Any discussion about the use of hypertext for learning or in the broader context of information retrieval seems incomplete without some mention of the disorientation problem. There seems to be a growing consensus of opinion that disorientation is the major limiting factor of hypertext (Kim & Hirtle, 1995; McDonald & Stevenson, 1996). Apart from the subjective feelings of bewilderment or confusion that the user might have, the problem seems to result in a measurable decline in performance. In general, disoriented users may encounter problems in deciding where they want to go in hypertext and how to get there (Kim & Hirtle, 1995). Consequently, the problem has the potential to simultaneously disrupt browsing and navigation in hypertext. In terms of browsing, users stop reading too soon when they are left to decide what, and how much, information to read (McDonald & Stevenson, 1996). In terms of navigation, users typically take longer to find information (Mohageg, 1992) and are unable to plan and execute direct routes through the text to reach desired information (McDonald & Stevenson, 1996).
A major aim of this study is to examine the effects of different hypertext topologies on navigation performance and user disorientation. The topologies examined in this study were hierarchical, nonlinear, and mixed. In hierarchical text the nodes were linked to form a strict hierarchy, for which a node at one level could access only those nodes directly above and below it. In nonlinear text the nodes were connected to form a complex network based on a large number of referential links. Mixed text had a basic hierarchical structure with a number of referential links that allowed users to jump across the branches of the hierarchy. We anticipated that nonlinear hypertext would produce the most disorientation in both browsing and navigation. We also anticipated that mixed text, which contains both hierarchical and cross-referential links, would facilitate efficient navigation better than purely hierarchical text.
Hypertext's flexibility allows nodes and links to be arranged in a variety of information structures. The most popular structures are hierarchical and network structures. Hierarchical structures connect nodes in a strict hierarchy, for which a node at one level can only access nodes directly above or below it. Network structures connect a node to any other node to form a complex structure with many links. The supposed advantages of nonlinear, or network, structures are twofold. First, this type of structure is intended to make information more accessible to the reader. For example, in hierarchically structured hypertext, moving from the top of the hierarchy to a node at the bottom may require the user to traverse several links, whereas the identical trip in a network structure of the same document might require only a single link. Second, the network structure allows nonlinear access to the information. Readers can choose to follow a variety of paths through the document, increasing their control over the sequencing of information.
However, this increase in control may have negative consequences. The advantages of nonlinear text may be severely limited if users are unable to find their way around unfamiliar and complex information structures without experiencing disorientation. Therefore, we need to identify structures that reduce the possibility of getting lost but that still embrace the essential goal of hypertext - to allow users some control over how they access information.
Mohageg (1992) examined the question answering performance of beginners using an information retrieval system with one of three structures: hierarchical, network, and a combination of hierarchical and network structures. He found that performance was poorest with the network structure but found no performance differences between the mixed and hierarchical structures. Mohageg concluded, therefore, that a mixed structure did not benefit beginners. …