A Heterogeneity-Based Temporal Stability Assessment of Pedestrian Crash Injury Severity Using an Aggregated Crash and Hospital Data Set

This study utilized a unique approach to crash data analysis by examining the temporal stability of pedestrian crash injury severity and its contributory factors. Police-reported crash data and EMS-related injury data from Nebraska were obtained from 2014 to 2018, and random parameter ordered probit models for injury severity were estimated for each year to account for unobserved heterogeneity. Four discrete levels of injury severity were considered for model estimation: fatality, disabling injury/suspected serious injury, visible injury/possible injury, and no injury. Data were filtered based on several important variables of interest, such as pedestrian characteristics, crash characteristics, environmental and weather characteristics, road surface characteristics, pedestrian location of crash, pre-crash pedestrian conditions, contributory circumstances of a crash, presence of work zones, and time gap between actual crash-time and police-reported time. A series of likelihood ratio tests were used to determine the temporal stability of factors over the course of two consecutive years and then over all individual time periods. The likelihood ratio tests showed temporal instability among explanatory variables for different time periods as well as for consecutive years. The random-parameters ordered probit models estimated a random distribution for the following indicators: old pedestrian indicator, pedestrian not visible due to dark clothing indicator, marked crosswalk at intersection indicator, time gap of 10-30 minutes between actual crash-time and police-reported time, chest area injury, work zone indicator, and ice on road indicator. This exploratory research suggests significant policy implications to help improve pedestrian safety.

• Record URL:
  • https://annualmeeting.mytrb.org/OnlineProgram/Details/19395
• Supplemental Notes:
  • This paper was sponsored by TRB committee AED60 Standing Committee on Statistical Methods.
• Corporate Authors:

  Transportation Research Board

  ,    
• Authors:
  • Farooq, Muhammad Umer
  • Khattak, Aemal J
• Conference:
  • Transportation Research Board 102nd Annual Meeting
  • Location: Washington DC, United States
  • Date: 2023-1-8 to 2023-1-12
• Date: 2023

Language

• English

Media Info

• Media Type: Web
• Features: Figures; References; Tables;
• Pagination: 20p

Subject/Index Terms

• TRT Terms: Crash data; Data analysis; Heterogeneity; Injury severity; Pedestrian safety; Police reports; Time of crashes
• Geographic Terms: Nebraska
• Subject Areas: Data and Information Technology; Highways; Pedestrians and Bicyclists; Safety and Human Factors;

Filing Info

• Accession Number: 01874166
• Record Type: Publication
• Report/Paper Numbers: TRBAM-23-01089
• Files: TRIS, TRB, ATRI
• Created Date: Feb 23 2023 9:21AM