Innovative methods for influencing behaviours & assessing success: 'Nudging the S-curve'

Start date:
September 2005
January 2006

This project explored possible models for explaining how and why sustainable behaviours may, or may not, percolate through the population

Defra commissioned 'Nudging the s-curve' to explore the potential implications for policy makers of recent developments in theories of behaviour change and social networks. The project was intended both to provide argument and evidence in support of a proposition that radical policy options for producing pro-environmental behavioural change need to be considered.

The project was commissioned as a 'think piece'. The content of the work was based on a period of research and reflection by the study team at Brook Lyndhurst, and fleshed out mind map analysis to generate possible interventions that might nudge a behaviour change system along an S-curve development path.

Three models of diffusion were considered during the research, each emerging from the work of a modern thinker:

1. Malcolm Gladwell‟s Tipping Points
In his book 'Tipping Points', Gladwell notes that ideas, products, messages and behaviours spread in a viral manner. As a result, little changes have big effects and such changes happen “in a hurry”. Gladwell proposes that the operation of these Tipping Points is dependent on the nature of relationships between three key types of individual - connectors, mavens and salesman – whose interaction brings about the kinds of diffusion with which we are concerned. Whilst superficially appealing, the 'tipping point' argument stops short of being able to say just how many such individuals are 'required', and in what settings, to bring about rapid diffusion. It provides the basis for a powerful backward-looking description of change, but little in the way of transferable policy conclusions.

2. Phillip Ball‟s Critical Mass
A deeper analysis by Phillip Ball, drawing parallels between emergent properties in the domain of physics and in social settings, finds that:

  • Non-equilibrium states are normal (that is, social systems are usually in flux);
  • Conditions in such states can be either stable, or “metastable” [ready to change, but not actually changed];
  • Metastability means that systems can persist in arrangements [over time] well beyond what might be supposed from an assumption of smooth transition; and that
  • Such systems do not exhibit smooth transition, nor even rapid transition of an S-curve form, but 'jump', extremely rapidly, through a phase transition [in the way water turns to ice, for example]

What is distinctive about Ball's analysis is not merely that he proposes change happening very suddenly once the 'critical mass' is achieved, but that there is no way of knowing, in advance, exactly when the change will occur. It could occur relatively early as the pressures for change build, or it could take a great deal longer.

3. Mark Buchanan‟s Ubiquity
Drawing on cross-disciplinary research, Mark Buchanan concludes that in all complex systems there are always small numbers of big things/events and large numbers of small things/events. More importantly, he concludes that the magnitude of an ensuing event is formally unpredictable at the outset. His analysis suggests that, within and across each and every group within society, there is the possibility of dramatic change, but little or no way in advance of knowing how and where such change will occur. He suggests that we may have a situation in which [dramatic] change can occur in some parts of the system, with little or no immediate impact on others. Rather than a smooth pattern of change – the S-curve – or a sudden jump – the phase transition – what we may in fact be dealing with is something akin to 'punctuated equilibrium', in which change happens in fits and starts.

Our interpretation of these analyses identified three important, common factors:

  • Inherent Uncertainty: The Gladwell analysis (S-curve) cannot tell us precisely how many and which Connectors, Mavens and Salesmen will be needed to catalyse a transformation. The Ball model (phase transition) cannot tell us precisely when a change will occur. The Buchanan approach (punctuated equilibrium) cannot tell us which steps will be large and which small. These limitations are not an outcome of weak analysis, or poor understanding, or a shortage of data: they are inherent, emergent properties of the large, open, complex systems that our societies are;
  • Networks: The mechanisms by which pro-environment behaviour change will take place will be among groups in society that are network-based. Better understanding of which networks are functioning, in what ways, with respect to different types of environmental behaviour – switching off lights, recycling, walking rather than driving, and so forth – will add powerfully to the ability of policy makers to target their endeavours; and
  • Broad targeting is not merely appropriate, it is essential. Within all network systems, there will be foci of greater influence, either because of the strength of linkages, the frequency of linkages, or linkages to other networks. Identification of key intervention points will not merely be more resource efficient, it will maximise the chance of having any influence at all on the system in question. Rather than focusing on key groups qua groups, policy should be focusing on groups whose network properties best lend themselves to the diffusion of change, in order to boost the probability of success.
  • Taken together, these three elements – inherent uncertainty, network based, targeted intervention – culminate in the idea that policies to promote pro environmental behaviour cannot, in fact, aim to 'nudge an S-curve'. Such an idea, in fact, falls into the trap of presuming a potential linear connection between cause and effect, between intervention and outcome.
  • Instead, the radical idea presents itself that policy should be attempting perpetually to ‘seed’ or catalyse change, through a wide variety of mechanisms, in a wide variety of places. A range of fundamental features of the social system mean that a model of policy intervention predicated on the steady refinement of interventions towards a set of policies that „work‟ may be ill-founded. Rather, given the complexities of 'behaviour change', a model of ceaseless innovation, within broad parameters of focus and in a network setting, offers a potentially valuable conceptualization of how to move forward.

Against this background, our research identified six implications for those concerned with policy-making in the domain of pro-environmental behaviours:

  • Tolerate waste and take risks – ceaseless innovation implies both risk-taking and 'waste' [since it cannot be known ex ante which schemes will work, although it is known that many of them will not]. The seemingly 'wasteful' expenditure of public money on behaviour change policies which appear not to work is an inevitable by-product of the inherent uncertainty within the social system.
  • Build on successes but avoid over specifying policies – a broad, strategic, pragmatic approach, rather than a narrowing, tactical, empirical approach to specifying policy should be used, focusing on case studies of both successes and failures, rather than searching for replicable cause and effect rules for social behaviour change.
  • Statistic rather than ballistic approach – Finding out the relative importance of the many factors at play is not practical. The relative weights assigned to factors is likely to vary between individuals, between groups and over time; and, in the absence of causality, tailoring policies even to the extreme 'segment of one' has no guarantee for success. Instead, policy should aim simultaneously to intervene on multiple levels in order to maximise the probability of behaviour change.
  • Focus on the common denominator of groups and networks – The issues that provide the basis for belonging to a group – the nature of the network linkages – are the appropriate locus for intervention with that group, or network. If the primary linkages are, for example, inherently spatial, then interventions that are spatial will have the greatest chance of 'traction'; if the linkages are predominantly concerned with belief, then interventions will need to be structured accordingly.
  • Hitchhike on consumer marketing – the greatest expertise in behaviour change is undoubtedly in the field of marketing. Policy should continue to make explicit and strategic use of this kind of resource.
  • From Pilots to Crystals - finally, we suggest that interventions need to progress from a model in which novel ideas are piloted, refined, and then (if successful) rolled out, to one in which novel ideas are used perpetually to seed, or catalyse, or act as crystals for, pro-environmental behaviour change. This is akin to the way in which, in a liberalised market economy, commercial success emerges not from 'perfecting' a product/service then rolling it out, but rather through the process of endless innovation.

Given the context for these conclusions, much remains to be done. The basic thinking needs to be tested, and thoroughly. Further consideration of network theory, and the practical requirements of analysing networks in the 'real world', needs to be undertaken. The potential implications for policy formulation, for monitoring and evaluation, for measuring 'success', need to be fully thought through and explored. In our judgment, further research and analysis of this kind would be worthwhile, on two main grounds. On the one hand, the scale of behaviour change required to bring a genuinely sustainable society into being is, it would seem, enormous. On the other, the range and complexity of factors required to explain behaviour (and the subset 'pro-environmental behaviour') is such that a longitudinal approach (considering diffusion over time) appears to have some useful advantages over a cross-sectional approach. In short, it seems that, whilst it may not be possible to 'nudge an S-curve', it might be possible to ‘coax diffusion’ – and that still seems preferable to old-fashioned carrots and sticks.

Brook Lyndhurst Blog

  • Energy efficiency: behaviour, rationality, economics and politics

    I had the pleasure of joining some 300 researchers and academics from around the world a couple of weeks ago to discuss the latest thinking on persuading consumers to use less energy.  The BEHAVE2014 conference took place in Oxford at a time when it is increasingly appreciated, by businesses, governments and civic society, that any [...] 

  • Herd behaviour amongst sports fans

    We had a conversation in the office the other day about herd behaviour and the difference between football and cricket crowds. Why is it that spectators at a football match can occasionally get aggressive and abusive, but spectators at a cricket match tend to act more like naughty schoolboys: boisterous but essentially good-natured? It’s a [...]