In 1981, Daniel Kahneman and Amos Tversky presented subjects with a choice about a hypothetical disease expected to kill 600 people. Two versions of the problem, identical in outcomes, produced opposite majorities. The framing effect they documented became one of the most cited findings in behavioural science and a cornerstone of the case against strict rational-choice models.

The Disease Problem That Split Two Minds

The task looked simple. Subjects were told that a disease outbreak would kill 600 people and that two programs were available. In the gain frame, Program A would save 200 people for certain, while Program B carried a one-third chance of saving all 600 and a two-thirds chance of saving none. Most subjects chose the certain option.

In the loss frame, the same outcomes were described differently. Program C would result in 400 deaths for certain, while Program D carried a one-third chance that nobody would die and a two-thirds chance that all 600 would die. Now most subjects chose the gamble.

The study drew on roughly 150 respondents, mostly university students, across several versions of the problem. The tension with rational choice theory was direct: if preferences depend only on outcomes, the two frames should produce the same choices. They did not.

How the Asian Disease Problem Worked

Kahneman and Tversky built the problem around a single manipulation. The gain frame described the outcomes in terms of lives saved. The loss frame described the identical outcomes in terms of lives lost. The numbers were arithmetically equivalent: saving 200 people for sure is the same as losing 400 people for sure.

In the gain frame, most respondents picked the certain option. In the loss frame, most picked the gamble. The reversal was not a small shift in the margin. It was a change in the majority preference, produced by wording alone.

The result mattered because it showed that the description of a choice, not just its consequences, could determine behaviour. A decision that looked risk-averse under one description looked risk-seeking under another. The 600 lives were the same in both versions.

Prospect Theory Behind the Reversal

The reversal made sense within prospect theory, which Kahneman and Tversky published in 1979 in Econometrica. The theory proposed that people evaluate outcomes as gains or losses relative to a reference point, not as changes in final wealth. Losses also loom larger than equivalent gains, a pattern now called loss aversion.

In the gain frame, the certain option of saving 200 lives sits in the domain of gains. People tend to be risk-averse there. In the loss frame, the certain option of 400 deaths sits in the domain of losses. People tend to be risk-seeking there, because the gamble offers a chance to avoid any loss at all.

The reference point does the work. Change the starting point from which outcomes are measured, and the same objective facts move from the gain side to the loss side. Prospect theory explained the flip without requiring any change in the facts.

The Paper That Made Framing Famous

Kahneman and Tversky published the framing experiments in 1981 in Science under the title 'The Framing of Decisions and the Psychology of Choice.' The paper has been cited many thousands of times across psychology, economics, medicine, and public policy. It became the standard demonstration that preferences are constructed, not merely revealed.

Kahneman received the 2002 Nobel Memorial Prize in Economic Sciences, shared with Vernon L. Smith. The Nobel citation referenced the framing work and prospect theory. Amos Tversky had died in 1996, so the prize recognised a collaboration that had already ended.

The paper's influence spread because the demonstration was easy to reproduce and hard to dismiss. A short questionnaire produced a large reversal. That combination made it a teaching staple and a target for critics.

What Replications Say and Do Not Say

Framing effects of this kind have been observed in many student samples and in professional groups, including physicians and policymakers. The direction of the effect is robust in the original paradigm. The size of the effect, however, varies widely across contexts.

Some larger, pre-registered replications have found smaller shifts than the original studies reported. The effect does not always flip a majority. In some settings it moves a modest fraction of respondents. That is a real finding, and it complicates any claim that framing always dominates.

Boundary conditions remain an open question. Framing effects appear stronger when the choice is hypothetical and weaker when people have experience with the domain. The same site has argued in a related piece on archived code and supernova results that replication depends on hidden analysis choices, and that lesson applies here too.

Framing in Medical and Policy Decisions

Medical decisions provide a natural laboratory for framing effects. Consider a hypothetical treatment that reduces the risk of a heart attack from 10 percent to 5 percent over five years. Described as a 50 percent relative risk reduction, it sounds substantial. Described as an absolute risk reduction of 5 percentage points, it sounds more modest. Both descriptions are accurate, yet they can lead to different choices.

Studies with physicians have found that they are not immune to this effect. In one classic study, physicians were more likely to recommend a treatment when its benefits were framed as survival rates rather than mortality rates. The framing influenced not just hypothetical choices but actual clinical recommendations, though the magnitude varied by specialty and experience.

In public policy, the framing of statistics can shape support for programs. A policy that saves 200 lives out of 600 at risk may be described as saving 200 lives or as leaving 400 to die. The two framings can lead to different levels of public support, even when the underlying facts are identical. This has implications for how risks are communicated in areas from public health to environmental regulation.

These examples show that framing effects are not confined to the lab. They appear in high-stakes decisions where the consequences are real. The challenge is to recognise when a frame is influencing judgment and to seek out alternative descriptions before committing to a course of action.

Framing in Financial and Consumer Choices

Beyond medicine and policy, framing effects shape financial decisions. A credit card that charges a 1.5 percent fee on foreign transactions may be framed as a surcharge or as a discount for using domestic cards. The two descriptions can lead to different perceptions of cost, even though the amount paid is the same. Similarly, a mutual fund that advertises a 5 percent annual return may be framed as a gain or as a loss relative to a 7 percent benchmark, influencing investor satisfaction and subsequent choices.

In consumer behaviour, the classic example is ground beef labelled as 75 percent lean versus 25 percent fat. Studies have found that consumers rate the former more positively, even though the two labels describe the same product. This framing effect extends to product pricing, where a discount framed as a percentage off versus a dollar amount off can lead to different purchase intentions. The size of these effects varies, but the direction is consistent: the frame matters.

These examples highlight that framing is not limited to life-and-death scenarios. It permeates everyday decisions, from retirement savings to grocery shopping. Recognising the frame can help consumers make more deliberate choices, though it does not eliminate the effect entirely.

Using Framing Awareness in Practice

Framing awareness is not a cure for bias, but it gives a reader a few concrete moves. The goal is to slow down the moment of preference long enough to see how the options were described.

• Notice whether the options are presented as gains or as losses, and write down which side of the reference point you are on.
• Restate the same choice in both directions before deciding, converting 'saved' into 'lost' and back again.
• Ask for absolute numbers rather than percentages alone, since a 50 percent reduction of a small base can sound larger than it is.
• Check the base rate behind any framed statistic, because a relative risk change without a baseline is hard to act on.
• Treat your first preference as frame-sensitive and revisit it after the reframing exercise.

A related piece on the redshift drift test makes a similar point about measurement choices shaping conclusions. The objection from some economists is that these lab demonstrations say little about market behaviour, where prices and repetition discipline choices. That objection has force in settings with feedback. It has less force in one-off medical or policy decisions.