Eye Gaze Tracking for Web Usability Evaluation
5.3 Web Usability Research with Eye Tracking
During the past few years, there have been a significant number of research studies concentrating on the usability of Web sites. However, only a few studies utilised the eye gaze tracking technique as a means of gathering information from users. This could be because eye tracking is a fairly new technique compared to other observation methods (e.g. videotaping, keystroke recording, etc.), and the
price of most eye tracking systems available on the market is still high, starting from US$4,000. Nevertheless, it is notable that the price of eye tracking systems has decreased over time. For example, the price of one system was US$250,000 in 1988 and US$20,000 in 1996 (Joch, 1996). The functionality and usability of eye gaze analysis software is also steadily improving. Therefore, in the near future, eye tracking technologies will be more popular among usability experts in research and commercial domains as their prices become cheaper and a wider range of hardware and software products are available.
One popular approach used to analyse user’s gaze behaviour is to superimpose gaze trails over a screen capture, as shown in Figure 5.3.
Figure 5.3 Gaze trails provided by GazeTracker (Eye Response Technologies, 2003)
For example, Newman (2001) conducted Web usability testing by using this technique. The eye tracker hardware used in Newman’s study was the model 504
from Applied Science Laboratories, while the software used for the analysis was GazeTracker from Eye Response Technologies. The participants were asked to visit four Web sites: CNN; CUinfo (Cornell University); Yahoo; and Amazon, and complete different tasks on each of these sites, except for the CNN Web site where the users did not have to complete any task. The GazeTracker software automatically detected user eye movements and provided gaze trails as shown in Figure 5.4.
Figure 5.4 Gaze trails for CNN Web site (Newman, 2001)
The software also generated multiple graphs representing different characteristics of users’ gazing behaviour. One example is a three-dimensional bar graph. The height of each LookZone represents total time spent looking at it as shown in Figure 5.5.
Figure 5.5 Bar graph provided by GazeTracker (Newman, 2001)
From the visual analysis of gaze trails and graphs, eye gaze patterns on the CNN Web site show that users tend to focus on new information and overlook objects that always stay the same or do not contain much information, such as the navigation bar and pictures. For the other three Web sites, it was found that the Yahoo Web site was clearly the most difficult to find information needed to complete the tasks.
In addition to Newman’s (2001) study discussed above, there are also other examples of Web usability research incorporating the eye gaze tracking technique.
For instance, Stanford-Poynter (2000) conducted a research project that used eye tracking to observe users while reading online news. The research was conducted using the EyeLink eye tracking system from SensoMotoric Instruments (later transferred to SR Research). The EyeLink head mounted tracking device was used in conjunction with software written by Stanford to capture both eye movements and
screen content. The participants were allowed to read news on their favourite Web sites. The findings from this research included:
• At the first glance of a page, users focus on text before graphics or photos.
• Users look at banner advertisements and perceive them even though they do not click on them.
• Users scan the whole page first to find topics of their interest and then focus on them.
• Users switch back and forth between multiple Web sites instead of reading them serially.
Cowen (2001) used eye movements to evaluate the usability of the Orange company’s homepages from four different countries: Belgium, India, Switzerland, and the United Kingdom. The eye tracking hardware used was SensoMotoric Instruments’ Head Mounted Eyetracking Device II (HED II) with Scene Camera.
The software used was iView developed by the same provider. In the experiment, participants were asked to complete two tasks on each of the four homepages. Two types of data were gathered from the experiment: performance data and eye movement data. Performance was measured by response scores and task completion times, while eye movement data was analysed using four measures: average fixation duration, number of fixations, spatial density, and total fixation duration. Then analysis of variance was performed for each measure. Inferring from the performance data analysis, the author found that there were significant interactions between the usability of a given page and the task performed on it. However, no
significant interaction between page and task could be inferred from the eye movement data analysis.
Goldberg, Stimson, Lewenstein, Scott, and Wichansky (2002) conducted an eye tracking study to evaluate the usability of Oracle’s Web portal prototype software. The eye tracking hardware used in this study was a SensoMotoric Instruments’ EyeLink eye tracker (version 1.1), while the software used was the iView on-line analysis software. The participants were asked to complete three training tasks and six experimental tasks on custom-made Web pages, using Oracle’s Web portal prototype. The screen was divided into small areas (called portlets3), in which Web content was displayed. The results of the study showed that for a Web page with two columns of portlets, users preferred to search horizontally across columns instead of searching vertically within a column. Another finding was that the participants did not always rely on a portlet’s header bar for navigation. In addition, this study also implies several recommendations for portlet design, such as that portlets in the left column receive more attention than those in the right column and that the two most important portlets should be placed in the top two positions in the left column.
3 Portlets are the visible active components end users see within their portal pages. Similar to a window in a PC desktop, each portlet owns a portion of the browser or PDA screen where it displays results. (IBM, 2003)
5.4 Procedures for Web Usability Evaluation with Eye