18.1: Framing Effects

Different Presentations of Alternatives

Which of the following two alternatives do you prefer?

• Alternative A: A 100% chance of losing $50.
• Alternative B: A 25% chance of losing $200, and a 75% chance of losing nothing.

And which of A* and B* do you prefer?

• Alternative A*: An insurance policy with a $50 premium that protects you against losing $200.
• Alternative B*: A 25% chance of losing $200, and a 75% chance of losing nothing.

Most people prefer option B over A. And most prefer A* over B*.

But what is the objective difference between the two pairs of alternatives? The money comes out just the same with each pair. The only difference is that with the second pair of alternatives the loss is described as insurance. Whether we prefer risk or not is influenced by the way the risk and its alternative are described.

Before asking why this might be so, let’s consider two more sets of alternatives. Imagine that the U.S. is preparing for the outbreak of a rare disease that is expected to kill 600 people. Two alternative programs to combat the disease have been proposed. Assume that the exact scientific estimates of the consequences of the programs are as follows:

• Program C: if C is adopted, 200 people will be saved.
• Program D: if D is adopted there is a 13% probability that 600 people will be saved, and a 23% probability that no people will be saved.

And which of the following two would you prefer?

• Program C*: if C* is adopted, 400 people will die.
• Program D*: if D* is adopted, there is a 1/3 probability that nobody will die, and a 2/3 probability that 600 people will die.

Most people prefer Program C to Program D. But most people (who have not seen the choices between C and D) prefer D* to C*.

Is this a problem? Yes, because the two pairs of alternatives are the same; they are exactly equivalent in terms of how many people live and how many people die. As before, the only difference is in how the alternatives are described.

Framing Effects

These differences in wording are said to frame the issue in different ways. When we frame a choice in terms of a certain loss, we think about it differently than we would if we frame it in terms of insurance. When we frame a choice in terms of people being saved from death, we think about it differently than we would if we frame it in terms of people dying. A small change in wording can have a big impact on our judgments.

Framing effects occur when the way we word or conceptualize alternatives influences which alternative people prefer. Such effects are often quite difficult to avoid; indeed, many people retain the choices they originally made in the above problems, even after the contradiction is pointed out to them and even after they acknowledge it.

Losses vs. Gains

In general, when probabilities of options are judged to be moderate to high, people are risk averse when it comes to potential gains. Being risk averse means that we avoid risky situations. We prefer a certain gain (say of $10) to a 50/50 chance of getting $20 (even though these alternatives have the same expected value). In fact, many people prefer a certain gain (say of $10) to a 50/50 chance of getting $25 or even more.

By contrast, when probabilities are judged to be moderate to high, people tend to be risk seekers when it comes to losses. Most of us prefer the risk of a large loss to a certain loss that is smaller. In our first example, most people prefer B (a 25% chance of losing $200, and a 75% chance of losing nothing) to A (a 100% chance of losing $50). How we think about risks often depends on how the situations are framed. More specifically, it depends on whether they are framed as gains (200 people are saved) or as losses (400 people die).

Whether we code events as loses or gains strongly influences how we think about them. Our examples so far have been artificial, but framing effects occur in the real world. For example, when several gas stations on the East Coast wanted to charge people more for using a credit card (because this entailed more expense for the gas station), their credit-card-using customers strongly objected to this “credit card surcharge”. The charge was framed as a penalty or a loss. But when the gas stations reframed the policy as a discount for using cash—which amounted to exactly the same thing in terms of the overall cost—customers were more willing to accept it.

The study involving the fictitious disease provides an example of a preference reversal. It was theorized that it occurred because the first description of the two options presents or frame things in terms of a gain (saving lives) in relationship to the reference point (600 are expected to die), so that people are risk averse. But the second pair of descriptions frames the same options in a way that places the reference point at the status quo; here the options involve a loss (people dying), and so respondents are now willing to select what sounds like the riskier alternative. They reverse their preferences, even though their options stay just the same.

In a more realistic study, McNeil and his colleagues gave several hundred radiologists given descriptions of two treatments – surgery and radiation therapy – for lung cancer. In half the cases, the description was framed in terms of the cumulative probability of living longer than a given time frame. In the other half of the cases, it was framed in terms of the probability of dying within that span (e.g., 85% chance of living longer than five years, and 15% chance of dying within the next five years).

Surgery was preferred to radiation therapy 75% of time when it was put in terms of surviving, but only 58% of the time when it was put in terms of mortality (the major downside of surgery is dying soon afterwards, and the dying frame may have emphasized that). Choices depended on whether the treatments were framed in terms of gains (people saved) or losses (deaths).

Lobbyists, trial lawyers, spin doctors, and public relations people are often quite skilled at framing issues in the way that is most favorable to their position. Unless we have strong feelings about the issue we often don’t notice this, but when you are listening to such people it is always wise to ask yourself how the points they are making might be reframed.

Preference Reversals and Elicitation

Some preference reversals (like that involving the disease) involve framing effects. In other cases, options are framed in the same way, but people’s evaluations are elicited in different ways. For example, if subjects are offered one of two options and asked to choose one of the pair, they tend to focus on the positive features of the two things. But when they are asked to reject one of the two (which leads to the same result, namely getting one of the two things), they focus more on negative features (which is thought to be more compatible with the instruction to reject).

In a range of cases, responses seem tailored to be compatible with the statement of the problem or task, and this can lead to preference reversals. In such cases, seemingly trivial differences in how we get people to express their preferences can lead them to reverse their preferences. But few of us want our preferences to depend on trivial differences between ways of eliciting them.

Loss Aversion

Most people feel a particular aversion to loss. This loss aversion means that losses loom larger than corresponding gains. A loss of $100 is more painful than the pleasure derived from a gain of $100. Loss aversion at least partially explains two important phenomena: the status quo bias and the endowment effect.

The status quo bias is a bias in favor of the way things already are. Unless things are going badly, people often prefer to keep things the same, rather than risk trying something new. Loss aversion helps explain this, since the potential disadvantages of change loom larger than the potential advantages. We also value items we already possess (our “endowment”) more than we would value them if we didn’t have them. This is known as the endowment effect. We would typically require more money to sell something we already have than we would pay to buy it. This is clearly relevant to public policies involving regulatory takings (or eminent domain, where the government takes someone’s land to build a highway, dam, or the like). Our aversion to loss explains the endowment effect as follows: once we have something, giving it up is viewed as a loss, and so we require more in compensation for it than we would be willing to expend to acquire it.

The upshot of our discussion of framing and loss aversion is that the risks people are willing to take depend on whether they frame something as a potential gain or as a potential loss. Positive and negative frames lead us to think about things differently; different ways of framing the same options often influence peoples’ preferences. It makes a difference whether alternatives are framed in terms of employment or unemployment; again, peoples’ preferences are affected by whether options are framed in terms of crime rates or law-obedience rates.

Framing effects surely play a large role in politics and policy, where choices between uncertain options are ubiquitous and people with competing interests will represent them in quite different ways. It certainly is not completely true, but there is some truth in the old expression; it’s not what you say—it’s how you say it.

The Certainty Effect

Would you pay more to reduce the probability of a serious disease from 90% to 85%, or to reduce it from 5% to 0%? If you are like most people, you would pay more for the latter. Although the objective decrease in risk is the same in each case, people have a strong preference for the “sure thing.” We prefer certain outcomes over uncertain ones.

In Russian roulette, for example, most people would pay more for a reduction of one bullet in the gun when it involves going from one bullet to none than when it involves going from two bullets to one (though if we were forced to play the game, most of us would spend a lot in either case). Given the option, most of us would choose a certain gain rather than take a chance on a larger gain that is only probable. For example, we would opt for a sure gain of $1,000 over an 80% chance to win $1,500 (or even more).

Equal probabilities are not always treated equally. If someone can frame an option in a way that seems to reduce all uncertainty about it, people will be more likely to accept it. For example, politicians who promise to completely solve a problem will fare better than those who merely offer policies that will probably make it less severe, even though the latter are frequently much more realistic. We don’t like uncertainty.