Javascript DHTML Drop Down Menu Powered by



Simplicity Versus Emergent Complexity

(published 11/6/17).




Unrelated to the above diagrams, Raouf (International Labor Organization (ILO) 1998) introduces Heinrich’s 88% - 10% - 2% and dominos and moves immediately on to say...


“multiple causation theory is an outgrowth of the domino theory, but it postulates that for a single accident there may be many contributory factors, causes and sub-causes, and that certain combinations of these give rise to accidents”.



Firstly, Heinrich (1941) did not put forward a domino theory. Heinrich’s ‘theory’ relates to his finding of human behaviour as the predominant cause of the accidents in his study and his finding is now referred to as the common cause hypothesis (Wright & Van der Schaaf 2004). For some, this means that Heinrich’s is “very much a theory of single causation” (Stranks 2007). Whilst the term ‘single causation’ is somewhat inelegant, it will suffice here to describe a ‘theory’ that has not been refuted and that Dekker (2015) now concedes to be correct in the vast majority of (if not all) work environments.


Secondly, multiple causation theory (MCT), which has its “basis in epidemiology” (Ridley 1990), was introduced by Petersen (1971) and he made it clear that D.A. Weaver (see Fig.D5. above) had helped him to “develop many, perhaps most, of the ideas” in his book. Furthermore, Petersen (1971) rejected, albeit erroneously (Davies et al 2003), Heinrich’s common cause hypothesis en route to MCT. In short, multiple causation theory assumes that behind every accident there lie many contributing factors, causes and subcauses and these “combine together in random fashion, causing accidents” (Petersen 1971).


At this stage, we might compare Heinrich’s dominos at Fig.D1. above with Weaver’s (1971) at Fig.D5. Similar to Heinrich, Weaver’s first two dominos relate to social environments, heredity and personal factors. However, he then offers that accidents etc are “symptoms of operational error”; or, as Petersen (1971) put it, accidents, unsafe acts and unsafe conditions are all “symptoms of something wrong in the management system”. Shortly after Petersen, Bird’s (1974) “updated domino sequence” (see Fig.D3.) arrived and the following from Bird & Loftus (1976)...


“Re-emphasis is directed to the fact that the domino effect is not necessarily a direct chain reaction involved with single events. It is rather a reaction involving the potential of multiple events at each stage, with each established causal factor capable of continuing the reaction itself and of interacting with other factors to continue the domino effect”.



Already noted elsewhere in our discussions, Bird’s (1974) view is that all accidents result from management failures. Similarly, Reason et al (2006) believe that accidents are the result of ‘latent weaknesses in the system’ (Reason et al 2006) and that errors are but ‘consequences’ of ‘upstream systemic factors’ (Reason 2000). Accordingly, the Swiss cheese model (see Fig.D4) assumes that an organisational accident requires the ‘rare conjunction of a set of holes in successive defences’ (Reason 1997: 2004) or, a ‘concatenation of multiple factors’ (Reason, Hollnagel & Paries 2006). In short, Reason (1990) believes that all catastrophic events arise from the adverse conjunction of “several distinct causal chains”.



Interim Comment:

Having gone from “single causation” to multiple causation and its multiple chains of multiple causes and from human behaviour causing accidents to things causing accidents, the prevailing paradigm could not be further away from Heinrich.  


For Heinrich (1941), the explanatory phase will show the accident sequence having occurred in a “fixed and logical order” with agency at its root. From there, deeper analysis follows a logical path to Amalberti’s (2001) commonsense solutions. At the other end of the spectrum, the number of factors available for ‘selection’ has the potential for chaos. As Reason (1990) says regarding his ‘causal chains’, “if these are traced backwards in time, we encounter a combinatorial explosion of possible root causes, where the elimination of any one could have thwarted the accident sequence. There are no clear-cut rules for restricting such retrospective searches”. Of course, the known result of such a philosophy is that a ‘single root cause’ will never be available and, in consequence, “emergence thus forces it way to the front” (Hollnagel 2014) in preference to a simpler explanation. Consequently, it may now be easier to see how, for instance, a.). the approach is wholly subjective, b.). all accidents can appear dissimilar, c.). a ‘but-for’ approach to causation emerges, d.). causal over-determination occurs, and e.). the organisation can appear causatively responsible for any accident even though salient cause may not be amongst the ‘causes’ that were actually ‘found’ or ‘selected’.  


With the potential for unlimited factors to emerge (many of which may be extremely remote in space and time and have uncertain, even tenuous, relationships), it is unsurprising that simple linear explanations are impossible for a particular school of thought. Similarly, with multiple causes (each so very different, yet so capable in hindsight of having prevented things) to choose from, it is understandable that effective remedial selection and the prediction of future failure continues to be problematic for them. Of course, it might be recalled that simplicity of explanation is a great virtue in science. Where two different explanations arise for the same phenomenon, the simpler is generally to be preferred. However, “a simpler explanation is not a simplistic explanation. It is an explanation that leaves fewer things unconnected and explains more things with fewer principles” (Weinart 2009).   




The Institute of Industrial Accident Investigators. All rights reserved.