Safety Effects of Shoulder Paving for Rural and Urban Interstate, Multilane, and Two-Lane Highways

Abstract

This paper introduces an Empirical Bayesian (EB) before-after analysis method for assessing the effects of highway shoulder paving on safety performance. In the first step, multi-year data on field-observed crash frequencies for highway segments with pavement resurfacing and shoulder paving (Type I treatment) and highway segments with pavement resurfacing only (Type II treatment), as well as untreated highway segments, on the Illinois state–maintained rural and urban interstate, multilane, and two-lane highways are collected. Next, data on untreated highway segments are used to calibrate safety performance functions (SPFs) and to predict crash frequencies for Type I and II treated highway segments, in the cases where respective treatments not been implemented. After computing the EB-adjusted crash frequency as a weighted sum of field-observed and SPF-predicted crash frequencies, the EB crash frequency for the after-treatment period is established by adjusting the EB crash frequency for the before-treatment period according to changes in traffic volumes and segment lengths between before- and after-treatment periods. The safety effects of shoulder paving are determined as the difference in the EB-adjusted Type I and II crash frequencies for the after-treatment period. It is revealed that reductions in shoulder-related fatal, injury, and property damage only (PDO) crashes from shoulder paving varies greatly by highway functional classification; and adding a new paved shoulder tends to be more effective in terms of crash reductions than paving the same width of the existing paved shoulder that has deteriorated over time or widening the paved shoulder.