Carl Westin

iv Summary controllers’ solutions depended on how the conflict was represented in the interface. Since conformal (and nonconformal) advisories were based on solutions when us- ing the heading band SSD, these advisories may not have been representative as conformal when using the triangle SSD. The fourth and final Consistency study analyzed controllers’ manual conflict solution data (no explicit resolution advisories) collected in the above three empiri- cal studies. The objective was to determine whether the problem-solving mismatch between controllers and CD&R automation could be explained by controllers solv- ing conflicts inconsistently. The study investigated the degree to which controllers consistently had solved the repeated conflict (four repetitions) over time (intra-rater variability), and to what extent they agreed on solutions (inter-rater variability). Based on a review of ATC conflict resolution strategies, a solution classification framework was developed against which controllers’ solutions objectively could be qualitatively coded and analyzed. Results revealed that controllers were consis- tent, but disagreed on how to solve conflicts. However, consistency was limited to higher-level decision stages, such as whether to vector an aircraft in front or behind an other, or interacting with both or only one aircraft. Controllers were inconsistent in relation to more detailed solution parameters, such as the direction of a solution (e.g., vector left or right) and the exact directional deviation value (e.g., right vector of 035 degrees). Consistency and agreement was not higher for biased conflicts that favored a certain type of solution. A difference, however, was noted in regards to overall solution strategy. With biased conflicts, the majority of controllers agreed on a shared solution geometry, while with the unbiased conflict, the majority solved the conflict according to the control problem classification. Experienced controllers were slightly more consistent than trainees in terms of the control problem classifi- cation. Taken together, this thesis has contributed to the knowledge of what drives con- trollers’ acceptance of resolution advisories in particular, and human-automation collaboration and automation acceptance in general. Empirical results showed that conformal ATC automation, solving conflicts like the controller, can benefit accep- tance and agreement of that system’s advisories, as well as reducing response time. These benefits were observed across varying expertise levels, particularly in rela- tion to expert operators. Strategic conformance may be most beneficial during the introduction of new automated decision aids, as a means for gaining acceptance. The development of conformal automation, and other personalized decision support, requires that the operator is somewhat consistent in her/his problem solv- ing. However, designing for conformal automation, or other personalized systems, requires an ethical consideration since such systems have the power to influence acceptance and trust independent from the system’s actual performance and relia-