- Journal List
- Psychiatr Psychol Law
- v.29(3); 2022
- PMC9225701
As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsem*nt of, or agreement with, the contents by NLM or the National Institutes of Health.
Learn more: PMC Disclaimer | PMC Copyright Notice
Psychiatr Psychol Law. 2022; 29(3): 471–486.
Published online 2021 Jul 19. doi:10.1080/13218719.2021.1926366
PMCID: PMC9225701
PMID: 35756709
Chantal Chevroulet,a Helen M. Paterson,a Angus Yu,b Emily Chew,b and Richard I. Kempb
Author information Copyright and License information PMC Disclaimer
Abstract
Two studies investigated the impact of recall timing on eyewitness memory. In Study 1, participants viewed a crime video and then completed the Self-Administered Interview (SAI©) either immediately, after a 24-h delay, after a 1-week delay, or not at all. All participants completed a final recall questionnaire 2 weeks after they had viewed the stimulus video. Study 2 aimed to determine how long the beneficial impact of the SAI© on witnesses’ long-term memory lasts. Participants watched a crime video and then either completed the SAI© or did not engage in an immediate recall attempt. Participants then completed a final recall questionnaire after a delay of 24 h, 1 week, 2 weeks, or 1 month. The results indicated that initial recall should be completed within 24 h of an incident and that under these conditions, the beneficial impact of early recall on long-term memory endures for at least 1 month.
Keywords: delay, eyewitness memory, eyewitness testimony, guided recall, self-administered interview
Within the judicial system, eyewitness evidence can be critical to an investigation. However, the quality of such evidence is impacted by the fallibility of memory, which is known to fade rapidly over time (Ebbinghaus, 1913; Murre & Dros, 2015). Furthermore, the longer the delay between the crime and recall, the more likely it is that a witness will encounter misinformation about the event (e.g., through the media) (Paz-Alonso & Goodman, 2008). Several decades of research has demonstrated that when witnesses encounter misinformation after an event, they often incorporate it into their memory, a phenomenon known as the misinformation effect (see Loftus, 2019, for review). As a witness’ memory of the event fades, they become increasingly susceptible to the misinformation effect, as they are less likely to notice the discrepancy between their original memory and any inaccurate post-event information (PEI) they encounter (Paz-Alonso & Goodman, 2008). Therefore, it is important for law enforcement officers to interview witnesses as soon as possible after a critical incident, when their memory is most robust and less likely to have been impacted by exposure to PEI. However, police officers report that this often is not feasible due to limited resources (Wright & Holliday, 2005). When police arrive at a crime scene, they need to secure the area, ensure the safety of everybody present, and perhaps pursue the perpetrator. Therefore, witness interviews are often delayed, sometimes for many weeks (Brown etal., 2008).
Clearly, such delays within the justice system are not ideal and can lead to the degradation of eyewitness evidence. To counteract this issue, researchers have developed guided recall tools that help witnesses to record their own accounts after an incident. For example, Gabbert etal. (2009) developed the Self-Administered Interview (SAI©). The SAI© is a paper-based tool that contains instructions on the use of memory-enhancing techniques, followed by open-ended questions about the event. Witnesses are encouraged to record details about the setting, sequence of events, and people involved. The SAI© is designed to be distributed by police at crime scenes and completed by witnesses independently. This allows eyewitness evidence to be collected soon after the event without placing a strain on police resources. More recently, researchers have developed iWitnessed, a smartphone application that uses guided recall to collect and preserve witness evidence (Paterson etal., 2018a, 2018b).
Research has demonstrated the effectiveness of guided recall tools at capturing detailed incident accounts. Hope etal. (2009) found that mock witnesses who used the SAI© after viewing an event produced more detailed contemporaneous notes than those who engaged in a free recall attempt without any cues or prompts. This finding has been replicated in more recent studies (Hope etal., 2014; Krix etal., 2016; see Horry etal., 2021, for review), showing that the SAI© helps to facilitate eyewitness memory immediately after a critical incident. Furthermore, Matsuo and Miura (2017) found that eyewitness accounts collected immediately using the SAI© are more detailed than accounts collected after a 1-week delay using the cognitive interview procedure.
Researchers have also investigated whether using the SAI© can help to preserve witnesses’ long-term memory of an event. According to Anderson’s (1983) spreading activation theory, memory is stored in the form on interconnected nodes. When one node is activated, this activation spreads through the surrounding network, strengthening the connections between the nodes and facilitating future retrieval of the same information. Furthermore, research on the testing effect has shown how testing as a form of early recall enhances later retention of the tested material (see Roediger & Karpicke, 2006, for a review). Consistent with the testing effect and the spreading activation theory, researchers have found that mock witnesses who engaged in an immediate recall attempt using the SAI© had better long-term memory for the event than either those who did not make an immediate recall attempt (Gittins etal., 2015; Hope etal., 2009; see Horry etal., 2021, for review) or those who engaged in an immediate free recall attempt (Krix etal., 2014). Some researchers have also found that the SAI© reduces witnesses’ susceptibility to the misinformation effect, possibly because they have had the chance to consolidate their memory of the event and are therefore more likely to recognise and reject inaccurate PEI (Gabbert etal., 2012; Gittins etal., 2015). However, other studies have failed to replicate this finding, (e.g., Mackay & Paterson, 2015; Wang etal., 2014).
In addition to memory benefits, research also shows that there are important practical advantages of guided recall tools. For example, police officers in the UK have responded positively to the SAI©, reporting that it helps to capture valuable information at the scene while reducing the burden on police resources (Hope etal., 2011).
As highlighted above, there are clear benefits of guided recall tools such as the SAI©. Nevertheless, further investigation is required to explore the potential limitations of such tools. For instance, as the SAI© is a paper-based tool that must be distributed by law enforcement officers, it is not always possible for witnesses to access it immediately after a critical incident. For example, victims of domestic violence may not come forward to authorities for many months. Additionally, police officers have reported that witnesses are not always able to complete the SAI© at the scene, for various reasons (Hope etal., 2011). In such events the officers asked the witnesses to work on the SAI© at home and return it to police as soon as possible (Hope etal., 2011). This begs the question: how might this delay impact the utility of early recall tools like the SAI©? Because memory fades so rapidly (Murre & Dros, 2015), it is possible that even a short delay could decrease the effectiveness of the SAI©. Additionally, researchers have found that a recall attempt only helps to protect individuals from the misinformation effect if it is completed before they are exposed to misleading PEI (Centofanti & Reece, 2006; Memon etal., 2010). Therefore, delayed completion of the SAI© could also decrease the likelihood that the recall tool would reduce eyewitness suggestibility to PEI.
The existing literature on the impact of timing on the utility of the SAI© is limited. Mackay and Paterson (2015) showed mock witnesses a video of a car accident before randomly allocating them to complete the SAI© immediately, after 24 h, or not at all. One week later, all participants were exposed to misinformation about the event before completing a delayed recall test. Mackay and Paterson (2015) reported that the SAI© enhanced participants’ delayed recall for the event, but only when it was completed immediately. Even a delay of 24-h nullified the benefit of the SAI© as a memory-preserving tool. Additionally, the SAI© was found to have no effect on participants’ susceptibility to the misinformation effect.
Paterson etal. (2015) conducted a further study into the impact of timing on the effectiveness of the SAI©. They showed a sample of university students a stimulus video depicting a crime and then randomly allocated participants to either complete the SAI© immediately, after a 24-h delay, after a 1-week delay, or not at all. Two weeks after viewing the initial event, all participants listened to an audio recording containing misleading PEI, and then completed a free recall account. Unsurprisingly, participants who completed the SAI© immediately reported more accurate information about the crime than those who completed it after a 24 h or 1-week delay. However, when tested 2 weeks later, the participants in the immediate SAI© and 24-h delay groups performed equally well, and better than those in the other conditions. There was no difference between the quality of free recall reports produced by participants in the 1-week delay group and the no SAI control group. Additionally, although the SAI© did not completely protect participants against the misinformation effect, it did reduce the amount of misinformation the participants reported, provided it was undertaken within 24 h of viewing the event.
In sum, while Mackay and Paterson (2015) found that a delay of just 24 h renders the SAI© ineffective at preserving witnesses’ long-term memory for an event, Paterson etal. (2015) found that the SAI© is equally effective at preserving memory if it is administered immediately or after 24 h. Furthermore, these two studies are also inconsistent with regard to the ability of the SAI© to protect witnesses against exposure to misinformation. The aim of the present project was to further examine issues surrounding the timing of guided recall tools such as the SAI©. Study 1 is a replication of the experiment conducted by Paterson etal. (2015) using a slightly more diverse community sample. The use of student participants to evaluate immediate recall tools may be problematic, given that students tend to be young and studying undergraduate psychology. Such students may therefore not be representative of the general population that may use such tools. Consistent with previous studies, it was hypothesised that the positive effect of the SAI© on both immediate and delayed recall would be greater for participants who completed it immediately rather than after a delay. We also sought to determine the level of protection against PEI afforded by the SAI© when administered 24 h after the event.
Study 1
Method
Design
Participants were randomly allocated to one of four SAI© conditions. After watching the stimulus video, they completed the SAI© either (1) immediately, (2) after a 24-h delay, (3) after a 1-week delay, or (4) not at all. Two weeks after viewing the stimulus video, all participants then listened to a taped conversation containing PEI and completed a final recall account. The dependent variables measured were the completeness and accuracy of the participants’ SAI© accounts and final accounts.
Participants
A US-only sample was recruited online using Amazon Mechanical Turk (MTurk; an online workplace). All participants were required to have a working sound system (headphones/speakers) to view the videos. Participants received monetary compensation for their time. They were told that the study would take no more than 2 h in total, across two or three sessions. Because this multi-session experiment was conducted online, there was some participant attrition. Out of 104 participants who accessed and completed the first session, 102 (98.1%) were judged to have watched the video and therefore eligible to progress to the subsequent sessions. Of these, 102 participants, 85 (83.3%) attempted all of the subsequent sessions. Another five participants were excluded from the final analyses, as they made no genuine attempt at recalling what they had seen in the video (e.g., they responded ‘I don’t know’ or ‘No idea’), leaving a final total of 80 participants (50 female).
Participants ranged in age from 18 to 62 years (M = 34.90; SE = 1.28). Participants were predominantly Caucasian (61.3%); however, other ethnicities were also represented (22.5% Asian; 6.3% African; 5.0% Latin American; 5.0% other). Their self-reported nationalities were 86.3% American, 10.0% Indian, and 3.8% other. The participants reported a range of different education levels (8.8% senior high; 17.5% diploma; 3.8% trade, 37.5% bachelor, 23.8% graduate; 8.8% other). Following attrition, there were 23 participants in the immediate SAI© group, 20 in the 24-h delay group, 20 in the 1-week delay group, and 17 in the no SAI© (control) group.
Materials
Stimulus video
Participants viewed a 3-min video depicting a theft within an apartment block (see Paterson etal., 2011, for details).
Filler videos
During the first session of the experiment, an 8-min video on the sport of Canal Jumping was viewed immediately after participants had watched the stimulus video. During the final session, a 5-min sport video on the Highland Games was viewed after participants had listened to the taped conversation. The filler videos were selected for their dissimilarity to the stimulus video, in order to reduce the likelihood of participants confusing the videos.
Self-administered interview (SAI©)
The SAI© is a guided recall tool that elicits details about the event, perpetrator, scene, witnesses, and vehicles that were present (see Gabbert etal., 2009, for details). Within the present study, the participants’ SAI© accounts were collected through an online survey. One section of the SAI© requires witnesses to sketch the scene of the crime. Some adjustments had to be made in order to collect this information remotely. Participants were instructed to complete their drawing on a piece of blank paper. They then entered a typed description of what they had drawn within the digital survey. Only the typed responses were scored for accuracy, as the experimenters did not have access to the original sketches.
Audio recording containing post-event information
Participants listened to a brief audio discussion between two actors playing the role of residents in the apartment block where the theft had occurred. The residents discussed what they heard had happened. The discussion contained eight accurate details and eight items of misinformation (see Table 1 for lists of accurate and inaccurate PEI). This recording was played at the beginning of the final session.
Table 1.
Accurate and inaccurate PEI presented in the audio recording.
| What participants viewed | What audio discussed |
|---|---|
| Accurate PEI | |
| The building main entrance had glass doors | Correct items discussed as consistent with video |
| Thief said he was looking for Paul | |
| Thief entered unit number 2 | |
| Man and woman preparing food in kitchen | |
| Thief was wearing white sneakers | |
| Thief stole wallet | |
| The wallet was in the handbag | |
| Man yelled out ‘Have you seen the salt?’ | |
| Misinformation (inaccurate PEI) | |
| Thief said his name was John | Thief said is name was James |
| Thief said he was from the maintenance company | Thief said he was from the water board |
| Garbage day is Thursday | Garbage day is Tuesday |
| Woman was not wearing a jacket | Woman was wearing a jacket |
| The door was closed but unlocked when woman left | Thief kept door open as woman left |
| Thief was wearing a grey shirt | Thief was wearing a striped shirt |
| Thief did not steal a watch | Thief stole a watch |
| Man yelled ‘Bastard’ at the thief in the end | Man yelled ‘dickhe*d’ at the thief in the end |
Open in a separate window
Free recall questionnaire
Two weeks after viewing the stimulus video, participants completed a free recall questionnaire. They were instructed to record as much as they could remember about the event and were prompted with the following questions: (1) What was the sequence of events from beginning to end? (2) What did the characters look like, and what were they wearing? (3) What dialogue was exchanged between the characters? (4) What did the setting look like? (5) What was stolen from the flat? (6) Write about any other details that you can remember.
Manipulation check
At the end of the experiment, participants were presented with a questionnaire asking them to guess the purpose of the study. No participants reported being aware that they were presented with misinformation.
Procedure
The study largely followed the procedure employed by Paterson and colleagues (2015) except that in the study reported here participants were recruited online using MTurk rather than being tested in person. At the start of the experiment participants consented to taking part in a study on eyewitness memory for a crime. After completing a brief demographic questionnaire, they viewed a stimulus video depicting a theft within an apartment building. Next they watched an 8-min filler video. Participants then completed the SAI© either immediately, after 24 h, after 1 week, or not at all (depending on the SAI© condition to which they were randomly assigned). Participants were allowed to work through the SAI© at their own pace, and no time limit was imposed for the completion of these accounts.
Between sessions, participants were instructed, ‘You will be contacted by email regarding your participation in the subsequent session which you will have 24 h to complete.’ The final session took place 2 weeks after participants had viewed the stimulus video. At this time, participants in all conditions were asked to listen to the audio recording of residents discussing the event. They were told that this recording would refresh their memory of the event. After listening to the audio discussion, the participants watched a 5-min filler video and then completed a free recall questionnaire about the event they had viewed 2 weeks earlier. Finally, all participants completed a manipulation check and were debriefed. Participants were given a 24-h leeway in which to complete the final session.
Data coding
The participants’ SAI© accounts were coded by one research assistant who was blind to the hypotheses and experimental conditions. Each account was coded in terms of the number of accurate and inaccurate details reported. There was a finite list of 411 accurate details that could be reported, and only details that matched that list were coded as accurate. Responses received higher scores, the more details they recalled. For example, if a participant said that the thief was wearing a denim jacket, they would score two points (one for denim and one for jacket). The number of inaccurate responses possible was unlimited, since participants could introduce anything they wanted in their recall. Subjective details (e.g., ‘he looked suspicious’) were coded as ‘subjective’ but were not analysed. The numbers of accurate and inaccurate details reported in each SAI account were used to calculate the percentage accuracy of the account [(accurate details/the sum of accurate and inaccurate details)× 100].
The participants’ final accounts were scored in a similar way. However, a distinction was made between details that were present within the PEI (audio recording) and those that were not. Therefore, when scoring the participants final accounts, the following information was recorded: the number of accurate details reported (excluding PEI), the number of inaccurate details reported (excluding PEI), the amount of accurate PEI reported, and the amount of misinformation (inaccurate PEI) reported. This was done in order to distinguish between information that participants recalled from the stimulus video and information that they may have learned from the audio recording they listened to 2 weeks later. The accuracy of the final accounts was calculated using the same method as was used for the SAI accounts. PEI was included within the accuracy calculations.
In order to establish inter-rater reliability, a second coder independently scored 20 randomly selected transcripts. Pearson correlations between the scores given by each of the coders were .948 for accurate details and .796 for inaccurate details (both correlations significant at p < .001). This indicates a high level of agreement (Cohen, 1988).
Results
The statistical test that was utilised within the present study is the one-way analysis of variance (ANOVA). Omega squared (ω2) was selected as an estimate of effect size because it corrects for the overestimation bias that is seen in eta-squared (η2) estimates. Numerous researchers have found that ω2 is less biased than η2, particularly when sample sizes are small, which is the case within the present study (Keselman, 1975; Okada, 2013). Within the social sciences, ω2 values of 0.01, 0.06 and 0.14 are classified as small, medium, and large effects respectively (Keppel, 1991; Kirk, 1996).
Initial recall using the SAI©
Three ANOVAs were conducted in order to analyse the impact of SAI© condition on the participants’ SAI© accounts with regard to the number of accurate details reported, the number of inaccurate details reported, and the overall percentage accuracy of the accounts. The group means and standard errors for each of these dependent variables are reported in Table 2. Significant ANOVAs were followed up with two planned orthogonal contrasts. The first contrast compared recall outcomes for participants who completed the SAI© within 24 h vs. those who completed it after 1 week. The second contrast compared the immediate and 24-h delay groups.
Table 2.
Group means for each of the recall variables measured from the SAI© accounts and final accounts produced by participants within each SAI© condition in Study 1.
| Self-Administered Interview© condition | ||||||||
|---|---|---|---|---|---|---|---|---|
| Immediate SAI© | 24-h delay | 1-week delay | No SAI© | |||||
| M (SE) | % (SE) | M (SE) | % (SE) | M (SE) | % (SE) | M (SE) | % (SE) | |
| Accurate details reported in SAI© account | 66.30 (4.81) | 67.00 (4.92) | 51.50 (4.15) | — | ||||
| Inaccurate details reported in SAI© account | 8.52 (0.97) | 11.15 (1.16) | 9.90 (1.24) | — | ||||
| Overall accuracy of SAI© account | 88.50 (1.27) | 85.20 (1.53) | 84.09 (1.34) | — | ||||
| Accurate details reported in final account (excluding PEI) | 53.91 (4.09) | 62.00 (4.98) | 48.65 (5.25) | 41.29 (6.00) | ||||
| Inaccurate details reported in final account (excluding PEI) | 7.98 (1.02) | 13.18 (1.57) | 9.78 (1.65) | 12.88 (2.38) | ||||
| Accurate PEI reported in final account | 2.91(0.40) | 2.85 (0.37) | 2.15 (0.33) | 2.29 (0.38) | ||||
| Misinformation |(inaccurate PEI) reported in final account | 0.67 (0.20) | 0.78 (0.21) | 0.63 (0.20) | 0.82 (0.25) | ||||
| Overall accuracy of final account | 86.33 (1.43) | 81.43 (2.13) | 84.69 (1.54) | 73.68 (2.76) | ||||
Open in a separate window
Accurate details
A one-way ANOVA revealed a significant main effect of SAI© condition, F(2, 60) = 3.42, p = .039, ω2 = .071. Subsequent planned orthogonal contrasts revealed that participants who completed the SAI© within 24 h (i.e., both the immediate and 24-h delay conditions) reported a significantly higher number of accurate details than did those who completed it after a 1-week delay, F(1, 60) = 6.83, p = .011, ω2 = .086. The amount of accurate information reported by participants in the immediate SAI© and 24-h delay conditions did not differ significantly, F(1, 60) = .011, p = .916, ω2 = −.015.
Inaccurate details
No differences were found between the number of inaccurate details reported across the SAI© conditions, F(2, 60) = 1.42, p = .250, ω2 = .013.
Percentage accuracy
The effect of SAI© condition on the accuracy of the SAI© accounts approached significance, F(2, 60) = 2.87, p = .065, ω2 = .056.
Final recall
Five separate ANOVAs were conducted in order to investigate the effect of SAI© condition on the quality of the participants’ final accounts. The five dependent variables used within these analyses were: the number of accurate details reported (excluding PEI), the number of inaccurate details reported (excluding PEI), the amount of accurate PEI reported, the amount of inaccurate PEI (misinformation) reported, and the percentage accuracy of the accounts. The group means and standard errors for each of these variables are reported in Table 2. Significant main effects were followed by a set of three planned and mutually orthogonal contrasts. The first contrast compared the recall outcomes of participants who had completed the SAI© with those who had not. The second contrast compared the recall of participants who had completed the SAI© within 24 h with those who had completed it after 1 week. The third contrast compared the immediate and 24-h delay groups.
Accurate details (excluding PEI)
A one-way ANOVA revealed a significant effect of SAI© condition on the number of accurate details reported in the final accounts, F(3, 76) = 2.83, p = .044, ω2 = .064. The first of the planned orthogonal contrasts was significant, showing that participants who completed the SAI© reported more accurate details than those who did not, F(1, 76) = 4.91, p = .030, ω2 = .046. However, participants who completed the SAI© within 24 h did not report a significantly different number of accurate details (M = 21.03, SE = 3.21) than those who completed it after a 1-week delay, F(1, 76) = 2.36, p = .129., ω2 = .016. There was also no significant difference between the amount of accurate information reported by participants in the immediate and the 24-h delay groups F(1, 76) = 1.40, p = .241, ω2 = .004.
Inaccurate details (excluding PEI)
The difference between the number of inaccurate details reported by participants in the different SAI© conditions approached significance, F(3, 76) = 2.45, p = .070, ω2 = .052.
Accurate PEI
The amount of accurate PEI reported by participants did not differ significantly across the SAI conditions, F(3, 76) = 1.08, p = .361, ω2 = .003.
Misinformation (inaccurate PEI)
Overall, approximately half of the participants reported at least one piece of misinformation from the audio discussion within their final account (43.48% in the immediate SAI© group; 50% in the 24-h delay group; 50% in the 1-week delay group; and 58.82% in the no SAI© group), demonstrating that the SAI did not provide complete protection against PEI even when completed immediately.
An ANOVA was conducted to determine whether the SAI© decreased the amount of misinformation reported by participants. There was no main effect of SAI© condition on the amount of misinformation reported F(3, 76) = 0.17, p = .915, ω2 = −.032.
Percentage accuracy
There was a significant main effect of SAI© condition on the overall accuracy of the participants’ final accounts, F(3, 76) = 7.51, p < .001, ω2 = .196. The first planned contrast was significant, meaning that the participants who completed the SAI© had more accurate final accounts (M = 84.25, SE = 1.00) than those who did not complete an SAI©, F(1, 76) = 18.77, p < .001, ω2 = .179. However, there was no significant difference between the accuracy of the accounts produced by participants who completed the SAI© within 24 h (M = 84.05, SE = 1.29) and those who completed it after a week, F(1, 76) = 0.12, p = .735, ω2 = −.009. Finally, thedifference between the accuracy of the accounts produced by participants in the immediate SAI© group and the 24-h delay group approached significance, F(1, 76) = 3.29, p = .074, ω2 = .023.
Discussion
The findings of Study 1 generally indicate that the SAI© is an effective memory-preservation tool when completed within 24 h of the critical incident. With regard to the quality of the participants’ initial accounts, those who completed the SAI© within 24 h produced more complete accounts than did those who completed it after a 1-week delay. This effect was moderate in size, according to the guidelines outlined by Keppel (1991) and Kirk (1996). However, there was no difference between the number of accurate details reported in the initial accounts produced by participants in the immediate SAI© and 24-h delay groups. This contrasts with the finding of Mackay and Paterson (2015), who reported that the SAI© is most effective at eliciting a detailed account when it is completed immediately.
The SAI© had a large effect on the quality of the final accounts produced by participants 2 weeks after viewing the critical incident. Participants who completed the SAI© wrote more complete and accurate final accounts than those who did not. Additionally, the SAI© was equally beneficial at preserving memory, whether it was completed immediately or after a 24-h delay. This is consistent with the findings of the original study by Paterson etal. (2015) and contradicts Mackay and Paterson’s (2015) finding that a 24-h delay renders the SAI© ineffective. It is possible that the discrepancy is due to the nature of the stimuli used within each study. Specifically, participants in Mackay and Paterson’s (2015) study viewed a video of a traumatic car accident, leading to an increase in their self-reported state anxiety. Stress has been found to impair eyewitness memory (see Deffenbacher etal., 2004, for a meta-analysis). Therefore, it is possible that the presence or absence of stress could be a boundary condition which impacts the relationship between timing and the efficacy of the SAI©. However, this is beyond the scope of the present study and further research is required to investigate this.
Although the present study found a positive effect of the SAI© on the participants’ long-term memory for the event, the recall tool did not appear to prevent or reduce the impact of the misinformation effect on the final recall. The small sample size may have limited our ability to obtain a statistically significant result; however, the very small effect size (ω2 = −.032) suggests that this may not be the case. Alternatively, the non-significant finding may be due to a floor effect, as very little misinformation was reported by participants in any of the recall groups.
A community sample was recruited for the present study. However, the participants reported much higher levels of education than anticipated. Therefore, the sample is not representative of the general population with regard to their literacy level. However, the participants were more diverse in age than a typical student sample, which adds to the external validity of the study.
In summary, the present study adds to our understanding of the effects of delay on the impact of the SAI© as a memory-enhancing tool. Although only three delay durations were incorporated in the design (immediate, 24-h delay, and 1-week delay), this has helped to clarify the impact of recall timing on the effectiveness of the SAI. Greater sampling along the timescale in future studies would allow researchers to more closely observe how the effectiveness of the SAI© changes over time.
Based on the findings of Study 1, along with the pre-existing literature (Mackay & Paterson, 2015; Paterson etal., 2015), it seems that the SAI© should be completed as soon as possible after an event, and certainly within 24 h, in order to maximise its beneficial effect. However, even in the optimal case when witnesses are able to complete the SAI© immediately after the incident, it is unclear how long the resultant memory benefits will last. In Study 1 these benefits were demonstrated after a delay of 2 weeks, but in some situations the time before a full police interview is completed may be considerably longer than this. However, most existing studies that have found a beneficial effect of the SAI© on witnesses’ long-term memory have only incorporated a delay of 1 or 2 weeks between the stimulus event and the final recall attempt (e.g., Gittins etal., 2015; Hope etal., 2009; Krix etal., 2014). Although cued recall tools have been seen to strengthen the memory trace, it is doubtful that they protect memory from decay permanently. Therefore, it is possible that over time a floor effect will emerge such that even individuals who have completed the SAI© will not be able to recall much about the critical incident. Study 2 was designed to investigate whether the effectiveness of the SAI© declines with time, and the rate at which this decline occurs. This has important practical implications for the organisation of witness interviews within police investigations and is thus an important practical question.
Study 2
Study 2 participants watched a stimulus video (followed by a filler video) and were then randomly allocated to one of two immediate recall conditions (SAI© or no immediate recall). After this, participants experienced a delay of 24 h, 1 week, 2 weeks, or 1 month, depending on the delay condition to which they had been randomly assigned. Following their allotted delay, each participant completed a final recall questionnaire. It was hypothesised that there would be a main effect of immediate recall condition such that the participants who completed the SAI© would produce more complete and accurate final accounts than those who did not. Additionally, it was hypothesised that there would be a main effect of delay condition such that the participants who completed their final recall attempt earlier would report more accurate information than those who completed it after a longer delay. Of particular interest was whether an interaction effect would arise such that the beneficial impact of the SAI© on the participants’ long-term memory would decline with greater delays between the stimulus event and the final recall questionnaire.
Method
Design
Study 2 used a 2 × 4 between-subjects design. The independent variables were the immediate recall condition (SAI© or no immediate recall) and the time between the stimulus and the final recall attempt (24 h, 1 week, 2 weeks, or 1 month). Dependent variables were the completeness and accuracy of the final accounts produced by the participants.
Participants
A sample of 160 participants (82 female) was recruited using MTurk. Participants received financial reimbursem*nt for their time. No participants reported being aware that they were presented with misinformation, so all participants were included in the analyses. The sample had an age range of 19 to 67 years (M = 33.71, SE = 0.81). Their self-reported ethnicities were 77.5% Caucasian, 10.6% Asian, 5.0% African, 3.8% Latin American, and 5.0% mixed. Most participants were American (95.0%); however, some other nationalities were also represented within the sample (1.9% Chinese; 1.3% Italian; 1.8% other). Participants reported a wide range of educational backgrounds (0.6% junior high; 18.8% senior high; 11.9% diploma; 5.6% trade; 41.3% bachelor; 14.4% graduate; 7.5% other). Each participant was randomly allocated to one of the eight experimental groups, with 20 participants in each group.
Materials
Study 2 employed the materials from Study 1.
Procedure
The study consisted of two sessions. During the first session, all participants watched the stimulus video of a theft followed by an 8-min filler video. According to their randomly allocated condition, participants then either completed the SAI or did no immediate recall. Following this, all participants were randomly allocated to undertake the second session after a delay of 24 h, 1 week, 2 weeks or 1 month.
In the second session, participants listened to the audio discussion about the theft they had viewed during Session 1 and then watched a 5-min filler video. Participants then completed a free recall questionnaire in which they responded to open-ended questions about the theft. Finally, they completed the manipulation check and were debriefed.
Data scoring
As in Study 1, a research assistant reviewed the participants’ final accounts and recorded the number of accurate details reported (excluding PEI), the number of inaccurate details reported (excluding PEI), the amount of accurate PEI reported, and the amount of misinformation (inaccurate PEI) reported. The overall percentage accuracy of each participant’s account was calculated. Details from the PEI were included within the accuracy calculations.
Twenty of the participants’ accounts were also scored by a second coder in order to establish inter-rater reliability. The scores given by the first and second coders were compared, and the results of the correlational analyses are displayed in Table 3. Pearson correlations between the scores given by the two coders ranged from .823 to .958, showing a high level of agreement (Cohen, 1988).
Table 3.
Pearson correlations between scores assigned by Coder 1 and Coder 2.
| Coders 1 and 2 (Pearson correlation) | |
|---|---|
| Number of accurate details reporteda | .921 |
| Number of inaccurate details reporteda | .823 |
| Amount of accurate PEI reported | .958 |
| Amount of misinformationb reported | .944 |
Open in a separate window
All correlations are significant at p < .001.
aExcluding PEI. bInaccurate PEI.
Results
A series of 2 (immediate recall condition: SAI© vs. no immediate recall) × 4 (delay: 24 h, 1 week, 2 weeks, 1 month) ANOVAs were conducted to investigate differences in final recall with regard to each of the following variables: the number of accurate details reported (excluding PEI), the number of inaccurate details reported (excluding PEI), the amount of accurate PEI reported, the amount of misinformation (inaccurate PEI) reported, and the percentage accuracy of the account.
When a significant main effect of the delay condition was found, three planed and mutually orthogonal ‘delay’ contrasts were conducted. The first contrast compared the recall outcomes for participants who completed their final account after 24 h vs. those who experienced a longer delay. The second contrast compared the recall of participants who experienced a delay of ≤1-week vs. those who experienced a longer delay. The third contrast compared the 2-week and 1-month delay groups.
Accurate details (excluding PEI)
The mean number of accurate details reported in the participants’ final accounts within each experimental condition is displayed in Figure1. The two-way ANOVA revealed a significant main effect of the immediate recall condition, such that the participants who completed the SAI© recalled more accurate details (M = 59.30, SE = 2.47) than those who did not (M = 44.83, SE = 2.46), F(1, 152) = 17.79, p < .001, ω2 = .092. Additionally, there was a main effect of delay condition, F(3, 152) = 3.12, p = .028, ω2 = .035. Subsequent contrasts revealed that participants who had completed their final account 24 h after viewing the theft reported more accurate details (M = 61.08, SE = 3.94) than those who had completed it after a longer delay (M = 49.06, SE = 1.99), F(1, 152) = 9.20, p = .033, ω2 = .045. No significant difference was found between the number of accurate details reported by participants in the 1-week delay group (M = 50.23, SE = 4.06) and those who experienced a longer delay (M = 48.48, SE = 2.21), F(1, 152) = 0.17, p = .678, ω2 = −.005. Similarly, the amount of accurate information reported by participants in the 2-week (M = 48.48, SE = 3.02) and 1-month delay conditions (M = 48.48, SE = 3.27) did not differ significantly, F(1, 152) < 0.01, p = 1.00, ω2 = −005. There was no interaction between the immediate recall and delay variables, F(3, 152) = 0.60, p = .617, ω2 = −.007.
Figure 1.
The mean number of accurate details (excluding PEI) reported in the final accounts of participants in each experimental condition in Study 2. Error bars represent standard error.
Inaccurate details (excluding PEI)
A two-way ANOVA analysed the number of inaccurate details reported by participants in the different experimental conditions (see Table 4 for means and standard errors). There were no main effects of the immediate recall condition, F(1, 152) = 0.68, p = .412, ω2 = −.002, or delay condition, F(3, 152) = 0.67, p = .572, ω2 = −.006. The interaction effect was also non-significant, F(3, 152) = 0.11, p = .953, ω2 = −.017.
Table 4.
Mean number of inaccurate details and misinformation reported in the final accounts of participants within each experimental condition in Study 2.
| Immediate recall condition | |||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| SAI© | No immediate recall | Total | |||||||||||||
| Delay condition | |||||||||||||||
| 24 h | 1 week | 2 weeks | 1 month | Total | 24 h | 1 week | 2 weeks | 1 month | Total | 24 h | 1 week | 2 weeks | 1 month | Total | |
| Number of inaccurate details reporteda | 6.55 (0.99) | 6.80 (1.07) | 7.20 (1.06) | 7.75 (0.73) | 7.08 (0.48) | 6.80 (0.96) | 7.15 (1.91) | 8.75 (1.58) | 8.50 (1.26) | 7.80 (0.73) | 6.68 (0.68) | 6.98 (1.08) | 7.98 (0.95) | 8.13 (0.72) | 7.44 (0.43) |
| Amount of accurate PEI reported | 3.85 (0.39) | 2.65 (0.38) | 3.00 (0.32) | 2.75 (0.40) | 3.06 (0.19) | 2.25 (0.33) | 2.15 (0.45) | 1.75 (0.20) | 2.10 (0.37) | 2.06 (0.17) | 3.05 (0.28) | 2.40 (0.29) | 2.38 (0.21) | 2.43 (0.27) | 2.56 (0.13) |
| Amount of misinformationb reported | 0.50 (0.18) | 0.70 (0.26) | 0.75 (0.22) | 1.00 (0.19) | 0.74 (0.11) | 0.30 (0.15) | 0.50 (0.18) | 0.60 (0.22) | 0.95 (0.26) | 0.59 (0.10) | 0.40 (0.12) | 0.60 (0.16) | 0.68 (0.15) | 0.98 (0.16) | 0.66 (0.08) |
Open in a separate window
Note. Means. Standard errors in parentheses.
aExcluding PEI. bInaccurate PEI.
Accurate PEI
Table 4 shows the mean amount of accurate PEI reported by participants in each of the SAI© and delay conditions. The two-way ANOVA revealed a significant main effect of SAI© condition on the amount of accurate PEI reported in the participants’ final accounts, F(1, 152) = 15.33, p < .001, ω2 = .081. Participants who completed the SAI© reported significantly more accurate PEI than those who did not. However, the effect of delay condition did not reach significance, F(3, 152) = 1.63, p = .186, ω2 = .011. The interaction effect was also non-significant, F(3, 152) = 1.02, p = .388, ω2 < .001.
Misinformation (inaccurate PEI)
In accordance with Study 1, Study 2 also demonstrates that the SAI© does not offer total protection against the misinformation effect, with 50% of the participants who completed the SAI© reporting at least one piece of misinformation within their final account.
Table 4 shows the mean amount of misinformation reported by the participants within each experimental condition. The two-way ANOVA found no main effect of immediate recall condition on the amount of misinformation reported, F(1, 152) = 1.01, p = .317, ω2 < .001. The effect of the delay variable approached significance, F(3, 152) = 2.55, p = .058, ω2 = .029. There was no interaction between the immediate recall and delay variables, F(3, 152) = 0.06, p = .983, ω2 = −.018.
Percentage accuracy
Figure 2 shows the mean accuracy of the final accounts produced by participants in each experimental condition. The two-way ANOVA revealed a significant beneficial effect of immediate recall on the overall accuracy of the participants’ final accounts, F(1, 152) = 9.87, p = .002, ω2 = .050. There was also a significant main effect of delay condition, F(3, 152) = 4.12, p = .008, ω2 = .053. Subsequent contrasts revealed that participants in the 24-h delay group produced more accurate accounts (M = 89.83, SE = 0.86) than those who experienced a longer delay (M = 85.46, SE = 0.84), F(1, 152) = 8.46, p = .004, ω2 = .042. The other two contrasts were non-significant. That is, there was no significant difference between the accuracy of the accounts produced by participants who completed the SAI© after a 1-week delay (M = 87.28, SE = 1.59) and those who experienced a longer delay (M = 84.56, SE = 0.98), F(1, 152) = 2.91, p = .090, ω2 = .011. Additionally, accuracy did not differ between the participants in the 2-week (M = 85.47, SE = 1.34) and 1-month delay groups (M = 83.64, SE = 1.41), F(1, 152) = 0.99, p = .321, ω2 < .001. Finally, no interaction was found between the immediate recall and delay variables, F(3, 152) = 0.54, p = .654, ω2 = −.008.
Figure 2.
The mean accuracy (%) of the final accounts produced by participants in each experimental condition in Study 2. Error bars represent standard error.
Discussion
The anticipated effects of completing an immediate recall attempt and of the duration of the delay prior to final recall were observed. In accordance with prior research, participants who completed the SAI© produced more accurate and complete final accounts than those who did not engage in an immediate recall attempt. In addition, participants who waited 24 h between the stimulus and their final recall questionnaire produced more detailed and accurate accounts than those who experienced delays of a week or longer. These effects were generally moderate in size.
Study 2 also investigated the ability of the SAI© to protect eyewitnesses from the misinformation effect. Participants who completed the SAI© did not differ significantly from those who did not complete the SAI on their reporting of misinformation. The small sample size may have limited our ability to obtain a statistically significant result; however, the very small effect size (ω2 < .001) suggests that this may not be the case. Alternatively, .it is likely that the effect of the SAI© on eyewitness suggestibility could have been masked by a floor effect, as the reporting of misinformation was very low across all the recall groups.
Study 2 found no evidence of an interaction between the immediate recall and delay conditions with regard to any of the memory outcomes. That is, the beneficial impact of the SAI© on the completeness and accuracy of the participants’ final accounts did not vary with the length of the delay before the final recall and was still evident after 1 month. Therefore, the advantages of the SAI© appear to be remarkably robust over time.
General discussion
Both studies provide further evidence that immediate recall tools such as the SAI© improve the quality and completeness of both contemporaneous and delayed eyewitness accounts. However, contrary to the previous findings of Gabbert etal. (2012), the SAI© did not have a statistically significant impact on the participants’ susceptibility to the misinformation effect. This could be partially due to the limited power created by the small sample sizes, which was a weakness of both studies. Alternatively, the lack of statistical significance could be due to a floor effect, as little misinformation was reported in either study. This suggests that the participants did not find the misinformation presented within the audio recording very convincing. Different sources of misinformation have been shown to vary in degree of influence (e.g., Gabbert etal., 2004; Paterson & Kemp, 2006), and it may be that the audio recording of a discussion of the event between two non-involved parties had little impact. However, it is important to note that other studies have also failed to find that the SAI© reduces witnesses’ susceptibility to the misinformation effect (e.g., Mackay & Paterson, 2015; Wang etal., 2014). This suggests that the SAI© may not reliably protect witnesses from misinformation.
Based on the findings reported here, the clear recommendation to emerge is that witnesses should be encouraged to complete an immediate recall tool such as the SAI© as soon as possible, and certainly within 24 h after the event. Under these circ*mstances it appears that the beneficial effect of the SAI© on later recall lasts for at least one month – which should provide investigators with sufficient time to arrange a formal interview. This has important practical implications for investigators, allowing them to obtain more complete and accurate witness statements.
Ethical standards
Declaration of conflicts of interest
Chantal Chevroulet has declared no conflicts of interest.
Helen M. Paterson has declared no conflicts of interest.
Angus Yu has declared no conflicts of interest.
Emily Chew has declared no conflicts of interest.
Richard I. Kemp has declared no conflicts of interest.
Ethical approval
All procedures performed in Studies 1 and 2 were in accordance with the ethical standards of the University of New South Wales [Approval no. 1972] and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.
Informed consent
Informed consent was obtained from all individual participants included in the study.
Funding Statement
This research was supported by an Australian Research Council Linkage Grant [grant number LP110100220].
References
- Anderson, J. R. (1983). A spreading activation theory of memory. Journal of Verbal Learning and Verbal Behavior, 22(3), 261–295. 10.1016/S0022-5371(83)90201-3 [CrossRef] [Google Scholar]
- Brown, C., Lloyd-Jones, T. J., & Robinson, M. (2008). Eliciting person descriptions from eyewitnesses: A survey of police perceptions of eyewitness performance and reported use of interview techniques. European Journal of Cognitive Psychology, 20(3), 529–560. 10.1080/09541440701728474 [CrossRef] [Google Scholar]
- Centofanti, A. T., & Reece, J. (2006). The cognitive interview and its effect on misleading postevent information. Psychology, Crime & Law, 12(6), 669–683. 10.1080/10683160600558394 [CrossRef] [Google Scholar]
- Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). John Wiley. [Google Scholar]
- Deffenbacher, K. A., Bornstein, B. H., Penrod, S. D., & McGorty, E. K. (2004). A meta-analytic review of the effects of high stress on eyewitness memory. Law and Human Behavior, 28(6), 687–706. 10.1007/s10979-004-0565-x [PubMed] [CrossRef] [Google Scholar]
- Ebbinghaus, H. (1913). Memory: A contribution to experimental psychology (Ruger H. A. & Bussenius C. E., Trans.). Teachers College, Columbia University. (Original work published 1885). [Google Scholar]
- Gabbert, F., Hope, L., & Fisher, R. P. (2009). Protecting eyewitness evidence: Examining the efficacy of a self-administered interview tool. Law and Human Behavior, 33(4), 298–307. 10.1007/s10979-008-9146-8 [PubMed] [CrossRef] [Google Scholar]
- Gabbert, F., Hope, L., Fisher, R. P., & Jamieson, K. (2012). Protecting against misleading post‐event information with a self‐administered interview. Applied Cognitive Psychology, 26(4), 568–575. 10.1002/acp.2828 [CrossRef] [Google Scholar]
- Gabbert, F., Memon, A., Allan, K., & Wright, D. B. (2004). Say it to my face: Examining the effects of socially encountered misinformation. Legal and Criminological Psychology, 9(2), 215–227. 10.1348/1355325041719428 [CrossRef] [Google Scholar]
- Gittins, C. B., Paterson, H. M., & Sharpe, L. (2015). How does immediate recall of a stressful event affect psychological response to it?Journal of Behavior Therapy and Experimental Psychiatry, 46, 19–26. 10.1016/j.jbtep.2014.07.006 [PubMed] [CrossRef] [Google Scholar]
- Hope, L., Gabbert, F., & Fisher, R. (2009). The Self-Administered Interview (SAI): Capturing eyewitness accounts at the scene of an incident. Society of Applied Research in Memory and Cognition. [Google Scholar]
- Hope, L., Gabbert, F., & Fisher, R. P. (2011). From laboratory to the street: Capturing witness memory using the Self‐Administered Interview. Legal and Criminological Psychology, 16(2), 211–226. 10.1111/j.2044-8333.2011.02015.x [CrossRef] [Google Scholar]
- Hope, L., Gabbert, F., Fisher, R. P., & Jamieson, K. (2014). Protecting and enhancing eyewitness memory: The impact of an initial recall attempt on performance in an investigative interview. Applied Cognitive Psychology, 28(3), 304–313. 10.1002/acp.2984 [CrossRef] [Google Scholar]
- Horry, R., Hughes, C., Sharma, A., Gabbert, F., & Hope, L. (2021). A meta‐analytic review of the Self‐Administered Interview©: Quantity and accuracy of details reported on initial and subsequent retrieval attempts. Applied Cognitive Psychology, 35(2), 428–444. [Google Scholar]
- Keppel, G. (1991). Design and analysis: A researcher’s handbook (3rd ed.). Prentice– Hall. [Google Scholar]
- Keselman, H. (1975). A Monte Carlo investigation of three estimates of treatment magnitude: Epsilon squared, eta squared, and omega squared. Canadian Psychological Review/Psychologie Canadienne, 16(1), 44–48. 10.1037/h0081789 [CrossRef] [Google Scholar]
- Kirk, R. E. (1996). Practical significance: A concept whose time has come. Educational and Psychological Measurement, 56(5), 746–759. 10.1177/0013164496056005002 [CrossRef] [Google Scholar]
- Krix, A. C., Sauerland, M., Gabbert, F., & Hope, L. (2014). Providing eyewitnesses with initial retrieval support: What works at immediate and subsequent recall?Psychology, Crime & Law, 20(10), 1005–1027. 10.1080/1068316X.2014.902456 [CrossRef] [Google Scholar]
- Krix, A. C., Sauerland, M., Raymaekers, L. H., Memon, A., Quaedflieg, C. W., & Smeets, T. (2016). Eyewitness evidence obtained with the Self‐Administered Interview© is unaffected by stress. Applied Cognitive Psychology, 30(1), 103–112. 10.1002/acp.3173 [CrossRef] [Google Scholar]
- Loftus, E. F. (2019). Eyewitness testimony. Applied Cognitive Psychology, 33(4), 498–503. 10.1002/acp.3542 [CrossRef] [Google Scholar]
- Mackay, T. L., & Paterson, H. M. (2015). How does timing of recall affect eyewitness memory and psychological distress?Journal of Police and Criminal Psychology, 30(4), 242–253. 10.1007/s11896-014-9156-z [CrossRef] [Google Scholar]
- Matsuo, K., & Miura, H. (2017). Effectiveness of the self-administered interview and drawing pictures for eliciting eyewitness memories. Psychiatry, Psychology and Law, 24(5), 643–654. [PMC free article] [PubMed] [Google Scholar]
- Memon, A., Zaragoza, M., Clifford, B. R., & Kidd, L. (2010). Inoculation or antidote? The effects of cognitive interview timing on false memory for forcibly fabricated events. Law and Human Behavior, 34(2), 105–117. 10.1007/s10979-008-9172-6 [PubMed] [CrossRef] [Google Scholar]
- Murre, J. M., & Dros, J. (2015). Replication and analysis of Ebbinghaus’ Forgetting Curve. PloS One, 10(7), e0120644. 10.1371/journal.pone.0120644 [PMC free article] [PubMed] [CrossRef] [Google Scholar]
- Okada, K. (2013). Is omega squared less biased? A comparison of three major effect size indices in one-way ANOVA. Behaviormetrika, 40(2), 129–147. 10.2333/bhmk.40.129 [CrossRef] [Google Scholar]
- Paterson, H. M., Eijkemans, H., & Kemp, R. I. (2015). Investigating the impact of delayed administration on the efficacy of the Self-administered interview. Psychiatry, Psychology and Law, 22(2), 307–317. [Google Scholar]
- Paterson, H., & Kemp, R. (2006). Comparing methods of encountering post-event information: The power of co-witness suggestion. Applied Cognitive Psychology, 20(8), 1083–1099. 10.1002/acp.1261 [CrossRef] [Google Scholar]
- Paterson, H., van Golde, C., Devery, C., Cowdrey, N., Kemp, R. (2018b). iWitnessed: A Smart Phone Application to Record Eyewitness Memory. Australian Police Journal, June, 60–64. University of Sydney. (2018). iWitnessed. Retrieved 10 November 2018, from https://sydney.edu.au/science/psychology/iWitnessed/
- Paterson, H. M., Kemp, R. I., & Ng, J. R. (2011). Combating Co‐witness contamination: Attempting to decrease the negative effects of discussion on eyewitness memory. Applied Cognitive Psychology, 25(1), 43–52. 10.1002/acp.1640 [CrossRef] [Google Scholar]
- Paterson, H. M., van Golde, C., Devery, C., Cowdery, N., & Kemp, R. (2018a). iWitnessed: Capturing contemporaneous accounts to enhance witness evidence. Current Issues in Criminal Justice, 29(3), 273–281. 10.1080/10345329.2018.12036102 [CrossRef] [Google Scholar]
- Paz-Alonso, P. M., & Goodman, G. S. (2008). Trauma and memory: Effects of post-event misinformation, retrieval order, and retention interval. Memory (Hove, England), 16(1), 58–75. 10.1080/09658210701363146 [PubMed] [CrossRef] [Google Scholar]
- Roediger, H. L., & Karpicke, J. D. (2006). The power of testing memory: Basic research and implications for educational practice. Perspectives on Psychological Science: A Journal of the Association for Psychological Science, 1(3), 181–210. 10.1111/j.1745-6916.2006.00012.x [PubMed] [CrossRef] [Google Scholar]
- Wang, E., Paterson, H., & Kemp, R. (2014). The effects of immediate recall on eyewitness accuracy and susceptibility to misinformation. Psychology, Crime & Law, 20(7), 619–634. 10.1080/1068316X.2013.854788 [CrossRef] [Google Scholar]
- Wright, A. M., & Holliday, R. E. (2005). Police officers’ perceptions of older eyewitnesses. Legal and Criminological Psychology, 10(2), 211–223. [Google Scholar]
Articles from Psychiatry, Psychology, and Law are provided here courtesy of Taylor & Francis
