rural crash severity modelling at marginal areas around cities in iran using ordinal logistic regression and partial proportional odds modelling approaches

Proof from earlier investigations indicates that many rural crashes happen in marginal areas around cities. therefore, exclusive crash severity models should be developed to pinpoint the factors linked to the increased likelihood of injury and fatality in these segments of rural roads. for this purpose, a partial proportional odds (ppo) model alongside the traditional ones including ordered logit (ol) and multinomial logit (mnl) models was utilized in this study to develop crash severity models for these segments of roads. the authors applied rural crash data gathered from highways that lead to isfahan for modelling. the ppo model outperforms the traditional models, as demonstrated by comparing developed models. also, the results indicate that rural crashes are more likely to be severe when the average speed exceeds 95 km/h, in multi-vehicle type crashes, in overturn-type crashes, when the at-fault vehicle is a truck/trailer, and when the at-fault or not-at-fault vehicle is a motorcycle. on the other hand, severe crashes in marginal areas around cities tend to decrease when a foreign vehicle is at fault and when the driver of the at-fault vehicle is 30 to 40 years old.