|< Day Day Up >|| |
H1 and H2 were strongly supported. High-demand tasks resulted in higher error rates and longer task completion times as compared to low-demand tasks, regardless of the application being used. Interestingly, the only difference between the high- and low-demand tasks was the quantity of information involved in the modifications that were performed. However, by requiring participants to complete larger modifications we increased both physical and cognitive demands. A direct increase in physical demands results from the additional characters that must be inserted or removed. A direct increase in cognitive demands results from the additional information users must remember while completing the tasks. Indirect increases in physical demands that occur as a result of the increased cognitive demands may affect all three components of the task completion time.
More importantly, we can isolate the direct increase in physical demands as affecting only the data entry component (tD) of the task completion time. The two remaining components of the task completion time (tN and tO) are not directly affected by the additional physical demands associated with the high-demand tasks. Therefore, any additional physical activities that occur during non-data entry portions of the task completion times can be considered indirect results of increased cognitive demands. Therefore, we can conclude that any increases in tN or tO result either directly or indirectly from increased cognitive demands. In contrast, increases in tD may result from increased physical demands or increased cognitive demands.
The results reported above confirm that both tN and tO were significantly longer for the high-demand tasks. This confirms that increased cognitive demands explain a significant portion of the increase in task completion times. More importantly, tN and tO account for 74% to 80% of the total task completion times as well as 68% to 83% of the difference in task completion times when the high- and low-demand tasks are compared. These results confirm that the increases in cognitive demands had a greater affect on task completion times than the increase in physical demands, that non-data entry activities dominate the total task completion time, and that non-data entry activities dominate the increase in task completion times observed when comparing the high- and low-demand tasks. These results strongly support H3 and are important since they confirm that, even in the context of clerical tasks such as those performed in this study, cognitive demands should be given the first priority when designing tasks.
H4 effectively states that there is a one-to-one mapping between significant reductions in total task completion time (t) and significant reductions in navigation time (tN). In other words:
significant reduction in t ⇔significant reduction in tN
As noted earlier, our participants were free to adopt any strategy they felt would be appropriate when completing their tasks and that strategy selection will be based on previous experiences with similar tasks. However, it is also important to note that participants were randomly assigned to one of the three platforms, minimizing the opportunity for systematic differences in strategies between platforms. Therefore, we believe it is appropriate to associate significant differences with differences in the computing platform being used as opposed to the underlying preferences of the individual participants.
The 400 platform resulted in significantly shorter task completion times for one task/application combination when compared to the 133 platform: the Excel-based high-demand task. Importantly, the 400 platform also resulted in significantly shorter navigation times (tN) for one task/application combination when compared to the 133 platform: the Excel-based high-demand task. Interestingly, the navigation subcomponent of the task completion time accounted for more than 75% of the reduction in task completion times for this task/application combination. These results support H4.
It is important to note that the navigation required in Word and PowerPoint tasks is different from that required in Excel-based tasks. In both Word and PowerPoint, navigation is typically accomplished in two stages. First, the user navigates to the correct page within a document and then they locate the desired item on that page. Locating the correct page requires scrolling in one-dimension. Locating an item on a specific page may involve scanning the entire page, but scrolling is typically still one-dimensional. Each page is of limited size, further limiting the complexity of the second navigation activity. In Excel, navigation activities typically involve locating a specific item on a single page. This page may vary dramatically in size between navigation events and navigation often involves scrolling in two dimensions. We conjecture that these fundamentally different requirements for navigation may explain the differences observed between the different applications, but additional research is necessary to validate this conjecture.
The Excel low- and high-demand tasks also resulted in different navigational outcomes. Navigation was not effected by the platform for the low-demand task, but did improve significantly for the high-demand task. We conjecture that this difference is due to the increased cognitive demands of the high-demand task. While the exact cause of the different performance is not known, those individuals using the 400 platform appear to have adopted different strategies than individuals using the 133 platform when completing the high-demand Excel task. At this time it is not possible to explain why they adopted different strategies, the details of their strategies, or the potential benefits of employing the same strategy regardless of the platform. Additional research is necessary to determine why users adopt different strategies when utilizing different computing systems. Results of such research could allow systems to be designed such that users are actively encouraged to adopt more effective techniques for navigation. The results of the current study confirm that decisions to upgrade computing platforms should not be driven by desires to reduce data entry times. Instead, such decisions should be motivated by the possibility that more powerful computing platforms will enable more efficient navigation, given the applications being used and the tasks being performed.
H5 is strongly supported. Several details of the experimental conditions are important when interpreting these results. First, the participants were never provided with any details regarding hardware being used (e.g., processor speeds or available memory). Second, all of the components used to complete the tasks (e.g., mouse, keyboard, and monitor) were identical in both appearance and functionality. Therefore, any differences in user perceptions resulted from differences in how the system responded to keyboard and mouse activity.
The general trend is for more favorable ratings as hardware performance improved, but significant differences were not apparent until the hardware performance exceeded that which the participants were accustomed to. At the time data collection occurred, our participants regularly experienced systems similar to the 133 and 266 platforms, but were unlikely to have had experience with systems as powerful as the 400 platform. As a result, participants who used the 400 platform provided positive evaluations while participants using the 133 and 266 provided ratings that tended toward negative to neutral. These results confirm that user perceptions are not an effective metric for assessing increases in productivity when upgrading computing systems. Users may notice and appreciate the additional computing power available from more powerful computing systems, but this does not necessarily map to increases in productivity.
|< Day Day Up >|| |