Carl Westin

4-4 Results 77 X Number of participants 1 2 3 4 5 6 7 Strongly disagree Neutral Strongly agree 0 4 3 2 1 (a) I found the resolution advi- sory interrupting X 1 2 3 4 5 6 7 Strongly disagree Neutral Strongly agree 0 4 3 2 1 (b) I accepted resolution advi- sories even though I did not agree with them. X 1 2 3 4 5 6 7 Strongly disagree Neutral Strongly agree 0 4 3 2 1 (c) I accepted resolution advi- sories without inspecting the conflict. F IGURE 4-5: Bar chart showing simulator questionnaire responses to three state- ments ( N = 5) in an underhanded manner, that they were wary of the advisory system, or that they were suspicious of the advisory systems intent, action, or outputs. Disagreement between participants was recorded for only one statement. Three participants re- sponded that they were familiar with the advisory system, while two disagreed. Responses for this statement did not vary with advisory source. Source bias VAS. SBVAS responses are reproduced in Figure 4-6. Because of the small sample and relative small spread in responses, a boxplot chart was im- practical. Instead, responses are provided in a high-low representation, showing the maximum, minimum, and average response. Results indicate that participants per- ceived the two advisory sources differently, with a slight preference for the human (i.e., air traffic controller) adviser as opposed to the automation adviser. Most responses fell within 10% of neutral, in one direction or the other, with two participants providing neutral answers to all statements. Responses from the other three participants indicates that they considered human-based solutions safer and slightly more reliable and efficient. The largest average deviation was to the question most associated with the fundamental difference between human and au- tomation: “Which source provided solutions more similar to how you would have solved the conflict?” In this case, participants showed a relatively large deviation toward the human source. Furthermore, participants indicated a preference for working with the human source in the future. In contrast, the automated source was perceived more risky and difficulty to work with. However, trustworthiness, helpfulness in solving conflicts, workload, and quality of solutions did not vary notably with advisory source.