A problem in assessing the effectiveness of small roundabouts over time

Abstract

In this study, a generalised linear modelling technique is applied to the number of accidents and traffic flow (total flow into the roundabout) for several small roundabouts in Newcastle upon Tyne for the years 1974 to 1992. There is a multicollinearity problem since there exists a high correla­tion between traffic flow variable and time variable as there has been an increase in traffic flow over time. Traffic flow is an important explanatory variable and cannot be dropped from the model. The conclusion regarding the effectiveness of small roundabouts over time can only be made under a weak assumption of accidents are proportional to flow. Based on this assumption, there appears to be no evidence of the reduction of the effective­ness of the small roundabouts over time.