Abstract
In this online study we examined the retention of recent personal events using an Internet-based diary technique. Each participant (N=878) recorded on a website one recent personal event and was contacted after a retention interval that ranged between 2 and 46 days. We investigated how well the participants could recall the content, time, and details of their recorded event. We found a classic retention function. Details of the events were forgotten more rapidly than the content and the time of the events. There were no differences between the forgetting rates of the “who”, “what” and “where” elements of the content component. Reminiscing, social sharing, pleasantness, and frequency of occurrence aided recall, but surprisingly importance and emotionality did not. They were, however, strongly associated with reminiscing and social sharing.
Acknowledgements
This research was supported by a Cognition grant to JM and a Rubicon grant to SJ, both from the Netherlands Organisation for Scientific Research (NWO). The experiment complies with the current laws of the Netherlands in which it was performed. We would like to thank Prof. Berntsen, Dr Meeter, and two anonymous reviewers for helpful suggestions.
Notes
1We compared the gender of the participants who dropped out (N=449, males: 32.7%) and those who completed the study (N=1316, males: 31.3%), but the difference between the two groups was not significant (p =.573). However, there was a significant difference in age, F(1, 1763) = 14.22, MSE=3119.91, p <.001. The mean age of the participants who dropped out of the study (M age = 45.3, SD=16.3) was lower than the mean age of the participants who completed the study (M age = 48.4, SD=14.3). We also found a difference between the levels of education, F(1, 1763) = 17.19, MSE=72.31, p<.001. The group that completed the study had on average a higher level of education than the group that dropped out. The two groups did not differ significantly on cue order presentation (p=.322), but we did find a difference on the retention intervals, F(4, 1760) = 5.10, MSE=0.96, p<.001. There was a larger proportion of participants who dropped out of the experiment in the longer retention intervals. Completion rate after 2 days of retention was 0.82, after 7 days it was 0.80, after 15 days 0.74, after 30 days 0.72, and after 45 days 0.70. The two groups did not differ significantly on emotionality (p=.395) and importance (p=.157), but we did find a difference on frequency of occurrence, F(6, 1758) = 4.07, MSE=0.76, p<.001, and pleasantness, F(6, 1758) = 2.14, MSE=0.41, p=.046. Participants who had completed the study had recorded events that occurred less frequently (M=4.33) and were more pleasant (M=5.35) than the events that participants who had dropped out had recorded (M=3.99, M =5.15). We also looked for interaction effects between the two groups and the retention intervals, but we did not find any interaction effect (emotionality p=.900; pleasantness p=.850; importance p =.809; frequency of occurrence p=.144).
2These exclusions did not lead to significant changes in the completion rates. There was still a significant difference in the completion rates for the five retention intervals, F(4, 1322) = 6.84, MSE=1.51, p <.001. The adjusted completion rate after 2 days of retention was 0.77, after 7 days it was 0.75, after 15 days 0.64, after 30 days 0.63, and after 45 days 0.59. The two groups did not differ significantly on the six presentation orders (p =.973), emotionality (p=.544), pleasantness (p=.300), and importance (p=.122), but we did find a difference on frequency of occurrence. F(6, 1320) = 2.72, MSE=0.22, p=.013. Participants who dropped out recorded events that occurred more frequently (M=3.99) than participants who had recorded a recent event and had responded in time (M=4.14). There were no interaction effects between the two groups and retention interval (emotionality p=.098; pleasantness p=.886; importance p =.303; frequency of occurrence p=.933).
3We assume that answers can be correct or incorrect and guesses or non-guesses. Guesses can be correct or incorrect, but non-guesses are assumed to be correct. Furthermore, we assume that guesses are made randomly. The proportion of guesses that is correct is 1/n, while the proportion of guesses that is incorrect is (n−1)/n. The proportion of answers that are incorrect guesses is known, because there are no incorrect non-guesses: 1 − p(c). The proportion of answers that are guesses is then (1 − p(c))*(n/(n−1)). The proportion of answers that are non-guesses is therefore 1−((1 − p(c))*(n/(n−1))).