Javascript DHTML Drop Down Menu Powered by



(re safety differently ADQ5.a – c.)

(published 3/4/17).


5. The article incorporates a piece on Dekker’s book, ‘Safety Differently’. It says “a wave of world class thinkers at the cutting edge of human factors research...are leading in new directions our thinking about why people behave as they do” (psychologists Hollnagel, Reason and Dekker are named). It then says, “At the same time, they are challenging many of the safety profession’s most fundamental articles of faith, such as the “Domino” theory of accident causation and the “Heinrich triangle” model of accident numbers and severity”.    



a). What is currently accepted as regards “why people behave as they do” at work?

b). What is wrong with the contemporary view identified at a). above?

c). What “new thinking” on human behaviour is being advocated and why?





“It would do scant justice to the improvements in safety and health...if a general impression were to be created that modern production processes are particularly vulnerable to capricious upset and major accident through ‘human’ causes that are uncontrollable – when the reverse is nearer the truth” (HSE 1989).



ADQ5a-b) can be addressed initially by looking at some of the general factors that forward thinking organisations and practitioners deal with when seeking to reduce the safety critical effects of human behaviour or performance variability. Of course, such factors applly to the maintenance and/or improvement of both physical and mental well-being but (and relevant for purposes here), changes designed to reduce stress and improve job satisfaction and or general health “must be desired by the employees” (Faragher et al 2005, italics in original).


First, industrial safety legislation in the United Kingdom is the result of centuries of learning and development. Learning came, sadly, at great cost to our fore fathers and many (e.g. Brown, Carlisle,) who spoke out on their behalf. Acts of Parliament came and went but, it was not until the Factory Act of 1833 that provision was made for our first Factory Inspectors. Those four (yes, the Act provided for just four Inspectors) initiated the mammoth task of improving conditions and the guarding of dangerous machinery. Many more Acts followed over the next 140 years and, informed by such as the Robens Committee report, Parliament eventually enacted the Health & Safety at Work etc Act 1974 (HSWA). From that enabling Act, we now have a core suite of Regulations that are clarified and extended by Guidance and Approved Codes of Practice from the Health & Safety Executive (HSE). The overlap of the Regulations etc. is such that when tied together (often via the Management of Health and Safety at Work Regulations 1999), a comprehensive management system results. 


In 1989, the HSE released HSG48, ‘Human Factors in Industrial Safety’. HSG48 effectively gave legal expression to an approach that many had already adopted when it identified “human factors as a distinct element” susceptible to the combined influences of “the organisation, the job and personal factors” (note: in the HSWA, the words damage and injury also mean any impairment to a person’s mental condition). In HSG48, the term human factors covered a range of issues including “perceptual, mental and physical capabilities of people and the interactions of individuals with their job and working environments, the influence of equipment and system design on human performance, and above all, the organisational characteristics which influence safety related behaviour at work” (HSE 1989). 


To help deal with such behavioural influences, the 2nd Edition (HSE 2000) and all subsequent ones elaborate around factors such as ‘social and family life’, ‘psychosocial risks’, ‘mental well-being’, ‘human reliability assessment’, human ‘performance influencing and shaping factors’, ‘stress’, ‘depression and anxiety’, ‘biological rhythms’, ‘personal attitudes, habits skills and personality’, ‘bullying’, ‘violence’, ‘psychological hazards’, ‘individual stressors’, ‘environment stressors’, ‘social and organisational stressors’, ‘psychosocial factors’, ‘job satisfaction’, ‘safety culture’ and ‘safety climate’.



“Individual characteristics influence behaviour in complex and significant ways. Their effects on task performance may be negative and may not always be mitigated by job design” (HSE 2009).


Interim Comment for ADQ5a:

Whilst some adhere to a philosophical notion that behaviour is ‘caused’ by matters beyond our individual control, the above allows those who must deal with it (daily, on the ground and in real time) to address those things that are factually and predictably known to influence our behaviour.


Interim Comment for ADQ5b:

Provided that the search strategy is not one that forces the analyst towards causal over-determination (which causes prioritisation difficulty, predictive inability and, with salient cause unaddressed, allows accidents and disasters to repeat), there is nothing wrong with the contemporary view of ‘why’ we behave as we do.


Interim Comment for ADQ5c:

From the other discussions, it is clear that a particular school is advocating that human behaviour, error and violation are, in certain circumstances, unpredictable, emergent properties of a system. Consequently, they are calling for a rejection of linear cause and effect (by way of ‘attack’ on Heinrich’s dominos) in favour of emergence and a view that ‘cause’ results from circular causal chains (of course, emergence does not reject linearity, emergence may be claimed if linear cause and effect cannot be found). However, the notions of complexity and emergence in safety differently (and resilience engineering and Safety I & Safety II) are underspecified, if specified at all.   


First, it should be noted that the cause of the underlying circular thought process is the embrace of ‘a systems perspective’. Also, there may be support from within systems theory for the perspectivist approach (Davies et al (2003) but, any such support is subject to certain expectations and requirements.


Clearly, writers such as Woods, Cook, Dekker and Hollnagel have now accepted that the common cause hypothesis is valid in the vast majority of cases (e.g. up to Dekker’s (2015) 1 risk in 9,999,999 safety units). What is required, therefore, are concise, explanatory reasons as to how and why, according to them, the common cause hypothesis applies where it does. Next, they should explain how and why the newly hypothesised relationship applies where it does and also explain a). why it does not apply in cases where the CCH applies, and b). how and why the CCH does not apply in the situations where the newly hypothesised relationship applies (Gholson et al 1989). The Professors need to provide something that is propositionally based and should also explain how any accidents they refer to meet the emergence requirement for radical novelty since, none so far have. Also, we need to know, by way of a clear, concise statement, what the difficulty and problem is according to them; i.e. as they have found it from their experience in the field. Furthermore, insight into how they intend to resolve the highly problematic matter of boundaries (Midgley 2000: Bammer 2003) is needed along with a model showing how and why the events are not linear or, how and why the un-forecasted outcomes are nonlinear (Vucic et al 2003). 



Notwithstanding that the paradigm underlying safety differently requires “living laboratories” (Nemeth (2008)  (i.e. some answers may need to await the outcome of Laing O’Rourke’s experiments), answers to the above questions are needed. Not least, they are needed in case the newly assumed relationships might be a figment of the analyst’s existing bias (Helmreich 1998). Furthermore, they might relieve some of the tension in those (e.g. Galea et al 2009) who see emergent properties as system wide phenomena since, at this time, antithetical approaches to causation are being held simultaneously. On that point, and as Heinrich (1941) also noted, assistance might lie in the fact that whilst “multiple causes are always present, they are not equivalent in role” (Feyer and Williamson, in ILO 1998). Indeed, applying Occam’s razor perhaps, “humans reach consensus regarding the directionality and nonspuriousness of causal relationships, in the face of opposing alternatives, perfectly consistent with experience” (Pearl & Verma 1994).


As it stands, the cited Professors are claiming that when they can identify the cause (e.g. as in Hollnagel’s example of a man falling from a ladder in Dekker et al 2008), the common cause hypothesis applies. However, when they cannot identify the salient cause from amongst the many ‘causes’ their search strategy compels, the circumstances and relationships are deemed to be “far more complex”. In short, and without any explanation as to how, why or according to who, certain problems (arbitrarily (until the above answers) determined) will be due to emergent properties that result from an infinite number of system wide influences and circular causal chains.

Returning to Hollnagel’s earlier referenced attack on the common cause hypothesis, a high degree of confusion was clearly expressed there around the matter of causation in general. Unfortunately, confusion seems to be breeding confusion as was recently revealed by Gantt (2017) (Gantt is co-editor at Safety According to Gantt, an accident wherein a man (for reasons known only to himself) collides with a lamp-post (one he has avoided 100s of times before) represents an accident without a cause!




The Institute of Industrial Accident Investigators. All rights reserved.