The confidence-frequency effect: A heuristic process explanation

Abstract

Abstract People’s feelings of confidence in the correctness of their knowledge while answering a knowledge test can be inferred in two ways: either by averaging the values of specific confidence values assigned to each item in a test (local confidence) or by asking after the termination of the test for an evaluation of the number of correct answers regarding the entire test (global confidence). Surprisingly, when local and global confidence values of the same test are compared, global confidence tends to be significantly lower than local confidence (the confidence frequency effect). In the present study a heuristic process explanation for the effect is presented and its validity is empirically tested. The global confidence heuristic (GCH) process is based on the ability of participants to recall, after a test was completed, the frequencies of specific confidence values which were assigned to the test’s items. Participants build their global confidence by adding about half the number of their guessed answers, to the number of questions with sure answers. The proposed GCH process was supported quantitatively. A content analysis on retrospective explanations provided by participants indicated that this process was feasible. Further research is needed in order to fully explore the power of the explanation suggested here for the confidence-frequency effect