A bivariate random effects spatial model of traffic fatalities and injuries across Provinces of Iran

From 2005-2015, Iran has experienced a 41.3 % decrease in road fatalities and an 11.1 % increase in non-fatal injuries. However, the trend differs across Iran provinces, and hence identifying factors that relate to road fatality and injury counts is an essential tool for improving road safety management programs and policies in the provinces. In this study, a statistical model was developed within a Bayesian framework with the aim of examining the annual fatal and non-fatal injury counts in the provinces of Iran during the period 2005-2015. Specifically, a bivariate spatial negative binomial Bayesian model with random effects was specified and estimated to account for unobserved heterogeneity due to the simultaneity effect between fatal and non-fatal injuries, the presence of province-specific factors, and the spatial correlation between neighboring provinces. All the three effects were found to significantly relate to the frequency of both injury types. Results also indicated that overall fuel consumption and share of diesel fuel consumed were positively related to fatal and non-fatal injuries. Higher population proportions of under 15, and 15-30 years of age were found to be positively associated with fatalities and negatively with non-fatal injuries. Furthermore, the annual number of hot-spots modified per 100 km of rural roads is associated with a decrease in fatalities. Results also suggest that the number of speed cameras operating on rural roads (within a province) might significantly decrease both fatal and non-fatal injuries. Accordingly, the implementation of active and targeted hot spot programs as well as speed camera programs are likely to improve safety performance of the provinces, and help to prioritize area-wide safety initiatives and programs.