users to install. Android users had an easier experience in both elections. The VoteHub app was available through the Google Play Store, and emails with instructions on how to participate included a link to install the app.
Voters received similar notices in both elections with instructions on how to vote. The email communication was designed to mimic the type of communication an election official may provide voters eligible to use mobile voting. The email included instructions (and links, when available) to install VoteHub, as well as their assigned "mock" voter registration information needed to use the app. The data included the last four digits of their mock social security number, driver's license number, and address. The email also contained a voting checklist and other helpful links to the verifier website to perform the ballot check and track their ballot (see a sample from December in Appendix A).
In the December election, voters also received a pre-election email one day before voting opened that contained instructions on how to install the app as well as helpful information about what to have ready when voting opened, such as access to their email, a second device to perform the verification (if available), and links to video demos showing how the app works. See Appendix B for a sample of this email. This email was sent based on feedback from the August election in which users indicated they would have preferred to know ahead of time what resources they needed to vote.
Also in December, voters who opted in received text messages notifying them when voting opened as well as a reminder text on the last day of the election. The text messages directed voters to check their email for instructions.
Results and Findings
For the August election, voting occurred on a single day, with the election opening at 8:00am ET, when voters received email notification, and closed at 7:00pm ET that evening. Over the course of the day, a total of 52 voters submitted ballots, or 25% of the voters who signed up.
During the election, we received a handful of reports by email from voters experiencing problems using the app. The biggest problem voters experienced was during ballot marking. The voting app requires voters to scroll through a complete list of candidates or ballot measure text before they can advance to the next contest, regardless of whether or not they have made the maximum number of selections. For voters using voice-over, this action requires a three-finger swipe gesture to scroll. However, many voters reported the gesture failed to activate the next button and they became stuck.
​
After investigating the issue, our developers discovered the swipe gesture would sometimes stop just above the bottom of the screen. They were able to fix the issue, and subsequent testing in December showed the improved usability.
Another issue reported over the course of the day involved communication from TestFlight through Apple's platform. Voters reported they never received a separate email from TestFlight inviting them to install VoteHub. We experienced this problem in other mock election tests with young voters. The use of a public TestFlight link in December resolved this problem.
​
For the December election, voting opened at 8:00am ET on Tuesday, December 5 and closed at 5:00pm ET on Monday, December 11. Over the course of the six days during which the election was open, a total of 63 voters submitted ballots, or 38% of the voters who signed up. We did find evidence that voters who signed up to receive text alerts about the election voted at a slightly higher rate (39%) than voters who did not (35%).
​
​Overall, we received fewer reports of issues from users in the December election. The testing did reveal one issue related to an accessibility feature. Early on the first day of voting, we heard from several voters who reported being unable to activate the menu to choose their state and county. After investigating, including through a video call with one user, we discovered the problem was caused by an error in the app when users had turned on reduced motion in their accessibility settings. Voters were able to fix the issue by turning that feature off. Our developers also fixed the issue quickly and we released an updated app before the end of the first day.
​
One of the most impressive findings in the mock elections was the number of voters who successfully performed the optional ballot check. Seventy-two voters completed the check across the two elections. That's over 62 percent of the 115 who submitted ballots, including 32 of the 52 voters in August and 40 of the 63 voters in December. This figure is far above the average 43% we have seen in mock elections with other voting groups. And it is well beyond the minimum any security expert has suggested would be needed to be assured of the security of the ballots cast using VoteHub.
Survey feedback in both elections further reinforced our conclusion that the ballot check process was relatively easy for voters to complete. Voter responses in the surveys noted that it was very or somewhat easy, particularly when using a second device to perform the check. Several voters commented in the survey and in subsequent focus groups that the process would be easier if we provided a direct link to the verification site through the app, and if we enabled voters to copy and paste the checking code. The app was designed intentionally to not link directly to the verification site, in part to encourage the use of a second device and because a malicious attacker could misdirect voters to a corrupt verification site, nullifying the benefit of the check. Similarly, we have not enabled a copy and paste of the code again to encourage users to perform the check on a second device.
It is worth noting that we do not expect to see ballot check rates at these levels in public elections. Voters in these mock elections were aware they were testing new voting equipment, and consequently may have felt compelled to test every feature. Some users reminded us in focus groups that they are also highly skilled and experienced technology users and therefore may not be representative of the broader blind and low vision population. Further study in public elections with a broader set of users will be necessary in order to determine if we can maintain necessary levels of ballot verification to ensure the integrity of elections. But these early tests provide encouraging evidence that we have not only made the process usable for first-time young voters, but also for blind and low vision voters.
Complete survey feedback and findings are presented in Appendix D.
Next Steps
Following these tests with blind and low vision users, Free Democracy Foundation's engineers and designers continue to make improvements to the user experience to enhance the accessibility and continue to improve the application. Several of the participants have offered to serve as beta users and provide continued retesting and feedback as we make further updates to the technology. Additional testing is also underway with voters with physical and cognitive disabilities in collaboration with the University of Colorado at Denver.
Free Democracy Foundation is also investigating opportunities to collaborate with digital identity service providers to explore alternative voter verification tools that could be offered in place of signatures. Given the signature process remains the source of the most friction for voters with vision impairment, we hope to collaborate with election officials, policy makers, and disability advocates to explore alternative strategies that will not only make VoteHub even more accessible, but will also provide stronger verification methods to ensure only eligible voters are able to vote.