Applying random parameters model to evaluate the impact of traffic, geometric and pavement condition characteristics on accident frequencies occurred at a-roads networks in the uk

Abstract

This study proposed a fixed and random parameters model to analyze multiperiod dry and wet accidents. Applied random parameters model as a new approach in the UK to estimate accident frequencies provides a reasonable understanding of the main factors that affect accident frequencies. Therefore, this approach allows the research to identify and control for contribution factors that may be bias estimation. To demonstrate the application of the proposed models, a six-year (2005–2010) of accidents data from traffic accidents on Norfolk, Oxfordshire and Nottinghamshire A-road network has been used. The estimation results show that random parameters model performs better than fixed parameters models in terms of Akaike information criterion, log-likelihood test, and chi square test. According to parameter estimates the logarithm of the annual average daily flow, percentage of heavy vehicles, gradients and rut depth are positively associated with more dry and wet accidents. On the other hand, speed limit, radius of curvature, number of lanes, skid resistance and texture depth are found as the significant variables decreasing the numbers of dry and wet accidents.