due tomorrow

PRINCIPLES AND PRACTICE OF Child and Adolescent Forensic lvi ntal Health

Edited by Elissa R Benedek, M.D.

Peter Ash, M.D.

Charles L. Scott, M.D.

Washington, DC London, England

Chapter ] 6

Reliability and Suggestibility of Children's Statements

From Science to Practice

Sarah Kulkofsky, Ph.D. Kamala London, Ph.D.

n spring 1985, a 4-year-old student at the Wee Cale Nursery School was having his temperature taken with a rectal thermometm when he commented, "That's what my teacher does to me at nap time at school." The child's comment led to an mvestlgation of Kelly Michaels, a young teacher at the nursery school All children at the nursery school wele lepeat- edly questioned about potential abuse by their parents (who were informed about sexual abuse suspmlons in the school), by investigators, and in individual and gtoup therapy sessions. The chfldlen eventually made a number of horrendous and sometimes bizalre alle- gations, including that they were raped with spoons, knives, and Lego blocks; that Kelly Michaels fmced them to lick peanut butter off her naked body; that she forced them to eat feces and drink urine; and that she forced them to play games in the nude. Despite the fact that there was no physmal evidence m the case and that Kelly Michaels passed a polygraph test, she was indicted on 299 charges and, after an 11-month trial, was sentenced to 47 years m prison. She served 5 years m prison before her case was overturned on ap-

peal, largely as a result of the unreliability of the chil- dren's testimony, which was elicited in a highly sug- gestive manner.

In the United States, more than 4 million cases of child maltreatment ale investigated each year (Pipe et al. 2007) Histormally, children rarely provided uncor- roborated testimony in legal settings. However, be- cause of society's greater recognition of the prevalence and problems associated with child maltreatment, partmulmly with child sexual abuse (CSA), several ju- dicial leforms were enacted starting in the 1970s and 1980s that brought more children at younger ages into the courts. It is estimated that more than 100,000 childlen m the United States testify annually m crim- inal and civil proceedings (Cecl and Bruck 1993).

As children began flooding the legal system, con- cerns were raised about their ability to provide com- plete and accmate accounts about past events. Many of these concmns were raised in lesponse to some

high-profile sexual abuse cases, such as those revolv- ing Kelly Michaels, as well as similar cases involving Gerald and Cheryl Amlrault and the McMartm Pre- school. These cases were characterized by bizarre alle- gations of abuse made by multiple children as well as by aggressive and overzealous interviewing tech- niques. When many of these cases started to come to light, very little research existed regarding the reliabil- ity of children's statements or appropriate interview-

217

21 8 Principles and Practice of Child and Adolescent Forensic Mental Health

ing techniques. In the years since, a vast literature has amassed outlining chlldreffs abilities to provide re- ports of past experiences as well as Interviewer behav- iors that can compromise the integrity of chlldlen's statements.

In this chapter, we provide an overview of the liter- ature on the factors that can affect the accuracy of chil- dren's statements. In the first malor section, we pro-

vide a review of the llteratule on children's accuracy when leporting past events In the second major sec- tion, we address disclosure patterns among sexually abused children, as beliefs about how and when chil- dren disclose abuse often influence lntervlewels' deci-

sions to pursue more aggiesslve questioning strate-

gies Finally, we end the chapter with an overview of recommendations for ptactitionels.

Children's Accuracy When Reporting Past Events

Memory Development First and foremost, understanding children's abihty to provide complete and accurate reports of past events in fmensic contexts requires knowledge of the develop- ment of memory about personally experienced events. Research has shown that by around age 2 years, chil- dren begin to talk about past events (Nelson and Flvush

2004). At this young age, however, children's reports often require a great deal of adult prompting; their re- sponses to open-ended questions such as 'q-hll me what happened" tend to include very little detail (Fivush 1993). Because many of the studies of children's per- sonal memory focus on naturally occurring events, ac-

curacy is difficult to assess because the researcheIs do not kmow what actually happened. However, maternal reports tend to confirm children's statements (e.g, F1- vush et al. 1987), and work involving staged events for which statements can be venfied also has shown high rates of accuracy (e.g., Leichtman et al. 2000). As such, young cfuldren's spontaneous lecall of past events is of- ten chaiacterized as accurate but Incomplete.

The conclusion that children's spontaneous re-

ports are largely accurate has led some clinicians to conclude that if a child makes a spontaneous state- ment, it can be assumed to be true {Cecl et al. 2007a). However, it is certainly not the case that all spontane- ous statements are accurate. In particular, if the child is interviewed about confusing events or events that run counter to his or her knowledge, accuracy may be

compromised. This is particularly important because in many forensic contexts, such as CSA, children are indeed asked to report about confusing events.

For example, Ornstein et al. (1998) had children experience a mock medical examination In which some common features (such as listening to the child's healt) were omitted and atypmal features (such as wiping the child's belly button with alcohol) were added. When children were interviewed about the event after a 12-week delay, 42% of the 4-year-olds and 74% of the 6-year-olds spontaneously repolted that at least one of the missing common features had been a pint of the examination. In a similar study, Kulkofslc/ and colleagues (2008) had preschool-age children en- gage in a pizza-baking activity that included several unusual elements (e.g., the pizza was cut with chop-

stinks). When interviewed 1 week later, on average, 24% of the childlen's bee-recall statements were clas- sified as incorrect. Furthermore, it is impmtant to note that these errors were not inconsequential In na-

ture. For example, more than a quarter of all children recalled a nonpresent knife in response to an open- ended question, and many children added further de- tail and embellishment. To illustrate, one child re- called, 'ÿ very sharp knife.., it cuts us... I telled Aaron

'Don't use that sharp lÿife!'"

In actual forensic intmviews, the interviewer would not lÿow which statements made by a child were inaccurate. Thus, children may be generally ac- curate about previously expeiienced events, but the in- clusion of key inaccurate details may lead interviewers down a dangerous road. As we outline in the next sec- tion, erroneous beliefs on the part of an interviewer about what might have happened can negatively influ- ence the accuracy of children's statements.

FurthermoIe, although children are able to report memories of childhood experiences and may report memories from younger ages than adults are able to recall (Fivush and Schwarzmueller 1999), there does appeai to be a limit to how early in childhood children can lemember. Specifically, children show difficulty remembering events that occurred prior to the onset of language. For example, Petelson and Ridehout (1998) interviewed young children about a wslt to an emer- gency department that occurred when children were between 13 and 34 months old Only children who were 25 months and older at the time of the injury were able to recall verbally any details of the event at later intelvlews. In a related study, Simcock and Hayne (2002) exposed 2- to 3-year-old children to a novel event and then tested their memories 6 months and 1 year late1. At both the inmal exposure and the

Reliability and Suggestibility of Children's Statements: From Science to Practice 219

memory mtervmws, parents reported children's vocab- ulary, mchidmgwords that were pertinent to the novel event. At both time points, no child used words to de- scribe the event that had not been palt of the child's vocabulary at the time of the original event. Taken to- gether, these results indicate that later verbal recall of an event is, in part, dependent on childlen's language abihty at the time of encoding.

To summarize, the research on memory develop-

ment in young children suggests that, m general, young children's spontaneous reports of personally ex- permnced past events are largely accurate but can be qmte sparse. However, accuracy is impaired when chil-

dren are asked to recall confusing or ambiguous events, and their enors are not necessarily limited to inconse- quentlal or peripheral details. In addition, throe is a limit to how eally in childhood children can remember, and thus, the veracity of memories lecalled before the onset of language should be considered suspect.

e e e8u ,gestlbd ty Cecl and Bruck (1993) defined suggestibihty as "the degree to which children's encoding, storage, letrieval, and reporting of events can be influenced by a range of social and psychologmal factors" (p. 404). This broad deflmtion of suggestibility allows for mformatmn that is presented both before and after an event to tarot childlen's recall and further allows for the possibility that children's reports may be inaccurate even without any underlying memory impairment.

Although children's spontaneous repmts are gen- erally accurate, reports that emerge as a lesult of sug- gestive interviewing techniques tend to be qmte error- prone. In the classic sense, suggesttve interviewing in- volves aslÿng leading questions. Studms of actual in- vestigative interwews indicate that interviewers fre- quently ask children leading questmns (Gecl et al. 2007bJ. Moreover, training programs designed to teach best practmes for interviewing young wÿtnesses do not appear to be effective in reducing the number of leading questmns interviewers ask (Sternberg et al. 2001 b). In general, childlen are less accurate when an- swering direct questions compared with open-ended questions (Ornstem et al. 1998; Peterson et al. 1999), and young children ale less hkely than older chlldlen and adults to respond to leading questmns with "I don't kmow" (Hughes and Grieve 1980). Leading ques- tmns are particularly problematic because the rater- viewer presupposes that certain events occurred (e.g.,

"He took your clothes off, didn't he?'); however, with-

out knowing exactly what happened, whmh is ahnost always the case m forensic mtervlews,'an inter-

viewer's leading question may actually be misleading. The suggestiveness of an intervmw goes beyond

simply indexing the number of leading questions. Rather, one must consider how interviewer bias plays out in the intervmw. Interwewer bias characterizes those mtercieweÿs who hold a priori behefs about what has occuned and mold the interview to maximize dis- closures that are consistent with those beliefs. As such, intervmwer bias has a negative effect on the accuracy of child witnesses' statements. Experimental resealch has shown that when mtervmwers are misinformed themselves about what has occmred, children's reports me less accurate and tend to incorporate the mismfor- matron given to interviewers (Goodman et al. 1995; White et al. 1997). Table 16-1 provides categorms and examples of suggestive interwewing techniques.

Interviewer bias may be commumcated through other suggestive techmques in addition to asldng lead- ing questmns. These include providing pomtive rein- forcement and negative feedback, using peer or paren- tal pressure, creating a negative or an accusatory emotional tone, inducing stereotypes about the ac- cused, and repeating questions or interviews until the child provides the desired answer. Resealch indicates that, compared with using a single suggestive tech- nique, combining suggestive techmques tends to re- sult in heightened levels of suggestlbihty For example, Graven et al. (1998, 2000) examined how the tech- niques that were used by investigators in the McMar- tin Preschool case tainted children's testimony beyond that of misleading questions alone. In one study (Gar- yen et al. 2000), the researchers asked kindergarten children to recall details from when a visitor named Paco came to their classroom. Half of the children were given interviews that included misleading ques- tions about plausible events (e.g., "Did Paco break a toy?'J and bizarre events (e.g., "Did Paco take you to a farm m a hehcopter?'). In this group, children as- sented to 13% of the plausible questions and 5% of the fantastic questions. A second group of children were given negative feedback to theil "no" responses and positive feedback to their "yes" responses. This latter group falsely assented to the plausible items 35% of the time and the bizarre 1terns 52% of the time. Fur- thermore, these group differences remained when chfldlen were interviewed neutrally 2 weeks later. These data indicate that interviewer bias m earlier in- terviews can'tamt children's later reports even if these later interviews are conducted in an unbmsed manner.

1

i

] ,

Iij ,II I 1

I

'I!

,II

220 Principles and Practice of Child and Adolescent Forensic Mental Health

TABLE 16-1. Examples of suggestive interviewing techniques

Suggestive technique Example

Leading questions

Repeating questions

Positive reinforcement

Negative feedback

Peer or parental pressure

Intervmwel: He took your clothes off, didn't he?

Interviewer: Where did he touch you? Child. He didn't. Interviewer: Tell me where he touched you

Interviewer: Did he touch you on your bottom? Child: Yes Intervmwer: That's right. You're doing a really good lob.

Interviewer: Did he lrass you? Child: No Interviewer: You're not doing very well.

Interviewer: The other kids told me he did these things. I just need you to tell me.

Creating a negative or accusatory emotional tone

Inviting children to pretend or speculate

Interviewer These are bad guys. We're going to put these bad guys in jail.

Interviewer' What do you think it would have been hke? What could have happened?

To illustrate how leading questions may be com- bined with other techniques {such as question repeti- tion), consider the following interview examples from the Bernard Baran sexual abuse investigation.

Child A

Morn' No. Did he give you a box of donuts? Child: WhaU Morn: Did he give you a box of donutsÿ Child No Morn: No Did he give you anything or say any-

tlnng? Child: He gave us a birthday.

Interviewer. Did you play, did you play a game called "The Touching Game" at ECDCÿ

Child, No

Intermewer. Yeah So I was remembering, I lmow a game that I used to play called "The Touching Game." I wonder if you ever played that at school.

Child. I didn't Intervmwer. You don't remember that game? Child: We didn't do it Interviewel: You didn't? Child: No.

Child B Mom: You can't remember anything he said to you?

Did he say, "wake up," or he didn't say any- thing at all?

Clfild" No Morn: He just went and did wheat? Child: Nothing Mom: And after he pulled, after he pulled on your

pee pee, did he tell you anything then? Did he give you anythingÿ

Child: No.

The defendant In this particular case was convicted of eight courÿts of CSA and was sentenced to thiee con- current life terms in prison, based almost exclusively on the testimony of the young children in this case.

Although a combination of highly suggestive tech- niques is more likely to introduce false memory, such highly suggestive techniques are not necessary to taint children's reports. Children can incorporate mislead- lng infolmation into their accounts even after a single suggestive interview (Ceci et al. 2007b). Furthermore, other milder forms of suggestmns have been shown to influence the accuracy of children's reports. For exam- pie, m a recent set of studies, Prinmpe et al. (2006, 2008) showed that rumors spread among peers can lead to false repoits. Principe and colleagues found that many children not only reported false details as a result of rumor transmission but also maintained that they had actually seen the false events themselves.

In most typical suggestibility studies, there are re- liable age differences, with younger children being mine suggestible than older children and adults (Cecl and Bruck 1993). In fact, age appears to be the single best predictor of suggestibility (Geddie et al. 2000).

Reliability and Suggestibility of Children's Statements: From Science to Practice 221

i'

However, this is not to say that only young children are suggestible. A great deal of evidence indicates that older children and adults can fall prey to suggestive techniques (Finnila et al. 2003). Furthermore, in some situations, older childien actually may be more sug- gestible than younger children. In particular, in some cases, older children's more advanced cognitive capa- blhties actually led to increased incorporation of false information {Cecl et al 2007b; Pnncipe et alo 2008).

One argument that is often made against much of the research on suggestibility is that children are only suggestible about inconsequential, peripheral details of events. However, just as is the case with children's spontaneous eIiors, tile effects of suggesuve questmn-

ing are not limited to irrelevant and penpheral details of unemotional events. Studies showing the ddeterlous ef- fects of suggestive techmques have included central de- tails to negative and painful events, such as doctor's of- flee and emergency department visits (Bruck et al. 2000; Burgwyn-Bales et al. 2001) and other forms of bodily touching (Poole and Lindsay 1995; White et al. 1997}.

Finally, it is important to note that childten's false reports that emerge through suggestive techniques may be indlstmguishable from true statements. Both Cecl et al (1994} and Leichtman and Cecl (1995} had legal and psychological experts watch videotapes of children's true and false reports that emeiged as a re- sult of suggestive techniques. In both cases, tile profes- sionals were no better than chance at distinguishing true from false memories. Furthermoie, Bruck et al (2002} systematically compared childien's true and false narratives and found that false narratives con- tained mote spontaneous details, more temporal markers, more elaboiations, and more aggressive de-

tails than did true nairatives (see also Powell et al. 2003). Even when experts attempt to apply mole sys- tematic me%ods to distinguish true from false leports elicited through suggestive questions, then decisions are not reliable. For instance, criterion-based content

analysis has been touted as one way to d,stinguish true from false repoIts in forensic contexts (Vril 2005) In criterion-based content analysis, experts code the wit-

ness's statement for the presence of specific contents that are expected to occur more frequently in true re- ports. Although some limited evidence indicates that criterion-based content analysis can sometimes distln-

guish truthful statements from intentional lies, it can- not reliably distinguish true statements ftom false statements that were developed as a result of sugges- tive questioning techniques (Kulkofslo7 2008).

Taken together, the hterature shows that childIen are vulnmable to leading questions and other sugges-

tive techniques, including some very mild forms of suggestion Although young children appear to be the most suggestible, older children are also susceptible to suggestive techniques. Furthermore, children may be suggestible about central details of events and events that involve pain or bodily touch. Finally, children's re- ports that emerge through suggestive questioning of- ten appear quite credible.

If suggestive techniques are so dangerous, why do interviewers use them? Seveial possible leasons for their use exist. First and fmemost, children come Into contact with an interviewer only if there is some rea- son to suspect that thy have been abused This suspi- cion obviously colors the entiIe interview and may lead to intervmwm bias. Given that young childIen, in particular, often give very scarce leports, forensic in-

terviewers may be tempted to resort to moIe sugges-

tive techmques in an attempt to verify their suspicions of abuse. Commonly held beliefs about how children disclose abuse duiing interviews also may influence an interviewer's decision of how to question them. Some

intervIewms believe that the use of suggestive tech- niques is necessary because children may be reticent to disclose sexual abuse and may disclose abuse only in a lengthy process, if at all. Belief in this pattmn may lead interviewers to aggressively pursue children who they suspect to be victims of sexual abuse. We now turn to the Issue of whethm this belief in these disclo- sure patteins is actually warranted.

Disclosure Patterns Among Sexually Abused Children

Child Sexual Abuse Accommodation Syndrome In 1983, Summit wrote "The Child Sexual Abuse Ac- commodation SyndIome," a theoretical view based on his clinical experiences with his adult psychiatric pa- tmnts. He postulated that children who have experi- enced intrafamihal sexual abuse may be reluctant to disclose abuse because of motivational reasons such as being ashamed, scared, or embanassed. As a result, he

argued that abused children may delay abuse disclosure, deny abuse when asked, make pmtlal disdosures, and retract abuse disclosures. He latel extended the theory to include children who have expeiienced extlafamihal sexual abuse (Summit 1992}. Summit's theory (Sum- mit 1988, 1992) has exerted a tIemendous influence on

il

222 Principles and Practice of Child and Adolescent Forensic Mental Health

forensic interview pracUces with children and contin- ues to be taught internationally In many contemporary tlaining seminars for child abuse professionals.

Although Summit {1992) later cautioned practitio- ners that GSAAS is a clinical opinion, not a scientific or diagnostic instrument, some investigators have

latched onto the idea that sexually abused children deny abuse or even that denial is diagnostic of abuse. In many instances of alleged satanic ritualistic abuse cases from the late 1980s, themes emerged whereby the investigators reported refusing to believe children despite their repeated denials of abuse (London et al. 2005) Many clinicians and forensic interviewers con- tinue to intelpret abuse denials or inconsistencies in children's statements as consistent with the children passing through the stages postulated by CSAAS

The tenets postulated by CSAAS have undergone little scientific scrutiny to date. This is perhaps be- cause Summit's theory was seen as consistent with fo-

renslc interview practices and general beliefs about abuse disclosure (Summit 1992). However, in several recent publications, London and colleagues (2005, 2007, 2008) reviewed the literature on disclosure pat- terns among sexually abused children The goal of these reviews was to examine the contemporary em-

pmcal findings legardIng the nature and timing of children's sexual abuse disclosmes. Two main sources of data exist: 1) adults' retrospective accounts of CSA and whether they disclosed the abuse to anyone and 2) case records horn children undergomg contemporane- ous forensic evaluation. We briefly review the major findings below.

perienced CSA indicated that the abuse was brought to the attention of authorities, with 4 of the 7 studies re- porting rates ftom 10% to 13%. Although the retrospec- tive accounts are subject to problems inherent in any retrospecUve account, London et al. (2008) concluded that extant data suppolt Summit's notion of secrecy among sexually abused chIldien' only about one-third to one-half of clfildren ever tell anyone, and even fewer cases come to the attention of authorities.

Many of the retIospective studies repolted very long delays between the abusive episodes and children's dis- closures, sometimes of several years Although data are limited at present, the trend seems to be that some children disclose lelatively close in time to the abuse (e.g., within the filst 6 months), whereas others walt manyyeals or never tell anyone during childhood (Lon- don et al. 2008).

Research on Abuse Denials During Formal Interviews

Research on Secrecy and Delayed Disclosure Data from the retrospective accounts indicate that many adults reported that they never told anyone din- mg childhood about the CSA they experienced. Even fewer repoRed that the abuse came to the attention of authorities Across 13 retrospective abuse disclosure studies levlewed, 21%-87% of the participants reported that they disclosed the sexual abuse during childhood (for studies and citations, see London et al. 2008). Of the 13 studies reviewed, 11 found that between 34% and 54% of their adult sample who experienced GSA reported that they ever told anyone about the abuse during childhood. Fewer studies reported data on adults' retrospective lepolts of whether the abuse dis- closure involved authorities such as police or social workers. Across 7 studies to plovIde data on disclosure to authorities, 5%-18% of the adults who reportedly ex-

Although victims may not readily report sexual abuse, this does not mean that they will deny abuse ff asked directly about it. Studies of children undergoing as- sessment for suspected CSA provide the second source of data on disclosure pattelns of abused children. These studies generally examined archival records from children who weie seen by police, social workers, physicians, or assessment teams. Use of samples un-

dergoing forensic assessment allows an exploration of the extent to which childlen make denials and recan- tations during forensic assessment.

Unlike the letrospective studies reviewed earlier, London and colleagues (2005, 2007, 2008) reported a wide range of disclosure rates actoss 21 different studies to examine abuse disclosure during forensic or medical interviews. Disclosure rates ranged from a low of 23% (Sorensen and Snow 1991) to a high of 96% (Bradley and Wood 1996). Methodological features, paiticularly sample choices and intelview methods, appear to play a plimary lole In accounting for these discrepant rates. Because of its impoltance, we focus on this issue.

To calculate uue rates of disclosures and denials during forensic interviews, mfolmation is needed that accurately classifies childlen as abused or nonabused regardless of whether they make an allegation during the intelariew. At the same time, the chosen sample should be representative of all children to come before forensic interviewers. For example, sampling methods that ehminate childlen horn their sample who readily disclose to forensic interviewers would not provide ac- curate estimates of the overall rates of disclosure. At

T I Reliability and Suggestibility of Children's Statements: From Science to Practice 223

the same time, because abuse substantlation is often reliant on the child's dlsclosme, samples that include only highly probable or prosecuted cases may exclude possible true cases in which the child demes abuse during interviews

London et al. (2005) argued that disclosure rates during forensic mtervmws vary systematmally accord- ing to the certainty with whmh children in the study samples were abused. They divided the hterature into four major groupings:

Group i--cases of dubious vahdity Group 2--select subsamples

Group 3--all childlen to come before fmensm m- terviewel s Group 4--cases that come before fmenslc inter-

viewers that are rated as founded or highlyprobable

The lowest disclosure rates came from studms in Group 1, with very dubious or overtmned cases and documented poor interview techniques (Gonzalez et al. 1993; Sorensen and Snow 1991). In these studies, the abuse denials may have been true denials rather than evidence of reluctant disclosure. Many of the children from Gonzales et al.'s sample wele from the McMamn Preschool case. Children in the Sorensen and Snow (1991) study wele from a rash of neighborhood ritual- lstic satanic abuse cases, most of which either were not prosecuted or were later thrown out of court. Because of the documented highly suggestive techniques used in these studms (Schleiber et al. 2006), we argue that these studies do not provide any information about dis- closure patterns among abused children.

The second major group of studies repmted disclo- sure rates among select subsamples of chlldlen who come before authorities. These studies provide the second tier of disclosure rates--between 43% and 61% of children disclosed abuse when interviewed (for the study cltatmns, see London et al. 2008). Two types of cases are included in this grouping" 1 ) children under- going extended evaluauon for nondisclosure with high

suspicion of abuse and 2) children coming to the at- tention of authorities because of strong evidence of abuse (e.g., wdeotaped abuse evidence m sexually transmitted disease diagnoses). Although these stud- ies yield nnportant information about abuse disclo- sure among their specific subsamples, cauuon is war- ranted in generalizing the results beyond the context undel which these mterwews occurred. Rather, the disclosure rates m this second malta grouping are lep- lesentative of the minority of childlen who do not make disclosures during lmUal interviews

In the third major grouping of studms, lesearchers repmted data among all childlen to come before foren- sm mtervmwers legardless of abuse substantmtion. In sevelal newly published studies among all chddren wlth CSA suspmlons to come before highly trained fo- rensic mtmariewers who used a speclahzed interview protocol, between 71% and 83 % made disclosures (Hershkowitz et al. 2005; Pipe et al. 2007). Impor- tantly, m these studms, no efforts were made to weed out cases in which suspmmns arose but abuse really did not occur. One important caveat to this finding is that most of these children had made disclosures pnor to the intervmws, which was the impetus to the inves- ugation. This factor probably contnbutes to the much higher rates found m these studms.

The final grouping of studies report dlsclosme rates among highly probable cases that come befme ln- vestigatms. In these studies, efforts were taken by abuse assessment teams to rate the celtainty of abuse m hght of all the case materials available. The highest rates of disclosure, 85%-96%, were found among these studms reporting disclosure rates among cases classified as highly pmbable. We argue that these rates provide the best esumate of disclosure rates among general samples of abused children who come before forenmc interviewers because some effort must be made to determine which children wine abused before evaluating disclosme rates among abused children. Table 16-2 summarizes findings from prospective studies of disclosure.

TABLE 1 6-2. Summary of prospective studies of disclosure

Methodo{ogy Disclosure rates [%)

Children from dubious satanic ritual abuse or large-scale child care cases 24-58

Select subsamples of chlldlen to come to the attention of authorities 43-61

All children to come before highly trained mtervmwers using the National Insutute of 71-83 Child Health and Human Development mtervmw protocol

Highly probable cases (based on case materials) to come befole mvesugators 85-96

224 Principles and Practice of Child and Adolescent Forensic Mental Health

Research on Recantations During Formal lntel iews

Recommendations for Practice

Once children have disclosed abuse to authorities, how likely are they to recant then allegations ? Like abuse disclosure, lecantation rates depend on sam-

plIng method and abuse substantiation. Two of high- est recantation lates wele repmted by Gonzales et al. (1993 ] and Sorensen and Snow ( 1991 ). As discussed earlier, because of serious concerns about the forensic interview methods used and uncertainty of abuse sub- stantlatlon in these cases, we argue that these rates aie

not representative of cases that come before forensic interviewers who have been trained with contempo- rary interview protocols. Recantation rates of 5%-9% are reported In genelal samples of highly ptobable abuse cases. A higher rate was reported by Malloy et al (2007), who found that 23% of the ctnldren wlth sub- stantlated CSA cases facing dependency heanngs (for removal from the home for leasons stemming from the CSA) recanted their abuse allegations.

Ovelall, extant data on recantation of CSA disclo- sures suggest that 1) recantation occms in a minority of cases and does not typify abuse disclosure patterns, and 2) recantation may be more common in certain groups (e g., one with pressule or motivation to make recantation, such as having a nonsupportlve, non-

abusing caIegiver). When evaluating the minority of cases in which abuse recantation occurs, hke evaluat-

ing disclosure evidence, all of the evidence in the case, including potential motivators for recantation, should be consideied.

Summal ] of Research on Child Sexual Abuse Disclosure Taken together, the lesearch on childlen's disclosure patterns suggests only limited support for CSAAS-- namely, that many chMIen delay reporting or fail to le- port abuse. Howevel, much less empirical support ex-

ists for the Idea that chfldien who come before forensic interviewers and are asked directly about sexual abuse are likely to deny or recant abuse. The research thus dispels the notion that children who deny abuse must be bombarded with highly suggestive techniques in or- der to produce disclosures. Rather, as we outline in the next section, with caleful interviewing, children can provide detailed information in lesponse to open-ended prompts. We now turn to specific recommendations for forensic investigators to obtain the most accurate and complete reports from suspected wctims of abuse.

Although our first major section painted a somewhat grim picture of the reliability of children's statements, children certainly can provide accurate information about past events. Specifically, when interviews are neutral in tone and little or no suggestive questions are used, children can provide accmate and useful Infor- mation about past events, including traumatic events (Fivush 1993; Peterson and Bell 1996). Because of tlus, best practice guidelines that have been in effect for al- most two decades encourage forensic interviewers to

iely on open-ended questions as much as possible (AmmIcan Professional Somety on the Abuse of Chil- dren 1990, "Memolandum of Good Practice" 1992). However, for a variety of reasons outlined earher` in real-world contexts, forensic intel-viewers often fail to

follow these guidelines, even when they have under- gone extensive training (Sternberg et al. 2001 a)

Because of this issue, researchers at the National Institute of Child Health and Human Development (NICHD) developed a structured interview protocol for fmensic Interviewers (Orbach et al. 2000). The in- terwew protocol is not simply a set of best practices or training guidelines. Rathel, it prowdes a sclÿpt that guides the interviewer through the various stages of the interview. The plotocol begins with a truth-he dis- cussion to orient children toward truth-telling, then moves to non-abuse-lelated questions about the chil- dren's own life as a means to give childien practice talking about the past before asking about the target event, and finally moves through a series of open- ended plompts to obtain information about the alleged abuse. Only if key elements are still missing fiom the child repolt after going through the open-ended plompts are more specific questions allowed.

The NICHD intelarIew protocol has been shown to be effective even for young children (Lamb et al. 2003; Sternbelg et al 2001b). Fm example, In one study com- paring protocol interwews with standard interviews (Sternberg et al. 2001b), more Infmmation was elicited in lesponse to open-ended prompts in protocol inter-

views, and 89% of allegations were made in response to these prompts, compared with only 36% m standard interviews. Furthermore, although intervmwers in both conditions made suggestive statements, interviewers in the protocol condition made significantly fewer. Thus, use of scripted protocols will likely increase the reliabil- ity of children's statements m forensic contexts.

Reliability and Suggestibility of Children's Statements: From Science to Practice 225

However, it is lmpmtant to keep in mind that chil- dren's reports may reflect other contextual factms out-

side of the formal mvestigauve intm-vmw. As noted eal- lier, children may be inaccurate even without any suggestive influences, and even mild suggestions in the chlldlen's environment, such as lumors spread among

classmates, can negatively affect the accuracy of chil- dren's statements Furthermore, children often are in-

telÿciewed multiple umes by concerned parents, teach- ersÿ therapists, or social workels prior to their first formal fmensm interwew Thus, mvesUgators need to consider the full context of the child's disclosure when malong judgments about statement rehability. Finch- sic practitioners should lefram from the temptation to

make claims about speclahzed knowledge that allows them to determine the reliability of a child's statement

Forensic practitioners should remain aware that rumply because a child has been suspected of being maltreated, this does not mean that the child has nec- essarily been maltreated or that the suspect is neces- sarily the perpetrator In other winds, practmonels are advised to keep their own bmses m check This in- cludes the bias to assume that if a child demes that abuse has occuned, then the child must have CSAAS. By remaining as neutral as possible and conducting in- terviews m a neutral manneÿ; forensic mtelvlewers

will be able to get the most lehable statements horn young witnesses.

i i

I

--Key,Points

Children often are able to provide relatively complete and accurate accounts of past experiences, provided that they are interviewed in a neutral, supportive manner.

Simply because a statement is spontaneous does not mean that it is accurate. In particular, when young children are reporting about confusing or unusual events, their accuracy may be impaired.

Suggestive interviewing techniques can negatively affect the reliability of children's statements. Furthermore, other suggestive influences outside the interview context can also have negative effects. Importantly; false statements that are induced by suggestive interviewing techniques may appear quite credible.

Evidence indicates that children often delay or fail to report abuse, but little empirical support currently exists for the notion posited in CSAAS that denial and recantation typify children's reports. Although denial and recantation may occur in a minority of cases that come before forensic interviewers, the scientific evidence does not support a "syndromelike" disclosure pattern. Thus, aggressive questioning techniques designed to obtain disclosures from children who have denied abuse are unwarranted.

Forensic interviewers should follow best practice guidelines by relying on open-ended questioning whenever possible. The NICHD interview may be a useful model for obtaining reliable information from children in the most neutral manner possible. However, investigators need to take into account the full context of disclosure, nondisclosure, and recantation.

!

226 Principles and Practice of Child and Adolescent Forensic Mental Health

References

American Plofesslonal Society on the Abuse of Children' Guidelines for Psychosoclal Evaluation of Suspected Sexual Abuse in Young Chlldlen. Chicago, IL, Amerman Professional Socmty on the Abuse of Children, 1990

Bradley AR, Wood JM' How do chlldlen tellÿ The disclosure process in child sexual abuse. Child Abuse Negl 20.881-891, 1996

Bruck M, Cecl S, Flancoeur E. GhlldrenJs use of anatomi-

cally detailed dolls to leport genital touching m a med- ical examination' developmental and gender compari-

sons. J Exp Psychol Appl 6:74-83, 2000 Bruck M, Gecl SJ, Hembmoke H' The nature of children's

true and false memolles. Dec Key 22.520-554, 2002 Burgwyn-Bales E, Baker-Ward L, Gmdon BN, et al' Chil-

dren's memory for emelgency medical treatment after one year the nnpact of lndlwdual difference variables on lecall and suggestlbihty Appl Gogn Psychol 15'$25- $48, 2001

CecI SJ, Bruck M' Suggestiblhty of the child witness a his- toncal rewew and synthesis Psychol Bull 113 403-439, 1993

Ceci SJ, Loftus EF, Leichtman MD, et al. The possible role of source mlsatmbutlons in the creation of false behefs

among pleschoolers. Int J Clln Exp Hypn 42'304-320, 1994

Ceci SJ, Kulkofslcy S, Y, Jemfuss JZ, et al Unwarranted as- sumptions about children's testlmomal accuracy Ann Rev Chn Psychol 3:311-328, 2007a

Cecl SJ, Paplerno P, Kulkofsky S" Representatmnal con- stlamts on children's suggestlbthty. Psychol Scl 18:503-

509, 2007b Finmla K, Mahlberga N, Santtilaa P, et al" Validity of a test of

clul&en's suggestibility for predicting lesponses to two interview SltUatmns differing m their deglee of sugges- tiveness J Exp Chdd Psychol 85.32-49, 2003

Fivush R Developmental pelspectives on autobiogIaphmal recall child wctims and child witnesses, in Under- standing and Imploring Testimony. Edited by Good-

man G, Bottoms B New York, Guilford, 1993, pp 1-24 lÿlvush R, Schwarzmueller A Children remember child-

hood lmplicatmns for childhood amnesla Appl Cogn Psychol 12 455-473, 1999

Fivush R, Gray JT, Fromhoff FA Two year olds talk about the past. Cogn Dev 2 393-410, 1987

Garven S, Wood JM, Malpass R, et al. More than suggestmn. consequences of the interviewing techniques from the McMartln preschool case. J Appl Psychol 83:347-359, 1998

Garven S, Wood JM, Malpass RS: Allegatmns of wrongdoing. the effects of reinfmcement on clÿldren's mundane and

fantastm claims J Appl Psychol 85.88-49, 2000

Geddle L, Fradin S, Beer J' Child characteristics which nn-

pact accuracy of recall in pleschoolers' is age the best predmtor? Child Abuse Negl 24'223-235, 2000

Gonzalez LS, Waterman J, Kelly RJ, et al' Children's patterns of dmdosures and recantatmns of sexual and ntuallstm

abuse allegations m psychotherapy Child Abuse Negl 17:281-289, 1993

Goodman GS, Sharma A, Thomas SF, et al. Mother lrmows best effects of relatmnshlp status and interviewer bias on children's memory J Exp Child Psycho160:195-228, 1995

Herslfl<owltz I, Horowltz D, Lamb ME' Trends of children's disclosure of abuse m Israel' a national study Child Abuse Negl 29 1203-1214, 2005

Hughes M, Grieve R. On asldng children bizarre questions. First Language 1"149-160, 1980

Kulkofsky S' Credible but inaccurate' can critmion-based

content analysis (CBCA) distinguish true and false memories? in Child Sexual Abuse' Issues and Chal-

lenges. Edited by Smith MJ. New York, Nova Scmnce Publishels, 2,008, pp 9,1-42

Kulkofsky S, Wang Q, CecI SJ' Do better Stolies make better memories? Narrative quality and memory accuracy In preschool chlldlen. Appl Cogn Psychol 22'21-38, 2008

Lamb ME, Steinberg KJ, Orbach Y, et al: Age differences m young children's responses to open-ended Invltatmns in the course of forensic interviews J Consult Chn Psy- chol 71"926-934, 2003

Lemhtman MD, Cecl SJ' The effects of ste, eotypes and sug- gestions on preschoolers' reports Dec Psychol 31 568- 578, 1995

Lemhtman MD, Pfllemer DB, Wang Q, et al When Baby Maisy came to school, mothers' mtervmw styles and children's event memories. Cogn Dev 15:99-114, 2000

London K, Bruck M, Cecl SJ, et al' Disclosure of child sexual abuse' what does the research tell us about the ways that children tell? Psychol Pubhc Policy Law 11:194- 226, 2005

London K, Bruck M, Ceci SJ, et al. Disclosure of child sexual abuse a rewew of the contempolary hteratuie, m Child Sexual Abuse. Disclosme, Delay, and Denial. Edited by

Pipe ME, Orbach Y, Lamb ME, et al. Mahwah, NJ, Lawrence Erlbaum, 2007, pp 11-40

London K, Bruck M, Wright DB, et al. How children report sexual abuse to others findings and methodologmal 1s- sues Memory 16 29-47, 2008

Malloy LC, Lyon TD, Quas JA Filial dependency and recan- rattan of child sexual abuse allegations J Am Acad Child Adolesc Psychiatry 46' 162-170, 2007

Memorandum of Good Practice. London, Her Majesty's Sta-

tmnery Office, 1992 Nelson K, Flvush R. The emergence of autobmgraphical

memory: a social cultural developmental theory. Psy- chol Rev 111.486-511, 2004

Reliability and Suggestibility of Children's Statements: From Science to Practice 227

Orbach Y, Hershkowltz I, Lamb ME, et al. Assessing the value of scripted protocols for fmensic interviews of al-

leged abuse vmtlms. Child Abuse Negl 24 733-752, 2000

Ornstem PA, Mernt KA, BalKer-Ward L, et al Children's

lmowledge, expectation, and long-term letenuon Appl Cogn Psychol 12:387-405, 1998

Peterson C, Bell M' Children's memory for traumatm injury Child Dev 67:3045-3070, 1996

Peterson G, Rideout R. Memory for medmal emergencies ex- perienced by 1- and 2-year olds Dev Psychol 34' 1059-

1072, 1998 Peterson C, Dowden G, Tobm J. Interviewing preschoolers:

compansons of yes/no and wh- questions. Law Hum Be- hav 23:539-555, 1999

Pipe ME, Orbach Y, Lamb ME, et al' Seehng lesolutlon in the &sclosure wars: an overview, m Child Sexual Abuse. Disclosure, Delay, and Demal. Edited by Pipe ME, Or- bach Y, Lamb ME, et al. Mahwah, NJ, Lawrence Er-

lbaum, 2007, pp 1-10 Poole DA, Lmdsay DS. Interviewing preschoolers: effects of

nonsuggesuve techmques, parental coaching, and lead- mg quesuons on reports of non-experienced events. J Exp Chdd Psychol 60 129-154, 1995

Powell MB, Jones CH, Campbell C: A comparison of ple- schoolers' recall of expenenced versus non-expeuenced

events across muluple mterv,ewers. Appl Cogu Psychol 17"935-952, 2003

Punclpe GF, Kanaya T, Cecl SJ, et al: Believing is seeing, how rumors can engender false memories m preschoolers Psychol Sct 17:243-248, 2006

Prmclpe GF, Guihano S, Root C' Rumor mongermg and re- membering' how rumors ongmatmg m children's infer- ences can affect memory J Exp Child Psychol 99 135- 155, 2008

Schrelber N, Bellah LD, Mamnez Y, et al Suggesttve mter- wewmg m the McMartm Preschool and Kelly Mlchaels daycare abuse cases, a case study. Social Influence 1.16-

47, 2006 Simcock G, Hayne H. Breakang the barrier? Children fad to

translate their preverbal memories into language. Psy- chol Sm 13.225-231, 2002

Smensen T, Snow B. How children tell' the process of disclo- sure m chdd sexual abuse. Child Welfare 70.3-15, 1991

Sternberg KJ, Lamb ME, Dawes GM, et al The memoran- dum of good practme, theory versus apphcatmn. Child Abuse Negl 25.669-681, 2001a

Sternberg KJ, Lamb ME, Orbaeh Y, et al: Use of a structured invesugatlve protocol enhances young children's re- sponses to free-recall prompts in the course of fmenmc mtelÿaews J Appl Psychol 86.907-1005, 2001b

Summit PC. The Child Sexual Abuse Accommodauon syn- drome. Child Abuse Negl 7 177-193, 1983

Smmmt PC' Abuse of the child sexual abuse accommoda-

tmn syndrome J Child SexAbus 1'153-164, 1992 Viii A' Crlterm-based content analysis a qualitative review

of the hrst 37 studies Psychol Pubhc Pohcy Law 11 3- 41, 2005

Wtnte TL, Leichtman MD, Ceci SJ. The good, the bad, and the ugly: accuracy, maccuracyÿ and elaboratmn m pre- schoolers' reports about a past event Appl Gogn Psy- chol 11:$37-$54, 1997