Forensic Developmental Psychology

Misconception #1: sexually abused children do not disclose their abuse

A highly influential assumption is that sexually abused children do not readily disclose their abuse because of shame, guilt, and fear; consequently, there may never be a disclosure or else there is a long delay between the abuse and disclosure. In the latter case, it is asserted that children will initially deny their abuse when questioned, but will slowly divulge its details with repeated questioning. It is further claimed that these disclosures are frequently recanted, but will be reinstated with supportive questioning. The most popular embodiment of this idea is Summit's (1983) term child sexual abuse accommodation syndrome (CSAAS).

A corollary of CSAAS is that, in order to overcome emotional and motivational barriers that inhibit spontaneous disclosure of their abuse, children must be asked specific questions about it over a period of time. In part, this assumption is supported by the findings that children will provide more detailed answers to specific or cued questions than to open-ended questions. With age, children acquire cognitive structures that assist the organization of events into coherent narratives, and the need for specific questions declines. But when specific questions must be asked, such questioning comes with a cost; answers to specific questions are more error prone than those to open-ended questions (see Ceci & Bruck, 1995). The clinical practice of asking leading questions to children suspected of abuse is nonetheless defended by claiming that children will not falsely report abuse when asked specific questions (e.g., Rudy & Goodman, 1991).

Because the CSAAS model was based not on empirical data but on clinical intuitions, we recently reviewed the literature to determine its empirical support (London, Bruck, Ceci, & Shuman, in press). We identified 10 studies in which adults with histories of childhood abuse were asked to recall their disclosures in childhood. Across studies, an average of only 33% of the adults remembered disclosing the abuse in a timely fashion. These data support the view that sexually abused children are silent about their victimization and delay disclosure for long periods of time.

Although these studies are informative on the issue of delay of reporting, because the participants were never asked, “As a child, did anyone ever ask you or question you about abuse?” the data are silent on the phenomena of denial and recantation. Another set of studies provides some data relevant to this point. We identified 17 studies that examined rates of denial and recantation by sexually abused children who were asked directly about abuse when they were assessed or treated at clinics. The rates of denial at assessment interviews were highly variable (4% to 76%), as were the rates of recantation (4% to 27%). We found that the methodological adequacy of each study (sampling procedures, validation of sexual abuse) was directly related to the denial and recantation rates observed; the weakest studies produced the highest rates. For the 6 methodologically superior studies, the average rate of denial was only 14%, and the average rate of recantation was 7%. Thus, although the retrospective studies of adults show that children do not disclose abuse, the studies of children's response patterns indicate that if they are directly asked, they do not deny, but tell.

In part, the myth about children's patterns of disclosure has persisted because documentation of the first stage of the CSAAS model (children are silent and delay disclosures) has been interpreted as evidence for the full model, according to which denial and recantation are common. Also, as shown by our recent review and analysis, the most commonly cited studies in the literature are those that support the model—but sadly, these are the methodologically weakest of the studies.

Even if it is conceded that children will not deny their abuse when asked, the question of how to elicit disclosures remains. Even though there are known risks of using leading or specific questions, perhaps these are necessary to elicit reports or details from sexually abused children who feel frightened, ashamed, or guilty. This claim has recently been challenged by Lamb, Sternberg, and their colleagues, who constructed a structured interview protocol and then trained interviewers in its use. The protocol requires trained interviewers to encourage suspected child abuse victims to provide detailed life-event narratives through the guidance of open-ended questions (e.g., “Tell me what happened”; “You said there was a man; tell me about the man”). The use of specific questions is allowed only after exhaustive free recall. Suggestive questions are highly discouraged. In their latest study, Lamb et al. (2003) examined the interviews of police officers trained on the protocol with 4- to 8-year-old children who had made allegations of sexual abuse. Lamb et al. found that 83% of all allegations and disclosures were elicited through free-recall questions (78% for preschoolers), and 66% of all children identified the suspect through open-ended questions (60% for preschoolers). These data dispel the belief that interviewers need to bombard children with suggestive techniques in order to elicit details of trauma; rather, children can provide detailed information through open-ended prompts, and if a child denies abuse when asked directly, there is no scientifically compelling evidence that the child is “in denial.” Abused children usually disclose the abuse when directly asked.

Misconception #2: suggestive interviews can be indexed by the number of leading questions

According to our model of the factors that influence suggestibility (Bruck, Ceci, & Hembrooke, 2002), the suggestiveness (and thus the risk of eliciting false information) of an interview is not directly reflected by the number of leading questions, but rather is indexed by how interviewer bias plays out in the target interview, as well as in all previous interviews. Interviewer bias characterizes interviewers who hold a priori beliefs about the occurrence of certain events and, as a result, conduct their interviews so as to obtain confirmatory evidence for these beliefs without considering plausible alternative hypotheses. When children provide such interviewers with inconsistent or bizarre evidence, it is either ignored or interpreted within the framework of the biased interviewer's initial hypothesis. According to our model, interviewer bias influences the entire architecture of interviews and is revealed through a variety of suggestive interviewing techniques, including the use of repeated specific questions (some of which may be leading) within and across interviews; implicit or explicit threats, bribes, and rewards for the desired answer; stereotype induction (e.g., telling children the suspected perpetrator “does bad things”); and guided imagery (asking children to create a mental picture of a specific event and to think about its details; see Ceci & Bruck, 1995). Although each suggestive technique is associated with error, the risk for false statements is greatly augmented when interviews contain a combination of suggestive techniques, which increase the salience of the interviewer's bias.

There is considerable empirical support for this model. When children are questioned about events they did not experience (e.g., seeing a thief steal food from the day-care center; Bruck et al., 2002) or about nonoccurring details within experienced events (e.g., “the man put something yucky in your mouth”; Poole & Lindsay, 2001), their reports are more error prone if these techniques are used than if the questioning takes place in a neutral, nonsuggestive manner.

Sometimes suggestive interviews can be void of leading questions; and sometimes leading questions may not pose a risk to the reliability of children's reports in the absence of interviewer bias. For example, Garven, Wood, and Malpass (2000) asked kindergarten children to recall details about a visitor named Paco who came to their classroom and read a story, gave out treats, and wore a funny hat. The children were asked misleading questions about plausible events (e.g., Did Paco break a toy?) and about bizarre events (e.g., Did Paco take you to a farm?). Some of the children were given selective feedback after their answers to the misleading questions. “No” responses were negatively evaluated, as in the following exchange:

“Yes” responses were positively evaluated, as the following example illustrates:

This group of children provided the desired but false answer to 35% of the plausible questions and to 52% of the bizarre questions. In contrast, a second group of children who did not receive this selective feedback falsely agreed with 13% of the plausible and 5% of the bizarre questions. Thus, a simple count of misleading and leading questions would not reflect the suggestiveness of the interviews in this study; the children frequently agreed with the false suggestions only when the selective reinforcement provided sufficient information concerning the bias of the interviewer. Two weeks later, when the children were asked nonleading questions with no selective feedback, the same level of between-groups differences was obtained. Thus, interviewer bias in a prior interview has long-lasting negative effects on accuracy in a later unbiased interview.

Misconception #3: suggestibility is primarily a problem for preschoolers

Although much of the literature pays lip service to the concept that suggestibility exists at all ages, including in adults, the primary view is that preschool children are disproportionately suggestible, and that there should be less concern about the tainting effects of suggestive interviews with older school-aged children. The focus on younger children reflects the disproportionate number of studies of preschool children at the end of the 20th century. This practice was directly motivated by forensic concerns of the day; in a number of high-profile criminal cases, preschool children made horrific claims about sexual abuse. Although the case facts showed that these children had been subjected to highly suggestive interviews, at that time there was no relevant body of scientific literature to indicate the risk of such interviews in producing false allegations about a range of salient events. When researchers began to fill in this empirical void, most of the studies focused on preschoolers, and few examined age-related differences. Those that did include age comparisons usually found that the older children rarely fell sway to suggestion, leading to the conclusion that only preschoolers are suggestible (e.g., Ceci, Ross, & Toglia, 1987). However, this conclusion is discrepant with the findings of another body of literature showing that many of the suggestive techniques used in the child studies also produce tainted reports or false memories in adults (e.g., see Loftus, 2003). By inference, one might assume that children in middle childhood must also be quite suggestible, given that both younger and older groups are.

Recent evidence supports this view: Susceptibility to suggestion is highly common in middle childhood, and under some conditions there are small to no developmental differences in suggestibility. For example, Finnilä, Mahlberga, Santtilaa, and Niemib (2003) staged an event (a version of the Paco visit we described earlier) for 4- to 5-year-olds and 7- to 8-year-olds. One week later, half the children were given a low-pressure interview that contained some misleading questions with abuse themes (e.g., “He took your clothes off, didn't he?”). The other children received a high-pressure interview; they were told that their friends had answered the leading questions affirmatively, they were praised for assenting to the misleading questions, and when they did not assent, the question was repeated. In both conditions, there were no significant age differences in the percentage of misleading questions answered affirmatively, although a significant number (68%) were assented to in the high-pressure condition (see also Bruck & London, 2003; Zaragoza, Payment, Kichler, Stines, & Drivdahl, 2001). It has also been found that under some conditions, older children are more suggestible than younger children (e.g., Finnilä et al., 2003; Zaragoza et al., 2001).

Misconception #4: multiple suggestive interviews are needed to taint a report

The final misconception is that it is very difficult to implant memories or to taint reports, and that false reports occur only when multiple suggestions are repeated over time (e.g., Ceci & Bruck, 1995). However, many studies have reported that children can incorporate suggestions about salient events after a single interview. In the study by Garven et al. (2000), for example, children's reports were significantly tainted after a single suggestive interview. Moreover, one interview had lasting effects: Children's initial inaccurate responses to the suggestions may have reflected social pressure; however, their continued false reports when queried by different (neutral) interviewers at later sessions reflected their false belief that the events planted by suggestion had actually occurred.

Recent evidence also suggests that, contrary to common psychological principles, there are a number of circumstances in which one suggestive interview produces the same amount of taint as two or more suggestive interviews. The risks that a second interview will increase suggestibility depend on the spacing of the interviews and also on the memory strength of the original event (Melnyk & Bruck, in press).

Finally, there can be significant tainting of reports and production of false beliefs when interviews are only very mildly suggestive. For example, Poole and Lindsay (2001) had parents read their children short narratives that outlined the children's previous encounters with a character known as Mr. Science at the researchers' laboratory. Unknown to the parents, some of the details in the stories were inaccurate, and thus were not experienced by the children when they met Mr. Science. Nonetheless, even under these mildly suggestive conditions, significant numbers of children (4- to 8-year-olds) later told an interviewer that they had experienced the suggested events (e.g., “The man put something yucky in my mouth”).

Conclusion

We have reviewed some recent advances in the field of forensic developmental psychology that challenge four common misconceptions, some of which have acquired the status of urban legends in the field of clinical practice and forensic psychology. The data indicate that there should be greater concern that interviews with possible victims of child abuse are conducted using scientifically validated methods and less concern that true victims will deny that they were abused.

Theoretically, these new findings challenge current views of the developmental trends in suggestibility, and thus of the developmental mechanisms underlying children's suggestibility. Traditionally, researchers studied candidate mechanisms (e.g., theory of mind—knowledge that other people may have feelings, intentions, and beliefs different from one's own; social compliance) that were known to develop by the end of the preschool years (because it was thought that suggestibility was greatly reduced by that time). However, if suggestibility levels remain relatively high throughout childhood, a new perspective is required. For example, the relationship of suggestibility to skills that develop throughout childhood (e.g., resolution of conflicting information, insight into a questioner's motives) should become the focus of future study. It may also be useful to examine whether there are developmental changes in the mechanisms underlying suggestibility, with different mechanisms playing a causal role at different developmental levels.