Outlines of a sensitising model for industrial safety assessment
Highlights
► Future practices of industrial safety assessment should rely on the insights of social sciences. ► There is a need as a consequence to produce models reflecting these insights and it is argued that there is not one single way of doing this. ► Four different research traditions are therefore identified. ► A classification framework helps to locate them and to specify potential synergies between works in research traditions. ► A sensitising model is introduced, resulting from the combination of a normative model and a descriptive one.
Introduction
Certainly one of the most challenging tasks in safety science today is to develop ways of assessing systems in order to capture the patterns identified by social sciences following major accidents, to prevent them from causing disasters. In other words, the challenge is to develop ways to better grasp in foresight what is being interpreted in hindsight, or, as shown in Fig. 1, to move from a study of past failures (1) to an anticipation of future ones (2).
This challenge is empirical, methodological, theoretical and epistemological (Le Coze, 2011a). This paper deals more specifically with the theoretical aspect of this challenge, setting aside the empirical, methodological and epistemological ones. A recent event such as the Macondo well disaster (Chief Counsel’s Report, 2011) demonstrates again the necessity of moving beyond traditional methods and tools designed for assessing safety. So far, these have been strongly dominated by technologically and quantitatively oriented rationales (e.g. probabilistic safety assessment), with attempts to complement them from other backgrounds such as human factors (e.g. human reliability assessment). This technological and quantitative rationale can now be considered too simplistic (although certainly prerequisite for any safety approach). It does not incorporate what makes these systems so highly complex, i.e. their dynamic and systemic properties (Le Coze, 2005, Vaughan et al., 2005). Yet, the current and next generation of technological developments require enhanced abilities, both from states, public and private companies, to better anticipate technological, human, organisational and socio-cultural types of failures (Evan and Manion, 2002). This calls for a better interface between technology and social sciences and their translation into safety assessment practices.
If one wants to tackle this challenge, one has to deal with a theoretical problem: data collection and interpretation are always, whether implicitly or explicitly, and to a large extent, knowledge-driven. A safety assessment must rely on some form of indications about where to look and what to derive from observations. As a consequence, one has to specify key dimensions indicating relevant areas to be considered and investigated for safety assessment during empirical phases (Le Coze, 2011b). This issue has been clearly formulated by Bourrier and Laroche (2001, p. 49) ‘One needs to determine where to look (…) it is difficult to study the entire functioning of an organisation or a system of organisations. One has to make choices and decide to approach the organisational reality through a series of slices, perspectives and angles of observations’.
Although some authors have now started to move in this direction and provide some guidance (e.g. Berman and Ackroyd, 2006, Dalzell and Hopkins, 2006, Reimann and Pia, 2007, Reimann and Pia, 2009, Boin and Schulman, 2010), this research area still remains without a strong base. For instance, no systematic and analytical overview of models is available so far for this particular subject. Yet, this is a necessary step in clarifying different design rationales for safety assessment. The first purpose of the paper is thus to offer such an overview. It identifies and classifies a variety of models available from a social science background (Sections 3 Safety management system (SMS), 4 Safety culture (SC), 5 High reliability organisations (HROs), 6 Accident models and investigations). A historical and ‘genealogical’ approach has been followed and has helped to delineate different ‘research traditions’. The second objective (Sections 7 Combining research traditions, 8 Outlines of a sensitising model) is to present the design rationale of a sensitising model established for industrial safety assessment. The theoretical aspects of this model are then discussed (Section 9).
Section snippets
Four research traditions
The last two or three decades of research by the social sciences in both safety and accident investigation offer many empirical and theoretical insights. Four historical traditions of empirical and theoretical research are distinguished in this paper: safety management system (SMS), safety culture (SC), high-reliability organisations (HROs) and accident investigations and models (AIM).
Introduction to the research tradition
The historical background of the ‘safety management system’ research tradition has been reviewed, for instance, by Hale, 1985, Hale et al., 1991, and more extensively in Hale et al. (1997). It includes empirical and conceptualised knowledge from industrial and consulting practices (e.g. Petersen, 1978), guidance promoted by control authorities (e.g. the Health and Safety Executive – HSE in the UK), standards from international bodies across industries (e.g. the International Standards
Introduction to the research tradition
‘Safety culture’ is very often referred to as a concept introduced by accident investigation reports. For instance, “the term ‘safety culture’ first came to prominence as a result of the International Atomic Energy Agency’s (AIEA) report on the Tchernobyl nuclear accident (…) since then it has been discussed in other major accident enquiries and analyses of safety failures, such as Piper Alpha oil platform explosion in the North Sea and the Clapham Junction rail disaster in London. In both
Introduction to the research tradition
What has been labelled as high reliable organisation (HRO) is an empirical and theoretical research tradition launched by scholars in the mid-eighties. These researchers decided to better understand the specificity of what were seen as technically and organisationally complex high-risk systems considered to perform with a high level of success. Aviation, the army and nuclear organisations were then investigated by a multidisciplinary team representing a mixture of psychological, engineering,
Introduction to the research tradition
The history of investigating organisational dimensions of accidents combined with production of models is several decades old. In the seventies, Turner (1978) was able to analyse more than eighty cases of accident investigation reports in order to elaborate what he called the ‘incubation model’ of disaster. Turner then put forth an empirical and social science mode of research favouring the study of many different cases in search of patterns likely to be shared between them. This strategy has
Different options
In the present situation, one option is to emphasise the specificity and diversity of the research traditions, a strategy followed for instance by Rosness et al. (2004). One therefore selects which tradition is suitable for his/her purpose (e.g. accident investigation). With this option, research traditions are not merged. Instead, they are differentiated to stress what distinguishes them. A different option is to attempt to combine this diversity of traditions into a global model. This second
Design rationale
Fig. 3 reveals the potential for many combinations. For instance, Weick and Sutcliff (2001), but also Hopkins (2005), show how compatible the concepts of ‘safety culture’ (as defined by Reason (1998)), and ‘collective mindfulness’ (as characterised by Weick et al. (1999)) are. They compare these two, then combine them for the purpose of providing design principles for organisations wanting to sustain or reach high levels of safety (an earlier example of comparison is Pidgeon, 1991, Pidgeon, 1998
Discussion
This sensitising model is designed for the purpose of exploring possibilities for improving industrial safety assessment. It is based on a design rationale (i.e. four requirements) that is summarised in relation to the selected models and their combination in Table 2.
This sensitising model is designed as a variation from the practices of technological safety assessment in order to incorporate knowledge from the social sciences. It addresses the theoretical problems identified in the
Conclusion
This paper introduces a sensitising model which aims to provide a perspective to improving safety assessment beyond a technologically oriented approach unable to incorporate the dynamical and systemic side of safety. One core issue is to indicate in foresight what to look at and where to look in attempting to capture patterns that one finds theorised in hindsight by social sciences. For this purpose, the paper specifies a design rationale based on four requirements leading to the selection and
Acknowledgements
The author would like to thank one reviewer for his comments which helped to sharpen the arguments of this paper.
References (121)
The relationship between culture and safety on offshore supply vessels
Safety Science
(2009)Safety culture and the issue of power
Safety Science
(2009)- et al.
Perspectives on safety culture
Safety Science
(2000) The nature of safety culture: a review of theory and research
Safety Science
(2000)Culture’s confusion
Safety Science
(2000)- et al.
Modelling of safety management systems
Safety Science
(1997) Theories of (safety) culture revisited—an anthropological approach
Safety Science
(2008)Studying organisational cultures and their effects on safety
Safety Science
(2006)Implementing a safety culture in a major multi-national
Safety Science
(2007)Organisations and disasters: from lessons learnt to theorising
Safety Science
(2008)