With the notable trend of motor vehicle crashes, a goal for any licensing agency is the ability to identify high-risk drivers. The higher the efficiency of identifying these drivers, the better the results achieved by their driver control programs aimed at preventing road crashes will be. To this end, a significant number of studies were carried out to establish detailed estimations of a driver's future crash potential on the basis of both individual factors such as prior driving record histories and aggregated factors such as driver age and gender. However, the degrees of precision of identifying risky drivers from the driver population using the available models are still not high. Although this may be attributed to the statistical nature of driver crash involvements which makes it difficult to accurately predict, some methodological deficiencies in these studies were also identified to have negative effects on the research outcome.; Kentucky data shows that a significant number of drivers are repeatedly involved in crashes, and a previous Kentucky study revealed that a significant relationship between crashes and convictions exists. Thus, the objectives of this study are the development of crash prediction models that can be used to estimate the likelihoods of a driver first being involved in a crash and then being responsible for a near future crash occurrence. Multiple logistic regression techniques were employed using the available data for Kentucky licensed drivers. This study considers as crash predictors the driver's total number of previous crashes, citations accumulated, the time gap between crashes, and demographic factors. The driver's total number of previous crashes was further disaggregated into the driver's total number of previous at-fault and not-at-fault crashes. In addition, optimal study data periods were selected to extract data for the predictor variables. Moreover, crash type and driver inexperience were taken into account as necessary. Sensitivity analysis was used to select an optimal cut-point for each model. Study results show a significant improvement in identifying risky drivers, especially from the inexperienced Kentucky driver population and in determining the culpability given the fact that a crash has occurred. Thus, the models determined here enable licensing agencies to develop possible remedies and alert risky drivers of their risk potential to society.
展开▼
