Localized Safety Performance Functions for Rural 3-Leg Stop-Controlled Intersections in Alabama

Safety Performance Functions (SPFs) are often used to predict the expected crash frequency for a proposed highway facility (intersection or road segment). Highway Safety Manual (HSM) provides a set of SPFs for practitioners to use and also recommends the development of jurisdiction-specific SPFs using local safety data. Because only limited variables are available in the data, SPF models need to handle the potential unobserved heterogeneity in data. Random-parameter models are often applied to develop SPFs; but the model outcomes are still to represent the regionwide relationships between crash frequency and factors. Given the interactions between traffic safety and the environment (social, culture as well as geography), the unobserved heterogeneity is likely related to the space. This study employs a spatial modeling approach, namely Geographically Weighted Negative Binomial Regression (GWNBR), to incorporate the spatial heterogeneity into SPF model specification. In contrast with models that provide regionwide SPFs (e.g., a state), the GWNBR model can generate an SPF for local areas (part of the region), called Localized SPFs (or L-SPFs). This study uses the 2014 to 2018 geo-referenced crash data from Alabama to develop L-SPFs for rural 3-leg stop-controlled intersections. The results show the L- SPFs estimated by the GWNBR model vary substantially across Alabama. For example, the coefficients for traffic volume (AADT) range from 0.121 to 0.919 across different parts of the state. Practitioners and decision-makers could use the L-SPFs to predict crash frequency in their local area, so the countermeasures and funds could be better allocated to reflect the local situations.


  • English

Media Info

  • Media Type: Digital/other
  • Features: Figures; Maps; References; Tables;
  • Pagination: 19p

Subject/Index Terms

Filing Info

  • Accession Number: 01764275
  • Record Type: Publication
  • Report/Paper Numbers: TRBAM-21-03672
  • Files: TRIS, TRB, ATRI
  • Created Date: Dec 23 2020 11:24AM