Probabilistic modelling in the analysis of traffic crash data: a study of bus drivers’ injuries in Turkey

The first aim of this study is to introduce the probability models in modelling the traffic crash data and present comprehensive interpretations of the statistical findings. The second aim is to reveal the risk factors that have important effects on the driver injury severity in bus accidents in Turkey. Data collected by The Department of Traffic Training and Research of General Directorate of Public Security and General Command of Gendarmerie were used. By reference to the ordinal nature of the dependent variable, Ordered Probit Model (OPM) and Ordered Logit Model (OLM) were constructed. The application results show that the average risk of injuries increases along with a 56-year-old or older driver. Other factors leading to the increases in the probability of injury severity are traveling on province road, rainy weather conditions and the accidents occurring as bumping to a stationary material. Accidents occurring as bumping from rear-end or one side of vehicle are is all associated with less severe injuries. Besides the remarkable interpretations of the model results, we also tried to find out which probabilistic model in what conditions fits well in the analysis of a crash data?