Motivated Mislearning: The Case of Correlation Neglect

Abstract

We design an experiment to study the role of motivated reasoning in correlation neglect. Participants receive potentially redundant signals about either an ego-relevant state—their IQ test performance—or an ego-irrelevant state. A simple hypothesis based on motivated reasoning predicts (i) asymmetric updating about signal redundancy and (ii) asymmetric updating about the focal state only in the treatment with ego-relevance. We find evidence for prediction (i) of our hypothesis: participants generally underappreciate the extent to which identical signals are more likely to come from the same source (and thus contain redundant information), but the bias is significantly stronger for ego-favorable signals than for ego-unfavorable signals. This asymmetric effect disappears in the treatment without ego-relevance. These results suggest that individuals may neglect the correlation between desirable signals to sustain motivated beliefs. However, in contrast to prediction (ii), we do not find significant asymmetric updating about the ego-relevant state (own IQ test performance), which may be explained by the moderate magnitude of the asymmetric updating effect on signal redundancy.

Keywords: Beliefs, cognitive errors, correlation neglect, motivated reasoning, belief-based utility

JEL Classification: C91, D83, D91