Historically, major accidents have been claimed to be the initial change tool. From the establishment of investigation boards the transport sector has developed systematic approaches to investigations. Especially in aviation, but also in other transportation sectors, in-depth accident investigation has led to multiple technical design improvements. The major goal has been to reveal different patterns in the learning processes from accident investigations, with emphasis on the transport sector. This part of the project has been finalized. The main hypothesis in ACCILEARN states that accident investigations play an insignificant role in terms of learning. The research work so far can neither confirm nor reject the hypothesis. However the research work in this project has shown the following results:
- A new tool to analyse the potential of learning.
- A historical analysis of the organizational development of accident investigation, revealing that even though learning has been referred to no attempt has been made to define either what constitutes learning or how learning processes should be designed.
- The historical study on the development towards multi-modal accident investigation boards concerns the lack of references to different investigation methods or contemporary risk and safety research.
- The hypothesis that unique major accidents had an important impact on political decisions has proved probable and could also contribute to explaining the chronological discrepancy between Norway and Sweden.
- The structure and mandate of the investigation boards puts limitations on the scope of investigations and hence on the potential for learning from accidents. Moreover, the investigation reports generally lack, or do not specify, the intended process of actions at different societal levels. The Swedish board has the broadest mandate.
- The most important factor for learning from an incident or accident is how the information and knowledge generated by the accident itself is dealt with immediately after the event has occurred.
- Our scrutiny of single investigation reports (mostly ad-hoc based) tends to come very late, and the contents are only partly or occasionally directed toward learning.
- Investigation processes or systems for investigation of accidents are not designed on the basis of risk analyses. (The integrity of the investigators, which may be important for securing an objective approach, may be in contradiction to an active learning approach, which should include relevant parties as early as possible.)
- By taking on a multilevelled approach to learning, accident investigation is in this part of the project considered ‘by default’ as part of a bigger picture. Here, learning was pursued in and between several interdependent elements. Our findings suggest the significance of a multi-scope to understand learning, in which the accident investigation plays a part. Patterns of influence in and between different organizations or systems are intricate and multilevelled. It must be emphasized, however, that the aim was not here to identify or ‘reveal’ the true order of things. Instead, our intention was to invite interviewees to review their experience and reflect on their views and interpretations of learning in this elaborate context, in other words to identify how learning, relating in some way to the given accident was ‘produced’.