Do High-Potential Employee Programs Work?

Despite its massive impact on a global scale, Twitter is a relatively small company. With five thousand employees, Twitter has to find the right potential leaders in their organization to invest in and promote. If Twitter chooses 5% of its workers to label as “high-potential”, they can spend millions of dollars on this set of 250 future leaders. Let’s take a look at what might happen inside a company like Twitter when putting its high potential program into place:

Sarah and Jennifer are two equally good employees managed by the same boss, Alex. One morning in June, after their performance reviews, Alex finds out Twitter labeled Sarah as a high-potential employee. Months later, the two women learn how their performance changed since their June review. Sarah’s performance skyrocketed, while Jennifer is doing just as well as before. Nothing changed around them except their boss knew about the high-potential label.

What happened?

High-potential programs in companies are built with good intentions. Firms wish to find talented people to invest in so they can put these employees on a fast track to leadership roles. These programs have a lot of power. What if they are simply creating self-fulfilling prophecies?

The Pygmalion effect

The Pygmalion effect describes a type of self-fulfilling prophecy that happens when other people’s expectations of us lead us to perform better. This self-fulfilling prophecy happens frequently in the workplace, but it was discovered first in elementary schools.1 In 1968, researchers told primary school teachers that some kids were “late bloomers” and others were not. Even though kids in both groups started at the same IQ level, the late bloomers blossomed.


The AI Governance Challenge

Here’s the catch: The researchers picked the late bloomers at random — they had no special qualities or high potential compared to the rest of the students. But the teachers’ belief in the late bloomer group improved their intelligence. By the sheer force of high expectations, a randomly-selected group made the researchers’ prophecy come true.

Simply naming someone as “high potential” changes their performance, by changing others’ expectations of them

Soon enough, studies showed the same Pygmalion effect with therapists, nurses, and managers.2 Clients in treatment for alcohol abuse were more likely to succeed if they were labeled “motivated” compared to others who were “unmotivated” — even if the label wasn’t true.3 Nurses and aides treated nursing home residents differently if experts said they would progress faster in their rehab compared to those labeled as “slower progressors”.4 These fast-healing residents had fewer hospital admissions and had fewer depressive symptoms, despite the labels meaning nothing at the start.

How does this work? For the most part, it occurs under our conscious awareness. Teachers, nurses, therapists, and managers don’t even remember treating one group differently than the other. Despite this, professionals of all kinds act differently towards people who have higher potential. Teachers give high-potential students more challenging work, more constructive feedback, and act friendlier to them.5 Leaders act differently toward high-potential teams by showing them more enthusiasm, encouragement, coaching, and confidence.6

Our expectations of others are below our conscious awareness, but they change how people perform.

Leaders influence their employees’ actions directly with differential treatment. They also create a self-fulfilling prophecy by changing workers’ expectations about themselves. When researchers take employees’ self-expectations into account, the Pygmalion effect goes from explaining 32% of the improved performance to only 6%.6 This means high-potential labels can raise leaders’ expectations, which in turn raise workers’ expectations, which improves their performance. Sometimes this process is even shorter: When companies tell employees they are on a high potential track, this works to increase the employees’ own expectations. This means the high-potential label is enough to improve any randomly selected employee’s performance. Below is a visual explanation of Sarah and Jennifer from our story above.

Telling employees or their managers about their high-potential status raises employees’ self-expectations

Now that we know high-potential programs can create a self-fulfilling prophecy through the Pygmalion effect, what can we do about this? Companies can take three big steps to strengthen their high-potential system:

1. Identify the right people by making better decisions about who should be labeled as high-potential

Any program in your organization is only as good as the data you put into it. If your methods for choosing talented workers is flawed, then your program will be flawed in the same ways. This means you need more accurate, relevant data on the skills and success of applicants during your hiring process and from your performance management system. For example, performance reviews are influenced by how confident and busy an employee seems. These aren’t good measures of potential or performance. Instead, measure performance through team members’ evaluations and objective progress towards employees’ goals.

If you choose to create (or improve) your high-potential process, focus on the quality of your decisions. After all, people who look like leaders are still seen as having more leadership potential than others who don’t fit the stereotype.7

2. Question your assumptions about what it means to be talented.

Is talent something that only some people have? Is talent innate or can it be developed? Your company’s views on these two questions shape the way they might approach high-potential programs, as their goal is to find and invest in talented employees. Below are the four ways companies think about talent and their implications for development programs.8

This talent matrix shows the underlying assumptions that come from believing talent is exclusive vs inclusive and static vs dynamic.8 If your company follows the static and exclusive approach, you can design a limited high-potential program that focuses on keeping these employees happy but doesn’t try to change their performance. Instead, a company following the dynamic and inclusive approach would invest in all its employees with customized programs that match their interests and career path. When developing a talent program, companies should choose whether they will train everyone, a few select people, or none at all.

3. Consider who needs to know about the high-potential label

So far, we can see how high expectations help employees by increasing their motivation and performance. Yet high expectations can backfire or have other unintended consequences. That’s why you should match your talent development approach in section number 2 with the transparency level that works for your company’s culture. If your company has a competitive, cut-throat culture, sharing employees’ high-potential status with each other could backfire. As well, if you have an exclusive approach to talent, transparency might hurt your overlooked employees even more.

High-potential programs can shape companies’ and employees’ futures. They are risky because the large investment in workers can become a self-fulfilling prophecy if programs are not designed well. This raises questions about how useful these expensive programs are. Why are employees spending so much time in leadership training academies, if telling their boss about their high-potential label works just as well, and for free? If anyone can rise to our expectations, it’s time to question our beliefs about who has talent and who doesn’t.

Can’t Say No To Promotional Offers?

Six months into this bizarre year — and after four months of living in an actual pandemic — we are beyond the point of debating whether or not the COVID-19 crisis will change human behavior. Yes, our behaviors have changed substantially, and no, not everything is a permanent change. But experts have come to the consensus that one behavior has seen a definite change: online shopping. As per the Adobe Analytics Digital Economy Index (June 2020), US consumers spent $73.2B in online spend in June alone, up a whopping 76.2% from last year.1

As someone who has spent nearly every penny I’ve saved from transportation on online shopping, I don’t disagree with the numbers. However, as an eternal optimist, I must add in the same breath, the experience of online shopping during a pandemic has been an opportunity for the behavioral scientist in me to learn a little more about the weird attraction we have towards “discounts”. You might argue I could have learned this by reading a few papers, but where’s the practical experience in that?

Sale! Sale! Sale!

Let’s start our journey into the world of discounts with the very anti-climactic story of JC Penney. For the longest time, JC Penney, a famous American retailer, gave out discounts like candy.2 Of course, the retailer used the age-old trick of increasing the prices of the items and then discounting them to create a sense of a bargain. But, people loved shopping there because everything was at a ‘low price’. 

All was fine, until the day Ron Johnson joined as the CEO of the company in 2012. He decided the company must be fair and honest. He introduced the no-coupons, no-discounting “fair and square” pricing, removing all the lovely discount tags, instead opting to show the actual price of the product. That’s where the anti-climax happens. Instead of being grateful that a retailer was being honest with them, customers hated it. What’s the fun in shopping if you can’t hunt for a bargain? With a dramatic fall in sales, all the company could do to save itself was fire Ron Johnson and bring back the discounts. That’s right — they increased their prices and put them on discount again, and customers flocked back.

So, what subjective gain do people get from a discount that they don’t get from a low price? Turns out, it’s all in our brains!

Bargain hunters

Here’s a little thought experiment. Which of these would you prefer?

Case 1: a 10% discount on an item causing a reduction of $5

Case 2: a 5% discount on an item causing a reduction of $5

Rationally speaking, it should not matter, since the quantitative discount received is the same. But we know it matters, right? A 10% discount just sounds so much better.3 That’s what Kahneman and Tversky found in their 1984 study about choices, values, and frames. What does this tell us? 

It points us to the idea that the discount provides an additional utility, which is not accounted for in the traditional utility. Traditional utility acquired from material consequences of exchange (for instance, giving money to get a product) is called acquisition utility. The utility derived from the psychological aspects of a transaction is called transaction utility — that’s what makes a discount special. 

If you now compare case 1 and case 2, the acquisition utility is the same $5 discount, but what makes case 1 preferable is the transaction utility derived from the term “10% discount”. 

So, what makes a discount special?

That’s the obvious next question. What are the psychological factors that drive a discount’s attraction and it’s transaction utility? In this article, I use the theory of causal attribution and its components to explain different aspects of discounts and how they are presented.

Research in social psychology has identified 3 dimensions of causal attributions: locus of control, controllability, and stability.4

1. Locus of control — i.e. whether one attributes the discount to internal factors or external factors

When a person attributes a discount to internal factors, such as their own effort and skill, the satisfaction from the discount is higher. Also called the “smart shopper hypothesis”, this refers to the act of patting ourselves on the backs for earning a discount. In other words, a discount feels better to consumers when they view themselves as responsible for having obtained the discount.5 When a discount is attributed to external factors such as luck, it is referred to as the “Lucky Shopper Hypothesis”.5

How this looks in real life:

The left side offer is an offer I received from a food delivery company in India. They sealed the deal with the words “In honour of your great taste in food” – Of course I have great taste in food, and of course I am a true foodie, hence why I deserve the coupon. The other extreme end of this spectrum is attributing discounts to external factors such as pure luck. 

2. Controllability –  This refers to the degree to which the promotion can be controlled by someone (controllable) or no one (uncontrollable)

If a customer perceives the factors related to the discount as under their control, they feel more responsible for the outcome, as opposed to a case when they have no control over how, when, or why they receive a discount. Take for example a promotional voucher for $10 off a cab ride — which a customer saves and uses on a day when cab prices are surging high due to rain — versus receiving an offer which a customer is forced to use right away because of an expiring time limit. In both cases, there is a discount involved, but the satisfaction from applying the discount when the customer wants to apply it is different.

How this looks in real life:

On the left side of the spectrum, Amazon is giving the customers a feeling of greater controllability by letting them “collect” the offer now and decide when they wish to use it. On the right side of the spectrum is an offer from a food delivery company that comes with a time-bounded pressure of 14 minutes. If I really want that burger, but I just had lunch, my only option is to let go of the deal and feel remorse.

3. Stability – This refers to how stable and predictable the discount is perceived to be over a period of time

When the reason for receiving a discount is stable over a period of time and is completely predictable, it drives repeat behavior related to this purchase. 

How this looks in real life:

The most common and relatable example of a completely predictable voucher is something we have all seen (and enjoyed) — the happy hour discount! If you walk into a bar during happy hour, no one can deny you the discount. The left side of the spectrum shows a Flash Sale, which is common in e-commerce platforms. Nothing about the Flash Sale is predictable — how much discount will I get? What products will be on discount? If I come back in an hour, will I still get this discount? If I come back tomorrow at the same time, will the same products be discounted? And that’s the effect the business is going for — they are using curiosity to get people to engage with the product.


The AI Governance Challenge

Implications of this knowledge

Having information about which factors make up the attractiveness of a discount gives businesses the ability to frame the same discounts in different ways based on their objectives. Let me explain this with an example:

Say a metro rail company is looking to shift the peak demand to off-peak hours with the purpose of easing the crowds. Just playing around with these different factors, what are the various discount mechanics they can use? Here’s a cheat sheet that you can use for reference.

As a customer, knowing companies have this information is important. At the very least, next time you are stuck in a bottomless scroll of attractive items on e-commerce and you are forced to add them to your basket just to unlock a discount, consider two outcomes:

  1. When you see a discount, you will objectively think about the effect the discount has on you and what made it seem attractive to you. Hence, you can decide whether it is worth the amount you will be saving.
  2. All of the thinking will shake you out of the discount trance, thereby saving you some money.

Either of those outcomes are ones I can live with.

TDL Perspectives: What Are Heuristics?

Julian Hazell, an Associate at The Decision Lab, sat down with Sekoul Krastev, a Managing Director, to learn more about his perspective on heuristics.

Some topics we covered include:

  • What are heuristics?
  • Why do they exist?
  • Are they good or bad?
  • How do they apply in the real world?
  • How were they discovered?

Julian: Let’s start with the basics. What are heuristics?

Sekoul: A heuristic is a simple rule that we use to solve more complex problems. They are mental shortcuts that allow us to make a decision about something without having to take a quasi-infinite amount of time to consider every single aspect of it. 

Julian: Can you give an example?

Sekoul: A common example of a heuristic in the decision-making literature is the availability heuristic. The availability heuristic is the tendency to view events or instances that are more salient as representative of an entire group. For example, if you’re thinking about the likelihood of dying on a plane due to a plane crash, you might think about all the times you’ve seen plane crashes appear in the news.

Now, plane crashes are very unlikely to happen. By some statistics, it would take 15,000 to 30,000 years of flying once a day to be in a fatal plane crash. Nevertheless, when you think about the likelihood of your plane crashing, you won’t recall all the flights that had ever been successful; instead, you will likely focus on the ones that did crash. 

This is the availability heuristic. It means that the events that are easiest to recall, and in this case, those that are emotionally salient, replace the process of forming a statistic about the likelihood of something happening.

Julian: Would we be better off without heuristics?

Sekoul: Heuristics are often talked about in the context of behavioral economics as irrational or biased ways of thinking or making decisions. Without heuristics, however, it would be very difficult for us to make any kind of decision or to function at all. 

Take the example of deciding between buying apples versus oranges at the supermarket. You might have a heuristic that you always buy oranges. That might be a habit that you’ve developed. Now, if you were to pull away from that and actually analyze the situation and think about every single factor that should go into that decision, it’s likely that you would never actually reach a decision. 

You might think about the prices of the two, the macronutrients, every single experience you’ve ever had with apples, every single experience you’ve ever had with oranges, et cetera. In order to avoid wasting resources or potentially having to face a problem that is completely unsolvable, the brain takes shortcuts. It basically uses heuristics in order to save us time and energy and to make the world simpler for us so that we can make more complex decisions faster.

Julian: What makes heuristics so important? When do they have the greatest impact?

Sekoul: What makes heuristics impactful is how useful they are in our decision-making process. At the same time, heuristics can sometimes lead to suboptimal decisions. So because they’re imperfect shortcuts to solving problems, they do sometimes lead us to make mistakes. 


The AI Governance Challenge

So, for example, you might think about discrimination as a type of heuristic, rather than thinking about it as something where people deliberately want to hate on a group. Another way to explain it and the way that the psychology literature might explain at least some types of discrimination is through the concept of a heuristic. 

Rather than trying to understand a group of people, one might instead think about every salient instance that that group contains or that is contained in the news regarding that group, and then formulate a mental shortcut that evaluates the group based on certain characteristics. 

That evaluation is obviously not fair because you’re taking a very small data set that’s been filtered through the media and using it to evaluate a whole group. And in those cases, that can lead to poor decisions at an individual level, but more importantly, it can lead to systemic problems.

Julian: Very interesting. I’d like to get at the nuts and bolts a bit more. Why do heuristics exist? 

Sekoul: Heuristics are a cognitive tool we use to facilitate decision-making. Essentially, our brains are optimized to, as far as we know, solve problems in a finite amount of time. When we’re faced with a decision such as the example of choosing an apple versus an orange, we tend to have a time limit that’s set by our brain during which the brain wants that decision to happen. 

So it’s very likely that as time goes forward, we use a simpler kind of rule to make that decision to the point where we might ultimately make it completely random. So what happens in the brain as we face a decision, we start by deliberating on it, and as time goes forward, we gravitate towards a simpler solution that ultimately unblocks us.


The AI Governance Challenge

Julian: Can we really avoid using heuristics?

Sekoul: As we touched upon earlier, heuristics are a useful tool for our brains to save energy and time. Without heuristics, we would not be able to solve many of the complex problems that we’re required to solve. As the world around us becomes more and more complex, heuristics actually become more and more prevalent and become larger and larger shortcuts. 

One example of this is in the political sphere. If we’re in 20,000 BC, our social circle might be composed of a group of, let’s say, 100 people. If we have to choose a political leader between two of those individuals, we might go for someone that we know better and think about our memories of that person and ultimately make a decision based on some heuristics, but also some deliberate thought.

In today’s context, where we have no personal relationships with political leaders, know very little about what they actually do on a day-to-day basis, who they are as people, what decisions they’ve made politically, the complexity of the decision of choosing who to support is much higher than before. 

So with that complexity comes a greater need for shortcuts. Something like political parties is, for many partisans, a heuristic that simplifies that problem. So, rather than understanding the individuals supporting what that person has done or is likely to do, we might make the much simpler decision of always supporting a particular political party.

Julian: Hypothetically, what would the world be like without heuristics?

Sekoul: Without heuristics, the brain would have a very hard time solving even the simplest of problems. If we just think about someone asking us, do you want to go out for lunch? That’s a question that might be solved quite easily with a heuristic. The heuristic might be something like if it’s after twelve o’clock, then I’ll say yes. 

On the other hand, if we don’t have a heuristic, even simple questions like that might paralyze us. One might, for example, weigh the downside of interrupting their current task against their hunger and the likelihood of getting hit by a car as they go to the restaurant. There’s an almost infinite number of factors that could be counted in almost every decision. Without heuristics, everyday life would be complicated to the point of paralysis.

Julian: Can we attribute moral qualities to heuristics? Are there instances where they are definitively good or bad?

Sekoul: Heuristics can be both good and bad at the same time. Because they allow us to function in the world and solve complex problems, they’re ultimately responsible for much of the progress that we make as individuals, groups, and as a society. At the same time, some problems need to be thought about more deliberately and our tendency to use heuristics rather than considering evidence is something that holds us back as individuals, as groups, and as a society. 

What distinguishes a good from a bad heuristic at the end of the day is our set of aspirations as a society, our values, and our moral system. The way that we judge the outcomes of our decisions is ultimately what should decide whether a more deliberate process is required. 

In the context of a complex issue such as discrimination, for example, we might, as a society, decide that being more deliberate and evidence-based is a better approach than simply using heuristics and discriminating against a particular group. In other cases, things may be more difficult to determine.

Julian: Okay, what about specific cases where we can isolate the effect of heuristics and decide how to adjust our behavior to accommodate them?

Sekoul: Well, on the other hand, in corporate settings, for example, speed is often important. In other cases, accuracy is very important. So depending on the consequences of a given choice, a combination of more deliberate and careful thought based on evidence will compete with heuristic, rule-based decision-making.

Rather than thinking about whether heuristics are good or bad, it’s important to think about the overall choosing process. Figuring out when evidence can be incorporated into decision-making without stalling the process is part of what applied behavioral scientists do, both on internal company projects and external projects facing clients. 

Julian: Let’s talk about how heuristics apply in the real world.

Sekoul: Behavioral scientists research the details of the heuristics we use every day. In applied work, they try to harness heuristics to make life more fair, efficient, and rewarding. For example, when a bank builds a product, they want to remove as much friction as possible from the customer experience. 

Heuristics will expedite the decision-making and reduce the effort of using the service. On the other hand, there are instances where the user must deliberately make a decision. Here, creating an interface that has users acknowledge, and work around, their heuristics will make a more effective product. 

There are places where friction should be reduced and heuristics should be used. There are other places where friction should be increased and the use of evidence should be higher and the use of rules lower.

As a whole, heuristics are a very influential concept in decision science and something that governs our lives as individuals, as groups, and as a society. The extent to which we use them influences the speed and quality of our decisions. At the same time, we must be somewhat deliberate with the heuristics we do use, even in cases where we do rely on rules, being more deliberate about the rules that we rely on as something that ultimately leads to outcomes that are aligned with our preferences.

Julian: This has been an insightful look into one of the core areas of behavioral science. Thank you for your time.

TDL Perspectives: Addressing The Climate Crisis

Sekoul Krastev, a managing director at The Decision Lab, sat down with Jayden Rae, a senior consultant with expertise in environmental policy work, to discuss some of the following topics: 

  • The collective action issue that makes climate change so difficult to tackle
  • Hyperbolic discounting and our myopia towards the future
  • What environmental policies have worked in the past
  • The political feasibility of environmental policies
  • How framing can dramatically impact the effectiveness of interventions
  • Behavioral levers that are useful for tackling climate change
  • The financial incentives that corporations face when reducing their emissions
  • How behavioral science can drive climate action as we go forward
  • How the COVID-19 pandemic may act as a catalyst for future change 

Sekoul:  Today’s chat is going to discuss behavioral science in the context of climate change. Let’s start with why climate change is such an important issue to tackle.

Jayden:  Climate change is one of the defining environmental challenges of our time. We’ve known about climate change for a long time but it has created a lot of challenges in the behavioral and political space for reasons that are rooted in human behavior. Climate change is a result of collective action problems, where consumption at the national level is perpetuated by individual choices that favor carbon intensive goods and materials. Climate change is a particularly challenging behavioral problem because of challenges arising primarily from psychological distance.

The idea of hyperbolic discounting essentially suggests that we value things in the present more than we value things in the future. Applied to climate change, this means that it is hard to realize how consumption in the present, like taking a long-haul flight, will translate downstream into damaging environmental outcomes. 

This also applies to spatial as well as temporal distance. It is known that a lot of the most significant climate challenges like sea level rise, food shortages, water scarcity, are going to take place in regions of the world in which the people who are consuming the most today do not live. So for people, it’s much easier to see what’s in our immediate environment and a lot of these environmental and climatic challenges are not being felt in the present and in our immediate environment. 

Sekoul: Why do you think this is a challenging issue at a policy level?

Jayden: There are a few reasons for this. The first one is related to collective action problems. So at the national level, as carbon emissions accumulate, the nation or the country is not necessarily experiencing negative consequences. This is one of the reasons international climate regimes have been largely ineffective, is that there’s no real incentive for a single country to reduce their individual impacts on climate change. 

However, the impacts of climate change will be felt later on into the future. So that’s one of the main challenges, and then, of course, some incentives in the energy industry make it quite challenging to reduce emissions at the national level.

Sekoul: What are some of the more successful policies that have actually achieved results regarding the climate?

Jayden: Probably the most effective environmental policy has been the Montreal Protocol, which addressed chlorofluorocarbons (CFCs) in the 1980s. At that time, scientists discovered the ozone hole over the Antarctic. The main reason for this was CFCs, which are a chemical used in refrigerants and industrial goods. It was believed, at the time, to be the main source of the ozone hole. So in the 1980s, policymakers from around the world came together to address this and actually created a legally binding protocol that restricted the global production of CFCs.

Because of the strict nature of this policy, essentially the industry is no longer able to produce CFCs. This method definitely favors the stick over the carrot but it was extremely effective, and today the ozone hole has largely recovered as a result of this policy effort. 

This policy worked for a few reasons. One is that it was really specific, we knew the exact nature of the problem. The second is that it was legally binding, and there was a lot of leadership from some key nations. So the US, for example, was one of the leading countries in that effort, which set a precedent for a lot of smaller nations that were not as responsible for the pollutant source.

Sekoul: If these hard-line approaches to policy making have been successful in the past, why don’t governments enforce them and create such policies at a more individual level?

Jayden: One reason is that it’s politically unpopular. A good example of this would be carbon taxes. They have been historically politically unpopular because of an essentially individual aversion to taxes in general. Plastic bag taxes, for example, have been largely unpopular in some places when they’ve been advertised as being so. So I think there’s definitely a political disincentive to implement more stick-related approaches.

Sekoul: What are policies that have failed and why?

Jayden: I think one interesting example, going back to plastic bag taxes, has been efforts in some places to actually create reward programs. So one key principle in behavioral science is loss aversion. So plastic bag taxes have been effective in some settings because when people have to start paying for something that was previously free, they feel like they’ve experienced a loss. If you’ve never paid for a plastic bag before, and then you start paying 10 cents, you feel that loss. But in some settings, policymakers have tried a different approach by using a reward.

Rather than paying 10 cents for a plastic bag, you would be reimbursed five cents, or you would save five cents off a purchase. Empirical studies have shown that the reward option of this policy has been completely ineffective — essentially, it doesn’t reduce consumption whatsoever. 

But if you have this exact same amount, but it’s a loss, you’re paying a tax on that, it can actually effectively reduce consumption. So it really shows how the intervention is framed to the consumer plays a really important role in whether it’s successful or not.

Sekoul: Can you talk a bit about the types of behavior levers that could be useful in creating effective policies that combat climate change?


The AI Governance Challenge

Jayden: So far, most environmental interventions have been applied to energy usage. One example that has been really effective in different settings is smart meters for regulating energy consumption. 

Essentially, smart meters show the consumer the quantity and price of the energy they are using in real-time. Some versions of this intervention can actually show the consumer what others are using in terms of energy, so they’re incentivized to use less energy than their neighbors. The smart meters have been really effective, and this really shows the value of making information salient to the consumer and linking it to their individual actions. 

Another effective intervention revolves around defaults in the environment. A lot of individual behavior and a lot of firm behavior is directly a product of the defaults in that setting. So if you think, for instance, to take a flight and you want to have a vegetarian option, you would have to opt into that. But you could also imagine a different reality where you have to opt-out of the vegetarian option.

Jayden: So if you want to change behavior, you have to make it extremely easy for people to exercise that behavior. A lot of our default settings are unsustainable options. From a policy perspective, it’s really helpful to change defaults to be more sustainable behaviors.

Sekoul: Since climate change is a game theory problem at an international level, what behavioral science approaches could a national leader use to lead the charge towards better policies?

Jayden: So there are a few. I think probably the most important one, which I briefly mentioned, is the simplification and framing of information. The more traditional approach to changing behavior is around regulation. In this case, simplifying information could be making environmental regulations or compliance really clear to firms to ultimately try to improve compliance. Then, of course, creating social norms around sustainable behavior; however, this usually takes a longer time to change.

Sekoul: Social norms ultimately become the biggest driver in any kind of collective behavior change. While it’s the more challenging approach, it’s probably the more sustainable one as well. How do you see social norms shifting? What drivers can shift social norms towards more sustainable behaviors?

Jayden: The first driver is probably around education and awareness. Any sort of social norm depends on an understanding of the underlying issues. In more recent years, environmental education has become a core part of school curriculums and that leads to more intergenerational change in norms. 

Unfortunately, given the pressing nature of environmental issues, this might be a bit too slow. Changing norms at the industry level is more challenging but decisions made at this level significantly perpetuate the crisis. 

Corporate social responsibility has become a norm in a lot of settings. In some ways, it’s exercised in meaningful, authentic ways, while other times it may not be. Even creating a culture at the firm level in which addressing the carbon footprint of the firm or having sustainability targets and initiatives as a part of core strategy is something that is new.

Sekoul: There has been a trend towards more corporate social responsibility programs in recent years, which could be a signal that companies see those programs as something that’s ultimately profitable, since companies are naturally focused on optimizing profits. What do you think drives that?

Jayden: It comes from the bottom but also from the top. With consumer behavior, a lot of consumers are now demanding more from the companies that they purchase from; furthermore, they have higher ethical standards for their goods. 

Then from the top are a lot of policy incentives. So, for example, in Canada, with the carbon tax system, it’s actually profitable for companies to cut down on their carbon emissions. So not only does it allow them to fulfill their social responsibility commitments, it’s also financially viable.

A lot of companies that are looking into the future can see that their investments or dependence on fossil fuels are not going to be sustainable and actually transitioning to more sustainable, lower energy intensive goods is going to be how they stay viable into the future.

Sekoul: Thinking about countries where some of these policies around climate change have been more successful and countries where they have been less successful, is there an inherent difference in either how they’ve been implemented or in the populations that they’re targeting?

Jayden: Often, countries that are most successful at addressing climate change have already gone through their phase of fossil fuel permanence and have industrialized earlier. So you have a lot of countries, like in Scandinavia, who have cut down their emissions. They are much more reliant on renewable energies today, but they had their period of economic growth and are now on the other side of the transition. 

The countries that are currently most dependent on fossil fuels are industrializing, developing countries. They are at a different point in the stages of growth. They are focused on economic development and human development. A lot of these richer countries, like Norway or Iceland, that are almost carbon neutral, have already passed through those stages of development and can focus their energy on being “green”. 

So there’s definitely some level of economic prosperity that can predict whether a country can or cannot actually implement sustainable policies directly. There’s also a geographical element as well. For some countries, it’s harder to make a transition to renewables given the natural resources that they can use to generate renewable energies, whether that’s hydro or wind power.

Sekoul: Where can behavioral science go from here? Are there ways behavioral science can be applied to environmental protection that aren’t prevalent enough?

Jayden: Yeah. A lot of past behavioral interventions have been focused on energy consumption. However, energy consumption is only responsible  for 24% of global emissions. A lot of work needs to be done creating and testing interventions that relate to other sources of emissions like agriculture, shifting global transportation, creating more local sources of some consumer goods.

The second key challenge, which is a challenge of all behavioral interventions, is scale. So a lot of these interventions have been tried in small scale settings, but it’s unknown how generalizable they are to other settings. So there needs to be more rigorous testing and evidence collected on these interventions can be implemented at scale and effectively implemented into policy and how these also relate to international environmental commitments.

The third challenge is around the effectiveness of some of these behavioral interventions over time. With carbon taxes, for example, some evidence has suggested that once consumers become sensitized to paying a tax, they will actually shift back to prior behavior. There has to be some long-term thinking about whether the behavioral intervention is actually going to create a sustainable change, or whether defaults are going to be sustained. It takes time to ensure that there are no unintended consequences of that intervention. 

Given the pressing nature of current environmental issues, we have seen that a lot of the traditional approaches to addressing them can be ineffective. Behavioral science relates to environmental challenges as our actions have direct consequences on the environment. 

Sekoul: One last question. We are currently confronted with the kind of large scale global event that becomes a common experience for pretty much everyone in the world with the coronavirus pandemic. Do you think that kind of collective experience and struggle and an alignment between what countries are doing to some extent is potentially a precedent or something that will facilitate climate change policy unions or alignments as well?

Jayden: I think there are two potential outcomes. One is pessimistic and one is optimistic. The pessimistic answer is that the economic stress that this particular global pandemic has created will incentivize national leaders to revert to the status quo and essentially have an emergency response to the current situation, which is possible. There’s a lot of pressure from citizens to essentially have a quick response that may just result in, for example, bailing out industries that are harmful to the environment.

The second, more optimistic answer is that the world has been completely destabilized, so we can take this kind of crisis to create new norms. Of course, the shared experience of going through this has created a sense of global solidarity, which has been missing in the climate change conversation and would be really critical to creating some of this international cooperation that is really necessary to overcome the collective action problem which is climate change.

There are reasons to be hopeful that this destabilizing situation has created the conditions to imagine new realities. The world that we have lived in is not just or healthy and it will not sustain us in the future. In that way, it opened up imaginations to what a new future could look like.

Why We Sometimes Favor Aggressive Political Leadership

A few weeks ago, Canada lost its bid for a seat on the United Nations Security Council. For some, it was a non-event. But for others, it reminds of a moment from five years ago when the new government promised Canada a fresh, progressive, and multilateral foreign policy. After a similar UN Security loss by the Conservative government in 2010, Justin Trudeau stated rather bluntly: “Canada’s back.” Canada would supposedly regain its voice on the world stage with a new diplomatic approach. Now, many are left questioning how that policy has worked for Canada today.1

Leader persuasion is an essential skill for gaining popularity and enacting policy. There exists a body of political research that focuses on how, when, and why political leaders succeed. Much of the research explores “hawkish” policy, which is when leaders take an aggressive approach to international relations. Hawkish leaders are typically perceived to be stronger and more uncompromising than their dovish counterparts.2 Examples include Winston Churchill, Richard Nixon, and Margaret Thatcher, who once famously reminded George Bush not to “go wobbly” in response to Sadam Hussein’s invasion of Kuwait in 1990.

These leaders stand opposite to “dovish” leaders, who usually advocate for more peaceful or diplomatic measures, such as Jimmy Carter, who focused on human rights with his national security policy back in 1977.1 These categories do not necessarily define partisanship, as hawkish democrats and dovish Republicans do exist. These terms do, however, help describe the typical approaches that leaders use, especially in foreign policy matters.

Trudeau’s (failed) attempt to gain a seat at the UN Security Council highlights a misperception of hawkish policy’s success. A significant part of Trudeau’s early campaign was his promise to bring Canada a new foreign policy that contrasted the hawkish strategy used by previous governments.1 Reportedly, “Trudeau has repeatedly pointed to the 2010 failure to win a seat as a sign the Conservative approach to more hawkish foreign policy was not as effective as his own focus on multilateral and quieter diplomacy.”1

When Canada lost, critics were quick to blame the government’s “dilettante” strategy and lack of a coherent foreign policy. His “quiet and multilateral” position was perceived to be a passive approach.1,3 It’s plausible to think that an amicable demeanor might fair better in foreign policy, but research demonstrates the opposite to be true in many cases.

Why hawks win

Experts have consistently found that hawkish policies in fact do better in foreign policy, especially in reconciliation.2,4,5 The popular phrase “Only a Nixon could go to China” describes this phenomenon. Many believe that the hawkish demeanor of US President Richard Nixon was necessary for the successful rebuilding of US/China relations in 1972.4

To test this effect, an experiment asked individuals to rate a fictional leader who is attempting to reconcile with another foreign leader. In the experiment, the leader was either hawkish or dovish, a republican or democrat, and either enacted a policy change or went with the status quo. When enacting change, the hawkish leader was viewed favorably, while the dovish leader was ridiculed.4

When an aggressive leader enacts new changes, people perceive them as acting moderately, so would seem. When a dovish leader does the same thing, individuals perceive it as a passive measure.4 For a hawkish leader, a policy change signals a major decision made in the nation’s best interests — a luxury that dovish leaders do not necessarily enjoy. Cognitive biases can explain why.

This behavior is a function of fundamental attribution error. With this bias, we tend to inaccurately attribute success (or failure) to the person, not the situation. For a hawkish leader who is known to be more hostile, people will attribute their hostile actions to their character. When they engage in moderate behavior, like attempting to reconcile, we perceive it as a deviation from their normal behavior in response to the situation.2,6 On the other hand, a dovish leader who attempts to reconcile does not receive this advantage, as it is consistent with their behavior.

Positive illusions — or unrealistically favorable attitudes towards ourselves or others — also contribute to our preference for hawkish leaders. Because of it, our faith in beneficial outcomes is increased with an apparently strong leader. For example, we tend to go to war because, rather logically, we think we can win. A leader who is confident in going to war is more assuring than a passive one, all else equal.2,5,6

The hawkish advantage in elections

Hawkish leaders also perform well during election season — though only in certain contexts. In 1968, the level of violence in times of protest shaped how voters perceived leadership styles. In times of peaceful protests during the civil rights movement, swing states preferred Democratic leaders who supported the movement. In times of violent protests, these states preferred Nixon’s style of leadership, which contained promises of “restoring law and order”.7 Over 50 years later, history is reoccurring.

Source: Will protests help Donald Trump as they did Richard Nixon in 1968? (2020, June 8). The Economist.

Some researchers make the case for all politicians to take hawkish approaches to reap advantages. Recent findings say otherwise. A different study looked at perceptions of the 2008 US Election candidates and found that when an individual believed that a democrat leader was more hawkish than themselves, they preferred them less.

The driving explanation for this is that in the post-9/11 US, people were more hesitant about war and preferred a leader that did not invoke an urge to engage in combat.5 Although war-averse, the participants still believed that the hawkish leaders were better equipped to handle matters of foreign policy and security and preferred them for dealing with those matters.

The studies present a notable paradox. On the one hand, already elected hawkish leaders have an advantage over dovish leaders when enacting forms of reconciliation. Yet before elections, more peaceful candidates can’t take advantage of this preference. Context is therefore a major player in these perceptions. The research shows that bias plays a role in our perception of good leadership, but so does our exposure to war or violence. The influence of context on our decision-making deserves recognition, especially in light of current events.


The AI Governance Challenge

Regarding the UN Security Council vote, critics everywhere contribute differing explanations to Canada’s UN Security Loss.1,3,9,10,11,12 The persistent belief five years ago that a dovish strategy would benefit Canada’s foreign policy is now up for question. The research, and the result, say that the strategy has failed. But given that dovish leaders can’t take advantage of the “hawkish benefit”, it’s hard to say if a different strategy would yield different results for Canada, or if the government would have been elected in the first place.

What a dovish leader can do

Researchers from Cornell, Stanford, and Georgetown agreed that hawks have an advantage in ratifying political bargaining and arms deals. But, dovish leaders can overcome this. In ratifying arms deals, they found that dovish leaders can overcome their disadvantage by paying a “premium”. In an arms deal treaty, doves’ tendency to cut military effort may cause other legislators voting on the issue to believe that the other party will not follow through with their obligations without a more aggressive stance.

However, if a dovish leader works with hawkish legislators on the deal who act as endorsers, they can successfully ratify the deal. To gain this endorsement, the dovish leader can spend money on areas outside of the deal, such as increased military spending. It isn’t that dovish leaders are unable to successfully enact foreign policy deals, but they do so at a greater price than hawkish leaders.8

“Nice guys finish last”, or so it seems. In the realm of security, people perceive hawkish leaders as more successful. Our cognitive biases contribute to this perception, such as when we give hawkish leadership undeserving advantages.2,6 These biases are context-dependent and beneficial for elections, though this depends on certain factors. Dovish leaders can overcome this “advantage”, but their success in foreign policy may come at a higher cost.

Over 10 years ago, Daniel Kahneman put it best in saying: “Understanding the biases that most of us harbor can at least help ensure that the hawks don’t win more arguments than they should.”6 Aggressive policy certainly shouldn’t be avoided in all situations. But in a time where we are increasingly evaluating policy and leadership in several contexts, it’s imperative to assure our decision-making gives credit only when it truly is due.

Using Digital Health Support And Behavioral Science Principles To Help Treat Tuberculosis


At TDL, our role is to translate science. This article is part of a series on cutting edge research that has the potential to create positive social impact. While the research is inherently specific, we believe that the insights gleaned from each piece in this series are relevant to behavioral science practitioners in many different fields. As a socially conscious applied research firm, we are always looking for ways to translate science into impact. If you would like to chat with us about a potential collaboration, feel free to contact us.


Behavioral science insights can profoundly impact health outcomes — from encouraging prosocial handwashing behaviors during a pandemic to increasing the number of individuals who sign up for health insurance. As a socially-conscious applied research firm, TDL is interested in using empathy, technology, and design-thinking to promote better outcomes in many aspects of society, from health to education to the economic empowerment of disadvantaged groups. To amplify these impacts even further, leveraging digital tools to create health solutions can scale and achieve these desired outcomes more cost-effectively than traditional interventions.

The Decision Lab reached out to Dr. Erez Yoeli of MIT to learn more about his work on a project involving digital health tools for Tuberculosis (TB) patients and the future direction of similar areas of research in behavioral science.

Dr. Yoeli is a research associate at MIT’s Sloan School of Management and co-director of the Applied Cooperation Team (ACT), a team of researchers that applies insights from the social sciences towards increasing contributions to real-world public goods. 

In this study, Dr. Yoeli and a multidisciplinary team of researchers developed a behavioral science informed digital health platform that provided TB patients with support and increased adherence to treatment plans.

A full version of the paper is available here:


Julian: What is the focus of your research?

Dr. Yoeli: My research focuses on altruism — understanding how it works and how to promote it.  I collaborate with governments, non-profits, and companies to apply these insights to address real-world challenges like improving antibiotic adherence, reducing smoking in public places, increasing energy conservation, and promoting philanthropy.

Julian: What process did you follow for this piece of research?

Dr. Yoeli: Tuberculosis is the world’s deadliest infectious disease. It kills roughly 2 million people each year despite the fact that it has had an effective cure since the mid-1940s. Unfortunately, this cure requires a long treatment (6+ months), during which patients are required to take a strong antibiotic on a daily basis and return to a clinic for regular (typically weekly) visits. 

Many patients, quite reasonably, stop this treatment prematurely with the hopes that they are cured. Unfortunately, they often are not by the time they stop, and this can result in the transmission of TB to others in their communities as well as the development of drug resistance. A further problem arises as the consequences of this resistance are borne not by the individual, but by the community as a whole.

We believed that individuals needed better support motivation to consider the consequences that ceasing treatment has on their communities. We built a digital health platform based on principles from behavioral science for this purpose. 

The platform’s three guiding behavioral principles were: 

  • Increase accountability
  • Reduce plausible deniability
  • Normalize adherence

Each morning, the platform would send patients a reminder to take their medication and to log in and verify that they’d done so. If patients failed to verify, they’d receive another reminder an hour later, and then an additional reminder an hour after that. 


The AI Governance Challenge

If at this point patients still did not verify that they took their medicine, they were classified as non-adherent, and a team of supporters would contact them by text or phone. There are also some additional features that the platform offers: occasional motivational texts, the ability to chat with supporters at any time, information about TB, and a bit of gamification that further amplifies feelings of accountability.

Behavioral science principles played a key role in informing the platform’s design. For instance, by insisting on verification — which might at first seem like it’s burdensome for the user — we killed two birds with one stone: We generated accountability and eliminated plausible excuses like ‘I didn’t see the reminder’ or ‘My phone battery was dead.’ 

Julian: What did you end up finding out?

Dr. Yoeli: We tested the platform in a year-long field trial with roughly 1,100 TB patients in Nairobi, Kenya.  One-half of the patients were supported via the platform (the treatment group), while the other half continued to receive the standard of care with no changes (the control group). Amongst the control group, roughly 13% failed to complete their treatment, a number that is typical of Nairobi.  Amongst the treatment group, though, only 4.3% failed to complete treatment, a two-thirds reduction — the greatest such reduction in the literature to date. The platform helped those with higher non-completion rates (men and patients who are co-infected with HIV) the most, bringing their non-completion rates down nearly to the same level of everyone else.

Julian: How do you think this is relevant to an applied setting?

Dr. Yoeli: Medication adherence is an important and difficult problem for a range of diseases throughout the world. Most people think of this as a problem for the patient, but we think it is helpful to focus the individual’s attention on the implications of adherence for their community. Behavioral science can then be used to successfully motivate individuals to be more adherent and to take into account the impacts of their actions on others.

Julian: What are some exciting directions for research stemming from your study?

Dr. Yoeli: We’re currently wrapping up a three-year follow up involving roughly 20% of Kenya’s TB patients throughout the country. We’ve also adapted the platform for HIV patients, and are gearing up to test it in South Africa. And, we’re incorporating a bit of Artificial Intelligence to help prioritize support to make sure our team of supporters reach out to those who need the most help.

We are always excited to work with new partners. If you have a project in health, sustainability, philanthropy, or any other field with which you could use some help, don’t hesitate to reach out by emailing me at

Remaining Vigilant In The Era Of Information Overload

Four months into the pandemic, a counterintuitive phenomenon has emerged. In March, when the risk of COVID-19 infection was at its lowest, the public’s motivation to follow prosocial pandemic behavior appeared to be at its highest. In the United States, the risk of infection is higher than ever, yet the motivation to adhere to public health recommendations seems to be at an all-time low. This phenomenon is known as caution fatigue and poses severe health risks to communities. 

Caution fatigue, which was coined by Dr. Jacki Gollan, an associate professor of psychiatry and behavioral science and a clinical psychologist at Northwestern University, can explain a lot of the recent behavior we have seen in the news. Gollan suggests that the initial burst of energy at the beginning of the pandemic helped us approach the public health guidelines optimistically. However, as we began to find difficulty in foreseeing the end, our energy banks became depleted, and we grew more relaxed about the steps we were taking to mitigate risk. Various behavioral insights can describe what may be causing caution fatigue.

Adapting to the threat

Threat habituation describes when we become less sensitive to threats after repeatedly encountering them. The idea is similar to that of fear-extinction training, a form of psychological training that aims to help individuals with anxiety by repeatedly exposing them to a fear-eliciting cue that is not accompanied by an aversive event. Eventually, the individual’s fear decreases as they learn that there is no real reason to be fearful.1

In this context, the threat is the highly-contagious COVID-19 infection. Nowadays, we are presented with the danger of COVID-19 nearly everywhere — the news, conversations with friends and family, social media feeds, and even work. If we aren’t directly affected (or, rather, infected) by COVID-19 during this, then we may adapt to the threat and gradually become desensitized. Since our brains cannot handle persistently high levels of stress, it is simply more comfortable for us to ignore the threat and return to healthier levels of stress. 

Uncertainty of the threat

The COVID-19 pandemic is considered abstract, especially as we cannot easily calculate the risk associated with our actions and environments. To a certain extent, our mind is often unable to comprehend the actual severity of the pandemic unless we are affected by it ourselves.

While initial bursts of energy and motivation can help us follow safety guidelines, the inability to tangibly identify risk can instill a sense of hopelessness and can contribute to caution fatigue.

One interesting concept concerning risk perception is voluntariness, which describes when risks taken voluntarily are perceived as lower, while risks originating from external forces (or out of our control) are seen as greater.2 While early in the pandemic, the risk of COVID-19 appeared to be out of our control, the level of panic was at its highest. However, as we were repeatedly exposed to suggestions of how we can reduce our chance of contracting and spreading the infection, our risk perception of COVID-19 may have gradually shifted towards feeling as if we were in control. As a result, we may perceive the risk of not adhering to guidelines to be lower. 

Information overload

We are constantly bombarded with information through various channels in situations as noteworthy as COVID-19. Our brains are more likely to remember emotionally salient information, especially adverse events, as these allow us to recognize potential threats. This is known as pessimism bias, which describes how people often overestimate the likelihood and consequences of negative future events. However, with an overload of information, this can become overwhelming.

One major problem that comes with information overload is misinformation. In fact, regarding the myriad of sources out there (especially ones on social media), much of the information surrounding COVID-19 is conflicting and inaccurate. Having to filter out the correct information adds a whole new layer of uncertainty, and can become mentally exhausting for us. As a result, caution fatigue can begin to set in.

Psychological reactance

Reactance is a theory in psychology that describes how individuals are motivated to regain freedoms when they feel the threat of losing them.3 Classic examples of reactance can be drawn from children’s behavior — you tell a child that they cannot play with a particular toy, and all of a sudden, the child only wants to play with that specific toy. Reactance also applies to adults — during the pandemic, it has caused deadly consequences by contributing to our lack of motivation to adhere to public health recommendations. For example, as the public is repeatedly reminded to wear a mask, some individuals may react by not wanting to wear a mask to unconsciously assert a sense of personal choice. 


The AI Governance Challenge

Although counterintuitive, the constant reminders to engage in safe behavior can encourage people to ignore guidelines and behave even riskier. Interestingly, reactance may be more common amongst individuals living in countries that boast individual freedom, like the United States.4 Reactance also contributes to the growing antipathy that our country is observing towards experts; put blankly, reactance allows individuals to feel in control of their own lives again, even if this means dismissing strongly grounded evidence and knowledge.

How to mitigate caution fatigue

Ameliorating caution fatigue is not a helpless cause. Firstly, bringing awareness to the concept of caution fatigue and its contributing factors can help promote self-awareness of our actions.

The following are more concrete strategies to combat caution fatigue:

Create a practice of periodically visualizing hypothetical situations where your risky behavior results in you or your family being adversely affected by COVID-19. 

This can help in perceiving the risk of the pandemic more tangibly. By placing a greater focus on the long-term consequences of your actions (and overriding our natural tendencies to overvalue short-term benefits), you can also motivate yourself to continue taking the proper precautions and avoid desensitizing yourself to the threat.

Remember that risk is compounded. 

Not having been affected by previous risky behaviors does not indicate a lesser risk of acting the same way. Instead, a decline in safety behaviors only adds to your overall risk of being adversely affected.

Focus on receiving news from only one or two reliable sources and limit the frequency at which you check the news to ensure you don’t feel burned out.

This can make the process of understanding information less overwhelming and more comfortable to digest.

With the paradox that psychological reactance presents, it is of utmost importance to remind ourselves that saving lives takes priority over a personal illusion of agency. We must recognize our lack of control over the situation. By doing so, we can maintain our motivation to continue staying safe.

Sports Leagues Aren’t As Competitive As You Might Think

Professional sports leagues are all about winning. Yet perhaps more importantly, they are about money (and the talent that can be bought with it). With Russian oligarchs and Emirs buying up teams in many sports leagues worldwide, a schism is being wedged between teams — those who are backed by the fat cats and are oozing with talent at the top of the table, and the have-nots at the bottom who find themselves barely able to stay in the league.

Competitive balance — considered a central requirement for maintaining spectator interest and therefore the success of leagues — is weakening in many leagues. Behavioral science and the results of experiments shed some light on how team ownership can erode competitive balance, and why leagues may not wish to push for reforms.

Team ownership and competitive balance

Professional team sports are classic examples of business cartels. However, while sports leagues face a host of incentive and enforcement problems discussed often in literature, they are different in one important and paradoxical respect: Sports leagues are in the business of selling competition. To be successful, a league has to ensure a competitive balance between the teams while keeping the spectators happy. Spectator preferences can be defined as being related to the performance of teams vis-à-vis each other. In other words, it is competitive balance, or the quality of contest between teams, that matters to viewers.1,2,3,4

Competitive balance is dependent on the motivations of team owners. While we cannot be certain what objectives (profit-maximization, win-maximization, or utility-maximization subject to minimum profits) team owners individually and jointly have, we can at least be sure that they all want to put together a team that will simultaneously maximize performance and improve attendance and revenues when they compete.5

Money talks

Wealthy owners for whom winning is a matter of pride over anything else — or, more specifically, wealthy owners with a philanthropic bent who wish to restore pride to, say, an impoverished city — might focus purely on winning by buying the best players on the market. So long as the money is being spent, team losses can be subsidized through other means such as the side business ventures that provided the owners to capital to buy the teams in the first place.6,7

Such owners can then put together superstar teams that out-compete teams with owners who cannot spend as much, thus reducing competitive balance.

Take the English Premier League (EPL), for example. Sports analysts identified a sharp increase in spending on players in the last two decades, especially by the wealthiest clubs, leading to a decline in competitive balance in the league starting around 1994-95. Research has shown that during the 1980s and early 1990s, when spending was relatively lower, the competitive balance of the league was average or higher than average compared to other large European football leagues.8

Decoding imbalance: underdogs, superstars, disengagement

In the EPL, the competition for top honors has become relevant only to the top quartile of the league where teams spend extraordinary sums to get top talent and play at a ‘different level’. The lower tier cannot realistically compete with the top teams and instead choose to focus on avoiding relegation.9,10 The superiority of the top clubs is remarkable no matter which statistic we observe.

The gulf between the haves and have nots is immense. In the 2018-19 season, Manchester City accrued 98 points, behind the 100 points it accrued the previous season. Manchester City won ten matches by at least three goals – meaning 25 percent of its games were effectively a no contest.

In 67 matches last season, one of the top six teams had the ball for around 70 of the 90 minutes, essentially dominating the flow of the game. The FA Cup final, an important fixture in the English football season, was one-sided with Manchester City thrashing Watford, a team who finished mid-table, 6-0.

Top dogs, top dollars

It comes as no surprise that the top teams make much more money than the bottom teams — the best teams fill seats in stadiums, get global audiences, and earn more television revenue. From 2008-09 till 2017-18, collective revenues for the top six teams rose by £247 million and revenues for the remaining 14 clubs fell by over £10 million.11

Some analysts argue that a talent gap can lead to improved performance by the weaker team, as explained by the ‘underdog effect’, which states that weaker teams work harder to beat lower expectations.12 Whether this could happen in a sustained manner through an entire season is, however, up for debate. When individuals with unequal talents compete, the less talented competitors may ‘give-up’ while the high-ability players waltz to victory. In other words, the relatively less talented athletes wilt under a ‘superstar effect’.


The AI Governance Challenge

For example, data suggests that Tiger Woods’ PGA Tour earnings were higher by $6 million between 1999 and 2006 because his competitors suffered the superstar effect.13 A similar effect is observed in competition between teams as well. Experiments have shown that if the gap in talent is large, the weaker team disengages.14 With respect to leagues like the EPL, these effects could explain the drop in competitive balance.

Unbalanced leagues: overcoming bottlenecks for change

The million-dollar question then is why do highly imbalanced leagues exist? For one, there could be a status quo bias for the top teams. Playing against (and dominating) weaker opponents could make good television in today’s world, where video bytes of Messi dribbling past a hapless opponent is what a global audience wants to see. It might then be the case that competitive balance is not what the top teams seek.

Another reason why both strong and weak teams might not wish to reform a league can be explained by the sunk cost fallacy. While it might be in the interest of all teams to break up a league and find alternative structures, existing investments in the league and revenue streams may stop teams from doing so, even if they are not making profits.

For example, although there has been chatter about the top teams from the big European leagues breaking away and forming their own ‘super’ league — one that will boast the best talent and the highest competition — putting a league together with the right contracts, TV rights, and fan support, among other considerations, is hard to accomplish.

There are measures sports leagues like the EPL can take to restore competitive balance if they choose to do so — after all, spectator or viewer contentment is key to a league’s success. Some popular measures include:

  • Cross-subsidizing weak teams: League revenues can be redistributed to compensate the weaker teams for their low incomes. League revenues could include national and domestic TV revenues, gate fees, etc.
  • Managing team inputs: Controlling team inputs can have a strong effect on competitive balance. For example, reserve clauses in player contracts allow teams to retain rights to players even upon the contract’s expiration; essentially, players cannot act freely to choose who they will play for.
  • Salary caps: Limiting what teams can spend can also go a long way in ensuring top players are available, at least in theory, to all teams in the league.
  • Rookie drafts: Drafts are another popular method to ensure talent is distributed equally amongst teams in the league, such that teams that finish lower in the league tables get to choose new talent entering the league sooner than the better performing teams.

While it is true that almost all leagues take some measures to maintain competitive balance, which ones are chosen and to what degree they are complied with will ultimately depend on how well the interests of different teams and the league itself are aligned.15,16

TDL Perspectives: Becoming A Behavioral Scientist

This discussion between Sekoul Krastev, a managing director and co-founder of The Decision Lab, and Julian Hazell, an associate at TDL, addresses what applied behavioral scientists do, and what are some of the things that are needed to successfully become one. This conversation covers:

  • What applied behavioral scientists actually do
  • What the typical day looks like for an applied behavioral scientist
  • What background is needed to enter (and succeed in) the field
  • If graduate school is necessary to become an applied behavioral scientist
  • Important trends in the field
  • Important skills to learn outside of the classroom
  • What skills prospective applied behavioral scientists often lack
  • What is disrupting the field (and how to prepare for this disruption)
  • What organizations an applied behavioral scientist can work for

Julian: Let’s start by asking what it means to be an applied behavioral scientist. 

Sekoul: An applied behavioral scientist is someone who is in charge of reaching some level of understanding of how people are behaving in a system and then modifying that system to better accommodate those within it.

Typically, the work done by an applied behavioral scientist focuses on customers (external), or employees (internal).

A customer focused project would look at how customers are behaving in order to better understand them and potentially improve their outcomes. 

The other case is when an applied behavioral scientist would look at how employees are behaving internally. They look not just at how employees are behaving, but what kinds of incentive structures might motivate them and how to change cultures, among other things. 

An applied behavioral scientist is somebody who understands those types of systems of behaviors in order to improve them or to align behaviors with some sort of optimal outcome.

The reason they are called applied behavioral scientists is because, in order to accomplish these things, they’ll leverage insights from many different fields

They will usually have a background in some aspect of behavioral science, meaning a sub-field such as psychology, neuroscience, or sociology, among others.

Someone from one of these fields will go through the process of taking all the insights they have from a theoretical level and figure out how to then apply them in a real-world setting — this is where the applied portion comes in.

An applied behavioral scientist takes techniques, such as creating surveys or running experiments, and adapts them in a way that answers useful questions in the context of customers and internal employees.

Julian: What is a typical workday for a behavioral scientist, let’s say in a large company?

Sekoul: A typical workday for an applied behavioral scientist depends on the kind of context that they work in. For example, if they work in an enterprise context where you have users, what they might do on a day-to-day basis is look at the particular Key Performance Indicators (KPIs) that they’re in charge of changing in a certain direction. 

Their day might start off in a standup meeting where they briefly discuss how the KPI is doing. For example, an applied behavioral scientist might talk about sales numbers, or engagement numbers, or number of downloads of a product, et cetera.

In that meeting, they also identify certain key goals, either in response to a KPI not moving into the direction they want them to, or maybe a more aspirational goal. 

Then, they’ll work throughout the day to either dig into insights, to understand why things are happening and the way they’re happening — for example, why sales numbers might not be going up or why engagement might be going down. 

Then, they’ll spend another part of their day designing interventions that are aiming to change that situation.

An applied behavioral scientist probably spends at least a third of their day in conversations with various teams in order to understand how a particular intervention they’ve designed might be implemented and tested in the real world, and then managing and coordinating that implementation and testing.

A final, smaller part would be to communicate all of that with senior stakeholders, depending on their level of seniority.

Julian: Okay, what about in a more niche environment, like a smaller company?

Sekoul: In a smaller company, an applied behavioral scientist — somebody who is maybe the only applied behavioral scientist on a team — might be asked to do different things, such as content production, or evangelizing behavioral science, helping to do user experience design, or helping to inform product design, among others. 

There is a lot more variety of tasks throughout the day. 

At the same time, the kind of data that they would be getting is typically less well-structured than in a large company. So they would end up with a lot more leeway in terms of designing the interventions, which usually means that they’d need a stronger emphasis on creativity and on being able to ideate and iterate.

Julian: Is there a right background for becoming an applied behavioral scientist?

Sekoul: There are a variety of backgrounds that are useful in becoming an applied behavioral scientist. Ultimately, someone who is very good will learn about various fields that inform the kinds of work that they have to do at a company. 

So backgrounds that can be useful are ones such as:

  • Psychology
  • Neuroscience
  • Cognitive science
  • Decision systems (which is a field within management)
  • Anthropology
  • Economics
  • Sociology

Sekoul: That being said, people will usually come in with one of those fields with some sort of expertise and will then learn how to use that as a base in order to learn what the other fields can do to contribute to their work. 

It’s very rare that someone will just use psychology, for example. They might start with a psychology background, but eventually will need to understand how people behave on a group level. So they’ll study some sociology, management, organizational behavior, or other fields to gain those necessary skills and insights.

In a nutshell, a good background is one that gives someone an idea of how people make decisions from the point of view of a single field, but also one that they can then leverage to pull insights from different fields.

Julian: How necessary is graduate school for pursuing this career?

Sekoul: Deciding whether to pursue graduate studies really depends on the kinds of skills that someone has built in their undergraduate program — a lot of people come out of undergrad with a pretty strong grasp of a particular field. 

For example, somebody studying psychology might already have a lot of knowledge that they’ve picked up on various kinds of studies and theories. They might also have a strong background in statistics or things like programming. 

So if someone comes in with those skills, even an undergraduate degree can be a very good starting point for becoming an applied behavioral scientist.


The AI Governance Challenge

That said, certain graduate programs, especially ones focused on research, can be very useful. It ultimately depends on what someone did in their undergraduate program and what the graduate program is.

Julian: Can you elaborate on the types of graduate programs relevant in this area?

Sekoul: There are two general types of graduate programs:

  • Ones that are course focused.
  • Ones that are research-focused.

A class-focused program will give you more knowledge within a specific field. It will usually focus on a subset of the fields. So, for example, if someone studied economics in undergraduate, they’ll then focus on something more granular, like behavioral economics.

A research-focused master’s or PhD program on the other hand is predominantly focused on producing some sort of research outputs. With one of these, one is essentially practicing the kinds of skills that would be useful in applied behavioral science, such as:

  • Running experiments
  • Learning about research methodologies
  • Recruiting participants
  • Incentivizing participants
  • Checking if participants are doing the task that you set up in a good manner
  • Creating pilots and then adjusting your task
  • Doing power analyses
  • Calculating what sample size you might need
  • Creating engaging experiments

All those things are learned through practice. Graduate programs that focus on giving someone that kind of practice are very useful. 

One caveat is that if someone spends too long in the graduate program, for example, a PhD, they might double down on a particular topic and then become a bit myopic when it comes to other topics. 

It’s very important that when someone starts in applied behavioral science, they’re able to pull insights from different places. And that’s not something one typically learns as they’re doing something like a PhD. Usually, someone specializes more and more.

Julian: Are there any notable trends in the field of applied behavioral science that someone just getting started might be interested in?

Sekoul: One interesting trend is that people hiring for applied behavioral science positions a few years ago were really focused on the behavioral science portion of the equation. 

Hiring managers essentially started by populating these nudge units or behavioral science teams with people who have a PhD in psychology or neuroscience. 

And they assumed that these people, because they’ve studied these fields at a high level and had peer-reviewed papers, would be able to then take those skills and insights and transfer them into a real value add within the company in order to change, for example, sales numbers, or improve engagement in a product.

That has shifted with time. The idea that you can just take somebody with a PhD and get them to deliver real value on day one has really changed in the last few years. 

And now there’s a trend towards looking for people who are essentially at the kind of meeting place between the more theoretical academic background and a more applied background

So the optimal candidate shifted from being a PhD in neuroscience towards being someone who’s kind of a hybrid. For example, somebody with a master’s in neuroscience and a few years experiences in a business-focused position.

Julian: What about technical skills?

Another big trend is actually the overvaluation of technical skills in the field. It used to be important to hire somebody with very strong statistical skills or programming skills. In the real world, however, those kinds of skills don’t necessarily lend themselves to real value on day one. 

Now, there’s still that kind of emphasis on some level of technical skills, but this has gone down with time. Instead, there’s a bigger emphasis on things like:

  • Being able to follow certain design methodologies
  • Being able to brainstorm
  • Being able to work with others

And other skills on the softer side.

A lot of the initial behavioral science teams that were created didn’t perform as well as they could have, simply because they didn’t necessarily have the right balance of hard and soft skills. They overemphasized hard skills to the detriment of soft skills.

Julian: To expand on that, are there important skills that come from outside the classroom for someone looking to be an applied behavioral scientist?

Sekoul:  The most important skill when one is applying behavioral science is the ability to tap into evidence, which splits off into two main categories:

Being able to read what’s already out there, and being able to run experiments to gather evidence yourself.

From a practical point of view, being able to read what’s out there means being able to read through papers like the ones on Google Scholar. Another way is by searching for a particular topic by opening up the most relevant papers and being able to read not only the abstracts, but also the methodology sections in order to understand what the key insights from the field are.

The other type is around testing by running experiments yourself. Once an experiment has been designed, one has to be able to validate it in the real world. One has to be able to:

  1. Design an experiment
  2. Code it
  3. Find an appropriately sized sample
  4. Run it
  5. Analyze the data
  6. Write it up
  7. Pull actual insights that one can then use to change an organization or a product

To develop those skills, what you might want to do is find some sort of behavior change goal that is considered interesting. 

For example, say you might be very interested in fitness. In that case, you could focus on a subset of fitness. You could say, for example, that you want to develop better posture. 

One good kind of activity to practice for being an applied behavioral scientist is a 3-step process:

  1. Go on Google Scholar
  1. Find different kinds of research talking about improving your posture from a behavior change perspective
  1. Design an experiment that tests out different kinds of interventions in order to see which one is most likely to improve posture over time

That process is something that you do nonstop in an applied behavioral science position. Finding an interesting and fun way to do that on the side while you’re still in school is probably the best kind of preparation you could do.

Julian:  What is the most important part of the hiring process that applicants often lack?

Sekoul: A lot of people coming into the field are trained in the science part of applied behavioral science. They come in with many hard skills and a good familiarity with a particular field. 

There’s usually a lack on the applied side, however, which is extremely important.

It’s pretty rare to find someone who is coming in with a balance of both science and applied experience. For most people, the lack of experience in going through the process of identifying a behavior change goal, looking for evidence, designing interventions, and then testing those interventions in the real world.

That process is something that not many people have gone through simply because the process itself is a combination of those technical academic skills and applied skills.

So, a lot of candidates will have experience on the technical side and will be able to read technical journals and examine evidence and design experiments and run them, but they won’t necessarily have an idea about how the results of those experiments might be implemented in the real world. 

They won’t, for example, know about agile methodologies. 

They may not know how to work with the UX designer, how to design features that are engaging to users, how to iterate through different versions of something. 

The applied aspect — that connection to the real world, and ultimately the users — is what’s lacking in most candidates.

Julian: Could you discuss disruption in the field of applied behavioral science and how to be positioned against it?

Sekoul:  There is a shift happening in applied behavioral science towards thinking about it as evidence-based design.

For some context, applied behavioral science is something that started off probably about `5 to 20 years ago. What happened was that the field of economics combined with psychology, forming what is known as behavioral economics. And this behavioral economics work started being more and more prominent with time, especially when it is applied in the organizational context. 

People have focused more and more on this. They’ve taken insights from behavioral economics and tried to translate them into real-world change, which has had some success. And a lot of what behavioral science has done so far is based on insights from behavioral economics.

Julian: Are there pitfalls from focusing too much on the granular details?

Sekoul: One potential pitfall for people who are starting off in applied behavioral science is to focus too much on a particular field, and to think that the field is what’s delivering the value they’re seeking to deliver as an applied behavioral scientist. 

And what evidence-based design means is less where the insights come from, and more about the ability to use the scientific method, to create interventions and then test them in the real world.

As the field becomes more agnostic to the bodies of knowledge that it might leverage, the skill set that’s needed from a person will be less reliant on a particular sub-field of behavioral science as well.

In order to respond to this trend, anyone looking to break into this field and stay in it for a long time should think about detaching from any particular fealties to neuroscience or psychology or sociology, for example, and think about what evidence-based approaches in design really look like.

Julian: What kind of organizations an applied behavioral scientist can work for?

Sekoul: The value of applied behavioral science is really dependent on the extent to which the behavior change goals can be shifted. 

Some behaviors can only be shifted through systemic change. It’s important to identify whether the company offers an environment that lends itself well to the kinds of insights that behavioral science provides. Typically, what that means in the real world is essentially looking at companies where behavioral science is able to add value to the lives of users or staff inside.

A good test for that might be to think about what kinds of goals a company has and then to see, to what extent applying empathy, understanding users better and designing with their preferences, can be augmented through behavioral science. 

Take banking, for example. There’s an enormous upside for somebody applying behavioral science in banking, because the kinds of decisions that go into an interaction with a bank are typically affected by a lack of information, a lack of understanding on the consumer side, a lack of resources in terms of time, in terms of the money, or a need to make certain decisions. 

Behavioral science can alleviate or unblock a lot of the barriers that might be present for good decisions.

In other companies, however, the kinds of change that you would need could be systemic. And in those cases, an applied behavioral scientist might be more focused on things like marketing rather than actually helping consumers make better decisions. 

And if you wish to apply behavioral science ethically, you should go towards companies that are able to actually improve the lives of those that they’re targeting, as opposed to just focusing on KPIs.

Julian:  Okay, to conclude, what would you say are the 3 most important pieces of advice for becoming an applied behavioral scientist?

Sekoul: Number one: Think very early on about the kind of change that you’d actually want to see happen in your career. 

More specifically, think about what kind of behavior change you actually want to create and what kinds of issues in the world matter to you. That will allow you to think more deeply about what fields would be relevant, what methodologies, even things like what statistical tasks or what software you might want to use. Everything should start with this idea of a cause or something in the world that you want to tackle in some way.

Number two: Once you have an idea of what your specific cause might be, think about all the practical implementations that are associated with it. 

So for example, if you are interested in something like fitness, you might think about what particular tests are involved. 

What types of interventions might be useful in raising the fitness level of someone? What are the most common pitfalls? 

That will then allow you to think more critically about practical implementations of behavioral science principles, and to think about what kinds of tools and interventions you might want to design. And so within that category, actually getting your hands dirty and designing tools and testing them in the real world and something that should happen as early as possible, as often as possible, because it’s ultimately what you’d be doing on the job.

Number three: Stay field agnostic, and approach problems with an open mind. 

If you’re looking to create behavior change in fitness, for example, just because you’ve studied psychology, it doesn’t necessarily mean that you should be tapping into insights from psychology to achieve your goal. 

You could, for example, tap into insights from AI, you could use things like organizational behavior, etc. You could look at exercise science. There are many fields from which your insights might be pulled.

Think about what types of products have worked in the past. What types of features of them successful in different products is something that’s a very useful exercise. 

Ultimately, a good applied behavioral scientist is somebody who is able to pull intervention ideas to open design opportunities in a wide variety of fields, pull them all together and then test them in the real world. 

So becoming field agnostic, rather than being attached to a particular set of insights, is very important early on — and even more important later on.

Julian: Thank you for sharing your expertise, this has been really insightful and I hope that many future behavioral scientists will take this advice to heart.