How Framing Changes Risk Perception

In a landmark 1981 study published in Science, psychologists Amos Tversky and Daniel Kahneman asked people to choose between two medical treatment programs for a disease expected to kill 600 people. When the options were described in terms of lives saved, most people chose the safe, certain option. When the exact same mathematical outcomes were instead described in terms of lives lost, most people switched to the risky option, despite the underlying probabilities and outcomes being completely identical. This is the framing effect, and it shapes decisions in hospitals, boardrooms, and grocery aisles every day.

The exact wording that flipped people's choices

In Tversky and Kahneman's original experiment, one group of participants was told that Program A would save 200 people for certain, while Program B carried a one-third probability of saving all 600 and a two-thirds probability of saving no one. A large majority chose Program A, the certain option, even though Program A and Program B have identical expected values (Tversky & Kahneman, 1981, Science). A second group received options with mathematically identical underlying numbers but different wording: Program C would result in 400 people dying for certain, while Program D carried a one-third probability that no one would die and a two-thirds probability that all 600 would die. Despite Program A and Program C describing the exact same outcome, 200 saved and 400 dead out of 600, the majority in this second group now preferred Program D, the risky option.

This reversal, choosing the safe option when outcomes are framed as gains but choosing the risky option when the identical outcomes are framed as losses, became one of the most replicated findings in behavioral decision research. It formed a central pillar of prospect theory, the framework Kahneman and Tversky introduced in a 1979 paper in the journal Econometrica titled "Prospect Theory: An Analysis of Decision Under Risk." That paper, along with decades of follow-up work, contributed to Kahneman's 2002 Nobel Memorial Prize in Economic Sciences, awarded jointly with experimental economist Vernon Smith. Tversky had died in 1996 and was ineligible, since the prize is not awarded posthumously, but the Nobel committee explicitly credited the Kahneman-Tversky collaboration as the intellectual foundation of the award.

What makes the Asian disease experiment (the informal name researchers gave the 1981 study, referring to the hypothetical disease outbreak in the vignette) so durable as a teaching example is its stripped-down clarity. Nothing about the actual medical facts changed between the two versions. Only four words shifted: "will be saved" became "will die." That single lexical substitution was enough to flip a majority preference from risk-averse to risk-seeking, which is precisely why the study still anchors nearly every psychology textbook chapter on judgment and decision-making more than four decades later.

Why losses loom larger than equivalent gains

Prospect theory's core insight is that people do not evaluate outcomes on a simple, linear scale of absolute wealth, health, or lives saved. Instead, they evaluate outcomes relative to a reference point, typically the status quo or an implied baseline, and losses relative to that reference point feel psychologically more significant than equivalent gains. Kahneman and Tversky termed this property loss aversion — the tendency for the pain of losing something to be felt more intensely than the pleasure of gaining the equivalent amount. Because losses are weighted more heavily than gains in this internal calculation, people become risk-averse when a choice is framed around securing a gain, but risk-seeking when the identical choice is framed around avoiding a loss, since gambling becomes psychologically attractive as a way to escape a certain loss.

Picture two people evaluating the exact same $50 outcome: one is told they are "keeping $50 out of $100," and another is told they are "losing $50 out of $100." Mathematically the two descriptions are identical; both people end up with $50. But the second framing anchors attention on the loss itself, which prospect theory predicts will feel more psychologically painful than the first framing's equivalent gain feels pleasurable. Kahneman and Tversky's original experiments estimated that losses are weighted roughly twice as heavily as equivalent gains in this kind of subjective evaluation, a ratio subsequent researchers have refined but consistently found to be greater than one (Tversky & Kahneman, 1992, Journal of Risk and Uncertainty).

A second analogy helps make the mechanism concrete. Think of a thermostat that reports temperature not in absolute degrees, but only as "5 degrees warmer than yesterday" or "5 degrees colder than yesterday." Even if yesterday's baseline temperature is arbitrary, the direction of change, not the absolute reading, is what a person reacts to emotionally. Prospect theory says human risk evaluation works the same way: the reference point does the psychological work, and the same objective destination can feel like a win or a loss purely depending on where the mental thermostat was set beforehand.

Kahneman and Tversky's 1992 refinement of prospect theory estimated a loss aversion coefficient of approximately 2.25, meaning losses are felt, on average, more than twice as intensely as equivalent-sized gains (Tversky & Kahneman, 1992, *Journal of Risk and Uncertainty*).

Why the effect isn't just about words, it's about reference points

The framing effect works specifically because human risk evaluation is anchored to a reference point, typically the status quo or an implied baseline, rather than an evaluation of final outcomes in absolute terms. Changing the reference point implied by wording — "saved" versus "died," "90% survival rate" versus "10% mortality rate" — changes which direction outcomes are perceived to move relative to that baseline, even when the actual, objective outcome described is identical.

This is a distinct claim from simply saying "wording matters." The framing effect is not a vague claim that positive language sounds nicer than negative language. It is a specific, falsifiable prediction: when the underlying math is held perfectly constant and only the reference-point-implying language changes, risk preference should flip in a predictable direction, and it does, reliably, across dozens of replications spanning medicine, finance, environmental policy, and consumer choice.

Framing	Reference point implied	Typical risk preference
"200 of 600 will be saved"	Gain relative to zero saved	Risk-averse (prefer certain option)
"400 of 600 will die"	Loss relative to zero deaths	Risk-seeking (prefer the gamble)
"90% survival rate"	Positive framing, implied gain	Perceived as safer
"10% mortality rate"	Negative framing, implied loss	Perceived as riskier, even though identical
"You keep $50 of $100"	Gain relative to $0	Risk-averse, feels like a win
"You lose $50 of $100"	Loss relative to $100	Risk-seeking, feels like a setback

Six paired examples showing how identical underlying facts produce opposite risk judgments depending on which reference point the wording implies.

The table above also illustrates something researchers call framing symmetry: for every gain-framed description, there exists a mathematically equivalent loss-framed description, and the pair will reliably produce different average choices in a large sample, even when every participant is shown only one version and has no way to compare it directly to the other.

The physicians who changed their treatment recommendations based on wording alone

Perhaps the most consequential replication of the framing effect came from medicine rather than economics. In 1982, researchers Barbara McNeil, Stephen Pauker, Harold Sox, and Amos Tversky published a study in the New England Journal of Medicine that asked 424 physicians, along with samples of patients and graduate students, to choose between surgery and radiation therapy for a hypothetical case of operable lung cancer (McNeil et al., 1982, NEJM). Half of the physicians received outcome statistics framed in terms of survival, for example being told that 90 out of 100 patients survive the immediate post-operative period. The other half received the mathematically identical statistics framed in terms of mortality, being told instead that 10 out of 100 patients die during or immediately after surgery.

When the data were presented in the survival frame, 84% of physicians recommended surgery over radiation therapy. When the same physicians' colleagues were shown the mortality-framed version of the identical statistics, a substantially smaller share recommended surgery, a gap the authors attributed directly to the framing manipulation rather than to any difference in the clinical facts, since the two versions described the same outcome distribution (McNeil et al., 1982, NEJM). The study's significance lies in the population being studied: these were not undergraduates in a psychology lab, but trained physicians making judgments in their area of professional expertise, and the framing effect swayed their treatment recommendations regardless.

This finding reshaped how medical communication researchers think about presenting risk to patients. Follow-up work has repeatedly confirmed that patients, like physicians, rate identical treatment outcomes as more favorable when described using survival statistics rather than mortality statistics (Armstrong et al., 2002, Medical Decision Making). Because informed consent depends on patients understanding their actual odds, not just the odds as flattered or worsened by whichever framing a clinician happens to choose, some medical communication guidelines now recommend presenting both survival and mortality framings side by side specifically to counteract the distortion either framing alone can introduce.

How does the framing effect show up in real medical settings, beyond the 1982 study?

Beyond McNeil's original 1982 study, researchers have documented the same reversal in contexts including cancer screening uptake, vaccination messaging, and surgical consent forms. A 2020 linguistic analysis of physician-patient conversations found that clinicians spontaneously shift between gain and loss framing depending on the treatment they favor, often without conscious awareness of doing so (Journal of Behavioral Medicine, 2020).

Why framing effects persist even when people know about them

A natural question is whether simply teaching people about the framing effect makes them immune to it. Multiple studies conducted after Tversky and Kahneman's original work have found that informing participants about the manipulation in advance, warning them explicitly that a framing trick is coming, only partially reduces its influence on subsequent choices. Debiasing attempts that ask participants to "think more carefully" or consider the decision from multiple angles tend to produce modest improvements at best, not elimination of the effect.

This persistence is one of the more unsettling findings in the framing literature, because it suggests the effect operates at a fairly automatic level of cognitive processing rather than purely through a simple lack of awareness. Kahneman later described this using his dual-process framework from his 2011 book Thinking, Fast and Slow, distinguishing between System 1 (fast, intuitive, automatic judgment) and System 2 (slow, deliberate, effortful reasoning). Framing effects appear to exploit System 1's rapid, reference-point-dependent evaluation before System 2 has a chance to intervene, which is consistent with why simply telling someone "watch out for framing" does not reliably engage enough deliberate override to cancel the bias.

There is a useful analogy here to an optical illusion like the Müller-Lyer illusion, where two line segments of identical length appear different because of arrowhead shapes at their ends. Knowing intellectually that the lines are the same length, and even having measured them yourself with a ruler, does not stop your visual system from perceiving one as longer the next time you look. Framing effects behave similarly: knowing the trick exists does not fully override the automatic perceptual and evaluative machinery that generates the biased judgment in the first place.

That said, "only partially reduces" is not the same as "does nothing." Structured decision aids, decision trees, and formats that force side-by-side comparison of both framings have shown more success than simple warnings, because they change the information environment itself rather than relying on the decision-maker's willpower to resist a bias in the moment.

Common misconceptions about framing effects

A common misunderstanding is that the framing effect only applies to dramatic, high-stakes scenarios like medical treatments involving hundreds of lives. Tversky and Kahneman, along with later researchers, demonstrated the same reversal pattern across far more mundane decisions, including consumer product descriptions ("90% fat-free" versus "10% fat"), discount versus surcharge framing for the identical price difference, and financial investment choices framed as potential gains versus potential losses. This shows the effect is a general property of human judgment rather than a quirk specific to life-and-death scenarios.

Another misconception is that framing effects reflect careless or unintelligent thinking that disappears with more expertise or education. Studies have found the effect persists even among physicians, statisticians, and other domain experts making judgments squarely within their own field, as the McNeil study on lung cancer treatment recommendations demonstrates. This indicates framing sensitivity is a structural feature of how human cognition processes gains and losses, not simply a product of unfamiliarity, inattention, or low numeracy.

A third misconception worth flagging: framing is sometimes conflated with outright lying or fabricating statistics. It is not. In every classic framing study, both versions presented to participants are numerically true. The manipulation lies entirely in which true fact is emphasized and which implicit reference point that emphasis constructs. This is what makes framing effects ethically tricky in fields like marketing and public health messaging: a communicator can technically tell the truth in two different ways and predictably steer an audience toward opposite conclusions, without ever stating a false number.

Where framing effects matter beyond the lab

Financial disclosures are one domain where framing has direct regulatory consequences. Presenting an investment's historical performance as "70% of years showed positive returns" versus "30% of years showed losses" describes the identical track record, but research on investor decision-making has found the two framings produce different risk appetites among retail investors evaluating otherwise identical funds. This is one reason financial regulators in multiple jurisdictions require standardized risk disclosure formats rather than leaving framing choices entirely to the discretion of whoever is marketing the product, since standardization removes one degree of freedom that could otherwise be used to nudge investor behavior.

Public health messaging is a second major application area. Vaccination campaigns, smoking cessation programs, and screening-uptake initiatives have all been studied through a framing lens, generally finding that loss-framed messages (emphasizing what you risk by not acting) tend to be more persuasive for detection behaviors like cancer screening, while gain-framed messages (emphasizing what you achieve by acting) tend to work better for prevention behaviors like sunscreen use, a distinction sometimes called the "prospect theory of health messaging" in public health research. This is not a universal rule and effect sizes vary considerably across studies and populations, but it illustrates that framing is not simply "more positive is always more persuasive."

Retail and e-commerce pricing offers a lower-stakes but pervasive example. A surcharge for paying by credit card and an equivalent discount for paying by cash describe the same price difference, but behavioral economics research going back to the 1970s found that consumers react more negatively to surcharges than they respond positively to equivalent discounts, a finding significant enough that it shaped early credit card industry lobbying over how card-processing fees could legally be disclosed to consumers.

The takeaway

Framing changes risk perception because human judgment evaluates outcomes relative to a reference point rather than in absolute terms, and losses relative to that reference point are weighted more heavily than equivalent gains, a pattern Tversky and Kahneman's prospect theory formalized in 1979 and demonstrated dramatically in their 1981 Asian disease experiment. The same underlying facts, described as lives saved or lives lost, survival rates or mortality rates, discounts or surcharges, can reliably flip which option feels safer, and this holds true even for trained experts like physicians and even when people are warned the manipulation is coming. Recognizing framing effects will not make you immune to them, but it does mean that whenever a risk is presented to you in only one direction, gain or loss, it is worth deliberately asking what the same numbers would look like framed the opposite way.

---

Read next:

• How priming changes what you notice
• Why confirmation bias is hard to notice
• How correlation differs from causation

FAQ

Q: What was the original framing effect experiment?

In a 1981 Science paper, Amos Tversky and Daniel Kahneman asked participants to choose between medical treatment programs for a disease expected to kill 600 people. Framing outcomes as "lives saved" led most people to choose the safe option, while framing the identical outcomes as "lives lost" led most people to choose the risky option instead.

Q: What is prospect theory?

Prospect theory, introduced by Kahneman and Tversky in a 1979 Econometrica paper, describes how people evaluate outcomes relative to a reference point rather than in absolute terms, with losses relative to that reference point weighted more heavily than equivalent gains, a property called loss aversion.

Q: Does telling people about the framing effect prevent it from influencing their choices?

Only partially. Studies have found that warning participants in advance about the framing manipulation reduces but does not eliminate its effect on subsequent choices, suggesting the effect operates at a fairly automatic level of cognitive processing that deliberate reasoning does not fully override.

Q: Does the framing effect only affect high-stakes decisions like medical treatments?

No. The same reversal pattern has been demonstrated across mundane decisions, including consumer product descriptions, discount versus surcharge framing, and financial investment choices, indicating it is a general property of human judgment rather than something specific to life-and-death scenarios.

Q: Why do survival rates and mortality rates change how people perceive the same treatment?

Describing a treatment's outcome using a survival rate frames the information as a gain, while describing the identical outcome using a mortality rate frames it as a loss. Because losses are weighted more heavily in human judgment, the mortality framing tends to make the same treatment feel riskier, a pattern documented among physicians in the landmark 1982 McNeil et al. study published in the New England Journal of Medicine.

Q: Does the framing effect affect trained experts, or just laypeople?

It affects experts too. The 1982 McNeil et al. study found that 424 physicians shifted their treatment recommendations for lung cancer depending on whether outcome statistics were framed as survival rates or mortality rates, even though the physicians were evaluating a scenario squarely within their own clinical expertise.