To make safe road environments it is important to understand the relationship between safety and the properties of the road environment, and traffic characteristics and traffic control measures. This paper develops disaggregate models to represent 'slight injury' and 'fatal and severe injury' accident frequencies for signalized intersection approaches in Singapore. The Poisson probit exogenous underreporting model is best suited to represent slight injury accidents. This model takes into account both heterogeneity and underreporting. In the case of fatal and severe injury accidents, the random effect negative binomial model indicates the maximum goodness of fit to thus become the most parsimonious model. Statistics suggest that the overdispersion is arising in fatal and severe injury accident data mainly due to heterogeneity and not from excess zero observations. The models further indicate the significance of several intersection related variables and factors associated with the reporting mechanism.
Samenvatting