Choice-Induced Preference Change in the Free-Choice Paradigm: A Critical Methodological Review

TLDR

Choices influence preferences and behavior; the free‑choice paradigm has been used to study choice‑induced preference change, but a methodological flaw identified by Chen and Risen (2010) casts doubt on its validity, yet studies continue to use it. The authors aim to highlight this flaw, illustrate its impact through simulations, review solutions, and assess how it alters interpretation of past findings. They conduct simple simulation studies and a meta‑analysis to demonstrate how noise in each phase of the free‑choice paradigm can generate artificial preference change. Simulations reveal that noise in each phase can produce artificial preference change, the meta‑analysis shows the effect size after addressing the critique, and the authors conclude that the conventional paradigm should be avoided and past findings re‑examined.

Abstract

Choices not only reflect our preference, but they also affect our behavior. The phenomenon of choice-induced preference change has been of interest to cognitive dissonance researchers in social psychology, and more recently, it has attracted the attention of researchers in economics and neuroscience. Preference modulation after the mere act of making a choice has been repeatedly demonstrated over the last 50 years by an experimental paradigm called the "free-choice paradigm." However, Chen and Risen (2010) pointed out a serious methodological flaw in this paradigm, arguing that evidence for choice-induced preference change is still insufficient. Despite the flaw, studies using the traditional free-choice paradigm continue to be published without addressing the criticism. Here, aiming to draw more attention to this issue, we briefly explain the methodological problem, and then describe simple simulation studies that illustrate how the free-choice paradigm produces a systematic pattern of preference change consistent with cognitive dissonance, even without any change in true preference. Our stimulation also shows how a different level of noise in each phase of the free-choice paradigm independently contributes to the magnitude of artificial preference change. Furthermore, we review ways of addressing the critique and provide a meta-analysis to show the effect size of choice-induced preference change after addressing the critique. Finally, we review and discuss, based on the results of the stimulation studies, how the criticism affects our interpretation of past findings generated from the free-choice paradigm. We conclude that the use of the conventional free-choice paradigm should be avoided in future research and the validity of past findings from studies using this paradigm should be empirically re-established.