Carl Westin

104 Automation transparency effects o j HB cfY HB cfN TRI cfY TRI cfN 0 Response time [z-scored] 1.80 1.20 .60 -.60 -1.20 -1.80 9 (a) Response time o j HB cfY HB cfN TRI cfY TRI cfN 0 Difficulty rating [z-scored] 1.80 1.20 .60 -.60 -1.20 -1.80 (b) Difficulty F IGURE 5-6: Boxplots of response time (a) and difficulty ratings (b) ( N = 9). difficulty ratings obtained after each scenario. Although scenarios with nonconfor- mal advisories were rated slightly more difficult, the Friedman test did not detect any significant differences between conditions. The following interface usage results only apply to transparency effects. Unlike conformance effects (measured in relation to the resolution advisory), transparency effects reflects data collected from the entire scenario run. Figure 5-7 provides boxplots of (a) number of SSD inspections, and (b) type and number of conflict solution commands. The former indicated a trend with less inspections when using the TRI SSD (Wilcoxon: z = -1.72, p = .086). Total number of interactions (the sum of all heading, speed, and combined interactions) did not vary significantly. There was a trend for more use of speed commands with the TRI SSD (Wilcoxon: z = HB HB TRI TRI Average number 0 0 15 30 45 60 6 8 Average number 4 8 12 16 20 24 All Heading Speed Combined (a) Number of SSD inspections (b) Type and number of conflict solution commands F IGURE 5-7: Boxplots of SSD inspections (a) and commands (b) across transparency conditions ( N = 9).