This demonstration of web-evaluation was constructed using a few pages from RU Cook extension site. The task was to search for information on 'deer resistant plants.'
Images used in the demo along with videos show simulation of me looking for links on deer-resistant plants. The dynamic and hierarchical nature of this task made it difficult to show where my eyes were scanning so instead video show my mouse cursor moving around & then finally selecting the links. (I tried to synchronize the mouse cursor movements to the movement of my eyes.)

The actual eye-tracking experiment was tested once on the web designer of the pages-- Bonnie Wasielewski. As you might expect, finding the appropriate pages was very easy. Bonnie was very rapid finding and selecting appropriate links-- both of which accurately got her to information she was seeking. Furher, she used very few (saccadic) eye-movements and fixations to get to these links. A novice is likely to take much longer and produce many more eye-movements and fixations. This speaks to two important points relevant to evaluating web pages:

1) this methodology can serve to distinguish experts from novice searchers. As just noted, a number of eye-movement , response, time, and accuracy measures can be used as a tool to distinguish individual differences in expertise.

2) It is important to consider who your audience is when designing web pages. Are they web savy? As Bonnie pointed out, older audiences tend not be as computer savy as younger students; so a designer's choice of 'selection' modes is critical.. Drop down menus (without instructions) can be a problem for people unfamiliar with contemporary web design.