Beliefs about Gender

We conduct a laboratory experiment on the determinants of beliefs about own and others? ability across different domains. A preliminary look at the data points to two distinct forces: miscalibration in estimating performance depending on the difficulty of tasks and gender stereotypes. We develop a theoretical model that separates these forces and apply it to analyze a large laboratory dataset in which participants estimate their own and a partner?s performance on questions across six subjects: arts and literature, emotion recognition, business, verbal reasoning, mathematics, and sports. We find that participants greatly overestimate not only their own ability but also that of others, suggesting that miscalibration is a substantial, first order factor in stated beliefs. Women are better calibrated than men, providing more accurate estimates of ability both for themselves and for others. Gender stereotypes also have strong predictive power for beliefs, particularly for men?s beliefs about themselves and others? beliefs about the ability of men. Our findings help interpret evidence on gender gaps in self-confidence.