Levels of Safety on Interurban Roads

Abstract

The levels of safety of the main interurban road network in Israel are studied, and 70 percent of the total interurban roads are covered. A macro approach to the relation between road safety and various road categories as reflected in the overall geometric and traffic operational features is presented and is limited to road links. The classification and regression tree (CART) analysis, a nonparametric method that generates a binary tree structure from the data showing the criterion (variables) for each split and giving a pictorial representation of the data, is used as a preliminary tool to illuminate the relation between the variables and road accidents as a measure of safety and to identify meaningful candidate variables for deeper analysis. CART results portray the importance of average daily traffic (ADT) and the need for stratification of ADT in accident modeling. It also indicates probable power function models and highlights some interactive variables and the effect of shoulder type among single carriageways. However, no interactive variables were revealed because of the total dominance and masking effect of ADT. Multivariate multiplicative Poisson regression models with log links were fitted to the data set using the generalized linear interactive modeling package. The results again emphasize the safer nature of freeways as compared with conventional single- and dual-carriageway roads and show that the relation between ADT and safety is curvilinear. For the fitted parametric models, the effects of ADT were found to be most important, as in the case of the nonparametric CART analysis. In the case of models for single carriageway roads, other explanatory variables (free flow speed, type of shoulders, and junction frequency) showed some significance.