The diffusion of environmental behaviours: The role of influential individuals in social networks

In mainstream commercial marketing, the concept of a 'maven' is increasingly well established. A maven is an individual with two key attributes: they have specialist or detailed knowledge about consumer products; and they are perceived as a trusted and reliable source of information by other individuals in their social network. It is hypothesised - and increasingly supported by evidence - that a maven can play a critical role in the success or failure of new products, since their advice carries particular weight with consumers.

Following research into the nature of and drivers for pro-environmental behaviours over the past couple of years, the possible role of mavens has emerged as a concern. Are there, for example, 'green' mavens? If so, where are they, what are their characteristics, and how do they function? In the context of the wider forces shaping pro-environmental behaviour - social norms, mainstream advertising, levels of trust in the government and so forth - an understanding of the present and potential role of mavens would be extremely valuable to all those hoping to promote the rapid take-up of new pro-environmental behaviours among the British population.

This research - a collaboration between Brook Lyndhurst, Opinion Leader Research and Dr Julie Barnett of the University of Surrey (now at Brunel) - investigated the notion of 'green' mavens, with a view to identifying specific opportunities for communications and policy.

Methodology
The research comprised four main elements:

• A detailed programme of literature review, aimed at bringing together the present 'state of knowledge' on mavens generally;
• A compact programme of interviews with commercial marketing professionals to explore their experience of the role of 'word of mouth' and the significance or otherwise of mavens in the success or failure of new environmental products;
• A pilot qualitative research exercise, involving detailed face-to-face interviews with individual mavens, recruited using techniques already developed by OLR and adapted specifically for this project; and
• Producing a detailed synthesis report, pulling together the research elements, setting out our analysis and making recommendations for communications strategy, policy development and further research.

Findings
Numerous possible threads run through this volume of material; and, because the various issues are interconnected, there is no obvious, natural starting point either to any exposition of the material or to any analysis. In the end, we judged that the appropriate place to both start and finish these conclusions was with Defra’s headline pro-environmental behaviours themselves. Our conclusions take the form of ten propositions, which build sequentially and which are set out below.

Proposition 1 – the headline pro-environmental behaviours should be thought of as ‘social innovations’
The dozen headline pro-environmental behaviours comprise “new things to do” for the majority of the population. Although they are heterogeneous, none of the headline behaviours, not even recycling, are yet fully established as ‘social norms’. (Indeed, if they were, there would not be an issue of trying to promote them.) Relatively few studies have focused specifically on the way in which new environmental behaviours permeate through populations. However, lessons from other domains, notably health, suggest that conceptualising these behaviours as social innovations offers the most fruitful analytical path. By understanding them as ‘innovations’, the more general principles governing the diffusion of innovations can be brought to bear.

Proposition 2 – the general process by which such social innovations will permeate through the population is a process of ‘diffusion’; diffusion theory is therefore central
Diffusion theory draws on a range of disciplines, and the literature on diffusion abounds with examples of how novel products, beliefs and behaviours have permeated through populations around the world. Although relatively few of these examples are narrowly concerned with pro-environmental behaviours, once such behaviours are conceptualised as social innovations, it becomes more straightforward to see the generic linkages from the diffusion literature to the headline behaviours with which we are most especially concerned.

The literature review nevertheless highlighted the innovative nature of this particular research: we found no work directly relating to ‘pro-environmental mavens’ and remarkably little on the diffusion of pro-environmental behaviours. This was in large part why the literature review, in particular, was conducted (and reported on) in such depth. Further research, pilot work and/or policy development in this area will have to bear in mind the innovative nature of work in this area and, as a result, will need to lean carefully and thoughtfully upon the material presented in the report.

Proposition 3 – diffusion is the process by which novel behaviours move from ‘innovation’ to being ‘social norms’
The diffusion of successful innovations follows a predictable path over time. Although the precise naming and classification of this process varies between researchers, in broad terms an early adopter group within a population experiments with an innovation; on the basis of their experimentation, an early majority begin to take the innovation on board; a late majority follow the early majority; and laggards bring up the rear.

This pattern typically describes an S-curve, or sigmoid curve. Across a range of products, beliefs and behaviours, this pattern has repeatedly been observed. The broad socio-economic and psychological characteristics of individuals at the varying stages of the adoption process have been extensively researched. In broad terms, the process of adoption is the journey from innovation to normality, the path from a new idea first arriving in a social network [of whatever size] to it being a social norm that actually characterises that network.

It is clear that different types of ‘influencer’ are important at different stages of the diffusion of a new behaviour. Given the different stages of development and uptake of social innovations, the concept of a ‘maven’ is too narrow to apply to all of them.

Proposition 4 – specific factors can be identified that dictate the likelihood and pace of diffusion
The ‘science’ of diffusion is not yet sufficiently well developed for there to be a coherent explanatory model. There are many variables, interacting under conditions of complexity. It is not yet possible to say with great certainty whether any given innovation will, or will not, be successful. Nevertheless, the research identified a finite number of factors that, between them, shape the likelihood of diffusion taking place:

1. Relative advantage: the more potential value or benefit is anticipated from the innovation, the faster it will diffuse;
2. Trialability: ability to try the innovation improves the prospects for adoption and diffusion;
3. Observability: again the extent to which potential adopters can ‘see’ the benefits of the innovation improves the prospects for adoption and diffusion;
4. Communication channels: the paths chosen by opinion leaders to communicate an innovation affect the pace and pattern of diffusion;
5. Homophilous groups: innovations spread faster amongst homophilous [roughly, ‘like-minded’] groups;
6. Pace of innovation/reinvention: some innovations tend to evolve and are altered along the way of diffusion whilst others remain stable;
7. Norms, roles and social networks: innovations are shaped by the rules, hierarchies and informal mechanisms of communication operating in the social networks in which they diffuse;
8. Opinion leaders: opinion leaders [or, more generally, catalytic individuals] affect the pace of diffusion;
9. Compatibility: the ability of an innovation to coexist with existing technologies and social patterns improves the prospects for adoption/diffusion; and
10. Infrastructure: the adoption of many innovations depends on the presence of some form of infrastructure or of other technologies that cluster with the innovation.

The literature suggests that careful attention to each of these factors provides the route, on the one hand, to understanding the diffusion of any particular innovation and, on the other, to developing intervention strategies to promote more rapid diffusion than might otherwise be the case. Several of these factors are directly linked to the function of particular individuals operating in social settings. The importance of ‘homophily’ – the idea that individuals are particularly likely to be influenced by people they perceive to be ‘like me’ – is considerable. The literature strongly suggests that excessive ‘social distance’ depletes the impact of innovations. The role of ‘opinion leaders’, too, is highlighted. In the context of the other factors, these elements powerfully suggest that localised ‘opinion leaders’, operating in their own social networks, have an important role to play in the diffusion of innovations.

Proposition 5 – specific “catalytic individuals” have an important role to play in the generic diffusion of innovations AND Proposition 6 – particular functions of individuals are important at different stages of the diffusion process
Rather than simply focusing on ‘opinion leaders’ or ‘mavens’, many different types of influential individuals are identified in the literature. We came across some 40 near synonyms for ‘influencer’. This suggested to us that social influence is a bundle of characteristics and attributes that are spread across the entire population. Some individuals have more of certain characteristics. Depending on the research question, different studies investigate different subsets of these attributes, hence the large variety of terms and slight divergences in meaning. This in turn suggested to us that we needed to recast our thinking in terms of functions, rather than individuals. Instead of thinking about individuals as mavens per se, the focus should be upon ‘maven-like’ (and other) functions played by individuals.

There were several consequences of this alteration in our thinking, both theoretically and in terms of the methodology for the primary research. First of all it opened up the possibility that a single individual may play more than one function, at different stages of the diffusion process, and with respect to different subject areas. For example, an individual might play a maven-like role with regard to travel, and an opinion leader role with regard to politics. This was reinforced by the literature; several studies have shown that opinion leaders are more likely to have maven-like tendencies, and vice versa.

This change in perspective also enabled us to move away from the narrow ‘maven’ and to concentrate on what we termed ‘catalytic individuals’. ‘Catalytic individuals’ are those who play a particularly important role in the process of diffusion of an innovation, whether by being a trusted source of information and advice (roughly mavenism) or by ‘setting the tone’ of their social circles and, by their attitudes and behaviour, establishing an innovation as socially acceptable (roughly opinion leadership).

A further consequence of this approach is that we chose not to attempt to recruit ‘environmental influencers’ or ‘environmental mavens’. Neither category has previously been identified by the literature; and, as we deduced, the functionality of influence is not ‘either/or’, but a continuum, with every individual having a mix of potentially relevant attributes.

Furthermore, given the overarching objective of considering how particular individuals might ‘normalise’ pro-environmental behaviours, we judged that individuals that are already strongly pro-environmental would be perceived, in many cases, as insufficiently ‘like me’ i.e. too far from mainstream social norms to be effective influencers in the current climate.

Proposition 7 – “catalytic individuals” exist, can be found and have the potential to play an important role in the diffusion of pro-environmental behaviours
Although thinking in terms of functions rather than individuals is helpful for clarifying thinking, it makes the job of actually identifying these individuals much trickier. We concluded that a traditional questionnaire-based recruitment methodology would be inappropriate to locate catalytic individuals.

Instead, we devised an alternative recruitment methodology that combined ‘snowballing’ with socio-metric techniques. The review of the literature had led us to conclude that social networks (i.e. any group of interconnected individuals) are the context within which diffusion occurs and within which catalytic individuals operate. The concept of social influence is by definition an outward looking concept - it does not occur in a vacuum, but catalytic individuals influence those around them. We therefore decided that the best people to ask about who was influential and played a catalytic role were the members of the social networks within which catalytic individuals operate.

The literature indicated that catalytic individuals are to be found in all walks of life, in all socio-economic groups, and across all genders, ages and so on. We therefore identified a number of pre-existing social networks using criteria to ensure a spread of different types of people, and set about finding the catalytic individuals. The networks in which we searched included neighbourhoods, sports clubs, church groups, and school gate networks.

We conducted in depth interviews with 22 catalytic individuals. The 22 people we met shared some features and were completely different in other ways. No two individuals exerted exactly the same kind of influence.

Some individuals clearly fulfilled a maven-like role and were often asked for advice and information about their areas of expertise, which ranged from technology to travel to home improvement. Other individuals were more towards the opinion leader end of the scale, and gave many examples of when they had influenced others to change an attitude or behaviour just by expressing their opinion or through their own behaviour. Most of the individuals were in between the two extremes and had characteristics of both types of influencer.

Although we did not aim to recruit ‘environmentalists,’ many of the interviewees had already influenced those around them on environmental matters, from recycling to car use. It became apparent that the influence exerted by these individuals could very easily be turned to environmental matters. There were many examples of when they had played a part in the diffusion of an environmental behaviour or idea.

It seems that, in principle, the environment is no different to any other topic about which catalytic individuals influence others. It is subject to exactly the same conditions as other topics: the interviewees all stressed that they must believe in something and have enough knowledge about it in order to influence others. If these conditions are fulfilled, with regard to the environment or any other topic, these individuals are likely to influence others.

There is one factor that seemed to be the key to the interviewees’ areas of interest and knowledge: they were overwhelmingly motivated by altruism. The majority of the examples they gave about influencing others involved some element of helping them out through improving their circumstances in some way. Whether their focus was on their immediate circles or they had more global perspectives, helping others was by far the most important thing to them.

Overall, the type of social influence exerted by our interviewees would seem to transfer neatly to environmental matters in general. The main way to facilitate this transfer would be to convince the catalytic individuals that they would help others and improve their lives by encouraging them to be environmentally friendly. It would also be necessary to ensure they had sufficient knowledge to feel that they were expert enough to answer questions and to make a real case for why a behaviour was a good idea.

Proposition 8 – there are actual and potential barriers to any prospective engagement with catalytic individuals in the realm of pro-environmental behaviours
In the light of the literature and both waves of primary research, it became clear that there are, or are likely to be, a number of barriers that could limit endeavours to engage with catalytic individuals and/or to use such endeavours to accelerate the update of pro-environmental behaviours:

• The high number of factors influencing diffusion, and the formal complexity of the diffusion process, mean that, on the one hand, many other elements may need to be correctly in place for catalytic individuals to have a beneficial effect and, on the other, even a highly effective intervention associated with catalytic individuals may fail for ‘system-level’ reasons;
• Given a shortage of historic case studies on pro-environmental behaviour change, and a chronic shortage of “real world” data about social networks in general, interventions will necessarily have a highly innovative element to them;
• The importance of adaptability highlighted in the diffusion literature, as well as the consideration of ‘word of mouth’ as a marketing channel, highlight the lack of control that is inherent to a diffusion programme: the eventual ‘social norm’ may look quite different from the original innovation;
• It would seem that innovations characterised by ‘ambiguity’ – where the costs and benefits of adoption are not clear, or apparently contradictory – are much less likely to diffuse than others. This potentially poses considerable challenges for some of the pro-environmental behaviours, given persistent diffidence on the part of the general public about the seriousness with which climate change is being tackled;
• The literature suggests that the development of a ‘critical mass’ of adoption of an innovation is a key stage of the diffusion process – but it is by no means clear how to spot that a critical mass has been reached. This begs questions about for how long any intervention in support of an innovation may need to persist – other things being equal, it would be very possible either to under- or over-shoot;
• The evidence suggests strongly that ‘negative word of mouth’ can be a powerful force, capable of scuppering a diffusion strategy. The only antidote appears to be complete honesty about that which is being diffused: which could raise very challenging questions for some pro-environmental behaviours;
• Catalytic individuals appear willing to promote ideas they consider to be good ideas that are useful to people – but only if they themselves are (a) convinced and (b) sufficiently knowledgeable to be able to argue positively for the idea without jeopardising their own identity;
• They are also unlikely to promote an idea or behaviour that is ‘too far’ from the social norm(s) characterising their social group, not least because this would jeopardise the status (and future effectiveness) of the catalytic individual themselves; and
• There is also the question about whether it would be ethically proper for government to engage directly with catalytic individuals in pursuit of pro-environmental policy objectives. The research team began with mixed views on the ethical question, and these persisted throughout the study. In the event, it was the demeanour and character of the catalytic individuals themselves that, perhaps, provides the answer: these people cannot be ‘used’ in an unethical way. If they think the idea being promoted to them is a good idea, they will be more than happy to do your diffusion for you; and if it is a ‘good’ idea, then there is no ethical difficulty. If it is a bad idea, or a stupid idea, or an idea that will not actually help anybody, then catalytic individuals will not help anyway and the ethical difficulty does not actually arise.

At a general level, each of these issues can be considered as a ‘risk’, and will need to be managed accordingly. They do not, in our view, when balanced against the positive factors emerging from the research, constitute barriers that cannot be overcome.

Proposition 9 – it is possible to prioritise headline behaviours on the basis of the likelihood that catalytic individuals will have a useful role to play in their diffusion
In chapter six of the evidence report we derive from the literature a set of criteria that could be used to prioritise Defra’s headline pro-environmental behaviours. These factors combined the general factors thought to influence diffusion with some factors specific to pro-environmental behaviours. On the basis of a provisional analysis, a more complete version of which should, in our judgment, be conducted before any pilot work is conducted (see below), we have classified the behaviours into five groups, on the basis of (a) their current state of diffusion and (b) the potential for a useful role to be played by catalytic individuals operating in their social networks: We would suggest that this prioritisation could provide the basis for any follow-up work to this research.

Proposition 10 – it is possible to identify a number of ways in which to engage with catalytic individuals with regard to these behaviours
We have concluded very firmly that generally catalytic individuals, of a type not previously captured in the literature, have a potentially important role to play in promoting pro-environmental behaviours. The role they play, and precisely how they play it, will vary between the headline behaviours. The characteristics of the catalytic individuals we interviewed suggest that a programme of engagement is feasible. Such individuals would need to be convinced of the positive benefits of the proposed behaviour for the people in their social network; and would need to have sufficient knowledge and/or understanding to confidently promote the idea.

We have concluded, too, that there are a number of barriers and complexities to any intervention programme intended to make use of this conclusion. On balance, we feel these barriers can be treated as risk issues, and managed accordingly, and that pilot projects to explore how diffusion can be accelerated through engagement with such individuals should be attempted.

Whilst there remain some outstanding research questions – see below - the crucial questions for any pilot projects (in addition to precise decisions about which behaviour(s) should be the focus of any pilots, what level of resources should be committed, which networks should be targeted and so forth) focus on how, precisely, to reach and then engage with catalytic individuals.

We have made it clear that a questionnaire-based approach is unlikely to work: the idea of randomly sampling the population to find these individuals is misplaced. The approach used during this research – socio-metric snowballing – is remarkably similar to the techniques used in peer-to-peer commercial marketing; and some of the recent Defra-funded EAF projects have also relied upon ‘champions’ or catalytic individuals, accessed ‘the hard way’. This would therefore appear to be the most sensible way forward. Broadly speaking, a programme of this kind would comprise identifying catalytic individuals in specific social networks; persuading them of the benefits of the particular behaviour to members of their social group; providing them with tailored support material; and then allowing them the freedom to do what they do.

Such an approach would, certainly at pilot stage, be human resource intensive, and would also require quite particular skills (to find, recruit, engage with and support the individuals). Detailed work would need to be done on the ‘value for money’ questions (comparing hoped-for outcomes per £ with alternative approaches) but our research, and the experience in particular of the commercial sector, suggests that these calculations could well provide attractive results.

There are, nevertheless, alternative possibilities for access and engagement that could be considered:

• Green platforms – some of our interviewees, when invited to consider how they would promote the pro-environmental behaviours, spontaneously mentioned ideas like ‘environmental coffee mornings’ or ‘neighbourhood teams’. These kinds of platforms echo many of the schemes funded in recent years by, for example, the Environmental Action Fund (EAF). We know that catalytic individuals are keen to adapt or establish organisational infrastructure in pursuit of their objectives, so there may well be scope to disseminate lessons from programmes such as EAF within and, more especially, beyond the ‘environmental community’.
• Local delivery partners – given, on the one hand, the generalised nature of the influencers we believe should be targeted and, on the other, the fact that numerous existing governmental mechanisms reach down to the ultra-local level, it might be more appropriate to work through established network channels. These appear to be well developed in the fields of health and social care, and these are also fields where ‘social marketing’ more generally is increasingly well established. It may be possible, therefore, to elide the ‘pro-environmental’ element alongside other initiatives to access and engage with catalytic individuals.
• ‘Probabilistic marketing’ – traditional marketing segments populations on socio-economic criteria, considers which segments might be interested in a particular product or service, identifies where or when the target segments will be amenable to a message, then fires the message accordingly (during a particular television programme, on a poster at a particular road junction, in a particular newspaper and so on). We have seen that the catalytic individuals in which we are interested are scattered throughout the population – they have no common socio-economic characteristics, and could not therefore be targeted in this traditional manner. We do know, however, that they have similar personality traits – in particular, a strong desire to help others. An innovative possibility, mimicking certain commercial marketing campaigns, may therefore be to develop marketing messages that could be generically ‘fired’ with the intention that only particular types of individuals would notice or heed the message.

These approaches are all innovative, and further work would be needed fully to detail how any pilots could and should be done. The possibility of different approaches being trialled in different settings, for different headline behaviours, would seem – resources willing - appropriate.

Further work, too, could easily and usefully focus on particular outstanding research questions that have been highlighted by our work. Three, in particular, stand out:

• Firstly, our research has focused overwhelmingly on individuals that exert influence – we have attended very little to those that are influenced. Some recent research literature suggests that there are circumstances in which diffusion proceeds because a mass of easily influenced people influence other easily influenced people; whilst, more generally, even the influence of catalytic individuals is in large part a function of the views and perspectives of those that are being influenced.
  
  We do not believe that further desk research is required in this regard. However, there may well be scope for primary research to explore how and how often people are influenced. There would appear to be scope for both qualitative work (which would need to be carefully structured, since people are unlikely easily to know or to admit to being influenced) and quantitative work (suitably worked omnibus questions could, in principle, give very useful insights into the relative importance in people’s minds of catalytic individuals as opposed to other factors).
   
• The second area where relatively little is known concerns the composition and condition of actual social networks. We have seen that network conditions are a major factor influencing diffusion; we have seen, too, that there is very little ‘real world’ data on such networks. It would clearly be impractical to map the innumerable social networks that characterise modern Britain, but possibilities include:
  (a) action research projects associated with the pilot projects discussed above;
  (b) formal attempts to quantify the number of particular network types;
  (c) efforts to, in a particular location, understand the inter-relationship between different, over-lapping social networks; and
  (d) more theoretical work to identify the visible factors that could indicate a network’s readiness or otherwise to make more rapid shifts in social norms.
   
• Thirdly, we are aware through, for example, our work evaluating Defra’s Environmental Action Fund, that a number of projects in the UK in the very recent past have been pursuing behaviour change programmes that involve a focus on specific individuals. At the time of writing, very little of this work has been formally evaluated, and still less has it been drawn together in a manner that might complement the literature-based and interview based material presented in this report. An exercise might therefore usefully be undertaken, in due course, to ensure a dynamic process of learning lessons from these experiences.

There will, of course, always be more research that could be done. However, as Eric Beinhocker put it when speaking to Defra recently about his book “The Origin of Wealth: Evolution, Complexity, and the Radical Remaking of Economics”, there comes a point in the development of an idea when there is no alternative but to give it a go, and tinker. We have been persuaded, on the basis of our research, that there is mileage in the idea of engaging with catalytic individuals to accelerate the update of pro-environmental behaviours and that it is time to give it a go.