Interviewers can influence children via suggestive questioning called “modifications.” Consider, for example, the following example:
Int: Can you tell me what grandpa did to Sean?
Child: He sucked her.
Int: He sucked her. Okay. What did he to do Jean?
Child: (No response).
Int: Do you know?
Child: (No response).
Int: So he sucked them. Who did he suck?
Child: Both of them (Walker & Hunt).
Walker and Hunt (1998) analyzed the types of questions used by Child Protective Services personnel when interviewing children in cases of alleged sexual abuse. Walker and Hunt found that modifications occurred in approximately three-quarters of the interviews they reviewed, with each interviewer making approximately 2.5 modifications per interview.
Hunt and Borgida (2001) used experimental interviews to ascertain how young children (ages 3-5) respond to interviewer modifications. In their study, 22.6% of the children incorporated interviewer modifications into their subsequent responses. Hunt and Borgida explained: “… in the present study, each piece of modified information was presented only once, in the context of a single interview question. Given this subtle manipulation, the fact that even a small percentage of the modifications were incorporated into subsequent answers should be seen as consequential” (Hunt & Borgida, 2001, p. 599).
Hunt and Borgida (2001) concluded the discussion of their results indicating:
“This study is consistent with a new trend in research on the testimony of child witnesses, using controlled experiments to investigate specific interviewing techniques found in analyses of actual forensic interviews of children. Such research has demonstrated that commonly used interviewing techniques can have serious, deleterious effects on children’s testimony. Likewise, this study suggests that, despite the possibility that many modifications may reflect accidental mistakes made by interviewers, they can have important effects on investigatory interviews” (Hunt & Borgida, 2001, p. 599-600).
Children between ages 3 and 8 participated in a session that involved four science demonstrations conducted by “Mr. Science” (Poole & Lindsay, 1998). Immediately after the demonstration, children were interviewed non-suggestively about what they had seen. The children responded with a great deal of accurate information, and minimal inaccurate information. Three months later, a storybook was mailed to the parents of these children. The storybook described two science demonstrations the children had seen, and two demonstrations they had not seen. The storybook also described an instance of nonexperienced touching involving Mr. Science having put something “yucky” in the child’s mouth. After the parents read the storybook three times to each child, the children were interviewed again, initially in a nonsuggestive manner, and then in a leading manner. In response to the nonleading portion of this second interview, 114 children reported a total of 58 events they never experienced, including 17 reports of nonexperienced touching. The older children reported as many nonexperienced events as the younger children. In response to the leading, suggestive portion of the interview, 33-42% of the children in each age group falsely reported that Mr. Science (1) put something “yucky” in their mouths, or (2) hurt their tummies.
Because children use their imaginations to embellish their reports of fictional events with persuasive sounding details, their accounts of these “non-memories” sound very compelling. Not surprisingly, then, the relevant research demonstrates that the accuracy with which various professionals identify children who report factually accurate information – compared to those who do not – is at a level less than chance performance (Bruck & Ceci, 1995). Specifically, this particular study reported:
“Experts who conduct research on the credibility of children’s reports, who provide therapy to children suspected of having been abused, and who carry out law enforcement interviews with children, generally failed to detect which of the children’s claims were accurate and which were not, despite being confident in their judgments … Similarly, it may become difficult to separate credibility from accuracy when these children, after repeated interviews, give a formal video-taped interview or testify in court” (Bruck & Ceci, 1995).
Another study showed mental health professionals a video-taped interview of a child who might have been sexually abused (Horner, Guyer, & Kalter, 1992). In response to viewing the videotape, the professionals exhibited a wide range of opinion regarding whether the child had been abused. Some professionals were convinced the child had been abused and were confident in their opinions. Other professionals insisted the child had not been abused, and they were also quite confident in their opinions. Numerous other professionals expressed equivocal opinions in between these two extremes.
When interviewing children in cases of alleged sexual abuse, the most accurate information is obtained in the first interview. The information obtained in subsequent interviews is contaminated by repeated questioning effects (Ceci & Bruck, 1995).
The following organizations and well-respected authorities – footnoted below – have emphasized the absolute necessity of videotaping – or at least audiotaping – investigative interviews in cases of alleged child sexual abuse (McGough, 1991; Ceci, Ross, & Toglia, 1989)., , , In this case, however, there are no verbatim, word-for-word transcripts of these interviews that would have been available if the interview had been videotaped or audiotaped. As a result, we do not know whether the interviewer(s) directed leading and suggestive questions to the child that could have further distorted and confused her memory.
Dr. John Yuille, a well-respected psychologist on the faculty of the University of British Columbia, and his colleagues have also emphasized the necessity of recording investigative interviews in cases of alleged child sexual abuse when they wrote:
“The number of interviews can be minimized by recording the interview. Preferably, the interview will be videotaped but, in the absence of video equipment, an audiotape will suffice. If no electronic equipment is available, a verbatim record must be kept. There are many advantages to recoding an interview, although having such an objective record of their interview performance may be intimidating for some professionals. It has come to our attention that, in the wake of the McMartin case, a number of professionals have decided not to record interviews. Since poor interview quality was a central feature of the McMartin case, many have decided that it is better not to have a record of an interview. It would seem that such a decision is based on the fear of having one’s work publicly scrutinized. This is a most unfortunate development. Our goal should not be to hide poor interviews, for it is important that the adequacy of the methods used to obtain children’s evidence be assessed. A more appropriate goal would be to educate interviewers in order to enable them to conduct interviews that can withstand public scrutiny” (Yuille, Hunter, Joffe, & Zaparniuk, 1993).
Unfortunately, however, it appears that prosecutors are becoming progressively disinclined to videotape interviews of children in cases of alleged sexual abuse. 297 surveys were mailed to prosecuting attorney’s offices nationwide (Quas, DeCicco, Bulkley, & Goodman, 1996). One hundred fifty-three offices (52%), representing 41 states, responded. The survey was designed to answer the following the questions: 1) What innovations are being utilized in cases involving children? 2) Which innovations are no longer employed, and why not? and 3) Do prosecutors view innovations as helpful in reducing children’s stress?” (Quas, DeCicco, Bulkley, & Goodman, 1996, p. 5). Interestingly enough, over half of the respondents indicated that they no longer employed …, videotaped statements at trial as evidence …” (Quas, DeCicco, Bulkley, & Goodman, 1996, p. 5). “In fact, `hurts case’ was the main reason respondents no longer videotaped interviews” (Quas, DeCicco, Bulkley, & Goodman, 1996, p. 5). “For offices in which innovations have been used, percentage of respondents indicating reasons why various innovations are never or rarely useful.” Videotaping interviews: Lack of funds – 28%, Defense challenge or appeal – 18%, Hurts case – 40%, Not needed – 22%, other – 9%, N = 90.
A systematic evaluation of the effects of videotaping investigative interviews, done in California, reported consistently favorable results. “In Denmark and Sweden, the precautionary audiotaping of investigative interviews has been standard practice for more than 20 years” (McGough, 1995, p. 380). “Furthermore, England has now adopted a videotaping statute as well as standard interviewing protocols” (McGough, 1995, p. 380). In the California Project, “A clear picture emerges: Professionals are enthusiastic about videotaping” (McGough, 1995, p. 383). The California Project “found little to support the fear that defense attorneys received `too much ammunition’ to attack the child. The professionals surveyed believed that fact finders were fully capable of understanding minor discrepancies and confusion in a child’s account and that when interviews were conducted competently, there is nothing to hide.” (p. 383). The California project found that videotaping “did not affect the children’s willingness to respond” (McGough, 1995, p. 384).
As would be expected, interviewers typically insist that they carefully avoid directing any statements or questions at children that are leading or suggestive. The relevant research, however, does not support these claims. During their interviews, interviewers typically question interviewees in a manner that biases the information they obtain (Dallas & Baron, 1985).,
Assumptions about a child’s supposed abuse history, for example, increase the frequency of questions directed at that topic — and asking enough questions allows interviewers to think they have found the answers they expected (Arkes, 1981). These processes are known as “one-sided questioning,” or “confirmatory questioning.”
The expectations of interviewers can also lead them to believe that evidence consistent with their initial impressions were exhibited during an interview, when in fact, they were not (Arkes & Harkness, 1980). Conversely, they are also less likely to recall evidence actually present during an interview, but inconsistent with their original impressions. These judgmental errors are known as “confirmatory bias” (Faust, 1989).
Additionally, related research has demonstrated that psychiatric trainees overlook 50% of the important issues transpiring during their interviews (Muslin, Thurnblad, Meschel, 1981). Furthermore, another 50% of what these trainees did report to their supervisors was also distorted. These data suggest that the total accuracy of a non-recorded interview approximates 25% (.5 x .5 = .25). Moreover, related research has demonstrated that experienced therapists also do not accurately recall their own behavior during interviews (Truax, 1966)., When adults are asked to recall conversations they have just heard, their verbatim memory fades within seconds (Ceci & Bruck, 1995). Therefore, without videotapes of an interview, we can never know exactly how the interviewer might have influenced the child who was interviewed.
In particular, interviewers typically cannot recall the specific questions they asked which elicited the responses of the child they report. These are particularly sobering considerations. In particular, for example, Bruck and Ceci have proposed that “… interviewer bias is the central driving force in the creation of suggestive interviews” (Bruck & Ceci, 1997). Bruck and Ceci further indicate: “Interviewer bias characterizes an interviewer who holds a priori beliefs about the occurrence of certain events and, as a result, molds the interview to elicit from the interviewee statements that are consistent with these prior beliefs.”
In a 2001 study, participants read short passages and one day later, they answered questions via telephone about the passages (text facts) and about the experimental session (event facts) (Joslyn, Loftus, McNoughton, & Powers, 2001). Six weeks later, the participants were telephoned again and asked the same questions about text facts and event facts. The participants were 42 college students at the University of Washington (30 females and 12 males) ages 17 to 30 years. At the six week interval, participants claimed to remember what – in fact they did not accurately recall – almost half of the time. In other words, the accuracy of recall did not exceed what would be expected at a chance level of performance.
The experimenters observed:
“People consistently judged their memory to be more accurate than it actually was. Indeed, approximately half the memory failures were [mistakenly] judged to be successes. This suggests that, in many cases, people do not access explicit memories when they claim to have remembered, but rather they may make inferences from an inflated analysis of their own abilities. They merely assume they must have remembered” (Joslyn, Loftus, McNoughton, & Powers, 2001, p. 795).
A 2002 study reported: “Participants in our experiments asked to say ambiguous sentences while attempting to convey a specific intention to their addressee. When they estimated the addressee’s understanding of the intended meaning, they showed a consistent tendency to overestimate their effectiveness” (Keysar & Henly, 2002, p. 207). This study continued to indicate: “Our findings suggest that when speakers monitor their own utterances, they do not act as unbiased observers. Instead, they underestimate the ambiguity of their own utterances and overestimate the extent to which their disambiguating cues make their intention transparent” (Keysar & Henly, 2002, p. 207). Consequently, we should be quite skeptical when interviewers in child abuse matters claim that the children readily understood their questions.
A 1999 study examined how accurately 27 experienced interviewers recalled details from their interviews with children between the ages of three and five (Warren & Woodall, 1999). The ages of these interviewers ranged from 28 to 53, average age of 40.59. Their experience in forensic/child protective work ranged from 4 to 21 years, averaging approximately 10.9 years. More than half of these interviewers (57%) had earned masters degrees, 30% held bachelor’s degrees, and one had a doctoral degree. These interviewers reported a range of 3-400 training hours, or a range of 4-6 training days.
These interviewers conducted videotaped interviews of children who one month earlier had witnessed two events: (1) a magic show and (2) a silly doctor visit. The interviewers were given one of two cue questions for beginning their interviews: (1) “I understand that a magician came to visit your school. Tell me what the magician did?” or (2) “Tell me about the time you went with Tracy to play silly doctor.” Except for these cue questions, the interviewers knew nothing else about these two events.
After their videotaped interviews with the children ended, the researchers audiotaped their interviews with the interviewers. In comparison to the amount of information children related during their videotaped interviews, the interviewers’ hearsay accounts involved information loss. “We asked our interviewers during the audiotaped session what kinds of questions they had asked to elicit information from the children. Most answered that they had asked primarily open-ended questions, whereas few stated that they had asked close-ended or specific questions” (Warren & Woodall, 1999, pp. 365-366). “We found that most (over 80%) of the questions were specific or close-ended (and 16% were leading). Thus, it is clear that interviewers are incorrectly remembering their actual questioning styles” (Warren & Woodall, 1999, p. 366). “We asked interviewers if they recalled anything the children said in the child’s ‘own exact words.’
Very few were able to give even a 2- or 3-word phrase using the child’s exact words.” (Warren & Woodall, 1999, p. 366). “In the course of their jobs, however, they regularly interview many children about similar topics, and sometimes even the same child about the same topic more than once. It is conceivable that details of those interviews could interfere with another and become confused over time.” (Warren & Woodall, 1999, p. 369). “In summary, our results suggest that the hearsay testimony of children’s interviewers is degraded. Even immediately after an interview, important content was omitted by hearsay accounts, and the majority of the verbatim (specific wording and content of questions and answers) was lost. Our results also suggest that interviewers are unlikely to be able to accurately reconstruct verbatim information later.” (Warren & Woodall, 1999, p. 369).
Examples of interview problems (I = Interviewer, C = Child)
I: When did you have your birthday?
C: February 2.
I: February 2? Wow, and you’re 5 years old. That’s neat. So, did the clown come before your birthday or after your birthday? (Note: There had been no prior mention of a clown).
C: No clown come. I want to go back.
I: When the clown came here, was it just recently or was it back around your birthday?
C: Back around my birthday
I: Yeah? Did he have a clown face?
I: What did he look like?
C: He just had a white face. (Note: The magician did not have on any special face make up, but he was a White man).
I: A white face. I see.
C: It was a magician.
I: Oh, it was a magician, not a clown. That’s right.
C: He had a clown face.
I: So, he had he had a hat on?
C: He didn’t have a hat on. Some things were in his hat (Note: The child is correct that the magician used, but was not wearing, a hat).
Interviewer’s recall of above excerpt: “I asked how he was dressed – was he dressed like a clown – he said he had painted white face. I asked if he had clown clothes on.” (No answer to this questions was provided by the interviewer. (Warren & Woodall, 1999, p. 364). In this excerpt, the child did not say the magician had a “painted white face.” The interviewer was mistaken in this regard.
I: Wow, you know, I’m new here and I don’t even know who Tracy is. Do you know who Tracy is?
I: Is Tracy a boy or a girl?
C: A girl.
I: A girl? Is she a grown up girl?
C: She a big grown-up.
I: Big grown-up? Is Tracy someone you know from before?
Interviewers recall of above excerpt: “Asked her if she remembered Traci – is trace a boy or girl. She said girl. Is she a big girl or little girl? She said big girl. Asked her if Traci worked here. She said yes.” (Warren & Woodall, 1999, p. 364). In this excerpt, the interviewer did not ask the child if Traci “worked here?” This interviewer was mistaken regarding this issue.
I: So how did she check your arm? What did she do to check it?
C: She just had to put something on it.
I: Some paint on it? (Misleading question resulting from not understanding child’s answer). [In fact, Tracy placed “the child’s arms on a scale” to “measure muscle.” p. 360).
C: Yeah paint.
I: Paint. What color paint?
C: Black, but then it came right off
I: Oh really.
C: Yeah, it came right off real fast. Yeah, right off.
I: So she put paint on your whole arm?
C: Yeah, but it came right off.
I: Oh really?
C: It came off when I was asleep at home.
I: Uh-hmmm. So you didn’t wash it off.
I: You didn’t wash it off, it just came right off by itself.
C: Yeah, yeah.
Interviewer’s recall of above excerpt: “The child told me that he played doctor with Tracy and that she painted his arm black. His whole arm with stuff that just came off. He didn’t have to wash his arm. He just went home and it just came off.” (Warren & Woodall, 1999, p. 364-365). In this excerpt, the child only reported paint on his arm in response to the interviewer’s misinterpretation. This interviewer’s report was therefore factually mistaken.
A 2000 study investigated the accuracy with which trained, experienced interviews record verbatim notes of their interviews (Lamb et al., 2000). Specifically, this study compared the audiotaped recordings of 20 forensic interviews of alleged sexual abuse victims (5 male and 15 female 4- to 14-year-olds with the investigators’ verbatim accounts (notes) of the same interviews. 25% of the substantive (i.e. forensically relevant) details provided by the children were not represented in the investigators’ notes. The investigators’ notes reflected a total of 806 substantive interviewer utterances, whereas the audio recordings of the same interviews included 1889 utterances, leaving 1083 utterances (57.3%) unaccounted for by the interviewers’ supposed verbatim notes.
The interviewers’ notes misrepresented the utterances used to elicit information from the children. Only 44% of the interviewer utterances were accurately identified in the notes. In particular, there was a systematic tendency to mistakenly characterize interviewer questions as open-ended when, in fact, the questions were close-ended. These interviewers specifically failed to record 53% of their suggestive statements, “Even when they made contemporaneous verbatim notes, these investigators tended to understate their role in eliciting information and to ignore many of the details, including central details, reported” (Lamb et al., 2000, p. 705).
This study raises “… serious questions about the ability of interviewers to recall the content and structure of their interviews with the degree of precision needed for forensic purposes” (Lamb et al., 2000, p. 706). Consequently, Lamb and his colleagues disagree with the American Prosecutors’ Research Institute actively discouraging the electronic recording of investigative interviews.
The relevant research clearly demonstrates that hearsay accounts of investigative interviews in CSA cases warrant persistent skepticism. When interviewers offer hearsay testimony: (1) The accuracy with which they recall what children say is less than reliable. (2) The accuracy with which they recall their own questioning style is less than reliable. (3) Their attempts at relying on “contemporaneous/verbatim” notes are less than reliable, and (4) Their claims that they complied with an interview protocol are less than reliable.