It is widely acknowledged that eyewitness testimony plays a crucial role in the criminal justice system. There has been a significant public concern regarding the number of people being wrongly accused of certain crimes. An essential factor contributing to this issue is the high malleability and fallibility of memory due to various elements. This study focuses closely on the retention interval and its impact on recall confidence and eyewitness accuracy. Twenty participants were invited to take part in an experiment. They were required to answer open-ended questions at different retention intervals (immediate, 5-10 minutes, a day, a week) after watching a video about a crime. Our overall analysis indicates that the length of retention interval significantly affects both overall accuracy and confidence level; i.e., the accuracy and confidence level tend to decrease as the retention interval increases. The overall Confidence-Accuracy (CA) relationship shows a significant and positive correlation. However, when considering the retention interval, a non-significant relationship is observed for all CA relationships. Meanwhile, both eyewitness accuracy and confidence level decrease as item difficulties increase. Among all categories, “scene” items have the highest accuracy, while “action” items have the highest rate of “don’t remember”.
