Conclusions

In general, in all of the dependent variables measured, the expandable table of contents users performed worse and had less favorable attitudes towards the system than the traditional table of contents users. The provision of multiple-frame does not lead to superior performance or speed in searching and browsing task. The lack of this significant main effect might be because the affect of an expandable table of contents overshadowed the good attribute of the multiple-frame. Finding significance in the interaction between table of contents and frame on users' performance, speed, and user satisfaction has led to the idea of holistic comparison in which the ensemble of software and hardware making the system is tested without decomposing its component variables (Wiedenbeck & Davis, 1997). In fact, Whiteside, Jones, Levy and Wixon (1985) argue that a reductionist experimental strategy is futile in investigating realistic systems because differences are due to an inextricably complex interaction of causes. Under these circumstance and the results of this study, it must be reasonable to identify the experimental goal in the largest sources of variance and investigate how they affect users. This suggests a need for further study of a holistic comparison to determine exactly how the combination of table of contents and frame affects users' performance and perceptions.