Framing effect
Describe the same outcome as a survival rate or a death rate and choices flip.
Logically equivalent descriptions of the same options produce different preferences. The classic case is risky choice framing, where an outcome stated in terms of lives saved draws risk-averse choices and the same outcome stated in terms of lives lost draws risk-seeking ones. Framing works on the reference point, so it is the mechanism behind most of what loss aversion: Losing something feels worse than gaining the same thing feels good. predicts.
How it shows up in software
Framing lives in copy, units, and chart axes. 'Save 20 percent' and 'pay 20 percent less' land differently, and so do '2 seats left' and '98 percent booked'. It also shows up in metrics surfaces: an error rate of 0.4 percent and an uptime of 99.6 percent describe the same system and provoke different follow-up questions.
Using it well
- Pick the frame that matches how the user will actually experience the outcome, then keep it consistent across the product so numbers stay comparable.
- Show annual pricing as a monthly equivalent alongside the total charged, and state both. Hiding the total is a different problem.
- In risk and permission screens, state what happens and what does not, so the user is not left inferring the missing half of the frame.
- When reporting results internally, present the same number both ways once. If the team's conclusion changes, the decision was riding on the frame.
Where it turns manipulative
- Framing that suppresses information rather than reorganising it. 'From $9' where nobody pays $9 is a drip pricing pattern, not a frame.
- Using a percentage base that flatters you and never disclosing it, such as a discount off an inflated list price or a growth rate off a tiny denominator.
- Reframing a downgrade as an upgrade in release notes. Users find out, and the credibility cost outlives the metric win.
Where you have seen it
Spotify
Plan pages state a per-month price for annual and monthly options, and the total billed appears at checkout.
Apple App Store
Privacy nutrition labels describe data practices as categories collected rather than as categories not collected.
Booking.com
Property listings state scarcity as rooms remaining at a price rather than as rooms available overall.
What the research says
- Tversky and Kahneman, 1981Well evidenced
In the Asian disease problem, a majority chose the certain option when outcomes were described as people saved, and a majority chose the gamble when the identical outcomes were described as people dying.
- Klein et al., Many Labs 1, 2014Well evidenced
The Asian disease framing effect replicated across all participating samples with a large effect size, making it one of the better-established results in the set.
- Levin, Schneider and Gaeth, 1998Mixed evidence
A typology separating risky choice framing, attribute framing (75 percent lean versus 25 percent fat) and goal framing, showing the three behave differently and should not be pooled.
Attribute framing is reliable but usually small. Goal framing, where a persuasive message stresses benefits of acting versus costs of not acting, is the shakiest of the three. Being specific about which frame you mean matters more than the headline result.
Grades are a judgement about the evidence, not about how useful the idea is. Plenty of contested effects are still worth knowing, as long as you do not cite them as settled.