This cognitive bias is also called the discriminability effect or the difficulty effect, and was first introduced in 1977 by psychologists Sarah Lichtenstein and Baruch Fischhoff. Multiple theories have been put forward over the years to explain the causes of the bias, including systematic cognitive mechanisms. Studies have shown that when people were asked to answer questions, and then give the likelihood of their answer being correct, they repeatedly overestimated their accuracy for hard questions and underestimated their accuracy for easier ones. Although many psychologists have acknowledged the phenomenon and significant research and experiments have been done on the subject, some have voiced their doubts about the its existence, criticizing the methodology used in these studies and explaining that the bias could be caused by experimenter bias, random error, and statistical artifacts.
In an experiment, respondents had to choose between two answers for a series of questions and, for each one, give a rating of their confidence level on a scale from 50% to 100%. A comparison of the respondents’ confidence ratings and actual answers showed their overconfidence for harder questions, and underestimation for easier questions.