Investigating the Causes For Single Vehicle Accidents Using Computerized Accident Patterns and Logit Modelling

Investigating the Causes For Single Vehicle Accidents Using Computerized Accident Patterns and Logit Modelling


Y Shiftan, S Hakkert, Technion, IL



This paper aims at developing various tools to better analyze the causes of single vehicle accidents on interurban highways. One of the most important items in accident reconstruction and analysis is the sequence of events and manouevres prior to the collision. Whereas some of this information is available in the raw police file, it is generally lost in the process of coding the statistical computerised accident records which form the basis for most remedial work. It is therefore a widespread habit to return to the police files for detailed analysis. This is a time consuming and sometime imposible task, seriously limiting the quality of accident research. This papers use a computerized accident report which contains the vocabulary used by police investigators, and varies according to the accident pattern under investigation.

The paper focuses on single vehicle accidents with injuries on non-urban roads. Police accident records data for a period of four years were collected and analyzed for two interurban roads in Israel. One of the roads was a dual carriageway and the other a high quality single carriageway road. The data were analyzed statistically to characterize the main factors that affect the severity of such accidents. In the next phase a logit model was estimated to better understand the causes for different levels of accident severity. The model estimates the probability that a single vehicle accident will be with light, serious or fatal injuries. In addition to help understand the factors that influence accident severity, the model can also be used to forecast accident severity for new or improved road segments. Such estimates are needed to account for accident costs in cost-benefit analysis.

The main conclusions from the paper are that single vehice accidents have different characteristics depending on the type of road on which they occur. The presence and type of obstacles, lighting, margins and side clearance are all important factors affecting the severity of single vehicle accidents. To improve safety and minimize accident severity, roadside obstacles should be eliminated or moved avay from the pavement edge. Unmovable objects, should be fitted with energy absorbing furniture, so as to decrease vehicle impact on collisions. Safety barriers used on the roads in this study were not properly functioning as safety enhancing features as a third of all fatal single vehicle accidents( albeit a small number) on one of the two roads involved a collision with safety barriers.

The paper shows that the combination of various means including: the utilization of the computerized accident pattern recording, statistical analysis, and discrete choice modeling, can help to better analyze and understand accident causes and lead to the development of tools to improve safety and estimate accident costs.


Association for European Transport