Strength of the Variable

Abstract

The AASHTO Highway Safety Manual (HSM) provides methods for departments of transportation (DOTs) and other agencies to incorporate high-quality quantitative safety analyses into project development and decision making. Taking full advantage of the HSM, however, requires states to collect and maintain more detailed and comprehensive data on their roadway systems. Moreover, given that the methods in the HSM were based on research using select state databases, full use of the HSM requires that states calibrate the models to their own databases and systems. Calibration is needed to account for differences in state reporting thresholds, terrain, driver demographics, climate, and other unique crash attributes. This paper outlines the efforts undertaken by the Ohio DOT to implement the HSM through data collection and calibration. Historically, the Ohio DOT has not collected data for all attributes included in the Part C predictive models. During the past 2 years, the Ohio DOT collected additional data elements for its roadway system, including parking, driveways, and roadway curvature. Ohio DOT staff also undertook calibration of the HSM safety performance functions (SPFs) by using the collected data, historical observed crashes, and specialized spreadsheet tools. In some cases, calibration factors could not adequately adjust the HSM models, which might have instead required the development of agency-specific SPFs. Cumulative residual plots were used to evaluate how closely calibrated SPFs predicted crash frequency compared with the observed frequency. The Ohio DOT determined that the calibration factors were currently sufficient to adjust the HSM models and that future development of Ohio-specific SPFs could be prioritized on the basis of calibration evaluations.