The Effect of Zooming Speed on a Zoomable User Interface

Conclusions

Impact for practitioners

We can draw an optimum conclusion from consistent results for these three dependent variables of performance time, retention accuracy and subjective preference. Generally speaking, faster zooming speed statistically has more efficient performance and better subjective satisfaction. Slower zooming speed has less efficient performance and worse subjective satisfaction.

We concluded that efficiency for zooming speed had a saddle shaped pattern. The performance time was longest at the slowest zooming speed, 1.5 factor/second. The performance time was also long for the fastest zooming speed. The shortest mean time appeared at the zooming speed 8 factor/second in the middle of the zooming speed scope. The zooming speed of 4 factor/second and 12 factor/second also had the reasonably good performance time.

Against our second hypothesis, the zooming speed has little, or no significant effect on the retention accuracy. The sum of Accuracy has similar pattern for the six zooming speed from the Retention Accuracy chart. Since the retention has close relationship with memory and our experiment is randomly done with speed up and speed down. The retention accuracy is a little worse for the middle zooming speed treatments since they happen to be in the middle of the experiment.

For the third hypothesis, lower zooming speed will decrease the subjective satisfactory rate while higher zooming speed will increase it. We conclude that they prefer high zooming speed but we are not able to find the range for the best subjective preference. They feel not annoyed at all when the speed is the highest, in our case, 16 factor/second. The possible reason is that we didn't provide high enough zooming speed to clarify the issue.

From the analysis above, we narrow down a range of zooming speed at which recognition accuracy, efficiency and the subjective satisfaction are good. We suggest the ideal range is from 4 to 12 factor/second. This may be taken into account for the future research and implementation.

Suggestions for future researchers

JAZZ's performance depends on the platform greatly. The zooming speed varies although we set the zooming speed value to be fixed. The zooming speed is faster on the Microsoft Windows NT/98/95 than the Sun's Solaris Ultra 5. The Jazz's performance on Microsoft Widows NT/98/95 isn't reliable and our experiment application crashes from time to time. This is perceived in our pilot experiment. We decide to do all our following experiment on Solaris ultra 5 to avoid inconsistency.

The experiment subjects are mainly graduate and undergraduate students from the CS department of UMCP. Most of them are familiar with computers but not with JAZZ. The performance time is longer and the recognition accuracy isn't good for the first several questions, which reflects that our experiment is short of training. The zooming is achieved by pressing the right button and by sliding right or left at the same time, which is labeled as "strange" by some of our subjects. We perceived this problem in our pilot experiment. We include an introduction page with a JAZZ window for them to play around and a warm-up question. The training seems still not enough. Training is important to help users, even expert level computer users to get used to a novel graphic interface.

Another suggestion is that the performance of the subjects might be analyzed. We find in our raw data that people's performance times differ greatly. It's not easy to compare users' personal ability in searching a map but it's really an important factor affecting the overall outcome.

Refine the theory

However, our experiment results are not enough to conclude universal users performance since our subjects are mainly expert level computer users and young with good visual ability. The problem with this experiment is that we can't gather diverse subjects to participate in the experiment. Experiments with subjects of different computer backgrounds, ages and visual conditions, and experiments done on different platforms will lead to a more universal conclusion and reflect a more realistic situation.

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