The aim of this study was to examine the effect of the dynamic urban traffic control system SCOOT (Split Cycle Optimisation Offset Technique) on the number of road accidents. Six areas where SCOOT has been in operation since the early 1980's - Worcester, Westminster, Maidstone, Southampton, Cambridge and Coventry - were compared with six non-SCOOT areas. Accident data were obtained from the Transport and Road Research Laboratory's STATS 19 database. A preliminary consideration of the data gave no clear indication of SCOOT having an effect on accident statistics. Detailed analysis was carried out by fitting generalised linear models to the data for the SCOOT and control areas using the GLIM (Generalised Linear (Generalised Linear Interactive Modelling) package. The most appropriate control areas for each SCOOT system were chosen by considering the correlation of the trends in accidents/casualties before the introduction of SCOOT. The period during and immediately after the installation of each system was treated separately from the before and after periods to avoid misleading results due to drivers' unfamiliarity with the changes. Changes in accident and casualty reporting procedures for fatal and serious accidents and accidents at pedestrian crossings, particularly in the greater London area, over the period of the study have made it difficult to establish true trends in accident and casualty frequency. The raw accident/casualty data for each individual SCOOT area provide no clear evidence of any change in accident/casualty frequency as a consequence of the installation of SCOOT. In some areas e.g. Worcester there was some evidence of a temporary increase in accident/casualty frequency in the period immediately following the installation of SCOOT. Analysis using GLIM to fit models for each individual SCOOT area based on the assumption that any SCOOT effect is multiplicative showed no consistent SCOOT effect for either the period immediately following installation or longer term. For some areas SCOOT appeared to decrease accidents/casualties while in others there appeared to be an increase. For total accidents the only statistically significant changes (at the 5 per cent level) were reductions in accidents in Maidstone and Southampton.
Abstract