Investigating the Effect of Marginal Areas around the Cities on Rural Road Accidents in Iran Using Linear and Logistic Regression Approaches

According to previous studies, 60% to 70% of the total rural road accidents would occur at the city entrance zones in Iran. Therefore, the characteristics of these zones could be considered as effective parameters in rural road accidents. In all prior studies, a 30-km buffer of the cities' entrances has been assumed as the border of the entrance zone. The 30-km buffer could not be considered as the boundary of the influenced area (BIA) of the cites´ entrance for all types of the roads and cities, merely based on aggregate rural road accidents´ data and a traditional definition of the city entrance zone. Determining the BIA for various rural roads with different characteristics using the modelling approach is the innovative aspect of this research. Furthermore, according to their specifications, implementing safety improvements in these areas, not only reduce the number of rural road accidents and fatalities, but also prevent the loss of road safety costs due to the economic problems of Iran. Thus, this study aimed to develop linear and logistic regression models to predict the BIA of rural roads in Iran. The results of this study indicated a fit index value of 0.737 for the linear regression model, and 0.379 and 0.346 for the ordered probit (OP) and multinomial logit (ML) models, respectively. The analysis of significant variables at 95% confidence level, revealed that the access points' density, and the length of adjacent land uses are the most significant variables affecting the BIA.