Real World Data on Driver Behaviour in Accidents and Incidents: Evaluating data collection and analysis methods for car safety development

Licentiate thesis, 2012

Real world data is important for safety development within the road transportation system. For car safety development in particular, methods to collect and analyse real world data on driver behaviour from normal driving, incidents and accidents are needed to address safety in driving. This thesis investigates what different analysis methods applied to self-report and observation data can provide about driver safety issues (e.g., drowsiness, distraction) in accidents and incidents. Nonresponse analysis and adjustment in an accident mail survey was performed by using insurance data from 8519 survey recipients and mail survey data for the respondents in Paper I. Document case studies were performed for 158 accidents in Paper II by combining accident mail survey questionnaires and insurance documents. In Paper III, an incident causation analysis was performed based on video-recordings of 90 car-to-pedestrian incidents in a naturalistic driving study. The findings imply that self-reported and observation data collection procedures are both required as complementary sources of information for car safety development. Mail surveys can be used as a cost efficient method to collect general information from a large number of accidents as well as information on some driver safety issues. Valuable, additional information about accidents can be obtained by analysing written descriptions from mail survey and insurance documents. This can provide insights into how the driver experienced the accident, facilitate the interpretation of mail survey responses, and provide information that is not captured by the mail survey variables. Video-recordings from naturalistic driving studies can provide detailed information on many driver safety issues. This is especially valuable for aspects of driver behaviour that is difficult to capture with self-report methods. There is ample opportunity to improve the understanding of driver safety issues in accidents and incidents. By combining data from self-reported and recorded events, future studies can improve estimates of the occurrence of different driver safety issues and provide a wider picture of accident and incident causation. A combination of different types of data sources can also be used to further address the validity of accident mail surveys.