Making Data Visualization Accessible: Takeaways from Tori Clark

Tori Clark, a digital accessibility consultant at Airbnb, joined us on 3Play Media’s Allied Podcast to discuss accessibility in data visualization.

We were thrilled to welcome Tori to the podcast, as she has immense knowledge in multiple areas of accessibility, including data visualization, an often-overlooked element of digital access. 

In her episode, Tori explains the core principles of accessible data visualization, the challenges of making data accessible, and common myths and misconceptions.

Core principles of accessible data visualization

While WCAG hardly references data visualization besides the recommendation to add alt-text to charts, Tori recommends focusing on the spirit of WCAG and not using shape, color, or position alone to communicate information.

“There isn’t quite enough in the current WCAG guidelines to cover everything that’s important and meaningful about data visualization,” Tori says. “One thing I like to talk about, though, is the spirit of WCAG…. so even if it’s not an outright violation based on the wording of the standard or the criterion, you can still kind of look behind it and be like, why was this created?”

With data visualization still in its infancy, as Tori describes it, and without clear written guidelines, it can be challenging to know what to prioritize when making data visualizations accessible. Tori gives us the following core principles:

Color

• When creating data visualizations, ensure good color contrast. Specifically, aim for a contrast ratio of at least 3:1 for graphics and user interface components.
• Don’t use color alone to convey information, and don’t rely on patterns for differentiation. Colors can be inaccessible for people with colorblindness, and patterns can cause issues for those with multiple disabilities.
• Label your visualizations. Labels need to be clear, whether it’s a line indicating a point in a static image or a way to interact with an image to expose a data point. 

Keyboard-only and speech recognition users

• Don’t make users tab through an entire dataset. While this might not cause issues with small datasets, the problems multiply with larger datasets. Tori says that the repetitive action of tabbing through a dataset can be both frustrating and painful for keyboard-only users.
• Provide a method for users to avoid a graph. The option to bypass a diagram is critical for many keyboard-only users and those using devices like switch controls.

Non-sighted users

• Don’t use tables as an alternative to data visualization. If the table was enough to communicate the information, ask yourself: Why did you create a data visualization?
• Don’t rely on untested, new technology to communicate data visualizations. “There is a lot of talk, for instance, around interactive sonification and data sonification,” Tori says. “And it’s such a neat idea… but the problem is, you also need to pair that with education for blind and other vision-related disabilities.” While data sonification has enormous potential, it’s not yet a common way for blind users to absorb information. Until there is more education for blind people on comprehending data sonification, only use it as an added element.

Learn more about accessibility in data visualization

Data visualization myths

Tori also shared common myths and misconceptions about accessibility in data visualization, such as character limits on alternative text, using tables as an alternative to data visualization, and using patterns to address colorblindness.

Myth: Alt-text should only be 125 characters or fewer.

Referencing Eric Eggert, Tori says not to be afraid of going beyond 125 characters for static images when necessary.

“You should be as brief as possible,” Tori says, “but if [alt-text] needs to be long, it needs to be long.” The most important thing, Tori says, is that your alt-text communicates the information, trends, and shapes that happen as you go from the front of the graph to the end.

Myth: Tables are an alternative to data visualization.

Tables are not an alternative to data visualization as they are not communicating the same information. “If the table was enough,” Tori says, “you wouldn’t have made data visualization in the first place.”

Additionally, tables can become even more complicated when the dataset contains many values, making comprehension nearly impossible.

Myth: You must include patterns in visualizations when addressing colorblindness.

Many people assume that patterns are an acceptable way to address colorblindness in visualizations like circle charts or bar graphs, but Tori cautions against this approach because it excludes people who are multiply disabled.

While consulting at Wells Fargo, Tori did work around understanding the cross-section of low vision and colorblindness. “What we found is that we can use spacing and alternating contrast… And by providing the appropriate spacing, [the data visualization] was something that was easy to consume by both colorblind users and low vision users.”

Tori calls this the “traffic light principle.” People often question how someone with colorblindness can differentiate between red and green; however, on a traffic light, the red light is always on top, and the green light is always on the bottom. The consistency in spacing means drivers don’t need to use color for differentiation.

To learn more about accessibility in data visualization, listen to Tori’s Allied episode below or read the transcript.