Last week I celebrated having uploaded 40 interviews with prominent people in behavioural science. Quite the achievement I'd say! I hope you continue to read and enjoy these articles, as it means I can keep writing them! All interviewees get asked seven questions of which one question is: " With all your experience, what skills would you say are needed to be a behavioural scientist? Are there any recommendations you would make?" Let's see what the experts think about the skillset you need!
Hard Skills: Experimentation Unsurprisingly, when people think of skills that are needed to be a behavioural scientist, the first things that come to mind centre around the hard skills. This is no different for quite a few of the experts themselves, as about two thirds mentioned that having this type of technical skill or technical knowledge is a great idea, and quite frankly a necessity. Neela Saldanha argues that you need to learn the ins and outs of running experiments, and learn how different fields (economists vs. psychologists) set up their experiments and data collection differently She emphasizes the importance of running experiments, or replicating famous ones (and doing so correctly!). Colin Camerer also argues that “you really should be able to do lab (and online) experiments, know about quasi-experimental designs (IV, diff-in-diff, regression discontinuity) and know some machine learning.” Well, it can’t get much clearer than that can it? He also mentions that learning methods is like “packing for a long, long trip to a place where there are no stores in case you forgot to pack anything. Fill that backpack with methods.” Yes sir! Matt Johnson also argues that it is important to develop strong experimentation skills. But he argues it from quite a different angle: “Specific applications, especially in novel scenarios, should be tested. It’s important to understand the general principles of course, but in the real world, things are very messy. Hypotheses should be informed by the existing literature but whenever possible fresh data should be brought to bear on it.” Matt is keeping his eye on the future! Kelly Peters emphasizes the need for strong experimental skills: “I believe one of the most important skills is the ability to design experiments under very complex conditions. There’s so much passion for behavioral insights, but the insights continue to emerge and evolve and it is important to recognize that the insights are only valid in very specific circumstances. Having knowledge of the litany of BE tactics is not adequate in the complexity of the real world. The hard part of experiments though is having the discipline and patience to wait for a signal in the noise of the real world.”
Hard Skills: Coding Gordon DA Brown first mentions the skill that has helped him most: “the ability to develop, or at least understand, formal models is important.” Gordon isn’t the only one mentioning the importance of having an understanding of formal models. Several other experts hint at this by mentioning the need to be able to write and understand code. Dan Goldstein recommend taking all the stats and data science courses you can when in school. He argues that it is much harder to learn this once “out in the world”, mostly due to a lack of time or support structure. Paul Adams argues that you need to be skilled in data science as there are lots of applications where data science can help us understand current decision making patterns. But also because it seems to me that there will be an increasing volume of data that scientists of all kinds can and will use to understand human behaviour.” Dan P. Egan corroborates this advice by also putting the emphasis on coding. He soothes those that are running away screaming: “you don’t have to become a full-fledged software engineer. It helps a lot to know how code works, databases etc, and to have a little bit of experience so you can work with those people. It is also incredibly useful for running analysis; sometimes to show that a potential change would have value, sometimes to analyse the results of an experiment. Being able to code has unlocked many opportunities for me.” So, don’t panic yet. The reason I’m telling you not to panic yet is that coding isn’t the only skill relevant to becoming a good behavioural scientist, and most people have actually mentioned it in a list of skills, rather than having mentioned it as “the” skill. Joshua D Greene puts it best: “I think it's increasingly important to have strong technical skills--statistics, programming. As our field progresses with computational modelling, "big data", and more advance imaging techniques it demands more and more of this. But there is no substitute for good-old-fashioned philosophical exploration--for reading widely and thinking broadly. Scientists tend to emphasize (and too often fetishize) the methods they have for answering questions. But it's no less important to find good questions.”
Open-mindedness & Curiosity To calm those now rocking back and forth in a corner because they are not that confident in their current and/or future coding skills, like I mentioned before, this wasn’t the skill most frequently mentioned. The skills that were most frequently mentioned were open-mindedness and curiosity. I’m lumping them together as many of the experts used different definitions of these two terms, and they had quite a bit of overlap. Looking at open-mindedness first, we see that many experts use this term as a way of being open-minded to all types of research, quoting Greg B Davies: “It’s a field that requires constant juggling between disciplines, and there is always something that other fields can teach you, so willingness to step outside of your zone of comfort is important.” Nick Powdthavee , Andrew Oswald and Ralph Hertwig seem to maintain the same definition as well. Ralph: “I think the major skill is openness, by which I mean both openness to ideas across disciplines and openness to people. On the one hand, I think we should try to benefit as much as possible from what other fields are working on and thinking. On the other hand, it’s my impression that new ideas tend to come from interactions with others. So I’d recommend the cultivation of intellectual and social openness.” Silja Voolma uses a similar approach to open-mindedness: “For any upcoming behavioural scientists, I’d recommend being as interdisciplinary as possible, the most impact in behavioural science comes from integrating multiple streams of research and perspectives and translating this into engaging and actionable communication for people.” Christian Hunt relates openness to the origin of the field: I think there’s a reason the phrase “Accidental Behavioural Scientist” exists. So many of us working in this field fell into it by chance. I’d encourage people that are interested in it, to be open-minded and explore what is, after all, an emerging science. There isn’t a fixed route in, and I suspect that everyone has their own unique story.” Koen Smets, the ultimate accidential behavioural economist provides a similar perspective: “behavioural science is the ultimate science of humanity. It is about what we do, what we feel, what we want, how we make tradeoffs, and so on. So a behaviouralist needs to be curious and open to everything that helps further that understanding. Inspiration can come (and for me certainly does come) from anywhere.” Magda Osman would even go as far as to state that “there aren’t any skills that are specific to being a psychologist that I don’t see applicable more generally. What I can say is, which has helped me, is that I have developed a better understanding of my area of interest by deliberately talking, and eventually working with those outside of my discipline. Contextualising paradigms of research, research questions and debates, theory development, and the accumulation of an evidence base in one’s own field by looking at how other research communities work in other disciplines has been invaluable. So, the wider the horizons by which I was willing to explore beyond my own, the better equipped I was to understand and defend (or else critique) my own field of interest. The take home being, be brave, challenge what you know and trust.” Wise words.
Looking into curiosity, we see quite a few experts heralding it as the most important, if not only skill that is necessary to be a good behavioural scientist. Rory Sutherland quotes this skill as having no substitute. He quotes how David Ogilvy described one of the hallmarks of a good copywriter as being “an extensive browser in all kinds of fields” and the same applies to this field. Many of the best practitioners have weirdly disparate backgrounds. Greg B Davies, Nick Chater, Graham Loomes, Cass Sunstein, Kelly Shortridge and Tim Houlihan all agree. Tim even states that “curiosity is essential to being a scientist and is the hallmark of every successful researcher I’ve met. Our world’s greatest findings come from researchers whose thirst for understanding is never quenched.” Tim also quotes the “Loewenstein approach to research” : he finds things in himself that on a purely rational basis don’t make sense and turns them into research projects.
Bob Sugden takes curiosity and applies it slightly differently: “You need to be curious about how people really behave and to follow the evidence wherever it leads.”
Understanding the Human If by this stage you’re getting quite surprised that there has been almost no mention of actually understanding people, and their behaviour, don’t worry, that section is coming now. I’d also like to emphasize that I didn’t order these sections in terms of perceived importance, my writing process is random at best. I don’t personally value any of these skills higher than the others. I don’t think understanding psychology is ranked lower/higher than understanding economics, data science or statistics. As Daniel Read has put it: “You need to be at least mediocre as an economist, a psychologist, and a statistician, and for one of these you must be outstanding.” So if you’re a great psychologist by actually understanding human behaviour, half the battle is already won! Daniel isn’t the only one mentioning the need for a deeper understanding of psychological insights, or behaviour in general. Neela Saldanha argues that “you should know the theories, and not just from reading the books, but by going to the source papers. Take classes in social & cognitive psychology if you can in addition to economics classes.” The importance of the source material, and being able to read scientific papers is also mentioned by Michael Hallsworth: “the story that is told up front in academic articles is not always identical to the one that comes through on closer examination. And if you are being judged on results, you need to know what is really going on.” Continuing on the practitioner side, Sam Tatam argues that “understanding the foundations of human psychology, including its evolutionary origins, is vital. Insight into human biology and evolutionary psychology provides a consistent, unchanging reference to better understand the ‘real why’ of behaviour.” Going back to academia, Dilip Soman reiterates the definition of behavioural science and lets it guide his most important skill: “since our science is a science about humans and not about econs, the ability to keep an eye out on what it means to be human is definitely an important skill. I think there are a few other pillars to do this field well.”
Collaboration and Communication Most research papers or behavioural interventions are not run by one person. “No man is an island,” after all. So it is not surprising that collaborations with others, or at the least the skill of communicating your ideas and findings to others is heralded as a great skill to have. Chris Starmer contributes his success to it: “The bulk of my research output is the product of work within collaborative teams where team members typically bring different mixtures of complementary skills in experimental design, theory, data analysis and so on. So, I put lots of weight on working in an environment where there are people who I can collaborate with well.” George Loewenstein also mentions the skill of collaborating, or rather, being able to pick good collaborators: “To me the key to success as a behavioral economist is to pick collaborators who have skills that complement one’s own. It certainly is important to develop skills in oneself; but it is impossible to be good at everything. But, picking good collaborators is a skill that every behavioral economist needs to have a successful, enjoyable career. I have had the good fortune to work with an astoundingly wonderful range of students and colleagues.” Moving from academia into the side of the practice of behavioural science, Dan Bennett also mentions the value of collaboration, but most importantly the skill of being able to communicate articulately with those from different fields: “As a practitioner I think it’s quite rare you’ll be in a situation where you won’t now be working with others outside of your skill-base to achieve your desired outcomes. So a core requirement of a practitioner is to be able to translate behavioural science to those outside of the field … and also be good at working people that don’t think like you!” Several other experts have mentioned the importance of being able to “sell your ideas” to others, which heavily relies on communication and the ability to “translate your findings from academia to the mainstream.”
Design A skill which got mentioned quite frequently, to my surprise (but that just shows how in tune with the field I am…) was the skill of design. Paul Adams argues that having this skill will help us “understand user experiences and design techniques to prototype interventions before going into a full blown field experiment.” He also mentions that this is already happening, but that more could be done. Kristen Berman also thinks knowing design is very helpful: “Most things and details matter, designers usually try to control those very details. If you can get good at sketching or mapping things out those are very good skills to have and will put you in a very good position to be in.” Dan P. Egan argues a similar point: “Being able to design prototypes of UI’s, interactions etc. You can do this in PowerPoint, or use designer software like Figma, Sketch and Principle. A dynamic picture (gif) is worth a million words.” Jez Groom argues the need for design as a skill in behavioural science, as he thinks it’s currently lacking: “A lot of psychologists and economists I meet are great scientists but don't have the skills to bring their ideas to life. At Cowry, we place a lot of emphasis on Behavioural Design and nurture this within all of our people. A recommendation would be for Behavioural Scientists to often take stock of their scientifically robust ideas and reflect on how dull they are to people in the wider world and prompt themselves to make contact and work with a craftsperson skilled in Design.” Guy Champniss also argues for more, or at least better application of design to behavioural science: “We need to become more attuned to design, because I believe applied behavioural science has so much to give at a foundational level as we (re) design products, services, industries, sectors and cities for the better. Dan Ariely said it well at the Behavioural Exchange this year – ‘[behavioural science] can do a lot with small tweaks. But it can do much better by developing technologies that are built from scratch on the principles of [behavioural science].’ The opportunity is there, but we have to better at earning our right to be involved at the beginning, and that requires a host of skills around a good understanding of the discipline. It seems that design is mainly quoted by practitioners as an important skill. However, I think here it’s the terminology that is rather deceiving. For academics, to be good at design is a core part of the experimental method. So both sides of the “behavioural science coin” argue for this skill, they just refer to it differently.
Skepticism Where would we be without a healthy dose of skepticism? At least, that’s what I always ask myself, but I’m glad to notice some experts find this skill invaluable as well. Nick Chater mentions that we should continue to be sceptic about existing accounts. Kelly Shortridge approaches scepticism from a similar angle, also mentioning curiosity and the willingness to be wrong : “Curiosity will naturally lead you to challenge existing assumptions and the status quo. A willingness to be wrong ensures that you’re constantly seeking to understand reality as it is, not how you currently conceive it to be or wish it to be – that you’re willing to test your hypotheses and accept whatever outcome results. Just as much as you should challenge existing assumptions in an industry or problem space, you should also accept when your own assumptions are successfully challenged.” Similarly, Aline Holzwarth defends the need for skepticism: “I think skepticism is one of the more underrated qualities in people. I find myself coming across more and more examples of people blindly applying behavioral science findings without experimenting or even measuring the impact of an intervention, and this makes me worry about the long-term future of behavioral science. We can’t forget that the beauty of behavioral science is in its methods far more than in its findings, and that as contexts and implementations vary, so do our results; we can’t always generalize from one situation to another. And the more skeptical we are, the more I think we will stick to the scientific method rather than faulty intuitions.” Skepticism shouldn’t just be applied to our theoretical base. Evelyn Gosnell recommends looking at behavioural scientific applications and asking yourself whether it is well executed; whether you would have done it differently. And what would you have done differently? Would you have used a different principle, or just changed the execution of what they did?
Overall there are many different skills that get recommended by many different people. Don’t think that as you move into the field of behavioural science that you’ll need all of these, or even need to excel at all of these. This article is predominantly meant to showcase how people from very different backgrounds have approached behavioural science, and what they have learnt from their own path. I’m going to end this piece on two pieces of advice. The first is by Ganna Pogrebna, who states and exemplifies that passion is the most important driver of good research. “If I were an early career researcher in behavioural science, I would not worry too much about skills. I think the most important thing is to find an interesting problem you are genuinely passionate about, then use your imagination to develop a vision about solving this problem. Your passion will lead you to people who will help you acquire those skills or who will collaborate with you to solve your problem. Last summer a student came to me wanting to work on a problem in urban analytics. She previously studied history and had no knowledge of any statistical packages and had not written a line of program in her life. However, she was very passionate about the problem she wanted to solve. So, in 3 months she learned Python and found people in UCL and MIT who helped her get the right programming and data analytics skills to solve her problem. She now runs a very successful AI company in the Silicon Valley.” The second piece of advice comes from Michael Hallsworth: “More prosaically, you’ve got to be willing to put the hours in. Thinking up ideas for new interventions is the easy part. The real work is making sure you turn up at an intervention site every day, or investing time to understand the concerns of a sceptical collaborator, or working through the snags with an IT provider. That’s the stuff which makes or breaks your work.”
*It is possible that when counting the names and responses not all 40 names are mentioned, or mentioned equally. A lot of responses had a lot of overlap, and as such got lumped together. If I hadn’t done this, this article would have been even longer!!!
If you want to make sure you know the ins and outs of every single interview, please be my guest, you can find all the interviews here.
Comments