Time - Interface x Task
Time - Interface x Outcome
phosphor animation phosphor animation
Figure 17: (a) Task time in seconds and (b) error rate by interface and task. (c) Task time in seconds and (d) error rate by interface and outcome (+/- standard error of the mean).
outcomes were faster than the other two outcomes (Figure 17c). These two error conditions involved a wrong icon moving, which participants detected early on and dismissed these trials instantly. Correct and ErrorUndershoot out- comes, on the other hand, required participants to wait for the animated transitions to complete. Consequently, we saw higher task times overall and in particular a strong negative impact in the slower animation conditions.
Error rate: Similarly, for the error rate metric, we observed a main effect of Interface (F(5,55)=4.704, p<.001), driven mainly by differences between the 2000ms animation Inter- face and the Phosphor, 125ms, 250ms, and 500ms condi- tions. We also saw a main effect of Task (F(2,22)=16.985, p<.001). As expected errors in the TripleIcon condition were significantly higher than that of Distractor (p<.006) and SingleIcon (p<.001) tasks (Figure 17d). Finally we saw a main effect of Outcome (F(3,33)=21.243, p<.001), with all pairs significantly different except for the correct v. Er- rorUndershoot and ErrorNeighbor vs. ErrorOther. Error- Other and ErrorUndershoot were easier to detect and there- fore led to low error rates (Figure 17b). The error in the ErrorNeighbor task, on the other hand, was easily missed and led to an error rate even higher than the Correct out- come.
Subjective preference: Participants selected one “favorite” interface condition after completing each task. To simplify selection, we offered the four mnemonic choices ‘super fast’, ‘fast’, ‘slower’, ‘super slow’, and Phosphor. Prefer- ences varied between tasks. For the singleIcon and the Dis- tractor tasks, the fast condition was a favorite (7 out of 11 participants). For the tripleIcon tasks, preferences were more balanced with 3 participants each preferring the fast, slow, and phosphor. The reason most commonly given for the lower popularity of the phosphor interface was that it was more distracting, especially in the distractor task.
The slow animation conditions received very low satisfac- tion rating throughout, despite their superior error rate. Sev- eral participants expressed a strong dislike for the wait time caused by these conditions.
Discussion The findings of our second user study indicate that the per- formance of the phosphor interface is comparable to ani- mated transitions. For all pairs but the fastest animation condition, Phosphor was significantly faster. For the 125 ms animation condition, there was no significant difference between conditions. This suggests that the benefits of phos- phor do not come at the expense of reduced task perform- ance.