HPC › Decisions › Glossary
Framing Effect is a cognitive bias in which the description of identical information, not its content, determines the decision made. When outcomes are framed as gains, people tend towards risk aversion; when framed as losses, risk seeking emerges. The effect is grounded in prospect theory's core insight that losses are weighted more heavily than equivalent gains.
The framing effect comprises three distinct subtypes: risky-choice, attribute, and goal framing, each with a different psychological mechanism.
How it works
The framing effect was established empirically in Kahneman and Tversky's Asian Disease Problem 1. When participants were told that a medical programme would save 200 of 600 lives (a gain frame), 72% chose it over a risky alternative. When the identical programme was described as leaving 400 people dead (a loss frame), preferences reversed sharply towards the gamble. The description changed; the outcomes were mathematically identical. This preference reversal cannot be reconciled with rational-choice models that treat logically equivalent options as equivalent inputs.
Three functionally distinct types of framing have been identified 2: risky-choice framing, where gain-framed and loss-framed outcome descriptions produce preference reversals; attribute framing, where labelling the same property positively or negatively alters evaluation; and goal framing, where emphasising the benefits versus costs of an action shifts motivation. Each type operates primarily through System 1 intuitive processing, whereby surface-level descriptions activate emotional valence before deliberate probability comparison occurs. This allows presentation to override content. Susceptibility is not uniform: older adults and those with lower numeracy show larger framing effects, particularly on medical and financial decisions 4.
Example
A portfolio manager is reviewing two risk-management strategies following a significant drawdown. Strategy A is presented as retaining 40% of the portfolio's value with certainty. Strategy B is framed as a 60% chance of recovering the entire amount. When those same strategies are re-expressed as 'a certain loss of 60%' and 'a 40% chance of total loss,' the manager's preference reverses, despite both formulations being mathematically identical.
The label 'loss' activates risk-seeking behaviour that the mathematically equivalent description of 'retaining value' suppresses.
Why it matters
The framing effect carries a confirmed effect size of d = 0.52, established by a p-curve meta-analysis correcting for publication bias 3, making it one of the most robustly replicated biases in behavioural decision research. Its consequences are not confined to laboratory problems. Patients and clinicians shift treatment preferences when survival-rate data is re-expressed as mortality-rate data for the same procedure, with the effect documented even among experienced medical professionals 2. The bias extends across finance, law, and public health: the description changes the decision without altering the underlying reality.
For anyone making consequential decisions, the practical implication is structural: the frame you receive rarely reflects neutral information design. Standardising to natural frequencies ('5 out of every 100 patients') rather than percentages reduces the effect by making gain-frame and loss-frame equivalence more salient 4. Presenting both gain-framed and loss-framed versions of the same data side by side offers stronger protection and should be standard practice in high-stakes advisory settings.
What is the difference between the framing effect and anchoring bias?+
The framing effect occurs when logically identical options produce different choices depending on how they are described as gains or losses. Anchoring bias, by contrast, occurs when an initial numerical reference distorts subsequent estimates or judgements. Both are cognitive biases, but anchoring concerns reference-point adjustment while framing concerns description-to-description equivalence.
How does the framing effect influence medical decisions?+
Both patients and clinicians change treatment preferences when the same outcome data is presented differently {{cite:10.1006/obhd.1998.2804}}. A procedure described by its survival rate receives different acceptance than the identical procedure described by its mortality rate, even among medically trained professionals. This applies to medication consent, screening uptake, and surgical choice.
Is the framing effect equally strong in everyone?+
No. Individual susceptibility varies considerably. Older adults and those with lower numeracy show larger framing effects, particularly on medical and financial decisions {{cite:10.17179/excli2023-6169}}. Conversely, individuals with higher cognitive ability and statistical training show reduced but not eliminated susceptibility. Meta-analytic evidence confirms a medium effect size (d = 0.52) across populations {{cite:10.1027/2151-2604/a000321}}.
How can the framing effect be reduced in practice?+
Translating percentage descriptions into natural frequencies ('5 out of 100 patients') makes the mathematical equivalence of gain and loss frames more visible and reduces preference reversals {{cite:10.17179/excli2023-6169}}. Presenting both the gain-framed and loss-framed versions of the same choice side by side provides additional protection, particularly in clinical and financial advisory contexts.
