2
Disclosu,re of Child
Sexual Abuse:
A Review of the
Contemporary Empirical
Literature
Kamala London, Maggie Bruck
University of Toledo

Stephen J. Ceci
Cornell University

Daniel W. Shuman
Southern Methodist University

Although it is widely acknowledged that the sexual assault of children is a
major societal concern, it is not known how many children in the United
States are victims of sexual abuse (Ceci & Friedman, 2000). There are two
major reasons for this lack of data. First, current estimates of the incidence of
child sexual abuse (CSA) do not reflect the number of unreported cases or the
number of cases reported to agencies other than Child Protective Services
(e.g., sheriff's offices or professionals such as mental health diversion programs). Second, the diagnosis of CSA is often difficult because definitive medical or physical evidence is lacking or inconclusive in lhe vast majority of cases
(Bays & Chadwick, 1993; Berenson, Heger, & Andrews, 1991), and because
there are no unique psychological symptoms specific to sexual abuse (KendallTacket, Williams, & Finkelhor, 1993; see Poole & Lindsay, 1998; ]. Wood &
Wright, 1995). Given these limitations of medical and psychological evidence,
children's statements typically comprise the central evidence for judging the

II

 -.--·
-------------

12

LONDON ET AL.

occurrence of CSA In making these judgments, professionals often must address the thorny issue concerning how children disclose abuse.
According to some experts, a major problem with relying on children's
statements in forensic investigations is that many sexually abused children remain silent about abuse, they deny that abuse ever occurred, or they produce
a series of disclosures of abuse followed by recantations of these disclosures.
This hypothesized model of incremental abuse disclosure has been put forth by
a number of authors (e.g., Macfarlane & Krebs, 1986; Sgroi, Bunk, & Wabrek,
1988; Summit, 1983). The central tenet of these models is that sexually abused
children are reluctant to disclose sexual abuse. When children do reveal abuse,
disclosure will be incremental over time, a process that often includes outright
denials and recantations of prior disclosures and then reinstatements of the
abuse. Although this hypothesized pattern of incremental abuse disclosure
has taken on different names, we will refer to it as Child Sexual Abuse
Accommodation (CSM).
The CSM model has been endorsed by many clinicians and scholars and
has been the basis of clinical and forensic judgments. For example, Summit's
(1983) paper describing a model of incremental abuse disclosure was rated by
professionals as one of the most influential papers in the field of child sexual
abuse (Oates & Donnelly, 1997). The central assumptions of the CSM model
are echoed in some contemporary guidelines for assessment and diagnosis of
CSA For example, the National Children's Advocacy Center (Carnes, 2000)
states in one of its publications, "Forensic evaluation is a process of extended

assessment of a child when that child is too frightened or young to be able to
fully disclose their experiences on an initial forensic interview" (p. 14) and
"Reluctance is commonplace and difficult to overcome in suspected child sexual abuse cases" (p. 42). In fact, some professionals have gone so far as to rreat
behaviors such as denials and recantations of abuse as indicators that sexual

abuse likely occurred (Fisher, 1995; Kovera & Borgida, 1998; Robin, 1991;
Summit, 1992).
Given the widespread appeal and currency of CSM-type models in the
mental health community and their acceptance in the forensic arena, especially when used to rehabilitate the credibility of a child witness, it is important
to examine the empirical basis for these theories (Summit, 1983, p. 180; Summit, 1992, p. 156; and see London, Bruck, Ceci, & Shuman, 2005, for extended discussion). In this papei; we critically review the existing empirical
data to assess the scientific support for the behavioral components of CSMtype models-secrecy/silence, denial, and recantation. We draw on two major
sources of empirical data on children's disclosure patterns: (1) retrospective
accounts from adults who claimed to have been sexually abused as children,
and (2) examinations of children undergoing sexual abuse evaluations. Each
source contributes some information to our understanding of CSA disclosure
patterns although both data sources contain limitations. To foreshadow the results of this review, we conclude that although a substantial proportion of children delay reporting or altogether fail to report incidents of child sexual abuse

2. DISCLOSURE OF CHILD SEXUAL ABUSE

13

(the secrecy stage), there is little evidence to suggest that denials, recantations, and redisclosures are .typical when abused children are directly asked
about abuse dunng forensic interviews.
PATTERNS OF SEXUAL ABUSE DISCLOSURE AMONG
ADULTS IN RETROSPECTIVE SURVEYS

Disclosure Rates
We conducted a review of the literature to locate studies that yielded statistics
on the proportion of adults with reported childhood histories of abuse who disclosed thelf abuse during childhood (London, et al., 2005). We selected studies that were .published since 1990 in order to control for potential cohort
effects that might be present due to the many changes in education advocac
11
mcrea~ed se~sit~vity, and legal procedures related to sexual abuse ~nd sexuai
ab~se mvest1gauons. Table 2.1 lists 11 retrospective studies that fit these critena'. Studies that did not specify the rate of disclosure during childhood are
not hsted m the table but are cited when relevant for related topics (for example, predictors of disclosure patterns).
Before. beginning this review, it is important to point out that there are
several d1ff1culties m interpreting the findings of this literature. The first issue
regardmg a.c'.'uracy d the retrospective data lies in accuracy of the abuse reports. Specifically, It is possible that some adults in these retrospective studies
had been abused but continued to deny abuse (which would work to reduce
the overall c.sA prevalence rates and inflate the disclosure rates; see Lyon,
chapter 3'. this volume). Additionally, it is possible that some adults in these
retrospecttve st~1d1es had not been abused but claimed to have been (e.g., due
to false memones encouraged by suggestive psychological therapeutic practices). Such false allegations would inflate the incidence of CSA and render
the data on disclosure nonmeaningful. A second issue surrounding the accuracy of the retrospective data lies in the accuracy of adults' disclosure reports·
that is, whether and when they reported the abuse to others as children. Som~
adults might have disclosed abuse in childhood, despite their reports to the
contrary. In other words, they may in fact have told someone but failed to remember. having done so. A rich cognitive psychology literature demonstrates
the mynad retrospective biases, even when the events in question are highly
emotional (e.g., Freyd, 1996; Read & Lindsay, 1997· Neisser 1997· Ross
1989). In their investigation of flashbulb memories, S~hooler a~d colleague~
(Schooler, Ambadar, & Bendiksen, 1997; Schooler, Bendiksen, & Ambadar,
1997) comed the term forgot-it-all-along-effect to describe the finding that people. sometimes.inaccurately rec.all to whom, when, and whether they reported
an unportant hfe event. Adults denial of CSA reports that were actually made
during childhood would not affect prevalence rates of CSA but would lead to
an underestimation of disclosure rates. Despite these many confounding

 TABLE2.l
Childhood Disclosures of Sexual Abuse: Retrospective Studies

!"

Reports abuse
at survey
Study
Arata (1998)

N=

Sample Sourcea

Definition of CSA

(%)

860b (f)'

College sample

Unwanted contact
before 14 yrs

24

3,220 (f)

probability
sample

286 (f)

CSA hotline callers

31 (at time
of abuse)

10

8.5

23

34 (within

National
Smith et al. (2000)

Childhood Report to Ave. age at Ave. age
disclosure authorities
time of
of sample
(yrs)
(%)
(%)
abuse

6 months

9

of abuse)

12

10.9

45

before 16 yrs

100

36

n/a

6

41

Not specified

100

36 (by age 13)

n/a

8.15

30

be~ore 16 yrs

100

37

n/a

< 16

41

Unwanted sexual
experiences before
age 18

6

38

n/a

11

21

Rape
Intra~familial

Roesler & Wind (1994)
Lamb & Edgar·
Smith (1994)

48(£) 12 (m)" Newspaper ad

Roeslet (1994)

168 (f); 20(m) Abuse center

Genital contact ·
Hong Kong
Chinese college

Tang (2002)

1,151 (£)
887 (m)

Finkelhor et al. (1990)

1,481 (f)
National
1,145 (m) probability sample

Somer & Szwarcberg (2001)

41 (f)

students

42 within
Before 18 yrs

Israeli abuse center

27 (£) 16 (m) 1 year of abuse

CSA survivors

100

45 (by age 17)

n/a
n/a

Median= 9.7 30-39

7.11

32
(continued)

TABLE 2.1 (Continued)

Reports abuse
at survey

ChiUlhood Report to Ave. age at Ave.Age
disclosure Authorities time of
of Sample
(%)
(%)
abuse
(yrs)

Study

N=

Sample Source~

Definition of CSA

(%)

Ussher & Dewberry (1995)

775 (£)

1991 Women's
Magazine Survey

Unwanted sexual
attention

100

54

18

8.5

38

10

87 (by age 18)

n/a

< 16

18

8.5

n/a

13

< 18

37.5

Fergusson et al. (1996)

Hanson et al. (1999)

New Zealand
Unwanted experience
1,019 (m&f) longitudinal study
before 16 yrs

4,008(£)

National
probability
sample

'

\

Nonconsensual
penetration
assaults before age 18

a. Unless noted, all studies were conducted in the United States.
b. Some studies sampled general populations while others specifically sampled sexually abused. We have provided in this table the percentage of abused participants endorsing abuse. Hence, data on disclosure are based on the percentage of the total samples that endorsed CSA.
c. cFemale.
d. llMale.

Jl

 16

LONDON ET AL.

2. DISCLOSURE OF CHILD SEXUAL ABUSE

17

In addition to providing overall disclosure rates, some studies examined pre-

Some authors have posited that CSA disclosure may be mediated by the
nature of the relationship between the victims and perpetrators. Two research
groups (Hanson, Resnick, Saunders, Kilpatrick, & Best, 1999; Smith, Letourneau, Saunders, Kilpatrick, Resnick, & Best, 2000) reported that CSA
disclosure was more likely when the perpertator was a stranger rather than a
family member. Consistent with these findings, Ussher and Dewberry (1995)
reported longer delays to disclosure among intra- versus nonfamilial abuse. In
contrast, five research groups found no association between victim-perpetra . .
tor relationship and CSA disclosure (Arata, 1998; Kellogg & Hoffman, 1995;
Kellogg & Huston, 1995; Lamb & Edgar-Smith, 1994; Roesler, 1994). Taken
together, at least when there is an association between disclosure and relationship to perpetrator, close relationships lead to decreased disclosure. It is
impossible to know at this point why only some of the studies have found this
association, and there are seryeral possible explanations. The association be . .
tween disclosure and relationship to perpetrator may not be robust. Alternatively, some methodological factors might be suppressing the effect: The
samples may not allow adequate detection of the relationship due to sample
homogeneity on this variable; or there may be other abuse-related variables
that suppress the effect of relationship to perpetrator. The retrospective data
are insufficient at this time, then, to conclude whether there is a consistent as.sociation between relationship to perpetrator and childhood disclosure of sexual abuse.
Next, we examined the retrospective studies for trends in age at time of
abuse and abuse disclosure. Age at time of abuse has not consistently been associated with abuse disclosure. Although Smith et al. (2000) found that
younger victims were more likely to delay disclosure than older child victims,
other researchers (e.g., Arata, 1998; Kellogg & Hoffman, 1995) failed to find
any relationship between age and delay of disclosure. There is one important
caveat to this conclusion. When study subjects reported experiencing CSA
during adolescence, this was consistently accompanied by high disclosure rates
(Everill & Waller, 1995; Kellogg & Hoffman, 1995). For example, Kellogg and
Huston found that 85% of their sample of young adults (mean current age =
19.5 years, mean age of abuse = 14 years) had disclosed at some point in the
past. Adolescents most commonly disclosed to a peer (Lamb & Edgar-Smith,
1994; Tang, 2002). In contrast, adults reporting that they revealed CSA as
school-aged children did so to a parent rather than to a peer (Arata, 1998;
Lamb & Edgar-Smith, 1994; Palmer, Brown, Rae-Grant, & Loughlin, 1999;
Roesler, 1994; Roesler & Wind, 1994; but see Somer & Szwarcberg, 2001;
Smith et al., 2000). These studies, taken together, imply that disclosure rates

dictors of disclosure rates. That is, what kinds of abuse, victim, and perpetra . .

may vary as a function of age of CSA onset, which in turn is associated with

tor characteristics emerge in this literature that might help explain some of the
variance in CSA disclosure during childhood? In this section we examine associations of some of these predictors from data within studies and, when possible, across studies.

the availability of a same-aged confidante.
No systematic relationships have been reported between demographic
variables such as race and ethnicity and childhood disclosure rates (e.g., Arata,
1998; Hanson et al., 1999; Kellogg & Hoffman, 1995; Kellogg & Huston,

variables in the interpretation of the existing data, some common themes
emerge across studies, as we discuss below.
As shown in table 2.1, the modal childhood disclosure rate (in 6 out of 10
studies) is just over 33%. Three other studies (7, 8, 9) reported slightly higher,
but still low, rates of disclosure. These low disclosure rates are consistent with
the claims of the CSM model that nondisclosure of sexual abuse (silence) in
childhood is very common. Only 1 of the 11 studies in table 2.1 reported relatively high rates of disclosure. The study was carried out in New Zealand by
Fergusson, Lynskey, and Horwood (1996) and involved a longitudinal study on
1,265 children. Sexual abuse was defined broadly in their study, ranging from
noncontact activities such as indecent exposure or lewd suggestions (including experiences with same-aged peers) to rape before age 16. At 18 years old,
87% of the abused subsample reported that they had told someone about the
abuse. There are several reasons wl1y Fergusson and colleagues obtained
higher disclosure rates than other researchers. Fergusson et al. noted the hlgh
disclosure rates may partially reflect the young age of the adults in their sample: Possibly some were still denying the abuse, thus producing lower rates of
CSA with concomitantly inflated rates of disclosure. Another possibility is
that the high disclosure rates could be driven by unknown victim, perpetrator,
or abuse variables. For example, the high rates of disclosure could; in part, be
an artifact of their abuse definition. Many of their subjects reported noncontact activities such as lewd suggestions, which many of the participants reportedly did not consider as incidents of CSA. This could also explain why
many of Fergusson et al.'s (1996) subjects denied abuse history during a three
year follow up interview (Fergusson, Horwood, & Woodward, 2000).
·In summary, although one study yielded bigh disclosure rates (Fergusson
et al., 1996), the other retrospective studies indicated that just over one-third
of adults who suffered CSA appear to reveal the abuse to anyone during childhood. Furthermore, among children who do disclose during childhood, delay
of disclosure is common (also see Kellogg & Hoffman, 1995; Kellogg & Huston, 1995). As shown in column 7 of table 2.1, an even smaller minority reported that the abuse was disclosed to authorities during childhood. Given the
differences in methodology, definitions of abuse, and sample characteristics,
the general consistency of these findings across these studies is noteworthy.

Predictors of Nondisclosure/Disclosure

 18

LONDON ET AL.

1995· Smith et al., 2000), However, most of the retrospective studies have too
little'variability in their sample's demographic composition to test for differences. (For discussions on how demographic variables-race and .gendermight relate to CSA disclosure see Fontes, 1993; Kazarian & Kazanan, 1998;
Kenny &McEachern, 2000; Levesque, 1994; Toukmanian &Brouwers.' 1998.)
We examined the existing data to assess support for the assumpt10n that
disclosure is related to the amount of fear or violence associated with the
abuse. According to the assumption, children do not disclose because they fear
the perpetrators who physically coerced, harmed, or threatened them. In general the data do not support the hypothesis that disclosure rates are related to
sev~rity of abuse. Although Arata (1998) found lower disclosure rates for contact versus noncontact abuse, she found no relationship between disclosure
and method of coercion (e.g., threat, gift; curiosity, appeal to authority, physical force). Most researchers, however, have either found the opposite pattern
(i e higher disclosure rates are associated with incidents involving life threat
physical injury; Kellogg & Hoffman, 1995; Hanson et al., 1999), or they
have found no significant relationship between seventy or method of coercion
and disclosure (Lamb & Edgar Smith, 1994; Roesler, 1994; Smith et aL, 2000).
Another way to examine the relationship between seventy/coerc10n/phy.sical harm and disclosure is to compare the rates among studies in Table 2.1 rn
terms of the types of abuse that were included in the study. Some experimenters defined CSA broadly (unwanted sexual attention by anyone) and
some defined it more narrowly (e.g., forcible penetration). Despite the differences in definitions, disclosure rates reported across studies were very s1m1lar,
except in Fergusson et al.'s (1996) study.
.
.
Next we searched for studies that examined the relat10nship between
threats that were used to secure the child's silence ("Don't tell or else··.") and
disclosure. We only found a few such studies, but a common problem ';as tha~
in many cases it is not possible to determine whether the measure of threat
referred to statements or actions during the commission of the assault to engender physical compliance or to threats used to engender silence (e.g., s~e
Arata, 1998; Hanson et al., 1999; Roesler, 1994; Smith et al., 2000). This failure to provide operational definitions of threats is problematic on. methodological grounds (how did the study participant interpret the questl?n?) and
on interpretational grounds (how does the consumer of the luerature interpret
the statistics?). Hence, the extant retrospective data are not informative as tq
the nature of the association between threats to remain silent and childhood
disclosure rates. Future studies are needed that clearly define the term
"threaten" to participants and to readers.

,,;,cl,

summary
The results of the retrospective studies make two important contributions to
our knowledge about the patterns of children's disclosure of abuse. First, when
taken at face value, these data reveal that approximately 60-70% of adults do

2. DISCLOSURE OF CHIW SEXUAL ABUSE

19

not recall ever disclosing their abuse as children and that only a small minority of subjects (10-18%) recalled that their cases were reported to the authorities. Thus, the retrospective studies provide evidence to support the
assumption that many incidents of CSA go unreported and that the silence
component in CSAAS models has a strong empirical foundation.
Second, analyses of predictor variables in these retrospective studies provide few insights into the factors associated with disclosure. They do, however,
suggest that commonly held assumptions, such as fewer disclosures among
more severe cases of CSA or in cases of intrafamilial abuse, lack adequate empirical support. We must await further data to examine these issues more definitively. In general, the reviewed studies were not designed to examine
specific predictors of disclosure, and most of the analyses were post hoc in nature. Generally, the retrospective literature is limited in detecting associations
between abuse variables and disclosure due to factors such as insufficient sam~

pie size, lack of homogeneity of samples, varying definitions of the variables,
and a failure to provide clear operational definitions of the variables. Given
these limitations, it we conclude that the existing data are inconsistent and
these issues remain relatively unexplored.
Finally, it is important to stress constraints on the generalizability of these
findings. Although these studies yield data on disclosure rates, they do not provide any data on the two other main components of the CSAA models: the frequency of denials or of recantations of abuse. This is because the surveys in
these retrospective studies did not contain items that asked subjects if they
were directly asked about abuse during childhood. Thus, in terms of data on
denial of abuse, it is not known from the retrospective data if the nondisclosers
were asked about and denied abuse or if the nondisclosers were simply not
asked. In order to examine trends in abuse denials and recantations, the liter~

ature on children's patterns of CSA disclosure must be examined.
PATTERNS OF DISCLOSURE AMONG CHILDREN
TREATED OR EVALUATED FOR SEXUAL ABUSE

In this section, we review studies of disclosure patterns of children who were
specifically assessed or treated for sexual abuse in terms of delay of disclosure,
denial, and recantation. Each section also includes a discussion of the correlates of delay, denial, and recantation. As with the retrospective studies, we excluded studies published prior to 1990 because of possible cohort effects that
could be due to the changes in interviewing practices and prevention programs
(for children) that have occurred in the decade of the 1990s.
Most of the studies presented in this section involved "ch.art reviews" of

children who were interviewed by CPS, mental health, or medical professionals specializing in the assessment and treatment of sexual abuse. Children presented at these clinics or centers for a variety of reasons that included a prior
disclosure to an adult, a suspicion of abuse by an adult or an agency, or the

 20

LONDON ET AL.

need for a second opinion or more extensive interviewing. Thus, across and
within studies, there is often great variability in the methods by which children
were interviewed, in the information collected (including differences and
sometimes ambiguity in operational definitions of abuse-related terms such as
threats), and in the procedures of diagnosing child sexual abuse. Furthermore,
in some studies, as will be noted, researchers categorized the children according to the likelihood of abuse (e.g., highly probable, unclear, not abused); in
other studies only children who met some prespecified criteria for abuse were
included, and in still other studies, the certainty of abuse status was not specified. Of course, as with the adult retrospective reports, the diagnosis of sexual
abuse whether as substantiation or unfounded, almost always comes with
some 'degree of uncertainty. Some children may falsely claim to have been
abused after undergoing suggestive, coerciVe interviewing from a biased inves~
tigator or because they were pressured to do so from a parent undergoing a
messy divorce; alternatively, some children may falsely deny abuse because of
a variety of reasons including pressure from a parent to do so. Substantiation
of abuse is a thorny issue, and we return to this issue to discuss its impact on
the disclosure data in a subsequent section. As with the retrospective studies,
then, the data from the child samples are not without limitations; some common themes do, nonetheless, emerge across studies.

Delay of Disclosure (Silence)
The results of these studies on child samples echo the adult retrospective finding regarding delay of abuse disclosure; namely, when children do disclose, it
often takes them a long time to do so. For example, disclosure rates of children
whose cases were referred for prosecution were examined by Goodman, Taub,
Jones, England et al. (1992) and by Sas and Cunningham (1995) ..Although
37-42% of the children had disclosed within 48 hours of the abuse, it took between 6 to 12 months for many of the children to disclose abuse. Even higher
rates of delayed disclosure were obtained in Elliott and Briere's (1994) swdy,
in which 75% of children did not disclose CSA within the first year followmg
the abuse, and 18% waited more than five years to disclose the abuse. Similarly, Henry (1997) reported an average of two years delay between abuse and
disclosure.
Some of the variability in the length of delay in the child studies may reflect the settings in which the data were collected. Shorter periods of delay may
show up in surveys of children in criminal trials simply because case.s that do not
show this pattern are excluded from consideration because of the mherent difficulty in obtaining convicrions. Therefore, it may be that cases in the prosecutor'~ office are unrepresentative of those that never reach the courtroom.

Few of the studies on delay of disclosure examine associations between
abuse-related variables and latency to disclosure. There are some data on
gender differences, suggesting that males may be more reluctant W disclose
than females (e.g., DeVoe & Faller, 1999; Goodman-Brown, Edelstem, Good-

2. DISCLOSURE OF CHILD SEXUAL ABUSE

21

man, Jones, & Gordon, 2003; Gries, Goh, & Cavanaugh, 1996; Sas and Cunningham, 1995; Stroud, Martens, & Barker, 2000; but see DiPietro, Runyan,
& Fredrickson, 1997; Keary & Fitzpatrick, 1994, who report no gender differences). However, as Goodman-Brown et al. (2003) discuss, gender differences
in disclosure rates may be suppressed by other variables associated with gender
(e.g., prior disclosure, relationship to perpetrator). For example, males may be
more frequently abused by nonfamily members than females. Hence, potential
differences in abuse dynamics between boys and girls should be kept in mind
when considering any potential gender differences in disclosure patterns. Additionally, there are reasons to suspect that members of certain ethnic groups
might face additional and culture-specific barriers to CSA disclosure (see
Futa, Hsu, &Hansen, 2001; Rao, DiClemente, &Ponton, 1992; Shaw, Lewis,
Loeb, Rosado, & Rodriguez, 2001; Toukmanian & Brouwers, 1998; Wong,
1987). However, to date there is no coherent account of the effects of demographic variables on abuse discl7isure in childhood.
Some researchers have examined the association between different abuse
characteristics and disclosure. As was the case with the adult studies, the data
on "threats" are difficult to interpret because most studies fail to provide operational definition of threats (i.e., Were threats used during abuse or after
abuse? Were threats used to obtain compliance or silence?). When clearly defined data on abuse characteristics do exist, they are sparse and not very consistent. For example, Sas and Cunningham (1995) found that children waited
longer to disclose abuse when the perpetrator "groomed" them and established
a close relationship than if the perpetrator used force. Some researchers have
found that children who are victims of familial abuse tend to delay disclosure
longer than those experiencing extra familial abuse (DiPietro et al., 1997;
Goodman-Brown et al., 2003; Sjoberg & Lindblad, 2002). However, the majority of studies we examined either failed to find such an association or failed
to report an association. Note that none of the studies covered in this section
address issues concerning denial. These are addressed in the next section.

Rates of Sexual Abuse Disclosure (or Denials) in Interviews
With Child Samples
In this section we review 17 papers published since 1990 that contained statistics on the frequency of denial and/or recantations for children who were
questioned about abuse. Table 2.2 lists the 17 studies that provided childhood
disclosure rates, in ascending order of denials. Additionally, table 2.2 provides
data on some of the central characteristics of the studies. When relevant, we
cite other studies in this section that did not provide data on the rate of
childhood disclosure in their sample but that do throw light on the correlates
of disclosure.
The pooled mean of disclosures for the studies listed in table 2.2 is 64%
(range 24-96%), or, put another way, the mean of denials is 36%. For reasons
discussed below, however, these figures should not be viewed as the best

 2.

DISCLOSURE OF CHILD SEXUAL ABUSE

23

estimate of central tendency. We focus on four factors that may account for the

enormous between-study variability in order to highlight methodological and
design factors that need to be considered in evaluating the generalizability, validity, and reliability of the findings in table 2.2. These factors are age of the
child, previous disclosure of abuse, substantiation of abuse, and representative
nature of the selected sample. We conclude that when such factors are considered, mean disclosure rates are quite high when children are explicitly asked
about sexual abuse.
Developmental Differences. The wide variation in the ages of the children
both within and between studies (see table 2.2, column 3) could partially account for differences in the rates of disclosure across studies. In order to examine this hypothesis, age-denial associations were examined within studies.
Although no significant relationships between age and denial were found in
two studies (Bradley & Wood, 1996; DeVoe & Faller, 1999), the more common finding was that school-aged children were more likely than preschoolers
to disclose abuse during formal evaluation. For example, B. Wood, Orsak,
Murphy, and Cross (1996) found that older children made more credible disclosures of abuse than younger children (also see Goodman-Brown et al.,
1
2003; Mordock, 1996). Similarly, DiPietro et al. (1997) found that older children were more likely to disclose than younger children and that children generally became more likely to disclose abuse after age four (also see Candon,
Payne, & Erbaugh, 1996; Gries, Goh, & Cavanaugh, 1996; Keary and Fitzpatrick, 1994; Sas & Cunningham, 1995).
There are several possible explanations for these developmental differences in children's abuse disclosures. They could reflect the single influence or
combined influences of linguistic, cognitive, and social/emotional factors.
Thus, younger children may not have the same linguistic skills to convey their
abuse experience; or, younger children may not understand the "meaning" of

abusive acts and thus fail to make explicit disclosures. In support of these positions, it has been found that younger children are more likely to make accidental disclosures whereas older children are more likely to make purposeful
disclosures (Campis, Hebden-Curtis, & DeMaso, 1993; Fontanella, Harrington, & Zuravin 2000; Nagel, Putnam, Noll, & Trickett, 1997). That is,
younger children are more likely to make spontaneous statements about abuse
that are not consistent with the topic of conversation or of the ongoing activity (e.g., stating, while watching TY, "Uncle Bob hurt my bottom"). In contrast,
older children are more likely to report the abuse to an adult when asked.

1

B. Wood et al. (1996) defined a credible disclosure as one that ''was adequate for use

as evidence in a future legal and/or child protection proceedings" (p. 84). The not credible
category included cases "where the child did not disclose, denied sexual abuse, refused to
cooperate, provided insufficient in detail or was not believable" (p. 84). The authors did not
cite the number of children falling into each of the not credible subcategories.

22

 24

LONDON ET AL.

Although the conclusions are consistent across studies, it is important to point

out that the ages of the "younger" and "older" children are not the same across
studies. Thus, in one study the "younger group'' might be of the same age as
the "older group" in another study. Therefore, it is not clear at what age children become more likely to disclose abuse.
A second possible explanation for developmental differences in rates of
denial is that there may be higher rates of true denials among younger than
older children. This hypothesis is based on several interrelated findings.
Younger children may be more likely than older children to be brought for as·
sessment due to caregivers' concerns about behaviors (rather than an abuse
disclosure) that often are ambiguous and do not necessarily reflect CSA (see
Campis et al., 1992; Fontanella et al., 2000; Levy, Markovic, Kalinowski,
Ahart, & Torres, 1995; Nagel et al., 1997). Thus, in any sample there may be
a greater proportion of younger nonabused children than of older nonabused

children, and the higher denial rates by younger children would then reflect a
higher rate of denials that are true negatives. For example, Keary and Fitz·
patrick (1994) were less likely to categorize younger children as sexually
abused compared to older children; in addition, the younger children were less
likely to disclose abuse. Unfortunately, these researchers did not present data
on age differences in denial rates among children who were classified as
"founded" by the assessment team.

Although most of the data indicate that younger children may be less
likely to disclose than older children, upon closer investigation, there may also
be patterns specific to adolescents. At least among cases that reach authorities,
children are most apt to reveal the abuse to their primary caregiver (Berliner
& Conte, 1990; Campis et al., 1992; Faller & Henry, 2000; Fontanella et al.,
2000; Gray, 1993; Henry, 1997; Sas & Cunningham, 1993). However, adoles·
cents may have a greater appreciation of the consequences of disclosing in·
trafamilial abuse and thus withhold information. It is also possible that they
may not readily disclose extrafamilial abuse to family members or to investigators because they feel it is their own business or they have already disclosed to
peers, as noted in the retrospective studies reviewed in the first part of this
paper. Hence, the rate of CSA disclosure to parents and authorities may re·
semble an inverted u. .pattern, with an increase in disclosure as one moves
from preschoolers to school-aged children, followed by an apparent decrease
as one moves into adolescence. There are, however, few data on disclosure pat. .
terns in adolescence, and we must await these before drawing any definite

2. DISCLOSURE OF CHILD SEXUAL ABUSE

25

there was a suspicion of abuse that required further investigation. Thus, most

of the children in each study had been questioned by someone (e.g., teacher,
parent) about abuse prior to the formal interviews or therapy sessions. This
factor is important because, as shown in table 2.3, the most significant predictor of disclosure in the formal interview is whether the child had disclosed before (e.g., to a parent, a teacher, a CPS worker). For example, Keary and
Fitzpatrick (1994) reported that of the 123 children who had made a prior disclosure, 86% disclosed again during the formal interview; in contrast only 14%
of the 128 children with no prior disclosures disclosed at interview. 2 Similar
patterns of results were found by Gries et al., (1996), DiPietro et al. (1997),
and DeVoe and Faller (1999).
This pattern of consistency of disclosure is most common in older children. Among children who had disclosed prior to formal assessment, younger
children were less likely than older children to disclose again during formal assessment (Keary &Fitzpatrick, 1994; also see Ghetti, Goodman, Eisen, Qin, &
Davis, 2002).
In sum, several studies suggest that once children have made an abuse dis·
closure, they are likely to maintain their allegations during formal assessments.
This finding suggests that if children have already told a professional or a care·
taker about an abusive event, then they are more likely to disclose in a formal
investigation. Discrepant cases (where a child discloses before the formal in·
terview but denies at the time of the formal interview) represent a minority
and appear to occur most commonly among very young children.

conclusions.

Abuse Substantiation. The third and perhaps most important methodological factor that accounts for variation in disclosure patterns across studies
concerns the validity of the diagnosis of child sexual abuse. In conducting
studies of CSA disclosure patterns, it is of utmost importance to ensure that
the group under study had in fact experienced childhood sexual abuse; otherwise, counts of frequency of delay to disclosure, denials, recantations, and restatements are uninterpretable. That is, children may deny because they in
fact never were abused; children may take a long time to disclose because it is
only with repeated suggestive interviewing that they will make disclosures
which are false; and children may recant in order to correct their prior false
disclosures.
In order to address problems of substantiation of abuse, some researchers
have classified children in their sample in terms of the likelihood of abuse.
Children meeting one or more of the following criteria (depending on the

Prior Disclosure of Abuse Predicts Disclosure During Formal Assessment.
The studies included in table 2.2 focus on children's reports during forensic
interviews and psychotherapy. That is, the children in these studies were
specifically brought to a clinic, mental health professional, or law enforcement
agency either because they had previously made a claim of abuse or because

When children have made a prior allegation but do not repeat it during a formal in~
vestigation, thi.s should not be categorized as a recantation because it is possible that the
child's first allegation was incorrect or misinterpreted and the report during the formal in~
vestigation is accurate. In this paper, recantations are defined as those that are made to the
same assessment team who heard the disclosure.
2

 26

LONDON ET AL.
TABLE2.3
Rates of Prior Disclosure and Disclosure during Formal Interviews

Prior Disclosure (%)

Disclosure During Assessment
Study

Yes

No

Devoe & Faller (1999)
DiPietro et al. (1997)
Keary & Fitzpatrick (1994)
Gries et al. (1996)

74
77
86
93

25
7
14
40

study) are classified as abused: perpetrator convictions, plea bargains or confessions, medical evidence, other physical evidence, and children's statements.

Although the use of such criteria is a good start, it should be noted that there
are problems with each of these criteria. First, the accused may be persuaded
to accept a plea bargain due to the stress, financial burden, and uncertain outcome of facing trial. There are some accused who have been falsely convicted
despite the absence of direct evidence to prove child abuse, and on appeal
their convictions have been overturned (Ceci & Bruck, 1995). Although this
may not be common, it does happen. Next, medical evidence is not always an

accurate indicator of abuse. In the statistically rare case where genital or anal
abnormaliries are found, similar abnormalities can somerimes be found among
nonabused children (Berenson et al., 1991). Finally, in terms of the studies that
are included in this report, the children's statements at the time of formal interview are used as indicators of abuse: But this is a circular exercise whereby

children who make spontaneous disclosures with much elaborarion, for example, are categorized in the "high certainty" group. Then the analysis of the disclosure patterns of the high certainty group indicates that the children
disclosed spontaneously and/or with much elaborarion (or did not deny).
Notwithstanding these problems with the use of certainty criteria, there
must be some reliable bases to categorize the children in studies of disclosure
of CSA, lest the disclosure rates obtained merely reflect the overall responses
of children (abused and nonabused alike) who are assessed for sexual abuse.
Keeping these reservations in mind, we now review those studies that have ex~

amined disclosure patterns as a function of the certainty of abuse diagnosis. We
will argue that with a few exceptions, high rates of disclosure rates are obtained
in studies where the abuse status of the child is well defined, and low disclosure rates are associated with samples where the diagnoses of abuse are either
unknown or questionable.
Referring to table 2.2, the eight studies with the highest disclosure rates
(96% to 76%) contain the statements of children with high certainty diagnoses
of abuse (sometimes these cases are labeled "substantiated," and, in general,
the researchers or clinicians considered it highly likely that the children had
been abused). The rates of disclosure are greatly lowered in these same studies

2.

DISCLOSURE OF CHILD SEXUAL ABUSE

27

when the data from the unsubstantiated or unclear cases are averaged with the

high-certainty cases (see data in parentheses in column 4). Thus, although
only 62% of DeVoe and Faller's (1999) sample of 5- to IO-year-olds disclosed
abuse, when only substantiated cases are included, the disclosure rate rises to
87%. The overall rate of disclosure in the Keary and Fitzpatrick study (1994)
is 50%; however, when only the substantiated cases are included, the rate is
95%. DiPietro et al. (1997) classified each of the children in their sample who
were assessed because of suspicions of CSA as unfounded, possible, probable,
or definitive abuse. Rates of disclosure during the first visit increased as a function of abuse certainty with 7%, 8%, 59%, and 76%, of cases classified as unfounded, possible, probable, or definitive abuse, respectively, disclosing abuse.
The overall disclosure rate in Dubowitz, Black, and Harrington (1992) was
58%; however, among their case§ rated by an interdisciplinary team as holding
low to possible likelihood, the disclosure rate was only 19%, compared to disclosure rates of 83% for the moderate to high likelihood cases and 75% of cases
with abnormal medical findings seen as indicative of abuse (e.g., abnormal anal
or genital findings). Elliott and Briere (1994) examined the case records of399
8- to 15-year-olds who were seen at a child sexual assault assessment center.
Overall, 57% of the 399 cases disclosed abuse, with 20 of these children later
recanting. When only children who were in the "abused" category were included in the calculation, the rate of disclosure increased to 84%. It is interesring to compare the profiles of these children to the 20% of the sample who
were categorized as "unclear". The latter sample all made noncredible disclosures or noncredible denials of abuse. These children classified as unclear were
more likely to be referred by a mandated reporter because of a suspicion of
abuse, more likely to be male, and more likely to exhibit more sexual acting out
behavior.
Returning to table 2.2, studies that include cases without providing information on their diagnostic certainty (in ascending order Gordon & Jaudes,
1996; Stroud et al., 2000; Gries et al., 1996) yield disclosure rates (61 %-74%)
that are lower than those of the studies just discussed. In these studies, there
is no other evidence to confirm the abuse status of these children, and hence
the disclosure rates of corroborated abuse cases are not ascertainable from the
data.
Table 2.2 shows that the lowest rates of disclosure are provided by Sorenson and Snow (1991) and Gonzalez, Waterman, Kelly, McCord, and Oliveri
(1993). Based on our analysis of the cases included in these studies, we conclude that these low rates reflect the unreliable diagnoses of sexual abuse in
these two studies. Because the Sorenson and Snow study is most frequently
cited as supporting the notion that sexually abused children deny and recant
(see table 2.2 column 6), it is important to carefully review this study and the
characteristics of the sample.

 28

LONDON ET AL.

Sorenson and Snow (1991) selected 116 cases of "confumed" CSA from
a larger sample of 633 children who were involved in child sexual abuse allegations from 1985 to 1989. Sorenson and Snow reported that 72% of children
denied abuse when first questioned by either a parent or an investigative in.terviewer; 78% moved into a "tentative disclosure" stage with partial, vague,

or vacillating disclosures of sexual abuse. Eventually, 96% of children made .an
active disclosure that involved detailed, coherent, first-person descriptions of
the abuse.
Although Sorenson and Snow's data often are cited as supporting the notion that children commonly deny and recant abuse allegations, there is serious reason to suspect that their findings are an artifact of their therapeutic

practices. Sorenson and Snow's declared beliefs and practices that took place
in the late 1980's parallel those that we now know run a high risk of producing erroneous, suggestive reports from children (see London et al., 2005, for

further discussion of Sorenson & Snow's data). Perhaps the major concern
about this study is the fact that Sorenson and Snow assessed and treated a significant number of children for ritualistic childhood abuse which involved allegations of repetitive, bizarre, sexual, physical, and psychological abuse of
children that includes supernatural and/or religious activities (Snow & Sorenson, 1990). It appears that a number of such children were subjects in Sorenson & Snow's 1991 study (see London et al., 2005 for further discussion). The
problem with the inclusion of these types of cases into studies of disclosure patterns is that there is no evidence to support the once poplar belief that ritualistic sexual abuse is common and further that the large proportion of reported
cases of ritualistic abuse can be accounted for by the practices of a small minority of clinicians (see Bottoms, Shaver & Goodman, 1996; Goodman, Qin,
Bottoms, & Shaver, 1995; Lanning, 1991; Nathan & Snedekor, 1995). Because Sorenson and Snow diagnosed so many "ritually abused" children in
their practice, this, by inference, leads to the possibility that these children's
allegations were a product of the practices and beliefs of these clinicians.
Given the nature of the "validated" cases in the Sorenson and Snow sam.-

ple, as well as their use of apparently biased and suggestive interviewing/therapeutic techniques, the results of the study are uninterpretable. The patterns
of disclosure may merely be characteristic of children who come to make false
allegations as a result of suggestion. This would explain why these children
originally denied having been abused (because they were telling the truth),
why they eventually disclosed (because they were pressured into making allegations), and why they recanted (they wanted to restate the truth).
The Gonzalez et al. (1993) study suffers from many of the same problems.
These authors examined the disclosure and recantation patterns of 63 children
in therapy for sexual and ritualistic abuse in daycare. Gonzalez et al.'s source

of data was the therapists' retrospective accounts of the reports made by their
patients. They found that within the first 4 weeks of therapy, 76% of the children had made vague disclosures ("bad things had happened"); by 8 weeks,

2. DISCLOSURE OF CHILD SEXUAL ABUSE

29

45% of the children had disclosed highly specific terrorizing acts (killing of
adults, children, and animals); and by 20 weeks, 43% of the children had reported aspects of ritualistic abuse (organized cults). However, for the same reasons that apply to the Snow and Sorenson article, the findings of this study are
scientifically problematic. First, the children in this study were from the McMartin Preschool case and other cases that arose in the community at the
same time. The allegations in this case, which involved claims of ritualistic
abuse, arose after multiple highly suggestive interviews with evaluators and
therapists (see Nathan &Snedek01; 1995). At the time of their study, children
had been in therapy on average for over one year. There was no physical or corroborative evidence of abuse and the charges in these cases were eventually all
dropped. The interviewing methods used by children's therapists and evaluators have been documented elsewhere (e.g., Garven, Wood, Malpass, & Shaw,
1998), and the scientific evidence now shows that these methods can produce
erroneous reports when used in interviews with children.
Finally, the results of the Bybee and Mowbray study (1993) may be open
to the same criticism as detailed above. The subjects in this study were all involved in a Michigan daycare case that involved multiple perpetrators. The
case eventually resulted in only one conviction, which was overturned on appeal. Compared to the other studies in table 2.2, disclosure rates were quite
low; 58% of the 106 children disclosed abuse.
In sum, the three studies with the lowest disclosure rates (Bybee & Mowbray, 1993; Gonzalez et al., 1993; Sorenson & Snow, 1991) suffer some serious
methodological difficulties. There is reason to be concerned that the patterns
of disclosures made by children in the Gonzalez et al. and Sorenson and Snow
studies may represent those of children who make false disclosures as a result
of suggestive interviewing practices.

Representativeness of Selected Sample. Some studies reported data on
samples specifically selected because the children had not previously disclosed
sexual abuse. Such studies that focus on children who had not disclosed abuse
during an initial interview do not provide a representative view of the disclosure patterns during forensic interviews; rather these studies merely reveal the
disclosure rates of children who have previously denied abuse. Thus the studies with this type of design provide information on the degree to which deniers
disclose sexual abuse with repeated interviewing. Three studies in table 2.2
(Carnes, Nelson-Gardell, Wilson, & Orgassa, 2001; B. Wood et al., 1996; and
Lawson & Chaffin, 1992) involved such sampling procedures.
The Lawson and Chaffin (1992) study will be used to illustrate the point
because this sample included children with medical substantiations of sexual
abuse; thus, the degree of abuse certainty is high in this study. From a sample
of over 800 children who tested positive for a sexually transmitted disease
(STD) at a large pediatric hospital, cases that met the following criteria were
selected: the presenting complaint was solely physical, there was no prior dis-

 LONDON ET AL.

30

closure or suspicion of abuse, and the child was older than 3 and premenarcheal. A sample of 28 girls met these criteria, with a mean age of 7 years. These
28 children and their mothers were called back to the hospital after they tested
positive for an STD. During this interview, the mothers were given the diagnosis for the first time and then v.'ere interviewed about sexual abuse. Next,,
their daughters were interviewed by a trained social worker. Only 43% of the
girls made an abuse disclosure during this initial interview. 3 This rate, however,
is based upon a very different population than sampled in other studies, in
which children were brought in either because of a suspicion or disclosure of
abuse. Rather, in the Lawson and Chaffin study, children were selected because of their medical history and because there was no prior abuse disclosure
or suspicion of abuse. Because it is not known how many of the 800 children
in the larger sample had already disclosed abuse, this subgroup of 28 children
with no prior disclosure might comprise an unusual sample; that is, they may
represent the small hard core of children who do not disclose abuse when directly asked. If they are a small minority, then these results are not generalizable to the entire population of children with STDs. In addition, it should be
remembered that very few children who have been sexually abused have any
physical symptoms or STDs and thus this sample again is not representative of
the CSA population. There is a second factor that is important to consider. In
this study, when the children were called back to the hospital, their mothers
were first informed of the STD diagnosis of their children. Children whose
mothers accepted the possibility of abuse (the parents were labeled as supportive) were more likely to disclose (63% of this group disclosed) than children whose parents were not supportive and did not believe their child had
been abused (only 17% of these children disclosed). Thus, differences among
studies might reflect the role of parental support, which might be quite low
when parents are first confronted with the fact that their children were abused,
as was the case in the Lawson and Chaffin study (also see Elliott & Briere,
1994). 4
Two studies reported disclosure rates for children who had undergone
forensic interviews but had not disclosed. B. Wood et al. (1996) examined 55
videotaped interviews of children referred by CPS to a multidisciplinary assessment center. All 55 children had been interviewed previously by CPS or
law enforcement officials but had not disclosed. Hence, the disclosure rate of
3

In a follow up study, Chaffin, Lawson, Selby, and Wherry (1997) located 5 of these 28

subjects. Though not specifically asked about their children's disclosure, four out of five
mothers spontaneously mentioned that the child disclosed CSA subsequent to this initial
evaluation.
4
Although many mothers do not support their children's disclosures of abuse, many are
supportive, especially if the defendant is an estranged husband or partner rather than a cur-

rent one. In many studiesi the support rate is between 50o/o and 85% (see Lyons, 1999, notes

238-39 for details).

2. DISCLOSURE OF CHILD SEXUAL ABUSE

31

49% in table 2.2 is based on the percentage of children disclosing out of these

55 children who had not previously disclosed. The study does not report on the
number of children seen (and disclosing) from the general population of children presenting for CSA concerns. Finally, Carnes et al. (2001) reported that
their sample of children undergoing extended CSA assessment because of failure to initially disclose represented approximately 10-15% of the total population presenting for assessment to the clinics in their study. Thus, the results
of this study merely indicate the response patterns of children who had previously failed to disclose abuse. Furthermore, although this is not the case for the
Lawson and Chaffin study, there are no data on the number of children in both
the B. Wood and the Carnes et al. study who met acceptable criteria for diagnosis of sexual abuse. Thus, children who did or did not disclose with extra assessment may or may not have been abused.

Recantations. Eight studies have examined the frequency of recantations
of abuse reports (see table 2.2 column 5). All but one of these studies also included information on disclosure rates. For the one exception, Faller and
Henry (2000) examined the recantation rates of children who testified at trial
about their sexual abuse. Thus, all these children had made prior disclosures
that were considered credible by the prosecutors' office.
Before reviewing the findings reported in the studies, it is important to
point out that there could be two different interpretations of recantation. The
first is that the child is withdrawing a true statement of abuse. The second is
that the child is withdrawing a false allegation of abuse. The relevant evidence
is not available in most studies to determine the child's underlying motivation
for recantation.

Tbe recantation rates of the studies listed in table 2.2 range from 4% to
27%. Our analysis of the variability is very similar to that just carried out with
respect to the denial rates; namely, the highest rates of recantation are obtained for studies that have the least certain diagnoses of sexual abuse. The
two studies with the highest recantation rates were those of Gonzalez et al.
(1993) and Sorenson and Snow (1991) in which the recantation rates were
27% and 22%, respectively. Because of concerns about the actual abuse status
of the children in these studies, one might argue that recantation rates reflect
the number of children who attempt to discredit their own previous false allegations by setting the record straight. 5 (If this is indeed the case, then the Gonzalez et al. (1993) and Sorenson and Snow (1991) data suggest that these
attempts appeared to have failed, however, as most of the children reinstated
their earlier accusations.)

5

There were also issu.es concerning the validity of the sexually abused sample in Bybee

~nd Mowbray (1993) who reported a much lower recantation rate of 11%. Thus, recanta,
non rates do not necessarily have to be high for doubtful cases.

 LONDON ET AL.

32

The lowest rates of recantation are obtained for samples that have the
most certain diagnoses of sexual abuse [4% (Bradley & Wood, 1996); 6.5%
(Faller & Henry, 2000; 9% of cases classified as abused, Elliott & Briere,
1994) ]. The slightly higher rate of 15% reported by Gries et al. (1996) is difficult to interpret because there is no information on the number of children
who were diagnosed as clear or unclear cases of abuse.
Although our analysis shows that some children recant sexual abuse, the
results of this analysis show that recantation is uncommon among high probability sexually abused children. In fact, it shows just the opposite: When considering cases for which there is a high probability of abuse, a small percentage
of children in these studies recant.
CONCLUSIONS

We began this paper by describing the widespread belief that sexually abused
children do not readily disclose their abuse and that even when they disclose,
they commonly recant such disclosures. Given the frequency with >;hich these
claims are made in the literature (as well as in proffered expert testimony), we
sought to examine their scientific basis. A review of retrospective studies
showed that most adults with histories of CSA recall that they never told anyone about the abuse during childhood. This pattern confirms the view that
failure to disclose is common among sexually abused children. However, these
findings do not address the issue of whether children will deny abuse or recant
their disclosures once they have come to the attention of forensic interviewers. In order to examine these issues, it is necessary to study how sexually
abused children disclose abuse when asked directly. Because it is difficult if not
impossible to obtain accurate information if the first disclosure is made outside

a formal setting (e.g., to a parent, friend, or teacher), we have to rely on studies in which children are questioned in formal investigative interviews. We
identified 17 studies that contained relevant data and found, when focusing on
children with high probabilities of abuse history, most children do disclose
abuse within the first or second interview and that only a small minority of
children recant their abuse. Even if analyses were broadened to include children with less certain CSA diagnoses, in all but two studies the majority of
children disclosed abuse when directly asked, and only a small fraction of them
recanted their previous disclosures.
One of the basic problems in interpreting the literature on children's disclosure of sexual abuse involves the issue of the validity of the diagnosis of sexual abuse. As we stated above, in many of the cited studies, classification of
abuse was often based in part upon children's disclosures; consequently, the

conclusion that abused children do disclose abuse during formal interviews
may be circular. However, there is some evidence that when children are classified as abused based on medical evidence or on other nonchild factors (confession, material evidence) that most of these children do disclose abuse. In

2. DISCLOSURE OF CHILD SEXUAL ABUSE

33

the Elliot and Briere (1994) study, for example, 118 of the 399 children had
positive physical findings: 84% of these children at one point disclosed abuse
(although 20, or 17%, of these 118 children later recanted their abuse reports).
In Dubowitz et al. (1992), 28 of the 132 children had medical examination
findings considered indicative of abuse; 75% of this subsample disclosed abuse.
Finally, in the Gordon and Jaudes (1996) study, 78% of the children with medical evidence disclosed abuse. Before leaving this topic, it is important to point
out that the disclosure rates for the medical subsamples of these studies were
very close to the overall disclosure rates for the entire sample. Nonetheless, it
must be noted that medical evidence itself is not always be a reliable benchmark (e.g., genital redness may be due to causes other than abuse), which
raises the possibility that an unknown number of children with medical findings considered indicative of abuse actually were not abused.
The chapter by Lyon in this volume explicitly deals with methodological
problems of calculating reliable denial and recantation rates. His solution is to
select samples of children with confirmed sexual abuse history (children with
medical evidence) where there has been no prior suspicion of sexual abuse; according to this argument, disclosure rates will be untainted by "suspicion biasesn that tend to inflate disclosure rates. This solution, however, rhust be

considered in terms of the generalizability obtained from such findings and the
degree to which they can be applied to real world issues. We argue that the results of studies of disclosure patterns of children with medical histories of sexual abuse where there has been no prior suspicion or disclosure do not bear on
the most important issue that faces professionals in the forensic arena. Specifically, if a child is brought for CSA assessment because of concern (and probable prior questioning), how likely will it be that this child provides a credible
disclosure or a credible denial? In other words, the behavior of the 'highly selected samples proposed by Lyon does not pertain to the types of cases that require expert testimony or clinical assessment; namely, what are the disclosure

patterns among children who come to the attention of forensic interviewers?
We hope that our discussion makes clear some of the very difficult methodological and conceptual problems that are inherent to the study of disclosure
patterns of sexually abused children.
In most of the studies cited in this paper, there was little if any detailed information about how the children were interviewed and the degree to which
standardized and validated protocols were used. One is unable to glean from
the literature whether, for example, disclosure rates might be related to the
type of interviewing or whether the individual underwent psychological therapy. In future studies it would be important to compare the disclosure patterns
of children interviewed with contemporary standardized interviews (e.g.,
Sternberg, Lamb, Esplin, Orbach, & Hershkowitz, 2002). If these protocols do
in fact optimize the elicitation of reliable statements from children, then the
disclosure patterns produced by these instruments would provide the most reliable data to test various hypotheses about the disclosure patterns of sexually

 34

LONDON ET AL.

abused children and to explore the factors that distinguish disclosers from
nondisclosers.

Although there are a number of studies to address issues of patterns of disclosure, several overriding issues remain to be addressed. Specifically, although
the data clearly demonstrate that children involved in high probability abuse
cases do disclose and do not later recant when interviewed, there do exist a minority of children who fit the behavioral pattern of denials and recantations
that is put forth in CSM models. The outstanding issues thus focus on the
characteristics of these children and whether these children fit the psychological profiles of the CSM model (e.g., factors related to the characteristics of
the abuse, the perpetrator, and the victim). In terms of abuse and perpetrator

2. DISCLOSURE OF CHILD SEXUAL ABUSE

35

cause denial of sexual abuse to professionals is not as rampant as previously
suspected. At least among the subsample of sexually abused children to undergo forensic evaluation, our analysis indicates that when children who have
been abused are questioned in formal settings, they will usually tell, obviating
the need for suggestive questioning strategies. We believe that child abuse professionals should be aware of this information and incorporate it into their clinical practice as well as into their expert courtroom testimony. If the field is to
be guided by scientifically validated concepts, then it must be predicated on
the literature that comes closest to the standards of science.

ACKNOWLEDGMENTS

characteristics, further data are needed to examine whether factors such as use

of threats or relationship to perpettator are associated with abuse disclosure.
In terms of vicrim characteristics, there needs to be a greater focus on developmental differences in disclosure patterns. In many of the studies we reviewed, children ranged in age from early preschool to late adolescence.
Clearly, it is not very informative to provide group means when age ranges are
so great. Studies are needed to examine potential developmental ttends in loyalty to family and peers, reactions to fear, need for privacy, choice of confidants, and then to relate these factors to disclosure patterns in children of
various ages. Another important area concerns the potential role of threats. In
this future research venture, it is crucial to distinguish threats that were used
to coerce the child into molestation from threats that were used to secure the
child's silence.
The status of the scientific findings of disclosure patterns is of importance
not only for diagnostic and assessment purposes but also for issues regarding
interviewing of children. As mentioned above, the CSM model has provided
a basis for experts to advocate that when children deny abuse when directly
asked, they should be questioned further and even should be questioned suggestively (e.g., Carnes, 2000; Faller & Toth, 1995; Macfarlane & Krebs, 1986).
In order for such practices to be empirically grounded, it is important to
demonstrate first that children will commonly deny abuse when questioned
(thus calling forth the need for special strategies), and second that the use of
special strategies will lead to accurate reports of abuse. The findings presented
in this paper address the first issue only. The second issue has been addressed
by a multitude of researchers in the past decade (e.g., Ceci & Bruck, 1995;
Ghetti & Goodman, 2001; Poole & Lindsay, 2002; J. Wood & Garven, 2000).
Professionals need to be aware that although suggestive techniques might produce true abuse reports from otherwise silent children, these same techniques,
especially when used by biased interviewers, entail a risk of producing false allegations (e.g., Bruck, Ceci, & Hembrooke, 2002; Poole & Lamb, 1998). Part
of the bias may include the notion that when children deny abuse they must
be pursued until they disclose their abuse; however, as we demonstrated in this
chapter, the need for suggestive interviewing is probably overestimated be-

Portionsof this chapter appear in London, Bruck, Ceci, and Shuman (2005).

REFERENCES
Arata, C. M. (1998). To tell or not to tell: Current functioning of child sexual abuse sur~
vivors \vho disclosed their victimization. Child Maltreatment, 3, 63-71.

Bays,)., & Chadwick, D. (1993). Medical diagnosis of the sexually abused child. Child
Abuse & Neglect, 17, 91-110.
Berenson, A., Heger, A., & Andrews, S. (1991). Appearance of the hymen in newborns,

Pediatrics, 87, 458-65.
Berliner, L., & Conte, J. R. (1990). The process of victimization: The victims' perspective.

Child Abuse & Neglect, 14, 29-40.
Bottoms, B. L., Shaver, P.R., Goodman, G. S. (1996). An analysis of ritualistic and religionrelated child abuse allegations. Law & Human Behavior, 20, 1....:.34.

Bradley, A. R., & Wood, J.M. (1996). How do cbildren tell? The disclosure process in child
sexual abuse. Child Abuse & Neglect, 20, 881-91.
Bruck, M., Ceci, S. )., & Hembrooke, H. (2002). The nature of children's true and false narratives. Developmental Review, 22, 520-54.

Bybee, D., & Mowbray, C. T. (1993). An analysis of allegations of sexual abuse in a multivictim day care center case. Child Abuse & Neglect, 17, 767-83.
Campis, L.B., Hebden-Curtis,)., & DeMaso, D.R. (1993). Developmental differences in
detection and disclosure of sexual abuse. Journal of the American Academy of Child and
Adolescent Psychiatry, 32, 920-24.
Candon,)., Payne, G., & Erbaugh, C. (1996). Outcome-based practice: Disclosure rates of
child sexual abuse comparing allegation blind and allegation informed structured in,

terviews. Child Abuse & Neglect, 20, 113-1120.
Carnes, C. N. (2000). Forensic evaluation of children when sexual abuse is suspected, 2nd ed.,
National Children's Advocacy Center, Huntsville, AL.

Carnes, C. N., Nelson-Gardell, D., Wilson, C., & Orgassa, U. C. (2001). Extended forensic evaluation when sexual ahuse is suspected: A multisite field study. Child
Maltreatment, 6, 230-42.
Ceci, S. )., & Bruck, M. (1995).}eopardy in the courtroom: A scientific analysis of children's testimony. Washington, D.C.: American PsychologiC:al Association.

Ceci, S. J., & Friedman, R. D. (2000). The suggestibility of children: Scientific research and
legal implications. Cornell Law Review, 86, 34-108.

 LONDON ET AL.

36

Chaffin, M., Lawson, L., Selby, A., & \Vherry, J. N. (1997). False negatives in sexual abuse
interviews: Preliminary investigation of a relationship to dissociation. Journal of Child

Sexual Abuse, 6, 15-29.
DeVoe, E. R., &Faller, K. C. (1999). The characteristics of disclosure among children who
may have been sexually abused. Child Maltreatment, 4, 217-27.
DiPietro, E. K., Runyan, D. K., & Fredrickson, D. D (1997). Predictors of disclosure during

medical evaluation for suspected sexual abuse. Journal of Child Sexual Abuse, 6,
133-42.
Dubowitz, H., Black, M., & Harrington, D. (1992). The diagnosis of child sexual abuse.
American Journal of Diseases of Children, 146, 668-93.
Elliott, D. M., & Briere, J. (1994). Forensic sexual abuse evaluations of older children: Dis,
closures and symptomatology. Behavioral Sciences and the Law, 12, 261-77.
Everill, J., & Waller, G. (1995). Disclosure of s~xual abuse and psychological adjustment in
female undergraduates. Child Abuse & Neglect, 19, 93-100.
F::iller, K. C., & Henry, J. (2000). Child sexual abuse: A case study in community collabo~
ration. Child Abuse & Neglect, 24, 1215-25.
Faller, K. C., & Toth, P. (1995). Forensically defensible interviewing. American Professional
Society on the Abuse of Children.
Fergusson, D. M., Horwood, L. )., & Woodward, L. ). (2000). The stability of child abuse
reports: A longitudinal study of the reporting behaviour of young adults. Psycho"logkal

Medicine, 30, 529-44.

Fergusson, D. M., Lynskey, M. T., & Horwood, L. J. (1996). Childhood sexual abuse and
psychiatric disorder in young adulthood: I. Prevalence of sexual abuse and factors as~
sociated with sexual abuse. Journal of the American Academy of Child & Adolescent Psy~

chiatry, 35, 1355-64.

Finkelhor, D., Hotaling, G., Lewis, I.A., & Smith, C. (1990). Sexual abuse in a national sur~
vey of adult men and women: Prevalence, characteristics, and risk factors. Child Abuse

& Neglect, 14, 19-28.
Fisher, C. B. (1995). Ainerican Psychological Association's (1992) Ethics Code and the val~
idation of sexual abuse in day care settings. Psychology, Public Policy, and Law, l,

461-78.

Fontanella, C., Harrington, D., & Zuravin, S. J. (2000). Gender differences in the charac;
teristics and outcomes of sexually abused preschoolers. Journal of Child Sexual Abuse,

9, 21-40.
Fontes, L.A. (1993). Disclosures of sexual abuse by Puerto Rican children: Oppression and
cultural barriers.Journal of Child Sexual Abuse, 2, 21-35.
Freyd, J. J. (1996). Betrayal trauma theory: The logic of forgetting childhood abuse. Cambridge,
:tviA: Harvard University Press.
Futa, K. T., Hsu, E., & Hansen, D. J. (2001). Child sexual abuse in Asian American fami~
lies: An examination of cultural factors that influence prevalence, identification, and
treatment. Clinical Psycho"logy: Science and Practice, 8, 189-209.
Garven, S., Wood,). M., Malpass, R., &Shaw,). S. (1998). More than suggestion: Consequences of the interviewing techniques from the McMartin preschool case.Journal of

Applied Psychology, 83, 347-59.
Ghetti, S., & Goodman, G. S. (2001). Resisting Distortion. Psychologist, 14, 592-95.
Ghetti, S., Goodman, G. S., Eisen, M. L., Qin,)., &Davis, S. L. (2002). Consistency in children's reports of sexual and phyS!cal abuse. Child Abuse & Neglect, 26, 977-95.

2.

DISCLOSURE OF CHILD SEXUAL ABUSE

37

Gonzalez, L. S., Waterman,)., Kelly, R., McCord,)., & Oliveri, K. (1993). Children's patterns of disclosures and recantations of sexual and ritualistic abuse allegations in psy~
chotherapy. Child Abuse & Neglect, 17, 281-89.
Goodman, G., Qin,)., Bottoms, B., & Shaver, P. (1994). Characteristics of allegations of ritualistic child abuse. Final Report to National Center on Child Abuse and Neglect.
Goodman, G. S., Taub, E. P., Jones, D. P., England, P., et al. (1992). Testifying in criminal
court: Emotional effects on child sexual assault victims. Monographs of the Society for

Research in Child Development, 57, v-142.
Goodman~Brown, T. B., Edelstein, R. S., Goodman, G. S., Jones, D. P.H.,

& Gordon, D.
S. (2003). Why children tell: A model of children's disclosure of sexual abuse. Child

Abuse & Neglect, 27, 525-40.
Gordon, S., &Jaudes, P. K. (1996). Sexual abuse evaluation in the emergency department:
Is the hi5tory reliable? Child Abuse & Neglect, 20, 315-22.
Gray, E. (1993). Unequal justice: The prosecution of child sexual abuse. New York: Free Press.
Gries, L. T., Goh, D.S., & Cavanaug[i, J. (1996). Factors associated with disclosure during
child sexual abuse assessment. Journal of Child Sexual Abuse, 5, 1-20.
Hanson, R. F., Resnick, H. S., Saunders, B. E., Kilpatrick, D. G., & Best, C. (1999). Factors
related to rhe reporting of childhood rape. Child Abuse & Neglect, 23, 559-69.
Henry, J. (1997). System intervention trauma to child sexual abuse victims following dis~
closure. Journal oflnterpersonal Violence, 12, 499-512.
Kazarian, S.S., & Kazarian, L. Z. (1998). Cultural aspects of family violence. In S. S. Kazarian & D. R. Evans (Eds.), Cultural clinical psychology: Theory, research, and practice (pp.
316-4 7). New York: Oxford University Press.
Keary, K., & Fitzpatrick, C. (1994). Children's disclosure of sexual abuse during formal in~
vestigation. Child Abuse & Neglect, 18, 543-48.
Kellogg, N. D., & Hoffman, T. J. (1995). Unwanted and illegal sexual experiences in child~
hood and adolescence. Child Abuse & Neglect, 19, 1457-468.
Kellogg, N. D., & Huston, R. L. (1995). Unwanted sexual experi~~ces in adolescents: Patterns of Disclosure. Clinical Pediatrics, 34, 306-12.
Kendall~Tacket, K. A., Williams, L. M., &Finkelhor, D. (1993). Impact of sexual abuse on
children: A review and synthesis of recent empirical studies. Psychological Bulletin, 113,

164-80.
Kenny, M. C., & McEachern, A. G. (2000). Racial, ethnic, and cultural factors of childho"od sexual abuse: A selected review of the literature. Clinical Psychology Review, 20,

905-22.
Kovera, M. B., & Borgida, E. (1998). Expert scientific testimony on child witnesses in the
age of Daubert. In S. Ceci & H. Hembrooke (Eds.), Expert witnesses in child abuse case:
What can and should be said in court (pp. 185-215). Washington, DC: Ametican Psychological Association.
Lamb, S., & Edgar~Smith, S. (1994). Aspects of disclosure: Mediators of outcome of child~
hood sexual abuse. Journal of Interpersonal Violence, 9, 307-26.
Lanning, K. (1991). Ritual Abuse: A law enforcement view or perspective. Child Abuse &

Neglect, 15, 171-73.
Lawson, L., & Chaffin, M. (1992). False negatives in sexual abuse disclosure interviews: In~
cidence and influence of caretaker's belief in abuse in cases of accidental abuse dis~
covery by diagnosis of STD. Journal of Interpersonal Violence, 7, 53 2-4 2.
Levesque, J. R. (1994). Sex differences in the experience of child sexual victimization.

Journal of Family Violence, 9, 357-69.

 LONDON ET AL.

38

Levy, H.B., Markovic, )., Kalinowski, M. N., Ahart, S., & Torres, H. (1995). Child sexual
abuse interviews: The use of anatomic dolls and the reliability of information. Journal
of Interpersonal Violence, 10, 334--53.
London, K., Bruck, M., Ceci, S. )., & Shuman, D. W. (2005). Disclosure of child sexual
abuse: What does the research tell us about the ways that children tell? Psychology,
Public Policy, and the Law, I 1, 194--226.
Macfarlane, K., & Krebs, S. (1986). Techniques for interviewing and evidence gathering.

In K. Macfarlane & J. Waterman (Eds.), Sexual abuse of young children (pp. 67-100).
New York: Guilford Press.
Nagel, D. E., Putnam, F. W., Noll, j. G., & Trickett, P. K. (1997). Disclosure patterns of
sexual abuse and psychological functioning at a l~year follow~up. Child Abuse & Ne~
glect, 21, 137-47.
Nathan, D., &Snedekor, M. (1995). Satan's silenc" Ritual abuse and the making of a modem
American witch hunt. NY: Basic Books.
Neisser, U. (1997). Jane Doe's Memories: Changing the Past to Serve the Present. Child
Maltreatment, 2, 126-33.
Oates, R. K., & Donnelly, A. C. (1997). Influential papers in child abuse. Child Abuse &
Neglect, 21, 319-26.
Palmer, S. E., Brown, R. A., Rae-Grant, N. I., &Loughlin, M.j. (1999). Responding to children's disclosure of familial abuse: What survivors tell us. Child Welfare, 78, 259-82.
Poole, D. A., & Lamb, M. E. (1998). Investigative interviews of childreno A guide for helping
professionals. Washington DC: American Psychological Association.
Poole, D. A. 1 & Lindsay, D. S. (1998). Assessing the accuracy of young children's reports:
Lessons from the investigation of child sexual abuse. Applied & Preventative Psychology,

7, 1-26.
- - - . (2002). Children's suggestibility in the forensic context. In M. L. Eisen, J. A. Quas,
& G. S. Goodman, (Eds). Memory and suggestibility in the forensic interview. Personality
and clinical psychology series (pp. 355-81). Mahwah, NJ; Lawrence Erlbaum Associates.
Rao, K., DiClemente, R., & Ponton, L. E. (1992). Child sexual abuse of Asians compared
with other populations. Journal of the American Academy of Child and Adolescent Psy~
chiatry, 31, 880-86.
.
Read, J. D., & Lindsay, D.S. (Eds.). (1997) Recollections of traumao Scientific research and clinical practice. NY: Plenum.
Robin, M. (1991). Beyond validation interviews: An assessment approach to evaluating
sexual abuse allegations. Child and Youth Services, 17, 93-113.
Roesler, T. A. (1994). Reactions to disclosure of childhood sexual abuse: The effect on
adultsymptoms. The]oumal of Nervous and MentalDisease, 182, 618-24.

Roesler, T. A., & Wind, T. W. (1994). Telling the secreto Adult women describe their disclosures of incest. Journal of Interpersonal Violence, 9, 327-38.
Ross, M. (1989). Relation of implicit theories to the construction of personal histories.

Psychological Review, 96, 341-57.
Sas, L. D., & Cunningham, A.H. (1995). Tipping the balance to tell the secreto The public discovery of child sexual abuse. London, Ontario: London Family Court Clinic.
Schooler, J. W., Ambadar, Z., & Bendiksen, M.A. (1997). A cognitive corroborative case.
In D.S. Lindsay (Eds.), Recollections of trauma: Scientific reseprch and clinical practice

(pp. 379-88). New Yorko Plenum.

2.

DISCLOSURE OF CHILD SEXUAL ABUSE

39

Schooler,). W., Bendiksen, M.A., &Ambadar, Z. (1997). Taking the middle lineo Can we
accommodate both fabricated and recovered memories of sexual abuse? In M. Conway

(Ed.), Recovered and Fa~e Memories (pp. 251-92). Oxfordo Oxford University Press.
Sgroi, S. M., Bunk, B. S., & Wabrek, C. J. (1988). Children's sexual behaviors and theirrelationship to sexual abuse. In S. M. Sgroi (Ed.), V'ulnerable populations, Vol. 1: Evalua~
tion and treatment of sexually abused children and adult survivors (pp. 1-24). Lexington,
:MA, England: Lexington Books.
Shaw, J. A., Lewis,). E., Loeb, A., Rosado,)., & Rodriguez, R. A. (2001). A comparison of
Hispanic and African~American sexually abused girls and their families. Child Abuse
& Neglect, 25, 163-1379.
Sjoberg, R. L., & Lindblad, F. (2002). Delayed disclosure and disrupted communication
during forensic investigation of child sexual abuse: A study of 47 corroborated cases.

Acta Paediatrica, 91, 1391-96.
Smith, D. 1 Letourneau, E. J., Saunders, B. E., Kilpatrick, D. G. 1 Resnick, H. S., & Best, C.
L. (2000). Delay in disclosure Qf childhood rape: Results from a national survey. Child

Abuse & Neglect, 24, 273-87. ,
Somer, E., & Szwarcberg, S. (2001). Variables in delayed disclosure of childhood sexual

abuse. American Journal of Orthopsychiatry, 71, 332-41.
Sorensen, T., & Snow, B. (1991). How children tell: The process of disclosure of child sex~

ual abuse. Child Welfare, 70, 3-15.
Snow, B., &Sorensen, T. (1990). Ritualistic child abuse in a neighborhood setting.Journal

oflnterpersonal Violence, 5, 474--87.
Store v. Hadfield, 788 P.2d 506 (Utah 1990).
Stroud, D. 1 Martens, S. L., & Barker, J. (2000). Criminal investigation of child sexual abuse:
A comparison of cases referred to the prosecutor to those not referred. Child Abuse &

Neglect, 24, 689-700.
Summit, RC. (1983). The child sexual abuse accommodation syndrome. Child Abuse &

Neglect, 7, 177-93.
- - - . (1992). Abuse of the Child Sexual Abuse Accommodation Syndrome. Journal of
Child Sexual Abuse, l, 153-63.
Tang, C. S. (2002). Childhood experience of sexual abuse among Hong Kong Chinese col~

lege studenrs. Child Abuse & Neglect, 26, 23-37.
Toukmanian, S. 0., & Brouwers, M. C. (1998). Cultural aspects of self~disclosure and psy~

chotherapy. In S.S. Kazarian & D. R. Evans (Eds.), Cultural clinical psychologyo Theory, research, and practice (pp. 106-26). New York: Oxford University Press.
Ussher, ). M., & Dewberry, C. (1995). The nature and long-term effects of childhood sexual abuse: A survey of women survivors in Britain. British Journal of Clinical Psychology,
34, 177-92.
Wood, B., Orsak, C., Murphy, M., &Cross, H.J. (1996). Semistructured child sexual abuse
interviews: Interview and child characteristics related to credibility of disclosure. Child
Abuse & Neglect, 20, 81-92.
Wood, J.M., & Garven, S. (2000). How sexual abuse interviews go astray: Implications for
prosecutors, police, and child protection services. Child Maltreatment: Journal of the
American Professional Society on the Abuse of Children, 5, 109-18.
Wood, J.M., & Wright, L. (1995). Evaluation of children's sexual behaviors and incorpo~
ration of base rates in judgments of sexual abuse. Child Abuse & Neglect, 19, 1263-73.