Impact of universal design ballot interfaces on voting performance and satisfaction of people with and without vision loss

Since the Help America Vote Act (HAVA) in 2002 that addressed improvements to voting systems and voter access through the use of electronic technologies, electronic voting systems have improved in U.S. elections. However, voters with disabilities have been disappointed and frustrated, because they have not been able to vote privately and independently (Runyan, 2007). Voting accessibility for individuals with disabilities has generally been accomplished through specialized designs, providing the addition of alternative inputs (e.g., headphones with tactile keypad for audio output, sip-and-puff) and outputs (e.g., audio output) to existing hardware and/or software architecture. However, while the add-on features may technically be accessible, they are often complex and difficult for poll workers to set up and require more time for targeted voters with disabilities to use compared to the direct touch that enable voters without disabilities to select any candidate in a particular contest at any time. To address the complexities and inequities with the accessible alternatives, a universal design (UD) approach was used to design two experimental ballot interfaces, namely EZ Ballot and QUICK Ballot, that seamlessly integrate accessible features (e.g., audio output) based on the goal of designing one voting system for all. EZ Ballot presents information linearly (i.e., one candidate’s name at a time) and voters can choose Yes or No inputs that does not require search (i.e., finding a particular name). QUICK Ballot presents multiple names that allow users to choose a name using direct-touch or gesture-touch interactions (e.g., the drag and lift gesture). Despite the same goal of providing one type of voting system for all voters, each ballot has a unique selection and navigation process designed to facilitate access and participation in voting. Thus, my proposed research plan was to examine the effectiveness of the two UD ballots primarily with respect to their different ballot structures in facilitating voting performance and satisfaction for people with a range of visual abilities including those with blindness or vision loss. The findings from this work show that voters with a range of visual abilities were able to use both ballots independently. However, as expected, the voter performance and preferences of each ballot interface differed by voters through the range of visual abilities. While non-sighted voters made fewer errors on the linear ballot (EZ Ballot), partially-sighted and sighted voters completed the random access ballot (QUICK Ballot) in less time. In addition, a higher percentage of non-sighted participants preferred the linear ballot, and a higher percentage of sighted participants preferred the random ballot. The main contributions of this work are in: 1) utilizing UD principles to design ballot interfaces that can be differentially usable by voters with a range of abilities; 2) demonstrating the feasibility of two UD ballot interfaces by voters with a range of visual abilities; 3) providing an impact for people with a range of visual abilities on other applications. The study suggests that the two ballots, both designed according to UD principles but with different weighting of principles, can be differentially usable by individuals with a range of visual abilities. This approach clearly distinguishes this work from previous efforts, which have focused on developing one UD solution for everyone because UD does not dictate a single solution for everyone (e.g., a one-size-fits-all approach), but rather supports flexibility in use that provide a new perspective into human-computer interaction (Stephanidis, 2001).Ph.D