Do Choices and Ratings Measure the Same Thing?

Two ways to ask a question

Does choosing the darkest square measure the same thing as rating how likely a highlighted square is to be the darkest? In this coauthored study, 80 participants completed both formats using the same visual stimuli.

Comparing representations

We fitted models that either shared a perceptual representation across tasks or allowed separate representations. Bayesian model comparison favoured the shared model for 57 participants. When separate representations were allowed, the choice format could distinguish stimuli more sharply.

What follows for measurement?

The results support shared representations for many participants under this model, while showing why task format deserves attention. They do not establish that the remaining participants used fundamentally different strategies: alternative response rules and modelling assumptions could also affect the comparison.

This is evidence from a single-attribute perceptual task. Whether the same conclusions hold for choices involving several attributes, such as consumer preferences, remains a separate question.