Because visualisation researcher Frank Elavsky noticed a big gap when it came to guidelines on accessibility of data visualisations, he developed Chartability. As Chartability describes itself:
Chartability is a set of heuristics (testable questions) for ensuring that data visualizations, systems, and interfaces are accessible.
From this definition it is clear that Chartability goes beyond the accessibility of just simple charts. It can evaluate any kind of data experience, including charts, graphs, bespoke and custom graphics based on data, models, algorithms and other data driven interfaces or systems.
Chartability is based on WCAG and its principles, and is not meant to replace these internationally recognised guidelines. But the WCAG were not developed with data visualisation in mind, so even WCAG compliant data visualisations and interactive charts can still be inaccessible to a lot of users.
Chartability describes itself as having 3 modes of use.
The fast pass Chartability audit involves 5 steps to quickly check a handful of access barriers quickly.
The first step is visual testing. This entails calculating the contrast ratios between text and data marks with their background colour, the use of colour blind safe colours and a check on the font sizes used.
The second step is keyboard probing. Interactive elements should be reachable and controllable using a keyboard alone, and this keyboard navigation and activation shouldn’t take too much time or effort.
Related to keyboard probing is screen reader inspecting. The key aspect of this third step is wether a text description of what is visible in a chart is provided and can be read out by the screen reader. But other elements in a visualisation should also be accessible to screen readers, like data values, category labels and axes titles. Furthermore, if a chart is interactive, screen readers should be able to recognise and announce the interactive features.
In the next step, checking cognitive barriers, the reading level and clarity of the text is evaluated. This step also involves checking wether or not the main message the chart is trying to convey is also present in written text (like in the chart title for example). On top of that, striking patterns in the data should be explained with callouts or annotations: it shouldn’t be up to the user to spot ant try to explain these patterns.
The fifth and last step is called evaluating context. This involves checking if the design and styling of a visualisation respects user settings like dark mode and high contrast mode, and checking whether alternative ways of accessing the data (with a simpler chart, or with a table for example) are provided.
In total, Chartability contains 50 tests to identify accessibility failures in data experiences. But 14 of those 50 tests are considered critical. In the intermediate Chartability audit, a data visualisation (or other kind of data experience) is tested against those 14 critical accessibility features.
Organised by the POUR-CAF principles, the 14 critical tests are the following.
Perceivable: