A Methodology for Assessing Safety Impacts of Highway Shoulder Paving

This paper introduces and applies an Empirical Bayesian (EB) analysis methodology for assessing highway shoulder paving impacts using Illinois data. Adding new outside paved shoulders is found to be more effective than widening existing paved shoulders and repaving the same width of paved shoulders. Shoulder paving is most effective for multilane highways, followed by two-lane and Interstate highways. Shoulder paving is effective for highways with per lane daily traffic not exceeding 10,000. It is found to be more effective in reducing injury and PDO crashes when the paved shoulder width does not exceed 8ft for Interstate highways and is between 4ft and 8ft for multilane and two-lane highways; and when the combined lane and outside paved shoulder width does not go beyond 20ft. No clear crash pattern is identified among lane widths of 11ft, 12ft, and 13 ft for equal total combined lane and shoulder widths ranging from 12ft to 24ft.