Modeling Multiple –Vehicle Property Damage Collisions inUrban Signalized Intersections

Development of disaggregate models for estimating different property damage collision type frequencies in urbanintersections has rarely been studied, particularly in Iran. It seems that very little research work has been implementedfor studying the factors affecting the collision type frequency at intersections. The main objective of this paper is todevelop suitable statistical models to predict types of property damage accident frequencies at signalized intersectionsapproaches in the City of Mashhad, Iran, based on geometric, traffic and regulatory control characteristics. Three negativebinomial models are estimated for collisions occurred in four-leg signalized intersections in the city of Mashhadand their results were compared. These models are total, rear-end and right angel collision models. The goodness offit was assessed by statistic tests. The Incidence Rate Ratio is used to assess the effectiveness of independent variableson frequency of property damage collision. Validation of models was controlled using paired samples T-test Method.Modeling collision types showed a strong relationship between frequency of property damage collision types andindependent variables such as road geometry, the type of control system and traffic characteristics. Results revealedthat seven of independent variables considerably affect the safety of signalized intersections.

Modeling Multiple –Vehicle Property Damage Collisions inUrban Signalized Intersections Keywords:

Property damage collision , signalized intersection , negative binomial model