Carl Westin

106 Automation transparency effects Boxplots of VAS questionnaire responses are shown in Figure 5-9. Values above 50 mm indicate a preference for the TRI SSD while values below indicate a pref- erence for the HB SSD. Overall, responses support the hypothesis that the TRI was perceived as a more transparent interface than the HB. More specifically, the TRI was perceived as more “helpful for conflict detection and resolution”, providing a better “overview of the solution options available”, and facilitating a better under- standing for “why the system suggested a certain solution”. In contrast to these positive perceptions, participants also found the TRI “more cluttered”, and slightly more “difficult to work with”, than the HB SSD. These responses were likely driven by the additional amount of meta-information provided in the TRI, which in turn can be considered a characteristic of increasing interface transparency. 5-7 Discussion The present study investigated the effect of interface transparency and strategic con- formal decision support (i.e., personalized) on controller trainees’ performance in CD&R, in addition to their acceptance of resolution advisories. Furthermore, two complementary questionnaires were used to examine differences in participants’ transparency perceptions of two CD&R tools which varied in the degree of trans- parency. Of particular interest were the potential interaction effect between trans- parency and conformance on the acceptance of resolution advisories, and whether an increased degree of transparency would reduce the importance of conformance. Neither interface transparency nor conformance significantly affected accep- tance or performance measured in simulation data. Still, there were some interesting trends and differences between conditions. Some of these findings were supported by questionnaire results, which indicated differences in participants’ perceptions of the two interfaces. From this study, two findings are of particular interest. First, acceptance results suggest that the benefits of conformal advisories do not lessen with increased interface transparency. Therefore, the hypothesis is rejected. However, participants’ preference for advisories conformal to their own solution style corresponds with findings attained previously. 101 This result, although not sig- nificant, provides additional support for the positive effect of individually tailored recommendations on the acceptance of decision aid outcomes. With the small sam- ple size, however, these results must be considered with caution. Furthermore, this study included trainees, in contrast to the larger sample of experienced controllers used in Hilburn et al. 101 Controller trainees may be less consistent in their conflict solving style than experienced controllers, which potentially can explain the lack of a stronger conformance effect.