Accessibility Education for Software Engineers: Evaluating the Impact of Game-Based Learning
P D Parthasarathy, Swaroop Joshi · 2026 · ACM Transactions on Computing Education · doi:10.1145/3785367
Summary
Parthasarathy and Joshi address a persistent industry problem: despite 25 years of WCAG, only 5.2% of top web homepages fully conform, and organizations consistently report that staff lack accessibility skills. Academic curricula are slowly integrating accessibility, but training strategies for practicing software engineers remain underexplored. To tackle this, the authors developed two open-source serious games — A11yMythBuster, a short arcade-style game for debunking accessibility myths across 60 facts/myths statements, and A11yBugHunter, a 10-level puzzle game in which players inspect real front-end code to identify WCAG 2.1 Level A violations, name the POUR principle breached, and (for bonus points) the specific success criterion. Both games were designed using Kiili's experiential gaming model, incorporating feedback, challenge, progressive difficulty, and gamification elements such as points, levels, and leaderboards, with adaptive difficulty in BugHunter. The games were built collaboratively through a two-day Stanford d.school-style design jam, accessibility-audited with Axe and WAVE, and assessed using the Serious Game Design Assessment (SGDA) framework. The authors used a Design-Based Research (DBR) methodology with this study representing the first iteration. They deployed the games at a multinational software company of 10,000+ employees during Global Accessibility Awareness Day (GAAD) 2024, capturing pre/post surveys, gameplay telemetry, and semi-structured interviews with 25 of the 125 participants. Participants spanned Novice to Expert self-rated accessibility knowledge across multiple regions and product teams.
Key findings
Awareness of accessibility myths more than doubled from 30.4% to 72.0% after playing A11yMythBuster, and participants who reported no myth awareness dropped from 31.2% to just 5.6%. Confidence in identifying WCAG issues shifted dramatically: Very/Extremely Confident rose from 36.0% to 76.8%, while Not/Slightly Confident fell from 47.2% to 12.8%. Confidence in creating accessible products rose similarly (Very/Extremely Confident: 42.4% to 60.8%). On self-reported learning gains, 35.2% reported High improvement and 60.8% Moderate; none reported no improvement. Engagement metrics were strong — 87.2% rated A11yMythBuster Engaging or Very Engaging; 88.4% rated BugHunter the same. No participants rated either game Poor. Instruction clarity for BugHunter was rated Excellent by 72.8% of participants. Gameplay telemetry showed 22 participants replayed MythBuster voluntarily; BugHunter averaged 17 minutes 41 seconds per session with a 5,500-point ceiling. Qualitative analysis (Krippendorff's alpha 0.84) surfaced seven themes including a shift from compliance-driven to user-centered framing, near- and far-transfer behavior change (fixing labels in real projects, championing accessibility in meetings), reduced dependency on central accessibility teams, and value as onboarding material. Novices benefitted most but sometimes needed more scaffolding.
Relevance
For accessibility practitioners and L&D teams, this is a rare empirical study of industry-facing training rather than university coursework, and it provides two open-source games (a11ymythbuster.learndigitalaccessibility.com and a11ybughunter.learndigitalaccessibility.com) that organizations can adopt or adapt. The finding that short, low-stakes, gamified sessions shift attitudes from "accessibility is compliance" to "accessibility is user-centered" is practically important — it suggests that myth-busting content belongs early in onboarding, before technical training. BugHunter's code-triage format also mirrors the real bottleneck in many teams: engineers who cannot interpret automated-scanner output. Limitations are significant: participants self-selected during GAAD (likely already accessibility-curious), outcomes are mostly self-reported, only 3.2% of participants identified as disabled, and the SGDA evaluation used a single external rater. There is no measurement of downstream bug-fix rates or code-review data, so claims of real-world behavior change rest on interviews. Still, for organizations trying to move accessibility out of a "central team of heroes" model, this is a concrete, evidence-backed intervention worth piloting.
Tags: accessibility education · game-based learning · serious games · WCAG · software engineering · gamification · workplace training · design-based research
Standards referenced: WCAG 2.1 · WCAG 2.2 · ARIA · POUR