Measuring Website Usability for Visually Impaired People - A Modified GOMS Analysis
Henrik Tonn-Eichstädt · 2006 · Proceedings of the 8th International ACM SIGACCESS Conference on Computers and Accessibility (Assets '06) · doi:10.1145/1168987.1168998
Summary
This paper adapts the GOMS (Goals, Operators, Methods, Selection rules) model — a well-established HCI method for estimating task execution times — to measure website usability specifically for blind screen reader users. The author argues that accessibility is fundamentally a category of usability, and therefore usability criteria like efficiency (measured through task execution time) can be applied to evaluate accessibility. The model was built on empirical findings from observations of 14 blind users and two field studies with 270 and 150 participants respectively. Key observations informed the model: blind users are experts with their screen readers even if they only use a subset of features; they employ multiple navigation strategies (heading lists, links lists, find feature, line-by-line browsing); they spend significant time verifying each element they land on; and what the author calls "microscopic navigation" — navigating within a single page to find relevant content — is a critical and time-consuming activity distinct from site-wide navigation. The paper uses the NGOMSL notation (a structured variant of GOMS) and extends classic GOMS with new constructs: verification steps with positive/negative outcomes that branch execution flow, and "choice sets" that represent equally valid alternative strategies rather than rational selection rules. Templates are provided for common screen reader interactions including browsing by quick key, using list modes, using the find feature, acquiring content via speech or Braille, and activating elements.
Key findings
The example analysis applied the model to the ASSETS '06 conference homepage, measuring execution time for finding and clicking a specific link using four strategies (next line, next link, find feature, links list) across two output modalities (speech and Braille). Results showed substantial time differences: the find feature was fastest at 19.43 seconds (speech) and 24.50 seconds (Braille), while line-by-line navigation was slowest at 68.36 seconds (speech) and 85.04 seconds (Braille). The next-link and links-list strategies fell in between at roughly 29 seconds (speech). Braille consistently took longer than speech for all strategies due to the sequential character-by-character reading process. The analysis revealed that verification — checking whether the focused element is the target — accounts for a large proportion of interaction time, with negative verification (rejecting irrelevant elements) typically faster than positive verification (confirming the target). The model is currently qualitative, as specific operator times for assistive technology users have not been empirically established; the paper uses standard GOMS timing values as approximations.
Relevance
This paper makes an important conceptual contribution by bridging the gap between accessibility evaluation and usability engineering. Most accessibility evaluation in 2006 (and still today) relies on guideline checklists and automated scanning tools, which cannot capture the efficiency of actual user interaction. The GOMS-based approach offers a way to quantitatively compare design alternatives from the perspective of blind users before user testing, which is expensive and time-consuming. The concept of "microscopic navigation" — the within-page navigation that blind users must perform — remains highly relevant and is often overlooked by designers who think of navigation only in terms of moving between pages. The model's insight that verification dominates interaction time reinforces the importance of clear, meaningful headings, link text, and semantic markup. For practitioners, the key takeaway is that supporting multiple navigation strategies (headings, links lists, search) is essential because different users prefer different approaches, and the efficiency gains from good page structure are dramatic.
Tags: screen readers · usability · web accessibility · GOMS · braille · blind users · evaluation methods · navigation strategies · efficiency
Standards referenced: WCAG 1.0 · ISO 9241-11 · ISO/TS 16071