Addressing spatial heterogeneity of injury severity using Bayesian multilevel ordered probit model

The objective of this study is to investigate the impact of influencing factors on crash injury severity at signalized intersections by employing a multilevel ordered probit model, which captures the spatial heterogeneity. The estimation was performed with Bayesian approach via Markov Chain Monte Carlo (MCMC) sampling. The crash data of 262 signalized intersections were used for two years from Hong Kong. The findings indicated that the multilevel ordered probit model can effectively address the spatial heterogeneity and provide better model fit than conventional ordered probit model. The results showed that at intersection-level number of conflicts and reciprocal of the turning radius are statistically significant for the crash injury severity while at arterial-level lane width, proportion of commercial vehicle and the presence of tram stops increases the injury severity. More importantly, about 38.3% of the spatial heterogeneity in crash injury severity is attributed to the arterial-level variables. The authors' findings can help designers and management departments develop a better understanding on intersection and arterial design and operation.


  • English

Media Info

Subject/Index Terms

Filing Info

  • Accession Number: 01720442
  • Record Type: Publication
  • Files: TRIS
  • Created Date: Oct 28 2019 10:27AM