Carl Westin

3-5 Discussion 55 use in en-route airspace). Several participants, however, also noted the tendency to feel “driven” by the SSD interface, become reactive, and to curtail their con- flict assessment under high workload situations. It is possible that this influenced participants’ response time to advisories in scenarios that were perceived as more difficult. In terms of experimental design, one bit of post-session feedback was grat- ifying: controllers reported that they had not recognized the conformal resolution advisories as replays of their own previous solutions. 3-5 Discussion A main effect of conformance was observed on acceptance, agreement, and re- sponse time. Conformal advisories were accepted more often, rated higher, and responded to faster than were nonconformal advisories. Complexity, on the other hand, showed a main effect on acceptance, agreement, and difficulty: all three in- creased with complexity. Acceptance was higher for conformal advisories, and this effect was slightly more pronounced under high complexity conditions. One plausi- ble interpretation of the agreement effect is that controllers, under the time pressure of complex conditions, did not fully evaluate a resolution advisory. In terms of response time, controllers tended to act fairly quickly (on the order of 5-6 seconds), and seldom timed out at fifteen seconds. Response time was lower (i.e., controllers responded faster) to conformal advisories, and under complex con- ditions. To summarize our findings in terms of our original research questions, we can state the following: • Conformance and acceptance controllers were indeed more likely to accept conformal advisories (replays of their previous performance) than nonconfor- mal advisories (replays of a colleague’s different solution); • Complexity and acceptance everything else being equal, controllers were more accepting of advisories under complex conditions. This finding is per- haps an indication of premature closure, a tendency toward incomplete eval- uation under time pressure; • Conformance and other effects again, response time was indeed lower for conformal advisories, and this effect was more pronounced under complex conditions (perhaps another indication of time pressure). 3-5-1 Levels of automation With no differences between LOA conditions detected, we decided to collapse the two before running statistical analysis. Although the theoretical distinction between