TITLE: Understanding overconfidence: Theories of intelligence, preferential attention, and distorted self-assessment AUTHORS: Joyce Ehrlinger, Ainsley L. Mitchum, & Carol S. Dweck ABSTRACT: Knowing what we don’t yet know is critical for learning. Nonetheless, people typically overestimate their prowess—but is this true of everyone? Three studies examined who shows overconfidence and why. Study 1 demonstrated that participants with an entity (fixed) theory of intelligence, those known to avoid negative information, showed significantly more overconfidence than those with more incremental (malleable) theories. In Study 2, participants who were taught an entity theory of intelligence allocated less attention to difficult problems than those taught an incremental theory. Participants in this entity condition also displayed more overconfidence than those in the incremental condition, and this difference in overconfidence was mediated by the observed bias in attention to difficult problems. Finally, in Study 3, directing participants’ attention to difficult aspects of the task reduced the overconfidence of those with more entity views of intelligence. Implications for reducing biased self-assessments that can interfere with learning were discussed. - Cleaned data files with composite variables are shared for open use. - Please contact the first author (Joyce Ehrlinger) at ehrlinger [at] wsu.edu to obtain raw data for replication of analyses. Some rights reserved with respect to raw data for any purpose other than replication of analyses.