Skip to main content

Integrating Accessibility into Data Science Curricula

Mine Dogucu has coauthored a book chapter on Data Science + Accessibility, sharing resources to foster inclusivity in computing.

An illustration of two people staring at a data visualization above a stack of hardware with many controls. Below the picture reads: “Making data accessible manes more than just sharing it online.”
Chapter artwork by DO-IT (Disabilities, Opportunities, Internetworking & Technology).

“We firmly believe that accessible education is the cornerstone of a more inclusive future, and it’s high time we brought this narrative into the realm of data science.”

This assertion appears in Data Science + Accessibility, one of 16 chapters in the new book Teaching Accessible Computing, edited by Alannah Oleson, Amy J. Ko and Richard Ladner. As the editors state in their introduction, their goal is to help educators feel confident introducing topics related to disability and accessible computing and integrating accessibility into whatever computer science course they teach.

JooYoung Seo headshot
JooYoung Seo

The chapter on data science courses is written by JooYoung Seo, an assistant professor in the School of Information Sciences at the University of Illinois at Urbana-Champaign, and Mine Dogucu, a professor of statistics in UC Irvine’s Donald Bren School of Information and Computer Sciences (ICS).

Mine Dogucu headshot
Mine Dogucu

“Accessibility is more than just a principle — it’s a practice,” explain Seo and Dogucu in their chapter. “It’s a commitment to ensuring that everyone, regardless of their abilities, can participate in and contribute to data science. It’s about recognizing and valuing diversity, and about striving for inclusivity in all aspects of our work.”

An Impactful Partnership

Dogucu recounts how she first met Seo through a post on GitHub. “I was using a tool widely used by data scientists, but it didn’t support alternative text. So I went to GitHub to submit a feature request, asking the team that developed the tool to support alt text.” Seo seconded her request. “He came to GitHub and commented on it, saying he really would like this feature as well, and that’s how JooYoung and I first met.”

Both are recipients of grant funding from Teach Access, which supports faculty efforts to teach undergraduate students about accessible technology design and development. “Our work is partially a result of Teach Access,” says Dogucu. “JooYoung and I think a lot about how to teach accessibility to students,” she explains. “If educators can’t teach accessibility, it’s probably because they themselves never learned it.” So how do you break that cycle?

Seo and Dogucu aim to address the issue by developing course materials, and they started by writing a paper together called “Teaching Visual Accessibility in Introductory Data Science Classes with Multi-Modal Data Representations.” The paper provided the framework for writing the Data Science + Accessibility chapter.

Data Science + Accessibility
Seo and Dogucu identify in the book chapter three key aspects to building accessibility in data science: computational reproducibility, data representation, and social and cultural value.

“I think perhaps the biggest contribution of this chapter is the focus on reproducibility,” says Dogucu. “Thinking of reproducibility from an access perspective is very important.” Of course, that then leads to the issue of data representation.

“For instance, a blind scientist might not understand everything in a report if the images aren’t in an accessible format,” explain Dogucu. “Or if a report is provided in pdf format, a screen reader would have problems reading it.” So the chapter introduces a variety of ways to ensure data access and representation.

For example, readers learn about an open source multimodal access and interactive data representation (MAIDR) system developed by Seo. The system has the potential to offer customizable options across braille, text, and sonification, providing new levels of access for people with varying disabilities.

Dogucu points out the long-term impact of such education and tools, touching on the social and cultural value. “When you teach it, these options are then built into the future.” The result is more inclusive data science, which leads to more inclusive insights and outcomes.

“This gets into the ethical implications of our work as data scientists,” says Dogucu. Data can perpetuate certain biases and assumptions depending on how the data is gathered, analyzed and presented, having lasting impacts on entire communities. Seo and Dogucu are working to remove bias, expand access and share new perspectives.

“Together, we can shape a future of data science that is not only technically robust,” they conclude in their chapter, “but also ethically responsible and universally accessible.”

Shani Murray