Comparison of Reading Accuracy between Tactile Pie Charts and Tactile Band Charts
Kosuke Araki, Tetsuya Watanabe · 2013 · Proceedings of the 15th International ACM SIGACCESS Conference on Computers and Accessibility (Assets '13) · doi:10.1145/2513383.2513404
Summary
This short paper compares the accuracy and efficiency of reading tactile pie charts versus tactile band charts for determining proportional data through touch. The authors are developing software that automatically generates tactile charts from CSV data and needed to determine which chart type is more suitable for representing proportional information in embossed format. Pie charts encode proportions through the angle and area of circular segments, while band charts (similar to stacked bar charts) encode them through the length of rectangular segments. Two experiments were conducted. The first involved 20 sighted undergraduate and graduate students who were blindfolded and presented with 20 tactile pie charts and 20 tactile band charts created with a Braille embosser. Each chart contained two data divisions, and participants were asked to estimate the proportion of the first element. Charts were presented without labels, titles, or legends to isolate the tactile reading ability from other cues. The second experiment involved one blind graduate student familiar with Braille, using 99 stimuli per chart type (1% to 99% in 1% increments) to more finely examine the error patterns observed in the first experiment.
Key findings
For both sighted (blindfolded) and blind participants, pie charts produced smaller reading errors than band charts across most proportions. Both chart types showed error patterns related to "reference points" — specific positions on the chart that are easily identifiable by touch. Pie charts have five reference points (0%, 25%, 50%, 75%, and 100%, corresponding to the top, sides, bottom, and middle positions of the circle), while band charts have only three (0%, 50%, and 100%). Error sizes decreased near these reference points and increased with distance from them, creating a characteristic ripple pattern. The sighted participants also showed a tendency to round answers to 5% increments. Mean reading times were similar: 12.1 seconds for pie charts versus 11.8 seconds for band charts among sighted participants (not significantly different). The blind participant read both chart types substantially faster — 4.09 seconds for pie charts and 3.54 seconds for band charts — roughly one-third the time of sighted participants, likely reflecting their tactile expertise. The authors suggest that adding grids, ticks, and labels (as in real-world tactile charts) could further reduce errors, and that 10% step ticks might keep reading errors below 5%.
Relevance
This paper contributes to the evidence base for designing accessible data visualizations for blind and low vision users. As data literacy becomes increasingly important, ensuring that blind students and professionals can access graphical information is essential. The finding that pie charts offer more tactile reference points — and therefore better accuracy — than band charts provides practical guidance for educators, textbook publishers, and developers of tactile graphics software. The dramatic speed difference between the blind participant and blindfolded sighted participants underscores the expertise that experienced tactile readers develop, suggesting that chart design should be optimized for skilled tactile users rather than defaulting to assumptions based on sighted perception. However, the study is limited by having only one blind participant, and the authors acknowledge that more blind participants are needed. The work supports the broader goal of making STEM education accessible by providing empirical data on which chart representations work best for tactile reading.
Tags: tactile graphics · data visualization · blind users · braille · pie charts · band charts · tactile perception · accessible graphics · education