Modelling Road Work Zone Crashes’ Nature and Type of Person Involved Using Multinomial Logistic Regression

• Autor(a) principal: Vieira, Adriana
• Data de Publicação: 2023
• Outros Autores: Santos, Bertha, Picado-Santos, Luis
• Tipo de documento: Artigo
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10400.6/13366
• topic: Road traffic safety; Work zone; Multinomial logistic regression; Crash nature; Type of person involved; Type of person involved

The sustainable development goals “Good health and well-being” and “Sustainable cities and communities” of the United Nations and World Health Organization, alert governments and researchers and raise awareness about road safety problems and the need to mitigate them. In Portugal, after the economic crisis of 2008–2013, a significant amount of road assets demand investment in maintenance and rehabilitation. The areas where these actions take place are called work zones. Considering the particularities of these areas, the proposed work aims to identify the main factors that impact the occurrence of work zones crashes. It uses the statistical technique of multinomial logistic regression, applied to official data on road crashes occurred in mainland Portugal, during the period of 2010–2015. Usually, multinomial logistic regression models are developed for crash and injury severity. In this work, the feasibility of developing predictive models for crash nature (collision, run off road and running over pedestrians) and for type of person involved in the crash (driver, passenger and pedestrian), considering only one covariate (the number of persons involved in the crash), was studied. For the two predictive models obtained, the variables road environment (urban/rural), horizontal geometric design (straight/curve), pavement grip conditions (good/bad), heavy vehicle involvement, and injury severity (fatalities, serious and slightly injuries), were identified as the preponderant factors in a universe of 230 investigated variables. Results point to an increase of work zone crash probability due to driver actions such as running straight and excessive speed for the prevailing conditions.