Document Type
Article
Publication Date
2-20-2024
Abstract
Overconfidence research often involves estimating performance within a fixed range. Recent studies suggest that overprecision is affected by central tendency bias, the tendency to estimate near the center of a perceived distribution. This paper examines whether a similar pattern influences the other two overconfidence types, overestimation and overplacement. In a laboratory experiment (N = 120), participants estimated their own and the average performance over five rounds of ten Rock–Paper–Scissors games. Each round earned for up to 40 points, but average performance varied across three treatments, with means of 16.6, 20, and 23.3 points. The main question was whether this manipulation would result in the sample appearing overconfident, underconfident, or well-calibrated. The results suggest that calibration depends on the relative position of mean performance to the midpoint of the range. When the mean aligned with half of the range maximum, participants appeared well-calibrated. A lower mean resulted in apparent overestimation, while a higher mean led to apparent overplacement. Experience shifted some estimates toward the actual mean and improved calibration, but only when feedback was not overly noisy. Monetary incentives and gender were controlled for but showed no significant effect on estimation accuracy. This study provides evidence that the one-half heuristic—the tendency to estimate at half the maximum of a given range—can mechanically bias confidence judgments. While further research is needed to confirm its effect in more traditional overconfidence tasks, caution is warranted when interpreting studies in which mean performance deviates from the range midpoint.
Recommended Citation
Zika, V. (2024). One-half heuristic in overconfidence research. ESI Working Paper 24-08. https://digitalcommons.chapman.edu/esi_working_papers/404/
Comments
ESI Working Paper 24-08
Previously titled "One-half Heuristic in Overconfidence Research".