Improving the Accessbility of Graphs for Users Who are Blind
The following is the summary of a study for improving the accessibility of graphs for users who are blind.
How do users who are blind currently interact with graphs? What challenges do users who are blind currently face when encountering graphs? How can we improve written descriptions and develop guidelines for web developers and designers when presenting graphs in web-based technologies for screen readers?
- Literature Review
- Moderator guide - writing and editing
- Interview - conducting and affinity diagramming
- User testing - conducting and affinity diagramming
- Report writing and editing
Graphs are a common method for communicating information to audiences, but are inaccessible to users who are blind. This study explores the challenges faced by these users through usability tests of graph descriptions with varying levels of detail. Users then answered questions about the graph content and gave satisfaction scores. We also conducted interviews with participants about their prior experiences with graphs. Overall, participants’ comprehension and satisfaction increased directly with the graph description level. We found that although screen readers are the tools primarily used to access graphs, some graphs are inaccessible by screen readers or have text alternatives that provide limited or no descriptions. Graph descriptions must include the graph type and attributes as well as the context of use for each graph. Descriptions must not burden users with remembering information or frustrate them with repetitive data. This paper proposes description guidelines for designers to create accessible graphs.
Data Collection in Two Parts
Usability tests on three graph descriptions that we wrote.
Interviews about their experiences with graphs.
- Live in Chicago
- Familiar with common web & accessibility technologies
- General understanding of bar & line graphs
For our usability test, we found three graphs that had a similar complexity and theme, but different content. We then wrote a description for each that had three different levels of detail: simple, medium, and thorough
- Simple - Graph Overview
- Medium - Graph Overview; values per category
- Thorough - Graph Overview; values per category; trends associated with the graph
We asked our participants these four questions to see if the descriptions effectively communicated the graph details.
We then asked the participants to rate the clarity of the descriptions provided for each graph based on a Likert scale and then in more open-ended questions.
Interviews on experiences with graphs
We asked additional questions in order to get feedback about our descriptions as well as the design of our study.
We assessed the effectiveness and clarity of each graph description by tallying the correct answers to each assessment. We computed the average clarity rating across all participants and compared these for each graph.
Each team member individually extracted quotes from the transcribed interviews for common and salient themes. We pooled our data using an affinity diagram and identified common themes.
(Featured video is time lapse of our affinity diagram process using Stormboard, courtesy of team member Casey Hudetz)
All participants used mobile phones and computers on a daily basis. Participants used their phones to access social media, email, call people, browse the internet and check their calendars. Participants used their computers to check email, access social media, work on application forms and gather news. All participants used at least one accessibility device such as a screen reader and/or mobile voice-over.
Experience with graphs
Participants had varied experiences with graphs. They interacted with graphs in paper and digital mediums within various settings including school, work, and as a hobby. Participants mentioned both positive and negative experiences with graphs. Two participants mentioned that Excel allowed them to navigate through the finer details of graphs. Some negative graph experiences reported by participants included nonexistent screen reader support, the lack of accessibility tools for graphs, and the limited description provided by screen readers. Many participants reported that the most detailed description provided by a screen reader was “This is a graph”
Challenges When Trying to Access Graphs
Participants also mentioned various challenges when trying to access graphs. Participants mentioned that if they had no prior exposure to graphs earlier in life, they might not understand the visual concept of the graph and that lack of spatial understanding was a challenge. Participants also mentioned that because we hadn’t specified the objective or context of the graph, they did not know what information they needed to memorize and they felt the burden to memorize the information we presented especially when we presented the specific values of each category. Many participants also felt that parts of our descriptions were repetitive and they felt that added to the challenge of trying to get to the meat of the description.
Most Important Features in a Graph Description
To our participants, the most important features in a graph description include the type of graph, graph scope, graph objective or context and sections of distributed content/categories/values and trends. Participants wanted to know the type of graph (i.e. bar or pie). They also wanted a way to verify that the description covered all content present in the graph description. Participants also felt that the objective or context that the graph was presented in was necessary for understanding which information needed to be gleaned. Participants expressed a desire to have graph descriptions broken down into sections that they could return to if they wished to review particular details about a graph; this is contrary to our graph descriptions that could only be repeated in full.
State the Context
First, designers should make the context of the graph clear in their description. When asking our participants to describe the trends of our graphs, many of them responded that it would depend on how the author intends to use the graph. For instance, a graph may serve to prove one particular point or serve to display detailed findings, neither of which we specified in our descriptions. Designers must explicitly state the graph’s purpose and context.
Provide access to the graph’s data on demand
Second, designers must provide access to the graph’s data on demand. Our participants were able to successfully compare the exact figures we provided, but did not want us to re-read any data that they missed. We gave our participants the option to re-read the descriptions, but only one took it. Several participants noted having experience with Excel to access structured data. Designers should include raw data that breaks down the graph into sections with a clear hierarchy so that users can skip to different sections. This hierarchy should follow common accessibility techniques.
Minimize Cognitive Load
Last, designers must pay close attention to the cognitive load of their descriptions. All participants struggled to remember details in the graphs with long descriptions. Designers must follow best practices and techniques to increase memory retention and reduce cognitive load. These techniques include chunking information together, using clear declarative writing, and removing repetitive descriptions.
CLICK HERE to download the SIGCHI Extended Abstract