Cognitive biases
The framing effect: why the same numbers sound different
You are choosing ground beef for burgers. One package says in big letters: “75% lean.” The one next to it says: “25% fat.” Your hand reaches for the first one on its own. Yet both packages hold the same thing: out of every 100 grams, 75 are lean meat and 25 are fat.
This is the framing effect: the same information affects choices differently depending on how it is worded. The fact stays the same; only the “frame” changes: what you choose to talk about, the good part or the bad one, the people saved or the people lost.
Below, we look at the classic experiments, where the frame shows up in everyday life, and how to see the whole number. In the middle of the article there is a short game with a package of ground beef.
Contents
In short: how the framing effect works
What the framing effect is
The best-known experiment was described by the psychologists Amos Tversky and Daniel Kahneman in a 1981 paper. Students were asked to imagine a made-up epidemic: 600 people are expected to die. There are two programs to fight it.
The first group got the programs described like this. Program A: 200 people will be saved. Program B: there is a one-in-three chance that all 600 will be saved, and a two-in-three chance that no one will be saved. Most people (72%) chose the sure Program A.
The second group got the same programs in different words. Program C: 400 people will die. Program D: there is a one-in-three chance that no one will die, and a two-in-three chance that all 600 will die. Now most people (78%) chose the risky Program D.
But A and C are the same thing: out of 600 people, 200 survive and 400 die. B and D are the same, too. The only difference is in the words: in the first case people spoke about those saved, in the second about those who die. The authors write that they saw the same reversal among university faculty and among physicians.
The epidemic experiment is about a choice involving risk. There is also a simpler kind of framing: the same feature of a thing is described from its good side or its bad side. “75% lean” or “25% fat.” “The glass is half full” or “half empty.”
Irwin Levin and Gary Gaeth ran such an experiment in 1988. Participants rated ground beef. Some were told it was “75% lean,” others that it was “25% fat.” The beef labeled “75% lean” was rated better. And when the participants tasted the meat, the difference got smaller: their own experience diluted the effect of the label.
How reliable the effect is. For problems like the epidemic there is a large review by Anton Kühberger: 136 papers, almost 30,000 participants. The effect turned out to be reliable, but small to medium in size, and it depends strongly on how the problem is built. A recheck in 2018 took into account that failed experiments are published less often, and put the effect even higher. We do not cite such a review for labels like “lean” and “fat.”
Try it yourself: a package of ground beef
The game takes a couple of minutes. Here is what will happen.
Step 1. You are in a store, choosing ground beef for burgers. The package has one sticker. Some players get one label, others get the other. The computer picks which one you see. The beef itself is the same for everyone. You rate it with a slider, and the package moves toward the basket or back to the shelf.
Step 2. You turn the package over. On the back is a second sticker. You rate the beef again, now knowing both labels.
Step 3. You change how much fat is in the package and watch how both labels change together. At the end we show what the trick was.
Why do players see different stickers? The 1988 experiment was built the same way: some participants saw one label, others saw the other, and the meat was the same. If the rating depends on which label you got, that is the framing effect. The second sticker on the back serves a different purpose: so that you can see both sides of one number yourself.
What usually happens. In the study, beef labeled “75% lean” was rated better than the same beef labeled “25% fat.” When participants tasted the meat, the difference shrank.
What the game can and cannot show. You gave your first rating seeing only one label. How you would have rated the same package with the other sticker cannot be known from one person: you gave the second rating knowing both. The effect shows up when you compare the first ratings of people who got different stickers. But if your rating moved when you turned the package over, that is a hint: perhaps you felt the frame at work on you.
Why the frame changes the choice
Picture the same painting in a gilded frame and in a plain wooden one. The painting is the same, but the impression is different. This is an image, but it captures the point well: the frame does not change the fact, it changes how the fact looks. Usually we see only one wording and do not know how the fact would look in another frame.
Words carry feelings. “Lean” sounds like a benefit, “fat” like a harm, “saved” like hope, “will die” like a disaster. Levin and colleagues believe that a good or bad presentation changes how we perceive and remember the information itself. Scientists are still discussing exactly how this happens. But the direction of the shift is steady: a good label, a better rating.
There is an honest caveat, too: a frame can be a hint. If a seller writes “75% lean,” they report not only the number but also that they themselves count it as a plus. In a study by Craig McKenzie and Jonathan Nelson, people said “half full” about a glass that had been empty before, and “half empty” about one that had been full. And listeners, from such a phrase, correctly guessed what the glass had been like before. Reacting to a hint is not entirely foolish. The mistake begins where the wording replaces the number itself.
Where it shows up
Shopping. “97% fat-free” on a yogurt and “3% fat” are the same thing. The manufacturer will almost always choose the first label. It is not deception: both are true. But on the shelf you see only one, and it has already been chosen for you.
Health. In a 1982 experiment, patients with chronic illnesses, students, and physicians were asked to imagine that they had lung cancer. They had to choose between surgery and radiation therapy. The treatment data were the same. Among other things, what changed was how the data were presented: as a chance to survive or as a chance to die. In all three groups, surgery looked more attractive when the results were described as a chance to survive. Physicians did not resist the frame either. We do not give the numbers from this experiment. An illustrative pair: “out of 100 people, 75 survive” and “out of 100 people, 25 die.” It is the same number, but it sounds different.
News. Imagine two headlines about the same poll: “One in five city residents bikes to work” and “Four out of five city residents do not bike to work.” It is the same share. The first sounds upbeat, the second gloomy. The author of a text picks the frame to fit the idea they want to get across.
How to see the whole number
The framing effect is one of the biases on the cognitive bias map, in the section on decisions. Next to it is the anchoring effect: there, the first number we see sets the frame.
Frequently asked questions
How is the framing effect different from loss aversion?
Loss aversion means that in decisions a loss weighs more than a gain of the same size. The framing effect means that the same fact can be described as a gain or as a loss, and that changes the choice. The two are connected: Tversky and Kahneman explained the reversal in the epidemic problem by saying that people treat risk differently when it comes to gains and when it comes to losses.
So is the “75% lean” label deceptive?
No. Both labels are true and say the same thing. The manufacturer simply picks the one that sounds better. It helps to remember that the choice has already been made for you, and to read the second label in your head.
If you know about the effect, does it stop working?
Not completely. Tversky and Kahneman wrote that people usually do not notice the other wording or how it could have changed their choice. When you see both sides at once, it is easier to rethink the decision. That is why the habit of flipping the phrase works better than the knowledge alone.
Is a frame always a trap?
Not always. A frame can also be a hint: if a seller writes “75% lean,” they are telling you what they themselves count as a plus, and listeners can work out what the starting point was. The mistake begins where the wording replaces the number itself.
