218. Do Nudges Work? with Michael Hallsworth

Welcome to episode 218 of the Brainy Business Understanding the Psychology of why People Buy. In today's episode, I'm excited to welcome back Dr. Michael Hallsworth from the Behavioral Insights Team. Ready? Let's get started.

You are listening to the Brainy Business Podcast where we dig into the psychology of why people buy and help you incorporate behavioral economics into your business, making it more brain friendly. Now, here's your host, Melina Palmer.

Hello. Hello everyone. My name is Melina Palmer and I want to welcome you to the Brainy Business Podcast. In today's conversation, I am joined by Dr. Michael Hallsworth, managing Director for the Americas Division of the Behavioral insights team, or bit. Michael was on the show nearly two years ago in episode 125 when he was sharing about his book, which is aptly named Behavioral Insights. The reason he's here today is to talk about a debate that has been going on across the behavioral science community and beyond for the bulk of this year. And as with any field, I'm sure.

In pockets for much longer than that.

But this is the Brainy Businesses addition in entering into a conversation of the Do Nudges Work debate, Michael gives a great backstory in our conversation. So I'm not going to get into too much detail here before just jumping into the show, but this has has.

Been something I've been watching for many.

Months now and trying to decide if it made sense to talk about on the show and if so, what the best approach should be. It didn't feel like something I wanted to do a solo episode about, and a lot of what I saw being posted on the topic was a bit sensationalized for my taste.

This is the same reason I don't like watching the news.

The framing just feels so aggressive and when a balanced conversation is pretty much always possible, it's.

It just isn't for me.

So when I read a recent article that Michael wrote in Behavioral Scientist which laid out all the points in a very clear way and talked about the real question we should be asking as well as next steps for the field into the future, I knew he was the perfect guest to come on and speak about this. As I said, he does a great job summarizing everything here and I hope you get value from this conversation. It can always be hard to look at ourselves, whether it's the fields we're in or ourselves personally and embrace opportunities for improvement, but it's so necessary in order to grow and change and thrive. The field is built on solid science and we have an opportunity to do even more and better going forward, but I'm going to wait to share my thoughts until later on in our conversation and in the closing of the episode. For now, before we jump into the interview, I want to be sure you know that the links to those mentioned articles as well as past episodes and lots of other content are all waiting for you in the show notes for the episode, which you can find within the app you're listening to and@the brainybusiness.com, 218. Now let's jump right in. Dr. Michael Hallsworth, welcome back to the Brainy Business podcast.

Thanks very much. It's great to be talking with you again.

Yeah, as we were saying in kind of our pre interview recap, getting to say hello again, it's been almost a hundred weeks since you were on the show and I guess based on pre recording, potentially it's been a full hundred weeks since we actually talked to each other. So it's been an interesting two years, but we don't need to get into all that, right?

It's been an interesting two years. Let's just, let's just leave it there. Yeah?

Yes. Yeah. Everyone knows what's been going on there. So I asked you to come onto the show based on an article that you recently wrote and we'll jump into that for everyone though who isn't familiar, who hasn't heard your episode yet. Of course it's linked for them in the show notes.

But can you share a little bit.

About who you are and what your background is in behavioral science?

Sure. I'm the managing director of the Behavioural Insights Team for the Americas. The Paper Insights Team was created in 2010 as the first dedicated government unit looking at behavioural science, how it can be applied to policy. Often people refer to us as the Nudge unit. I've been with the team pretty much since its creation. I've published various studies in academic journals and also kind of various frameworks for policymakers to help them use behavioral science more effectively and improve what they do. So yeah, I've been involved in this field for around, I think 13 years now. Did a PhD in the field, wrote a book with Elspeth Kirkman as an introduction to applying behavioral science in practice.

And that book, when that was coming out, that was when you were on the show, we were talking about that book. So we can hear a little bit more about that in that other episode as well. And so with having you on now, like I said, there was an article that you recently wrote that came out in behavioral scientists on August 2nd here of 2022, it looks like when that was published and it is on the debate that's been around for a bit now, this about nudges and whether or not they work, can you provide some background on the discussion that's been happening and some of your perspective on the debate?

Absolutely. So many of your listeners will be familiar with the idea of nudge. Very famous book published in 2008 by Richard Thaler and Cass Sunstein. I won't go into the details, but essentially it's a way of applying findings from behavioral economics, psychology in practice to guide choices without mandating choices, to guide people in particular directions while retaining their freedom of choice. This idea has been very popular. Various people have applied nudging and practice over the last 14 years. And what's happening as the field of applied behavioral science, which is bigger than nudging alone, as that field has matured, people have started to run these kind of overview studies, these meta analyses, to say what are the overall effects of these kinds of interventions. And one of these meta analyses came out in January 2022 in the journal PNAS Proceedings, National Academy of Sciences, high profile journal. And this analysis said that nudges, and I'll get into the fact that they just took nudges alone later, that these nudges have an overall effect size of what we call D 0.43. And to explain that that's a fairly big effect in the article, I give the comparison that that's about the same as the effectiveness of, say, interventions to increase motor skills in children. So it's like a kind of a moderate effect size. Now also, that study said, you know, we've seen some evidence of publication bias in the literature, and publication bias is a problem that affects many disciplines in the scientific literature. It is not specific to behavioural science. What it basically means is that you may have in your overall body of studies more studies that are likely to be showing, say, a positive result to show that something happened, or there was a new finding that, that we didn't just see no effect, we saw some effect. And so what they conclude here is that we probably think we've got too many of these studies which are showing an effect in our body of studies. Right. And so the effects may be smaller than we're saying than this figure we gave. And then recently, and this is what kind of triggered the debate really, there were a few responses to this study and one of them went quite a lot further and said, actually, we think that if you account for publication bias properly, there is no overall effect from these nudge studies. And bear in mind that in both cases, we're talking about the overall effects. Like if you average out everything, what is the effect? And this got a lot of people interested because if you just read it at superficial level, it says, oh, there's no evidence for nudging. And as I mentioned, nudging has been really popular over the past 14 years. So a lot of people have been interested in it. You have a lot of listeners to your podcast. And so I think there was a bit of a tendency to say, oh, well, okay, well, maybe we want to try and sort of respond to that and puncture the bubble a bit and say, oh, it's been overhyped and so on. And so some of these articles that were based on this response took a fairly sweeping view of it and said, oh, there's no evidence for nudging. And oh, is this the end of the road and all this kind of thing? And what I'm basically saying is this is not a very sensible debate. And I'm not just saying this because I've worked in this area. It just doesn't make very much sense on its own terms. And I think we want to move away from this black and white for and against debate about nudges as a whole. And I can go and explain why that is, but I feel like I've just gone through a whole range of things there. It's pretty, you know, there's a back and forth debate and I'm really not sure it's the right one.

Yeah, no, and thank you for the backstory on it being able to provide some of that context. Of course, we'll link to the articles that you referenced in the show notes as well. So people can do their own research, which is a big piece of all of this that is very important. And the audience, assuming this isn't your first episode, you're familiar. I constantly am talking about how singular data points are not generalizable in other contexts. And that context, context, context, context is so key and people should be doing their own tests is something that I am preaching all the time, up and down. And so the. I've, I've had a similar thought as I mentioned to you in the debate, as I've been kind of watching the discussion happening there and I felt that your article really finely articulated some of the thoughts that I had had. And, and you have so much more background and knowledge in a lot of these other pieces of research and things that I loved the way that, that you showcased just the discussion that's happening. And like you said, it's not to say there shouldn't be any sort of debate or conversation or shift. It's not to say that the field needs to stay exactly as it is forever. There is, you know, there may be some growing pains, but this isn't necessarily the discussion that we should be having, is what I've gathered from some of what I've seen you talking about.

Yeah, And I think it's a more realistic discussion. I see this as an evolution in the conversation. You know, I spend a lot of time, I spent 12, 14 years applying this stuff in practice and I've seen that there are real world results from taking these approaches. They can make a difference. But I think what we need to be doing is being realistic about what we actually know here and the size of effects we're likely to expect. And when I saw the original study, I felt that it probably overstated the effect size. I think that other people said it seemed quite large. And maybe at the time I should have said something about that because what I'm trying to get at, it's not like I'm cheerleading and saying on the nudge side, we must support every study that shows big effects and say that the other studies opposing that are not right. I'm saying that we need to first understand what's a realistic effect size for some of these interventions and secondly, can we get a better understanding of context and how context affects results? Because we need to move beyond saying context is important and try to understand why it's important in what ways, how we can develop the skills to understand that, and how to develop skills to adapt things sensitively and appropriately to different contexts. And I think these are kind of underdeveloped actually in behavioural science because I think we've tended to privilege or tend to think of behavioral science as about knowing things like knowing findings and so on, and thought less about those skills. In terms of. It's about how do you put that intervention together in a way that really works. And so I can give you some examples jumping ahead here. You know, you can have an intervention that in a way is completely in line with the theory, with what we know about what seems to work. But it doesn't work because there's something about the way you translate it into concrete terms, into language, say that means it doesn't play out that way and you need to make some adaptations for the context, being sensitive to how people are likely to react that actually make it more likely to work. So, for example, I'VE seen in the past various studies that make claims that so and so such intervention doesn't have an effect. And I thought, well, that's kind of interesting, maybe that's not what I expected. And then you go and look at actually what the people did and you understand, well, if, well, of course it didn't have an effect because the intervention is actually quite poor. Like theoretically it's testing this thing. But you look at it, it's deeply unconvincing. Maybe it's got, maybe it's too long, maybe it's got lots of other elements in it that are muddying the waters. We're not paying enough attention to that kind of thing. So just going back to my statement about do these kinds of things work? So it's not just taking my word for, for it. One thing I point out in the article is that something that is a missing piece from this kind of study or this debate around oh, nudges, do or don't work is, well, if publication bias matters, can we look at some evidence that isn't so affected by publication bias? And we do have that. So the study came out earlier this year which looked at a whole range of studies that my organization in the US did and also the US Office of Valuation Sciences in the federal government. And what we did was we actually gave access to the whole range of studies that we did over the course of four years, 165 trials combined from the two organizations. Now, this wasn't the one, you know, we didn't select these for publication. This was everything. So publication bias is not a factor here. And what this analysis showed is if you look at all these studies, the projects produce an average improvement around 8% compared to the control. It's around one and a half percentage points on the baseline. So the authors conclude this is highly statistically significant. These are well powered studies. We can be really quite sure of these effects. So it's not the case, if you just take the headline from this latest study, that there's no evidence for nudging. We do have evidence for the real world effects which are not affected by publication bias. And those effects are smaller than the ones in the original study, but they're still meaningful. They're still meaningful because often these are things that can be implemented at scale. They're often low cost and therefore have a decent return for people working on practical issues. And moreover, in this period when we really were focused as an organization on low cost, rapid randomized trials with cities, actually mainly with cities in the U.S. so the kind of work we were doing was fairly circumscribed on these low cost, relatively modest interventions. Usually we're not testing these big structural changes in that period. That's not to say you can have bigger structural changes based on behavioral science, but we just weren't doing it then. So it's a kind of really, I'd say, reliable estimate of the kinds of things you can get from these more modest interventions, which we can also call nudges, if you like. However, the problem with all this is that that label of nudge, well, what really has happened is it's a very compelling word and concept and Thaler and Sons doing a great job explaining it. But. But pretty much everything in the field of applied behavioral science got attached to that label for a better or worse. And so when you actually start looking at some of these headline studies, they really bring together a lot of different kinds of interventions in different settings and so on and so on. And what this means is that we put a lot of things under this label of nudge and then we say that bunch of things does work or doesn't work. And so you're simplifying it twice over by the time you catch an answer. I'm not sure this answer of it not working means very much. Instead, what we see actually is there's a lot of variance in results. Human behavior is complex. You've made this point many times in your podcast. We can see that results vary by context and results vary by group within the same kind of sample. We're calling this heterogeneity. Many studies have shown really kind of rigorous ones, that there's a lot of heterogeneity because of all these factors that go into our behavior. And so even if you get something like replication studies, and the whole point of replication studies is that they are meant to try and keep things as similar as possible. They try and say, oh, okay, we're trying to replicate this and keep things as similar as the original study. Even then you can't get rid of the heterogeneity. You have this variance in effects that cannot be explained away. With that in mind, if we're taking that seriously, then thinking about these overall average treatment effects is really not the way to be going in the future. I know that we talk about them for good reason because sometimes we have to simplify. And I've been a policymaker myself, and sometimes you have to just look at the overall effect because you cannot tailor things anymore. But other people have been talking about idea of a heterogeneity revolution. So we stop thinking about the overall effect as the only thing that matters. We accept that effects vary because behavior is complex. That means that there are factors going into a result which are meaningful and which may mean that something seems to not work in a different setting. Rather than saying it doesn't replicate in this really binary way of. Did it replicate? No. Well, yes. We're saying, can we go and see what is driving these patterns of behavior in different settings and get to a point where we understand why it's happening? Not just like saying, oh, it didn't replicate. Okay, well, let's just move on. There might be more to it than that. So where does this take you? I mean, what it does say is that we should start thinking about tempering some of the. The claims that we're making here. And again, this comes back to the evolution of applied behavioral science. Over the past 10 years, there was a period of making the case for behavioral science in applied settings. Like, is any of the stuff real? Okay, supposing it's real. If we've done some studies, then, okay, I don't have very much time. What do I need to bear in mind? What are my takeaways from this? And so you get frameworks. I've written some of them which say general statements. And we have those out there. But we know, of course, that it's a bit more complicated than that and they do a job. But if you're careful, you end up back in this debate of like, oh, well, no, actually that doesn't work. And saying, let's go down a level of detail here if you really care about getting the best results. So we need to start tempering our claims here and say that when we're talking about, say, take social norms, we've often made statements saying people are influenced by what others do in the same situation. And I think that statement has value. I don't think we should be getting rid of that, but I think we should be building in some caveats to it, saying it really matters who the comparison group is. It really matters what kind of norm we're talking about. Is it about what people do or what people say? And those are not just like kind of interesting variations. They're kind of fundamental, really. And so instead of making this overall claims so clearly, instead I think we should be talking about some of these ideas a bit like behavioral scientists have talked about incentives. And to explain what I mean here, if you're an economist, obviously incentives are extremely important. And incentives are important. We know they're important. But certainly a lot of what behavioural science has done has said, well, yes, they're important, but we can make them more effective. There are some things that you can enhance by, for example, making group incentives, but there are also ways in which incentives backfire and you need to bear that in mind or there are ways you can, it really matters how you present incentives in the wrong context. They can really not work so well. So I think that's more of a template for how we should be talking about behavioral science findings as well. We don't need to oversimplify, we don't need to oversell. The results are there, it's just that there may be slightly smaller than some of the bigger claims and they vary in ways we don't fully understand. And I think the next phase is getting that understanding and taking it seriously so we can give, you know, better recommendations.

Yeah, absolutely. And thank you for the explanation. I think it's something that everyone, regardless of their background in the space can get an understanding of, you know, the complexity of what we're talking about here and the importance of it and how we can be thinking about not just this sort of debate or the, you know, headlines that we see. Like you were saying, it's, you know, the six o' clock news will sensationalize things or you know, in this, in a similar way, their headlines are often.

Not the full story.

And so it is important to dig a little deeper, ask our own questions and look to see what more is there. Whether that's from an article that comes out, whether that is from results that are there or not there. You know, just dig a little bit deeper, ask why and see what more you can find instead of just kind of accepting the first thing that pops into your brain when you see that.

Right, yeah, sure. And I think there are some concrete ways forward as well. It's not just like the way we communicate things and the way we think about things. There are a few things we can do. So one, and I won't go through all of them, but the three I think about are running multi site studies so we can explore these differences more systematically. One is getting better at understanding or identifying when results are likely to be dependent on a particular context, that kind of skill, and then developing the ability to adapt ideas to context better. And so the first one around multi site studies, it's partly about getting a greater variety of data. So we're not testing on just a really homogeneous group, but I think some really interesting ways forward around things like crowdsource studies and so what These are. These are where you take different research teams or different kind of organizations, and they're all asked to test a particular idea or hypothesis. And they don't know what each other is doing. And they'll come up with a different way of testing the same hypothesis. And then participants are randomly assigned to get one of those treatments or tests from the same group. And so what you can then do is you can see which of these hypotheses are being supported. Suppose you tested, like, five ideas. Like, one would be, say, people try less hard if they're given very small incentives compared to no incentives. Right. You can choose different things here. What kind of task are we talking about? What kind of incentive are we talking about? How often do they get the incentive? These are all choices that really could make quite a big difference. And then you might have another thing which is testing people believe a belief more strongly if someone argues against it. Just very simple one, okay, you have a collection of these different ideas, and then for each one, different teams come up with different ways of testing it. When you take the results together, you should be able to see which of these ideas are kind of more supported, which come out through the stress test of being subjected to different approaches through different teams trying to test it. And we see this does happen. There's a study that did this with 15,000 people, and they found that they had five ideas and two of them were supported. It was, generally speaking, good evidence that they held up under this stress testing. Three didn't. And I think that's a much better way of doing it systematically to test these different choices that people are making running studies like what kind of incentive. And those details are often lacking when we read about studies, but we know they're important. We don't go into that level of detail. But it's not just more and more detail. It's then trying to sum it up and say, well, okay, two of these five hypotheses did seem to come through the stress test. Now it's going to be hard. Doing that kind of work is difficult. I'm not just coming out here and saying, oh, it's easy. Just get a load of different organizations and do a coordinated. But there are projects like Behavior Change for Good, which have brought together a whole set of different institutions to run these kind of mega studies. It can be done. What we should be doing is doing it. So we systematically kind of testing and varying these different factors to get to a better place with what works and for whom. So then the second thing was around, can we Predict if we have result, how, like how context dependent is it likely to be? So there is some reason to be optimistic here. We have studies that ask people to say, how dependent do you think this result is on context? Like how context dependent do you think this finding is? And people. Actually, we found that. Sorry, I did not conduct the study. J. Van Bavel, I believe, conducted the study. They. They found that if you thought something was more context dependent, indeed those studies were less likely to replicate successfully. And what I mean by replicate successfully, from what I said earlier, what this probably means is there were other things going on that meant it looked like they didn't replicate in different setting. This language of replication is kind of a bit problematic when you have so much variance going on. And yeah, I just mentioned that study around crowdsourcing. Well, they also crowdsourced people's expectations and they found that people were able to anticipate the results of an intervention just by looking at the intervention themselves, like how big the sample was, what were the materials like? So they seem to be good at predicting what would actually work in a particular context. So what I'm trying to get at here is this kind of understanding is there in behavioral science, and we could use it and cultivate it to predict which of these interventions are likely to be broadly applicable and which are probably going to need some adaptation and put some warnings there saying, think carefully about this one. This one probably needs some adaptation and some additional resources. Well, what are the core elements and what are the things that. That probably can be adapted to different contexts? And I do think that behavioral science can learn a lot from implementation science here, which is an adjacent field which I think we don't pay enough attention to because it really looks at these factors of saying, what are the things that play out to make something work in practice? And then finally, this skill about adapting, it's not just about what you know, it's about how you match it to context to produce the result. It's not like we can just take a nudge off a shelf and open it and it works. And this is something I think that we already use. That is a skill that's at the heart of good applied behavioral science. Anyway, whenever you create an intervention, you're making choices, you're taking a concept and you're saying, how do I realize this in practice? Which word should be used in a particular setting? What time of day should something be offered? And we're just not really paying attention to those things. I think I'm generalizing. Some people are, obviously, but when we come to step back and see what do we learn about these things, we're not capturing those factors. The skill of adapting is quite like the skill of creating an intervention in the first place. And I think, you know, when you see it, when you see an intervention, you think that is, well, that's been well constructed, that can work. And ideally, of course, when the results show that as well. But I feel that our understanding of what goes into that skill and what makes that intervention come together is still not very well set out. And I think we know that as practitioners, but not doing very much about it.

Right.

So you mentioned that some people are doing a better job of that. Is there anywhere, and you had talked about, you know, implementation science. Is there maybe a way or a space to be looking at where we can be building that skill up? You know, is it just, you know, we're talking about the adaptation and that it's not documented well enough? It's. It's kind of getting lost in the shuffle or whatnot. So is there some sort of advice from what you've seen that people, you would recommend people should do as they're working on testing or anything else?

Again, you know, this is hard. I don't think it's just me saying this as well, though. Sure. There's a new book edited by Nina Mazar and Dil Soman called Behavioral Science in the Wild, where this is kind of a big theme point being that Behavioral Science in the Wild is kind of different from behavioural science in the laboratory when you're applying in different contexts. So I think part of the issue is that it doesn't really make a lot of sense for individual people or organizations running studies to provide this level of detail. And it's hard enough to prefer something to be published, let alone putting all the context in there. We're not incentivized to do that. So maybe there's something about expectations, there's something around a low effort way of capturing some of these things. And I do think there are initiatives and organizations that are beginning to do this and beginning to try and say what were some of the ingredients here that we're not normally capturing. So there is hope that that's happening, but it kind of requires someone with some funding to do the hard work and bridge between organizations, get that trust and then say, well, what did you do in this study? And then start to draw those patterns. And I think even within organizations, it can be really hard to start tracing the threads between studies and saying, okay, interesting. So we saw this thing here and we saw this thing here and maybe there is a connection. Should we just explore that a bit more? We do it sometimes. We had a collection of studies where we found that surprisingly, a really kind of formal letter was doing better from, say, some government institutions compared to an apparently more engaging, well designed, brightly colored communication. And we started digging into that bit and we thought there was actually something going on here that was quite interesting in terms of people's expectations of government, people's expectations of what seems to be a credible message versus one that's a bit suspicious. And so you kind of do that, but quite often it gets lost because you've then got to move on to the next study. And so you never build up that cumulative understanding. And I have to say, I don't think academics are really incentivized to build up that cumulative understanding either because instead what you focus on is the headline outcome and the contribution to theory. But that theory really is not very separate from the specific choices you made when creating your intervention. Even though we privilege the theory much more, I would say, in academia.

Yeah, thank you, thank you for those points. And so everyone knows too, like you had mentioned to me at the beginning, you know, you've been thinking about a lot of this for a long time and then had the opportunity with that article. And in addition, it's not just kind of off the top of your head thoughts. You've been working on a manifesto that is upcoming soon that everyone will be able to see which. And I got a little bit of a sneak peek, peek on. Can you share a little bit of what this manifesto is and what people will be able to get from that?

Yeah, sure. So what this tries to do is draw together, I would say, the things that have come across my radar, bit's radar over the past 12 years as things that we need to be doing more of. Ways we need to just like re think or reinterpret what behavioural science is, new ways of applying behavioural science, new opportunities and you know, responding to some of the criticisms that are out there, which I think we all have to take seriously and we can engage with and we should. And there are answers, there are proposals for addressing these. Some of the criticisms which we set out in this manifesto. So we have 10 proposals which are things around how we use our CTs, how we explore people's situation and settings, how we enable people to use behavioral science themselves, how we reflect on our own situation and maybe the perspectives and biases that we bring as behavioral scientists rather than just pretending with some objective Lens looking at people. We're not. We are. There is no view from nowhere as we say, what are we going to do about that? Take that seriously. Are we going to get better at predicting our results beforehand and then feeding back the findings to people so we get a better understanding of our own skills and our own abilities and whether say a finding is truly unexpected or not? I touch on how do we take all this variation in results seriously? Like we just discussed, what future does that have? And how can we use data science in a way that is supportive rather than exploitative? Because we talk about data science a lot, machine learning, tailoring, interventions. I think you run pretty quickly in some big ethical questions there about how you're going to use that, how you can use those new techniques. I think there's a positive vision and there's a more concerning vision. And then I think the overriding kind of question is how do we help behavioral science tackle some of those bigger questions that people say? Oh, you just tinker around the edges of the fundamental questions and there may be some truth in that, but there are kind of good reasons why that may have happened. And it's not all that behavioral science can do. So we also talk about in a very kind of crunchy and specific way, how do you deal with complex systems that are really all around us, like cities, like markets? How can behavioral science help a better understanding of the interactions in those systems and how the systems can be set up to get the best possible results for society? So I'm energized by that. I think there are lots of ways forward and that's also where I'm coming from in this debate of saying, oh, nudging. There's no evidence for it. This is the wrong debate. It's not about like, does it or doesn't it work for a particular label? We've put a load of stuff in. There are ways we can take behavioral science forward. We have some concrete proposals and yeah, the manifesto tries to get those out there, go into a lot of, you know, some detail because didn't want to just make some big statements and without doing the hard work below that. So we try and be quite specific about what you're going to do to make this a reality. So that's coming out. Yeah, pretty soon. Trying to chop it up into bite sized pieces because there's a lot in there. But I think it's really exciting and I hope it paints a hopeful vision for behavioral science to go into its next phase and really address some of the big issues and fulfill Its potential.

Yeah, absolutely.

Well.

And I am honored to be one of the limited few who have had a chance to read through it so far. And I think it was energizing to me in reading it. I. It made me excited about the next phase and the opportunities that exist.

So I think, at least with this.

Reader, it's doing what you say. So I'm. I'm very excited about that. And of course, we'll be linking to the BIT website, which I'm assuming it will show up there when it's out and.

And published when it's out.

Yeah, yeah, perfect. And. And thanks for giving just a little glimpse of that so that people can be on the lookout for it, because as we've said here, I think it's really valuable stuff, and I think it really will be helping to shape the field moving forward in a way that is building on what has been done so far. And as you said in your article and within that manifesto, really, the. It's not to say anything that has happened so far is bad or wrong. We're learning from what has happened and now we move forward.

Yeah, yeah, exactly. We are learning from what's happened. And, you know, it is a field that covers a lot of area, a lot of areas. Like, there's a lot going on and it's had a lot of impact in that process. You have to simplify messages necessarily, because otherwise you don't get traction. It just gets lost. So I think with all these things where you're trying to improve practice in the public sector, the private sector, wherever it is, what is that balance between the level of simplification to get traction versus the level of complexity where it's like. But okay, here's what's going on below the surface. And what I really hope we can do as a field is get to a good place in understanding that complexity as experts, really, and then we can. I just think there's a lot more that can be done. And it's not that by thinking about the way that results vary, that doesn't necessarily mean that you have to then just break things down into smaller and smaller pieces. And this is where I think it comes back to the idea of a system, because you need to understand what's going on, the variety of different things, and then how they interact and they interact. And by interacting, they create a bigger thing. That's how markets function. But you can only really understand a market if you understand the different parts of that market. And I think it's true with behavior as well. We need to Go from that kind of really high level. Here's the overall effect. Hold onto that for when it's important as it is and when you just have to look at the overall effect but then start building up that understanding of well, there's a lot more to it than that. And here are the ways those different pieces come together so we can get a grip on some of the bigger issues facing society.

Yeah, and looking for those spillovers too, the things that you weren't even thinking about that happened. What can you learn from that for the future? What happened over there that you weren't thinking about pulling this lever and what did that do?

Yeah, and there were some interesting tensions there because a lot of the, in the replication crisis there's been a big move towards pre registration of outcomes. But isn't that. So I understand entirely why that's happened but it shouldn't become something that constricts inquiry because you may not be able to predict in advance all the potential outcomes, particularly if you're intervening in a complex environment with lots of things going on. So we just need to make sure that that very good impulse doesn't take us away from saying, well okay, but actually our prior understanding wasn't complete and actually something else is going on that we need to go and look at here. So I think it can be done. But I think these are the kind of next level debates and tensions that we need to be thinking about.

Wonderful.

Well, thank you again for coming on the show to talk about this very complex topic. Of course. And like I mentioned, we'll be linking to the Behavioral Insights Team website. Where else? For everyone who is now wanting to.

Learn more, to connect with you, to.

Ask questions, whatever that is, where should they go to be able to hear more from you?

So an easy way if you want to summarize to go to my website michaelholsworth.com because you can also contact me through the form there. That might be helpful to some people if they want to get in touch and discuss these things. But yeah, those are the main ways.

Perfect. Well we will link to that in the show notes as well and your.

Your book and the articles that we've discussed.

So much, great stuff. So yeah, thank you again for joining and having this discussion with me today.

I really appreciate it.

Thanks so much for having me.

Thank you again to Dr. Michael Hallsworth for joining me on the show today. What got your brain buzzing in today's conversation for me, I'm so excited about the next steps and what the field of behavioral economics and behavioral science will be able to do and achieve in its next phase of life. Yes, it is hard to have people question you or your past work, whether it's you personally or the field you work in. No matter what, it's normal to have an immediate visceral reaction to any criticism, but it's also important to take a step back and a deep breath and consider what can be learned from this experience. I did an episode last year on how to deal with and respond to negative reviews, which is feeling very related to this topic.

So you know it's definitely linked for.

You in the show notes. Yes, there are trolls out there, but there are less of them than you think. And there's typically a learning lesson or opportunity to grow in any criticism, critique or question. You just have to dig a little deeper, take that step back and be open to learning a little bit more. The summary of what you heard from Michael is that yes, nudges do work. And as I've always said on this show, nothing is perfectly generalizable, so context is always important. Just because something worked in a lab test or for another company doesn't mean it will apply across every situation, and that is across any field, not just the behavioral sciences. We need to look at the intent behind the question, questions being studied, and look for what that might mean in our particular context. Is there anything that needs to be changed as we go about applying this information to our particular situation? What confounding factors should be on our radar? What should be part of our testing? And yes, again, this is why ongoing.

Testing is so critical.

This is a field in its toddlerhood. Right now I think we're beyond infancy, definitely, but likely not far full on into adolescence yet. And maybe it's just because I have.

My own toddler here at home, but.

There'S so much ahead of us and so many amazing things to be done. Michael brought up some really cool opportunities in today's discussion, and there are many more in his fantastic manifesto, which I highly recommend you reading when it comes out. I'll be promoting it around for sure and I'll go back and add it.

Into the show notes for this episode for sure, so don't worry about that.

This idea of looking into the possibilities of more complex systems and being able to be even better at predicting what will happen when and why is fascinating and something I look forward to being part of researching and reading about into the future. I hope you are energized by this conversation to look within your own work, whether you're within the behavioral sciences or not and reflect on any less than stellar feedback or questions you may have gotten over time. Yes, they can be hard to hear, but they are so valuable when you take the time to thoughtfully learn from them. As a reminder before we close out the episode, as always, there's a lot of great additional reading and listening waiting for you inside the show. Notes for the episode Related books like Michael's, articles also like Michael's and many.

Others, as well as past episodes, also.

Like Michael's and many others. It's all waiting for you within the app you're listening to and at the brainy business.com 218 and thank you so much again to Dr. Michael Hallsworth for joining me on the show today. It was a delight to chat with and learn from you. Join me next week to learn from Nancy Harhut about her new book, Using Behavioral Science in Marketing. It's going to be a lot of fun. You won't want to miss it. Until then, thanks again for listening and learning with me, and remember to be thoughtful.

Thank you for listening to the Brainy Business podcast. Molina offers virtual strategy sessions, workshops and other services, but to help businesses be more brain friendly. For more free resources, visit thebrainybusiness.com.