What does it mean to make data visualization inclusive?
A brief from the 2023 Information+ Conference
The Information+ biennial conference is becoming the most accurate re-presentation of “Interdisciplinary practices in information design & visualization”, as per its claim: it brings together a growing number of scholars and practitioners from a very broad range of disciplines and practices, exposing the audience to a variety of presentations, topics and players, from academic research groups and companies to freelancers; it also reflects how diffused and diverse the opportunities of building value through data and information are, when design plays a leading role. Being a thorough and comprehensive report of the event too long for these columns, I’ll focus on a topic that was well represented this year in Edinburgh, and it’s also a current research interest of mine: how to make data visualizations more accessible and inclusive, both from a business and a research perspective; a topic addressed at Information+ by the “Inclusive Perspectives” dedicated session.
According to Donna Klama’s presentation — she is part of the “Viz Squad” at Celonis, a company developing solutions for process mining — there are four main categories of users for whom we should design more accessible interfaces: the ones who experience constrained mobility (1), and that’s where navigating each and every components through keyboard interactions might help; (2) visually impaired users, in need of appropriate color combinations and contrast, and deserving better screen-reader-based experiences; (3) customers requiring specific standard and certifications for auditing purposes; (4) Everyone! Meaning that accessible interfaces are more likely to enable better experiences and make users more comfortable.
I’d add that visually impaired, or even blind users would also benefit from the integration of other data representation modalities in current interfaces: data sonification and/or haptic interfaces would help reduce — or eliminate — the load on the visual channel and enable a more effective user experience.
Keyboard interaction with data visualizations was at the core of another presentation, given by Jonathan Zong, Graduate Researcher at the MIT Visualization Group / CSAIL and design fellow of the MIT Morningside Academy for Design, together with Daniel Hajas, Innovation Manager at the Global Disability Innovation Hub / UCL Interaction Centre. Daniel lost his sight at 16 because of retinal detachment and since then works on fostering innovation initiatives to address accessibility issues, with data and information especially. An issue, or limit, I see with keyboard interaction with data visualizations is that it works very well with simple visual models, like bars, pies and trend charts — Power BI Desktop introduced this feature in March 2019, and Microsoft has a dedicated division for Inclusive Design — but it might become complicated as the complexity of the visual model increases, from scatter plots with many (close) points represented, to network visualizations. This is where Olli, the open-source Javascript library introduced by Zong and Hajas might help: it converts existing visualization specifications into keyboard-navigable hierarchical structures to support self-guided data exploration, producing an accessible HTML tree containing textual descriptions. Olli currently supports Vega-Lite/Vega (from the same research lab), and Observable Plot, and — according to the authors — can be extended to support other Javascript visualization libraries and toolkits by means of a dedicated function that “adapts existing chart instances into a standard OlliVisSpec interface”, in accordance with the Accessible Rich Internet Applications (ARIA) standard. The most relevant feature of Olli is the hierarchical structure that enables impaired users to explore a visualization from the overview down to the details, back and forth: see here an example. Olli also features multiple ways to navigate the visualization and allows for different levels of information granularity. The authors also highlighted the importance of co-design in developing the library: there is no equity nor inclusion in a process that doesn’t systematically and consistently involve the end-users in the design process. If you want to know more about Olli, see the poster presented at IEEEVIS 2022, and the paper at EuroVis 2022.
Olli was originally created by Matt Blanco, candidate for a B.Sc. in Computer Science and Design at Northeastern University and currently working as researcher at the Center for Design I direct, during his co-op (internship) at the MIT Visualization Group: those kinds of experiences may make the life of a student at NU quite special.
While the session at Information+ was a key moment, opening to a thoughtful discussion, we need, as a community, to recognize that inclusion is a broad concept, not limited to blind users or people who experience low vision: making sense of the abstractions and symbolic conventions used in visualizations might be an issue for people with cognitive and learning disabilities as well — as Marriott et al. point out in their Inclusive Data Visualization For People With Disabilities: A Call To Action. Similarly, “people who lack fine motor coordination or have tremors may be unable to operate interactive visualizations”. Moreover, inclusion is not only about understanding and interpreting data: users with disabilities may also face issues in making use of tools and applications to design and develop data visualizations. A broad interpretation of ‘disabled’ users is also required: inclusion is a matter of perception, operations (interactions), cognition and culture(s); tackling it requires strong commitment across several organizational divisions, and the integration of different competences, with many of them being already available within design agencies and/or companies.
If you are interested in digging deeper into accessibility, inclusion and equity in data visualization, here is a list of some initial resources, as a starting point:
- Data Visualization Society’s Fireside Chat on “Accessible Data Visualization”
- Dr. Sheelagh Carpendale’s talk at the Canadian Research Software Conference (CRSC2021) — Data Visualization for Empowerment and Inclusion
- Microsoft Research’s Inclusive Data Visualization Initiative
- The Inclusive Design: Making Dashboards Engaging, Informative, and Accessible presentation at the 2021 Tableau Conference.
Donna Klama’s conclusion at Information+ works quite well as a reminder for what we should fundamentally be committed on, both as practitioners and researchers: inclusion should be the norm in the design process, not an exception for a specific target users or use cases, and this implies continuous learning — in conversation with the end-users and all potential stakeholders — and dedicated initiatives to foster organizational culture.