Disjunction-effect triad test

Description

Pose one decision three ways (outcome known one way, known the other, still unknown) to show whether uncertainty itself, not the outcome, drives deferral or choice.

The sure-thing principle says that if you would take an action whatever an uncertain event's outcome, you should take it while the outcome is unknown. Tversky and Shafir showed people violate it. Students bought a discounted holiday whether they had passed or failed an exam, but when the result was pending most paid a fee to defer. A second gamble was accepted by 69% after a win and 59% after a loss, but only 36% when the first result was unknown. The triad test operationalises this: the same decision is presented under known-A, known-not-A and unknown versions, between subjects or within subjects spaced among fillers. A disjunction pattern (same action under both known outcomes, a different one when unknown) and paying for non-instrumental information show that uncertainty is driving behaviour, which points to design responses that resolve or reframe the pending event.

Attributes

Loss type
Unmigrated
Effort
Days

Steps

  1. Write the decision and the pending uncertain event that precedes it.
  2. Create three versions: outcome known to be A, known to be not-A, and unknown.
  3. Administer them between subjects, or within subjects spaced among filler items with the unknown version first.
  4. Optionally offer a small fee or delay option to wait for or learn the outcome first.
  5. Classify each person's triad and compare action shares across versions; count disjunction patterns and the rate of paying for information that does not change the choice.
  6. Interpret with care: some deferral is rational when the outcome genuinely matters.

Sources

  • Preference, Belief, and Similarity (Tversky, 2004)

Also known as

  • Disjunction-effect triad test (known versus unknown outcome)
  • Sure-thing principle test
  • Known-unknown triad
  • Uncertainty deferral test

Categories

  • Analytics & Behaviour
  • Experimentation & Causal Inference