Art Beyond Sight
Museums want to be inclusive. But when it comes to visual art, "accessible" usually means a ramp at the entrance. Not a meaningful experience for blind and partially sighted visitors. I ran three research studies across 20 weeks to find out: can art experts and non experts create artwork descriptions good enough to actually serve blind and partially sighted users? And what guidelines would make that possible at scale? Methods: Delphi method, semi-structured interviews, statistical analysis (Friedman's ANOVA, Wilcoxon tests). Participants: 18 art professionals, 10 blind/partially sighted users, 23 members of the public. Outcome: Evidence-based guidelines adopted as a framework for museum description programmes
Client:
University of York
Role:
Lead UX Researcher
Year:
2023 (20 weeks)
The Aim
The objective of this project was to develop a comprehensive set of guidelines for describing artworks, primarily paintings, intended for blind and partially sighted people. The research was guided by three key questions:
Can the current guidelines be improved by asking a range of relevant stakeholders (e.g., people working in the art world, blind and partially sighted people) to review them?
How usable are the resulting guidelines to interested members of the public in creating descriptions of works of art?
Do blind and partially sighted people find the resulting descriptions interesting and useful?
STUDY 1: Expert Evaluation
I designed a multi-round online questionnaire and recruited two distinct participant groups — 18 art professionals and 10 blind and partially sighted individuals — to evaluate an initial set of artwork description guidelines I had developed from existing literature. Rather than a single round of feedback, I used the Delphi method: an iterative consensus-building approach where participants review, respond, and refine across multiple rounds until agreement stabilises. This ensured the guidelines weren't shaped by one loud voice or one perspective, but by genuine convergence across people with very different relationships to art and visual impairment.
Strong consensus (over 80%) emerged for Language (85%) and Practice Run (82%). Moderate consensus (50–80%) was reached across Overview, Size, Composition, Medium, Style & Technique, Locations & Directions, Colour, and balancing Objective and Subjective description. These results directly shaped the refined guidelines carried into Study 2.
STUDY 2: Public Implementation
With refined guidelines in hand, I designed a user study to test a question that mattered for real-world scalability: could ordinary members of the public — with no specialist training — use these guidelines to create descriptions that were actually useful? I recruited 23 participants and asked each of them to create descriptions for six artworks using the guidelines, then collected both the descriptions and their reflections on the process.
Participants averaged 111 words per description. 60% felt confident using the guidelines. Around 35% found it difficult — particularly when describing colour and fine detail — which told me where the guidelines still needed work, not where the participants had failed. The most consistently applied aspects were Objectivity & Subjectivity (95.9%) and Language (93.2%), confirming that the refined framing from Study 1 had translated into something non-experts could actually follow.
STUDY 3: User Evaluation
The final study brought the work back to the people it was designed for. I recruited six blind and partially sighted participants to evaluate the descriptions created in Study 2 — giving them three descriptions per painting (short, medium, and long) across three artworks, and asking them to rate and reflect on each. This closed the loop: real users, judging real outputs, produced by real non-experts using guidelines I had developed and refined across the two previous studies.
I used Friedman's ANOVA and Wilcoxon signed-rank tests to analyse the rating data across description lengths. Longer and more detailed descriptions were generally preferred, though medium-length descriptions were favoured for certain paintings — suggesting that the right description length isn't fixed, but varies with the complexity and nature of the artwork itself. This nuance directly informed the final version of the guidelines.
Key Findings
Guideline Refinement: Expert feedback led to more concise, practical guidelines balancing detail with usability.
Description Creation: Non-experts could successfully create useful descriptions with the refined guidelines, though some found describing colors and specific details challenging.
User Preferences: Blind and partially sighted users generally preferred medium to longer descriptions with rich details on perspective, size, and color.
Practical Application: The guidelines provide a framework for volunteers and museum staff to create meaningful art descriptions without specialized training.
What this means beyond the research
This project demonstrated that accessibility at scale doesn't require specialists at every step, it requires well-designed systems that empower non-experts to contribute meaningfully. The guidelines developed here offer museums, galleries, and digital platforms a practical framework for expanding accessible experiences without prohibitive cost or expertise barriers. For me, it solidified a core design belief: inclusion isn't a feature you add at the end. It's the research question you start with.