Is the self-confrontation method applicable to Naturalistic Driving Studies?

Abstract

Since the year 2000, two methods are being increasingly used in driver behavior studies: on the one hand, Naturalistic Driving Studies (NDSs) that consist in gathering behavioral data during drivers’ usual trips over long periods of time; on the other hand, video confrontation methods, whose principle is to display sequences of usual driving activities to the driver to spell out subjective experience. Both practices aim to track inner user experience, but linking NDS and confrontation methods is a challenge. Indeed, displaying to a participant almost all traces of an activity (self-confrontation method) is not feasible for a large volume of data over extended periods of time collected in NDS. Therefore, setting up self-confrontation with NDS recordings raises the issue of identifying relevant video sequences before the interview. It also relies on the possibility for participants to relive sequences of a non-immediate past during the self-confrontation. The study of safety-critical situations and on-board secondary tasks in the first European large scale NDS (UDRIVE) was an opportunity to test video confrontation method within the framework of a NDS. The Knowledge Data Discovery (KDD) method was used for pre-selecting video sequences for 16 interviews. During each interview, three were presented to the participant. Efficiency was evaluated with qualitative indicators derived from explicitation interview. The priming revealed to be efficient for safety critical situations, but not for secondary tasks. This statement is in line with experimental work on memory processes which shows a correlation between memorization and emotional content. Consequently, the use of autoconfrontation methods in NDSs seems well suited for investigating specific driving situations, with limitations for more casual ones. At the same time, the knowledge gained from using self-confrontations will help refining indicators used in the data mining (KDD) process.