Summary19

From EQUIS Lab Wiki

modern eye tracking makes it possible to track things like - where do users look first? what do users look at most? which elements are not seen? researches can check iteratively if modifications of the graphic design lead to a wanted change in visual scan patterns

while eye tracking analysis is increasing within the area of web usability, the methods are far from being a broadly recognised part of the usability toolbox. it has been criticised for being costly and tedious and not delivering enough benefit to warrant that.

the paper looks at investigating the validity of users’ ability to report their own gaze patterns after looking at a user interface

self reporting requires that users reports are sufficiently reliable although not 100% complete. People’s patterns of fixation are known to be highly predictive of what they can remember afterwards and people tend to have a consistent viewing pattern when they re-visit something that they have seen previously

they investigated two low-cost alternatives to state-of-the-art eye tracking that could possibly deliver a cheaper and more handy way of getting access to eye movement data: 1)prompting users to report on their own eye movements from memory, and 2) asking web designers to predict the eye movements of a typical user.

they asked users to fill out questions which required them to find information on websites, they eye tracked this then asked the users to report their eye movements by repeating them while looking at the same websites.

Secondly, 17 web designers (all of them with more than 18 months of professional experience in web page design) were asked to predict the eye movements by marking a “typical user scan path” on print-outs of screen dumps from the 8 web pages the users were asked to scan. The designers’ predictions were compared with typical user scan paths constructed by n-gram analysis of the total eye tracking data from the 10 subjects for each web page.

they divided each webpage into a number of Areas of Interest (AOI´s) on basis of the gestalt laws i.e. “law of good continuation”, “law of proximity”, “law of similarity”, “low of closure”, and “law of symmetry”

Each AOI was classified according to a list of common web page elements, such as banners, contact info, drawings, email, input fields, logos, mixed content, navigation elements, pictures, search fields, text blocks and URL´s.

RESULTS The users could remember 70 % of the AOI´s they had fixated on for more than 125 ms. The average number of AOI´s users remembered having seen did vary across different web pages (from 4 to 16), as some of the web pages had more elements than others. Also some of the task questions were harder to answer than others, thereby initiating a broader visual search of the web page. The amount of AOI´s that the user could remember having seen depended on how many AOI’s they had actually seen, as there were no differences in the relative memory between simple and complex web pages.

On average, users had 1.9 false memories per web page web pages with many AOI´s had more false memories

On average, the group of 17 web designers could predict 46 % (SD = 22.1 %) of a typical user scan path, when told about the question that the user had to answer.

http://www.amber-light.co.uk/CHI2006/Alstrup_CHI_PositionPaper_EyeTracking.pdf