Viewpoint: Three steps forward – a research agenda for behavioural change

three steps forward 2Alongside the blog reboot, we also have a new contributor, Tom Wein, who is the founding partner of Aware International, a social enterprise offering behavioural science for international development. Tom will be writing about the practice of applying behavioural science. 

Behavioural change has come a long way fast. In recent years, we have developed a far more fine-grained understanding of different behaviours in different contexts, and of the psychological processes behind them. We have a huge list of revealed heuristics and biases. We have too a healthy range of well-evidenced models for grouping and prioritising factors in behaviour, in relatively parsimonious fashion. We have begun (though there is much work to do) to think about how to consistently and transparently link our understanding of a behaviour to recommended interventions to change that behaviour. Neuroscientists and evolutionary psychologists are enthusiastically attacking the problem of where these psychological processes might come from, and work has begun on the effort to validate and investigate findings across cultures.

Of course there is vastly more to do in all these areas. For instance, there are new contexts to examine (the field remains dominated by health and financial behaviours) and our models need further refinement. Doubtless there are new heuristics and biases to uncover too. Yet these are areas that are receiving plenty of researchers attention; there are a number of crucial topics beyond this that remain under-examined.

This piece outlines three interwoven lines of work that seem particularly pressing. These are:

  1. linking linear behaviour change with our understanding of complex systems;
  2. developing a good basis for selecting behaviours to change; and
  3. understanding what we mean by behaviour in the first place.

Underlying all of this, a constant presence, is the need to check our work and replicate our findings.

Behaviour in complex systems

Those interested in behaviour change often contrast their work with that of traditional communications professionals. One of the key differences, they claim, is that behavioural change offers a much tighter focus on achieving a particular, observable change. Focus is surely a good thing, and many ordinary communications campaigns have hopelessly, unmeasurably broad aims – so broad that even if they could possibly succeed, one could never accurately judge their success. Yet the behaviouralist reaction to this can often lead to an overly linear approach. Too often, behavioural change posits a simple causal relationship between a specific campaign and an isolated behaviour. This is not how the world works.

Take, for example, preventing corruption: even the most well-researched, well-designed, tailored campaign to stop civil servants taking bribes will fail, if it fails to address the political and social pressures those civil servants are under from patronage networks below, corrupt benefactors above and co-conspirator peers. These groups have a strong interest in maintaining the corrupt behaviour, and will surely respond to our campaign. Or take obesity: our tightly-focused campaign to improve eating by these people now must be complemented by an understanding of the likely responses by peers and food companies, and by a consciousness of secondary and tertiary consequences.

Economists know quite a lot about the effect of incentives in complex systems; behavioural change advocates will need to match that expertise. A more focused approach to behaviour change brings benefits – but we need a much clearer and better evidenced picture of the downsides, of how and when targeted approaches succeed or fail, and of what the best response might be. Integrating the behavioural change approach with what we know of complex systems and how they come to change is an enormously important, daunting challenge. As a practical means of summarising how change happens in complex political contexts, international development’s Theory of Change approach (which allows actors to make their assumptions about social change explicit) might supply a starting point.

Choosing a behaviour

Most behaviour change approaches assume an objective has already been selected. The behaviours that will achieve that objective are then picked based on some combination of informal ethnography and political acceptability. (Good qualitative work happens sometimes, but it is certainly not the norm). Pragmatic flexibility must be preserved for behaviour change to thrive, but it is surely odd to devote so much detailed effort to determining how to change a behaviour, but so much less effort to figuring out if that is the most sensible thing to do. At the very least, some rules of thumb must be developed. What behaviours are to be candidates, and how are we to prioritise among them?

Just as we use search strategies for identifying the relevant papers, we need to establish search strategies for listing the behaviours we could seek to change in order to deliver a given policy objective. Then, we need to choose between the various options. How to do so? We need some criteria on which to judge behaviours, which at the very least makes assumptions explicit. As an initial set of considerations, those criteria should probably include the relevance of the behaviour to the eventual policy objective, the measurability of the outcome, an estimate of the likelihood of achieving the outcome and the ethics of altering this behaviour.

Without some more rigorous basis, we will design the interventions that suit us, or that suit the powerful. What is more, we will never have a justificatory basis for selecting a non-obvious route to change. This work will of course have to be linked to the understandings of systems called for above.

What is a behaviour?

If we are to list and choose between behaviours,  then we must know what a behaviour is; it is not clear that we currently do. Some suggested objectives are clearly broad and non-behavioural in nature. Similarly there is clearly a lower-bound – no one advocates three separate behaviour change campaigns to address reaching for the seatbelt, pulling the seatbelt down and securing it – but many cases are more marginal. Yet there is surely an ideal level of human action – not too vague, not too specific – at which change campaigns should aim. We just don’t know what that level is.

Without a good definition of the behaviour that we claim is at the centre of our approach, what response are we to give to those who suggest that we use a national communication campaign to ‘promote healthy lifestyles’, other than our general intuition that this is too vague to do much good? How are we to review the effectiveness of interventions, if we are less than certain what level of human action it is supposed to alter? This is more than a philosophical loose end; it is a gap at the heart of the enterprise. Much has already been written on the question of individuating actions in the philosophy and psychology of action literatures, but the behavioural change field has not examined or incorporated it.


We press always ahead into new areas, as we should. But our excitement at breaking new ground must be tempered by a realisation that much of our current knowledge is in doubt. Psychology is no science at all until it can replicate its findings, and sort the true from the merely promising. Too many of the most exciting discoveries of recent decades remain for now in epistemological limbo, neither false nor true, and we can only move forward with a solid base, and with healthy institutions that ensure that future results will receive the same scrutiny. This is not the place to propose how that should work, but any articulation of the discipline’s future that omits this duty is inadequate.

Practical, usable, explainable solutions in these three areas of research, plus the task of replication, would put us well on the road to an ‘end-to-end’ behavioural change approach, and begin to provide a transparent and explicit basis for decision-making at every level below that of high level policy and priority selection (which will properly remain the domain of consensus politics, not technocracy). Get to work!

You can learn more about Tom on our ‘Who we are’ page.

Tom is grateful to Dr. Chris Mills and the UCL Centre for Ethics and Law, who organised the November 2015 workshop on ‘Behavioural Public Policy: Theory and Practice’, at which many of these thoughts were crystallised. He is further grateful to both Chris and Dr. Jeroen Nieboer of LSE for their valuable comments on an initial draft.



Want to say something?

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s