The performance on the perspective display remains above the plan-view performance until 105 seconds into the experiment, at which time the mean for the perspective view is 9.1 aircraft correctly marked, and the mean for the plan-view is 9.0. At this point, 90 percent of the descending aircraft have been identified; it is only the remaining 10 percent where the plan-view display allows faster completion of the task.
Unlike Scenario A, Scenario B showed virtually no difference in performance between the plan-view and perspective display. The plots of mean performance are almost identical. The plan-view performance plot for Scenario B is similar in its linearity to the plan-view performance in Scenario A. However, at each time step, the mean number correctly marked in Scenario B was approximately twice as high as in Scenario A. Given an exhaustive search strategy and twice as many descending aircraft to be found, it is to be expected that the descending aircraft will be found at a much higher rate. The perspective display did not exhibit this dramatic increase because the participants were not employing an exhaustive search strategy. They were finding the clearly-descending aircraft first, moving as quickly as they could, and having more descending aircraft didn't allow them to scan the display any faster.
When using the perspective display, the time-to-complete plots for both scenarios showed long tails due to the fact that the participants had considerable difficulty finding the last few descenders. There were several reasons for this difficulty.
Pre-test trial runs suggested, and the actual test runs confirmed, that participants had considerable difficulty finding the last few descenders. We therefore decided to analyze task performance for the time required to detect only 70 percent of the total number of descenders. This analysis confirmed the speculation that performance for the perspective display would be better early in the trial run. The participants could very quickly find most of the descending aircraft because their visual cues were fairly obvious. When the obvious descenders were exhausted, the participants had to resort to other strategies, such as hooking to read the textual display. When considering the two scenarios separately, the advantage enjoyed by the perspective display was not so clear-cut. For Scenario A, the perspective display was clearly better, but it was only slightly better for Scenario B. There may be two reasons for this result;
In general, the participants mismarked very few non-descenders on either display for either scenario. Nearly all mismarkings were a result of an error in using the mouse, not an error in the participants' judgment.
After completing the experiment the participants were asked a series of questions to garner subjective responses:
Did you prefer one type of display over the other?
The majority (9) of the participants, even some who did not perform well on the perspective display, preferred it over the plan-view for a variety of reasons. Most cited that they could identify many descenders quickly. Many stated it may have been easier to miss a descender but it was far less painful to rescan the display. The task on the plan-view was considered uninteresting because hooking was viewed as tedious while the task on the perspective display was considered engaging. After the experiment, several participants wanted to repeat the task on the perspective display. Most agreed their performance would increase greatly with some practice on the perspective view, but very little with the plan view. Two participants noted that the perspective view had advantages in an evolving situation because more aircraft motion could be detected and because more distinct, detailed, and differently oriented symbols made it easier to keep track of what aircraft were already examined.
Two participants gave a slight edge to the plan-view because of familiarity and the greater ease in executing an exhaustive search. One participant expressed a very strong preference for the plan-view. That participant complained that the perspective view did not offer enough cues, that little confidence was held for the existing cues, and the cues were lost in clutter. The participant claimed to trust and feel more comfortable with numbers than with images. One participant did not express a preference.
What problems did you have?
Nearly all participants complained about the position of the hook readout. Most of the difficulty stemmed from having the hook readout on a different screen (the 19" monitor) than the perspective view (which was on the HDTV). Participants said looking too far away from the perspective representation to read the hook readout hampered the search. Though none of the participants indicated that the placement of the hook readout hampered searches on the plan-view display, most would have preferred information to pop-up in the vicinity of the hooked track with either display type.
Two participants found that because the perspective display was so large that they could not see everything at once. This made it harder to search the display space exhaustively without losing track of what had already been examined.
Did you find the perspective cues for ascent and descent effective?
Except for one participant, all other participants found the perspective display cues to be very effective in all but a few aircraft. Most commented that the cues made rescanning the display less tedious, but some cues were diminished by distance and clustering.
Was hooking easier on one of the display types?
Half of the participants found ease of hooking to be independent of the display type. They stated that hooking with the plan-view display was more difficult in symbol clusters, but hooking with the perspective display was more difficult with small, distant symbols. The other half of the participants found hooking to be better on the perspective display because the symbols made for bigger targets for the mouse.
