A growing body of research suggests that confidence judgements can provide a useful indicator of memory accuracy under some conditions. One factor known to affect eyewitness accuracy, yet rarely examined in the confidence-accuracy literature, is retention interval. Using calibration analyses, we investigated how retention interval affects the confidence-accuracy relationship for eyewitness recall. In total, 611 adults watched a mock crime video and completed a cued-recall test either immediately, after 1 week, or after 1 month. Long (1 month) delays led to lower memory accuracy, lower confidence judgements, and impaired the confidence-accuracy relationship compared to shorter (immediate and 1 week) delays. Long-delay participants who reported very high levels of confidence tended to be over-confident in the accuracy of their memories compared to other participants. Self-rated memory ability, however, did not predict eyewitness confidence or the confidence-accuracy relationship. We discuss the findings in relation to cue-utilization theory and a retrieval-fluency account.
