Performance was rated by two independent “backseat evaluators” at 41 locations along the route on a previously developed 3-point scale,
20 with respect to eight different behaviors/skills; 28 locations were in suburban streets, 11 on a multilane highway, and 2 in city streets. One backseat evaluator, designated as the primary evaluator, sat in the middle of the backseat (positioned so that the evaluator did not obscure the driver's view of the rearview mirror), with the second evaluator sitting behind the driver. At each location, the eight driving behaviors evaluated included the following:
scanning and attention to other road users, signs, and markings;
lane position of the vehicle;
steering steadiness, involving smoothness of steering at any point of the drive; appropriate use of
speed relevant to road conditions and the speed limit;
gap judgment between the driver and other cars when entering traffic flow or intersections or passing moving or parked cars and following distance; appropriate use of
braking to allow smooth driving and stopping as required;
directional indicator use to signal to other road users intention to change direction; and
obeying signs and signals. If a given maneuver was not relevant at a given location, it was not rated (e.g., using the indicator signal would not be relevant if there was no turn or lane change involved at a given location).
Pedestrian detection was determined by identifying the number of times that participants correctly reported the presence of a pedestrian, road worker, or cyclist encountered when they were considered relevant to the driving task and this was scored after testing from the video recordings. In real-world driving assessments such as these, it is impossible to control pedestrian events; for the 46 drivers, there were a total of 77 pedestrians (45 for the bioptic drivers and 32 for the control drivers).
Road sign recognition was also recorded for each of the 58 traffic signs encountered along the driving route.
Traffic light detection rate was determined by calculating the percentage of the 25 traffic lights where the signal light was correctly reported.