U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590

Skip to content U.S. Department of Transportation/Federal Highway AdministrationU.S. Department of Transportation/Federal Highway Administration

Federal Highway Administration Research and Technology
Coordinating, Developing, and Delivering Highway Transportation Innovations

This report is an archived publication and may contain dated technical, contact, and link information
Publication Number: FHWA-RD-98-133
Date: October 1998

Accident Models for Two-Lane Rural Roads: Segment and Intersections

2. Literature Review

Intersection Variables

Traffic Flow

Traffic flows (ADT) have often been used as measures of exposure or as explanatory variables in modeling accidents at intersections. Many accident studies have used intersection accident rates in the form of accidents per million entering vehicles (Kuciemba and Cirillo, 1992). This type of rate has been used for safety performance evaluations and safety comparisons even though it does not take into account the magnitude of conflicting movements. Another common way to measure intersection accident rates is in accidents per unit time. McDonald (1966) exhibited a model relating accident frequency (accidents per year) to a product of powers of the cross-road and major road entering ADT.


  • N = number of accidents per year
  • Vm = major road ADT in vehicles per day
  • Vc = cross-road ADT in vehicles per day.

Leong (1973) proposed comparable but simpler models of the form:

A method for handling exposure measures developed by Surti (1965) was applied by Hakkert and Mahalel (1978). The latter authors proposed that accident frequency is linearly related to an exposure measure X, called index flow, calculated as the sum of the products of the flows at each of 24 conflict points defined by Surti. The model for urban intersections is as follows:

Hauer et al. (1988), as already noted, used traffic flows for each conflict pattern to predict accidents, found different functional forms and coefficients for different patterns, and addressed the short- comings of simple models of intersection accidents in terms of flows. The need for detailed models by pattern is presumably greater for signalized intersections than it is for stop-controlled minor roads with low traffic.

Control Type

The safety effect of converting to all-way stop was contradictory in two papers (Lovell and Hauer, 1986; Persaud, 1986). Lovell and Hauer affirmed the benefit of converting to four-way stop, while Persaud rejected its effectiveness. King and Goldblatt (1975) concluded that signalization reduces right-angle accidents but increases rear-end accidents, with no significant change in total accident-related disutility.

Sight Distance and Alignment

Three reports relate intersection sight distance (ISD) to accidents (David and Norman 1975; Wu, 1973; Moore and Humphreys, 1975). David and Norman reported that an increase in sight radius reduces the number of accidents. Sight radius was defined to be an average of all intersection sight distances at 50 feet from the intersection. Thus sight radius is not equivalent to the ISD defined in the AASHTO Design Manual, the so-called "Green Book." Wu cited the safety effect of clear vision and poor vision at both rural and urban signalized intersections. Clear and poor vision are qualitative descriptors as opposed to precise quantitative measures of ISD. Bared and Lum (1992), in a presentation on the safety effectiveness of intersection design elements, concluded that sight distance and other alignment variables are important at intersections. Among others, Urbanik et al. (1989) affirmed the significance of sight distance on crest vertical curves at intersections. Intersection sight distance will be indirectly considered in this study by surrogate variables: horizontal curvature, vertical curvature, and Roadside Hazard Rating.


Previous    Table of Contents    Next
United States Department of Transportation - Federal Highway Administration