“Being Really, Really Certain You Know the Main Idea Doesn’t Mean You Do”, 1990 (; backlinks):
[40 undergraduates were assigned to a high-certainty condition or a 1-reading condition and asked to read 10 200–500 word passages, answer a question about the main idea of each, and rate the certainty of their answer. Subjects in the high-certainty condition were told to read and reread the passages as many times as necessary in order to correctly answer the questions. The other Subjects were told to read each passage just once.
While high-certainty Subjects spent more time reading and were more certain of their answers, they did not show statistically-significantly better performances than Subjects who read passages once.
Thus, asking high-certainty Subjects to be very, very sure of their answers had 2 negative effects: slowed response time and increased confidence in incorrect interpretations.]