Data When You Can’t See
How to design visualization for blind individuals.
Have you ever had a hard time reading and comprehending visualization work? Well, you definitely are not alone — complex visualizations can be challenging to read at the best of times. But consider someone who has a visual impairment or is entirely blind. While the visualization field has grown rapidly over the years, it is only recently that researchers have begun to pay full attention to how blind individuals can access all of its wonderful advances.
In a recent paper at IEEE EuroVis 2021, Nam Wook Kim, Shakila Cherise S Joyner, Amalia Riegelhuth, and Yea-Seul Kim discuss the issue of inaccessible visualization and investigated approaches to support people with visual impairments in particular. They aimed to examine the knowledge gap in visualization accessibility by analyzing relevant research papers published since 1999. Inspired by the grounded theory method, they conducted thematic analysis through open coding 56 papers within their literature collection. After the iterative coding process, the researchers mapped a design space for visualization accessibility.
The researchers derived seven dimensions of accessible visualizations (user, task, chart type, interaction, granularity, modality, and technology) in three categories (why, what and how) in their analysis of design space.
The researchers found that most papers do not specify various types of visual impairments. However, the spectrum of people with visual impairments — from low vision to blindness — has diversified needs of accessing information. Therefore, the researchers suggested the first step of accessible visualization is to understand the needs and motivations of people with different types of visual abilities.
The researchers observed two main tasks of visualization literacy: reading and creating visualizations. Visually impaired people rely on alternative sensory channels, such as audio and tactile perception, to access information. However, they found a lack of understanding on what and how people with visual impairments perform non-visual tasks and suggested further study in this area.
The authors pointed out two types of visualizations: the basic and advanced charts. With the growth of more advanced visualizations, we need to understand more about how visually impaired people comprehend unfamiliar visualizations, and what information order makes sense to them.
The researchers used the existing interaction taxonomy by Heer & Shneiderman to analyze the interaction dimension within their paper collection. They observed that navigating within a chart was the most common interaction. While interactive visualizations have enormous growth, there are possibilities that the complex interactions might exclude visually impaired audiences and worth further understanding.
The authors proposed three levels of information granularity: existence, overview, and detail of a visualization. Users might notice the existence of a chart, want an overview, and explore in detail. This dimension provided useful guidance on how to structure accessible information in order. Yet, we still need further understanding of how to present information within each level.
Selecting alternative sensory modalities is essential for accessible visualization for visually impaired people. Audio and tactile perceptions are the common alternative channels. Among all modalities, speech is the most common and low-cost accessible modality. It provides precise values and details, but the prolonged descriptions can be frustrating. On the other hand, tactile graphics provide faster and nonlinear data exploration, but require larger spaces for high-resolution information. The multi-sensory perception is often deployed to overcome the limitations of different modalities.
Besides alternative sensory modalities, the existing assistive technologies also play important roles in accessible visualization. Screen readers are most widely used and accessible, while tactile devices are costly but provide more immersive experiences of graphics. The researchers suggested that visualization designers consider the compatibility with diverse assistive technologies.
Preliminary model for visualization accessibility
From the design space analysis, the researchers also identified a four-stage preliminary model for visualization accessibility based on the user’s flow of accessing visualized information. Firstly. the model starts with notifying users of the existence of a chart. It can then provide an overview of the visualization, including the intended message, visual structure, and orientation. The details such as individual data points should be provided only when requested to avoid overwhelming the user. Last but not least, it would be helpful to provide context information such as adjacent data points and spatial context.
If you’re a visualization designer, I highly recommend you read this paper and include people with different visual abilities into your design considerations. Kim et al. have paved a solid road to accessible visualization by analyzing research papers over the last two decades. They presented the design space of accessible visualization, pointed out the current gaps, and suggested a useful preliminary accessibility model. As with the increasing growth of the visualization field in recent years, I am looking forward to the development of accessible visualizations that support a wider spectrum of people with disabilities.
This blog post is based on the following paper:
- Nam Wook Kim, Shakila Cherise S Joyner, Amalia Riegelhuth, Yea-Seul Kim. Accessible Visualization: Design Space, Opportunities, and Challenges. In Proceedings of IEEE EuroVis 2021.