This paper describes an approach to the study of road accident causation which combines the traditional method of in-depth research from on-the-spot information with the traditional quantitative aggregation of police road accident files in the STATS 19 database. Groups of cases to be studied, such as 'overtaking accidents' are selected and assessed by researchers. Human interpretation is used to retain some of the information lost by other methods. A prototype sequence is built up which can be used for a variety of purposes. The method is outlined. The reliability of researchers' interpretations is examined. An acceptable level for the similarity of researcher judgements is suggested. The results of a study of researcher reliability are discussed.
Abstract