Amos Tversky and Daniel Kahneman showed in their paper The Framing of Decisions and the Psychology of Choice that people judge objectively equivalent problems differently depending on how they are worded. This becomes particularly clear when the same decision is presented once as a gain and once as a loss.
Students at Stanford University and the University of British Columbia were given two linguistically different versions of the same decision problem:
| Problem 1 152 participants | Problem 2 155 participants |
|---|---|
Imagine that the United States is preparing for the outbreak of an unusual Asian disease that is expected to kill 600 people. Two programmes are available: Programme A: 200 people will be saved. Programme B: With a probability of 1/3, 600 people will be saved; with a probability of 2/3, nobody will be saved. |
Imagine that the United States is preparing for the outbreak of an unusual Asian disease that is expected to kill 600 people. Two programmes are available: Programme C: 400 people will die. Programme D: With a probability of 1/3, nobody will die; with a probability of 2/3, 600 people will die. |
Programmes A and C are objectively equivalent: in both cases exactly 200 people survive and 400 die. B and D likewise describe the same probability distribution.
Nevertheless, the decisions differed sharply. In the first version, 72 percent chose the certain Programme A and 28 percent Programme B. In the second version, only 22 percent chose Programme C, while 78 percent preferred the riskier Programme D.
The first formulation emphasises lives saved and is perceived as a gain situation. The second emphasises deaths and is experienced as a loss situation. Tversky and Kahneman thus demonstrated a framing effect: when gains are possible, people often behave more risk-aversely; when losses threaten, they are often more willing to take risks.
The example shows how strongly linguistic framing can influence decisions even though the mathematical structure does not change. Especially with statistics, political messages and risk communication, it is therefore worth asking whether the same information could also be framed differently.