14. Safety and Behaviour Change

Whilst increasing amounts of resource are ploughed into initiatives to improve industrial safety there appears not to be a corresponding return on this investment as manifest by outcomes such as a reduced frequency of accidents or major hazards. Human behaviour has been identified as a major modifiable determinant of exposure to risks and hazards and is widely agreed to be a legitimate target of interventions to improve safety. Whilst numerous theories and frameworks have been used to understand and intervene with the behavioural determinants of risk the field suffers from a proliferation of models which are difficult to compare and contrast. This limits the accumulation of a coherent body of knowledge and expertise about what works, in what situations, for what problems, both within and across sectors. Furthermore, practices in the field may or may not reflect what is known scientifically, and the rise of the ‘safety industry’ and ‘safety professional’ means that the theories and techniques behind methodologies may be obscured by commercial interests.

This chapter examines the potential contributions of the emerging science of behaviour change to the field of industrial safety. It will outline the principles of understanding and changing behaviour using the Behaviour Change Wheel (BCW); a theory and evidence-based framework for designing behaviour change interventions that is gaining traction in cross-disciplinary research (Michie et al. 2014).

The study of behaviour change has its roots in experimental psychology. As such, the tools of the scientific method—theory, hypotheses, experimentation, evaluation—are at the heart of the approach. Whilst there has been a great deal of empirical research into behaviour change across a range of sectors—including industrial safety—the field has lacked a unifying framework by which findings from studies employing different theories and methodologies can be integrated. In many areas, this has resulted in a fragmented research literature that can be difficult to pull into a coherent body of knowledge for the purposes of designing interventions.

One recent approach to reducing this muddle and providing coherence is the Behaviour Change Wheel (Michie et al. 2014). The Behaviour Change Wheel is a synthesis of 19 frameworks of behaviour change identified across a range of behavioural and social sciences (Michie et al. 2011). The BCW consists of three layers (Fig. 14.1).

The approach begins at the hub of the wheel where the sources of the behaviour that could prove fruitful targets for intervention are identified (i.e. a behavioural analysis). A simple model of behaviour, COM-B, is used to conduct the behavioural analysis. COM-B is an acronym for ‘Capability’ (physical and psychological), ‘Opportunity’ (physical and social) and ‘Motivation’ (automatic and reflective), conceptualised as the three essential conditions for behaviour. Surrounding the COM-B model is a layer of nine intervention functions that can be used to address deficits in one or more of capability, opportunity or motivation. These intervention functions can then be linked to the behaviour change techniques (BCT’s; i.e. the active components of an intervention) described in published taxonomies of BCTs (Abraham and Michie. 2008; Michie et al. 2013). Finally, the outer layer, the rim of the wheel, identifies seven types of policy that one can use to deliver the intervention functions.

In order to translate general intervention functions identified using the Behaviour Change Wheel into a practical intervention for the given context, relevant behaviour change techniques are subsequently identified (for more information about behaviour change techniques, see Michie and Johnston 2011). Further consideration as to how the behaviour change techniques will be delivered within organisations or communities is taken into account by considering criteria such as the affordability, practicality, effectiveness and cost-effectiveness, acceptability, side-effects/safety, and equity of the intervention (i.e. the APEASE criteria; Michie et al. 2014). The BCW is designed to be a pragmatic framework that can be used to systematically design and support the evaluation of behaviour change interventions (see Michie et al. (2014) for more details).

The behaviour change approach described in this paper should be differentiated from the ‘safe behaviour,’ ‘behaviour modification’ or ‘behavioural safety’ approaches described by Hopkins (2006). Whilst behaviour and how to change it is at the heart of both approaches, ‘behavioural safety’ programmes are more narrow in focus and deal primarily with downstream causes of accidents. Theorists have criticized the behavioural safety approaches for falling foul of the ‘fallacy of monocausality’, which is the idea that there is often a single root cause—in this case, behaviour—of an event. Since human factors are often implicated at some point during the causal chain of events leading up to an industrial accident, behavioural safety programmes focus on understanding and modifying the associated behaviours. This approach has been heavily criticized by unions and academics alike because of the perception that it unfairly blame workers for the accidents that befall them, misdirecting attention from other factors that may play a role in the complex causal chain of events that leads up to an accident. In contrast, a behaviour change approach acknowledges that whilst unsafe behaviour may trigger an accident, the behaviour itself may be better viewed as something requiring an explanation rather than itself being the explanation (Hopkins 2006). As such, the behaviour change approach outlined in the BCW acknowledges that changing individuals’ behaviour requires making changes to the contextual influences on that behaviour. This may include the way people are organized, managed, motivated and rewarded, as well as their physical environment and the tools that are available to them (Fleming and Lardner 2002).

The starting point for any attempt to change behaviour is to determine what it is that needs to be different (the outcome) and then identify the behavioural determinants of that outcome. For example, if a company wishes to have fewer on-site accidents then it would need to specify the specific types of accidents (e.g. fewer machine tool injuries) and the behaviours that are related to this (e.g. using tools correctly, wearing personal protective equipment). Specificity is important since all behaviour is context dependent. The behavioural determinants of one type of machine tool injury may or may not translate to a different machine because different machines require different types of complex motor movements to operate them. Similarly, the behavioural determinants of a particular type of machine tool injury on the same machine may be different between sites because the context in which the machine and the worker interact may be different. There may be commonalities across contexts but one should always be open to the possibility that the behavioural determinants of outcomes could vary in important ways between them.

Accidents caused by deviations to established protocols for routine tasks are different to accidents caused by workers’ behavioural responses to unexpected events that have never happened before, or cannot be foreseen. Interventions to prevent accidents arising from the former are referred to as Safety-I approaches, whilst interventions directed towards limiting the damage caused by the unexpected and unforeseen are termed Safety-II approaches (Hollnagel et al. 2015). A behaviour change approach can be applied within both approaches since they share an emphasis on understanding behaviour in context. For example, being able to specify the behavioural determinants of effective responses to unexpected events can inform how organisations train individuals to respond to similar circumstances. In both cases the behaviour change approach is based on an understanding of how people actually carry out their work (work as done) as opposed to how they are supposed to do it (work as imagined). Interventions developed with the BCW always start from examining the behaviour in context and in many cases this requires changing the context to enable the behaviour rather than changing behaviour to fit the context. Within the industrial safety sector, an intervention based on behaviour change principles may well involve restructuring the nature of work itself in order to bring it in line with the known limits and constraints on human performance, rather than attempt to modify human performance in order to bring it in line with unattainable production targets.

Whilst behaviour change principles are amenable to looking at safety through both Safety-I and Safety-II perspectives, the BCW may at first seem more aligned with Safety-I approaches on the basis that the primary unit of analysis is behaviour. Nevertheless, by understanding behaviour as context-dependent, and broadening the definition of context to include the organisational, cultural and linguistic determinants on behaviour, models such as the Behaviour Change Wheel have the potential to be applied within Safety-II frameworks. Safety-I looks at why things go wrong and tries to identify and eliminate the causes of error, whereas Safety-II looks at why things go right and tries to ensure that they happen again, often by promoting organisational resilience. In both cases a focus upon describing the situation in behavioural terms and identifying the determinants of what goes wrong, or what goes right, is likely to generate helpful new insights for targeted interventions.

Once the behavioural targets have been identified the next step in designing a behaviour change intervention is to identify the determinants of the behaviour(s). The Behaviour Change Wheel uses a simple model of behaviour, COM-B (Michie et al. 2011) as a framework to categorise the various influences on the behaviour. COM-B is an acronym for the three essential conditions for behaviour; ‘Capability’ (physical and psychological), ‘Opportunity’ (physical and social) and ‘Motivation’ (automatic and reflective).

Once a behavioural diagnosis has been identified, the next stage is to design an intervention to influence the behaviour in the desired direction. The BCW describes nine categories of intervention classified by their function. These intervention functions and their definitions are described in Table 14.1. When designing intervention using the BCW the designer is encouraged to think about the entire range of possible ways to influence behaviour, not just the obvious ones or those with which they are most familiar as a result of professional training or experience. Nevertheless, it is clear that different intervention functions are more suited to influencing different forms of behavioural influence (e.g. training is an appropriate way to help people acquire the physical capability to perform a behaviour whereas persuasion will be ineffective). More information on the relationship between the COM-B domains and the interventions functions can be found in Michie et al. (2014).

Williams (2015) describes how the BCW was used to understand the positive impact of an environmental restructuring intervention on the incidence of road traffic accidents in a manufacturing company. The company fitted vehicles with driver-performance trackers as a means to increase productivity. Driver performance measures included ‘fuel efficiency’ and ‘sympathetic braking/accelerating’ and each driver’s performance was visible to their co-workers on a live display screen in the transport office. The unintended impact of this was a dramatic reduction in the occurrence of road traffic incidents. Interviews with the drivers in the company revealed that this change to the work environment (an environmental restructure intervention) changed the norms by which drivers judged their driving (tackling social opportunity) as well as providing motivation to be the ‘best driver’ (boosting reflective motivation) as well as providing feedback to drivers about their performance which they used to improve their performance (improving capability).

Organisational context is a powerful determinant of safety culture and decisions made by those in power will have an important influence on the expression of safety-related behaviour. The BCW identifies seven distinct types of influence which can be leveraged by authorities to influence behaviour. These include:

As with intervention functions, the relevance of each different policy function will vary according to the nature of the behaviour to be changed and the context in which it occurs. Nevertheless, the inclusion of a policy level of influence is an important differentiator for the BCW compared to other behaviour change frameworks, especially when applied to a sector where organisational culture has been shown to have such an important influence on the uptake and spread of initiatives to improve safe practices.

Intervention functions describe the different ways in which behaviour is influenced (e.g. through coercion and training) but they fall short of describing the specific techniques that are employed to bring about change. Behaviour Change Techniques (BCT’s) are the observable, replicable, irreducible components of an intervention designed to change behaviour, for example, goal setting and self-monitoring. It is possible to specify the content of behaviour change interventions by listing these active ingredients using a hierarchical taxonomy of behaviour change techniques that has been scientifically developed through a rigorous process of expert consensus review (Michie et al., 2015). The Behaviour Change Technique Taxonomy V1 (BCCTV.1) is available online as a web and app-based resource as well being described in Michie et al. (2014). Specifying behaviour change interventions at the level of BCT’s has the potential to allow researchers and practitioners to understand with more precision how different techniques may be related to outcomes. This may be useful for designing more cost-effective interventions since elements that are proven to be unrelated to outcomes can be discarded, or for identifying why similar interventions have different results in different contexts of delivery.

As a relatively new framework the utility and the effectiveness of the BCW is yet to be established in the field of industrial safety. Whilst it is for researchers and practitioners in this sector to decide for themselves the potential applications of this framework the following options may be useful starting points for further development based on its application to other areas:

Like many other fields in applied behavioural science, the field of industrial safety is in danger of being overwhelmed by the proliferation of theories and frameworks that can be brought to bear on the perennial challenge of ensuring that workers behave in ways that minimize the risks to self, others and the environment. The emerging science of behaviour change, and specific frameworks such as the Behaviour Change Wheel, create opportunities to integrate the valuable insights from diverse disciplines using a single framework that has cross-disciplinary appeal.