Section Three

The Evidential Child Interviews


(3) The Evidential Child Interviews

(3.1) An effective "truth" test
(3.2) Open and closed questions
(3.3) Repeating a question
(3.4) Length of interviews
(3.5) Number of interviews
(3.6) Use of imagination to "aid recall"
(3.7) Little challenging of the children’s statements (source monitoring)
(3.8) Leading questions
(3.9) The use of body-part charts
(3.10) Anatomically correct dolls
(3.11) Accepting denial
(3.12) Verbal and non-verbal reinforcers
(3.13) Priming of Interviewers
(3.14) Conclusions

Next Section
Previous section
Document Contents | Intro | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | References | Appendices |

(3) The Evidential Child Interviews

"I could help you [to remember]..."

Sue Sidey, evidential child interviewer, to a child, Ellis case.

Section one of this critique has already exposed indicators of bias within the child interviewing teams in the Ellis case. Changes to the protocols regarding training of interviewers since the case will be examined later in this report. However a recurring observation from academic commentators of the case is that those involved in the interview processes of this case (and others) where inadequately qualified to fully understand the nature of the work they were undertaking. Auckland University professor of psychology, Michael Corballis:

"I think too much credence was given to clinicians ignorant of the nature of memory and of the persuasive influence of certain interviewing techniques. Perhaps there also needs to be a better appreciation of which research is bad and which is good.

At the time such credulity extended beyond the boundaries of the investigation and trial process itself, even into academia. Retired reader in psychology, Jim Pollard comments on the prevailing attitude at the time within the Canterbury University psychology department:

".... the Psychology Department was in the throes of extreme political correctness. And that included believing what rather ignorant social workers and clinical psychologists claimed about children's testimony."

Corballis again;

"I think the widening gap between clinical and academic psychology is partly responsible for the failure of psychologists to see and arrest what was happening. If clinical psychologists and counsellors thought about the nature of memory at all, they probably held to a 1970s model in which memory for attended events is permanently stored, and failures of memory are regarded as failures of retrieval.  Hence it may have seemed reasonable to use whatever means were available to help people ‘recover’ (retrieve) memories for early events.  Since the 1970s, however, psychological science has shown that memories are not permanently stored; they are fallible and subject to distortion.  It became clear to experimentalists that it is easy to induce false memories of events that never happened."

It is generally accepted that children's memory processes differ from those of adults, and that they are more prone to suggestion through a variety of factors. I will discuss the evidence for this more fully later in the scientific analysis section. However, it is important to realise that once memories of events (that never occurred) become firmly established in the minds of children (and some adults), they can become so strongly held that, with even the very best interviewing techniques, it may not be possible to discover from the interviews alone that the testimony is not reliable. For example, Ceci et al in 1997 found that once false suggestions had been implanted in pre-schoolers,

"These children's false [videotaped] statements were quite convincing to professionals, who were unable to distinguish between true and false accounts…some children appeared to internalise the false suggestions and resisted debriefing."

The widespread use of faulty interview techniques that inadvertently reinforce memories of non-existent events could nevertheless by itself destroy or distort existing accurate memories. This would then discredit the interviewers, and make any resulting convictions unsafe.

It is worth bearing in mind that the first official evidential interviews occurred months after the alleged events. The convictions were based on statements obtained from between six and over 50 months after the alleged events (see appendix 2). As discussed elsewhere, the intervening months were filled with many sources of contamination (i.e. possible misinformation and fear from other people including parents, therapists and police). These informal interviews or therapy sessions were not systematically recorded, and the court heard only conjecture about what had taken place. Some parents admitted asking leading questions, but appeared unable to recall the exact details. This information may well have been more significant than all the videotaped interviews that came much later. In addition, little is publicly known about the first interviews (December 1991) conducted by Sue Sidey, which failed to find "any statements consistent with abuse." (p38 Davies)

The following interview techniques and practices are the ones that the scientific literature has identified as being important in influencing the reliability of children’s testimony. This section also provides insight into the flaws that occurred in the interview processes, citing specific interviews from the investigation itself.

Back to section contents

(3.1)     An effective "truth" test

At the beginning of each evidential interview, and again before cross-examination at the trial, a procedure called a truth test was used with each child. This was like a junior version of swearing on the bible, and was also designed to ensure that the child was able to discriminate between actual events and imagined ones. Children can find it difficult to distinguish reality from fantasy. This procedure did not always work in this case.
(Davies p28, 31,etc). On page 35 Davies states,

"... the necessity of children to tell only what they remembered and to make explicit the ‘don’t know’ option could have been more heavily emphasised from the start."

In relation to child Z/Kari Lacebark (who at the age of five was first interviewed six-months after the time of any of the alleged incidents), Davies (p28) notes that in the first interview, Kari "seems to have a rather uncertain grasp" on the idea of telling the truth

More recent research, I understand, suggests the use of practical demonstrations that can be shown to generally increase accuracy. I cannot say if the test used here was adequate, but quite possibly it was not effective enough. What is evident is that some of these children had an insufficient grasp of this important distinction, and this raises serious issues when regarding the credibility of their evidence. It may have helped if the truth testing procedure had been repeated at intervals (especially when children seemed to be losing the ability to distinguish between fantasy and reality).

Back to section contents

(3.2)     Open and closed questions

Many researchers strongly advise that open questions be used as the primary form of question in child abuse interviews. Free recall, where the interviewee is simply asked to tell what happened in his/her own words, is likely to produce the most accurate information. Open questions are those which begin with who, when, where, what and how. They do not suggest an answer, or imply that a given response may be right or wrong.

Closed (yes or no answer questions) are more unreliable according to research. For example, Poole and Lindsay (1995) confirmed that 4-7 year olds responses to open questions were more accurate than to closed questions. A high proportion of the children were willing to answer questions about events they had not experienced. In an interview situation, the interviewer often has no way of knowing what really happened other than a child’s statement. In this situation, the use of closed questions would seem to be pointless.

There is much evidence of closed questions being used in the Ellis case interviews and insufficient caution about the associated problem from Sas or Davies.

Similar problems can be noted for multiple-choice questions. The true answer may not be contained in the choices given.

Back to section contents

(3.3)     Repeating a question

This can give a child the impression that the answer just given is not what the interviewer wants to hear. This is a form of feedback, indicating a withdrawal of reinforcing smiles, etc until the "right" answer is given. Children are then likely to change their answer in order to please. Poole and White (1991) have shown that repeated questions are suggestive, and can lead to unreliability. Our own Law Commission paper (quoted by Eichelbaum p 40) stated.

"...young children [younger than 10] wish to comply with the suggestions of an adult authority, or because they interpret an adult’s repeated questioning as an indication that their first response was judged ‘wrong’."

The U.K. memorandum (Eichelbaum report page 43) is saying the same thing.

This information may not have been made particularly forcefully to the jury as with reference to Zelas at the trial the second Appeal Court noted [50 (v)]

"Dr Zelas stated that repeating questions could suggest to the child that the original answer was wrong. However, repeating questions may also be beneficial, since it may validate the response to the question, and may help elicit information from the child, particularly if the question is asked in a different way."

On page 27 of her report Sas states

"There were a few occasions when an interviewer had difficulty accepting the child’s information [denial], and repeated the question, ‘did anything happen?’ several times."

She goes on to say that in these cases the worst that could have happened was that it "may not have happened exactly as described [when finally an allegation was elicited]".

I cannot agree with Sas’indifference to the problem. This faulty technique was employed all too often, and its use reflects a bias on the part of the interviewer. On this issue, Ceci (McLoughlin 1996) had this to say:

"There wasn’t any effort to falsify the [interviewers’] hunches [that Ellis was an abuser]. Often there was a repetition of questions, it was almost as though the interviewers were surprised that the child said `that’s all’ Mr Ellis did to them and therefore they would repeat the same question over and over again."

Back to section contents

(3.4)     Length of interviews

In her Best Practice Model for Investigative Interviews Sas states (p29)

"The length of interviews should depend on the child’s age and tolerance, usually not more than an hour,..."

Although Sas subsequently omits to detail the interview durations in the Ellis case, Davies states (p36)

"...the interviews do go on too long in my opinion ..." In fact, more than 50% of the interviews detailed by Davies lasted longer than one hour, one even lasted 105 minutes (child Z/Kari Lacebark’s first interview) .

It was also obvious that in many interviews, children expressed a desire to terminate the interview. Such a desire can further motivate a child toward pleasing the interviewer, in order to escape. In Davies’ report, child Z/Kari Lacebark was clearly motivated to escape from the interview; she complained early on of wanting to be sick, and (p29) "Z [Kari] shows increasing signs of wanting to terminate the interview." Sas (p28) states that,

"... some children were tired and did not wish to continue, others resistant, others distracted, and many upset. Despite this, the interviewers attempted to persevere."

Apparently this was acceptable to Sas because on page 29 she states, "there were no blatant negative effects." If this is Sas’reasoning in this instance, then it is flawed. It is not justified to conclude that the length of interviews may not have a significant effect on the quality of the testimony merely because no ‘blatant’ effects were observed. Many interviews were clearly too long, and the allegations that arose late in them were increasingly unreliable, especially in instances where the children asked for the interview to finish.. In any case, allegations need to be freely made, without hint of duress. Allegations that arose under these conditions are not reliable.

Back to section contents

(3.5)     Number of interviews

Few of the allegations arose in the first interviews with each child. Three of the children involved in the trial had more than three interviews. Not all of their interviews were shown to the jury. There appears to be a steady increase, over time, of the more bizarre and unlikely allegations made by the children. Ceci et al (1997) found that in children aged three to six;

"The number of interviews and the length of the interval over which they were presented resulted in the greatest level of suggestibility. While some types of events (negative, genital, salient) were more difficult to implant in children's statements, some children appeared to internalise the false suggestions and resisted debriefing. These children's false statements were quite convincing to professionals, who were unable to distinguish between true and false accounts."

Sas (p57) observes this effect:

"The repeated interviews were an issue only in the case of child X [/Bart Dogwood], as it was evident in his fifth interview that the process was contributing to increasingly bizarre and disjointed accounts in the child." This is clear on pages 53 (the "circle incident" in the fourth interview) and 54 (fifth interview) with its "widening of allegations with respect to the number of teachers involved" and "some inconsistencies with respect to the other female teachers that were present."

However Sas’contentions seem to be at odds with what I observe; that in most cases the least likely allegations arose after the first two interviews:

o        Child R/Eli Laurel (p92 Eichelbaum), in the third interview bizarre allegations emerged of four men (with strange haircuts) and women being involved in poking sticks up kids’ bottoms.

o        Z/Kari Lacebark’s less credible story of abuse at Ellis’s house emerged in the fourth interview. Davies writes (p31),

"There is nothing in this interview which convinces me that Z [Kari] visited Peter’s house or was assaulted by a man called Joseph." Two further interviews, in Davies’ view, "subtract from, rather than add to Z [Kari]’s credibility as a witness."

o        S/Tess Hickory in her third interview states that Ellis drove a green car. We know that he didn’t drive, and she had said that they walked in the first interview. She also stated that Ellis’s mother kicked her. However, the allegation about being asked to drink wees, also bizarre, resulted from the first interview.

o        Q/Lara Palm spoke of going up escalators with Ellis to a room where lots of people were working at desks and did secret touching (Eichelbaum p91). Davies (p17) reports she talked of "little wee kids working there for the City Council." My theory may not apply in this case as this story was related during the second interview.

o        X/Bart Dogwood in his third interview described trapdoors, mazes, Ellis’s mother dispensing "yucky pills", the presence of Spike, Boulderhead and nineteen of their friends etc. Bart’s fourth and fifth interviews produced even more bizarre material. In Davies opinion (p23)

"These later interviews show a gradual spiral into more elaborate allegations embracing a wider and wider circle of helpers and teachers at the creche."

o        Child O/Molly Sumach was only interviewed once and so no conclusion in this regard can be made.

If correct, then this effect applies to five out of the six children discussed in the Eichelbaum report. As Eichelbaum does not examine N/Zelda Cypress, who retracted, the report provides no data for the seventh child whose testimony resulted in conviction.

In a study by Rawls 1996, (reported by McLauchlan1996), an attempt was made to reproduce the interview techniques used in the Civic childcare centre case. She concluded, "Their errors seemed to evolve over time with repeated interviews." This supports my hypothesis.

Davies (p36) notes there was no justification for repeat interviewing of children after a delay of 6 months or more (as occurred) and he notes that most guidelines recommend a maximum of one or two interviews. "Our concern about the research is the number of interviews that were carried out," wrote police legal adviser Mark Copeland and child abuse interview training coordinator Wendy Burgering in a letter to The Dominion on May 31 1996 after the paper published an article on Rawls’ research.

In conclusion, there are allegations that arose after too many interviews that were too widely spaced. These allegations and resultant convictions are therefore suspect.

Back to section contents

(3.6)     Use of imagination to "aid recall"

Asking children to imagine "what it feels like" before a "disclosure" may be a powerful means of creating a false memory. Such an approach recurred frequently in the taped interviews. The methodology of constantly using children’s imaginations in order to "aid recall" has no basis in science. The opposite is the case. One of the more powerful mechanisms of memory construction is termed imagination inflation. In this mechanism all that is required to significantly distort memory is to ask the subject to imagine an event. The study by Garry, Manning and Loftus (1996) provides evidence that the act of asking a subject to imagine an event (which in reality had not occurred) increases the probability that the subject will believe that it did take place.

Where the interviewers have asked the children to imagine something has happened, and/or describe how it feels is clearly a case of bad practice. An evidential interview should be designed to discover what a child has actually experienced, or at least believes was experienced. This imagination technique seems more likely to create memories, rather than record what is already there. Its frequent use casts serious doubt on some allegations and resultant convictions.

Back to section contents

(3.7)     Little challenging of the children’s statements (source monitoring)

It is crucial to probe the aetiology of the children’s allegations when they are made. That is the best time to take the opportunity to source test, by asking questions such as "Did that really happen? Did you talk to anyone about this?" If yes, and if the person happened to be the mother "What did mummy say?"

On p28 of Davies’ report, it is revealed that "... mummy has told her about ‘baby stuff" and the child [Z/Kari Lacebark] added, ‘... me knew the colour and it was plain white". This is significant, because the prosecution argued that these children had advanced knowledge of sex that could only have resulted from being abused (the jury apparently believed that in this instance the child was talking about semen). The interviewer did not explore the source of this serious allegation. There is acknowledgement in the report that the mother was repeatedly questioning her child (p29). Paragraph 48 (p 29 Davies report) clearly indicates that the child was rehearsed in what to say by the mother. The child’s statement, "I know lots of things about Peter", according to Davies "... may be significant in that it may not imply first hand knowledge." This is emphasised again on page 30: "The use of adult terms like ‘penis’ and ‘vagina’ may reflect conversations with her carers…[and] adult input." On page 31 Kari says she knows "... because mummy telled me." The evidence against Kari’s credibility carries on over the next pages, and is convincing.

So, the allegations from this child are unlikely to be spontaneous ones that would indicate a knowledge of sex that could only have come from an experience of sexual abuse. It is more likely that most of it came from information put to her by her parents, therapists, other children or police. Another possibility is that the child was referring to books for children describing bad touching, such as Katie’s Yucky Problem by Lynda Morgan (1986) and A Very Touching Book by Jan Hindman (1983). Hood (p206) describes how the complainant parents read these books to their children from an early age. It is not clear from the report whether Davies was aware of this.

It is surprising how the jury could have allowed Z/Kari Lacebark to be such an effective witness considering the unreliability of her testimony, and the fact that there was a seven month delay between this child’s attendance at the Civic childcare centre, whilst Ellis worked there (until 1992), and the child’s later allegations. She was the only child to obtain four guilty verdicts and no not-guilty returns.

An interesting example of this lack of source monitoring is found in Hood (p436 and p437). Crawford was interviewing a child. The child keeps telling her directly that he is making stories up, most of it is clearly fantasy, and yet Zelas, Detective Eade, and crown prosecutor Chris Lange apparently gave the allegations total credence. Significantly, Justice Williamson was not presiding when this joke (the child’s description) arrived in the depositions court. Had he heard this, he might have been less credulous. The point is that as this allegation was presented seriously in court (before the trial) and it could only have proceeded that far if the officials believed that it had some reasonable chance of being credible.

American research psychologist Stephen Ceci as reported by McLoughlin (1996) said,

"This case entailed an array of factors that give me cause for concern. Children frequently reported highly implausible events that were never checked, for example the presence of the defendant’s mother during baths, repeated sodomy occurring only minutes apart with other children who were said to be present, and they were never reined into reality. That some of their claims were plausible is no assurance that they did not emanate from the same sources as the implausible claims."

Sidey and her team were not maximising accuracy with their techniques. They didn’t query allegations to test their veracity, instead they repeatedly pressed for details of abuse.

Back to section contents

(3.8)     Leading questions.

The use of leading questions seemed to be scattered throughout the interviews. This is not surprising, as Zelas in court claimed that reliable evidence could still be obtained through leading questions (Hood p504, paragraph 7). On page 48, Sas states that child S/Tess Hickory (two guilty verdicts) was interviewed using "more leading questions" during the second interview.

On page 18 of the Sas report, she states that parents should be told to

"... try one leading question" if there is no initial "disclosure" prior to any expert interview.

Then again, on page 24,

"... professionals have to be realistic, and everyone recognises that some leading questions will need to be asked in order to assist children retrieving their memories."

Despite all this, on page 31 Sas states that there "... were occasions when the number of direct or leading questions employed was too high." Yet in her conclusions on page 58 she states,

"This [leading questions] likely served to trigger the children’s memories, or at times might have decreased their apprehension about talking about things."

In other words, she regards this is acceptable, because she goes on to conclude that the children’s evidence was reliable. It is difficult to determine exactly where Sas draws the line on this issue. Her apparent flexibility in regard to the use of leading questions allows her to draw any conclusion, provided it is in agreement with her bias.

Davies notes in his concluding remarks (page36)

"Children are sometimes asked leading questions...most codes of practice strongly discourage [this]".

He concludes that other factors such as the excessive numbers of interviews, the delays between interviews and the over-use of direct, closed and multiple choice questions is "of more concern".

Despite these major concerns, Davies concludes "the general standard of interviewing was good" (page 37)

Lindsay and Poole (2001) show that parents need only to use open- ended prompts in order to construct false memories in their children. Combined with the "frequent use of direct, closed and multiple choice questions" (Davies p36) the value of these interviews is very low indeed.

Another mechanism for constructing memories is the use of narrow questions. Davies notices (p36) that,

"...of more concern [than leading questions] is the frequent use of direct, closed or multiple choice questions in some interviews, where the age of the child and age of memory would be unlikely to provide an accurate response."

The use of leading questions is a form of extremely bad practice, and Eichelbaum should have paid more attention to the inconsistencies I have quoted from Sas and Davies. When these considerations are weighed his conclusion that the evidence was reliable seems rash.

Back to section contents

(3.9)     The use of body-part charts

Rawls study (1996a) showed that using charts of body parts can increase errors in recall. These kinds of charts were used in some of the civic interviews. This is yet another questionable practice that was accepted by Zelas, the jury, Eichelbaum and Sas. Furthermore, in the interviews, references to body parts formed multiple choice questions used by interviewers, compounding the problem.

Back to section contents

(3.10)   Anatomically correct dolls

Clearly, these should not have been used. Bruck et al (1995) showed that they increase the incidence of false statements about being touched genitally. In her report Sas appears ignorant of relevant research findings. For example, on p49 she even states:

"... in this interview, the use of anatomically correct dolls would have been an asset."

When I questioned (Crown expert witness and interviewer supervisor) Karen Zelas about this issue in January 1994, she provided me with copies of research by Canadian Gail Goodman that Zelas claimed provided a basis for the use of anatomically correct dolls. Zelas said that the dolls should only be used after a disclosure has been made. I alerted Zelas to a later study by Goodman that seemed to cast doubt on any use of these dolls. Zelas said that she had not seen this later study. In the later 1987 study, when asked objective questions, the 3 – 5 year olds scored worst in terms of accuracy when questioned using anatomically correct dolls, and best when there were no dolls at all. Goodman’s conclusion that the dolls of themselves do not lead to false reports seemed therefore to be at odds with some of her data. An example of the problems generated by the use of such dolls is well illustrated in the Civic childcare centre case. One of the charges made in court (during the depositions trial) alleged that Ellis had sex with a lesbian co-worker at the childcare centre. The child in the interview concerned made no allegation of this incident. The charge arose solely from the interviewer’s interpretation of a child’s play with dolls. (Hood p427)

The second appeal judges noted [50 (iv)]

"Dr Zelas gave evidence at trial as to the use of anatomically correct dolls, stating that the general belief is that anatomically correct dolls are suggestive, but that this is uninformed, as research evidence indicates that the dolls are not suggestive to young children. However, she accepted that they should be used sparingly to avoid any possible criticism of the interview techniques. While her recommendation that the dolls should be used sparingly is consistent with the evidence of the present experts, her opinion that they are not suggestive to young children is inconsistent with the current weight of opinion."

The use of these dolls, then, should have alerted Eichelbaum to the existence of a bad practice. In section (4) on changes to best practice, this critique makes clear that the use of anatomically correct dolls is no longer acceptable.

Back to section contents

(3.11)   Accepting denial

If a child says that something did not happen, then this statement needs to be accepted, because continual probing of the same material could produce unreliable testimony. The child may decide to provide what he/she thinks the interviewer requires. As the interview requests, the child’s story may become gradually more specific. Likewise, if a child retracts an allegation, then that must be the end of the matter, no matter what the belief of the interviewer, who has to remain impartial to the very end. If abuse did occur, then there is a good chance that another spontaneous disclosure will occur (note that in this statement the use of "disclosure" is in accordance with its dictionary meaning).

Back to section contents

(3.12)   Verbal and non-verbal reinforcers

One important mechanism by which inaccurate memories are constructed as in the response shaping paradigm described, is through the use of subtle (and possibly unconscious) reinforcers such as smiles, nodding, appearing to be about to terminate an interview, etc. This effect is in the category of established fact in psychology, backed by literally hundreds and probably thousands of studies, going right back to the work of the famous behaviourist, B.F. Skinner. As noted elsewhere in this critique, much memory construction could have occurred before the formal interviews, resulting from interactions with parents, therapists, police and probably the other children. In the Ellis case interviewers knew details of the allegations before they started each interview (Sas p27), There is a high probability that construction of inaccurate memories through reinforcers occurred and operated on the most susceptible young children.

The making of allegations may have been positively reinforced to the extent that children experienced a feeling of gratification for making them. Child R/Eli Laurel.

Sidey: "...what do you feel like when you think about that stuff?"

Eli:                  "Um, good."

Sidey:              "Good? What makes it feel good?"

Eli:                  "Because I told the teacher."

Hood (pp 436-437) gives an excellent and detailed example of the use and effect of reinforcement. Interviewer Crawford slowly helped the child to create an unlikely story. In his first interview, the boy stated that he couldn’t remember anything about the childcare centre. After having being told that Ellis killed "all the boys" with axes Crawford asked, "how did that make you feel?" The boy appeared in due course to become alarmed that Crawford believed everything he said. The boy then repeatedly stated,

"I’m just really joking ... I only joke, joking saying that, only joking he doesn’t, do you hear?... I was just really tricking " etc, etc.

It is probable that this interview actually provides a good record of the ontogenesis of a memory construction process. Yet Crawford in consultation with Zelas "... accepted that Ryan had made a genuine disclosure about sexual abuse" (p437). I think that this example reveals a strong bias in the interviewing team of a desire to produce allegations by any means.

Had the interview continued it is entirely possible that a point would have arisen where the child would have given up saying, "I was just joking /tricking." After all, Crawford was ignoring these statements, but positively reinforcing the allegations. The allegation even went to court (at depositions) where it apparently created some merriment at the expense of the prosecution.

It is possible (and given the record, highly probable) that in at least some of the other cases involved (perhaps all), parents had already completed this process.

When I interviewed Cathy Crawford, she clearly had no knowledge or understanding of memory construction or verbal reinforcers. Sas reveals such awareness once in her report (p44), when she describes child Q/Lara Palm (two guilty verdicts), who had stated that she could remember having been in an incubator as a baby. Sas indicates an understanding of how this memory was constructed (apparently through being told family history and through viewing a photograph.) Yet it does not seem to occur to her that some of the Civic childcare centre allegations may be unreliable, and could have been constructed through a similar process. By the trial, all the interviews conducted by Crawford had been eliminated from the evidence. This was unfortunate, because these interviews highlighted the problems that were present in the other interviews and they also exhibited evidence of allegations that arose from extraneous sources (i.e. Crawford passing on and creating unfounded rumours).

In the scientific literature, the misinformation effect is the reporting of false details that are believed to be true by a witness after the event. Loftus (1989) gives 21 references to experiments proving this misinformation effect. Younger children are more prone than older children. There are a variety of ways in which misinformation can arise. They include the mechanism already described. There is no reference to this proven phenomenon in the Eichelbaum report. Nor could I find a significant analysis by Sas and Davies to assess the responses of the interviewers immediately after each "disclosure." Saying "good" or "yes" or smiling, etc, are very-powerful reinforcers, and will result in a greater frequency of similar statements regardless of the veracity of these statements. Children are very strongly motivated to please adults in authority (Oshima Takane 1987). The avoidance of these reinforcers at these crucial moments in interviews should be laid down in guidelines as an essential element of good practice. This was not done. The NSW guidelines (Eichelbaum report p39) encourage interviewers to be supportive, "while not rewarding or punishing the child for giving details."

A detailed analysis of the effects of possible reinforcers and punishers of the children’s allegations may reveal significant effects if access to enough transcripts is given. If it can be shown that the interviewers were (consciously or unconsciously) rewarding and punishing the children to produce allegations, then it is fair to conclude that the interviews were biased, and the resulting allegations.

Certain words used by the interviewers may act as discriminating stimuli, signalling positive reinforcement (access to smiles, hugs, toys, etc), negative reinforcement (removal of an aversive stimuli such as terminating the interview itself) or punishment (frowns, increasing the length of a boring interview, etc)

One way to analyse this is to see what kinds of statements from the child precedes certain words or phrases used by an interviewer. It is perhaps most useful to look for a reinforcer (or discriminating stimulus that signals reinforcement). Unfortunately, the most powerful reinforcers (and/or discriminating stimuli for reinforcers) may have been non-verbal (eg smiles), and thus the videos would be needed for a meaningful analysis. An analysis of some of the transcripts published by Colman (2003) – suggests that "mmm, mm", "yeah" and "right" (from one interviewer) may signal punishment, thus suppressing the type of statements from the children that preceded these words.

I conclude from this that the strong bias operating on the part of the interviewers towards finding Ellis guilty was made manifest in the way that the interviewers directed and shaped the children’s responses. It may not have been deliberate, but it should have been deliberately avoided. The later interviews were all unreliable as by that stage the reinforcers had acquired the capability to be very powerful, and when considering this cause, they can produce the greatest distortions to the children’s memories. As I have noted earlier, most best practice guidelines do not recommend more than two interviews.

Back to section contents

(3.13)   Priming of Interviewers

I assume Eichelbaum believed that the interviewers were searching for the truth, and for reasonably spontaneous, unprompted allegations. Any alternative explanation would undermine his conclusion that the interview techniques were "of a good overall quality." It would therefore be safer for the interviewers to know only of the general nature of the allegations. The material gained during the interviews could then be compared, upon completion of all the interviews, with that it was alleged they had told their parents. Sas in her guidelines for best practice states

"The interviewer should not be aware of the specifics of the allegations about which they are to interview the child about..." (p26)

Yet on page 27 of her report, Sas makes a revealing admission (my emphasis):

"In all interviews I reviewed, it was obvious that the interviewers had been given information by parents prior to the interview with the child."

She goes on to say ("my clinical impression") that this practice was perfectly acceptable in the Ellis case.

Information also flowed in the other direction, again contrary to Sas’own guidelines (p21) regarding best practice:

"From the testimony, it was clear to me that the interviewers shared with the parents the outcomes of their children’s interviews each time they came." Sas (p21).

In her conclusion, (p58) she refers to claims by Ellis’s appeal lawyer that interviewers repeated details of interviews to parents after interviews. She admits,

"This was a valid criticism ... The effect may have been to lead parents to encourage their children to disclose, and then the additional interviews followed".

The end result, here alluded to by Sas, is that this process of information exchange, in both directions, between interviewers and complainant families set up a feedback loop whereby "disclosures" could be amplified by parents between formal interviews, and by the interviewers during the interviews.

I conclude from this that the procedures used were neither scientific nor fair. It can be argued that this behaviour is indicative of a situation where the interviewers, including Zelas (and even Sas), were only interested in proving a prior assumption of guilt.

Back to section contents

(3.14)   Conclusions

There is no reference in Davies’ report as to the use of the dolls in his summary of the techniques (pp35 – 37). He does go on to imply, as noted previously, that there were too many interviews. However Davies concluded (p35) "that by today’s standards, the quality of interviewing stands up surprisingly well."

Ceci (McLoughlin 1996) comments:

There wasn’t an effort to rein the children back into reality when they roamed into these fabulous claims. Whether or not the interviewers’ minds were made up prior to the interviews I can’t say, but what I can say [is] there was no serious attempt to test an alternative hypothesis to the [Crown’s] claim that Mr Peter Ellis molested these children."

Ceci’s conclusion is of interest, and succinct

" Some of the things the children said I would be exceedingly sceptical that they ever occurred. I don’t think the bizarre stuff happened. Does that mean nothing happened? Well, I simply don’t know. No one else knows either except God and Mr Peter Ellis."

One point is obvious, that throughout the interviews, and (apparently) regardless of the interview techniques, both credible and highly unlikely allegations arose. One would expect that the use of good practice might at least suppress the highly unlikely allegations.

If I am correct in that these inaccuracies occurred more frequently in the later interviews, then this implies that the interview techniques used were ineffective. Not only that, but the poor interview techniques could themselves be directly responsible for the bizarre allegations and distortions. A behavioural analysis of this would be that; as the interviews progress, the children’s’ responses are steadily shaped through use of verbal and non-verbal reinforcers. These reinforcers become increasingly more effective over time. This is a falsifiable hypothesis, as it predicts that when reinforcement for suspect allegations is reduced, or removed, then the rate of this type of allegation will reduce. This model also predicts the fact that the allegations became more varied, and as a result more bizarre, through response generalisation over time.

Thus it appears that we are left with acknowledgements from the two experts and Eichelbaum that the interviews all contained practices that could give rise to concern, but none seem prepared to conclude that these practices undermine the reliability of the evidence. Even leading questions were regarded as acceptable. I conclude otherwise.

I conclude that the interviewers began with a prior knowledge of the allegations and an assumption of guilt on the part of Mr Ellis. The interviews were also conducted poorly in a manner that encouraged unlikely allegations, and led the children towards sexual allegations. This was achieved by a variety of means that included:

o        closed and leading questioning,

o        the incorrect use of anatomically correct dolls

o        not exploring (and in certain circumstances challenging) allegations in order to identify their source and the possibility of distortion

o        a probable misuse of verbal and non-verbal reinforcers (smiling, saying "yes", etc)

o        continuing for too long even when children indicated that they wished to stop. This creates a negative reinforcement paradigm wherein the child’s wish to escape the disliked interview situation can lead to making allegations in order to escape.

o        conducting too many interviews

o        not accepting denials by the children

There are convincing explanations for the frequent departures from good practice on the part of the interviewers. The interviewers’ behaviour could be predicted from the preceding description (in section (1) on bias) of bias and false assumptions within the child protection movement.

It is difficult to understand how Eichelbaum managed in his conclusions to ignore so much evidence of poor interviewing techniques. This evidence was contained within the reports of Davies and to a lesser extent, Sas. Furthermore, inferred from his own comments, it was also extensively produced at the previous appeals.

I am at a total loss as to how Eichelbaum could have concluded (p4) that the interviewing was "of a good overall quality."

End of Section Three

Next Section
Previous section
Back to section contents
Back to document contents