To decrease the possible effect of random or spurious accidents on the identification of hazardous locations, a 0.3- mile (0.48-km) segment was chosen along with 1- and 2- year periods for accumulating and comparing accident data. An optimal method for identifying hazardous locations and sections was determined to be one which (1) maximizes benefits from improvements, (2) identifies locations with critically high accident rates, and (3) identifies potential hazards. A sample of 170 locations was used to compare several location-identification methods.
Abstract