Sound, Touch, or the Full Monty? A Comparative Study of Accessible Data Exploration Systems for Blind Users.

Blind and low-vision (BLV) individuals encounter significant challenges in accessing and interpreting data, a critical component in many data-intensive fields. Traditional screen readers, which convert text to speech, have been the primary solution, but recent developments such as the Monarch refreshable tactile display and hybrid approaches such as TactualPlot, which combines sound and touch, offer new possibilities for data accessibility. In this paper, we present a controlled, within-subject comparative study of three accessible data exploration systems: a screen-reader-based chart description interface (Olli), a tablet-based audio-tactile system (TactualPlot), and a multi-line refreshable tactile display (Monarch). Ten blind participants completed data analysis tasks involving bar charts, pie charts, line charts, and scatterplots, spanning single-point identification, pairwise comparison, and trend analysis. We evaluated performance using task accuracy and completion time, subjective workload using NASA-TLX, and qualitative feedback through think-aloud protocols. Results indicate that task accuracy did not differ significantly across systems, while completion time varied substantially, with participants completing tasks fastest on the refreshable tactile display. NASA-TLX ratings revealed significant differences in perceived workload across systems, particularly for mental demand, effort, and temporal demand. Qualitative findings highlight how system-specific interaction techniques, hardware constraints, and feedback mappings shape user strategies, learning curves, and perceived efficiency, especially for complex chart types such as line charts and scatterplots. Rather than isolating sensory modalities in abstraction, our findings demonstrate how accessible data exploration is mediated by device-level design choices and interaction techniques that operationalize audio and touch in practice. We conclude with design implications for accessible visualization systems and guidance for future research on integrating audio, tactile, and speech-based interaction techniques to better support non-visual data analysis.