Multivariate Model of Road Crashes for Vulnerable Users
Enregistré dans:
| Auteurs: | , |
|---|---|
| Statut: | Versión publicada |
| Date de publication: | 2026 |
| Description: | Different authors have analyzed the occurrence of crashes involving vulnerable users, but most studies focus on pedestrians and cyclists. Few have explored the correlation between crash types and neglected another vulnerable user: motorcyclist. This article investigates the correlation between crashes involving pedestrians, cyclists, and motorcyclists at the area level in Costa Rica. Univariate and multivariate models using the Full Bayesian hierarchical approach were developed to relate crashes in the 489 districts between 2016 and 2022 with sociodemographic and exposure variables. Results showed that the multivariate model was more parsimonious, exhibiting lower posterior deviance and DIC compared to the univariate model, also the inclusion of multivariate random errors impacted both model fit and parameter estimates. Key variables like vehicle-kilometers, population, and proportion of households with vehicles showed significant improvements in their estimation parameters. Additionally, the multivariate model was better suited for estimating excess crash frequency. These findings can be very useful for future research exploring the correlation between these crash types, particularly at smaller scales like intersections or highway segments. |
| Pays: | Portal de Revistas UCR |
| Institution: | Universidad de Costa Rica |
| Repositorio: | Portal de Revistas UCR |
| Langue: | Español |
| OAI Identifier: | oai:portal.revistas.ucr.ac.cr:article/2570 |
| Accès en ligne: | https://revistas.ucr.ac.cr/index.php/ringenieria/article/view/2570 |
| Mots-clés: | pedestrian motorcyclist bicyclist Area-wide Correlation |