Analysing Motorcycle Injuries on Arterial Roads in Bali using a Multinomial Logit Model

抄録

This paper aims to investigate the influence of accident related factors on motorcycle injuries on two arterial roads in Bali. Multinomial logit (MNL) models are estimated considering three severity classes such as slight injury, serious injury and fatal injury as response variables using local police data as explanatory variables. The analysis shows that there are four variables associated with motorcycle injuries. Sideswipe accidents involving motorcyclists were 51.7% less likely resulting in serious injuries than slight injuries. In addition, motorcycles collided with other vehicle(s), either motorist/motorcyclist failed to yield and motorcycle at fault were 89.1%, 60.7% and 44% respectively less likely resulting in fatal injuries than slight injuries. Probability analysis shows that a change in 1% of these variables could influence motorcycle injuries between 33% and 34%.