Can you remember back to 2010... David Cameron as the shiny, newly-elected Prime Minister, all guns blazing about his 'Big Society' vision? One of Cameron's projects, much talked about in those relatively halcyon pre-Brexit, pre-Trump, pre-Bake-Off days, was the cabinet office's Behavioural Insights Team (BIT), otherwise known as the 'nudge unit'.
The BIT was set up on the premise that people's decision-making is often irrational rather than based on straight logic and so changing the environment in which people make choices can bring about significant changes in their behaviour at very little cost. They introduce and test subtle interventions to encourage individuals to make different decisions. An example might be sending an SMS reminder about a dental appointment – it might be x% more effective at getting patients to attend if it mentions how much it will cost the NHS if they don't show up.
The team essentially work out the best way to 'nudge' people in a particular direction to help implement policy.
Since it was first established, the Behavioural Insights Team has moved away from 10 Downing Street, and is now a spin-off, social purpose company, jointly owned by Nesta, the UK Government and their employees. However, their goals and methods remain the same.
Here Ruth Doherty, BDJ Managing Editor, talks to Dr Michael Hallsworth, Director of Health and Tax at the Behavioural Insights Team, to find out more about what 'nudging' involves and how it can be effective in health policy.
Michael, in your own words, what does the Behavioural Insights team do?
We help government and policymakers get a more realistic understanding of human behaviour so they can build that into the way they go about designing services and policies. I think in the past we haven't been particularly good at doing that in the UK, partly because we tend to assume certain things about the way people make decisions. The assumption is that people take all the available information into account, weigh up the costs and benefits of taking each option, and then we pursue the one that maximises benefits to ourselves. In fact our behaviour is much more habitual and is governed by the cues in our immediate environment; these are often things we are not aware of in the moment.
We want the public sector, and anyone who is trying to achieve social goals, to take these factors into account when they are trying to influence behaviour. The main way we achieve this is through testing things. We run randomised trials in real world settings and then we show the difference that taking this approach makes.
We take the view that most government policy is intending to influence behaviour in some way. Therefore the scope is quite broad. It's not like you can just say we could do some behavioural stuff or we could do normal policy. Normal policy is about behaviour.
Have you ever done any work in relation to oral health?
We haven't, apart from thinking about sugar tax and how that would best influence behaviour. We have done work on missed hospital appointments and what are the messages you put in SMS reminders to get people to turn up, which might be relevant to dentists too.1
There's also a study carried out specifically with dentists in Germany called 'Nudges at the dentist',2 which looked at reminders to get a dental check-up. They found that a neutral reminder was most effective compared to messages which talked about the benefits of prevention, whether they were positively (if you do 'X' the benefits are) or negatively (if you don't do this 'X' will happen) framed. They found that the benefits message didn't really make a difference.
What about behaviour change around flossing?
Work has been done about flossing and helping people to make more effective plans to do so, but not by us. What I think is very important about things such as flossing is the extent to which you can integrate a behaviour into an existing habit. Doing this is essentially the key to making a sustainable change. For example, with a lot of people brushing their teeth has become automatic, so the idea is that you try to link a new behaviour, flossing, with an existing one, brushing, so the new element becomes a habit as well. The interesting thing about this is that the more steps that you incorporate the longer it takes to become automatic.
It's all about making an 'if/then' plan: so you say if X happens then I will do Y and you try to, in a very mechanical sense, link an occurrence and a response together. This has been shown to be very effective, more so than just creating a standard plan to do X. The idea being that if you make a plan to do something you underestimate the power of the situation itself, so your best intentions are led astray. For example, you might say I really want to eat more healthily so I'm going to avoid having cake at lunch but you don't realise that your having cake is actually being driven by you walking past a dessert display each day. These kind of influences can be very powerful, so you succumb to the temptation. So you need to design your behaviour in advance so that you are not exposed to the temptation and that's how you reclaim some autonomy over your automatic responses and behaviour.
Are nudges still effective if you know that they are happening to you?
Broadly speaking, yes. However, we would like more evidence on this. There is a recent study, called 'Warning you are about to be nudged', that explicitly tests this and they found it doesn't make much difference if people know in advance what's going on. If you reflect on this, there are many things which are automatic but we sort of know are happening.
In some instances you are aware of the whole thing. For example, in our work on communications to reduce antibiotic prescribing by doctors, one of the things involved in this was around comparisons of their prescribing practice with that of their peers. The letter sent explicitly persuaded them to prescribe fewer antibiotics and the comparison was not hidden, yet the intervention still works.
Another example is an interesting study in a more controversial area: end-of-life care. The study was carried out in the US and investigated the effect of changing the default setting on a form recording people's choices of palliative versus life-extending care. I think the default was at the top and may have even been pre-selected. They found that the default had a big effect on this very consequential decision but if you point this fact out and say 'look we know the default has a big effect on your decision, do you want to change your mind?', no one changed their minds. They stay with the default.
Are cognitive biases gendered in any way?
Not that I know of. We sometimes see differences by gender. For example, we had an intervention aimed at reducing alcohol consumption which linked the effects to appearances and women responded more than men to appearance-based messages than health-based messages. It wasn't a big difference though.
I think what's interesting is how universal some of these effects actually are. I'm often surprised by the fact that certain interventions don't seem to differ greatly by subgroups. There's a big question around cultural differences – are these really universal responses? A lot of the behavioural research has been done so far in wealthy, educated, post-industrialised countries with certain cultural aspects to them. However, now we have more than one office in more than one country and work with organisations like the World Bank, we can test to some extent whether the effects vary. We have found that broadly speaking some of the effects do hold up despite cultural differences. For example, we've seen similar effects on tax compliance in Guatemala as in the UK. But the jury is very much out until we collect more evidence.
In particular, research is needed to investigate the effect of social norms on the effectiveness of interventions – a lot of the research has been done in the US which is quite an individualistic culture, but in some countries, for example those in the Far East, there is a more collective culture – will that mean that the interventions are more effective there? Who knows? There is some evidence that it might be the case but we need to gather more evidence.
What about health inequalities – traditional methods of public health education can often exacerbate these, does nudging help or hinder?
The initial argument would be that 'nudging' is not as vulnerable to creating inequalities on the basis that it places less emphasis on giving information, which is the more traditional approach to public health. We know that it's generally the more educated, motivated, wealthy people who take up this type of information-based public health education. Now if we are looking less to information to convince people to change their behaviour and instead look more at the environments to make the behaviour easier to change, and if we say a lot of these cognitive effects are universal and non-conscious, then you could say they are equally affecting the more motivated and the less motivated. It's a more universal approach in that regard but I think we have yet to see conclusive evidence on that.
Another interesting aspect to consider is the idea that maybe people don't look at this information because they are under a lot of cognitive load from the very situation of being in poverty. There's a very interesting book called 'Scarcity' which looks at this idea. In that book they talked about an intervention where they gave a simple IQ test to farmers, I think in India, before and after the harvest. Before the harvest, the farmers did much worse at the test because they were concerned about limited means, but afterwards they were reassured of their income for a time and so performed better.
The idea is that if you are in poverty then you have to think about it all the time, and this doesn't leave the capacity to think about other things, like your health for example. So that's an interesting psychological explanation of some of the things you see and also some of the divides you see by socioeconomic status.
Appointment attendance is something you've worked on quite a bit in a hospital setting, do you have any advice for dentists on increasing attendance rates at their own practices?
Think about the messages you are sending out to remind patients about appointments because they are not all equal. And certainly making it as easy as possible for patients to choose times that will suit them makes it more likely that they will turn up.
There is also a really interesting question that has not yet been resolved about when you send appointment reminders for optimal effects. I'm amazed actually at the lack of research on this. It's a big question: what time of day? And also what medium? Though I think SMS is generally thought to be pretty good.
There is a question of how much money a dentist wants to spend on this problem and also what kind of relationship they want to have with patients. So for example, you can use fines to change behaviour but once you start fining people for non-attendance you enter a different relationship. You also have to ensure that you can collect that money effectively.
It's worth saying as well that we saw some indications in our hospital dataset, that if the hospital cancelled on you and you had to rearrange, then you were more likely to no show. That is very interesting because it implies that there is a reciprocity effect here. So if you are going to cancel on people be prepared for them to cancel on you.
Has anything in your work really surprised you?
One study we did quite a while ago now around organ donation involved placing many different messages on the screen that you went to after renewing your vehicle tax online. We tried various different messages around joining the organ donor register: some of the messages were social norm messages, eg 'X number of people sign up after seeing this message'. So in one variant there was just that message, in another one there was that message plus the NHS logo, and there was a third one with that message and a picture of a group of people, the idea that the image reinforces the message of the social norm, ie a lot of people do this too. And we found that the social norm message did increase sign-ups to the register by a fair amount, adding in the NHS logo made no difference, but adding in the picture of the people backfired enormously and it was the only one of the messages we tested which did worse than the standard message with no persuasive element at all and in fact would have reduced the number of people signing up. Now this was very unexpected, and a bit puzzling. The only explanation I could come up with is that the picture looked like stock imagery and that may have somehow triggered the idea that it was somehow untrustworthy or it was trying to persuade you in a too overt way and people reacted against it at some level.
So you could do a follow up study about what is this aversion to this kind of image, and to test if it is a real effect, it might not be. I think it is quite important to know because government uses stock images quite a lot in publications and websites and so on. What effect are they having on people's acceptance of the information?
What do you think is the most challenging policy area in which to affect change?
Loneliness and social isolation is a very hard one. This problem is particularly difficult to address because if people think they are being targeted because they are lonely, they will react against any attempt to help them. Also, even as a starting point, it's very difficult to identify when people are lonely. If you have a group of people who are all lonely, it can actually make things worse if they are then brought together. So some of the more obvious things you might do to help, like support groups and so on, are more difficult to achieve.
We did come up with an intervention idea for this area, in conjunction with Public Health England, but timing issues meant it was never implemented. It was based on integrating offers of help into the health system but crucially we tried to make it so it was an opportunity for them to help other people, so it was framed so that the 'lonely person' was not the passive recipient of help.
Is there anything frustrating about this work?
Sometimes interventions that have been shown to work do not get taken up as quickly as we would hope – particularly in health systems. This problem goes wider than behavioural insights interventions, of course. We are exploring whether behavioural science can help ideas to spread further and faster.
Do you ever 'take your work home' in the sense that you end up 'nudging' family or friends?
The nice thing about this thinking is that anyone can use it, in many different settings. My colleague Owain Service has written a book called 'Think Small' on exactly how to do this.
Personally, some years back my wife made a resolution to be on time more for meetings. So we created a commitment device – she wrote a cheque for £50 for a charity she didn't like. If she wasn't late all month, I would tear up the cheque. If she was, I'd post it!