Why being behaviour-smart trumps responding to demand

I’ve just come across a clear example of what I mean when I say that policy-makers and commissioners could be much wiser when it comes to using behavioural insights, and that being systematic about behaviour could lead to much better policy.  In a comment in response to a passing mention in a Bicycle Dutch blog on roundabout design, Patrick Lingwood (whom I believe is Walking & Cycling Policy Officer at Bedfordshire CC) gave this very detailed description (below) of how the authority arrived at a specific design.

It’s long, so I’m going to make my  response first; I don’t want to lose anyone here, so I’m not asking you to become experts in comparative cycle infrastructure and roundabout design. Before I do, can I just say that we should be grateful to Patrick for posting in such detail; it’s not often we get such a clear explanation of the rationale behind decisions.

Here’s my point: Bedfordshire’s design (effectively endorsed by the DfT) begins with an assumption about meeting demand for motorised traffic at the current level or higher, and of meeting current demand for cycling and walking, only more safely than at present. With this starting point, the resulting design is probably the best obtainable; personally, I’d love it, as I’m very comfortable taking prime position on my bike.

But I’m pretty sure that behaviour-smart policy-making would start with an analysis of current behaviours. A place-shaping local authority would identify the thousands of journeys of less than, say, 5km that pass through this roundabout daily and use a range of insights into the barriers to healthy, sustainable travel by the people who travel by car but are within walking or cycling distance of their destination. With a clear understanding of the behaviours it wants to encourage, and the design process would then begin. And insights from behavioural sciences, along with experience elsewhere, would be used to identify how to make the behaviours being encouraged both easy and normal. This would be reflected in the final design, whereas the Bedford plan is unlikely to reframe walking and cycling as anything other than (as at present) low-status. A behaviour-smart approach might seem radical at the moment, but it’s the future, as policy-makers become more exposed to behavioural insights.

Anyway, I’ve had my say – here’s Patrick’s explanation:

“with regards to the Bedford turbo-roundabout, it is easy to misunderstand the impact of a design on the basis of a short article in LTT (Local Transport Today). The overall concept indeed was inspired by the examples in The Netherlands, but the concept has been adapted to the local environment.

First you need to understand the current arrangement in Bedford. It is an English-style roundabout in the middle of a very Victorian town. The roundabout itself is relatively small but has a relatively wide unmarked circulating carriageway where traffic either circulates as single lanes or doubles up depending on flow. Vehicles circulate too fast, entries are too close and it is difficult to predict vehicle paths, so that other vehicles, including cyclists, find it difficult to safely enter the roundabout. There are several circulating vehicle-colliding-with-entry-vehicle collisions as a result.

Most cyclists are also following a different path from most motorised vehicles. So there are conflicts as cyclists move across the different paths of vehicles. And of course there are the typical entry – circulating collisions where vehicles enter the roundabout without noticing the circulating cyclist.

On top of this, there are high pedestrian flows. There are 2500 pedestrian crossing movements of the arms without any assistance apart from the central islands, with several accidents resulting as crossing pedestrians are hit mostly by exiting vehicles.

So with 32 injury accidents, including 8 serious, over 10 years, at a cost to society of nearly £2 million, there was an urgent need to do something. The roundabout lies on an important route to the town centre and station for cyclists with over 500 cyclists a day, but it is also on a key interurban and intra-urban route with 25,000 motorised vehicles a day.

The council’s first choice was single lane compact (Dutch-style) roundabout. This was modelled but was found not to cope with the vehicular capacity, because whereas compact roundabouts can cope with those kinds of flows, the junction flows at this roundabout are not balanced. This is where the turbo-roundabout concept came to our rescue.

Ignoring the type of cycle provision for the moment, the essence of a turbo-roundabout is that it can cope with higher flows than a compact roundabout (up to 35,000 vehicles per day) because of 2 lane entries and 2 lane circulation. Secondly the raised circulating lane dividers prevent straight-lining through the roundabout so reduce vehicular speed entering, circulating and exiting the roundabout, thereby creating a safety benefit. Finally the spiral lane marking through the roundabout reduces the conflict points as vehicles enter, circulate and exit.

Our estimate, based on the radius of curvature of vehicular paths, is that in Bedford this will reduce motorised vehicle speeds from current 25mph to around 10-15mph, approximating much more to cycling speeds. The lower vehicle speeds allow us to put Zebras on every arm which will further change the way motorists use the roundabout, becoming more aware of vulnerable road users, and further reduce entry and exit speeds.

So with motorised vehicles and pedestrians now catered for, that leaves what is the best solution for cyclists. First it should be understood, notwithstanding the Dutch style preponderance of cycle tracks, if vehicle speeds are low enough, it is safer for a cyclist to go round a roundabout in the primary position. Annular cycle tracks create additional conflict points. In primary position on the road, a cyclist is most visible and has least conflict points, especially so in a single lane compact roundabout or turbo-roundabout.

Secondly, you cannot legally create a non-signalised annular cycle track and a pedestrian crossing in the UK context. This is the significance of the TRL work. It is the first stage in seeking a change in Government regulations to allow this. So either pedestrians get priority at a Zebra or cyclists have priority using Give Way markings (not a feasible option in this context).

Thirdly and most importantly, a cycle track is neither the safe nor correct solution for the Bedford roundabout. The four roads that lead to the roundabout have very high flows of pedestrians with adequate footway widths, but certainly not shared path widths. Most cyclists approaching this junction are on-road, using the cycle lanes on the most congested link. Cyclists going through the roundabout in primary position, i.e. taking up the whole lane, will have no more conflict points than in a Dutch-style compact roundabout, and around a third of the conflict points under the existing design. A detailed analysis of the accidents suggests a 75% reduction in serious accidents and 40% reduction in slight accidents for all modes, including cyclists.

So as long as cyclists are happy to take primary position in front of traffic, they will be safer and get through the roundabout faster than in an annular design. The conflict is between perceptual safety and real safety. The big question is whether cyclists will feel safe in primary position in front of slow moving traffic.

This is where personal feelings often cloud professional judgements. An analysis of Bedford data is that there are 2 types of cyclists – “Quick” cyclists happy to share with traffic and “Quiet” cyclists who want to be segregated from traffic as far as possible. The current division of cyclists in Bedford is around 60% Quick cyclists and 40% Quiet cyclists (on the basis of a survey of station cyclists and an analysis of road usage). Currently at this junction, 350 are on-road and 200 off-road (a lot of those are child cyclists).

So for Quiet cyclists we are creating a cycle track which leads to the Zebras. They will be able to go round the roundabout using the Zebras, 2 of which have been widened to maximum of 4m (on the main cycle crossing flows) to create as safe a crossing environment as possible.

In summary, the roundabout should definitely be safer for all users – pedestrians, cyclists and motorised users, than the current layout. Secondly, unlike now, cyclists will have a legal choice, depending on their nature, whether to go through the roundabout on-road or off-road. Hopefully this will create a virtuous circle of happier cyclists, eventually leading to higher cycle usage. We are thankful that the DfT Cyclist Safety Grant has been awarded to allow us to trial this innovative design.”

Behavioural insights, better metrics and @rorysutherland

Practitioners using insights from behavioural sciences to improve service design is one thing, but we’re also now seeing the need for policy makers to be systematic about understanding behaviour. This has a number of dimensions. For example:

• it makes ethnographic observation more important than straightforward qualitative research. In short, you’re more likely to be able to influence behaviour if you gain insight into when and how it happens, than from focus group findings of people telling you what would make them change what they do.
• it makes experimentation and prototyping vital, because we’re not going to rely (are we?) on interventions based on a belief that people are ‘rational, economic’ beings. There’s a clear link here with design thinking.
• in designing interventions, it shifts the focus from ‘why’ to ‘what’: from ”why do people do this and why might they do what we want to encourage?” to “what it is that people actually do? what triggers it and what are the barriers to people doing what we want to encourage them to do?” I’ve

I should write about each of these, really. But not today. Today, I’m reflecting briefly on something Rory Sutherland said at an event last week that I hadn’t previously given much thought to. He suggested that we need new metrics in public services and public policy. He’s right of course, though there’s a strong case that we’ve got used to having so many metrics that we don’t take enough notice of the ones we do have.

The specific example I remember is about mobile coverage. In the early days of mobile networks, covering 90% of the population when your competitors only reached 75% was a way of differentiating one service from another and creating competitive advantage, in turn driving other companies to increase their coverage. From a public policy point of view, the main metric being used was in citizens’ (customers’) interests. But once every operator had 98% coverage, they began competing on price, and innovation focused on how to design and present specific packages of data/calls, etc. This hasn’t driven up service quality; why would it?

Rory suggests that, since then, the main differentiator between services (in practical experience, though not reflected in marketing) has been the reliability – or uninterruptedness – of service outside of city and town centres, especially in transit. And although we know this as users, we don’t have access to any way of comparing services on this. Competition on this would have driven up the standard of service – but there’s no metric for this. My take on this is that, well, we COULD invent a metric for this, if we were so inclined. And maybe Oftel could have made each network publish its indicator on this and include it in all publicity.

Now, how relevant is this to my main theme – that people need to be systematic about understanding behaviour? I appreciate that some people would agree with Rory’s point, and say that it’s simply about being more ‘customer focused’. Maybe so. But I’m pretty sure that those who saw ‘percentage coverage’ as a key metric weren’t trying to be anything other than customer centric. And I’m also pretty sure that if you’re being smart about understanding behaviour, you’re more likely to generate the sort of useful, clever metric Rory suggests because you’d be focusing on what people actually do.

Using behavioural insights: commissioners need to get sharp

I spent an afternoon last week at Information Is In The Eye Of The Beholder, an event organised by the Design Councils’ Behavioural Design Lab. Maybe I’m lazy, but the main thing I took from it was a confirmation of something that has been becoming very in my work: that practitioners who apply insights from behavioural sciences in the policy sphere and in public services (and there aren’t many of us) need to challenge our clients to be very clear about the behavioural outcomes they want.

This may seem an obvious point to make, but I don’t think it is. Why? First, there is no established market for public service commissioning of behavioural insights, so commissioners have little experience to draw on. Second, so much policy is designed on the false assumption that people are rational, economic beings that there are likely to be some false assumptions built in to any default view of behavioural goals.

Let me illustrate this.

Felicity Algate from the Cabinet Office’s Behavioural Insights Team showed a prototype smartphone app that would enable people to compare their energy bill with the putative bill from competing energy companies, and to switch supplier with one click. In terms of providing feedback, applying mere exposure effect, and reducing goal dilution (a real barrier to switching), it’s great. It’s a real “if only all public services could be like this” moment.

But it does raise plenty of policy issues. For one thing, what are the carbon implications (given that the UK has self-imposed carbon budgets to meet by law)? Well, first, let’s be clear that paying less for energy does not increase energy efficiency, which is improved by reducing the energy input required for a given output, not by making it cheaper! Plenty of switching campaigns make this mistake, which in policy or behavioural terms is a significant error (sorry Calderdale, I had to pick on someone). Second, switching suppliers to pay less could increase energy usage. Third, there is an (arguably underfunded) policy drive to help people reduce household energy bills be reducing usase, through the Green Deal, hence a real risk of lack of clarity on how to manage energy bills, with damaging behavioural outcomes.  (I should point out that, at the event, others raised risks around the potential behaviour of companies in a market operating in this way; I’m only thinking here about the behaviour policy-makers encourage in citizens). In summary, there’s a case to be made that this is great innovation using behavioural insights, to serve a policy goal that isn’t very smart behaviourally.

In contrast, Felicity also illustrated the use of text messages to encourage people to pay fines (already in the public domain). This is surely a good use of behavioural insights, bringing a public service into line with current knowledge, but also uncontestable in policy terms. Not only is justice being served, but ‘clients’ stand to save the £200 that would otherwise be added to their fine to pay bailiffs.

My point, in summary: taking behavioural sciences seriously means applying them to policy considerations, not merely applying them once the policy thinking is done. And for those of us working in the field, that means not always accepting a commissioner’s brief ‘as is’.

 

 

Will we do politics better if we understand behaviour?

If you’ve heard one of my talks recently, on public services using insights from behavioural sciences, you might have heard me say that I reckon this perspective can make some policy decisions easier to understand than conventional narratives, and that this understanding is becoming more mainstream. But, y’know, I used to work for a think tank, so I’m used to asserting things that I’m not entirely sure are true.

So it’s always useful when something happens that makes me think I might have a good point. And here’s what happened today.

I heard a speech by the Prime Minister today on immigration. I don’t make a habit of this; it was on in the background while I made a cuppa. The full transcript doesn’t seem to be available yet, so you’ll have to take my word that, early in the speech, before talking about measures on benefits and housing, he asserted that immigration to the UK had become too large, because we are a “soft touch”.

What the PM is doing is framing his following comments in a way that is likely to make me favour the measures he went on to announce. He does this by framing immigrants as freeloaders. Group co-operation having been so important to our success as a species and, hence, being such a strong driver of our instincts and emotions, this is significant. As a UK native, I’m being encouraged to identify the group being discussed (who are in fact, of course, a set of individuals) as a threat to our (and, therefore, my) resources. Loss aversion kicks in (I value things I might lose much more than I value things that aren’t mine yet but could be), and I’m primed to support measures against this group.

You might think I’m accusing the PM of dog whistling. But I think the behavioural analysis is less loaded politically because it clearly acknowledges our human nature and relies on this, rather than relying on your existing political view. If you understand some key insights from behavioural sciences, you could accept this analysis (and maybe think we could do this stuff better), whatever your politics. In the UK of the near future, I like to think, people would ask whether triggering loss aversion and perceptions of freeloaders is a useful way to frame a discussion about immigrants.

Even if my analysis is true, I can’t say for sure that it’s deliberate, but that doesn’t really matter, does it? If we can grow this understanding of our human instincts in an honest and open way, we’ll all be in a better position to make and discuss policy, and do politics better.

Audio: Can Public Policy Cope with Behavioural Sciences?

I recorded the talk I gave at the Political Innovation event at the end of September. But the sound quality isn’t great, so I didn’t do anything with it. Then, a couple of days ago, I listened to this newly-remastered recording of Joy Division playing ULU in 1980 and decided that, if I can enjoy JD through a bit of muffle, then why not make my talk available?

Here it is on Soundcloud. A bit of an experiment. Let me know if you think this works. There is a Soundcloud app for smartphones, so you can listen on the move if that’s your thing – or you can download it to your iPod.

I’m hoping this will appeal to anyone who is interested in both public services/policy and also behavioural insights. The final few minutes consists of me being pretty optimistic: I reckon that making the use of insights from behavioural sciences mainstream – and being open about it – can help create better designed services, a more effective relationship between citizen and state, and also, in the longer-term, a more mindful society.

Are some behavioural insights easier to stomach than others?

Every so often, when I’m running workshops on using insights from behavioural sciences, I’m struck by how the reaction to some effects differs from others. I’ve mentioned this before in relation to the use of smileys.

And I think I’ve noticed that, in public services and policy, some effects have been used more readily than others. The example of organ donation, with policy makers in England choosing to use exposure effect and those in Wales using defaults is a good example.

So I now have a theory: that there are some behavioural effects that can be understood and accepted even with the ‘rational-agent’ model. Of course people are more likely to do something if you make it the thing that happens unless they act! And of course people are more likely to do something if you ask them to do it! I can believe these two things without accepting that my instincts are perfectly adapted to an environment very unlike the one I live in.

I hadn’t really put two and two together when I did my talk last week for Political Innovation. I don’t think it would have changed my argument, but it might have made it more nuanced.

So, is there a set of insights from behavioural sciences that decision makers (and people generally) are more comfortable with, because they don’t fundamentally challenge our learned understanding of what people are like (that is, basically in control of our actions, and largely acting with purpose)? And a set that are more troubling, because they alert us to how driven we are by herd instincts and evolutionary imperatives? The latter set would include hyperbolic discounting, social proof and many others.

Am I on to something? Is this a new insight, or so obvious that no-one has bothered to point it out? Or does this sort of analysis already exist? Let me know.

Can Public Policy Cope With Behavioural Sciences?

Lord Krebs, incoming President of the British Science Association was reported last week as criticising government use of ‘nudges’ But it seems that this amounts merely to a reservation are that using behavioural insights shouldn’t be a “get out of jail card if the government wants to avoid tougher approaches like taxation and regulation”. Which shoots down in flames a case that precisely no practitioners are making in the first place.

In fact, making public policy choices informed by behavioural science is on the verge of being accepted into the mainstream – in principle at least. This is quite something for what was a trendy discussion topic only a handful of years ago.

In the long run, what will surely happen is that the main thread of the case for using behavioural insights (that we are “predictably irrational”, so policy and services need to reflect this) will change. I predict it will be turned on its head, and I’m going to do what I can to make sure it is. Why? Because our true nature, our loss aversion, our temporal discounting, the heuristics we use are, in fact, completely rational, given the way our species has evolved. We only appear to be irrational when viewed through the lens of economists and philosophers whose assumptions were never true. Sorry, but humans are not rational economic agents, operating in an economy heading for equilibrium (and credit is due to IPPR for bringing this understanding closer to the mainstream with its Complex New World report).

Putting this growing acceptance into practice, though, is being held back by a number of things. First, the number of practitioners (that is, people who help bodies apply behavioural insights) is small; second, and a chief cause of #1, there isn’t a clear market for commissioning behavioural work in public services where this is not well established; third (and, of course a chief cause of #2), it doesn’t ‘belong’ to any existing profession. So while behavioural insights have informed the work of designers and advertisers for decades, that’s just not true in public policy. So what’s becoming mainstream in principle is being held back in practice.

If I’m right, then over the next few years, we’ll have to face up to the real challenges of public policy catching up with the evidence base and making decisions based on what people are really like, not what Enlightenment thinkers and classical economists say that we’re like. These challenges should be much more substantial than the red herring debates to date (such as that around the House of Lords Select Committee report)

An event I’m doing with Political Innovation aims to highlight one such practical challenge, namely the tension between using insights from behavioural sciences and an approach that progressive voices, myself included, are rightly positive about: co-design. In participative democracy, participation and service design, co-design with citizens and users is surely a good thing. Is it not a way of cementing the move away from the ‘we know best’ style of decision-making which has rhetorically been a no-no for twenty years, but which we all privately admit still prevails?

Here’s the problem, though: Why would I, as a citizen, play a part in designing a service which aims to prompt certain behaviour on my part with cues which I consider irrational or childish? As a policy maker, what is my response to this: accept that the human inclination to think that we are much more in control of our actions than we really are trumps proven, scientific insight? The risk is that the use of behavioural insights becomes an extension of ‘we know best’ policy-making. And that’s what we’re trying to avoid, isn’t it?

I’m not sure I have the answers, but I’m pretty sure that the questions here are going to be important. Come along on 25 September if you can.

Is it childish to be influenced by smileys?

Have you ever wondered why smiley faces are so powerful? I ask because they’re significant in the world of behavioural economics and psychology.

Here’s one example you probably know already: the way that OPower reduces energy consumption in US communities by employing social proof in the way that energy use is communicated. The image here shows how it works in practice, with clear communication on household energy use in relation to neighbours reinforced by a ‘great/good/below average’ rating, illustrated with smiley faces as appropriate.

The key thing to note is the impact of the smileys and rating; without them, usage tends to converge on the mean, so high users reduce usage, and low users increase over time. The smileys stop this happening, saving lots of money and carbon, which is the point; the initiative is designed to reduce energy usage. (Click here to read a paper on this).

Why is this example important in the work I do? Because when I’m running a With The Grain workshop, introducing people to insights from behavioural sciences, prior to helping people use these insights to generate ideas on how they can encourage certain behaviours, I use the OPower example as a ‘jumping the shark’ moment. To most people, it seems so absurd that such a simple measure should have such an effect on our behaviour that it feels right to give participants the right to say that they aren’t convinced. I invite them to say so if they feel I am taking them down the same path that script writers took Happy Days viewers!

The objection (or at least surprise) that people have is generally that something so ‘childlike’ or ‘childish’ can have such an effect on adults.

An experiment at Yale (paper here) found that babies accepted a video of a ball striking disordered blocks and appearing to create order – but only if the ball had a ‘face’. This has provoked interest among those interested in why humans are disposed to believe in deities, but I think it’s also interesting for those of us trying to use insights from behavioural sciences in our work. It is surely less troubling to see the way that such an apparently simple thing can greatly influence our behaviour not as ‘childish’ but as being innate.

In my view, it’s easier to work with effects that we think influence us because of an innate pre-disposition, than it is to work with something that we see as childish, not least because the latter idea carries with it a sense of immaturity that should, over time, leave us.

It seems, humans aren’t like that; the evidence shows we are influenced as adults by smileys. We might as well embrace the fact, and work with the grain.

Getting Meta All The Time: a new type of behavioural insight?

This is worth a look. I know people have been using insights from behavioural economics and behavioural psychology in staff (or student) cafeteria – and then writing about it – for some time. But if you’ve not read a similar case study before, this is good.

6 Ways Google Hacks Its Cafeterias So Googlers Eat Healthier | Co.Design: business + innovation + design.

The one thing here that IS new to me, though, is the use of what’s called a ‘meta-nudge’. This might be more significant than it seems. Here’s why …

One objection to using behavioural insights in designing policy is that it’s somehow underhand. To some extent, this is a valid view, informed as it is by the use of behavioural effects by marketers, restauranteurs, phone companies and the like (check OFT’s 2010 report on pricing for examples), which seem designed essentially to ‘trick’ customers into parting with more cash than they otherwise might.

Well, informing staff in the canteen, as Google does, that if they use a bigger plate, they’re likely to eat more is not really applying a behavioural insight as such; instead, it’s explicitly telling people about the insight, and giving them the choice. So where we might usually use the ‘default effect’ by making only smaller plates available, here we are explaining how a larger plate primes a certain behaviour – and leaving people to make their own decision. Superficially, this may be similar to a sign by the lift and stairs stating that most people use the stairs (which utilises social proof to encourage people to walk, as referred to on p17 of this Cabinet Office report), but it’s quite distinct, really – it’s the equivalent of the stair sign saying “If we put a sign here telling you that most people use the stairs, then you are more likely to use them yourself”. And I don’t think that has been tried yet!

Exposed! How to encourage #behaviourchange when changing the default is a step too far

You might have noticed that England and Wales are taking different approaches to increasing organ donation rates. They’re both addressing the problem that the UK’s ‘opt-in’ system gives a much lower rate of donation than ‘opt-out’ systems in some other countries. You can guess which nations in the chart below use which approach! The difference in outcomes is stark.

In terms of behavioural effects, this is all about defaults. And the plan in Wales is to use this insight to change the default to presumed consent. In England, a different approach – just as legitimate – has been chosen: relying on Exposure Effect, rather than Default Effect. So, when you apply for a driving license, you will have to choose between being on the donor register, or not.

There won’t be a ‘don’t know’ option, which strengthens the effect, and we are sure to see donor rates increase.

So I was pleased to be reminded (by@alex_north) that Boots is helping the drive to improve rates, by asking people if they want to be added to the register when they apply for a store loyalty card. I’ve clipped their web page on this (right). Again, this is about employing Exposure Effect.

My reason for posting this is simply to flag up how straightforward it can be to use behavioural effects. It doesn’t require masses of academic input; it just requires a bit of literacy about behavioural effects.

Speaking of which, good to see Boots adding the clip below, using Emotional Engagement to encourage people to ‘do the right thing’.