Modeling the severity of road accidents at intersections

The main objective of this paper is to realize a micro-economic study of road accidents in Tunisia. Indeed, we opted a multinomial logit model to analyze the severity levels of road accidents at intersections. This severity, in the light of the road traffic safety literature, essentially depends mainly on three factors (human, vehicle, environment) relating to the road traffic system, as well as, road safety. These factors interact with space and time to provide an explanation of the occurrence and severity of an accident. This involves identifying and measuring several qualitative and quantitative variables to describe the different components. Therefore, this work aims to highlight the way and extent through which these explanatory variables affect the severity of a traffic accident. It's a question of estimating the probability that a driver is the victim of an accident of certain gravity, through the analysis of both his personal characteristic as well as those of the traffic environment. The results obtained confirm the effects of the factors mentioned on the severity of an accident and show that all the variables selected are theoretically and statistically significant and explanatory at different degrees.