Finding out your yearly bonus is $100 more than you had thought it would be is a pleasant surprise—but finding out it is $100 less is not just a bit of a letdown, it is infuriating. People tend to be more upset by a loss than they are pleased by an equivalent gain, a phenomenon known as loss aversion.
Social scientists have shown time and again that loss aversion happens in all sorts of situations, from yearly bonuses and supermarket sales to life-and-death medical decisions. We hate to lose what we have, or expect we will have, far more than we like gaining something extra. As well established as loss aversion is, however, it remains unclear what mental processes are behind it.
“Loss aversion is usually looked at as an outcome phenomenon,” says Ulf Böckenholt, a professor of marketing at the Kellogg School of Management. “You basically just see that people find losses unpleasant and try to avoid them. But there’s very little work on what triggers this reaction.” Illuminating the processes that underlie loss aversion could not only help researchers better understand how we make decisions, but also help us frame decisions in ways that would help us choose the option that is best for us, circumventing our instinctive loss aversion—or using it to our advantage.
Need for a New Model
Böckenholt and his colleagues Martijn Willemsen, an assistant professor at the Eindhoven University of Technology, and Eric Johnson, a professor at Columbia Business School, set out to investigate what drives loss aversion. They found that the usual model of loss aversion does not fully explain how people make decisions—and that, rather than tallying everything up as a gain or a loss right away, people are constantly comparing options and seeking out information that supports their inclinations.
That usual model is a process called value encoding: When a decision-maker first sees an option, he or she immediately compares it to a reference point—how much money they have, say, or what the price of an item usually is—and judges relative to that whether it is a loss or a gain. When it is time to make a decision, the person picks the option that, essentially, feels best compared to the reference point, with the sense that they have gained the most, or lost the least.
Another potential mechanism is called value construction. In this process, people do not just store, or encode, a value for each option right away. Instead, they build up a preference as they learn about the different options available. Decision-makers do not objectively assess their options, however: each new piece of information biases what information they look for next, and what they make of it. Earlier studies of value construction have shown, Böckenholt says, that “people try to find, as soon as possible, something that stands out. And then after they’ve found an option that stands out, they more or less try to confirm it” by seeking out supporting information. This can be useful, if they have honed in on the right option early on, or disadvantageous, if they are overlooking something better.
Böckenholt and his colleagues predicted what might happen if each of these processes led to a decision. If value encoding was at work, the decision-maker might tend to make comparisons with the reference point early on, rather than comparing options to one another, and would weight losses more heavily than gains. If value construction was behind the decision, on the other hand, the order of the options could impact which one the decision-maker landed on as the early leader—and thus influence both what information they looked for and what they ultimately chose.
Facing Difficult Decisions…
The researchers first looked at how people make risky choices, decisions where at least one of the options is a probability—presenting likelihoods of certain outcomes—rather than a sure thing. The classic example of risky choice is what is known as the Asian disease problem: the United States is preparing for an outbreak of a rare Asian disease, the decision-maker is told, that will kill 600 people if left unchecked. There are two treatment options. If the options are framed as gains, the decision-maker is told that Option 1 will definitely save 200 lives, while Option 2 has a one third chance of saving 600 lives and a two thirds chance of saving none. If the options are framed as losses, 400 people will surely die if the decision-maker chooses Option 1, while there’s a two thirds chance 600 people will die and a one third chance 200 will die if they go with Option 2. The numbers are the same, either way; what is different is that saving 200 people sounds a lot better than leaving 400 to face certain death.
The 232 volunteers in the study were seated in front of a computer screen and given this scenario along with a similar one about possible layoffs. To see each piece of information about the options, they had to mouse over it—letting the researchers record what information the participants looked at when. The researchers switched up the order of the options between participants, as well as the order of the scenarios and the loss or gain frames, to suss out whether order affected how decisions were made.
Overall, the results of the study suggested that value construction was playing a big role in the decision-making process. The decision-makers did not generally focus on the reference point less as time went on, as a value encoding process would suggest, and they showed a tendency to focus more on the option they eventually chose as time went on, an indication that they were likely honing in on it, comparing it directly to the other options.
The order of the options, too, had a big effect on choice. “The way these choices are represented, the order in which this information is processed, has a tremendous influence on what people end up choosing,” Böckenholt says. “That should not matter at all in value encoding, because there you look at each option separately.” In value construction, however, where decision-makers tend to stick with an option they see early on, order can have a strong impact.
The researchers found, too, that they could predict a person’s eventual choice from the information they looked at early on—a pattern that is seen in all sorts of decision-making. “I think it’s true, in general, that decision-makers make up their minds much earlier than you’d think on the basis of their behavior,” Böckenholt says. “They may still be looking and checking things out, but actually they already know what they want.” This sort of data mining could be applied to many sorts of scenarios, he says. “If you look how people browse [online], then you want to predict more or less which product are they going to buy, this result…suggests that very early browsing behavior already is sufficient to predict which product, let’s say, a person may end up choosing.”
…And Not-So-Difficult Decisions
It is not just in life-or-death scenarios that people shy away from losses, and Böckenholt and his colleagues wanted to investigate whether the same mental processes kicked in during other types of decision-making. This time, they presented 202 decision-makers with consumer choices: say, whether they wanted to buy Printer A, which was more expensive but faster, or Printer B—less costly, but slower. The decision-makers were shown one of two reference points: a printer that was yet more expensive and really fast, or one that was the cheapest and the slowest.
The researchers again could predict a decision-maker’s eventual choice from how they perused the options early on. And once more, the decision-making process suggested that value construction, not value encoding, was the primary driving force: participants did look at the reference point early on, but more so if an option framed as a loss was presented first, suggesting that order matters, too. Decision-makers tended to choose the option more similar to the reference—and they tended to keep looking back it, suggesting it was the favorite option, not the reference point, they were comparing things to.
While these two accounts often are presented as mutually exclusive—you are either encoding value or constructing it—the truth could be somewhere in between. In fact, these two studies suggest that the two mechanisms seem to be important at different stages of the decision-making process.
“There’s some initial support for value encoding, but for later stages it becomes more a constructive process,” Böckenholt says. “So value encoding more or less accounts for the first 10 seconds—or at least we can’t say it doesn’t—but then for the rest, it’s value construction.” Comparing options to the reference point may help decision-makers pick an early favorite, at which point a value construction mechanism kicks in.
This clearer understanding of the processes in play could eventually help researchers, marketers, or policy-makers design scenarios that would make decision-makers more or less loss-averse. “For example, we show the comparisons between safe and risky options in the first study is predictive of choice,” Böckenholt says. It is quite easy for participants in the study to toggle back and forth between options. But making the comparison harder—if they were in different stores, for instance—could change people’s choices. “If you make this comparison easier, or if you make it more difficult, then it will also affect the final choice.”
Related reading on Kellogg Insight