The Effect of Zooming Speed in a Zoomable User Interface

Results (statistical & objective report of what happened)

Note that the raw data, analysis tables are available in the Appendices.

Performance Time

We recorded the time that each subject spent to answer each of the twenty-four search questions (in raw data section of appendix B1). The questions were showed in the same order for each subject. Half of the subjects conducted the experiment from low zooming speed to high zooming speed. The other half of the subjects conducted the experiment from high speed to low speed. The following chart shows the mean performance times and the deviations of the performance time means for the twenty-four tasks. The purple data are the data from the subjects who did the experiment from high speed to low speed. The red data are the data from the subjects who did the experiment from low speed to high speed.

The twenty-four search tasks were organized into six groups that corresponded to six zooming speeds. The average difficulty of any group was similar to any other group. In the same group, the difficulties for different questions varied. We were more interested in the mean performance time for the whole group of tasks. The following chart shows the mean performance time for the six groups of tasks.

The one-way factor analysis of variance (ANOVA) on performance time for the six zooming speeds showed that as zooming speed increased from 1.5 to 16, the performance time varied (see appendix C1). This result was statistically significant at alpha = 0.01 (Fcrit(5,198) = 9.08). The performance time for the speed 8/sec seemed to be the best, so two more ANOVA tests were conducted to investigate whether the performance goes down if the zooming speed is too high or too low. Are the trends on both sides of 8/sec statistical significant? The one-way analysis of variance on performance time for the first three zooming speeds (16,12,8) showed that the performance time decreasing was not statistically significant at alpha = 0.01 (Fcrit(3,132)= 26.2). The one-way analysis of variance on the performance time for the speeds (8,4,2,1.5) showed that the performance time increasing was statistically significant at alpha = 0.01, (Fcrit(2,99) = 99.5).

Since ANOVA test didn't show significant trend of performance time as the zooming speed went up from 8/sec, all pairs of t-test would give sufficient statistical detail. Full set of paired t-tests were then run to determine whether the performance time variation was statistically significant between one zooming speed and another. Five paired t-tests between zooming speeds of (16,12) (12,8), (8,4), (4,2), (2,1.5) showed that the t-tests between zooming speeds of (16,12), (12,8), and (8,4) were not statistically significant (which is consistent with the second ANOVA result), while the t-test between zooming speeds of (4,2) and (2,1.5) was significant at alpha=0.01, (one tail, Tcrit(66) = -2.385 or 2.385) (see appendix C2). However, the differences between every other groups were statistical significant for pairs of (16,8), (8,2) and (4,1.5), except for (12,4). The t-test results showed that the performance differences also appeared to be statistical significant for pairs of (8,1.5), (12,1.5), (16,1.5) and (12,2).

Retention Accuracy

The retention accuracy was measured by assigning 1 point for a correct answer, 0.7 point for the answer that is close to the correct answer and 0 point for a totally wrong answer. The following chart shows the retention accuracy for the six different zooming speeds. We listed the retention accuracy data of the subjects who did the experiment with speed going up separately from the subjects who did the experiment with the speed going down. One-way ANOVA statistics were performed for the six questions. The result showed that the difference wasn't statistical significant (showed in appendix C3) with f=1.036 and Fcrit(5,198)=9.08 at alpha=0.01.

Subjective Preference

Each subject was asked to answer one preference survey question. The survey question asked subjects which speed was the most uncomfortable speed for them. The following chart shows the survey result. There were 25 subjects who thought the low speeds were annoying. Three subjects dissatisfied most with the fast speeds. Six subjects answered no preference. No subjects dissatisfied with the medium speeds.

Department of Computer Science: Direct questions and comments to the student editorial team