AN EVALUATION OF ACCIDENT SURROGATES FOR SAFETY ANALYSIS OF RURAL HIGHWAYS. FINAL REPORT. VOLUME II. TECHNICAL REPORT

The objective of this study was to validate the use of accident surrogate measures for analyzing safety on rural highways. Emphasized were inexpensive, quickly obtained measures. They were tested for isolated horizontal curves and unsignalized intersections on two lane state highways. From a literature review, 23 potential surrogates for curve accidents and 20 for intersection accidents were identified. They were measured in samples of 78 curves and 121 intersections in western New York. Predictive equations for accident rates were derived by multiple regression analysis. Degree of curvature and traffic volume were the best predictor variables for curves, while major and minor road traffic volume, minor road average stopped delay, and percent left turns were the best predictor variables for the intersections. The maximum variance in accident rates accounted for was 31 percent, but mathematical considerations suggested that stronger relationships were theoretically possible with more reliable and precise surrogate measures. Increasing reliability through sophisticated instrumentation, longer measurement periods, and other methods was recommended.