Donskoy, Misha and Kapetlinin, Victor (1997). Window navigation with and without animation: a comparison of scroll bars, zoom, and fisheye view.

By Reginald Riley Jr.

The basic idea behind this paper is that an attempt is made at an experiment to find the effect of animation on different types of navigation techniques, specifically: scroll bars, zoom, and fisheye view.

It begins with giving background information on animation, but more specifically, its role in the world of HCI. Even though animation is considered important resource, it is still largely unexplored. It continues to define an important benefit of animation, "object continuity" where the user can actually see transformations taking place as oppose to the user having to interpret the relationship between the initial state and final state, or between individual snapshots.

This view of "object continuity" is the basis for the experiment where it is hypothesized that "Introducing animation into window navigation can make navigation more efficient." It was believed that animation would be more helpful with fisheye navigation but less helpful with zooming techniques.

The experiment was conducted with twelve Swedish university undergraduate students from 19 to 24 years old all with prior experience with Windows 3.1. In every task, the students were presented with four windows; one of which was empty and the others consisted of 48 files each. Out of all 144 files, the students had to find the twelve files with either .txt, .doc, or .wrt extensions and drag them to the empty window. The file filled windows only displayed about 25% of their entire content at any one time, thus requiring navigation to accomplish the task.

As stated earlier, the three techniques were your standard scroll bar, zooming where the user could zoom out to get an overview of all the files, scaled down to fit the screen, and fisheye view where all items could be seen on the screen; however, the icons representing the files that were near the focus were larger, while those farther were smaller. Each version was tested with animation and without animation. With no animation, every navigational action resulted in displaying a new image of the window content. With animation, one transition state was added in between the initial and final displays.

In each of the 6 experimental conditions, the students worked through a series of six tasks, chosen according to a Latin Square plan, consisting of two learning tasks, and four main tasks. Data was recorded in a log, and each student was asked to evaluate each technique, based on his or her own preference, both with and without animation.

The results were not statistically significant. Zoom and scroll bars had slower performance with animation, and fisheye view demonstrated faster performance with animation. However, in all of the cases, students preferred animation to no animation. As discussed, these results may be a result of only one transitional state being implemented for animation, or that no transitional state was even needed for animation.

This study wanted to find the results of animation being applied to different navigation techniques. First, I find the hypothesis to be very weak; "Introducing animation into window navigation can make navigation more efficient." Maybe this is nic-picking, but the use of the word "can" really does not provide a concrete statement to prove as being true or false.

I do not understand why one would consider the addition of only one transition equaling animation. The results did not surprise me based on this, yet it seemed as though they were. To get true results, it seems that a totally animated transition should have been implemented, then compared to an initial/final state method. Additionally, because they developed the tasks, why would they test, and not know if additional steps were needed to animate the action? That is an important variable that should have been specifically controlled.

After many attempts, I still do not understand the entire ANOVA section, or the significance of the p variable.

It was odd that they implemented zooming, without panning. It seems very inefficient to have to zoom all of the way out, select an area, then zoom all of the way back in again. However, zooming proved to be the fastest technique. This is in contrast to fisheye view, which took the longest. Maybe fisheye would have been faster if the focus point changed constantly as the mouse moved. Additionally, scroll bars may have been faster if the students were able to utilize the wheels that we have on mice today. This all can be summed up in that, the resulting times can be attributed to the functionality behind the technique, or the lack there of.

If the notion of "object continuity" was presented and considered to be a benefit of animation, why were steps not taken to make the animation in the experiment as continuous as possible?

Are twelve students enough of a sample for this type of experiment.

What exactly is a Latin Square plan?

Which is more important, the time it takes to require a task or whether the users are completely satisfied with the available tools needed to accomplish the task?