Abstract: Research was carried out to investigate the concept of 'accident migration by type', whereby attempts to reduce one type of road accident at a specific location can lead to increases in other accidents at that location. No evidence for such an effect was found, and instead evidence was found that 'untreated' accidents can be reduced alongside the 'treated' accidents. An investigation was also carried out intp 'spatial accident migration', whereby attempts to reduce accidents at a specific location can lead to increases in accidents nearby, but no evidence was found for this effect. Problems due to regression-to-mean effects were not fully resolved but this would not have affected the conclusions of this research. The use of different control methods was also examined together with the effect of using two alternate definitions of an accident 'site'.

Data were collected from the following highway authorities: Birmingham, Buckinghamshire, Cheshire, Hereford and Worcester, Hertfordshire, Kent, Newcastle-under-Lyme, Nottinghamshire, Oxford, Oxfordshire, Sheffield, Staffordshire, Warwickshire, West Yorkshire.