“They Aren‘t Built For Me“: An Exploratory Study of Strategies for Measurement of Graphical Primitives in Tactile Graphics

Dylan Cashman

Brandeis University

Nam Wook Kim

Boston College

Lane Harrison

Worcester Polytechnic Institute

Areen Khalaila

Tulane University

Overview

In this study, we explore strategies used by Blind and Low Vision (BLV) people to read tactile graphics.

​

• Motivation
• Methodology
• Study 1: Replication of Cleveland & McGill
• Study 2: Group Interview
• Findings
• Takeaways

Visual Chart (Cleveland & McGill)

Tactile Version Used in Our Study

Group Interview

Why is Accessibility So Important?

• Visualization is ubiquitous in education and the workplace
• Visualizations have been part of the College Entrance Exams (SATs, ACTs) since 1994¹, with entire sections dedicated to data analysis in recent years
• Data skills are listed in almost any job requirements�
• 8.7 million individuals living in the U.S. with vision difficulty²
• Self-reported that they "are blind or have serious difficulty seeing, even when wearing glasses"

¹ Source: manhattanreview.com

​

²Source: American Federation for the Blind

​

Unemployment among blind and low-vision people is 2–3× higher than the general population

Journal of Visual Impairment & Blindness (2019)

34% job growth projected for data-science roles over the next decade�

U.S. Bureau of Labor Statistics (2019)

​

While visualizations have become integral to K–12 education and data-related professions, their accessibility remains limited�

Background

• Screen reader descriptions

​

​

​

​

• Novel media (sonification, etc.)

• Multimodal graphics

​

​

​

​

EUROVIS 2022

CHI 2023

CHI 2024

ASSETS 2002

CHI 2024

CHI 2022

VIS 2024

CHI 2024

VIS 2022

”Raised line drawings, called tactile graphics, are widely used in the education of BLV students and in orientation and mobility (O&M) training.” �— Reinders et al., IEEE TVCG 2025

Refreshable Tactile Display

Braille Embossers 

3D Printers

Swell Paper

Background

Nemzer (2018)

Should a tactile representation replicate its visual counterpart to be effective?

To investigate this, we observe and interview BLV participants reading graphical primitives.

​

Cleveland McGill used as controlled testbed

• Comparisons with previous replications
• Elementary perceptual tasks

Note: Visual and Tactile perception have different physical constraints and neural perceptual systems!

Experimental Design – Study 1

• 11 BLV participants
• Up to 80 trials each (n=651 total)
• 4 chart types
• 20 samples per type, 50-95% ground truth
• Designed with help from Perkins School for the Blind and printed on Swell Paper
• 30 minute sessions
• Quantitative + Qualitative data
• Error from ground truth
• Think aloud describing strategies
• Interview afterward

​

Experimental Design – Study 2

• 4 BLV participants
• 90 minute group interview
• Procedure:
• Talked about each chart type
• Each participant gave their strategies
• ...then responded to others' strategies
• Qualitative Data:
• Recordings
• Photos
• Video

​

Findings - Quantitative

• Performance appears similar to previous studies
• Laboratory study Cleveland McGill 1986
• Crowdsourced study Heer & Bostock 2010
• Observational data suggested followup studies on additional features
• Distractors (Talbot 2014)
• Education level
• Visual Memory

​

Log Error Score

Bar

Stacked Bar

Pie

Bubble

More Error

Less Error

Findings - Qualitative

• Thematic analysis
• Strategies
• Difficulties and Frustrations��
• Gaps in Current VIS
• Guidelines for Addressing Gaps

​

​

Findings - Qualitative

• Thematic analysis
• Strategies
• Difficulties and Frustrations��
• Gaps in Current VIS
• Guidelines for Addressing Gaps

​

​

Findings - Qualitative

• Thematic analysis
• Strategies
• Difficulties and Frustrations��
• Gaps in Current VIS
• Guidelines for Addressing Gaps

​

​

G1: Physical Constraints on Size and Orientation of Primitives

• Hand Size
• Number of Hands
• Orientation/Rotation

G2: No Clear Area Measurement

G3: Local Exploration Hinders Object Identification

• Related to “Midas Touch”�

Findings - Qualitative

• Thematic analysis
• Strategies
• Difficulties and Frustrations��
• Gaps in Current VIS
• Guidelines for Addressing Gaps

​

​

More Direct Encoding and Double Encoding

• Direct Encoding of Area as Number
• Add additional texture for popout

Explore Interactions to Address Tactile Noise

• Support Local + Global Exploration
• Multiple Views + Details on Demand

More Collaborative Design

“I’ve never seen a graph, I’ve never seen anything visually. I may be frank – graphics mean nothing to me. I have no context, so I would dismiss them out of hand if they aren’t built for me... If they’re built from a visual place, at least for me, they mean nothing, they have no resonance with me whatsoever.”

Quote from participant in both parts of our study, born blind.

In response to question about if they felt the graphics could be improved.

“They Aren‘t Built For Me“: An Exploratory Study of Strategies for Measurement of Graphical Primitives in Tactile Graphics

Dylan Cashman

Brandeis University

Nam Wook Kim

Boston College

Lane Harrison

Worcester Polytechnic Institute

Areen Khalaila

Tulane University

Thank you to the AccessVis workshop and the Best Paper Committee and Reviewers!

Preprint

Data + Analysis