If you tell a woman that girls are bad at math and then give her a math test, she’s not going to do as well as a woman who wasn’t exposed to that threat. This effect is astonishingly robust; it works at every age, in nearly every possible experimental scenario, and even when you’re not testing women.
