A brief pc-based hazard prediction training program improves young novice drivers’ hazard perception skills compared to a control group over time

Van der Kint, S.T.; Van Schagen, I.; Vlakveld, W.; Mons, C; De Zwart, R.; Hoekstra, T.

Young novice drivers are overrepresented in crashes. One of the causes is their poor hazard perception skills. Can a brief PC-based hazard prediction training speed up the acquisition of these skills? To investigate this a training was developed and tested. The training consisted of eight short videos from the driver’s perspective that ended in a crash or an emergency brake incident. Participants watched each video twice but only up until the moment the dangerous scenario started to unfold. During the following minute, they made as many predictions as possible about what would happen next. After this, the video was played again, including the remaining part with the crash or emergency brake incident. Thereafter, with the aid of stills from the video with highlighted areas and arrows, a voice-over explained which areas an experienced driver would have attended to, to avert the crash or emergency brake incident and which other latent hazards experienced drivers would have attended to that did not materialise. To evaluate the training, all participants first drove an approximately 15-minute simulator drive while an eye tracker recorded their gaze directions. In this drive seven latent hazards were staged that did not materialise. Then participants were randomly assigned to either an experimental group that completed the hazard prediction training (N = 19) or a control group (N = 20) that completed a placebo training with the same look and feel of the hazard prediction training but that did not train hazard prediction. Directly after, both groups drove the same simulator drive again but starting from a different location than the pre-training drive. Controlled for their gazes at latent hazards during the pre-training drive, the experimental group gazed at significantly more latent hazards during the post-training drive than the control group. Approximately five months after the intervention both groups completed an online hazard perception test consisting of animated videos. In these videos latent hazards were staged that were conceptually the same as those staged in the simulator drive. This online test was also completed by a group of experienced drivers (N = 19). The experimental group scored significantly better on this test than the control group. Experienced drivers scored better than the control group and worse than the experimental group. However, neither difference was significant. The results indicate that a short PC-based hazard perception training can indeed enhance hazard prediction skills and there are indications that these skills have retained.

Verschenen in
Transportation Research Part F
102 (2024)


Dit is een publicatie van SWOV, of waar SWOV een bijdrage aan heeft geleverd.