| Abstract: | This study examines critical aspects of both the ecological and the person‐oriented accounts of observed biases in confidence judgements on tests of cognitive abilities. These biases reflect metacognitive processes involved in test‐taking. According to the ecological approach, poor realism of confidence judgements is due to the nature of the items included in general knowledge tests (test‐driven biases). The person‐oriented approach, however, argues that biases in confidence judgements may be due to a general self‐monitoring trait. The present study employed the ‘de‐biasing’ procedure proposed by Juslin ( 1994 ) for the selection of general knowledge test items, and used a newly developed geographical knowledge test suitable for the Australian population. Two other cognitive tests (Raven's Progressive Matrices and Line Length) were administered in order to determine whether there is a consistency in confidence ratings across diverse tasks. Statistical procedures traditional to both approaches‐calibration curves and factor analysis ‐ were employed. The results, with minor qualifications, support both perspectives. The study found a separate confidence factor, indicative of a self‐monitoring trait. Two other potential metacognitive factors (i.e. ‘expectation’ and ‘evaluation’, corresponding to self‐assessment/planning and self‐evaluation) could not be separated from accuracy and speed measures. Copyright © 2001 John Wiley & Sons, Ltd. |
