ABSTRACT
There is a central issue in stimulus selection that has to do with the experimental context: is the experiment going to be a user study of an existing set of data (such as whether a particular webpage is effective in attracting the viewer’s gaze to specific parts on the screen), or is the experiment going to test a non-specific experimental question (such as which elements in general attract the viewer’s gaze when browsing)? The issue is therefore whether the possible test set is constrained from the outset or not. If it is constrained, then the matter reduces to which elements of the whole dataset become part of the test set in the experiment (for example, it might be unfeasible to test all individual pages contained in the webpage that one is interested in); that is, the question is how to sample properly. If the test set is unconstrained, the problem is more general, since then the researcher has to come up with a suitable set.
