Road safety performance index: A tool for crash prediction

Abstract

The relationship between road infrastructure parameters and road crashes is not widely investigated in developing countries. The study objective is to create a measure for rating road safety performance as a tool to predict road crashes. A checklist included 184 elements categorized into nine groups, a sample of 105 selected streets in two districts with different population densities and incomes in Amman. The road elements’ safety impact was rated by two groups (professionals and non-professionals). Other data include vehicular and pedestrian traffic, speed, and road crash data. The principal component analysis was used to reduce variables and generalized linear models for the crash prediction considered the parameters or the road safety performance index as the predictors (un/extracted). The results showed that the road safety performance index in the less dense area is better than that of the densely populated. The professionals’ and non-professionals’ safety ratings did not establish a significant consensus. Modeling results indicated that using unextracted predictors poorly predicts road crashes for both the parameters and the Road-Safety-Performance-Index (RSPI). The safety-weighted predictors used in the modeling provide a valid estimate for road crashes with a high correlation of 69% for the) RSPI. Additional traffic parameters improve, to some extent, the prediction power of the RSPI-based models. The study concluded with the need to develop policies and procedures to improve road conditions and their performance, thus enhancing road safety.