‘Accident migration’ following the implementation of highway engineering schemes

The observed ‘migration’ of traffic accidents from treated sites to their neighbours has been the subject of considerable controversy in recent years. Some authors have argued that the observed migration represents a real change in the underlying mean accident frequencies at the sites involved, while others have argued that it is merely a statistical illustion arising, like regression-to-mean, due to a bais by selection. In this paper a detailed data set is examined with a view to separating any changes in observed accidents which reflect real changes in the underlying mean accident frequencies from changes which are due to chance alone. It is shown that, although regression-to-mean effects can give an illusion of accident migration, regression-to-mean does not explain the observed accident migration at treated sites.