<<

APPLIED COGNITIVE PSYCHOLOGY Appl. Cognit. Psychol. 17: 203–214 (2003) Published online in Wiley InterScience 28 November 2002 (www.interscience.wiley.com) DOI: 10.1002/acp.863

Children’s and the Cognitive Interview: Do the Positive Effects Hold over Time?

ANNELI S. LARSSON,1*PA¨ R ANDERS GRANHAG1 and EMMA SPJUT2 1Department of Psychology, Go¨teborg University, Sweden 2Department of Psychology and Department of Computer Science, Sko¨vde University, Sweden

SUMMARY Most studies investigating how the Cognitive Interview affects children’s recall have employed short retention intervals (a week or less). In our study children (10–11 years old) saw a film picturing an extraordinary performance by a professional fakir. Half of the children were interviewed after seven days (n ¼ 24) and the other half after six months (n ¼ 25). At each test session, half were interviewed according to the Cognitive Interview (CI), and half according to the Structured Interview (SI). We found that: (a) the children in the CI condition recalled significantly more correct information than the children in the SI condition (both after seven days and after six months), and (b) the children interviewed after seven days recalled significantly more correct information, and less , compared to the children interviewed after six months. The results suggest that the CI can be used as an investigative tool both after short and long retention intervals. Copyright # 2002 John Wiley & Sons, Ltd.

INTRODUCTION

Many criminal investigations are dependent on information from . However, police officers commonly express that they find witnesses’ statements to be very general and incomplete (Kebbell, et al., 1999). This view is supported by archival studies focusing on the amount and type of information reported by witnesses (van Koppen and Lochun, 1997). Hence, it is important to develop interview techniques that facilitate the elicitation of eyewitness information. Since children are very vulnerable in the legal process, the development of such techniques is particularly important for this group. In the last years much has been paid to one such technique, the Cognitive Interview (see, e.g. Fisher and Geiselman 1992; Milne and Bull, 1999). Research shows that this technique is promising, but that many aspects still remain to be addressed. One is how the length of the retention interval influences the effectiveness of the technique. This aspect is important since it is not unusual that a long time passes before a is interviewed. Thus, the main question of investigation in the present study was: Do the positive effects of the Cognitive Interview hold after a longer delay?

Correspondence to: Anneli S. Larsson, Department of Psychology, Go¨teborg University, PO Box 500, SE-405 30 Go¨teborg, Sweden. E-mail: [email protected] Contract/grant sponsor: Swedish Crime Victim Compensation and Support Authority.

Copyright # 2002 John Wiley & Sons, Ltd. 204 A. S. Larsson et al.

The Cognitive Interview The Cognitive Interview rests upon two well-known principles. The first is the multi-component view of the memory trace (see, e.g. Bower, 1967). In brief, it is assumed that memory traces are multi-featured, and if one feature is not accessible using a particular type of retrieval probe, another probe might help unlock the memory. The second is the ‘-specificity’ principle, suggesting that the most effective retrieval probes are those facilitating the reinstatement of the original encoding environment (see, e.g. Tulving and Thomson, 1973). Using these principles as the point of departure, the founders of the CI, Geiselman and Fisher, suggested four (Geiselman et al., 1986): (1) the ‘mental reinstatement of context’ instruction, which asks the interviewee to reinstate mentally both the external (physical surroundings) and internal context (sub- jective states-of-mind) of the experienced event; (2) the ‘report everything’ instruction, which instructs the interviewee to report all details he or she can remember; (3) the ‘reverse-order-recall’ instruction, which encourages the interviewee to recall the event in an alternative temporal order; (4) the ‘change perspective’ instruction, where the interviewee recalls the event from an alternative perspective. Lately, different principles from the interpersonal communication paradigm have been incorporated into the original CI (see, e.g. Fisher and Geiselman, 1992; McCauley and Fisher, 1996). These principles and components meet under the term Enhanced Cognitive Interview. Since the mid-1980s a large number of studies examining the effectiveness of the CI have been conducted (for a meta-analysis see Ko¨hnken et al., 1999). The result pattern is very clear: the CI elicits more correct information than a comparison interview, both for adult and child witnesses. The increase in correct details is however often accompanied with an increase in incorrect details (for a discussion on this finding see Ko¨hnken et al., 1999). Both Geiselman and Padilla (1988) and Saywitz et al. (1992) noticed that children found it difficult to use components three and four of the original CI, and suggested that these should be excluded for situations where CI is used with younger children. To further facilitate memory performance it has been suggested that children should be ‘warmed up’ before the actual interview. This can be done by letting the child, in detail, talk about an event that s/he is very familiar and comfortable with. For a large selection of other potential ‘child-friendly modifications’ of the original CI, see Geiselman and Padilla (1988) and Granhag and Spjut (2001).

Children’s delayed recall The general finding is that children interviewed after longer delays report less correct and more incorrect information than children interviewed after shorter delays (Fivush et al., 2002). Critically, few studies have investigated how the effectiveness of different interview techniques varies with the length of the retention interval. Furthermore, very few CI studies have employed longer intervals than seven days (Ko¨hnken et al., 1999). To our knowledge, only two studies examining how the Cognitive Interview affects children’s recall have used a retention interval longer than two weeks. Memon et al. (1993) interviewed 6- and 7-year-old children after two days and then again after six weeks. Using two different interview techniques, Cognitive Interview and Standard Interview, they found no differences in accuracy or error rates between the two techniques. Furthermore, no differences were found as an effect of delay. In a study by Flin et al. (1992) both children (5–6 years and 9–10 years) and adults witnessed a staged event and were then interviewed both on the following day and after a

Copyright # 2002 John Wiley & Sons, Ltd. Appl. Cognit. Psychol. 17: 203–214 (2003) Children’s delayed recall in cognitive interviewing 205

five-month delay. The subjects were divided into three groups. The participants in group one were interviewed after one day according to an enhanced interview technique (incorporating the ‘mental reinstatement’ and the ‘report everything’ components of the CI), and on the same occasion the participants in group two were exposed to a cued recall. After five months the participants in groups one and two were questioned again, all according to a cued recall. The participants in group three were interviewed only after the five-month delay, and then according to a cued recall. The results indicate that while all witnesses forgot information over this period, the younger children recalled slightly less information than the older children and the adults. The general trend was that the error rate increased over time. In terms of correct and incorrect information, no differences were found as a consequence of the interview techniques employed.

The present study In the present study children watched a film featuring a professional fakir. After seven days, half of the children were interviewed either according to the principles of the Cognitive Interview (CI) or according to a Structured Interview (SI), used as a control interview. After six months the remaining half were interviewed according to either the CI or the SI. Our main research objective is of relevance from both a theoretical and an applied perspective. From a theoretical point of view it is of interest to map the extent to which the positive effects following the CI instructions deteriorates over time. From an applied point of view, practitioners may be interested in knowing whether the CI can be employed even after rather long retention intervals. We had four main hypotheses. First, we predicted that the children interviewed according to the CI would report more correct information—both after seven days and after six months—than the children interviewed according to the SI. This hypothesis was based on the fact that the CI has been found to give a strong overall increase in terms of correct information (Ko¨hnken et al., 1999). In addition, we think it is reasonable to assume that the mnemonics constituting the CI would be as helpful after six months as after a week (Fisher and Geiselman, 1992). Second, in line with general principles of memory (see, e.g. Baddeley, 1990) and previous findings (see, e.g. Ceci and Bruck, 1993) we predicted that the children interviewed after 6 months would remember less correct information than the children interviewed after 7 days, irrespective of the interview technique used. Third, and also in line with basic principles of memory (see, e.g. Ceci and Bruck, 1993), we predicted that the children interviewed after a long delay would report more incorrect information and more confabulations, than children interviewed after a short delay. Finally, based on previous research (see, e.g. Chapman and Perry, 1995; Granhag and Spjut, 2001), we hypothesized that there would be no differences in terms of incorrect information and confabulations between the CI and the SI.

METHOD

Design The design was a 2 2 between-subject factorial design. The first factor, the interview factor, consisted of two groups: (i) the Cognitive Interview, and (ii) the Structured

Copyright # 2002 John Wiley & Sons, Ltd. Appl. Cognit. Psychol. 17: 203–214 (2003) 206 A. S. Larsson et al.

Interview. The second factor, the delay factor, consisted of the two retention intervals (i) 7 days and (ii) 6 months. The dependent measures used were: correct information, incorrect information, confabulations, completeness, and the number and type of ques- tions put. Furthermore, we calculated the percentage accuracy, which is the proportion of correct details relative to the total number of details reported (i.e. output-bound memory performance). This measure is closely related to a statement’s degree of reliability. For an extensive discussion on input- and output-bound memory measures see, e.g. Koriat and Goldsmith (1996).

Participants The children, 29 boys and 20 girls, were recruited from a local state school. They were 10–11 years of age and the rationale behind this choice was that children this age tend to show fewer individual differences in terms of memory performance than younger children (see, e.g. Gee and Pipe, 1995). The 49 children were randomly allocated to one of the two interview conditions. Twenty-five were interviewed according to the Cognitive Interview (11 girls and 14 boys) and 24 according to the Structured Interview (9 girls and 15 boys). They were further allocated to be interviewed either after 7 days (n ¼ 24), or after 6 months (n ¼ 25).

Interview training Three female graduates of psychology conducted the interviews. They all had previous experience in interviewing children, both in using a CI and more standardised techniques. The three interviewers conducted both the CIs and the SIs. Before conducting the interviews they attended a one-day training programme in cognitive interviewing. The programme included a theoretical phase (e.g. lectures on developmental aspects on children’s memory), as well as a practical phase (e.g. receiving detailed feedback after conducting the CI).

Materials The subjects were shown a 15-minute film about a fakir. The fakir, among other things, placed himself on a bed of nails and played with fire. Judging from the children’s spontaneous reactions we were successful in finding an event which they found both fascinating and dramatic. Adding to the suspense the fakir managed to hurt himself during his performance, which caused a fair amount of bloodshed. The live event on which this film was used in previous research (Granhag and Spjut, 2001) and included a large amount of detail.

Procedure The film was shown to the children in school, and it was checked that all children could clearly see the screen. They were instructed to pay close attention to the film. Seven days after watching the film, half of the children were randomly allocated to either the CI condition or the SI condition. The children were interviewed individually. Six months later, the other half of the children were interviewed, this time, also either according to the

Copyright # 2002 John Wiley & Sons, Ltd. Appl. Cognit. Psychol. 17: 203–214 (2003) Children’s delayed recall in cognitive interviewing 207

CI or according to the SI. The interviews were conducted in rooms similar to, but not the very same as, the one where the children had watched the film.

The Cognitive Interview When conducting the CI the interviewers followed the instructions given in the Appendix. Since previous research (see, e.g. Saywitz et al., 1992) has shown that children can find it difficult to use the ‘reverse-order recall’ and the ‘change-perspective’ components, these were excluded. First, the interviewer gave the child some important pieces of information. For example, the child was informed that if a particular question was repeated during the interview this should not be taken as a sign that the previous answer to that question was incorrect (see the Appendix). The interviewers were instructed to ask open-ended questions based on the child’s free-recall, and as a minimum to include the following four: ‘What did the fakir do?’, ‘What different objects did the fakir use?’, ‘What did the fakir look like?’, and ‘What was the fakir wearing?’

The Structured Interview When conducting the SI the interviewers followed the instructions given in the Appendix. As in the CI condition, the interviewers were instructed to ask questions based on the child’s free-recall, and as a minimum to include the same four open-ended questions as mentioned above. All the interviews were audio tape-recorded and transcribed verbatim for coding.

Data coding procedure A comprehensive scoring procedure was used where information was scored as correct, incorrect or confabulated. Information was coded as correct if the child, for instance, reported that the fakir was wearing ‘black trousers’ (which he was). This information would then be scored once for ‘trousers’ and once for reporting the correct colour (‘black’). If the child reported that the fakir was wearing ‘blue trousers’, this information would be scored once as correct for ‘trousers’ and once as incorrect for reporting the incorrect colour (‘blue’). When a child falsely reported something that did not actually happen in the film, this was coded as a . Information that was mentioned several times was only scored once. The codings were based on a predefined scoring schedule used in earlier research (Granhag and Spjut, 2001), and consisted of 224 information units. One person coded all 49 interviews. Ten interviews were then randomly selected to be coded by a second coder. The interjudge-reliability proved to be 93%, that is, 336 of the total 361 information units found in the selected ten interviews were coded identically by the two coders. In order to analyse the proportion of open-ended questions asked in the interviews, the 49 interviews were coded a second time. We separated open-ended questions (e.g. ‘What did the fakir do?’) from closed questions, i.e. yes–no questions like ‘Did the fakir play with fire?’ (see Fivush et al., 2002). Fourteen interviews were then randomly selected to be coded by a second coder. The interjudge-reliability proved to be 93%, that is, 175 of the total 189 questions found in the selected fourteen interviews were coded identically by the two coders. The disagreements were solved in a conference between the two coders. It was also checked whether the SIs included any of the CI-specific components. No CI- components were found in the SI-interviews.

Copyright # 2002 John Wiley & Sons, Ltd. Appl. Cognit. Psychol. 17: 203–214 (2003) 208 A. S. Larsson et al.

Table 1. Means and standard deviations for correct information, incorrect information, confabulations and percentage accuracy for all conditions Interview technique

Cognitive Interview Structured Interview Delay Delay

7 days 6 months 7 days 6 months (n ¼ 12) (n ¼ 13) (n ¼ 12) (n ¼ 12)

Correct information M (SD) 45.92 (14.15) 28.15 (7.36) 32.42 (13.21) 11.83 (3.74) Incorrect information M (SD) 0.83 (0.94) 1.62 (1.39) 1.58 (1.44) 1.75 (1.71) Confabulations M (SD) 0.58 (0.90) 1.31 (1.03) 0.08 (0.29) 1.33 (1.15) Accuracy 97.0% 89.8% 95.8% 80.4%

RESULTS

Correct information, incorrect information and confabulations Three 2(Interview: CI versus SI) 2(Delay: 7 days versus 6 months) between-subjects ANOVAs were employed for the dependent measures correct information, incorrect information and confabulations (for means see Table 1). A significant main effect was found for the Interview factor, F(1, 45) ¼ 24.86, p < 0.05, 2 ¼ 0.36. That is, children in the CI condition reported significantly more correct information (M ¼ 36.68, SD ¼ 14.17), than the children in the SI condition (M ¼ 22.12, SD ¼ 14.16). A significant main effect was further revealed for the Delay factor, F(1, 45) ¼ 41.11, p < 0.05, 2 ¼ 0.48. Children interviewed after 7 days reported significantly more correct information (M ¼ 39.17, SD ¼ 15.06), than the children interviewed after six months (M ¼ 20.32, SD ¼ 10.14). We found no significant interaction effect for correct information, however it should be noted that after 6 months a much lower proportion of information was recalled in the SI condition than in the CI condition, compared to after 7 days. No main effects or interaction effect was found for incorrect information. For confabulations, no main effect was found for the Interview factor. However, a significant main effect was found for the Delay factor, F(1, 45) ¼ 14.40, p < 0.05, 2 ¼ 0.24. Children interviewed after 6 months reported significantly more confabulations (M ¼ 1.32, SD ¼ 1.07), than the children interviewed after 7 days (M ¼ 0.33, SD ¼ 0.70). No interaction effect was found for confabulations. Noteworthy, the proportions of both incorrect information and confabulations were very low.

Completeness and percentage accuracy (output-bound memory performance) Completeness refers to the proportion of correct information reported relative to the total amount of information found in the to-be-remembered event (in our case 224 information units). A 2(Interview: CI versus SI) 2(Delay: 7 days versus 6 months) between-subjects

Copyright # 2002 John Wiley & Sons, Ltd. Appl. Cognit. Psychol. 17: 203–214 (2003) Children’s delayed recall in cognitive interviewing 209

ANOVA showed a significant main effect for the Interview factor, F(1, 45) ¼ 24.86, p < 0.05, 2 ¼ 0.36. That is, children in the CI condition gave significantly more complete statements (M ¼ 16.38, SD ¼ 6.33), than the children in the SI condition (M ¼ 9.88, SD ¼ 6.32). Furthermore, a significant main effect was found for the Delay factor, F(1, 45) ¼ 41.11, p < 0.05, 2 ¼ 0.48. That is, children interviewed after seven days reported significantly more complete statements (M ¼ 17.49, SD ¼ 6.72), than the children interviewed after 6 months (M ¼ 9.07, SD ¼ 4.53). No significant interaction effect was found. Overall, in terms of completeness the children’s statements were rather low. It should, however, be noted that we used an elaborate scoring schedule, comprising a very large amount of details. The percentage accuracy rate (or output-bound memory performance) refers to the proportion correct information relative to the total amount of information reported (see Table 1). A 2(Interview: CI versus SI) 2(Delay: 7 days versus 6 months) between- subjects ANOVA revealed a significant main effect for the Interview factor, F(1, 45) ¼ 6.08, p < 0.05, 2 ¼ 0.12. That is, children in the CI condition gave significantly more accurate statements (M ¼ 93.27, SD ¼ 6.97), than the children in the SI condition (M ¼ 88.11, SD ¼ 11.67). A significant main effect was also found for the Delay factor, F(1, 45) ¼ 27.81, p < 0.05, 2 ¼ 0.38. That is, children interviewed after 7 days reported significantly more accurate statements (M ¼ 96.43, SD ¼ 3.30), than the children inter- viewed after 6 months (M ¼ 85.28, SD ¼ 10.89). No significant interaction effect was found. Importantly, the accuracy was overall very high.

The free-recall phase versus the question phase A pairwise T-test comparing the free-recall phase and the question phase showed no significant difference for correct information. However, for incorrect information the same test revealed a significant difference, t(48) ¼5.56, p < 0.001. The question phase (M ¼ 1.22, SD ¼ 1.25), generated significantly more incorrect information than the free- recall phase (M ¼ 0.22, SD ¼ 0.47). Furthermore, the two phases differed significantly concerning confabulations, t(48) ¼2.19, p < 0.005. The question phase (M ¼ 0.59, SD ¼ 0.86), generated significantly more confabulations than the free-recall phase (M ¼ 0.25, SD ¼ 0.63).

Number and type of questions A 2(Interview: CI versus SI) 2(Delay: 7 days versus 6 months) between-subjects ANOVAwas employed for the number of questions put in the question phase. A significant main effect for the Interview factor was found, F(1, 47) ¼ 4.44, p ¼ 0.041, 2 ¼ 0.09. There were more questions asked in the CI condition (M ¼ 13.96, SD ¼ 4.55), than in the SI condition (M ¼ 11.38, SD ¼ 4.15). No significant main effect was found for the Delay factor. Furthermore, the interaction effect proved to be significant, F(1, 47) ¼ 5.36, p ¼ 0.025, 2 ¼ 0.11. After 7 days, the interviewers asked more questions in the SI condition than in the CI condition. In contrast, after 6 months the interviewers asked fewer questions in the SI condition than in the CI condition. Finally, we investigated whether there was any difference between the conditions in terms of the type of questions asked (i.e. open-ended versus closed questions). A 2(Interview: CI versus SI) 2(Delay: 7 days versus 6 months) between-subjects ANOVA

Copyright # 2002 John Wiley & Sons, Ltd. Appl. Cognit. Psychol. 17: 203–214 (2003) 210 A. S. Larsson et al. revealed no significant main or interaction effect. In short, the relation between open and closed questions did not vary as a consequence of interview technique or retention interval. In short, about 9 out of 10 questions were open-ended.

DISCUSSION

Our prediction that children interviewed with the Cognitive Interview would report more correct information than children interviewed with the Structured Interview was con- firmed. Importantly, and in line with our prediction, this finding held for both the short delay and the long delay. Hence, our results support previous findings showing that the CI enhances children’s recall (Ko¨hnken et al., 1999), and expands the literature by showing that the technique can be helpful also after longer delays. Interestingly, the children interviewed with the CI after 6 months recalled almost as much correct information as did the children interviewed with the SI after one week. Differently put, our findings suggest that even if a rather long time has passed between the to-be-remembered event and the interview, a child’s memory can be facilitated if he or she is asked to mentally reinstate the external and internal context, and to report everything he or she can remember. However, future componential analyses are needed in order to map how the effectiveness of different CI mnemonics is affected by the length of the retention interval (see, e.g. Milne and Bull, in press). Flin et al. (1992) did not find an increase in correct information after using an enhanced interview after one day and then a cued recall after 5 months, compared to using a cued recall on both occasions. Discussing their findings the authors recognize that the result might have been different had they adopted a more sophisticated version of the CI. Our findings support their idea. It should also be noted that we used a CI after both the short and the long delay. This was not the case in the study reported by Flin et al. Recent CI research has shown that the two components used in the current study, i.e. ‘mental reinstatement of context’ and ‘report everything’, work well in tandem (see, e.g. Milne and Bull, in press). Importantly, recent research has also shown that the ‘mental reinstatement of context’ instruction can help children as young as 4–6 years (Herskowitz et al., 2001), and that the third component of the CI, the ‘change-perspective’ instruction, can be beneficial with children as young as 5–6 years (Milne and Bull, in press). Our second prediction, that children interviewed after the shorter delay would remember more correct information than children interviewed after the longer, was supported. This finding is well in line with previous eyewitness research (see, e.g. Ceci and Bruck, 1993; Quas et al., 1999). Our third prediction, that more incorrect information and more confabulations would be reported after 6 months than after a week, was partially confirmed. The children interviewed after 6 months showed a non-significant increase in incorrect information. Further, and in support of our prediction, we found that children interviewed after 6 months reported significantly more confabulations than the children interviewed after 7 days. These findings are probably due to the fact that memory traces weaken over time, and that the likelihood that different knowledge-driven constructive processes are used in order to fill memory gaps increases for longer delays (Ornstein and Haden, 2002). It should, however, be noted that the proportions of recalled incorrect information and confabulations overall were low; when collapsing conditions and occasions less than 5% of the recalled information was incorrect and less than 2.5% was confabulated. Put differently, the children’s statements were very high in terms of

Copyright # 2002 John Wiley & Sons, Ltd. Appl. Cognit. Psychol. 17: 203–214 (2003) Children’s delayed recall in cognitive interviewing 211 reliability, a finding which is supported both by general memory research on children (Fivush et al., 2002), and by CI research (Ko¨hnken et al., 1999). Finally, our fourth prediction, that there would be no differences between the interview techniques in terms of incorrect information and confabulations, was confirmed. Previous research on children’s recall and the CI shows a somewhat mixed pattern in terms of incorrect information, with some studies showing that the CI causes an increase in incorrect information (see, e.g. McCauley and Fisher, 1995; Memon et al., 1997), whereas other studies show no increase in incorrect information (see, e.g. Granhag and Spjut, 2001; Memon et al., 1993). We believe that one important reason why we were able to avoid an increase in incorrect information and confabulations was that we used an enhanced version of the CI, incorporating components that worked for a decrease in memory errors. For example, when helping the children to reinstate mentally, the interviewers were very careful not to use words like ‘pretend’ and ‘imagine’, or other words that could trigger fantasizing. In previous research where the CI has resulted in an increase of incorrect information (see, e.g. McCauley and Fisher, 1995), it has not been highlighted that it is important to avoid words that may trigger fantasizing. Our finding that the CI did not elicit more confabulations than the SI supports previous findings (see, e.g. Chapman and Perry, 1995; Memon et al., 1993). For an exception, see Hayes and Delamothe (1997), who found the CI to be associated with a significant increase in confabulations for children interviewed after a 3-day interval. In addition, we found that information gathered with the CI was significantly more accurate than information gathered with the SI, a finding which further speaks for the superiority of the CI. This result is in contrast to previous research, which shows that information elicited with the CI tends to be as accurate as information elicited by conventional interviews (Fisher et al., 2002). We also found that the question phase generated significantly more incorrect informa- tion and confabulations, compared to the free-recall phase. This result is well in line with previous research showing that as the interviewer moves from free-recall to open-ended questions (or for that matter, from open-ended questions to closed), the number of incorrectly reported items tend to increase (see, e.g. Dent and Stephenson, 1979). We found that more questions were asked in the CI condition compared to the SI condition, which supports previous findings (see Memon, 1998). Interestingly, for our study this result was mainly due to the difference found after the 6-month delay. That is, the number of questions asked in the question phase increased over time for the CI condition, whereas it decreased for the SI condition. Reasonably, this might have contributed to the finding of more correct information reported in the CI condition. However, it should be noted that the relationship between the number of questions asked and the amount of information reported is far from straightforward. For example, in the present study the number of questions asked after a week was about equal comparing the CI and the SI condition. Still, significantly more correct information was generated in the CI condition. Furthermore, for the CI condition more questions were asked after 6 months than after a week. Still, more correct information was generated after a week than after 6 months. Hence, it is reasonable to assume that the relationship between the number of questions asked and the amount of information recalled is moderated by a host of factors, for example (a) the degree of completeness and general quality of the previously given free-recall, (b) the extent to which the interviewer feels motivated to ask further questions and (c) the extent to which the mnemonics employed facilitate the activation of the interviewee’s memory. Finally, for both interview conditions the large majority of all questions asked were open-ended.

Copyright # 2002 John Wiley & Sons, Ltd. Appl. Cognit. Psychol. 17: 203–214 (2003) 212 A. S. Larsson et al.

CONCLUSIONS

Although the results in the present paper is based on a relatively small sample size we believe that the main conclusions to be drawn are of importance to practitioners who interview children in forensic situations. First, expanding on previous research, we were able to show that the superiority of the Cognitive Interview held after a rather long delay. Future research is needed in order to investigate if this finding holds for even longer delays. Importantly, we found that the use of the CI increased the children’s recall of correct information without causing an increase in incorrect information and confabula- tions. Second, the children’s statements were rather low in terms of completeness, but high in terms of reliability (for an overview see Ceci and Bruck, 1995). Third, after a delay of 6 months children recalled significantly less correct information and more confabulations, than the children interviewed after a week. This stresses the importance of interviewing a child as soon as possible after he or she has experienced the event of interest. Fourth, we showed that the significant part of the incorrect information and confabulations recalled stemmed from the question phase. Future research is needed to clarify to what extent this finding is due to that the question phase was preceded by a free-recall phase. In this connection it should be noted that most researchers on investigative interviewing advocate the idea that witnesses, whether adults or children, should be encouraged to give a free-recall (see, e.g. Lamb et al., 1999; Milne and Bull, 1999). Nevertheless, studies mapping real life police interviews (see, e.g. Fisher et al., 1987) show that many officers never allow for a free-recall, but instead ask a high number of specific questions. This exemplifies the challenge facing social scientists trying to filter down research findings to the communities that most profit from them, in this case police officers and social workers.

ACKNOWLEDGEMENTS

This research was supported by a grant from the Swedish Crime Victim Compensation and Support Authority given to the second author. A portion of the research was presented at the 11th Conference on Psychology and Law, Lisbon, Portugal (5–8 June 2001).

REFERENCES

Baddeley A. 1990. Human Memory—Theory and Practice. Lawrence Erlbaum: Hove. Bower GH. 1967. A multicomponent view of a memory trace. In The Psychology of Learning and Motivation, Vol. 1, Spence KW, Spence JT (eds). Academic Press: New York; 230–325. Ceci SJ, Bruck M. 1993. of child witness: a historical review and synthesis. Psychological Bulletin 113: 403–439. Ceci SJ, Bruck M. 1995. Jeopardy in the Courtroom: A Scientific Analysis of Children’s Testimony. American Psychological Association: Washington, DC. Chapman AJ, Perry DJ. 1995. Applying the cognitive interview procedure to child and adult eyewitnesses of road accidents. Applied Psychology: An International Review 44: 283–294. Dent H, Stephenson GM. 1979. An experimental study of the effectiveness of different techniques of questioning child witnesses. British Journal of Social and Clinical Psychology 18: 41–51. Fisher RP, Geiselman RE. 1992. Memory-Enhancing Techniques for Investigative Interviewing: The Cognitive Interview. Charles C Thomas: Springfield, IL. Fisher RP, Brennan KH, McCauley MR. 2002. The cognitive interview method to enhance eyewitness recall. In Memory and Suggestibility in the Forensic Interview, Eisen ML, Quas JA, Goodman GS (eds). Erlbaum: Mahwah, NJ; 265–286.

Copyright # 2002 John Wiley & Sons, Ltd. Appl. Cognit. Psychol. 17: 203–214 (2003) Children’s delayed recall in cognitive interviewing 213

Fisher RP, Geiselman RE, Raymond DS. 1987. Critical analysis of police interviewing techniques. Journal of Police Science and Administration 15: 177–185. Fivush R, Peterson C, Schwarzmueller A. 2002. Questions and answers: the credibility of child witnesses in the context of specific questioning techniques. In Memory and Suggestibility in the Forensic Interview, Eisen ML, Quas JA, Goodman GS (eds). Erlbaum: Mahwah, NJ; 331–354. Flin R, Boon J, Knox A, Bull R. 1992. The effect of a five-month delay on children’s and adults’ . British Journal of Psychology 83: 323–336. Gee S, Pipe M-E. 1995. Helping children to remember: the influence of object cues on children’s accounts of a real event. Developmental Psychology 31: 746–758. Geiselman RE, Fisher RP, MacKinnon DP, Holland HL. 1986. Eyewitness memory enhancement in the cognitive interview. American Journal of Psychology 99: 385–401. Geiselman RE, Padilla J. 1988. Cognitive interviewing with child witnesses. Journal of Police Science and Administration 16: 236–242. Granhag PA, Spjut E. 2001. Children’s recall of the unfortunate fakir—a further test of the enhanced cognitive interview. In Psychology in the Courts: International Advances in Knowledge, Roesch R, Corrado RR, Dempster RJ (eds). Routledge: London; 209–222. Hayes BK, Delamothe K. 1997. Cognitive interviewing procedures and suggestibility in children’s recall. Journal of Applied Psychology 82: 562–577. Herskowitz I, Orbach Y, Lamb ME, Sternberg KJ, Horowitz D. 2001. The effects of mental context reinstatement on children’s accounts of sexual abuse. Applied Cognitive Psychology 15: 235–248. Kebbell MR, Milne R, Wagstaff GF. 1999. The cognitive interview: a survey of its forensic effectiveness. Psychology, Crime and Law 5: 101–115. Ko¨hnken G, Milne R, Bull R, Memon A. 1999. The cognitive interview: a meta-analysis. Psychology, Crime, and Law 5: 3–27. Koriat A, Goldsmith M. 1996. Memory metaphors and the real-life/laboratory controversy: correspondence versus storehouse conceptions of memory. Behavioral and Brain Sciences 19: 167–228. Lamb ME, Sternberg KJ, Orbach Y, Hershkowitz I, Esplin PW. 1999. Forensic interviews of children. In Handbook of the Psychology of Interviewing, Memon A, Bull R (eds). John Wiley: Chichester; 253–277. McCauley MR, Fisher RP. 1995. Facilitating children’s eyewitness recall with the revised cognitive interview. Journal of Applied Psychology 80: 510–516. McCauley MR, Fisher RP. 1996. Enhancing children’s with the cognitive interview. In Psychology and Criminal Justice: International Developments in Research and Practice, Davies G, Lloyd-Bostock S, McMurran M, Wilson C (eds). Walter de Gruyter: Berlin; 127–134. Memon A. 1998. Telling it all: the cognitive interview. In Psychology and Law: Truthfulness, Accuracy and Credibility, Memon A, Vrij A, Bull R (eds). McGraw-Hill: Maidenhead; 170–187. Memon A, Cronin O, Eaves R, Bull R. 1993. The cognitive interview and child witnesses. Issues in Criminological and Legal Psychology 20: 3–9. Memon A, Wark L, Bull R, Ko¨hnken G. 1997. Isolating the effects of the cognitive interview techniques. British Journal of Psychology 88: 179–197. Milne R, Bull R. 1999. Investigative Interviewing: Psychology and Practice. John Wiley: Chichester. Milne R, Bull R. in press. Back to basics: a componential analysis of the original cognitive interview mnemonics with three age groups. Applied Cognitive Psychology. Ornstein PA, Haden CA. 2002. The development of memory: toward an understanding of children’s testimony. In Memory and Suggestibility in the Forensic Interview, Eisen ML, Quas JA, Goodman GS (eds). Erlbaum: Mahwah, NJ; 26–61. Quas J, Goodman G, Bidrose S, Pipe ME, Craw S, Ablin D. 1999. and memory: children’s long term, remembering, and suggestibility. Journal of Experimental Child Psychology 72: 235–270. Saywitz KJ, Geiselman RE, Bornstein GK. 1992. Effects of cognitive interviewing and practice on children’s recall performance. Journal of Applied Psychology 77: 3–15. Tulving E, Thomson DM. 1973. Encoding specificity and retrieval processes in . Psychological Review 77: 1–15. van Koppen PJ, Lochun SK. 1997. Portraying perpetrators: the validity of offender descriptions by witnesses. Law and Human Behavior 21: 661–685.

Copyright # 2002 John Wiley & Sons, Ltd. Appl. Cognit. Psychol. 17: 203–214 (2003) 214 A. S. Larsson et al.

APPENDIX: SPECIFICATIONS OF THE TWO INTERVIEW TECHNIQUES

SI CI Instructions given to the interviewers —Establish rapport x x —Do not interrupt the witness—listen actively x x —Avoid words like ‘imagine’ and ‘pretend’ x x —Avoid leading questions x x —Ask open-ended questions x x —Ask one question at a time x x —Pause after the interviewee’s response x x —Use age-appropriate language x x —Ask witness-compatible questions x x —Ask questions based on the interviewee’s free-recall x x Instructions given to the interviewee —It’s okay to answer ‘I don’t know’ x x —S/he should not make up any answers x x —S/he should ask the interviewer to explain if s/he doesn’t understand a question x x —The previously given answer isn’t wrong if a question is repeated x x —S/he should generate information on his/her own initiative x x —S/he should try to be concentrated throughout the interview x x —S/he should follow the ‘mental reinstatement of context’ instruction x —S/he should follow the ‘report-everything’ instruction x Order of the interview —A free-recall phase x x —A question phase x x

Copyright # 2002 John Wiley & Sons, Ltd. Appl. Cognit. Psychol. 17: 203–214 (2003)