CHAD 300


27W i n t e r 2 0 0 6 • J u v e n i l e a n d F a m i l y C o u r t J o u r n a l

II N T R O D U C T I O N Cases involving child sexual

abuse (CSA) are among the

most heartbreaking cases a

judge hears, for both emo-

tional and legal reasons.

How does the court evalu-

ate the testimony of the

young child who says it

never happened, the child

who discloses months

after the alleged event,

the girl who recants after

visitation with family mem-

bers, or the boy with an IQ of 51 who cannot

clearly articulate how the alleged abuse occurred?

Proving child sexual abuse in the absence of physi-

cal evidence or testimony of an eyewitness is difficult.

Children recant, child development issues intervene,

and cognitive limitations raise questions, while the testi-

mony of the adult perpetrator does not waiver. Indeed,

although errors in either direction can have devastating

consequences, finding the truth in CSA cases seems too

often impossible. Decision makers confront the twin

specters of leaving inarticulate children unprotected

from further traumatic sexual abuse on the one hand,

and subjecting innocent caregivers to criminal prosecu-

tion or the loss of parental

rights on the other.

Science can be a tre-

mendous asset to judges in

understanding and interpret-

ing the behavior of victims

of child sexual abuse, since

their behavior can often

seem counterintuitive. It is

essential that judges consider

children’s disclosure patterns

in light of current research in

order to have the greatest

chance to evaluate the facts

and find the truth. Children’s disclosure patterns are

crucial because physical findings are diagnostic of child

sexual abuse in 10% or fewer cases (Frasier & Makaroff,

this issue). Sexual abuse, especially when there is no

penetration, rarely results in physical trauma. Even when

there has been sexual penetration, the capacity for rapid

healing of the genital anatomy inhibits the detection of

evidence (The National Research Council, 1993, p. 72).

Therefore, children’s statements are central both to the

prosecution of the crime of child sexual abuse and the

protection of children from further abuse.

This review is intended to update criminal, juve-

nile, and domestic relations court judges who preside

The State of the Debate About Children’s Disclosure Patterns


A B S T R A C T In current research studies about the disclosure patterns of sexually abused children, experts agree that most victims delay disclosure for years, often until adulthood. Researchers disagree about dis- closure rates and recantation rates among children during formal interviews. Studies of children who had not previously disclosed but are known through corroborative evidence to have been sexually abused show lower rates of disclosure than do studies of children who had disclosed prior to the formal interview. Gradual disclosures among children are common, and more than a single interview may be necessary in some cases. Prior disclosure, level of support by non-offending parents, developmental level, and relationship to perpetrator affect children’s rates of disclosure and their disclosure patterns. More research is necessary to clarify children’s post-disclo- sure recantation rates and predictors.

Erna Olafson, Ph.D., Psy.D., is Associate Professor of Clinical Psychiatry and Pediatrics at Cincinnati Children’s Hospital Medical Center and the University of Cincinnati College of Medicine. She is Director of the Program on Child Abuse Forensic and Treatment Training at Cincinnati Children’s Hospital Childhood Trust. Her presentations and publications focus on domestic violence, sexual assault, and child abuse. Judge Cindy S. Lederman is the Presiding Judge of the Juvenile Court in Miami-Dade County, Florida. A member of the NCJFCJ’s Board of Trustees, she was instrumental in creating the Dade County Domestic Violence Court and served as the court’s first Administrative Judge. In 1999, Judge Lederman was awarded a Fellowship from Zero to Three: The National Center for Infants, Toddlers and Families in their Leaders of the 21st Century Initiative.

C h i l d r e n ' s D i s c l o s u r e P a t t e r n s i n S e x u a l A b u s e C a s e s

28 J u v e n i l e a n d F a m i l y C o u r t J o u r n a l • W i n t e r 2 0 0 6

over CSA cases about current areas of agreement and

disagreement among scientific researchers about the

disclosure patterns of CSA victims. A major volume on

abuse disclosure patterns is scheduled for publication

in 2006 (Pipe, Lamb, Orbach, & Cederborg, in press).

It contains chapters by researchers from differing

perspectives that we have drawn upon for this article

(London, Bruck, Ceci, & Shuman, in press; Lyon, in press).

Unfortunately for those charged with making decisions

about children’s welfare, no single school of researchers

has the last word on these controversial issues.

Brief History Many scholarly papers about children’s disclosure

patterns either begin with a discussion of Roland

Summit’s Child Sexual Abuse Accommodation Syndrome

(CSAAS) or structure their arguments around his model

of children’s behavior in these cases (London, Bruck,

Ceci, & Shuman, 2005; Lyon, 2002; Lyon, in press;

Summit, 1983). Summit argued that children often deny

being sexually abused, even when they are directly

asked, and that disclosing children often subsequently

recant their allegations. He based the accommodation

syndrome primarily on cases of intrafamilial child

sexual abuse (incest) rather than extrafamilial child

sexual abuse.

Although Summit’s “syndrome” has been litigated

with a variety of outcomes in many courts, it may have

become so controversial that it obscures rather than

clarifies the issues at hand. Judges should bear in mind

that for almost a century before Summit published his

influential paper, there was statistical evidence that

children often delay disclosure or remain completely

silent about sexual victimization. Indeed, this prior

literature was so extensive that a major psychological

journal rejected Summit’s accommodation syndrome

paper before it found publication elsewhere because,

the reviewers argued, it contributed nothing new

(Lyon, in press; Olafson, 2002). There have also been a

number of studies documenting children’s disclosure

patterns in otherwise corroborated child sexual abuse

cases since the 1983 publication of Summit’s paper

(Lyon, in press). Examining children’s disclosure pat-

terns one category at a time, without organizing them

around Summit’s now-controversial accommodation

syndrome, may clarify and simplify the issues.

The Issues What are the disclosure and non-disclosure patterns

among children known to have been sexually abused?

There are several issues:

■ Do most child victims delay reporting sexual abuse, sometimes until adulthood?

■ If directly asked, do most child victims disclose sexual abuse?

■ If directly asked, do some CSA victims initially fail to disclose or deny being abused, so that more than one formal interview becomes necessary?

■ How common is incremental abuse disclosure, from partial and fragmentary accounts to full disclosure over time?

■ Once children have disclosed sexual abuse, do a high percentage of known victims subsequently recant or retract their disclosures?

■ Are there factors such as gender, developmental level, culture, degree of abuse severity, parental sup- port, and relationship to perpetrator that influence disclosure patterns among CSA victims?

Sources of Information

The two most reliable sources of information about

disclosure patterns in CSA victims are:

■ Retrospective surveys of adults who report having been sexually abused during childhood; and

■ Research about children’s statements during evaluation and treatment in cases with cor- roborative evidence that is independent of children’s statements, such as videotapes of the actual abuse, physical findings, sexually trans- mitted diseases, and offender confession.

Both sources are imperfect. Cases that have indepen-

dent corroboration may be unrepresentative of sexual

abuse cases in general. Retrospective surveys depend on

human memory over time, so that under-reporting, over-

reporting, and inaccurate reporting may occur.

Nevertheless, cases with independent corrobora-

tion and retrospective surveys are superior to the

other sources sometimes used in literature reviews. For

example, studies that claim substantiation or conviction

rates as “independent” corroboration may significantly

inflate the percentages of actually abused children who

disclose their victimization during formal question-

ing. This is because substantiation, prosecution, and

E r n a O l a f s o n a n d J u d g e C i n d y S . L e d e r m a n

29W i n t e r 2 0 0 6 • J u v e n i l e a n d F a m i l y C o u r t J o u r n a l

conviction depend so heavily at all decision stages on

children’s statements. To argue that substantiation rates

that depend upon children’s disclosures proves that

most children make disclosures when interviewed is to

argue in a circle (Lyon, in press).

The definitions of key terms also affect research out-

comes, but researchers do not always specify their oper-

ational definitions. “Child sexual abuse” can include a

wide variety of behaviors, from non-contact exposure to

genital fondling to violent genital, oral, and anal rape. In

this article, we focus primarily on contact child sexual

abuse. “Disclosure” also has a variety of meanings. We

define disclosure to mean a clear verbal statement that

at least one abusive act took place, although a disclosure

need not be a complete report of everything that hap-

pened. Our definition does not include suggestive doll

play and other fragmentary “partial disclosures” that,

when included in research studies, artificially inflate

children’s “disclosure” rates (e.g. Dubowitz, Black, &

Harrington, 1992).

“Non-disclosure” can also vary in meaning depend-

ing on whether it refers to a child’s non-disclosure dur-

ing a single or initial interview or a child’s non-disclo-

sure maintained over six or more interviews. Children

questioned only once show higher “non-disclosure”

rates than do children questioned several times, so that

studies such as that by Sorenson & Snow (1991) that

show very high initial non-disclosure rates have an even-

tual disclosure rate of over 90%.

Child Sexual Abuse Disclosures Delayed Until Adulthood

There appears to be a consensus among research-

ers that most child sexual abuse victims delay disclos-

ing, often until adulthood. A number of well-designed

retrospective surveys now show that the great majority

of victims delay disclosing contact child sexual abuse

during childhood (Finkelhor, Hotaling, Lewis, & Smith,

1990; Smith et al., 2000). These surveys also indicate

that even when adults recall having told someone

about the abuse, the majority of these cases were not

then reported to the authorities. In one survey, 28% of

respondents stated that they had disclosed to no one

before telling the telephone interviewer about the child

sexual abuse (Smith et al., 2000); another survey found

that 42% of men and 33% of women first told anyone

about having been sexually abused as children when

asked during the retrospective telephone interview

(Finkelhor et al., 1990).

London and colleagues (2005) summarize the ret-

rospective literature by noting that the results of 10

retrospective surveys indicate that only one-third of

adults who suffered child sexual abuse revealed the

abuse to anyone during childhood. The study concludes

that “approximately 60%-70% of adults do not recall

ever disclosing their abuse as children, and only a small

minority of participants (10%-18%) recalled that their

cases were reported to the authorities” (London et al.,

2005, p. 203). Although London and colleagues note the

research limitations inherent in adult retrospective liter-

ature, they also write, “Given the differences in method-

ology, definitions of abuse, and sample characteristics,

the general consistency of these findings across these

studies is noteworthy” (London et al., 2005, p. 201; but

see Poole & Dickinson, 2005).

Judges and other fact finders can only adjudicate

those cases that come to their attention, and a child’s

prior disclosure to a caregiver or friend constitutes

the most common means by which child sexual abuse

comes to the attention of the authorities and thus to the

courts (Lyon, in press). Therefore, because it appears

that most people delay disclosing until adulthood, chil-

dren who decide to tell someone about being sexually

abused and whose cases therefore come to court are

not representative of sexually abused children in gen-

eral. In other words, child protection authorities and

the judiciary are likely to see only a minority of those

children who are actually being sexually abused. There

are, of course, some sexual abuse cases that are report-

ed for reasons other than a child’s prior disclosure, such

as children’s sexualized behaviors, physical findings,

and other external evidence. This review article focuses

on the disclosure patterns and behaviors among both

groups of sexually abused children, those who had pre-

viously disclosed and a smaller number of those who

came into the system in some other way.

Child Sexual Abuse Disclosures Delayed within Childhood

There appears to be agreement among researchers

from diverse perspectives that “when children do dis-

close, it often takes them a long time to do so” (London

C h i l d r e n ' s D i s c l o s u r e P a t t e r n s i n S e x u a l A b u s e C a s e s

30 J u v e n i l e a n d F a m i l y C o u r t J o u r n a l • W i n t e r 2 0 0 6

et al., 2005, p. 204). In a study of 399 children aged 8

to 15, Elliott and Briere (1994) find that of 248 subjects

assessed as having been sexually abused, 74.9% did not

disclose their abuse to anyone within the year that it first

occurred, and 17.8% had waited more than five years to

tell anyone. The courts are likely to see many such cases

in which children delayed reporting for months or even

years before telling someone about the abuse. It is also

not unusual for children to disclose the abuse long after

adjudication when they are in a safe environment and

the litigation is finished. Delays in telling anyone about

the abuse for several months, a year, or even longer

occur in a significant percentage of child sexual abuse

cases (Henry, 1997; Sas & Cunningham, 1995).

In weighing the evidence in child sexual abuse cases, judges and other fact finders should be aware that, in a high percentage of actual CSA cases, there will be delays of months or even years between the onset of the abuse and a child first disclosing to another person.

Children’s Gradual Disclosures during Formal Interviews

Many prosecutors are familiar with the problem of

incremental disclosure, in which a child may disclose

only aspects of an abusive event, such as genital fon-

dling, during the initial interview. Shortly before trial

is scheduled to begin, the child, perhaps during court

preparation with the prosecutor, describes new details,

such as penetrating oral sex, that necessitate postpone-

ments, the filing of new criminal charges, and concerns

about the child’s credibility and competence. In one

such case, a young incest victim, when asked why she

had not mentioned crucial additional information dur-

ing her initial advocacy center interviews responded,

“I just didn’t think of it.” This pattern of partial disclo-

sure can be explained by Summit’s classic child sexual

abuse accommodation syndrome, but it may also simply

reflect the usual patterns of recall in the very young.

In an experimental study, Dr. Robyn Fivush asked non-

abused children aged 3-6 about a known event on

two subsequent occasions (Fivush, 1994). On the two

recall occasions, children reported different but still

accurate information about the events, with an overlap

of details between the two retellings of only 20%. This

research about children’s normal patterns of recollec-

tion and reporting could in itself justify recommending

that children be given more than a single interview to

tell the authorities about the events in their lives.

In a summary of 21 studies from 1965 to 1993 of

children diagnosed with gonorrhea, Lyon finds gradual

disclosure by children to be very common (Lyon, in

press). In 118 CSA cases studied by Elliott and Briere

(1994), there was external evidence for the abuse,

including, for example, medical evidence diagnostic of

child sexual abuse, perpetrator confession, a witness

to the abuse, or pornographic pictures of the child.

In a number of these 118 cases, victims disclosed par-

tially in the first interview by mentioning fondling, but

when investigators confronted them with the external

evidence for more severe abuse (penetration), the chil-

dren then made more complete disclosures.

Thus, when questioned during formal interviews,

children may only partially disclose during the initial

interview. Because evidentiary studies show that trau-

matic medical evidence (such as a ruptured hymen) is

lacking in a significant number of cases in which per-

petrators have confessed to penile penetration, judges

should not prematurely regard children’s statements

as complete after a single interview (Muram, Speck, &

Gold, 1991). As Elliott and Briere (1994) write, “Forensic

evaluations that consist of a single interview may result

in incomplete disclosure and less accurate determina-

tions, especially in cases where medical or other exter-

nal data are lacking or inconclusive” (p. 274).

This recommendation does not contradict the long-

held principle in the child protection fields to avoid

subjecting children to repeated interviews by multiple

investigators from social services, law enforcement, and

the court system. The National Children’s Advocacy

Center has developed and tested guidelines for extend-

ed forensic evaluations with reticent children. If several

interviews become necessary, it is recommended that a

single interviewer conduct them and that the question-

ing be sensitively structured to build rapport over time

and avoid repetitive questioning and suggestiveness

(Carnes, Wilson, & Nelson-Gardell, 1999; Carnes, Nelson-

Gardell, Wilson, & Orgassa, 2001).

E r n a O l a f s o n a n d J u d g e C i n d y S . L e d e r m a n

31W i n t e r 2 0 0 6 • J u v e n i l e a n d F a m i l y C o u r t J o u r n a l

Because many sexually abused children in externally corroborated cases are known to disclose only gradually, more than a single interview may become necessary to serve children’s safety and justice. See the guide- lines by the National Children’s Advocacy Center (Carnes et al., 1999; 2001).

Non-Disclosure or Denial by Children When Interviewed about Child Sexual Abuse

The most troubling cases for the courts are those in

which there are red flags indicating a strong possibility of

child sexual abuse: The case is reported, the child inter-

viewed, and the child discloses no sexual abuse. There are

two classes of children to consider here:

■ Children who previously disclosed partially or fully to another person and thus precipitated entry into the system; and

■ Children who came into the system through other means, such as diagnosis of a sexually transmitted disease during routine medical care, extreme sexual- ized behaviors, or the discovery of videotapes docu- menting the abuse.

It is about children’s disclosure patterns once they

are in the system that the experts disagree, and these

cases are the most troubling to those responsible for

protecting children from abuse and protecting adults

from false allegations.

London et al. (2005) state that “the data clearly demon-

strate that most children who are interviewed about sexual

abuse do disclose and do not later recant…” (p. 217).

Lyon (in press) responds with a critique that reveals

problems with two kinds of case selection bias in many

of the samples upon which London and colleagues

based the above conclusion. Lyon argues that:

■ To avoid suspicion bias, one must examine cases that did not come to the attention of the authorities because a child disclosed to someone prior to the formal interview; and

■ To avoid substantiation bias, one must examine cases in which substantiation was completely inde- pendent of the child’s statements.

To understand how both forms of selection bias

artificially inflate the actual rates of children’s sexual

abuse disclosures, consider the following extreme case.

If we suspect sexual abuse only when a child has previ-

ously disclosed, then 100% of children in a sample of

children suspected of being sexually abused will have

disclosed at some point. If we substantiate child sexual

abuse only if a child discloses, then 100% of children

in a sample of substantiated cases will have disclosed.

The reality is only somewhat less extreme. The great

majority of suspected CSA cases come to our attention

only because a child has previously disclosed. Child

sexual abuse substantiation also depends most heavily

on children’s disclosures, because external evidence of

child sexual abuse (such as physical findings or offender

confession) is rare and generally detected only after

sexual abuse has been suspected.

London et al. (2005) seem to agree with Lyon about

suspicion bias by writing, “Prior disclosure of abuse pre-

dicts disclosure during formal assessment” (p. 209), but

they do not then systematically deal with the problem of

suspicion bias. London and colleagues also acknowledge

but do not fully address the substantiation bias problem

by writing, “In many of the cited studies, classification of

abuse was often based in part on children’s disclosures;

consequently, the conclusion that abused children do

disclose abuse during formal interviews may be circular”

(p. 217). They then base their conclusion that “the evi-

dence fails to support the notion that denials, tentative

disclosures, and recantations characterize the disclosure

patterns of children with validated histories of sexual

abuse” (p. 194) on their review of research studies that

are in many cases flawed by both suspicion and substan-

tiation bias. What do studies that avoid both biases tell

us about this area of contention?

Studies of Disclosure Patterns in Cases without Selection Bias

Nine boys and one girl were interviewed by police

after Swedish law enforcement discovered videotapes of

102 incidents of child sexual abuse, ranging from expo-

sure of the child’s genitals to oral/anal/vaginal intercourse

(Sjoberg & Lindblad, 2002). The perpetrator was either

related to the children or knew them through his work

at a day care center. Abuse severity was coded both from

the videotapes and from children’s statements. No child

had previously disclosed abuse nor had it been suspected.

Five children reported no abuse during police interviews,

for a disclosure rate of 50%. The child who had suffered

C h i l d r e n ' s D i s c l o s u r e P a t t e r n s i n S e x u a l A b u s e C a s e s

32 J u v e n i l e a n d F a m i l y C o u r t J o u r n a l • W i n t e r 2 0 0 6

the greatest number (60 incidents) and most severe

sexual assaults according to the videotaped evidence did

not disclose during the police interview. Two of the five

children who did disclose did so only in response to lead-

ing questions. No child reported any sexual behavior not

documented on the videotape.

Cases with children not suspected to be sexual abuse

victims who are diagnosed with sexually transmitted dis-

eases, who are too old to have acquired the diseases con-

genitally and too young to have acquired them through

consensual sex with peers, also avoid both suspicion

and substantiation bias. Confining this review to STD

diagnosis deals with the problem raised by London et al.

(2005) that “medical evidence” is not always a “reliable

benchmark” because, for example, genital redness may be

caused by many things besides sexual abuse.

Lawson and Chaffin (1992) found that among 28 chil-

dren in which STDs were medically diagnosed without

prior suspicion of abuse, only 12 children (43%) made

an allegation of sexual abuse during the initial formal

interview, and 16 children did not. Almost half of these

children had shown no physical or behavioral symptoms

of sexual abuse, so that there were no “red flags” that

would have otherwise brought these children into the

system as possible CSA victims. Maternal attitude influ-

enced disclosure patterns greatly. Among those children

whose parents were supportive, 63% disclosed abuse

during these initial interviews, whereas when caregivers

expressed skepticism, only 17% disclosed.

Of the 16 false negatives in the original Lawson

and Chaffin study, five were subsequently located and

consented to be interviewed. Four of these five had a

supportive parent and one a non-supportive parent.

Researchers presented the study to parents and chil-

dren as an evaluation of responses to prior emergency

room visits, and they never mentioned child abuse.

Nevertheless, four of the five parents spontaneously

told the researchers that their children had disclosed

sexual abuse some time after the initial hospital inter-

view, a finding that supports the idea that CSA disclo-

sure is often an incremental process that may require

more than a single interview (Chaffin, Lawson, Selby,

& Wherry, 1997). Upon psychological testing, the four

non-disclosing children whose parents had been sup-

portive at the time of the initial interview tested three

times higher on dissociative symptoms than did the dis-

closing children and nine times higher on dissociative

symptoms than non-abused control children. Because of

the nature of this study and the very small numbers of

children involved, these results are far from conclusive,

but they do suggest a possible link between dissociative

symptoms and non-disclosure among CSA victims.

London et al. explain the Lawson and Chaffin

results by describing this sample as “unusual” and as

representing “the small hard core of children who do

not disclose abuse when directly asked” (2005, p. 215).

Lyon argues in response that the Lawson and Chaffin

sample avoids the problems of suspicion and substan-

tiation bias that characterize many other samples. Lyon

then raises a concern about the many cases that are

closed as unsubstantiated after a single interview during

which a possibly sexually abused child without medical

evidence fails to disclose when formally questioned.

A number of other samples document similarly

low rates of disclosure in STD cases. Lyon examined 21

studies published between 1965 and 1993 of children

diagnosed with gonorrhea. In nine of these papers, the

authors referred to a “history” of sexual contact or sexual

abuse for some of the children with gonorrhea, without

clarifying whether this history came from children’s dis-

closures or from other sources (Lyon, in press). In most

of the remaining studies, the authors used words such as

“admitted” or “denied” sexual contact or referred even

more directly to children’s statements. Even when all

the cases of “history” were counted as actual child dis-

closures, Lyon finds that the average rate of “disclosure”

among the 579 children in these studies was 43%, or 250

children. Given the broad definition of “disclosure” that

he applies here, Lyon argues that this may actually be an

overestimate of disclosure rates. Most of these studies

indicated that the medical professionals questioned the

children, but the precise nature of these questions is not

known. When Lyon omits studies with children younger

than three years of age to control for developmental

limitations on narrative skill, he finds that 185 of 437

children, or 42%, disclosed.

To summarize this sample of disclosure studies that

avoid both suspicion and substantiation bias, Sjoberg

and Lindblad find a disclosure rate of 50%, Lawson and

Chaffin find a disclosure rate of 43%, and in a review of

21 studies of children diagnosed with gonorrhea, Lyon

finds a disclosure rate of 43%. London et al. (2005) assert

E r …