Empirical Bayes application on low-volume roads: Oregon case study

Abstract

ntroduction: This paper investigates the Empirical Bayes (EB) method and the Highway Safety Manual (HSM) predictive methodology for network screening on low-volume roads in Oregon. Method: A study sample of around 870 miles of rural two-lane roadways with extensive crash, traffic and roadway information was used in this investigation. To understand the effect of low traffic exposure in estimating the EB expected number of crashes, the contributions of both the observed and the HSM predicted number of crashes were analyzed. Results and Conclusions: The study found that, on low-volume roads, the predicted number of crashes is the major contributor in estimating the EB expected number of crashes. The study also found a large discrepancy between the observed and the predicted number of crashes using the HSM procedures calibrated for the state of Oregon, which could partly be attributed to the unique attributes of low-volume roads that are different from the rest of the network. However, the expected number of crashes for the study sample using the HSM EB method was reasonably close to the observed number of crashes over the 10-year study period. Practical Applications: Based on the findings, it can still be very effective to use network screening methods that rely primarily on risk factors for low-volume road networks. This is especially applicable in situations where accurate and reliable crash data are not available.