Now that we’ve defined the behavioral insights approach and charted its history, this chapter aims to give a better sense of its application to real-world problems. To do this we give a handful of concise examples that vary by place and issue. We seek to contrast each example with more common approaches to solving these sorts of problems.
The examples are organized by a simple framework for different kinds of interventions. This framework starts with the three basic categories of intervention that featured in figure 2: rules, incentives, and information. We give examples of how each of these interventions can be enhanced through behavioral insights. To these, we add two other categories that are important but often neglected: interventions that change processes or the decision environment. This framework is not exhaustive, but we think that it is useful.
Here and elsewhere in the book, we discuss many of the concepts within behavioral insights (e.g., defaults, social norms, framing effects). However, we will not cover every one of these ideas and their effects on behavior; this guide is not comprehensive, but rather offers the essentials. We offer a definition of many of these ideas in the glossary.
Switching to a greener household energy plan is one way that many of us could reduce our carbon emissions. Traditional approaches to increasing switching usually involve giving us more or different information to influence our decision. Switching rates remain low, though; energy can be a difficult market to shop and the perceived hassle of doing so is high. In Germany, in the late 1990s, an energy company broadened its service to include three plans: the cheapest and least environmentally friendly one; a mid-priced but greener plan; and one that was more expensive but even more eco-friendly. Since customers had to be on one of the three plans, the starting point—the preselected default—was the mid-priced option. Customers had to actively respond to the letter if they wanted to switch to either of the other two plans.
To highlight the rule change, 150,000 letters were mailed out informing customers that they could choose any of the three plans, but that doing nothing would keep them on the preselected mid-priced option. As you might expect from what you have already read, the results were striking: After two months, 94 percent of customers stuck with the default plan,1 with the effect of automatically flipping most of a market to a greener alternative overnight. Just 4.3 percent went with the cheaper option, 1 percent chose the greenest but more expensive plan, and 0.7 percent of customers switched to another supplier. It is not just consumers who are influenced by defaults: 80 percent of countries accept treaty adjudication by the International Court of Justice when it is the default option, compared with only 5 percent when the country has to actively choose it.2
Changing system-level laws or rules is not always possible. However, we can also try to improve the rules of thumb individuals use for making decisions. For example, teaching credit-card customers to replace their existing decision processes with a simple rule of thumb—use cash if the purchase is under $20—led to an average reduction of $104 in revolving debt six months later compared to a control group.3 Or entrepreneurs can be taught simple, reliable rules of thumb to help them follow basic accounting practices—such as calculating business profits by physically separating business and personal accounts into drawers, with a simple rule about transferring between the two. An RCT found that this approach improved financial management by around ten percentage points, compared to standard accounting training.4
In 2010, Zambia’s adult HIV prevalence was 14.3 percent—one of the highest rates in the world—but demand for protective contraception was low. To promote the use of contraception, researchers led by Nava Ashraf recruited influential messengers—specifically, hairdressers and barbers—to sell female condoms. Since resources are typically limited, it was important to test whether monetary rewards would be necessary to ensure that the messengers participated or whether nonfinancial incentives might suffice. The stylists were randomized to receive one of the following: a 90 percent commission on sales; a 10 percent commission; a nonfinancial incentive, consisting of a public progress chart that documented sales and emphasized the contribution sales make to the overall health goal; or a volunteer “contract” and no reward (the control group). The condom sales of the groups were then monitored.
While the financial incentives were no more effective overall than the control, over the course of a year the sellers with the progress chart sold more than twice the number of condoms than any other group. These results hold true for at least a year, so they are not just driven by novelty. The design of the experiment allowed the researchers to show that the progress chart worked by increasing the effort expended by the messengers, rather than by increasing demand from consumers. Had all stylists been offered nonfinancial incentives, they would have sold 22,496 condoms: 11,810 more condoms than if they had all been volunteers.5 The results of this experiment show that applying behavioral insights to incentives, such as incorporating public signals of success and outcome-oriented progression charts, can increase motivation in low-resource environments, even outperforming traditional financial incentives.
Financial incentives are undoubtedly a powerful influence on behavior. However, as suggested in chapter 1, behavioral insights can also be used to design them in more effective ways. For example, in one study giving financial rewards for group outcomes resulted in greater healthy weight loss compared with rewarding individual performance: the added motivation to not let down a teammate supercharged the promise of financial gain.6
Many of the best-known examples of behavioral insights concern changing the framing of information related to a decision. For example, many tax authorities around the world have run experiments showing that changing the presentation of forms or reminder letters can substantially increase tax compliance. These changes might involve introducing new information, such as stating that the recipient is in the minority of people who have not paid, or they can focus on making existing information clearer, like clarifying exactly what someone is required to do in a particular situation.7 While these kind of general effects of a message on a whole population matter, we should also consider how specific subgroups react. For example, research by Elizabeth Linos shows that job advertisements highlighting the challenging nature of police work are more effective at attracting applicants than those that take the more traditional approach of focusing on public protection and service. But in addition to this overall difference in effectiveness, the “challenge” message is three times as effective for people of color and women applicants.8 We say more on this point in the final chapter.
Just as changing a message’s presentation can affect results, so too can changing the messenger. For example, a series of studies in the United States compared the effect of a motivational message from a beneficiary of a worker’s efforts to the same message from their team leader. These studies found that when emphasizing the positive impact of a task, beneficiaries were significantly more effective than leaders when it comes to improving productivity and performance.9 Similar messenger effects have been found in charitable giving,10 smoking cessation,11 and promoting agreement with court rulings.12 Finally, the timing of a message matters more than we often realize. Prompting drivers to wear seatbelts immediately before driving increases seatbelt use, while reminding them five minutes beforehand does not.13
Globally, there are an estimated 340 million workplace accidents each year.14 These accidents are enormously damaging to both individuals’ lives and their contribution to the economy. While some accidents are a result of poor working conditions, others stem from the behavior of workers. For example, employees in a Chinese textile factory were in the habit of throwing waste scraps of cloth on the floor next to them, creating a slipping hazard. An explanation of why this habit had formed was that workers were financially motivated to continue working without breaks. Initially, the factory tried a traditional approach to influence behavior: offering monetary incentives to workers if they put waste in trash cans. The effect was disappointing: scraps were still thrown on the floor, and the danger remained.
Sherry Jueyu Wu and Betsy Levy Paluck, researchers partnering with the factory, thought that meaningful visual cues on the floor might help change behavior. Specifically, they introduced decals depicting golden coins on the production floors. Culturally, golden coins are considered to symbolize fortune and luck, meaning the employees would have a disincentive to cover them with waste. Introducing these decals led to a 20 percent decline in waste on the floor.15 A small, contextually meaningful change to the design of the environment was enough to overcome a seemingly entrenched habit.
Received wisdom in medicine is that it’s “better to get it over with.” This ethos, coupled with efficiency considerations, means that speed is often implicitly valued in medical procedures, especially where pain is involved. In 1995, a team of physicians and academics based in Toronto was grappling with a very specific research problem: decreasing pain perception among colonoscopy patients. Their research hinged on a phenomenon known as the “peak-end rule.” This rule states that our judgment and memory is disproportionately shaped by the peak moment and the very last part of an experience. Since colonoscopies at the time typically conformed to the notion of “better to get it over with,” they inevitably induced significant discomfort in the final moments. A slower “exit,” the team hypothesized, would be perceived as less unpleasant, even though it would make for a longer procedure. The insight proved correct: patients whose colonoscopies involved an additional three minutes of slow extraction rated their overall pain to be lower and had a more favorable view of the experience.16
These brief examples show that behavioral insights can have a substantial effect on important issues. They do not, however, shed light on how these interventions are designed or the effects measured. In the next chapter we turn our attention to the practical steps of applying behavioral insights.