Tactile graphics are one of the best ways for a blind person to perceive a chart using touch, but their fabrication is often costly, time-consuming, and does not lend itself to dynamic exploration. Refreshable haptic displays tend to be expensive and thus unavailable to most blind individuals. We propose TactualPlot, an approach to sensory substitution where touch interaction yields auditory (sonified) feedback. The technique relies on embodied cognition for spatial awareness - i.e., individuals can perceive 2D touch locations of their fingers with reference to other 2D locations such as the relative locations of other fingers or chart characteristics that are visualized on touchscreens. Combining touch and sound in this way yields a scalable data exploration method for scatterplots where the data density under the user's fingertips is sampled. The sample regions can optionally be scaled based on how quickly the user moves their hand. Our development of TactualPlot was informed by formative design sessions with a blind collaborator, whose practice while using tactile scatterplots caused us to expand the technique for multiple fingers. We present results from an evaluation comparing our TactualPlot interaction technique to tactile graphics printed on swell touch paper.

Inclusivity for blind and low vision (BLV) professionals in data science and analytics is limited by a gap in understanding their unique data analysis needs. We contribute to the literature by reporting on a two-step online survey delving into the experiences and challenges faced by BLV individuals engaged in data-related roles. Our fndings highlight that despite expertise in programming and GUI-based analysis tools, BLV professionals faced accessibility issues at various points in the data analysis pipeline---issues ranging from data loading and transformation, availability and compatibility of data tools with assistive technology, and visualization authoring. The prevalent use of tools such as Excel, Python, and SAS alongside heavy reliance on assistive technologies highlights persistent accessibility challenges. Furthermore, frequent collaboration with sighted colleagues indicates compromised independence. These results underscore the urgent need for "born accessible" tools that ensure the inclusivity and autonomy of BLV professionals in the feld of data science.