Information seeking in the WWW: Detecting T-patterns in eye movements and navigational behavior

D.C. Unz¹, B. Werner¹ and R. Mangold²

¹Media and Organizational Psychology, Saarland University, Saarbruecken, Germany
²Hochschule der Medien, Stuttgart, Germany

Using a search engine is one of the most commonly used services of the internet. Half the users spend more than 70% of their online-time seeking for information. But nearly half the users are frustrated about using a search engine. Therefore the question how to facilitate the processing of search result lists is posed. Whereas studies that address this question typically simply count certain aspects of user’s behavior, as the opening of documents, or look at the overall search times, the ongoing process over time is often not addressed along the timeline. The aim of our studies is to account for the temporal course of user’s behavior and address the process in time. In particular, we ask, what kind of heuristics user do pursue when inspecting result lists of search engines.

In a first study (42 subjects were presented with a search results page that listed 25 web pages retrieved for a query about ‘assessment centre building blocks’) we recorded subjects’ eye movements and mouse clicks and looked at the order in which the search results were processed. Most of the subjects seem to apply a sort of strict depth-.rst strategy, a minority seems to apply a breadth-first strategy, and another minority seems to apply a partially breadth-first strategy.

In a second study 50 subjects were asked to perform four tasks similar to those of the first study. Again, subjects’ eye movements and mouse clicks were recorded. The subjects were assigned to different ‘simple heuristics’ according their statements in a questionnaire. We asked if we can find different (observable) behavior patterns of persons that pursue different (unobservable) heuristics. To study the temporal course of processing the search result list in detail, the observational records of eye movements and mouse clicks are analyzed for its temporal and sequential structure by the means of the T-pattern detection method proposed by Magnusson and operationalized in the software Theme™ (PatternVision Ltd, Iceland).

Paper presented at Measuring Behavior 2005 , 5^th International Conference on Methods and Techniques in Behavioral Research, 30 August - 2 September 2005, Wageningen, The Netherlands.

© 2005 Noldus Information Technology bv