The Model of Evaluation of Crash Severity at Roundabouts (A Case Study: Tehran-Iran)

In this article the severity of 661 crashes on roundabout in Tehran - Iran for two consecutive years (2012, 2013) were investigated for finding the effective factors on severity of crashes, their injuries and the relation of these factors with existing scientific knowledge using Logistic regression and Probit regression models. The kind of road users, light condition at night, condition of road surface, Average daily traffic, and the daily average of number of motorcycles, the daily average of number of pedestrians, mode and angel of collision were identified as the most significant effective factors. In this article AIC standard (Akaike Information Criterion) and McFadden R square statistic tests were used for goodness-of-Fit of models. With using the model and roundabout which their data were available, the validity of model were evaluated. At the end two built models have been compared and factors affecting the severity of crashes at roundabout have been introduced.