Your critical thinking gym
Home
The Commandments
Join the Club
Your critical thinking gym
● Cognitive bias
← Named and demonstrated by psychologists Amos Tversky and Daniel Kahneman, 1981, building directly on their earlier Prospect Theory (1979).
The same facts, described in two logically identical ways, can flip which option feels smarter, safer, or more right, even though nothing about the facts changed.
Also called: message framing · valence framing. Not the same as loss aversion, the deeper value-function property that drives one type of it (Section 09).
A systematic shift in judgment or choice caused by how logically equivalent information is worded or presented, not by any actual change in the underlying facts, probabilities, or stakes.
Plain version
Say a treatment "saves 90 out of 100 people" and people like it. Say the exact same treatment "fails to save 10 out of 100 people" and people like it less, even though those are the same 90 and the same 10. Nothing about the treatment changed. Only the wording did, and the wording did real work on the decision.
Three variants, same underlying pattern
Risky choice framing: a choice between a sure thing and a gamble is described in terms of gains ("saved," "survive") or losses ("die," "lost"), and people get more cautious under the gain wording and more risk-tolerant under the loss wording, Tversky and Kahneman's original 1981 demonstration.
Attribute framing: a single quality is described in a positive or negative light with no risk or gamble involved at all, like ground beef labeled "75% lean" versus "25% fat." Same product, better ratings for the positive frame (Levin & Gaeth, 1988).
Goal framing: a persuasive message emphasizes what you gain by acting ("catch it early and protect your health") or what you lose by not acting ("skip this and put your health at risk"), most studied in health-behavior messaging like breast self-examination promotion (Meyerowitz & Chaiken, 1987).
All three were organized into one typology by Levin, Schneider & Gaeth (1998), whose review is the standard reference for treating "framing effect" as a family of related effects rather than one single trick.
Aliases
Message framing and valence framing are used interchangeably with framing effect in the research literature.
Not the same as loss aversion
These two get folded together constantly, and they're related but not identical. Loss aversion is the deeper claim that losses hurt psychologically more than equivalent gains feel good, full stop, no wording trick required. The framing effect is the broader, surface-level finding that wording alone shifts judgment, and it shows up even in cases like the beef label where nothing is actually being risked or lost at all. Loss aversion helps explain why risky-choice framing works; it doesn't explain attribute or goal framing on its own. Full distinction in Section 09.
Scope
Gets confused with
Loss aversion: precise distinction above and in Section 09.
Quiet, like most of the bottom half of this scale. A frame doesn't feel like a nudge, it feels like the obvious, natural way to describe the thing. There's always an easy cover story available ("I just liked that option better"), and going with the frame you were handed is mildly comfortable, since re-deriving the numbers yourself takes real effort. Nothing here trips a wire on its own.
Why your brain does this
Judging things in isolation is expensive. It's much faster to judge them relative to a reference point, is this better or worse than normal, than expected, than what I already have. Kahneman and Tversky's Prospect Theory (1979) showed people evaluate outcomes as gains or losses relative to a reference point, and weight losses from that point more heavily than equivalent gains. That's a genuinely useful shortcut: in everyday life, whoever is describing something to you is usually also signaling, on purpose, which reference point actually matters right now. Trusting the implied frame most of the time is a reasonable bet, not a glitch.
Where it misfires
The shortcut breaks down when the reference point being handed to you is arbitrary, cherry-picked, or deliberately chosen to move you, rather than genuinely informative. "90% survive" and "10% die" describe the identical outcome, but your brain still processes them as though they were different pieces of news, because the mental machinery for reacting to a stated frame doesn't pause to check whether the frame itself was neutral.
The live academic debate, briefly
The traditional Tversky and Kahneman framing treats this as a violation of "invariance," a basic rationality principle stating that logically identical options should be judged the same regardless of wording. A more ecological reading, in the spirit of Gerd Gigerenzer's work, notes that in ordinary conversation a speaker's chosen framing usually does carry real information about what they think matters (a doctor who says "90% survive" instead of "10% die" is often also signaling something about tone and intent), so leaning on the frame is a reasonable social inference more often than a lab experiment can capture. This page doesn't take a side; both readings agree the effect is real and worth watching for.
How it activates, step by step
1. Information that could be described several logically equivalent ways gets described one specific way.
2. That wording sets an implicit reference point, a gain frame or a loss frame, whether or not the speaker meant it to.
3. Judgment shifts to match the frame: more caution and preference for the sure thing under a gain frame, more risk tolerance under a loss frame, more favorable ratings under a positive attribute frame.
4. The underlying facts, numbers, and stakes never actually moved.
The framing effect isn't just something that happens to people passively, entire fields of professional persuasion exist specifically to choose the frame for you before you get a chance to choose it yourself.
Pairing a policy or fact with emotionally charged wording sets the frame before a reader has any chance to weigh the underlying substance neutrally. Calling an estate tax a "death tax" reframes an inheritance-based policy question as a question about taxing death itself, which tests dramatically better even though the underlying policy is unchanged.
The mirror-image move: softening language reframes something threatening or harmful as neutral or routine, "enhanced interrogation" instead of torture, "collateral damage" instead of civilian deaths. Same underlying reality, a gentler reference point installed on purpose.
Presenting only the statistic that fits the desired frame, the survival rate but never the mortality rate, the approval number but never the disapproval number, forces the reader into a one-sided frame without technically stating anything false.
Outside politics
Food labeling runs on attribute framing constantly and legally: "75% lean," "90% fat-free," "low sugar" are all frame choices, not new information, and they measurably move purchase intent (Levin & Gaeth, 1988). Medicine has the same problem in the opposite direction: physicians describing a treatment's survival rate versus its mortality rate shift both patient and physician preference for that treatment, even though it's the identical treatment (McNeil et al., 1982; Section 05).
What increases it
Strong: low numeracy. Less numerate people rely much more heavily on the stated frame itself, while highly numerate people's choices track their own calculated evaluation of the actual numbers far more than the wording used to present them (Peters and colleagues' numeracy research, published in Judgment and Decision Making).
Strong: low elaboration or quick, single exposure. Framing effects shrink substantially when people are prompted to deliberate, list the pros and cons of each option before choosing, or otherwise slow down and process the numbers directly rather than reacting to the wording.
Moderate: verbal rather than numerical quantifiers. Wording like "high % lean" produces a bigger shift than the equivalent numerical version like "75% lean," because verbal framing leans more on intuitive, less effortful processing.
What does NOT reliably predict it
Professional training doesn't make anyone immune. McNeil, Pauker, Sox and Tversky's 1982 study found the same survival-versus-mortality framing effect in practicing physicians and graduate students, not just ordinary patients, at a similar magnitude. Age also does not show a clean, consistent pattern in the research: some studies find older adults more susceptible under negative framing, others find no reliable age effect once general cognitive ability is accounted for, the literature genuinely remains mixed here.
State-dependent factors
Time pressure and cognitive load push people toward relying on the stated frame rather than working out the underlying numbers themselves, since deliberate recalculation takes spare mental capacity that a rushed or distracted moment doesn't have.
A transparency note: the political example above is a genuinely contested area of the research, not a clean, settled finding, and it's presented that way on purpose rather than smoothed over. The pattern across all three examples: identical underlying facts, a wording choice that costs nothing to make, and a measurable shift in judgment, whether the stakes are a life-or-death medical decision, a grocery purchase, or a policy opinion.
First named
Amos Tversky and Daniel Kahneman formally introduced the framing effect in their 1981 Science paper "The framing of decisions and the psychology of choice", building directly on the reference-point and loss-aversion machinery of their own Prospect Theory, published two years earlier (Kahneman & Tversky, 1979).
The landmark study
The "Asian disease problem": participants were told a disease outbreak would kill 600 people and asked to choose between two programs, described either as lives saved (a gain frame) or lives lost (a loss frame), where the actual numerical outcomes were identical either way. About 72% chose the certain option under the gain frame, while about 78% chose the gamble under the loss frame, a large, reliable reversal driven entirely by wording.
Since then
Levin, Schneider and Gaeth's 1998 review organized decades of follow-up work into the three-variant typology used on this page (risky choice, attribute, goal framing). A 2020 replication effort specifically tested a critique arguing the classic Asian disease wording contained a linguistic artifact that inflated the effect, and found the framing effect held up even once that artifact was removed, so this remains one of the more solidly replicated findings in the field, not a contested one. The genuinely unsettled parts of the literature are about susceptibility (Section 04) and specific applied domains like political labeling (Section 05), not about whether the core effect is real.
Cross-tradition note
See Section 02 for Gerd Gigerenzer's ecological-rationality framing, the relevant non-English research tradition for this page's adaptive-origin section.
Score: 2 out of 5 (see the breakdown in the card after Section 01).
The in-the-moment question
Before reacting to a number or a description, ask: is there an equally accurate, opposite-sounding way to say this exact same thing? "90% survive" and "10% die" pass that test instantly once you ask it. If you can restate the same fact in the opposite valence and it's still true, the wording you were just handed was a choice, not the only honest option.
Why this is hard to catch
The framed version doesn't feel like a nudge, it feels like simply how the fact was reported. There's no competing description sitting next to it for comparison, so nothing about the moment signals that a different, equally true version exists. And going along with the given frame is the path of least resistance: reconstructing the raw numbers and re-evaluating them yourself takes real mental effort that most moments don't invite.
Debiasing research note
This is one of the more responsive biases on the site to structured intervention. Experiments testing "consider the opposite" and deliberate elaboration strategies against the attribute framing effect found both reduced the bias, with "consider the opposite" outperforming elaboration alone. That said, most of this evidence comes from single-session lab and e-commerce settings; whether the effect holds up over months in real, high-stakes decisions (like the medical framing in Section 05) is less well established, so treat these as promising, not proven at scale.
Commonly co-occurs with
Framing effects and anchoring bias often work together: a frame can set which number gets treated as the salient anchor in the first place, then anchoring keeps judgment tethered near it.
Often mistaken for
Related but not the same thing. Loss aversion is the underlying claim that losses are felt roughly twice as intensely as equivalent gains, a property of how people value outcomes, full stop. The framing effect is the broader, surface-level pattern that wording alone shifts judgment, and it shows up even in cases with no real stakes or risk at all, like the "75% lean" beef label. Loss aversion helps explain risky-choice framing specifically; it can't explain attribute or goal framing on its own. See Section 01.
Compounding effect
In persuasive and political messaging specifically, framing effect and negativity bias reinforce each other: a loss-framed message is not just processed differently because it's a "loss," it also benefits from negativity bias's tendency to weight bad-sounding information more heavily than good-sounding information of the same size, making loss frames a especially reliable lever for anyone trying to move an audience.
Primary sources
Accessible reading