Carl Westin

48 First empirical insights Conformance analysis and programming Prequel simulation Experiment simulation Complexity High Complexity High Complexity Low Complexity Low LOA high LOA high LOA low LOA low Conformance high Conformance low Manual performance Decision aid support F IGURE 3-3: Experimental design. 3-3-6 Independent variables Two LOA (High vs Low) were represented by • Management by Consent (MbC): in which automation presented a resolution, and this would implement only if authorized by the participant within a 15 s interval; or • Management by Exception (MbE): in which automation presented a resolu- tion, and this would automatically implement after the same 15 s interval, unless the participant specifically vetoed it. Various LOA taxonomies have been put forth over the years, and they tend to characterize automation on a continuum from fully manual to fully auto- mated. 43–45, 51 The MbC and MbE levels chosen here seem to correspond best to the midpoint of such taxonomies. These selected levels also capture the realistic near term evolution of ATC automation, in which authority alternates between hu- man and machine. Two levels of traffic complexity ( CX high and CX low ) were defined. Various studies suggest that traffic complexity drives workload 5, 136, 137 and that, in turn, the benefits of automation are increasingly realized under high workload. 45, 51, 127, 129 Complexity was varied through means of aircraft count and calibrated in a series of developmental trials. On the basis of participant feedback and expert opinion, two realistically extreme levels of complexity were established. Scenarios 1 and 4 were high complexity with 10-14 and 8-11 aircraft, respectively, always present in the sector. Scenarios 2 and 3 were low complexity with 6-8 and 3-7 aircraft, respectively, always present in the sector.