Random Parameter Framework for Two-Lane Rural Roadways: Findings from Washington State

This paper discusses a modeling framework for Safety Performance Function (SPF) estimation, based on explanatory variable type and modeling technique, for 31 crash classifications. The scope of this research was approximately 5,443 miles of two-lane rural highways in Washington State, with SPFs analyzed in terms of crash data, roadway geometry, and roadside features aggregated between 2002 and 2010. Model evolution involved traditional and advanced fixed parameter Negative Binomial (NB) SPF estimations, followed by segment specific estimations using Random Parameter Negative Binomial (RPNB) models, built on two homogeneous segment datasets, one with pure mid-block segments, and the other including intersection ranges. Model behavior was analyzed considering roadside information, existence of intersection segments, specific route effects, and parameter heterogeneity. It was found that roadside features could have a significant impact on route specific model performance, and that the magnitude for Incident Rate Ratio for Average Daily Traffic (ADT) was lesser in the total accident models on pure mid-block segments as compared to those including intersection ranges. Through the RPNB estimations, the logarithm of ADT, logarithm of segment length, and eight different roadway geometry characteristics were found to have segment specific varying effects on the predicted crash frequency across the 31 crash types. Over-dispersion values were significantly reduced when considering roadway geometry parameter heterogeneity across the crash types on two-lane rural highways, indicating the erroneous inflation of alpha due to the assumption of fixed parameters in NB models.

• Supplemental Notes:
  • This paper was sponsored by TRB committee ANB20 Standing Committee on Safety Data, Analysis and Evaluation. Alternative Title: Heterogeneous Overdispersion Parameter and Random Parameter Negative Binomial Models for Two-Lane Rural Roadways: Findings from Washington State
• Corporate Authors:

  Transportation Research Board

  500 Fifth Street, NW
  Washington, DC  United States  20001
• Authors:
  • Hong, Jungyeol
  • Hariharan, Barad
  • Shankar, Venky
  • Venkataraman, Narayan
  • Huang, Shuaiqi
  • Kezhiyur, Alben Jose
  • Milton, John C
  • Van Schalkwyk, Ida
• Conference:
  • Transportation Research Board 95th Annual Meeting
  • Location: Washington DC, United States
  • Date: 2016-1-10 to 2016-1-14
• Date: 2016

Language

• English

Media Info

• Media Type: Web
• Features: Figures; References; Tables;
• Pagination: 23p
• Monograph Title: TRB 95th Annual Meeting Compendium of Papers

Subject/Index Terms

• TRT Terms: Average daily traffic; Crash rates; Geometric segments; Highway safety; Mathematical models; Rural highways; Two lane highways
• Geographic Terms: Washington (State)
• Subject Areas: Highways; Safety and Human Factors;

Filing Info

• Accession Number: 01592851
• Record Type: Publication
• Report/Paper Numbers: 16-5439
• Files: TRIS, TRB, ATRI
• Created Date: Mar 7 2016 10:19AM