Carl Westin

74 Source bias effects In the conformance design phase, participants’ solutions, as recorded in the pre- quel simulation, were analyzed in a three step process to determine their conformal and nonconformal advisories. First, solutions were decoded against a solution pa- rameter framework consisting of five decision stages. This yielded a solution pat- tern , describing each solution in detail. Next, a participant’s solution patterns across scenario repetitions were compared for similarities. The consistent patterns across repetitions were used to define the participant’s conflict solution style . Finally, the conflict solution style was used to define the participant’s conformal advisory. To ensure reliability, three researchers accomplished this process in parallel. In the final experiment simulation the same participants played an identical sim- ulation, with the same measurement scenario and designed conflict, as encountered in the prequel simulation. Only this time, they were supported by a decision aid that provided resolution advisories. Participants were divided into two groups that encountered advisory source condition runs in different orders. A condition run con- sisted of ten training scenarios and first five test trials pertaining to one of the source conditions. After a short break the procedure was repeated with another training session and five test trials, but this time with the other source. The decision aid automatically displayed the heading band SSD for one of the aircraft in conflict and visualized the suggested solution by plotting it within the SSD. An advisory was accompanied by a beeping sound and the advisory agree- ment dialog window that enabled participants to either ‘accept’ or ‘reject’ the advi- sory. Participants were at this stage also required to indicate their agreement with the advisory. A countdown timer showed that participants had 30 seconds to make a decision. A resolution advisory consisting of a heading, speed, or combination thereof. Unknowingly to participants, resolution advisories were only provided to the designed conflict. Advisories were either conformal or nonconformal as pro- grammed in the previous conformance design phase. Although conformance varied, all solutions were safe and solved the conflict. Participants were instructed to this fact. However, they were instructed that the suggested advisory not always would be optimal and that they may want to consider an alternative solution. An intermediate level of automation scheme was chosen (management by con- sent), whereby the participant had to accept the advisory to execute the advisory. Participants could freely interact with other traffic for the duration of the resolution advisory. This allowed participants to inspect other aircraft for alternative solu- tions. After each scenario, participants were asked to rate the subjective difficulty. Debriefings and source bias questionnaires were administered after simulations. The appearance of resolution advisories were configured based on participants’ first interaction to solve the designed conflict in the prequel simulation. The first time of interaction was on average 53 seconds after scenario start (ranging between