Talk presented at Human Factors and Ergonomic Society Houston Chapter Conference (May, 2006)
Background: Consistency in Web site design improves user performance (e.g. Detweiler, & Omanson, 1996; Nielsen, 1999), and users form expectations about navigation and the likely location of information based on previous use of the Web sites in general (e.g. Markum & Hall, 2003; Grier, Kortum, & Miller, in press). However, how does previous exposure to a specific site impact users' ability to find information and notice new content on that same site? More specifically, if users have been previously successful at finding information without a shortcut link, will they be less likely to notice new shortcut links to that information? What happens if a previously used link is removed? Will the location of the link influence its likelihood to be noticed?
Method: The current study employed a 2 (Shortcut Link Placement: left side or within body text area) x 2 (Shortcut Link Presence Consistency: yes or no) x 2 (Presence of Shortcut Link on Repeated Task: yes or no) mixed design. Forty-one participants completed two search tasks and then repeated both of them (four searches total). For each search task, participants started at the front page of a Web site and then explored the site until they found the answer to a question. The answer was located three clicks away (most efficient path) without the shortcut link, and only one click away with the shortcut link. Participants typed the answer into a response box at the bottom of the page. A difference score was calculated for the search times and for the number of steps in the search path between the first and the repeated search for each task. Participant comfort using the Internet was used as a covariate in the analyses.
Results: ANCOVAs showed significant main effects for Consistency (consistent link presence led to faster searches with fewer steps for the repeated task) and for LinkPresence on Repeat (faster and fewer steps on the repeat task whenthe link was present). However, significant interactions between these variables showed that the absence of the link on the repeated task was only detrimental if it had been present during the first search for that task. The largest gains in performance occurred when either the link was consistently not there, or when the link was only present on the repeated task. A significant 3-way interaction for the number of steps showed that removing or adding a link to the left navigation was more influential than when the link was in the body text area.
Discussion: Web site designers by necessity will make changes to a site, but users will often leave a site if information is not readily found (Nielsen, 2003, 2004). The current results show that even one previous experience finding information on a site can influence a user's subsequent performance, especially if relevant shortcut links are changed. Making changes in expected navigation areas (e.g. left side links) should be done with care because of the greater impact on performance.