Carl Westin

192 Conformance design the scenario (through interaction with noise aircraft) such that the scripted resolution advisory has become invalid by the time it appears. • A resolution advisory must be valid long enough for controllers to identify and investigate the conflict. • The timing of a resolution advisory must be kept constant within each sce- nario group. If not, individual timings for each controller and scenario could be a confound in the experiment. • Resolution advisories must provide a valid conflict-free solution. This means that resolution advisories cannot suggest resolution commands in a yellow or red no-go area in the SSD. Even though such resolutions could solve the designed conflict, they would cause a secondary conflict with a noise aircraft. This would be potentially detrimental for controller trust in the automation, and could provoke a “reject all” strategy. • Resolution advisories must appear when both aircraft are in the sector. This is important, as controllers should be able to explore alternative solutions by inspecting the SSD of both involved aircraft. In addition, they should be able to solve the conflict by interacting with any of the involved aircraft. One risk is that the timing for advisories can be perceived as disturbing or oth- erwise annoying by controllers and negatively affect the acceptance and agreement with advisories. This problem is generally known as poor automation “etiquette.” Here, etiquette refers to the degree to which automation adheres to human social behavioral “rules” for communication and interaction. Moreover, it should be noted that the use of a generalized resolution advisory by default is nonconformal and may constitute a confound. Therefore, analyzes were carried out as to determine to what extent controllers’ solutions for the same designed conflict varied depending on when it was solved (see Chapter 6). Note that several controllers’ problem-solving styles consisted of interacting with both aircraft to solve the conflict. The simulator, however, did not allow the presentation of a dual interaction as a resolution advisory. As such, only the first in- teraction of a controller’s problem-solving style was considered when determining a conformal solution. Finally, in the First empirical study (Chapter 3), controllers’ general problem-solving styles were not defined. In this study, the conformal and nonconformal resolution advisories were based on solutions specific to each sce- nario replicate. For the subsequent empirical studies (Chapters 4 and 5), however, the procedure with problem-solving style was implemented in order to ascertain that a conformal resolution advisory represented in fact represented a consistent problem-solving preference, and not a randomly selected solution.