Performance effects of keyboard navigation in web systems

Abstract

This article reports performance tests from users navigating in Web systems with the keyboard. Different Web interfaces were developed and tested with special and ordinary users. As initially supposed we verified that the two groups took longer time doing selection tasks and search tasks with the keyboard when compared to the traditional mouse. In general terms, selection of menu options was the most affected task. With the keyboard access special users needed 29% more time selecting options from a Web Interface that had an Horizontal Menu (Interface HM) and 200% more time with a Web Interface that had a Vertical Menu (Interface VM). The other users needed 71% more time, and 84% more time with the same interfaces respectively. In selection tasks the change in the number of necessary attempts for correct answers was not significant. We also verified that the effect of keyboard access on user performance was much smaller in search tasks than it was in selection tasks. Ordinary users with the mouse and keyboard needed the same time doing search tasks with a Web Interface that had a Single Page for each different topic (Interface SP). With a Web Interface that had each topic distributed across Multiple sequential Pages (Interface MP), ordinary users needed 15% more time in search tasks with the keyboard access. The special users needed 9% and 126% more time for the same interfaces, respectively. In search task experiments, we got a curious result, special users got a higher rate of correct answers with keyboard navigation. This result may show that keyboard navigation can help special users doing a more systematic search work. With one of the developed interfaces special users got a 33% to 63% improvement (a 88% change) in the correct answers. This effect was no noticed in the other groups.