A preference reversal occurs when the same person, evaluating the exact same two options, gives contradictory answers depending purely on the format of the question — directly choosing between them versus separately pricing each one. This is a serious problem for the basic economic assumption that people have stable, well-defined preferences that simply get revealed differently depending on how you ask; a genuinely stable preference shouldn't flip based on question format alone.
The finding suggests something more unsettling than simple inconsistency: that a preference isn't always a fixed thing waiting to be measured, but can be partly constructed in the moment by the specific way a question is framed — which means the method used to elicit a preference (a direct choice versus a price) isn't a neutral measurement tool, it's itself shaping the answer.
Lichtenstein and Slovic presented participants with two gambles: one offering a high probability of winning a modest amount, and another offering a low probability of winning a much larger amount. When asked directly to choose between the two gambles, participants usually preferred the safer, high-probability bet. But when the same participants were separately asked to state the minimum price at which they'd sell each gamble if they owned it, they often priced the riskier, high-payoff gamble higher than the safer one — directly contradicting the preference they'd just expressed when choosing between the same two options.