Overconfidence

Overconfidence is an inflated self-assessment of one's knowledge, answer, or estimate on a task. It is a bias in which a person's subjective confidence of performance is greater than his or her actual performance. This is in contrast to calibration, in which subjective confidence matches accuracy or performance.

Overconfidence has been studied in two distinct ways, either as overestimation of knowledge or performance (overestimation), or as excessive certainty that one's estimate or answer is correct and accurate (overprecision). For example, if a student estimates he or she scored 90 percent on a test but actually scored 75 percent, the student is overconfident through overestimation. With overprecision, a student would create intervals that they are 90 percent confident contain the correct answer, and research with overprecision has found that people create intervals that are too narrow.

Overconfidence is often undesirable because it causes decision makers to underestimate obstacles, to fail to engage in proper preparation and information gathering, or to fail to seek out information contrary to their knowledge. However, overconfidence can be beneficial for increasing persistence in a course of action in the face of failure or in persuading other people about the accuracy of one's ideas. Generally, overconfidence is seen as a form of self-deception or illusion that people sincerely believe, but in certain cases confidence can be strategically and deceptively inflated.

One traditional way of studying overconfidence in the laboratory is by having people answer general knowledge questions (for example, what was Martin Luther King, Jr.'s, age at death?) and providing a range that they are 90 percent confident contains the correct answer (overprecision), or providing a confidence assessment that their answer is correct (overestimation). People's confidence is usually only weakly related to accuracy. Other research has assessed overconfidence in students' rating their academic performance; workers' assessment of workplace performance' managers' view of their managerial skills; or nurses', physicians', and medical residents' assessment of their skills and knowledge. This research has also found that actual skill and knowledge falls significantly below self-assessment of skills and knowledge. Research on deception detection has assessed judges' confidence in their detection skill compared to their actual accuracy in detecting deception and found that people are generally overconfident.

Overconfidence has several hypothesized causes. One cause is that people lack the knowledge necessary for accurate self-assessment of performance. If someone lacks the knowledge to perform well on a test or if someone is using cues that lack predictive utility on a deception detection task, he or she would also lack the [Page 742]knowledge to accurately assess their poor performance, something researchers have termed the “double curse of incompetence.” For example, novice card players are less able to distinguish good from bad moves. Another cause is that people have little insight into errors of omission and cannot adjust their confidence accordingly. People over-rely on available evidence and fail to consider evidence or cues that they have missed in their judgment. For example, people may over-rely on only a few cues to deception and ignore valid cues. Further, for some topics or traits, competence is ill-defined, relative, and ambiguous, so that different people could judge competence and knowledge based on different criteria and choose those criteria by which they are more successful. People are often motivated to maintain flattering assessments of themselves and cherry-pick criteria and feedback that portrays them as the most capable and affable.

...