determination of factors affecting rural crash frequency at marginal areas around cities

The high rate of traffic crashes and consequent losses have caused many social and economic problems in developing countries like iran. according to previous studies, 60-70% of rural damages in iran occur in marginal areas around cities. hence, identifying effective factors in rural crashes around cities is of great importance to take proper safety measures in such areas. this requires developing exclusive models to predict rural crash frequency around cities. given the limitations of fixed-parameters models, such as poisson and negative binomial regression, the present study uses a random-parameter approach, including the random parameters negative binomial (rpnb) and random effects negative binomial (renb) models to determine the factors affecting rural crash frequency at marginal areas around cities. the data on rural crashes on roads leading to isfahan over 3 years, were used to develop the models. the random-parameter models were used to consider the unobserved heterogeneity among crash observations. the goodness-of-fit measures indicated the acceptable accuracy of the developed models. meanwhile, the likelihood ratio test revealed a slight difference between the rpnb and renb. according to the outcomes, the factors influencing crashes around cities included traffic volume, the ratio of local and non-local vehicles, the proportion of longitudinal distance-limit violations, access points controlling, and access to rest area facilities.