Framing Effects: How the Way You Present an Idea Changes the Answer

The framing effect is the shift in people’s choices when the same facts are described in different words: a plan that “saves 200 people” wins support that the identical plan that lets “400 people die” loses. Amos Tversky and Daniel Kahneman named it in a 1981 paper in Science.¹ When the Many Labs project repeated their problem in 2014 in 36 samples, 11 of them outside the US, framing replicated clearly, at about half the size reported in 1981.²

So the wording of a proposal, a status update or a request changes how it is received, though less than the famous version suggests. A budget can shrink or be mostly kept; a project can be late against the plan or early against last month’s estimate. Choosing between those descriptions is one of the quieter communication skills used at work, and this guide covers how the frame moves the answer and how to choose one honestly.

What is the framing effect?

The framing effect is a change in judgment or choice caused by how equivalent information is described rather than by the information itself. Tversky and Kahneman defined a decision frame as the way a person sees the options, outcomes and odds, shaped partly by how the problem is worded and partly by the person’s own habits.¹

“Framing” covers more than one thing, and mixing them up explains much of the confusion online. In a 1998 review, US psychologists Irwin Levin, Sandra Schneider and Gary Gaeth argued that studies had been testing different effects under one name, and sorted them into three kinds.³

Each kind behaves differently, so a rule learned from one (“people avoid risk when you talk about gains”) does not automatically carry over to the others. Before borrowing a framing tip, check which kind it came from.

The classic example: lives saved or lives lost

The best-known example of the framing effect is Tversky and Kahneman’s “disease problem”, in which two groups chose between the same pair of public-health programs and reversed their preferences because one group read about lives saved and the other about deaths.

The study

Moderate evidence

Same 600 lives, opposite majorities: the disease problem

Students at Stanford University (US) and the University of British Columbia (Canada) read about a disease expected to kill 600 people. When one program would save 200 people for certain and the other had a one-in-three chance of saving everyone, 72% chose the sure program. A second group read the same two programs as deaths: 400 people would die for certain, or a one-in-three chance that nobody dies. Now only 22% chose the sure program.¹

The two versions are logically identical, yet the majority flipped from playing safe to gambling. The authors described a common pattern: choices involving gains tend to be risk averse, and choices involving losses tend to be risk seeking. They reported the same reversal in other groups, including university faculty and physicians.¹ The disease problem was hypothetical, answered on paper with nothing at stake for the respondent.

Attribute framing is gentler but closer to daily work. In a 1988 study, Levin and Gaeth had consumers rate ground beef labeled either “75% lean” or “25% fat”; the lean label drew more favorable ratings, and the gap shrank once people actually tasted the meat.⁴ A label moves opinions most when people have little else to go on, so when you want a fair hearing for an idea, a demo, a pilot or the raw figures will do more than a flattering adjective.

The disease problem also appears in our guide to how cognitive biases work and which ones survived retesting, so the rest of this article stays with what the effect means when you are the one choosing the words.

Why does the framing effect happen?

The framing effect happens because people judge outcomes as gains or losses from a reference point, not in absolute terms, and because a speaker’s choice of words carries real hints about what the speaker thinks.

The first cause comes from prospect theoryprospect theory: Kahneman and Tversky's account of how people choose under risk: outcomes are judged as gains or losses from a reference point, and losses weigh more than gains of the same size. Its main patterns have held up when the original questions were run again in many countries.Full entry in the glossary, Kahneman and Tversky’s account of choice under risk. Outcomes are felt as positive or negative changes from a neutral reference outcome, and the value people place on them is S-shaped: steeper for losses than for gains.¹ Change the reference point and the same result moves from the gain side to the loss side. “Saved 200” invites you to count up from the 600 expected deaths; “400 will die” invites you to count down from nobody dying.

Levin and colleagues point to encoding: whether information is labeled in positive or negative terms shapes how people take it in, and that colors the judgment.³ “75% lean” makes you think about leanness; “25% fat” makes you think about fat.

Shlomi Sher and Craig McKenzie, psychologists at UC San Diego, argued in 2006 that equivalent frames are often not equivalent in what they reveal.⁵ Speakers tend to describe things in terms of what has increased relative to a reference point. In their first experiment, US undergraduates asked to hand over “a half-empty cup” were more likely to pick the cup that had started full than people asked for “a half-full cup”.⁵

1. Left (teal), filled up from empty: speakers lean toward calling it half full: the description follows what has increased.
2. Right (coral), poured down from full: speakers lean toward calling it half empty, so the words hint at where things started.

Dashed lines: where each glass started. Thin gray line: the level both glasses share now.

The same paper describes a 1989 study in which people playing research managers were more generous in funding a team described by its successes than one described by its failures, with the same record. Sher and McKenzie then showed that people describing a weak team were more likely to mention its failures than people describing a strong team with the same record.⁵ So a listener who reads “30 successes” as good news is not being foolish: the frame is a hint about the speaker’s view.

Framing a budget cut, a delay or a change at work

At work, framing is mostly a choice of reference point: what you compare the news to, and whether you describe what is kept or what is lost. The order in which options appear is a related influence. The illustrative examples below show each choice with a true version on both sides.

A budget cut. “Training keeps most of its budget” and “Training loses a slice of its budget” are both true. The kept frame invites acceptance; the lost frame invites a fight to win it back. If the people affected will feel the loss, name it: they will find the numbers anyway, and a sunny frame they later see through costs you credibility.

A project delay. The same new date can be “two weeks behind the plan” or “a week ahead of last month’s revised estimate”. The honest one is the date your reader committed to and planned around, with the more recent comparison added if it is genuinely useful.

A change to how the team works. Prospect theory treats outcomes as gains or losses from a reference point, often the status quo, so a new process can land as a list of losses unless you also show what people gain.¹ Describe the gain plainly, then acknowledge what people give up.

Order is not framing in the strict sense, but it works just as quietly. In a 2012 US study at a Boston train station, Dana Carney and Mahzarin Banaji found that people choosing between two near-identical pieces of gum within about a second mostly took the one placed first, while people told to think it over split roughly evenly.⁶ If you list options in a message that will get a fast reply, put your recommendation first and say that it is your recommendation, rather than letting position do the persuading.

Do not assume loss frames always hit harder, either. A 2007 meta-analysis of 93 studies of disease-prevention messages by Daniel O’Keefe and Jakob Jensen, known to us from its abstract, found gain-framed and loss-framed appeals almost equally persuasive.⁷ Pick the frame that is accurate and relevant to the reader.

How strong is the framing effect, and when does it fade?

The framing effect is reliable but modest, and its size depends heavily on how the choice is set up. A large retest and two pooled analyses agree that it is real and smaller than the 1981 numbers suggest; whether expertise weakens it is still unsettled.

The Many Labs project, whose first authors were Richard Klein and Kate Ratliff, replaced “the United States” in the problem with each sample’s own country. Neither the setting (lab or online) nor the location (US or elsewhere) changed the results much.² Framing is not an American classroom curiosity. An effect of this size works as a nudge, and wording is most likely to matter when two options are close.

The pooled studies point the same way. Anton Kühberger, at the University of Salzburg in Austria, combined 136 papers in a 1998 meta-analysismeta-analysis: A study that combines the results of earlier studies on the same question into one overall estimate. Pooling makes the estimate more precise, but it cannot repair the studies it pools: a meta-analysis of surveys is still survey evidence.Full entry in the glossary and found a small-to-moderate overall effect with large differences between designs; among the features that mattered most was whether a study shifted the reference point or simply made one outcome more prominent.⁸ A 2018 reanalysis by Alexander Steiger and Kühberger, correcting for the bias toward publishing significant results, put the effect higher and close to the Many Labs estimate, and found that its size changes with how the framing task is built.⁹ Both summaries rest on the papers’ abstracts.

Whether knowing a field protects people is less clear. In a 1995 US study of internists, residents and medical students given 12 hypothetical clinical cases, framing changed decisions on only 2 cases for the doctors and none for the students, according to the abstract.¹⁰ Yet Kühberger’s meta-analysis found no clear difference between studies of students and studies of other groups,⁸ and in James Druckman’s 2004 experiment, described below, expertise reduced framing only when people also discussed the problem.¹¹ Treat expertise as no guarantee: give specialists the same honest frame you would give a newcomer.

How can you avoid being influenced by framing?

The best-tested way to blunt a framing effect is to hear the other frame too. In a 2004 US experiment with 580 participants, most of them students, political scientist James Druckman found that giving people both versions of a problem removed the framing effect, and that discussion with people who had read the other version shrank it.¹¹

Druckman’s explanation is that the second frame prompts more deliberate thinking and offers another way of seeing the problem.¹¹ The train-station study points the same way for the order of options: people told to think it over showed no pull toward the first one.⁶ On your side of the desk, this turns into a few habits:

• Restate any key number in its opposite form before deciding: if you are told a plan is 90 percent on time, say aloud that it is late one time in 10.
• Ask what the comparison is: compared with what, and who chose that baseline?
• For a decision that matters, discuss it with someone who saw a different summary before you commit.

Framing honestly: where a good frame turns into spin

Framing is unavoidable, so using it ethically means choosing a frame that is true, relevant to the reader’s decision and one you would defend if they saw the alternative. It turns into spin when the reference point is picked to hide the comparison the reader would care about.

Persuasion through framing is not suspect in itself; the question is what the reader would conclude if they had both frames. It is the same test that closes our guide to ethos, pathos and logos: would the reader feel informed or handled if they saw how you built the message? Because frames carry information about the speaker’s view, a frame you would not stand behind is a small misstatement, even if every number in it is correct.⁵

Three practical lines help. Keep the reference point the reader cares about, usually the original plan or last year’s actual figure. Give both versions of any number a decision rests on, which is also what the counter-framing evidence suggests protects listeners.¹¹ And state your recommendation as a recommendation rather than burying it in an order or a label.

Before you send a framed proposal or update

• The comparison point is the one my reader planned around
• Any key number is also given the other way round
• What people lose is named, not only what they keep
• My recommendation is labeled as mine, not hidden in the order
• The reader has time to think before a decision that matters

If an item fails, the fix is usually one added sentence.

The bottom line

The framing effect has replicated across countries but works as a nudge, strongest when people have little else to go on and weaker when they hear both sides. Your frame tells your reader what you think, so pick the reference point they care about, name the losses as well as the gains, and give them the other frame yourself when a decision matters.