Carl Westin

A-1 Introduction 189 T ABLE A-4: Problem-solving style for controller B Aircraft ID Resolution type Direction Value Rationale QS1338 Combined Right in- crease 15 ◦ & 60 kts All solutions consisted of dual aircraft so- lutions. Interacted with QS first in three out of four solutions. In all three cases QS was turned right, once with a speed increase. Often used speed and heading combinations. Note that a controller’s problem-solving style for a specific conflict represents that controller’s consistent solution for that conflict provided that the same solution was implemented in at least three out of four scenario repetitions. The consistency and agreement study was based on the analysis of controller’s problem-solving styles (consistency measure) and how these differed between controllers (agree- ment measure). Note that the solution parameter framework was used as one of three solution classifications (the solution parameter hierarchy) in the consistency and agreement study (Chapter 6). The other two classifications were the control problem and solution geometry classifications. A-1-4 Define conformal and nonconformal solutions The strategic conformance of a conflict resolution advisory can be considered on a spectrum ranging from conformal to nonconformal. A controller’s perfect con- formal resolution advisory would exactly match all solution parameters defining that controller’s problem-solving style. Any deviation from the reference problem- solving style, at any step in the solution parameter framework, would make the resolution advisory less conformal and increasingly nonconformal. The definition of a controller’s conformal resolution advisory was based on that controller’s problem-solving style. The definition of nonconformal resolution advi- sories relied on identifying and comparing similarities in and differences between controllers’ problem-solving styles. Reliability was achieved by having three re- searchers agreeing on each controller’s conformal and nonconformal resolution ad- visory for each designed conflict. For the nonconformal resolution advisories we wanted solutions that contrasted a controller’s own solution. The initial idea was to simply look at each controller’s own solution and manually script a different resolution advisory. The issue with this approach, however, was that, since we are not controllers ourselves, we might script an unrealistic resolution advisory that could confound our experiments. Fur- thermore, we did not have time or resources to develop an algorithm that satisfied