Using galvanic skin response measures to identify areas of frustration for older web 2.0 users
Darren Lunn, Simon Harper · 2010 · Proceedings of the 2010 International Cross Disciplinary Conference on Web Accessibility (W4A) · doi:10.1145/1805986.1806032
Summary
This paper investigates how older web users experience stress and frustration when interacting with Web 2.0 dynamic content, using galvanic skin response (GSR) measurements combined with eye-tracking data. GSR is a physiological indicator that measures changes in skin electrical resistance caused by sweat gland activity, which increases with mental stimulus and stress. The study extended previous work with younger users (under 29) by recruiting 23 older participants across three age bands: 30-49, 50-59, and 60-69. Participants completed six web-based tasks across live websites including Google Search, Google Suggest (with auto-suggest lists), National Rail Search, National Rail Suggest, iGoogle, and Yahoo! portal pages. The tasks were designed to compare user stress levels when dynamic content such as auto-suggest lists (ASLs) were present versus absent, and when pages contained multiple competing dynamic content areas. Participants sat before a TOBII 1750 eye-tracker while wearing a MindWalker 3 biofeedback meter on their fingers. The GSR data was sampled every tenth of a second and smoothed using a Savitzky-Golay filter to identify meaningful peaks indicating stress responses.
Key findings
The ANOVA analysis across all six tasks showed that age was not a statistically significant factor in mean stress levels — contradicting the initial hypothesis that older users would find dynamic content inherently more stressful. However, the study revealed that older users were a markedly non-homogeneous group, with large variance in GSR measurements and no consistent patterns of interaction identifiable from the GSR graphs, unlike younger users who showed clear, consistent patterns. For younger users, the earlier study had found significantly less stress when auto-suggest lists were present (Google Suggest mean peaks: 5.30 vs. Google Search: 10.95, p=0.0004), but this significant difference did not replicate for the older age groups. The 50-59 age group showed significant benefit from auto-suggest on the National Rail tasks (p=0.035), possibly because the constrained input reduced typing errors — four participants made errors during search but none during suggest tasks. Key behavioral observations included: older users exhibited heightened cautiousness, frequently checking the page before beginning tasks; they made more typing errors; they were more likely to blame themselves when errors occurred ("Oh no. Look at that") while younger users blamed the computer ("It's the keyboard"); and some older users reacted strongly to asynchronously loading content like YouTube videos even when not looking at them.
Relevance
This study pioneers the use of physiological measurement for understanding accessibility-related frustration — moving beyond self-reported surveys and task completion metrics to capture unconscious stress responses. The finding that older users are non-homogeneous in their stress responses challenges the common practice of treating "older adults" as a single user group with uniform needs. This has direct implications for inclusive design: assistive features for dynamic content cannot be one-size-fits-all, as some older users are comfortable with dynamic features while others experience hesitancy and uncertainty. The observation about cautiousness — older users approaching tasks tentatively, checking pages before acting, and self-blaming for errors — has practical design implications: clear feedback, error prevention, undo mechanisms, and explanatory cues for dynamic content can reduce hesitancy. The behavioral contrast between age groups regarding error attribution (self-blame vs. externalizing) suggests that error messages and recovery flows should be designed to reassure users that mistakes are normal and recoverable. For researchers, the combination of GSR with eye-tracking provides a methodological template for evaluating emotional responses to interface design decisions beyond traditional usability metrics.
Tags: older adults · physiological measurement · galvanic skin response · eye tracking · Web 2.0 · dynamic content · cognitive load · user research · stress measurement