In this study we explored the use of log files as a window into the process of hypermedia navigation. Although there is a growing body of research addressing theoretical and design issues related to open-ended, non-directive technologies such as hypermedia, relatively few studies have attempted to explain navigational performance. Sixty-six undergraduate students used a multidimensional, computer-based kiosk that could be explored in a nonlinear fashion to find information in response to one of two information-retrieval tasks (simple or complex). Cluster analysis was used to generate performance profiles derived from navigational data captured in log files. Analyses of within-cluster performance profiles, combined with external validation criteria, led to the classification of four different types of navigational performance (models users, disenchanted volunteers, feature explorers, and cyber cartographers). These characterizations were consistent with information-retrieval scores and the external criteria (self-efficacy, perceived utility, and interest). For example, individuals who appeared to take the time to learn the layout of the kiosk also had the highest self-efficacy, while those who used the help screen and watched the most movies had the lowest self-efficacy. Results also demonstrated an interaction between various individual navigational profiles and type of information-retrieval task.
ASJC Scopus subject areas