Carl Westin

162 Discussion and recommendations needed to further investigate these three classifications and to what extent they ac- curately capture controller’s conflict solving preferences. In addition, longitudinal studies should explore the long-term effects of conformal automation on human- automation interaction. 7-3-2 Defining and displaying consistency A consistent problem-solving style was defined by the same solution being applied in three out of four, or four out of four, repetitions. The few solutions collected per controller limited the consistency study. Measures may have been insensitive to con- sistency patterns for more detailed solution parameters. A challenge when studying individual’s behavior is how to measure consistency. Statistical approaches have, traditionally, focused on the two-distributional moments of central tendency (e.g., mean) and dispersion (e.g., standard deviation) to describe the “average” behavior of a group. 237, 269 With the emphasis shifting from the group to the individual, the applicability of these methods is questioned. When we turn to individuals, however, the same principle can be applied. But instead of gathering a large group of people (although still necessary), a large number of recordings from the same individual is needed. Statistically, a much larger data set, say 20 repetitions of a conflict, would have been desirable as a foundation for determining consistency. Therefore, future research should consider larger sets of repeated stimuli. Consistency could not be established down to the detailed decision stage that specifies a directional value (such as a vector of 035 degrees). The simulation was limited in that the direct interaction with aircraft (i.e., by means of datalink) may have constrained consistency. The SSD interface facilitated precise speed and vector changes (by clicking and dragging the vector line) such as increasing the speed by 7 knots or turning 28 degrees right. In contrast to reality where traditional R/T is used, controllers generally provide instructions in average numbers, such as “QS1338, turn right heading 035.” However, results showed no increase in consistency when considering controller’s solutions in groups of 10 degrees or 10 knots. With consistency limited to high-level decision stages, results provided a lim- ited understanding of controllers’ conflict solving. A convenient inference would be that detailed solution parameters are of less importance for solving conflicts, such as the direction of a vector and deviation required to attain a desired separation dis- tance (i.e., decision stages 3 and 4 in Figure 6-1 in Chapter 6). However, previous research argues against this inference. For example, controllers have been found to aim for different separation distances depending on the current experienced work- load. 79, 80 Further research is required in order to better understand the relationship between detailed decision stages, for which controllers were found inconsistent, and the higher-level decision stages, for which controllers were found consistent.