Certainty-Effect (Allais Common-Ratio) and Pseudocertainty Test

Description

Present a sure-versus-risky choice, the same choice with all chances scaled down, and a two-stage version equal to the scaled one, to see whether real or apparent certainty is overweighted.

Expected utility says that if people prefer a sure $30 to an 80% chance of $45, they should keep that preference when both chances are divided by four. Most switch to the larger prize once neither option is certain: the certainty effect. In a two-stage version, where a first stage eliminates 75% of players and the choice is made in advance, people return to the 'sure' option even though outcomes equal the scaled problem: pseudocertainty. Tversky and Kahneman found the same pattern in physicians choosing treatments. The test shows whether a framing that makes an outcome look certain changes choices, which matters for guarantees, protection plans and conditional offers. Equating regret opportunities did not remove the effects, pointing to nonlinear decision weights.

Attributes

Loss type
Siloed
Effort
Days

Steps

  1. Problem A: offer a sure smaller gain versus a high chance of a larger gain (for example a sure $30 versus 80% of $45).
  2. Problem B: the same options with both probabilities multiplied by one factor (25% of $30 versus 20% of $45).
  3. Problem C: a two-stage version in which a first stage ends with 75% probability and the Problem A choice must be made before it resolves.
  4. Give the problems within subjects in separated order or between groups; optionally pay a random subset for real.
  5. Compare choice shares: switching from A to B shows the certainty effect; C matching A rather than B shows pseudocertainty.
  6. Optionally add a version equating regret opportunities to rule out regret as the explanation.

Sources

  • Decision Making: Descriptive, Normative, and Prescriptive Interactions (Bell et al., 1988)

Also known as

  • Allais paradox test
  • Common-ratio test
  • Two-stage pseudocertainty test
  • Certainty effect check

Categories

  • Analytics & Behaviour
  • Behavioral Science & Economics