REDUCING SUGGESTIBILITY IN CHILD WITNESS INTERVIEWS

Abstract

This study set out to test the prediction that a Cognitive Interview may increase resistance to subsequent misleading suggestions in child witness interviews. The misleading information was presented in the form of questions both prior to, and after, a cognitive or structured interview to 8 and 9 year old witnesses to a video-taped event. Use of the cognitive interview resulted in more correct responses to post-interview questions than did the structured interview eventhough there was not quite a significant effect of the cognitive interview on information recalled during the actual interview. On the basis of their interview performance the children were classified as `intruders' or `non-intruders' (i.e. those children who intruded pre-interview misleading items into the subsequent interview and those who did not). The `non-intruders' made significantly fewer errors on the post-interview questions indicating lower vulnerability to misleading information. Moreover, those children who selected the `don't know' option made fewer errors in the interview and were more accurate in their responses. Theoretical and practical implications of the data are discussed in the context of group differences in vulnerability to suggestion and techniques for increasing resistance to suggestion.

In the last two decades the increase in reports of child abuse has led researchers working in the area of children's testimony to focus on the vulnerability of child witnesses in interview situations. For example, a large body of research has set out to identify how easily children may succumb to misleading suggestions provided by an authority figure (Ackil and Zaragoza, 1995; Ceci and Bruck, 1995; 1993a; 1993b; Leichtman and Ceci, 1995; Marche and Howe, 1995; Poole and Lindsay, 1995). This type of research has provided some clues as to why and how misleading information effects may occur but has tended to be negative in its focus. With an increased concern about procedures for collecting evidence from child witnesses (Bull, 1995a; Davies, Wilson, Mitchell and Milson, 1995), there is a need for positive research directed at identifying situations under which testimony may be improved. Most scholars would concur with the view that the manner in which witnesses are interviewed can have a profound impact on how much they report. To echo recent reviews we can do much to improve the quality of interviewing and in doing so elicit valuable information from those children who can be competent witnesses (Bull, 1995; Lamb, Sternberg and Esplin, in press).

An important component in the evaluation of any investigative technique that is new to professional investigators is to demonstrate that it does not result in increased susceptibility to misleading information. The extent to which the Cognitive Interview (CI) may increase resistance to subsequent misleading suggestions is the primary question to be addressed in the current study. The effects of misleading suggestions presented prior to an interview on subsequent interview performance will also be examined.

The CI and resistance to subsequent misleading questions

The CI is a technique for interviewing witnesses that attempts to improve both memory retrieval and dyadic communication in a witness interview (Fisher and Geiselman, 1992; Koehnken, 1995; Memon and Koehnken, 1992; Memon, in press). One of the principal CI techniques is the mental reinstatement of the physical and personal contexts which existed at the time of the event (e.g. Eich, 1980). The other CI strategies include instructions to recall in reverse order, and to report everything even partial information in order to reduce the extent to which prior knowledge, expectation and schemata limit recall. In addition to the `cognitive' components, the CI in its current form places considerable emphasis on social communication techniques. The latter include the `transfer of control' of the interview from the interviewer to the witness. This technique is put into place during the rapport-building phase and allows witnesses to dictate the pace of the interview and structure their own recall. This is achieved through use of open questions, by not interrupting witnesses, by timing questions carefully so that they are related to witnesses' retrieval patterns and not to a set protocol of the interviewer.

The CI was developed, tested and refined in a series of studies employing trained interviewers (police and students), simulated incidents and police training films and was found to produce significant increases in the amount of correct information elicited (e.g. Geiselman, Fisher, MacKinnon and Holland, 1985). Only one study has investigated how an adult witnesses interviewed with a CI fare when presented with misleading suggestions after the interview. Geiselman, Fisher, Cohen, Holland and Surtes (1986) asked (a) whether CI increases or reduces suggestibility, and (b) how critical the timing of the CI was. They reasoned that the CI, by guiding the witness back to the original memory record, strengthened memory for original event details thereby `inoculating' against any subsequent misinformation effects. In the relevant study (Geiselman et al, 1986, Experiment 3) college students viewed a video recording of a crime and a few minutes later were instructed in the use of CI techniques. They were then asked to write an account of the event. The standard interview (control) group waited for six minutes and then wrote their narrative account. The witnesses were subsequently presented with a series of leading and misleading questions. The CI group were significantly less affected by the suggestive questioning and were more likely to respond `don't know' and to query the contradictory information that was provided. Geiselman et al concluded that experiencing a CI before suggestive questioning can reduce the effects of suggestion by facilitating the retrieval or original details (this will be referred hereafter as the 'Geiselman-effect'). The Geiselman et al study leaves two questions open: firstly, whether the inoculation effect with the CI would also appear with children who are known to be more vulnerable to suggestions (see Ceci and Bruck, 1993a) and secondly, whether any effect would still appear after a longer delay. Memory is weaker after a longer delay and, therefore, more vulnerable to suggestions (e.g. Ackil and Zaragoza, 1995). To answer these questions, the Geiselman et al. study was replicated with children using a longer delay between exposure and interview. An initial study conducted in our laboratory with child witnesses suggested the CI may increase resistance to suggestion (Milne, Bull, Koehnken and Memon, 1995). We set out to replicate this effect.

Identifying vulnerable groups and reducing suggestibility

It is clear from studies conducted with adults that situational pressure and characteristics of an interviewee will influence yield to suggestion (see Gudjonsson, 1992 for a review). There are two studies using children which have attempted to measure vulnerability using the Gudjonsson Suggestibility Scale (GSS), a research and clinical tool to assess responses to `misleading questions' and `negative feedback' in the context of recalling a particular event (Gudjonsson, 1987). Warren, Hulse-Trotter and Tubbs, (1991) administered the GSS to 7-year olds, 12 year olds and adults. The younger children's scores on the GSS indicated greater vulnerability to suggestion than older children and adults. Danielsdottir, Sigurgeirsdottir, Einarsdottir and Haraldsson (1993) obtained similar data in a study using Icelandic children. However, these studies do not address the more specific question of whether some children are in fact more likely to be affected by misleading suggestions than are others. It would be useful to have some data on this question as this knowledge may enable the police or jurors to estimate the potential effects of suggestions prior to an investigative interview.

In the current study, rather than using a scale to measure suggestibility, we classified children into `high' and `low' vulnerability groups based upon their responses to misleading questions presented prior to the interview. We then examined the impact of incorrect responses to misleading questions on interview performance as detailed in the next section.

Effects of Misleading information upon a subsequent interview

One part of this study (as described above) was concerned with the effects of a CI or SI on post-interview suggestions. A second purpose of the current study was to examine whether or not misleading information that was given prior to a properly conducted interview carries over into the interview. This question has been neglected in previous research. Leichtman and Ceci (1995) in their study of preschooler's (3-6 year olds) eyewitness reports examined the effects of repeated interviews with different degrees and types of suggestive influences. Following the visit of a stranger (Sam Stone) to their nursery, some children were interviewed with suggestive questions about the event once a week for the four weeks following the visit. The fifth interview experienced by all children was conducted ten weeks after the event by a new interviewer who was not present during the visit of the stranger or during the prior four interviews. In this final interview all children were asked for a free report and then probed with questions. The basic finding was that prior repeated suggestions increased false reporting. The effects were observed both in the free report and in the question phases of the final interview.

The Leichtman and Ceci (1995) study is one of the first to have noted effects of suggestions on free recall performance, which is generally assumed to produce an uncontaminated account (but see Warren and Lane, 1995; Poole and Lindsay, 1995). Moreover, the Leichtman and Ceci study demonstrated that an event which is inaccurately reported in free recall will continue to be inaccurately reported when probed with questions. In other words, a prior erroneous report increases the likelihood of making a subsequent inaccurate report. Furthermore, the children in the Leichtman and Ceci study tended to embellish their statements with more and more confabulated details after a series of interviews. The Leichtman and Ceci procedure is an example of quite intense suggestion. Although this may happen in real cases (see Ceci and Bruck, 1995 for examples) it would also be relevant to look at the effects of one inappropriate interview (i.e. containing some suggestive questions). This may, for example, happen if parents are suspicious that their child had been sexually abused and then try to verify this suspicion. Since they are unlikely to be experienced interviewers, the chances are that they will ask at least some suggestive questions. The question then is whether or not a single suggestive interview or an interview with a few suggestive questions will have the same devastating effects on the quality and evidential value of a later statement as the repeated suggestions of the Leichtman and Ceci type. Moreover, will the misleading information intrude into the subsequent `good' interview?

The effects of interview technique

Much of the early work on the CI has not specified exactly what techniques are used by the control groups who are referred to as `standard interviewers' (see Memon and Bull, 1991 review). It is important to compare the CI with a proper control group in order to understand what techniques are likely to influence yield to suggestions. In line with our previous work on the CI (e.g. Memon, Wark, Bull and Koehnken, in press; Mantwill, Koehnken and Aschermann, 1995) a Structured Interview (SI) was used as a control. Our SI interviewers received the same quality of training as the CI interviewers in communication and questioning techniques (but without the specific CI techniques). Moreover, the SI resembled procedures that are recommended to professionals who interview children (see Bull, 1995b; Yuille, Hunter, Joffe and Zaparniuk, 1993).

The current study focused on only one age group of children, 8-9 year olds. The reason for selecting this age group is that previous research has demonstrated that children of this age are old enough to benefit from the CI instructions (Milne et al., 1995; Memon et al, in press; Koehnken, Finger, Nitschke, Hofer and Aschermann, 1992; Saywitz, Geiselman and Bornstein, 1992). Children under the age of 7 years do not appear to have sufficient metamemory skills to use the CI effectively (Cronin, Memon, Eaves, Kupper and Bull, 1992; Memon, Cronin, Eaves and Bull, 1993).

Method

Participants

One hundred and thirteen children watched a short video clip of a magic show in small groups (see below). The children were 8-9 years of age with a mean age of 8 years 7 months. Ninety-seven children were subsequently interviewed (the remainder were either absent or their parents did not wish them to be interviewed). There were 50 children in the SI condition (25 girls and 25 boys) and 47 (25 girls and 22 boys) in the CI condition.

Interviewers and training

Four interviewers (three female and one male student,) were trained solely in the use of the Cognitive Interview (CI) and four (three female and one male student) in the use of the Structured Interview (SI). The interviewers questioned an approximately equal number of children. The training of CI and SI interviewers took place over two days in separate sessions. For each group the training began with an introductory lecture on the importance of the interview in psychological assessment and information gathering. Both groups were also given guidelines about non-verbal behaviour in the interview (e.g., seating position, eye contact, pauses and speech rate). For each phase of the interview (see below) there was a demonstration role-play excerpt which was followed by a live practice role-play (interviewers were asked to choose an event, playing the part of interviewee and interviewer) these latter being video-recorded. The interviewers then received feedback on their interviews. There were plenary discussion/question sessions at the end of the training. In addition to active role-plays, the interviewers were encouraged to rehearse mentally the various phases of the interview.

The SI and CI interviewers practised the `rapport building' phase. This entailed making casual conversation with the witness by asking open-ended questions (choosing a general topic of interest to the child), informing the witness that they (the interviewers) did not have any knowledge of the event and that it was the witness who should dictate the pace of the interview. The last two instructions are part of the `transfer of control' procedure.

In the free recall (FR) phase, the SI interviewers were asked to request a free narrative account from the witness. The CI group received identical instructions for the free report but in addition they were given training in encouraging witnesses mentally to reinstate context (as described in this paper's introduction) prior to obtaining a free report. The CI interviewers also incorporated at this stage the instruction to report everything.

In the questioning (QU) phase, the CI and SI interviewers were asked to use the information reported by that witness in free recall as the sole basis for the questions. (Both the CI and SI interviewers had also been similarly instructed in the use of appropriate types of questions). They were asked to begin with open questions and then follow these with closed questions. Interviewers were asked to use the QU phase to find out more about who was present at the event and what they did. Where a person was mentioned, interviewers were asked to elicit details about clothing. They were specifically instructed to avoid leading, misleading, and forced-choice questions. The CI interviewers received additional training in the activation and probing of images relating to various parts of the event. As part of their training both the CI and SI interviewers were asked to warn the witnesses that they must not fabricate responses. Thus, the training of the CI and SI interviewers differed only in terms of the techniques of context reinstatement, report everything and the activation and probing of images.

The Event

A short video clip of a magic show was chosen as the to-be-remembered event. The sequence showed the magician (a large bald headed man dressed in a purple suit) standing in front of a group of six children. He performed a series of tricks and for one trick enlisted the help of two children from the audience. It was felt that the event would be of interest to the children thus maximising the information available for recall. The video clip lasted for approximately five minutes and was shown to children in groups of 5-6. The children were asked not to talk about the content of the video.

The Misleading Questions

Each child was presented with a total of nine questions (misleading and filler) both prior to, and after the interview (see Appendix I for the questions). The delay between administration of pre and post interview questions was approximately 9 minutes (equal to the mean length of interviews). The questions were presented in a different random order to each child, such that some children had particular questions pre-interview whereas other children had the questions post-interview. The questions were presented by the interviewer who read out each item to the child and recorded the response given.

Procedure

Each child was individually interviewed twelve days after the event was shown (average delay was 12.02 days in the CI and 12.25 in the SI). Each interview was audio-taped. Children were randomly allocated to interviewers. The interviewers were personally introduced to the children by their teacher who then accompanied the children to the testing room. The interviewers engaged in rapport with the children along the way. Once seated the interviewer continued to build rapport and explained the interview procedure. Data collection began with the pre-interview questions being administered orally to the child. The interviewer explained that s/he had not seen the magic show video and so did not know the answers to the questions (this was true). Children were asked to report only what had really happened and not to guess or make anything up. The children were reminded several times that it was acceptable to say "I don't know[1]". Once the interviewer had gone through all the pre-interview items s/he told the child about what was going to happen next (i.e. the interview). The CI interviewers began with the context reinstatement instructions (e.g. "Picture yourself back in the room where you watched the video. What could you see and hear in the video...") and also requested that the child tell them everything they could remember. They then elicited a free report. The SI interviewers simply elicited a free report at this stage. The children were asked to tell the interviewers about the magic show in their own words. The interviewers explained that they were going to ask some questions about what the children had just told them. The CI interviewers included the "tell me everything" instruction at this point. During the questioning phase of interview (which focused only on things mentioned by the child in the free recall phase) the interviewers asked the children mainly open and closed questions rather than leading questions, with the CI group activating and probing specific images again using the context reinstatement technique (e.g. "You mentioned a magician, I want you to try and get a picture of the magician's face in your head? Can you see it? I want you to concentrate on the magician's face and tell me what it looked like."). At the end of the interview, the interviewer explained to the child in simple terms why they should not discuss the interview with the other children.

Coding and scoring of the transcripts

The interviews were transcribed verbatim and the resultant details coded and scored. This was done by reference to a full record of what had occurred in the witnessed magic show. The children's recall was classified as either correct, as error (e.g., saying black rope instead of the white rope which was present) or as confabulation (mentioning a detail that was not present).

The position the recalled information occupied within the interview was noted [free report (FR) or questioning phase (QU)]. After a child's FR was scored, information appearing in the QU of the interview was only scored if it was new. Given that information at the QU phase of the interview was related to what was reported earlier in FR, interview phase was not treated as a repeated measures factor in the analyses.

Any details mentioned by children during the interview which were not in fact present in the witnessed magic show but had been contained in the pre- interview suggestions were noted separately as `intrusions'. We were also interested in whether or not any pre-interview misleading information negatively influenced aspects of the subsequent statement beyond that misleading information (Leichtman and Ceci, 1995). To answer this questions we distinguished `intrusions' from `confabulations' (the latter comprised errors pertaining to details that were not part of the misleading suggestions).

There were two independent coders who coded/scored the transcripts. Eight transcripts were scored by both of the coders and compared. The correlations of the two coders' scores were for total accurate, r=.93 p<.001, for total error, r=.72 p<.05 and for total confabulations, r=.78 p<.05.

Results

The presentation of the results follows the sequence of questions being addressed in the current study as set out in the Introduction. The first question is whether or not the CI increases resistance to subsequent misleading questions with children.

The CI and resistance to subsequent misleading questions

In order to address the hypothesis that the CI inoculates against subsequent misleading suggestions (the `Geiselman-effect') it should be established that there were no pre-existing CI-SI differences in responses to misleading questions prior to the interviews. Thus it was necessary to compare group differences (CI/SI) in responses to the pre-interview misleading questions first. The second purpose of this analysis was to separate children into those who are more vulnerable to misleading questions from children who are less vulnerable.

There were three types of response to pre-interview responses: correct, incorrect or `don't know.' A series of t-tests showed no significant group differences on any of these measures (all ps>.05).

An examination of group differences in responses to post-interview questions did however, yield significant effects. There were significantly more correct responses to the post-interview questions in the CI condition as compared to the SI condition supporting the `Geiselman-effect' (t (98)=-2.99, p<.01) and replicating Milne et al, (1995). There were no significant differences in incorrect responses or `don't know' responses as a function of interview condition. The mean responses are presented in Table 1.

Table 1: Mean correct, incorrect and don't know responses as a function of interview condition.

		Cognitive	Structured
Pre-interview	correct	2.25	1.86
	error	2.71	2.83
	D/K	4.04	4.31
Post-interview	correct	2.79	1.74
	error	3.58	3.88
	D/K	2.73	3.18

Note D/K = "don't know" response.

Effects of Misleading information upon a subsequent interview

This part of the analysis asks whether or not misleading information given prior to a properly conducted interview carries over into the interview and whether children who are more vulnerable to misleading information differ in their interview performance from children who are less vulnerable.

It was found in the pre-interview analysis that there was considerable variation in how accurate children were in response to the misleading questions. The question was if this was just random variation or whether some children are habitually more vulnerable to misleading information and suggestions. If this were the case, the children who are more affected in the pre-interview questions ought to show poorer performance in the interview and post-interview misleading questions. To analyse this question children were classified into high and low suggestibility groups. This was operationalised by whether or not a child `intruded' items from prior misleading questions into the subsequent interview.

This section examines the responses of children who mentioned within their interviews incorrect information which related to prior misleading questions[2] (an `intrusion'). If a child made one or more intrusions in their interview, they were deemed an `intruder'. On average children made three intrusions. Of the 97 witnesses, 39 were `intruders' (20 were in the CI condition and 19 in the SI condition) and 53 were `non-intruders' (24 in the CI condition and 29 in the SI condition). Analysis of the proportion of details that were accurate (i.e percentage correct, incorrect or confabulated divided by total number of details reported) showed that the `non-intruders' recalled a significantly higher proportion of accurate details (means=.76 and .59 for non-intruders and intruders respectively, t (95)=-.61, p<.0001).

(i) Where in the interview do the intrusions appear?

Of the 39 intruders, 31 made at least one intrusion in the free recall part of the interview. This is an important finding since it is generally assumed that free recall yields accurate performance despite the findings of several recent studies (Warren and Lane, 1995; Poole and Lindsay, 1995; Leichtman and Ceci, 1995). The remaining 8 `intruders' made intrusions in the questioning phase only. Overall, there were were twice as many intrusions in the questioning phase as compared to the free recall. This is hardly surprising since interviewer questions were guided by what was reported by witnesses in the free recall stage (see method)

(ii) Intruders/non-intruders and interview technique.

Having established that some children's will intrude incorrect details arising from an earlier test into their interview responses, we wanted to see if such intrusions are more likely to occur in a CI as compared to an SI. We therefore used our classification arising from our first set of analyses, in order to look further for evidence of group differences on interview performance and interaction effects.

A series of 2 X 2 ANOVAs (intruders/non-intruders, CI/SI[3]) on the information provided in the interview found the intruders were significantly less accurate than non-intruders (F (1,93)= 36.90, p<.001). (Accuracy was defined as the proportion of correct details as a function of the total amount reported including errors and confabulations). This difference was due largely to the intruders making more confabulations (F (1,93)= 45.51, p<.001). (Note, as stated above, that confabulations do not include incorrect information related to prior misleading questions). There was no effect of interview technique and no significant interaction effects (F<1).

Our classification of children into two groups based upon the extent to which they `intruded' items from prior misleading questions into the subsequent interview provides us with one way of distinguishing high and low vulnerable groups. These data suggest that some children are more vulnerable to the effects of misleading questions than are others and that these children perform poorer in a subsequent interview (in free recall and in questioning). Further evidence of this is seen in the correlations between responses to prior misleading questions and subsequent interview performance.

(iii) Correlations between pre interview question responses and interview recall.

A significant positive correlation was found between the number of correct responses on the pre-interview questions and the amount of correct interview recall obtained (r = .35, p <.01). There was also a significant positive correlation between errors on the pre-interview questionnaire and (i) interview errors (r= .28, p<.05), (ii) interview confabulations (r = .27, p<.05). There were also significant correlations between the number of `don't know' responses on the pre-interview questions and the proportion of interview information that was accurate (r = .43, p<.001). Thus the more a child responded `don't know' to the pre-interview questions, the better was his or her interview performance.

(iv) Correlations between interview recall and post-interview question responses.

With regard to responses to the post-interview questions, we again find some differentiation across children. Not only were there significant correlations between the pre and post-interview for correct responses (r =.29, p<.01), for errors (r = .52, p<.001 ) and `don't knows' (r = -.43, p<.01), those children who were more accurate in their interview provided more correct responses to the post-interview questions (r = .32, p<.01). The children who gave the greater number of incorrect responses to the post-interview questions had more confabulations in the interview (r = .29, p<.01). Furthermore, those who had fewer `don't know' responses to the post-interview responses had produced more errors (r = .31, p<.01) and more confabulations in the interview (r = .38, p<.01). These latter two findings suggest that low frequency of `don't know' responses to subsequent misleading questions could be an index of likely error/confabulation in an interview, however well such an interview is conducted.

To summarise the children who were more accurate in the pre-interview questions were also more accurate in in the interview and in their responses to post-interview questions.

The effects of a Cognitive versus Structured Interview on recall

In order to compare CI and SI performance in the interview, a series of t-tests were performed. Contrary to expectation, there were no significant differences in recall performance across the Cognitive and Structured interview conditions in the amount of correct details (t (95) =-1.23) , errors ( t (95) =-.35 ) or confabulations (t (95) = -.07). Breaking down the analysis by stage of interview showed no significant differences in the FR stage. There was however, a non-significant trend with more correct details reported in the CI as compared to the SI condition in the QU phase (t (95)=-.1.68, p=.09). (In this respect, that is a CI/SI difference only in the question phase, the interview recall data in the present study resemble those in another of our studies (Memon et al, in press). There were no significant differences in errors (t (95) = -.67) or confabulated details (t (95) = -.56) in the questioning phase. There were no significant differences in the proportion of correct, incorrect or confabulated details (accuracy) as a function of interview type. The means are presented in Table 2.

Table 2: Mean recall in each interview condition across interview phases. Standard deviations in parentheses.

Discussion

Consistent with previous research, it was found that a cognitive interview conducted subsequent to misleading questions reduces the negative effects of such questions. This may be an interview technique that `inoculates' against the effects of subsequent suggestive questions which are often used by investigative interviewers. This is an especially important issue since many guidelines for investigative interviewers recommend that suggestive questions only be used as a last resort once all other interview phases (e.g. FR and QU) have been passed through (e.g. Memorandum of Good Practice, 1992). In the present study, use of the CI produced more correct responses to the post-interview misleading questions replicating the `Geiselman-effect.' This effect occurred despite the absence of a significant differences overall in the interview recall performance of CI and SI groups. These two findings: (1) the `Geiselman-effect' and (2) the lack of a significant CI-SI difference will be discussed in turn.

The CI aims to increase accessibility to stored information and facilitates retrieval by providing the witness with multiple routes to retrieval and through maximising overlap between encoding and retrieval environments (Tulving and Thomson, 1973; Tulving, 1974). The CI also encourages the witness to engage in multiple retrieval attempts, a condition which has been found to facilitate recall (see Brainerd, Reyna, Howe, & Kingma, 1990; Payne, 1987 for reviews). Assuming that the CI results in a more extensive search and effective retrieval strategy, a witness interviewed with the CI may well be better able to discriminate original event details from misinformation through more effective source monitoring (Johnson, Hashtroudi and Lindsay, 1993). While this may explain the "Geiselman-effect" it is as odds with the lack of a significant advantage of the CI found in this study. It is to this finding that the discussion now turns.

The absence of a significant effect of CI regarding the amount of correct information recalled in the interview at first sight seems at odds with the published literature (e.g. Saywitz et al, 1992). However, there are a number of methodological differences between previous studies and the current study such as the retention interval and control group. Each factor may contribute to the discrepancies in the results. Taking the effects of retention interval first. In the present study there was a delay of 12 days between experiencing the event and the interview (as compared to the 2 day delay in the Saywitz study). The CI relies heavily upon effective context reinstatement and it is possible that over time children had difficulty in reinstating the specific features or attributes of the magic show. Riccio, Rabinowitz and Axelrod (1994) provided examples of situations which may correspond with the forgetting of stimulus attributes such as the increase over time in susceptibility to memory distortion (Loftus, 1992) and source monitoring errors (Lindsay, 1994). Riccio et al also make reference to the well established effect that as memory for specific stimulus attributes fade what remains is the `gist' or `schema' for the event (Brainerd and Reyna, 1990). With respect to the inconsistency in the effects of the CI, the efficacy of the procedure over varying retention intervals has not as yet been systematically investigated. Thus it could be that the CI is only effective at delay intervals which permit good context reinstatement.

An alternative explanation for the failure to obtain significant effects of the CI rests with the use of a control procedure (the SI) which incorporated some components of the enhanced CI (as described by Memon and Koehnken, 1992; Fisher and Geiselman, 1992). Indeed, a meta-analysis has shown that when the SI as opposed to an untrained group of interviewers is used as the `control condition' the effect size of the CI is significantly reduced (Koehnken, Milne, Memon and Bull, 1994). The present study found no significant advantage of the CI. It may be the case that the efficacy of the CI and SI vary with different groups of children. In the current study, the SI was developed along the lines of a `child-centred' interview with a great deal of emphasis upon effective communication (as set out in the Memorandum of Good Practice, 1992). A future study comparing the version of the SI used in this study with various other control procedures (see McCauley and Fisher, 1995) and using different age groups would shed further light on the efficacy of the SI used here.

The use of the CI with children should not be ruled out on the basis of the present study. There was a trend in the means favouring the CI over the SI in the questioning phase. Interestingly, this is the phase of CI where several specific mnemonic aids are applied and is the phase that is most different from the SI. The tendency for the CI to produce more details in the questioning phase has also been noted in earlier research (Memon et al, in press; Milne et al, 1995). By isolating the techniques used in the questioning phase of the CI, it may be possible to identify the most effective components of the CI.

In this study the classification of the children into `intruders' and `non-intruders' revealed that the former make more error responses and fewer correct responses on the pre-interview questions and on the post-interview questions, plus fewer `don't know' responses on the post-interview questions. Intruders were also less accurate in the interview due to their having more confabulations and marginally more errors. These findings suggest not only that some children are more vulnerable than are others to misleading questions, but that this vulnerability affects their interview performance. The extent to which children go along with misleading questions presented prior to, or after, a well conducted interview, may therefore prove to be a useful guide to the validity of their interview recall. This is an important finding. Vulnerability to suggestion is something that has been neglected in research on child witnesses (see Gudjonsson and Clare, in press; Gudjonsson, Clare, Rutter and Pearce, 1993 for relevant comments on adult witnesses).

The manipulation of witness behaviour through use of interviewer instructions may go some way in reducing suggestibility effects. However, the situation is far more complex than it appears. For instance, the children who fell into the intruder category in the present study made significantly fewer `don't know' responses despite explicit instructions to say `I don't know' (cf. Saywitz et al, 1991; Saywitz and Snyder, 1993). The interpretation of a `don't know' response is open to question but one plausible hypothesis is that such responses provide some indication of the extent to which children can resist pressure to produce a response that may be an incorrect one. This is consistent with the increasing amount of evidence for the contribution of social mechanisms to suggestibility effects. For example, recent studies have shown that children can resist suggestibility if they are clear about the purpose and relevance of reporting only what they recall about the original event (Newcombe and Siegal, 1995; Siegal, in press). Other studies have illustrated how social factors such as demand characteristics and memorial factors such as source confusions (an inability to distinguish original and post-event details) may interact in complex ways (Lindsay, Gonzales and Eso,1995).

Our increased understanding of factors that may contribute to suggestibility effects provides us with some guidance as to how we may minimise inaccuracies in children's reports. The potential for child witnesses to be misled should not however be overlooked. Research which more closely simulates the conditions in real world investigations has indicated that when children are repeatedly provided with misleading suggestions (Ceci, Loftus, Leichtman and Bruck,1994) and when the suggestions come from a credible source such as a parent (Poole and Lindsay, 1995) they more readily accept misleading suggestions. Further research on the contexts in which suggestibility effects occur may provide us with more insights into the conditions under which children's performance is most likely to be impaired by suggestive questioning and which interviewing procedures may serve to reduce such effects.

References

Ackil, J.K. and Zaragoza, M. (1995). Developmental differences in eyewitness suggestibility and memory for source. Journal of Experimental Child Psychology, 60, 57-83.

Brainerd, C.J., & Reyna, V.F. (1990). Gist is the grist: Fuzzy-trace theory and the new intuition. Developmental Review, 10, 3-47.

Brainerd, C.J., Reyna, V.F., Howe, M.L., & Kingma, J. (1990). The Development of Forgetting and Reminiscence. Monographs of the Society for Research in Child Development, 55, (3-4 Serial No 222).

Bull, R. (1995a). Innovative techniques for the questioning of child witnesses especially those who are young and those with learning disability. In M. Zaragoza, J.R. Graham, G.C.N. Hall, R. Hirschman & Y.S. Ben-Porath (Eds.), Memory and testimony in the child witness Thousand Oaks, CA: Sage.

Bull, R. (1995b) Interviewing children in legal contexts. In R.Bull and D. Carson. Handbook of psychology in legal contexts., Chichester, Wiley.

Ceci, S.J. & Bruck, M. (1993a). Suggestibility of the child witness: A historical review and synthesis. Psychological Bulletin, 113, 403-439.

Ceci, S.J. & Bruck, M. (1993b) Child witnesses: translating research into policy. Social Policy Report. Society for Research in Child Development. Volume VII, (3).

Ceci, S.J. & Bruck, M. (1995). Jeopardy in the Courtroom, American Psychological Association: New York.

Ceci, S.J., Loftus, E.F., Leichtman, M. & Bruck, M. (1994). The possible role of source misattribution in the creation of false beliefs among preschoolers. International Journal of Clinical and Experimental Hypnosis, 42, 304-320.

Cronin, O., Memon, A., Eaves, R., Kupper, B. and Bull, R. (1992, May). The Cognitive Interview with child witnesses: a child centred approach? Paper presented at NATO Advanced Study Institute: The child witness in context. Italy.

Danielsdottir, G., Sigurgeirsdottir, S., Einarsdottir, H. R. & Haraldsson, E. (1993). Interrogative suggestibility in children and its relationship with memory and vocabulary. Personality & Individual Differences, 14, 499-502.

Davies, G., Wilson, C., Mitchell, R. & Milson, J. (1995). Videotaping children's evidence: an evaluation. London: Home Office Research and Planning Unit.

Eich, J.E. (1980). The cue dependent nature of state dependent retrieval. Memory and Cognition, 8, 157-173.

Fisher, R.P., & Geiselman, R.E. (1992). Memory enhancing techniques for investigative interviewing: The Cognitive Interview. Springfield III.: Charles C. Thomas.

Fisher, R.P., Geiselman, R.E., & Amador, M. (1989). Field test of the cognitive interview: Enhancing the recollection of actual victims and witnesses of crime. Journal of Applied Psychology, 74(5), 722-727.

Geiselman, R.E., Fisher, R.P., MacKinnon, D.P., and Holland, H.L. (1985) Eyewitness memory enhancement in the police interview: cognitive retrieval mnemonics versus hypnosis. Journal of Applied Psychology, 70, 401-412.

Geiselman, R.E., Fisher, R.P., Cohen, G., Holland, H.L., & Surtes, L. (1986). Eyewitness responses to leading and misleading questions under the cognitive interview. Journal of Police Science and Administration, 14, 31-39.

Gudjonsson, G.H. (1987). A parallel form of the Gudjonsson suggestibility scale. British Journal of Clinical Psychology, 26, 215-218.

Gudjonsson, G.H. (1992). The psychology of interrogations, confessions and testimony. Chichester: Wiley.

Gudjonsson, G.H., Clare, I. C.H., Rutter, S.C. and Pearse, J. (1993). Persons at risk during interviews in police custody: The identification of vulnerabilities. Royal Commission on Criminal Justice, Research study no. 12. London: H.M.S.O.

Gudjonsson, G. & Clare, I. C.H. (in press) The relationship between confabulation and intellectual ability, interrogative suggestibility and acquiescence. Personality and Individual Differences.

Johnson, M.K., Hashtroudi, S. & Lindsay, D.S. (1993). Source monitoring. Psychological Bulletin, 114, 3-28.

Koehnken, G. (1995). Interviewing adults. In R.Bull and D. Carson. Handbook of psychology in legal contexts., Chichester, Wiley

Koehnken, G., Finger, M., Nitschke, N., Höfer, E. & Aschermann, E. (1992). Does a cognitive interview interfere with a subsequent Statement Validity Analysis? Paper presented at the American Psychology and Law Society meeting, San Diego.

Koehnken, G. Milne, R. Memon, A. Bull, R. (1994, March). A meta-analysis of the effects of the Cognitive Interview. Paper presented at the Biennial Conference of the American Psychology Law Society, Sante Fe.

Lamb, Sternberg & Esplin, P. (in press). Making children into competent witnesses: Reactions to the Amicus Brief. In Re: Michaels, Psychology, Public Policy and Law.

Leichtman, M. & Ceci, S. (1995). The effects of stereotypes and suggestions on preschoolers' reports. Developmental Psychology, 31 (4), 568-578.

Lindsay, D.S. (1994). Memory source monitoring and eyewitness testimony. In J.D. Read., D.F Ross, & M.P. Toglia (Eds.) Adult Eyewitness Testimony: Current Trends and Developments. New York: Cambridge University Press.

Lindsay, D. S., Gonzales, V. & Eso, K. (1995). Aware and unaware uses of memories of postevent suggestions. In M. Zaragoza, J.R. Graham, G.C.N. Hall, R. Hirschman & Y.S. Ben-Porath (Eds.), Memory and testimony in the child witness (pp. 86-108). Thousand Oaks, CA: Sage.

Loftus, E.F. (1992). When a lie becomes memory's truth: memory distortion after exposure to misinformation. Current Directions in Psychological Science. 1, 121-123.

Mantwill, M. Koehnken, G., & Aschermann, E. (1995). Effects of the cognitive interview on the recall of familiar and unfamiliar events. Journal of Applied Psychology, 80, 68-78.

Marche, T.A. & Howe, M. (1995). Preschoolers report misinformation despite accurate memory. Developmental Psychology, 31, (4), 554-567.

McCauley, M.R. & Fisher, R.P. (1995) Facilitating children's eyewitness recall with the revised cognitive interview. Journal of Applied Psychology 80, (4), 000-000.

Memon, A. (in press). The cognitive interview. In O. Hargie. A Handbook of Communication Skills Routledge, London

Memon, A. , & Bull, R. (1991). The cognitive interview: Its origins, empirical support, evaluation and practical implications. Journal of Community and Applied Social Psychology, 1, 291-307.

Memon, A., & Koehnken, G. (1992). Helping witnesses to remember more: The cognitive interview. Expert Evidence: The International Digest of Human Behaviour, Science & Law, 1(2), 39-48.

Memon, A, Cronin, Ó, Eaves, R. & Bull, R. (1993) The Cognitive Interview and Child Witnesses. In G.M. Stephenson & N.K. Clark (Eds.) Children, Evidence and Procedure. Issues in Criminological & Legal Psychology. No. 20. British Psychological Society, Leicester, UK.

Memon, A. Wark, L., Bull, R. and Koehnken, G. (in press). Isolating the effects of the cognitive interview techniques. British Journal of Psychology

Memorandum of Good Practice (1992). Department of Home and Health. London: HMSO.

Milne, R., Bull, R., Koehnken, G. & Memon, A. (1995) The cognitive interview and suggestibility. In G.M. Stephenson & N.K. Clark (Eds.), Criminal Behaviour: Perceptions, Attributions and Rationality. Division of Criminological & Legal Psychology Occasional Papers, No. 22. British Psychological Society, Leicester, UK.

Newcombe, P.A. & Siegal, M. (1995). Where to look first for suggestibility in young children. Paper presented at the bineenial meeting of the Society for Research in Child Development, Indianopolis, March.

Payne, D.G. (1987). Hypermnesia and reminiscence in recall: A historical and empirical review. Psychological Bulletin, 101, 5-27.

Poole, D.A., & Lindsay, S. (1995). Interviewing Preshoolers: Effects of nonsuggestive techniques, parental coaching and leading questions on reports of nonexperienced events, Journal of Experimental Child Psychology, 60, 129-154.

Riccio, D.C. Rabinowitz, V.C. & Axelrod, S. (1994). Memory: when less is more. American Psychologist, 49, (11), 917-926.

Saywitz, K. & Snyder, L. (1993). Improving children's testimony with preparation. In G. Goodman & B. Bottoms (Eds.), Child victims, child witnesses: Understanding and Improving testimony. New York: The Guilford Press.

Saywitz, K.J., Geiselman, R.E. and Bornstein, G.K. (1992). Effects of cognitive interviewing and practice on children's recall performance. Journal of Applied Psychology, 77, 744-756

Saywitz, K.J., Moan, S.F., & Lamplear, V.S. (1991). The effect of preparation on children's resistance to misleading questions. Paper presented to American Psychological Association on Child Development Meeting, California

Siegal, M. (in press). Conversation and Cognition. In R. Gelman, and T, Au (eds.), Handbook of perception and Cognition: Volume 13. Perceptual and cognitive development. San Diego, CA: Academic Press.

Tulving, E. (1974). Cue-dependent forgetting. American Scientist, 62, 74-82.

Tulving, E., & Thomson, D.M. (1973). Encoding specificity and retrieval processes in episodic memory. Psychological Review, 80, 353-370.

Wark, L., Memon, A., Koehnken, G. & Bull, R. (1995) Uses and abuses of schematic knowledge. University of Southampton, Department of Psychology, Working Paper, RWPS/95/1.

Warren, A. & Lane, P. (1995). Effects of timing and type of questioning on eyewitness accuracy and suggestibility. In M. Zaragoza, J.R. Graham, G.C.N. Hall, R. Hirschman & Y.S. Ben-Porath (Eds.), Memory and testimony in the child witness Thousand Oaks, CA: Sage.

Warren A., Hulse-Trotter, K., & Tubbs, E.C. (1991). Inducing resistance to suggestibility in children. Law and Human Behaviour 15, 273-285.

Yuille, J., Hunter, R., Joffe, R. & Zaparnuik. J. (1993). Interviewing children in sexual abuse cases. In G. Goodman & B. Bottoms (Eds.), Child victims, child witnesses: Understanding and Improving testimony. New York: The Guilford Press.

Zaragoza, M., Graham, J.R., Hall, G.C.N., Hirschman, R. & Ben-Porath, R.S. (1995). (Eds.), Memory and testimony in the child witness Thousand Oaks, CA: Sage.

Appendix I: Questions (rotated pre/post-interview)

What was the last trick the magician did?

What type of hat did the magician wear?

Who was the girl that danced with the magician (name)?

After he magicked the white rabbit the magician did the rope trick.

What colour was the rope?

How many flowers did the magician have?

What was the first trick the magician did (name)?

What colour was the magician's guitar?

What was the name of the girl who helped the magician?

What was the magician's name?

What colour was the dove?

What song did the magician sing?

Was there a white rabbit in the magic show?

How many balloons were there?

What colour was the carpet?

What was the pig `s name?

Was the magician wearing a cloak?

Did the magician have a string of hankies?

Acknowledgements

This research was supported by a grant from the British Economic and Social Research Council (ESRC) R000234290. A version of this paper was presented at the 1st Biennial meeting of the society for applied research in memory & cognition, University of British Columbia, Canada, July 1995. We are grateful to the staff and pupils of Kanes Hill School, Hampshire. We thank Rebecca Milne for her contribution to the research, Tony Roberts for statistical advice and Annereike Oosterwegel and anonymous reviewers for comments on an earlier draft of this paper.

Footnotes

[1]Previous research has shown that specific instructions in the use of the `don't know' option can reduce errors in response to misleading questions although they also reduce correct responses (Saywitz, Moan and Lamplear, 1991' Saywitz and Snyder, 1993).

[2]It is impossible to separate a genuine intrusion from a spontaneous confabulation particularly with a familiar event such as a magic show (Wark et al, 1995). Is it possible that an unknown number of intrusions may have occurred even in the absence of misleading questions. As pointed out by an anonymous reviewer, the only way of checking this is by inclusion of a non-misled control group.

[3]The differences in sample sizes in the intruder (n=39) and non-intruder condition (n=53) is not likely to influence the reliability of the ANOVA. A binomial test showed no significant differences as a function of sample size ( p=.24).