Abstract
We examine how consumers update their confidences in ordinal (relative) judgments while evaluating sequential product-ranking and source-accuracy data in percentage versus frequency formats. The results show that when sequential data are relatively easier to mathematically combine (e.g., percentage data), consumers revise their judgments in a way that is consistent with an averaging model but inconsistent with the normative Bayesian model. However, when the sequential data are difficult to mathematically combine (e.g., frequency data), consumers update their confidence judgments in a way that is more consistent with the normative Bayesian model than with an averaging model. Interestingly, greater processing motivation for sequential frequency data leads to updated confidence judgments that are lower than normative Bayesian predictions but consistent with the averaging model. Overall, the results of the experiments reveal counterintuitive findings; updated confidence judgments are higher and more accurate when sequential data are more difficult to process and also when consumers have lower processing motivation.