Safety culture and beliefs in the nuclear industry — Looking forward: Looking back. Opening keynote address

Over the last three decades, the nuclear industry has come to accept both the concept of safety culture and the possible role that it might play in the aetiology of accidents and incidents and in safety management. This development is to be welcomed in general terms but is not without its shortcomings in practice. These largely concern the definition and subsequent operationalization of the concept and the way in which it has then been put to use. While many would agree that safety culture is an emergent property of a complex organizational system and is multifaceted, in the nuclear industry, its measurement and management have often been focused on self-reported beliefs, perceptions and behaviours. Interventions often target changes in these psychological (individual) outcomes at the expense of the other aspects of safety culture. Furthermore, many of the models that underpin this approach have assumed universality: one size fits all. While such models are useful in population level research, they can be seriously inadequate and even misleading in more particular situations. They are not sufficiently context specific at the plant level or in functions within plants. Unsurprisingly, therefore, improvements in safety culture, so modelled and measured, do not always strongly relate to improvements in safety management and safety performance. The paper explores the changes that might be necessary for a fair test of the utility of safety culture in determining the quality of safety management and actual safety performance. The final point raised in the paper is fundamental but missed by some. However cast, measured and managed, the concept of safety culture was never promoted as the sole determinant of safety management or the sole reason for safety failure. Therefore, judging the utility of the concept in relation to the quality of safety management in the nuclear industry can only be done logically in the context of the other factors involved. (author)