The present study assessed the validity of a laboratory simulation of the car-passing task by means of canonical correlation techniques. Twelve experienced drivers observed filmed situations in the laboratory and actual situations on the road. A canonical correlation of .96 was obtained between the laboratory performance measures and the on-the-road criterion measures. It was found that 41% of the variability accounted for in the laboratory was redundant with 60% of that accounted for on the road. Application to driver education is emphasized.
