The accident network (AcciNet): A new accident analysis method for describing the interaction between normal performance and failure

Abstract

Accidents continue to create an unacceptable personal, social, and economic burden in many domains. Various accident analysis methods exist; however, key limitations have been identified. This paper describes a new accident analysis method, the Accident Network (AcciNet), that was recently developed as part of an ongoing collaboration between Human Factors and Ergonomics research groups from Australia and the United Kingdom. The method is demonstrated via an analysis of the Uber-Volvo fatal pedestrian collision. The analysis demonstrates how AcciNet goes beyond current state-of-the-art accident analysis methods to consider the role of normal performance in accident causation and identify the interrelations between failures, normal performance, and both human and non-human actors in the system. We describe the implications for accident analysis in practice and outline the next steps of the research program, including formal reliability and validity testing of AcciNet and the development of practical training materials.