Observed and reported driver behaviour at junctions : implications for driver training

Over half of the accidents on British roads occur at junctions and it was the primary goal of this research to develop an increased understanding of the underlying factors behind these accidents. The vast majority of all road accidents are attributable to human error and the research investigated junction negotiation with respect to drivers' perceptions of the social and environmental components of driving. The first part of the research, an observation study, gathered basic information about actual driver behaviour at junctions. The progress of over 3600 vehicles at four junctions of differing styles was recorded and analysed with the aid of a timebase video facility. It was found that approximately 7% of all drivers were involved in some form of near-miss for which evasive action was necessary. In addition to basic descriptive information, inferential statistical techniques were used to identify factors contributing to near-miss incidents in addition to signalling, tracking and approach speed behaviours. The information derived from this first study was used, in conjunction with that obtained from group discussions, to develop a questionnaire. Using a postal distribution technique, the questionnaire was distributed to a random sample of British full driving licence obtained from the records of the Driver and Vehicle Licensing Authority. An additional sample was obtained from the Thames Valley Police accident records at Milton Keynes to ensure that a suitably-sized accident-involved sample was available for analysis. The various sections of the questionnaire were designed to reflect different aspects of driving at junctions. In addition, respondents were asked to provide details of the most recent accident, if any, which they had been involved in. Just over half of the 740 respondents to the questionnaire reported such accidents, and the information provided was used to establish factors implicated in accident-involvement, and particularly accident culpability, at junctions. In addition to sex and exposure factors, it was found that self-descriptive metavariables were the most effective at predicting aspects of involvement in accidents at junctions. In particular, those deemed to be accident-liable were more likely to describe themselves as self-centred and ill-mannered. Other metavariables, particularly those recording the subjective riskiness of various manoeuvres, were also found to be useful discriminators between various sub-groups of accident- involved drivers. Finally, the differences in responses made by drivers who had been trained by a variety of methods, or combination of methods, were investigated. It was discovered that those drivers initially trained by a qualified instructor were more likely to respond in similar ways to accident-involved drivers. In contrast, those who had taken some form of advanced tuition were more likely to report more considerate, attentive traits. Several suggestions for further research were made, particularly recommending the adoption of a longitudinal research design to enable causal relationships between accident-involvement and responses to questionnaire items to be determined.