Abstract
In the usability testing of Web sites, thinking aloud is a frequently-used method. A fundamental discussion, however, about the relation between the use of different variants of thinking aloud and the evaluation goals for this specific medium is still lacking. To lay a foundation for this discussion, I analyzed the results of three usability studies in which different thinking-aloud tasks were used: a simple searching task, an application task and a prediction task. In the task setting, the profile of the Web surfer, the communication goal of the Web site and other quality aspects are taken into account. The qualitative analysis of these studies shows that the task variation has some influence on the results of usability testing and that, consequently, tasks should be matched with the evaluation goals put forward.