EZ Ballot with Multimodal Inputs and Outputs
Seunghyun Lee, Xiao Xiong, Liu Elaine Yilin, Jon Sanford · 2012 · Proceedings of the 14th International ACM SIGACCESS Conference on Computers and Accessibility (ASSETS 2012) · doi:10.1145/2384916.2384960
Summary
This demonstration paper presents the EZ Ballot, a prototype accessible voting system that fundamentally reconceptualizes ballot design to provide an equal voting experience for all voters regardless of ability. Current accessible voting machines add accessibility features on top of standard ballot designs, which paradoxically makes the experience more complex and difficult — blind and visually-impaired voters take significantly longer (31 vs. 5 minutes), ballot navigation causes confusion, and cognitive accessibility is largely ignored. The EZ Ballot takes a different approach: instead of adapting a conventional ballot, it uses a simple, consistent linear structure that breaks the entire voting process (contests, candidates, review, and casting) into individual yes/no questions presented one at a time. Each screen displays a single question both visually and through audio, such as "Do you want to vote for Barack Obama & Joe Biden for president and vice president?" This reduces cognitive demand by eliminating the need to process complex visual layouts or navigate multi-candidate lists. The system was developed at Georgia Institute of Technology's Center for Assistive Technology and Environmental Access.
Key findings
The EZ Ballot's key innovation is its universal design approach — rather than creating separate accessible modes that layer onto a standard interface, the system provides a single simple interface that works for everyone. The linear question-based structure eliminates the visual complexity of full-face ballots and the orientation problems of page-by-page ballots. Multimodal input and output interfaces are seamlessly integrated: inputs include physical buttons, iPad touch screen buttons, and gestural input (with potential for sip-and-puff and other assistive devices), while outputs include visual display and synchronized audio. This flexibility accommodates voters with visual, cognitive, and dexterity limitations without requiring separate modes or unfamiliar assistive hardware. The prototype demonstrates that the same ballot structure can work across different I/O modalities, meaning poll workers do not need to set up specialized equipment for different voters. The authors note the approach could be generalized beyond voting to other applications such as electronic kiosks.
Relevance
This paper exemplifies the universal design principle at its best: rather than retrofitting accessibility onto an existing design, it starts from scratch with a design that is inherently accessible to the widest range of users. For accessibility practitioners, the key lesson is that sometimes the most accessible solution is not adding features to an existing interface but fundamentally simplifying the interaction model. The one-question-per-screen approach directly addresses cognitive accessibility — an area often neglected in voting technology and in digital accessibility more broadly. The work also highlights that accessible design can benefit all users, not just those with disabilities, as the simplified ballot structure reduces errors and confusion for everyone. The multimodal I/O integration demonstrates how a well-designed core interaction model can be flexibly adapted across different input and output methods without changing the user experience. Accessible voting remains a critical civil rights issue, and this research offers a practical model for inclusive civic technology.
Tags: accessible voting · multimodal interface · cognitive accessibility · visual impairment · dexterity · universal design · gestural input · iPad · civic participation