Carl Westin

54 First empirical insights A somewhat puzzling result, however, was that controllers’ response was slower when rejecting nonconformal advisories than when rejecting conformal advisories. We expected a quicker response time to rejected nonconformal advisories. That is, the “further away” an advisory is from the preferred solution (i.e., larger mismatch), the faster it will be rejected as the controller quickly can determine its nonconfor- mance. A likely explanation it that some of the conformal advisories rejected, were in fact not conformal. Rather, these advisories were a poorer match to controllers’ problem-solving style than nonconformal advisories were in average. The reason for this can be traced back to how conformal advisories were defined. Conformal solutions were based on the solution registered for the same scenario in the prequel simulation. Possible learning effects, or conflict solving inconsistencies, may have rendered a solution nonconformal and stored in memory as a poor solution. 3-4-4 Scenario difficulty Our main purpose in collecting difficulty ratings was to validate (via a proxy mea- sure) our complexity manipulation. To this end, difficulty ratings were obtained af- ter each session, on a scale of 0-100. Notice that these ratings referred to the entire scenario, not just the transient advisory. Figure 3-9 shows that high complexity sce- narios were rated higher than low complexity scenarios. Rated difficulty increased significantly with complexity ( F [l,15] = 179.95, p = .000). Neither the main effect of conformance, nor the interaction between the conformance and complexity, was significant. 3-4-5 Debrief interview feedback Two main themes emerged from post-session debriefs, one encouraging and one cautionary. First, several participants noted that the prototype SSD tool made pos- sible a new way of working (a sometimes-desirable result of new automation), by facilitating the use of speed adjustments (which controllers do not tend to currently 0.9 0.6 0.3 0 -0.3 -0.6 -0.9 Difficulty rating (z-score) CX high CX low F IGURE 3-9: Standardized difficulty rating, by complexity and conformance.