Access Differential and Inequitable Access: Inaccessibility for Doctoral Students in Computing
Kristen Shinohara, Michael McQuaid, Nayeri Jacobo · 2020 · Proceedings of the 22nd International ACM SIGACCESS Conference on Computers and Accessibility (ASSETS 2020) · doi:10.1145/3373625.3416989
Summary
This paper investigates why the pipeline of students with disabilities pursuing doctoral degrees in computing remains so thin, despite increased support at the undergraduate level. In 2018, the NSF reported that only about 3% of math and computer science PhD recipients identified as having a visual limitation and 1.2% as having a hearing limitation. Through interviews with 19 current and former graduate students who identified as blind or low vision (12 participants) or deaf or hard of hearing (7 participants) in computing and related doctoral programs at US institutions, the researchers uncovered pervasive inaccessibility across virtually every aspect of graduate education. The study used a grounded theory approach, with three researchers independently coding interviews and iteratively refining themes over the course of a year. Participants ranged from first-year to seventh-year PhD students, with five having already graduated or left their programs. The paper introduces two key conceptual frameworks: access differential — the gap between the access that nondisabled and disabled students experience, which participants constantly measured by observing their peers — and inequitable access — the degree to which existing accommodations fail to adequately address inaccessibility, meaning that having an accommodation does not equate to having access.
Key findings
Participants experienced inaccessibility in research tools (only Eclipse IDE was screen reader accessible, limiting tool choice to "1 of 3 things" while sighted peers could choose freely), reading and annotating research papers (PDF figures, equations, images, and highlighting/notes features all inaccessible via screen readers), data analysis and visualization tools (blind students wrote custom scripts to extract meaning from data their sighted peers could simply look at), reference management software (Mendeley, Zotero inaccessible — one participant maintained manual folder-and-Word-doc systems), and qualitative analysis tools (Dedoose installer inaccessible, forcing participants to use Excel spreadsheets instead). For DHH participants, challenges centered on interpreter availability and quality: interpreters were scarce for PhD-level courses (priority given to events with more deaf students), lacked domain knowledge for highly technical content, and required extensive preparation time. Captioning services had significant delays and domain-specific errors. Impromptu conversations — a core part of graduate life and research collaboration — were entirely inaccessible for DHH students. Faculty and disability services frequently confused inaccessibility with difficulty, dismissing accommodation requests. When accommodations were granted, they often created new burdens: human readers required scheduling and training, extended exam time was mathematically unrealistic (8 hours for a 4-hour advanced algorithms exam), and interpreter preparation consumed the student's own time. Students responded by developing workarounds (writing custom scripts, using manual spreadsheet analysis), accepting sub-par access (reading 75% of a paper and "taking their word" for figures), dropping courses, or leaving programs entirely.
Relevance
This paper is essential for anyone involved in computing education, disability services, or academic accessibility policy. It demonstrates that the accommodation model — where students must identify, request, justify, and then manage their own accommodations — is fundamentally broken at the doctoral level. The access differential framework provides a powerful conceptual tool: accessibility is not a binary (accessible/inaccessible) but a comparative gap measured against what peers can do. The finding that computing PhD students, who have the technical skills to code their own workarounds, still face insurmountable barriers underscores how deep the systemic issues run. For accessibility practitioners, the paper highlights that inaccessible research tools (MATLAB, NVIVO, Dedoose, Zotero, PDF annotation, data visualization) represent critical gaps that the computing community should address. The inequitable access concept — that accommodations can appear to "solve" problems while actually adding burden — challenges organizations to evaluate accommodation effectiveness, not just provision. Limitations include the small sample focused on two disability categories and the survivorship bias of interviewing those who made it to graduate school.
Tags: education accessibility · disability rights · deaf and hard of hearing · blindness and low vision · digital accessibility · screen readers · accessibility barriers · organizational accessibility