What are we to make of safe behaviour programs?

This 2006 discussion paper from Andrew Hopkins critically discusses some assumptions underlying safe behaviour programs and some of their limitations in the context of complex systems.

I can only cover some of the points, so recommend you read the full paper.

Note: Trigger warning for those who violently object to ‘unsafe behaviour’.

Overall, Hopkins argues:

·???????? “Safe behaviour programs run the risk of assuming that unsafe behaviour is the only cause of accidents worth focusing on”

·???????? “The reality is that unsafe behaviour is often merely the last link in a causal chain and not necessarily the most effective link to focus on, for the purposes of prevention”

·???????? “Conventional safe behaviour programs aimed at front line workers are also of no use in preventing accidents in which the behaviour of front line workers is not involved”

·???????? “behavioural safety observations are likely to have their greatest impact if directed upwards, at managers”

First, I’ve skipped the background on what safe behaviour programs are. Nevertheless, the programs are said to be “highly controversial”, and where unions argue that they “amount to a return to the strategy of blaming workers for the accidents that befall them,2 especially when they are associated with programs that punish workers who have accidents”.

1. The fallacy of mono-causality

The first challengeable assumption of safe behaviour programs is said to stem from the view that most accidents result from human action and error.

This philosophy is seen reflected in the DuPont approach, with one General Manager observing that they “focus on equipment, not on the acts of people. In our experience, 95 per cent of accidents occur because of the acts of people”.

DuPont is said to “assert strenuously” that their approach is more than a simple behaviour modification system, but even so “is undeniably on behaviour modification and that is how it is understood by many of its advocates as well as its critics”.

Hopkins argues that there is a basic fallacy is in prescribing behavioural approaches in response to the belief that human action ‘causes’ most accidents. That is, “The fallacy is the presumption that accidents have a single or a primary cause”.

Instead, contemporary accident analysis works on the assumption that there is a potentially wide network of contributory factors in an accident. So while a human action may have been the last element to interact with the hazard trajectory before the accident, “that behaviour is better viewed as something requiring explanation rather than in itself an explanation”.

Production pressures are said to routinely lie underneath behavioural decisions and acts. Aptly, “Despite all the company rhetoric about putting safety first, the experience of many workers … is that production takes precedence over safety”.

Hopkins discusses how the mono-causality principle plays out in a behavioural change document from oil & gas. This document discusses different eras of safety philosophy that an industry moves through, being engineering, safety management and then behaviour. He argues that their conceptualisation is contradicted by other research.

But nevertheless, he challenges the idea that focusing on behaviour occurs after exhausting engineering and safety management approaches first, since this assumes that engineering improvements have already been addressed.

But this is a fallacy, as a focus on behaviour could naturally lead to discussions on how to avoid the need for those adaptive behaviours in the first place via engineering and better work design. Hence, “we find ways of eliminating or reducing risky behaviour that are not dependent on behaviour modification techniques”.

He argues that human factors specialists don’t typically commit this fallacy – and an example is aviation which has long identified the need for improving systems factors, like cockpit layout, CRM and more.

Blame and causality assumptions

Hopkins also talks about the byproducts of safe behaviour programs being blame. Despite efforts to avoid blame, he asserts that “there is an almost universal tendency to allocate blame at the point where explanation comes to an end” and because “safe behaviour programs explain accidents in terms of unsafe behaviour, it is almost impossible to avoid attributing some degree of blame to the victim”.

The hierarchy of controls

The hierarchy of controls is also presented as an argument against the logics of safe behaviour programs as primary initiatives. Be observes, for instance, that PPE at the bottom of the hierarchy is “PPE is notoriously ineffective and should only be relied upon as a last resort.”

Further, “controls that depend upon front line workers doing the right thing can never be totally reliable”. Despite this, many safe behaviour initiatives rely on PPE and behavioural rules as the main or only controls that are seriously considered; and these are often adopted as an afterthought or after the machine or process has been designed.

Because behavioural approaches are generally concerned with the lower end of the hierarchy, “a focus on behavioural safety can lead to the abandonment of any commitment to the hierarchy of controls”.

Behavioural programs and their search for sharp end performance over blunt end

Some behavioural programs are said to be explicit about their narrow focus on behaviour on the frontline – he cites an example from a behavioural safety manual.

He discusses other methods that rely on people, like SLAM/Take 5 quite favourably, noting that they are like mini risk assessments, and are “immune from many of the criticisms made of behaviour modification programs” [*** I’m not convinced of his arguments here….but whatever].

Nevertheless, Take 5s and the like can be ineffective if they degenerate into an “ineffective ritual”.

Behavioural programs often focus on behaviour with certain characteristics, being:

1) directly observable,

2) occur frequently enough to be counted repeatedly

He observes that such a mechanistic ruleset on which behaviours to observe and count necessarily misses a lot of other actions, decisions, behaviours etc that aren’t directly observable, or frequent.

He discusses examples of adaptive routines resulting from resource or work plan mismatches, goal conflicts etc., and how these may not be readily observable. He also talks about adaptive behaviours towards things like process disruptions and the like.

He expands on the limits of observable behaviours to the naked eye – like with how permit to work systems fail to achieve their goals, plant isolations, or responses to alarms.

Hence, “The casual behavioural observer is not in a position to identify unsafe behaviour of this type. Indeed there may be whole sequences of behaviour that need to be observed or even studied before it can be said that the behaviour in question is unsafe”.

For one, some work tasks may require expertise domain expertise to recognise behaviours; he provides the example of the Esso Longford gas plant explosion, and how neither workers, supervisors or managers realised that their actions to re-start the pump given the vessel characteristics could result in such a catastrophic outcome.

Pointedly, using this example he argues that none of these parties could have assessed these actions as ‘unsafe’, and “no behavioural observer would ever have picked it up”. Hence, a “focus on behaviour that occurs frequently and that anyone can readily observe inevitably restricts attention to a very limited, even trivial class of behaviours”.

Next he discusses the limits of behavioural programs, as they typically focus on active behaviours and not so much latent conditions within the organisation.

He uses examples of the Challenger and Columbia accidents or the Moura mine disaster, and how the key precipitating factors didn’t relate to active behaviours from workers.

Active behaviour focus also omit consideration of corporate funding and resources, and what managers attend to and the examples they set. He proposes that unless behavioural programs can include the behaviour and decisions of managers, then they “will be missing crucial contributory factors”.

Do safe behaviour programs work?

This section briefly explores some of the evidence behind behavioural approaches (noting the substantial evidence published since 2006). He observes that certainly evidence does exist highlighting that behavioural approaches can improve many facets of performance (depending on how you define and classify performance).

In any case, he argues that they also have the capacity to work, if we use the perspective of accidents involving a multi-causal network.

Whether behavioural programs deliver positive benefits crucially depend on trust between workers and management, and this in turn depends on leaders being perceived to be genuinely committed to safety.

Where trust does not exist, “evidence is that they will fail”. In these cases, management must “concentrate on mending relationships with workers”.

Conclusion

In sum, Hopkins argues:

·???????? “Safe behaviour programs run the risk of assuming that unsafe behaviour is the only cause of accidents worth focusing on” – the fallacy of mono-causality

·???????? Behaviour of frontline workers, in contrast, tends to be the last interfacing point in a network of factors and “not necessarily the most effective link to focus on, for the purposes of accident prevention”

·???????? While they can form a reasonable component of any comprehensive approach, they “should never be the central component and care should be taken that they do not shift the emphasis away from potentially more important safety management strategies”

·???????? “behavioural safety observations are likely to have their greatest impact if directed upwards, at managers”

·???????? “One major drawback of behavioural safety programs is that they miss critically important unsafe behaviour, such as attempts by workers to re-start processes that have been temporarily interrupted”

·???????? Moreover, “Conventional safe behaviour programs aimed at front line workers are also of no use in preventing accidents in which the behaviour of front line workers is not involved”

Link in comments.

Author: Hopkins, A. (2006). What are we to make of safe behaviour programs?. Safety science, 44(7), 583-597.