Investigating Sighted Users' Browsing Behaviour to Assist Web Accessibility
Eleni Michailidou, Simon Harper, Sean Bechhofer · 2008 · Proceedings of the 10th International ACM SIGACCESS Conference on Computers and Accessibility (Assets '08) · doi:10.1145/1414471.1414495
Summary
This paper presents an eye tracking study investigating how sighted users browse web pages, with the goal of using these insights to improve web accessibility for visually impaired users. The core argument is that understanding sighted users' visual browsing patterns can inform how page content should be structured for screen reader access. Thirty-three participants browsed nine web pages (ranging from visually simple to complex) while their eye movements were tracked using a Tobii x50 eye tracker. Pages were divided into five analysis areas: header, main content, right column, left column, and footer. The study used serendipitous browsing (no specific task) to capture natural viewing behaviour unbiased by task demands. The authors contextualise this work by noting that sighted users can scan a page and comprehend it in approximately 5 seconds, while screen reader users may take 10 seconds to 3 minutes just to reach the main content — a dramatic disparity that proper content ordering could reduce.
Key findings
Statistical analysis (repeated measures ANOVA) revealed several significant patterns. Main content and header areas attracted first fixation most often (91% and 61% respectively), while footer received zero first fixations. Users spent the most gaze time on main content (5.6 seconds average) followed by header (2.5s), left column (2.3s), right column (1.3s), and footer (0.1s). The common gaze order was: main content, header, right column, left column, footer. Salient elements (large images, logos, animations) consistently attracted the most fixations as shown by hotspot visualisations. Crucially, users only fixated on the first 3-4 items in menu lists, never reading complete navigation menus. A significant positive correlation (r=0.6, p<0.5) was found between visual complexity scores and gaze time — more complex pages took longer to browse and produced more scattered, disordered scanpaths. The authors propose three design guidelines: (1) provide page title followed by main content first in the source code, placing salient elements like advertisements after content using CSS for visual positioning; (2) order content as header, main content, left menu, right menu, footer; (3) limit menu lists to 7±2 items with serial access to remaining links.
Relevance
This paper bridges the gap between visual design research and screen reader accessibility by empirically demonstrating how sighted browsing patterns should inform content ordering for non-visual access. The practical implication is straightforward: if sighted users look at main content first and spend the most time there, then screen readers should deliver main content first too — yet many websites in 2008 (and still today) force screen reader users through navigation, advertisements, and sidebars before reaching the content. The CSS-based solution proposed (maintaining visual layout while reordering source code for screen readers) remains a valid and widely recommended technique. The finding about menu length — users only fixate on the first 3-4 items — supports accessible design patterns like progressive disclosure and the "7±2" principle for navigation. For accessibility practitioners, this study provides empirical evidence to justify content-first source ordering, reduced navigation complexity, and minimised visual clutter, all of which remain foundational web accessibility principles.
Tags: eye tracking · web accessibility · visual impairment · web design · visual complexity · screen readers · user research · browsing behavior · transcoding
Standards referenced: WCAG · Section 508