An integrated survey–GIS and random-parameters ordered probit approach to modeling road accident severity in Amman

Abstract

Traffic crashes constitute a major global safety and public health concern, particularly in rapidly developing urban contexts. This study aims to identify and quantify the key determinants of road accident severity in Amman, Jordan, with a focus on both observable and unobserved contributing factors. The study adopts an integrated analytical framework combining crash data analysis, Geographic Information Systems (GIS)-based hotspot identification, field observations, and stakeholder surveys. A Random Parameters Ordered Probit (RPOP) model is employed to account for unobserved heterogeneity in the relationship between crash severity and explanatory variables. The analysis incorporates temporal factors (season, day of the week, crash time), crash characteristics (type of collision), driver attributes (age, gender, behavior), roadway and environmental conditions (lighting, weather, speed limits), and vehicle type. Hazardous locations were identified using Traffic Department records and validated through spatial hotspot mapping and on-site investigations. Additionally, structured questionnaires were administered to drivers and traffic police to capture behavioral patterns, knowledge of traffic regulations, and professional insights into crash causation. The results reveal that driver non-compliance, adverse weather conditions, older driver age groups, pedestrian involvement, and poor vehicle conditions significantly increase the likelihood of higher crash severity levels. The findings also demonstrate the presence of substantial unobserved heterogeneity across observations, justifying the use of the RPOP modeling approach. Crash severity in Amman is influenced by a complex interplay of human, environmental, vehicular, and infrastructural factors. Addressing these elements requires a multidimensional and data-driven approach that accounts for both measurable variables and latent behavioral differences. The study recommends a combination of short-term and long-term interventions, including targeted law enforcement, public awareness campaigns, roadway design improvements, stricter vehicle inspection systems, and the implementation of intelligent transportation systems (ITS). These measures can support policymakers and practitioners in developing effective strategies to enhance road safety and reduce crash severity in urban environments.