A core principle in decision science is that people choose according to theirsubjective values. These values are often measured using unincentivized scaleswith arbitrary units (e.g., from 0 to 10) or using incentivizedwillingness-to-pay (WTP) with dollars and cents. What is unclear is whetherusing WTP actually improves choice predictions. In two experiments, we comparethe effects of three different subjective valuation procedures: anunincentivized rating scale, the same scale with incentives, and incentivizedWTP. We use these subjective values to predict behavior in a subsequent binaryfood-choice task. The unincentivized rating task performed better than theincentivized WTP task and no worse than the incentivized rating task. Thesefindings challenge the view that subjective valuation tasks need to beincentivized. At least for low-stakes decisions, commonly used measures such asWTP may reduce predictive power.
