Do Web Users with Autism Experience Barriers When Searching for Information Within Web Pages?
Sukru Eraslan, Victoria Yaneva, Yeliz Yesilada, Simon Harper · 2017 · Proceedings of the 14th International Web for All Conference (W4A) · doi:10.1145/3058555.3058566
Summary
This paper provides empirical evidence that adults with high-functioning autism experience real barriers when searching for information on web pages, using eye tracking to reveal the underlying differences in visual attention patterns. The study employed a between-group comparison design with 18 adults with clinically diagnosed high-functioning autism (ADOS criteria, IQ>70) and 18 neurotypical controls, all daily web users with normal or corrected vision. Participants were shown screenshots of six web pages of varying visual complexity (Apple, Babylon, AVG, Yahoo, Godaddy, BBC — measured by ViCRAM tool) and asked two information-location questions per page (e.g., "Can you find a telephone number for technical support and read it?") with a 30-second time limit. Eye movements were recorded using a Gazepoint GP3 tracker at 60Hz. Areas of interest were defined using the VIPS (Vision-Based Page Segmentation) algorithm. The study investigated three research questions: whether success rates differ between groups, whether there is higher variance in individual scanpaths among the ASD group, and whether the trending (typical) scanpaths differ between groups. All materials, task descriptions, and eye tracking data are openly available.
Key findings
Participants with autism were less successful at locating information on web pages: mean correct answers of 10.83/12 (median 11.50, SD=1.72) versus 11.56/12 (median 12.00, SD=0.86) for controls. Critically, the scanpaths of the ASD group showed significantly higher variance than the control group. Using String-edit distance to measure scanpath similarity, the mean within-group similarity was 33.4% for ASD participants versus 37.8% for controls — meaning autistic participants followed more diverse, less predictable paths to find the same information. ScanGraph visualization confirmed this pattern: control group scanpaths formed a more tightly connected cluster, while ASD scanpaths were more dispersed. The trending scanpaths (typical paths identified by the STA algorithm) of the two groups were approximately 55% dissimilar, indicating substantially different search strategies. Despite this dissimilarity, approximately 45% commonality existed, suggesting the groups do not follow completely different strategies but rather that ASD users are more affected by irrelevant page elements (consistent with Weak Central Coherence Theory and stimulus overselectivity). The authors note that these barriers likely have much greater magnitude for autistic individuals at the lower end of the spectrum or children learning to use the web.
Relevance
This is among the first empirical studies demonstrating web accessibility barriers specifically experienced by autistic users, filling a significant evidence gap that has contributed to cognitive accessibility receiving lower priority in WCAG guidelines. The key finding — that autistic users follow more variable, less efficient search paths due to distraction by irrelevant page elements — has direct design implications: reducing visual clutter, creating clear visual hierarchies, and providing prominent landmarks could reduce the attentional demands that disproportionately affect autistic web users. The higher variance in ASD scanpaths is particularly important because it means there is no single "autistic scanning pattern" to design for; instead, the focus should be on making pages more forgiving of diverse attention patterns. The open data availability enables replication and further analysis. The authors argue that elements related to cognitive disability should be given higher priority in future WCAG revisions — a recommendation that has gained increasing support in the accessibility community. This paper serves as a precursor to the same team's 2018 W4A paper on using gaze data for autism detection, connecting the accessibility and screening research streams.
Tags: autism · eye tracking · web accessibility · cognitive accessibility · scanpath analysis · visual attention · neurodivergence · user study · information search · web searching · WCAG · open data · weak central coherence · stimulus overselectivity
Standards referenced: WCAG