Contrary to our expectation of an inverse relationship between mouse pad size and performance in mouse tasks, no significant differences were found in the performance of subjects using different sized mouse pads. In the Clicking task use of the medium size mouse pad allowed for slightly better performance than the smallest mouse pad, but the largest pad unexpectedly yielded the slowest performances. The dragging task conformed to our hypothesis that the smallest pad would result in the slowest performances and the largest in the quickest, but the performance difference was not large enough to confirm the hypothesis.
The operative hypothesis in this experiment was that there would be a significant difference in performance between the various mouse pads, but the failure to observe such a difference was not entirely unexpected. With the mouse tracking settings used in this experiment, the default Windows 95 settings which most users likely use, the entire screen could be traversed without having to pick up and reposition the mouse while using a surprisingly small mouse pad - 3.6 cm X 2.7 cm. Since all three of the mouse pad sizes used in this experiment were larger than the minimum needed to complete the tasks without having to reposition the mouse, that time consuming correction should not have impacted the results if subjects moved the mouse accurately. Outside of personal mouse manipulation style, in which case the occurrence of repositionings would be independent of mouse pad size, the need to reposition the mouse would only occur if subjects overran targets at the extremities of the screen. For example, by continuing to move the mouse to the left after the pointer has already reached its leftmost point, the position of the mouse on the mouse pad will no longer reflect the position of the cursor on the screen. While one such error is unlikely to lead to the need to reposition the mouse, several errors compounded could easily lead to that need. Unfortunately for the accuracy of this experiment, the subjects exhibited more careful mouse manipulation than they might in everyday usage due to their realization that they were participating in a serious experiment where irresponsible participation would adversely effect the findings of the researchers.
As a result of the careful mouse manipulation by the subjects, fewer gross movement errors were made, and since the number of such errors which could be accommodated without repositioning is the most significant difference between the three sizes of mouse pads, the primary expected origin of performance variation was effectively eliminated. It is possible that the 1.6 second difference in performance between the smallest and largest mouse pads in the Dragging task, the more complex of the two tasks performed by the subjects and thus more subject to induce errors, could be the result of the differing ability to accommodate errors. While that 1.6 second difference in task completion time between smallest and largest mouse pads may not be considered statistically significant, we believe that it is representative of a genuine advantage inherent in larger mouse pads in regular usage situations.
Fitt's Law would, upon initial consideration, not be of much use in analysis of this experiment. Though a predictive measure of time to point seemed to be appropriate for this analysis, the components composing the application of Fitt's Law in this case do not vary between the mouse pad sizes. The index of difficulty does not vary as the distance between and size of objects in the tests remain constant across the sizes of mouse pads. Given that the results showed no significant difference in performance between the mouse pads, the constants used in conjunction with the index of difficulty to predict pointing time must be the same for all mouse pads. Since the same mice were used and all properties other than the size of the mouse pads remained constant, these constants should be the same, as the pointing device in question did not significantly change (and even though the size of the pad changed, there was no reason that change need effect results).
Observations made during the testing, and comments of several subjects, suggest that Fitt's Law may have an application to this experiment. While most subjects manipulated the mouse the same whatever the size of the mouse pad, some of the subjects were observed to employ different mouse manipulation strategies depending on the size of the mouse pad. These subjects would move the mouse by just moving their wrist while using the small mouse pad, but when using the large mouse pad would move their arm at the elbow and sometimes shoulder. Though there was no reason physical reason for this change, several subjects did change to more complex movements as the size of the mouse pad increased, perhaps because as one subject expressed, "(the large pad) makes me feel like I have a long distance to move the mouse pointer." This complex movement cannot be captured by the same basic Fitt's Law equation which captures the simple, wrist only movement. The complex motion is better captured by a multiple component equation like that in Shneiderman (325)to describe precision pointing. Though the pointing tasks have not changed in this experiment as the size of the mouse pad changed, through their own changes in mouse manipulation style these subjects have effectively turned the tasks performed with the large mouse pad into precision pointing tasks that require both gross and precise movement.
If the results produced by this experiment are believed to be those which would be predicted by a Fitt's Law calculation (the required constant values have not been determined for this experimental environment, thus leaving us unable to perform the calculation), it would be implied that the simple Fitt's Law formula and the more complex multi- component variation produce identical predictions for the tasks performed in this experiment. For these equations to produce identical values for other tasks, ones in which the distance between objects and the size of the objects are changed from those found in this experiment, would require constant values which represent no difference in difficulty between simple and complex arm motions. Previous studies (Sears and Shneiderman) have shown that there is a definite difference between simple and complex motions, supported by the comments of subjects ("Takes more effort to move mouse"), hence the need for the multi-component formula. Thus it must be concluded that the tasks used in this experiment, to the detriment of the ability of the experiment to produce telling results, are among the few which provide the unique parameters which result in the same values being produced by both predictive theories. Other tasks with different distances between buttons and different sized buttons would be more revealing.
The subjective questions asked of the subjects provide far more revealing results than the performance metrics, though those results are not all clear. A majority (53%) of subjects favored the medium size mouse pad, but nearly as many subjects preferred either the smallest or largest pad (with 23% of subjects preferring each of those pads). The least favorite mouse pad was the smallest (53%), but the largest mouse pad was also disliked by a significant number of subjects (40%).
A broader question, "Would you like to use this size mouse pad on a regular basis", further illustrated the polarized feelings toward the extreme sized mouse pads. Only 1 subject who did not list the smallest pad as their favorite found the proposition of regular use appealing. Subjects felt very strongly about the smallest mouse pad, whatever their feeling happened to be, and their comments explained those feelings. Those subjects who liked the smallest pad recognized that it was large enough to accommodate any mouse movement with room to spare, while also appreciating the desk space savings. As one subject commented, "I just had no need for a larger ones, so why take up more space on my desk." The majority of subjects were uncomfortable with the smallest pad, expressing concerns that their motions were restricted and as one subject wrote, "I always felt as if I was going to fall off the edge." Subjects were only slightly more accepting of the idea of using the largest mouse pad on a regular basis, again believing strongly in their convictions. Those in favor enjoyed that there was "Lots of room to move around, never ever have to reposition mouse", but the majority felt it was a waste of desk space. Not surprisingly, subjects were most accepting of the medium sized mouse pad, not only because it alleviated the fear of falling off the edge present with the smallest pad while not wasting space like the largest pad, but also because it was the pad most like those commercially available.
When asked how they would change the size of the mouse pads subjects responded in such a way that conforms to the preferences discussed above. The average of responses showed a clear preference for a mouse pad just slightly smaller than the medium pad as used in the experiment. Not only is the suggested change for the medium pad small, the changes suggested for the small and large pads are similar in scale and point to an ideal size very near the size indicated in the suggestion for the medium pad. Unlike the numerical results produced by the tasks, these suggestions cannot be rejected as statistically insignificant, F(2,89) = 32.92 (p<0.05, F- critical=3.1).