| Faraday (2000) has proposed a model of how users search for information on web pages that emphasizes Salient Visual Elements (SVEs), which capture attention and draw the eyes to them. Six SVEs in hierarchical order of importance, motion, size, image, color, text style, and position are held to determine the first fixation on the page, and therefore, the entry point for the search. Once an entry point has been selected, Faraday asserts that users employ Gestalt grouping principles (e.g. proximity, similarity, etc.) to define the boundaries of the area surrounding the entry point and then scan that area in the normal reading order of the native language. The process is iterated with successive entry points until the required information is found. To date, the Faraday (2000) model is the only model of visual attention in web design, and although it has generated considerable interest, it has not been tested empirically. The present investigation provides the initial experimental evaluation of the applicability of the model. Four experiments are reported using the ERICA system (Lankford, 2000) to monitor eye movements.;The initial experiment utilized simple e-commerce web pages to verify predictions derived from the model regarding observed fixation sequences, the location of first ocular fixations, and transitions in gaze from one portion of the page to another when participants were set to search for designated targets. None of the predictions was supported, leading to questions regarding the fundamental adequacy of Faraday's hierarchal ordering of dominance among the SVEs. Two subsequent studies measuring initial fixations when participants were permitted to free-scan simplified non-web page displays indicated that while motion does attract first attention, it is rivaled by position. However, when used with realistic web pages in a final study in which participants searched for specific targets, neither motion nor position dominated target detection or gaze behavior. Given that expectancies have been found to guide many search activities (Wickens & Hollands, 2000), future research might well focus on expectations rather than upon SVEs in generating models of visual attention in web design. |
