On the Low Reliability of Sunk Cost Vignettes

Area of Science:

• Decision-making psychology
• Behavioral economics
• Cognitive biases

Background:

• Sunk cost bias, continuing failing endeavors due to prior investments, is extensively studied.
• Vignette-based research on sunk costs lacks systematic psychometric validation.
• Reliability and validity of commonly used sunk cost scenarios are questionable.

Purpose of the Study:

• To assess the psychometric properties of established sunk cost vignettes.
• To determine if sunk cost scenarios reliably measure the intended psychological construct.
• To examine the internal consistency and external validity of these measures.

Main Methods:

• Two experiments involving 395 participants.
• Utilized classic sunk cost vignettes and measures of cognitive reflection and social desirability.
• Assessed internal consistency, intercorrelations between scenarios, and external validity.

Main Results:

• Consistently poor internal consistency (ω = 0.14-0.57) across vignettes.
• Weak intercorrelations between scenarios, even for similar ones.
• Problematic external validity with inconsistent relationships with cognitive reflection and social desirability.

Conclusions:

• Current sunk cost vignettes demonstrate significant measurement failures.
• Unreliable behavioral measures may be misinterpreted as neural differences in neuroimaging.
• Need for systematic scenario categorization to link vignettes to specific psychological processes and neural circuits.