This paper presents a new approach to accident research, which provides some prospects for accomplishing some objectives, which have long been desirable, but have been difficult to achieve using other methods. These objectives include the following. (1) Setting the human factors in accident causation in the context of the overall causal pattern. (2) Showing how different causal patterns interact at the accident level, in terms of mediating processes as well as statistical co-occurrence patterns. (3) Enabling the more systematic and comprehensive identification of priorities for further work on accident causes and countermeasures. Much progress seems possible by linking advances in methodology with substantive research questions. Crucial problems that need to be addressed include: (1) the enormous loss of information necessarily resulting from the method used to code accident case files; and (2) the order of performing the different stages of accident analysis. The proposed new strategy uses human analysts to interpret a sample of accident case records. The analysts use various cumulative case study methods taken from other branches of the social sciences and medicine. A new 'micro-forecasting' technique is also included. Stages to be considered include: (1) data collection; (2) preliminary analysis; (3) general case-analysis procedure; (4) micro-forecasting; (5) analysis of countermeasures; and (6) generalising across cases.
Abstract