Framing Effect

A framing effect is a shift in preference produced by the way a decision is presented, where the alternatives themselves are unchanged. Amos Tversky and Daniel Kahneman set it out in "The Framing of Decisions and the Psychology of Choice," published in Science in January 1981, using the phrase "decision frame" for "the decision-maker's conception of the acts, outcomes, and contingencies associated with a particular choice." Their summary states the finding plainly: the principles governing how people perceive a problem "produce predictable shifts of preference when the same problem is framed in different ways." The reason this matters rather than merely amuses is the standard it violates. A rational choice is supposed not to reverse when only the wording changes, so a reversal is evidence about the reader rather than about the options. The paper is often credited to Kahneman alone; it has two authors, and Tversky is the first of them.

The best-known demonstration is a pair of problems about an outbreak expected to kill 600 people. One group chose between a program that would save 200 people for certain and a program with a one-in-three chance of saving all 600; 72 percent took the certain option. A second group chose between a program under which 400 people would die and one with a one-in-three chance that nobody would die; 78 percent took the gamble. The outcomes are the same in both versions. Only the accounting changed, from lives saved to lives lost, and with it the appetite for risk. The underlying machinery, a value function that is concave for gains and convex for losses, belongs to prospect theory; what the framing work adds is that the wording decides which of the two halves a person is standing in.

The second demonstration in the same paper is the more uncomfortable one, because the error survives being pointed out only until the two options are written next to each other. Respondents faced two simultaneous decisions, and 73 percent picked the combination of a sure gain and a loss gamble, while 3 percent picked the reverse pairing. Set out directly as a single choice between the two resulting prospects, a second group of respondents chose the pairing that 73 percent had assembled not at all, and chose the pairing 3 percent had assembled unanimously, because it wins by ten dollars in every branch. The frame had kept the two decisions in separate mental accounts, and nothing in either account revealed that one combination dominated the other.

The paper's own money example makes the same point at household scale. Respondents were told the calculator they were about to buy was $5 cheaper at another branch of the store, twenty minutes' drive away, and asked whether they would make the trip. Sixty-eight percent said yes when the calculator cost $15, and 29 percent when it cost $125. The saving is $5 either way and the drive is the same drive. What changed is the account the $5 was entered in, which is where framing and mental accounting meet.

Financial information is unusually easy to frame, because most of it can be stated correctly in more than one unit. A cost is the same money whether it is written as a percentage of a balance or as a number of dollars a year, and the two versions do not read the same way. A return is the same return whether it is quoted before or after what it cost to obtain, and only one of those numbers is the one an investor keeps. A record can be summarized as gains in seven years out of ten or as losses in three out of ten. A price gap between paying by cash and paying by card can be labeled a discount for one or a surcharge on the other; Thaler's 1980 paper records that when legislation touching the point was before Congress, the card industry's lobbying concentrated on which of those two words would be used. None of this requires anyone to mislead. It requires only that a true statement be available in more than one form, which for prices, fees and returns it almost always is. The countermeasure is correspondingly dull: convert the figure into the other unit before deciding, and do the conversion yourself.

One limit keeps the concept useful. Not every difference in wording is a frame. If one description tells the reader something the other omits, the two are not equivalent and a different choice is a response to new information. Applied loosely, "that's just framing" becomes a way of dismissing a distinction that is real.

A frame does its work in four steps. A description arrives and fixes a reference point, usually the current state or zero. Each outcome is then coded as a gain or a loss relative to that point rather than as a final position. Gains and losses are weighted differently. The choice follows from the coding, so a description that moves the reference point moves the answer without touching the options.

A worked example, using the numbers from the 1981 paper, because the arithmetic is checkable in a line. Respondents were given two decisions at once. In the first, a sure gain of $240 against a 25 percent chance of $1,000; 84 percent took the sure gain. In the second, a sure loss of $750 against a 75 percent chance of losing $1,000; 87 percent took the gamble. Combining the popular answers produces a 25 percent chance of winning $240 and a 75 percent chance of losing $760. Combining the unpopular answers produces a 25 percent chance of winning $250 and a 75 percent chance of losing $750. The second is better by $10 whichever way the coin lands. Asked to choose between those two prospects directly, a separate group of 86 respondents took the second unanimously and the first not at all, while the group that met the same choice as two separate questions had assembled the first by a margin of 73 percent to 3 percent.

A household version of the same discipline: write the number twice. If a cost is quoted as a percentage, work out the dollars it comes to on the balance it will be charged against. If it is quoted in dollars, work out the percentage. Neither figure is more honest than the other, and seeing both is what stops the presentation doing the deciding.

Pros

• The concept identifies a specific and checkable moment of exposure, which is any point where a number is being described to you rather than calculated by you.
• The countermeasure is mechanical and cheap. Converting a figure into a second unit takes one line of arithmetic and does not require any insight into your own psychology.
• It explains why two people can read the same disclosure and reach opposite conclusions without either of them misunderstanding it.

Cons

• The term gets stretched to cover any difference in wording, including differences that carry real information, which drains it of meaning.
• Knowing the effect exists provides very little protection, because the frame operates on how the option is scored rather than on what the reader knows.
• There is no neutral frame to fall back on. Every presentation of a number is a presentation, so the aim is to see more than one rather than to find the unbiased one.
• It cuts both ways in practice: a frame can be chosen to help a reader understand a cost as easily as to obscure it, and nothing in the effect itself says which is happening.