Behavior Problems in Sexually Abused Young Children

Abstract

A sample of 85 children, ages 3 to 12, and sexually abused within the previous 24 months, were avaluated with the Child Behavior Checklist (CBCL). Data relating to the nature of the abuse were also collected. This sample was clearly more deviant behaviorally as measured by the CBCL. Internalizing behavior was related to the frequency of abuse, the sex of the child, the relationship of the perpetrator to the child, and the severity of abuse. Externalizing behavior was related to the duration, perpetrator's relationship, time elapsed, and sex of the child. Sexual behavior in this sample was related to frequency of abuse and number of perpetrators.

Full text

Journal of Pediatric Psychology, Vol. 11, No. 1, 1986
Behavior Problems in Sexually Abused
Young Children
1
William N. Friedrich,
2 Anthony J. Urquiza, and Robert L. Beilke
University of Washington
Received May 30, 1984; revised November 19, 1984
A
sample
of
85
children,
ages 3 to
12,
and
sexually abused within
the previous
24
months,
were evaluated with
the Child
Behavior Checklist
(CBCL, Achen-
bach
&
Edelbrock,
1983).
Data
relating
to the nature of
the
abuse
were
also
collected.
This sample was clearly more deviant
behaviorally
as measured
by the CBCL. Internalizing behavior
was
related to the frequency of abuse,
the sex of
the
child,
the
relationship
of the perpetrator to the
child,
and the
severity of abuse. Externalizing behavior was related to the duration,
perpetrator's
relationship,
time
elapsed,
and
sex of the
child.
Sexual behavior
in
this
sample
was related
to frequency of
abuse
and
number
of perpetrators.
KEY WORDS: child sexual abuse; behavior problems.
The sexual abuse of children is a problem that was once thought to be rare
but has become recognized as occurring frequently (Russell, 1984). Victims
may demonstrate a variety of short- and long-term effects of this abuse. Yet
the empirical literature is sparse and is either marred by methodological pro-
blems or reports contradictory results. Some of the specific methodogical
problems involve the use of a variety of definitions of sexual abuse, sam-
1 We
gratefully acknowledge the assistance of Lucy Berliner, Tom Merrill, Barb Huffman, and
Annette Fehrenbach with subject recruitment, and Thomas Achenbach, Oavid Finkelhor, and
three anonymous reviewers for comments on earlier drafts of this article. A version of this
article, "Behavioral Impact of Sexual Abuse on Young Children," was presented at the Third
National Conference on the Sexual Victimization of Children, April 27, 1984.
'All correspondence should be sent to William N. Friedrich, Department of Psychology, Mayo
Clinic, Rochester, Minnesota 55905.
47
0146-8693/86/03004047103.00/0 © 1986 Plenum Publishing Corporation

48 Friedrich, Urquiza, and Beilke
pling techniques that rarely assess the child victim shortly after the abuse,
and the lack of standardized outcome measures.
Researchers on the outcome of abuse diverge widely: Some predict lit-
tle or no harm and others predict harm only when the child victim grows
older and begins to engage in intimate heterosexual relationships. Others
discuss possible short- and long-term effects in children which include pro-
blems in sexual adjustment, interpersonal problems, education problems, and
a variety of other psychological symptoms (Mrazek & Mrazek, 1981).
Largely clinical data suggest that greater trauma to the child is incur-
red when the perpetrator is more closely related (Burgess, Grot h, Holmstrom,
& Sgroi, 1978), the abuse has a longer duration (Burgess et al., 1978), that
force is used (Finkerlhor, 1979), the victim is older and can understand what
happened (Tsai, Feldman-Summers,
&
Edgar, 1979), and the abuse involves
penetration (Russell, 1984).
Mrazek and Mrazek (1981) have suggested that six primary
characteristics of the sexual abuse may be related to subsequent behavioral
sequelae. These include the extent of sexual contact, the age and developmen-
tal maturity of the child, the degree of relatedness between victim and
perpetrator, the affective nature of the sexual relationship, the age difference
between the victim and the perpetrator, and the length of the sexual rela-
tionship. They recommend that at least these characteristics be included in
studies of behavioral symptoms subsequent to abuse.
The present study was designed to overcome some of
the
methodological
shortcomings extant in the literature. An initial step in this direction was the
choice of a standardized measurement device, in this case a child behavior
checklist. Second, only children were studied, and these were children who
had a documented history of sexual abuse that had been ongoing no more
than 24 months earlier. Finally, by using only those cases where the offender
was at least
18
years old, we eliminated the age difference variable and were
able to focus on measuring those variables identified by Mrazek and Mrazek
(1981).
One of the only published studies similar to this had only
28
children
in the sample, ranging in age from
3
to 15. Additi onal ly, the time since abuse
varied more widely, and a child behavior checklist without sufficient items
pertaining to sexual behaviors was used (Adams-Tucker, 1982).
The measurement device chosen, the Child Behavior Checklist (CBCL,
Achenbach & Edelbrock, 1983), is a widely used, psychometrically sound
instrument that is completed by the parent. It provides measures of social
competence in the child and broadband factors for Internalizing and Exter-
nalizing Behavior. These two factors are the result of a factor-analytic studies
of children's behavior problems. Essentially, behavior problems fall generally
into one or the other of these two broad categories of internalizing and ex-
ternalizing. The former includes behaviors described as fearful, inhibited,

Sexual Abuse 49
depressed, and overcontrolled, whereas the latter includes aggressive, an-
tisocial, and undercontrolled behavior. The CBCL was also chosen specifically
for this study because it contains several items related directly to sexual
behavior. Sexualization or increased sexual behavior has been reported as
an outcome of sexual abuse (Kreiger, Rosenfeld, Gordon,
&
Bennett, 1980).
The CBCL allows the enumeration of these particular sexual behaviors, for
example, masturbation, preoccupation with sex, etc.
The study was designed to answer several questions. Initially, we wanted
to determine, using norms for the CBCL, what percentage of the children
in our sample could be described as having a significant elevation on the in-
ternalizing or externalizing scales. Second, we wanted to determine whether
any of the abuse characteristics were related in a significant fashion to inter-
nalizing, externalizing, or sexualized behavior. This was examined using multi-
ple regression analyses.
METHOD
Subjects
Eighty-five sexually abused children, male (n = 24) and female (n =
61),
were included in this study. They ranged in age from
3
to 12 (X = 6.2,
SD = 2.7), with 4-, 5-, and 6-year-olds constituting 59% of the sample.
Documentation of the abuse was done in each case by the local Children's
Protective Services office. They were derived from four sources: Consecutive
referrals to the first author for psychological evaluation (January 1983 to
September 1984) by the local Sexual Assault Center (n = 50), all children
whose mothers were in group therapy (during 1983) as part of their child's
treatment for sexual abuse at a local mental health center
(n
= 24), all children
whose mothers were in group therapy (during the first half of 1984) as part
of their child's treatment for sexual abuse at a private child treatment facili-
ty (n = 5), and all the sex abuse referrals to the outpatient department of
a children's hospital (January 1984 to June 1984) (n = 6). Although every
effort was made to include all eligible children available, our sample cannot
be considered representative of sexually abused children. Due to the fact that
children demonstrating more problems are more likely to be referred for treat-
ment, as a group these children are probably displaying a more pronounced
response to the abuse.
Mothers or mother figures completed the CBCL (Achenbach &
Edelbrock, 1983) in all cases, and a therapist/caseworker familiar with the
case completed a cover sheet. This contained the following information: the
relationship (weighted by "closeness" to the child) of the child to the

50 Fried rich, Urquiza, and Beilke
perpetrator (father or mother = 1, stepf ather = 2, brother = 3, grandfather
or grandmother = 4, uncle = 5, family friend = 6, other male relative =
7,
neighbor = 8, stranger = 9); the type of sexual abuse with each type
weighted (observes = 1, fondles breast = 3, fondles genitals = 4, cunnil-
ingus = 5, attempted intercourse = 7, fellatio = 8, vaginal/anal intercourse
= 12); the duration of the abuse (in months); the frequency of the abuse
(times per month); number of perpetrators; and whether physical force ac-
companied the abuse (yes or no). Up to three types of sexual acts per child
were coded. Since we wanted to develop some measure of the "severity" of
the abuse, the final score for severity of the act was the sum of the weighted
score for each act (possible range is 1 to 32). (The weights were arrived at
by the authors.) We also realize that the relationship closeness weights may
not fully reflect the degree of emotional involvement the perpetrator had with
the child. However, more specific information was not obtained at this time.
See Table I for abuse percentages and weights.
For the purposes of this study, sexual abuse was defined as sexual con-
tact with an adult, whether by force or consent, to include direct contact,
that is, genital, oral, anal intercourse, and observed contact (i.e., adult ex-
posing self to child). All children had a documented history of sexual abuse
in the previous 24 months (X = 6.1 months, SD = 5.6 months). The primary
perpetrator was the father (51%), followed by other male relatives (26%),
stepfather (10%) and smaller percentages of neighbor, family friend, and
mother.
Again, this sample is not considered representative of a random sam-
ple of sexually abused children. However, it does seem to be a primarily
middle-class or above sample, in that 93% of the fathers or father figures
and 63% of the mothers were in skilled trades or better. Almost one-third
of the mothers were full-time housewives. Regarding ethnic distribution, 8 5%
were white and the remainder distributed evenly among black, Asian, and
Native American.
Table I. Sexually Abusive Acts Committed
Type of abuse
Fondle genitals
Fellatio
Attempted intercourse
Observes
Other, e.g., pornography
Intercourse, vaginal
Intercourse, anal
Cunnilingus
Fondles breasts
Abuse
weight
3
5
6
1
1
7
8
4
2
Percentage of total sexual
acts recorded {N = 176)
44
14
13
II
5
10
4
3
2

Sexual Abuse 51
Instrument
The CBCL (using the revised profile) is a standardized checklist of
childhood behavior problems and social competence. It includes
118
behaviors
that parents score on a three-point scale as to their presence in the child in
the previous 6 mon ths. Scores are derived for up to three dimensions of social
competence (i.e., activities, social, and school). Behavior problems can be
scored along two broad factors, Internalizing and Externalizing. The former
consists of items related to depression, anxiety, somatic concerns, and so
forth, whereas the latter factor consists of items related to aggression, cruel-
ty, delinquency, hyperactivity, and so forth. For the purposes of this study,
we developed a third behavior score. This consisted of the sum of the
responses for six items of sexual behavior on the CBCL. These include the
items (with the percentage endorsement for each item in this sample) #5
"Behaves like the opposite sex" (16%); #59 "Plays with own sex parts in
public" (35%);
#60
"Plays with own sex parts too much" (52%); #73 "Sexual
problems"
(45 %);
#96
"Thinks about
sex
too much" (53%); and
#110
"Wishes
to be of the opposite sex" (11%). Only 16 of the 85 (19%) children did not
have at least one of these items endorsed.
The CBCL has norms for both male and female in the age ranges of
4-5 and
6-11.
The developers of the test indicate that it is reasonable to in-
clude 3-year-olds in the 4-5 group and 12-year-olds in the 6-11 group (T.
M. Achenbach, personal communication, August 31, 1984). We had eight
subjects age 3 and five subjects age 12.
RESULTS
Initial analyses focused on examining what percentage of our sample
could be described, using CBCL norms, as having a significant elevation,
that is, T >70, on either the internalizing or externalizing scales. Approx-
imately
35%
of the males and 46% of the females were significantly elevated
on the internalizing scale, and 36% of the males and 39% of the females
on externalizing. Given that only 2% of the normative samples would have
T scores >70, this clearly suggests much greater deviance in this sample than
in a normative group.
It appeared that there were some age level differences in the degree of
significant behavior problems. Children 5 and younger were compared to
children ages 6 through 12 and 51% and 36% had T scores >70 on inter-
nalizing, and 31% and 44% on externalizing, respectively. Apparently,
younger children looked more internalizing and older children looked more
externalizing.

I
Table II. Correlation Matrix of Sex Abuse and CBCL Variables
Sex
Internalizing
Externalizing
Use of force
Time elapsed
Duration
Frequency
Perpetrator
No.
of perpetrators
Severity
Sexual behavior
Age at onset
Age
.07
-.15
-.02
.02
.43'
.55'
.03
.18°
.24°
-.07
-.12
.92'
Sex
.04
-.04
.06
.19
-.08
-.26°
.10
-.07
.01
-.17
-.00
Inter-
nalizing
.74'
.02
-.33'
.21°
.37'
-.20°
.18
.21°
.54'
-.02
Exter-
nalizing
.02
-.29'
.31'
.30*
-.31'
.20°
.17
.58'
.11
Force
.20°
-.00
.29'
.07
-.09
.26'
-.03
-.06
Time
elapsed
.16
-.18
-.02
-.28'
.04
-.38'
.04
Duration
.57'
-.14
.34'
.16
.33'
.55'
Frequency
-.04
.37'
.20°
.54'
.09
Perpe-
trator
.04
-.03
-.22°
.19
No.
of
perpe-
trators
-.08
.32'
.33'
Severity
.13
-.10
Sexual
behavior
.03
°p < .05.
'p < .01.

Sexual Abuse 53
A Pearson correlation matrix was generated next (see Table II). Inter-
nalizing behavior is significantly correlated with externalizing, duration, fre-
quency, severity, and sexual behavior, and inversely with time elapsed and
relationship to perpetrator. Externalizing is directly correlated significantly
with internalizing, duration, frequency, number of perpetrators, and sexual
behavior, and inversely with time elapsed and relationship to perpetrator.
Sexual behavior is directly correlated significantly with internalizing, exter-
nalizing, duration, frequency, and number of perpetrators, and inversely with
time elapsed and relationship to perpetrator. The pattern of correlations is
remarkably similar for internalizing, externalizing, and sexual behavior.
The next analyses focused on determining how the sexual abuse
characteristics were related to the behavior problems exhibited in the child.
This was done using the stepwise multiple regression program outlined in
SPSS (Nie, Hull, Jenkins, Steinbrenner,
&
Brent, 1975). Because of the small
sample, male and female were not examined separately, but sex of child was
included in the analyses. We limited our independent variables to 10: sex,
current age, age at onset, relationship of child to perpetrator, the number
of perpetrators, severity of abuse, duration of abuse, frequency of abuse,
the presence of physical force, and the time elapsed since the last incident.
These
10
were then examined with the Internalizing and Externalizing T scores
(as recommended by T. M. Achenbach, personal communication, August
31,
1984) and Sexualization raw scores as the dependent variables.
Four variables were significant predictors of Internalizing, overall F(4,
79) = 5.4, p <
.001,
R2 = .32. These were the frequency of the abuse, the
sex of the child, the relationship of the child to the perpetrator, and the
weighted score developed for severity of abuse. In summary, female children
abused frequently and severely by a perpetrator emotionally close to them
exhibited more internalizing behavior. Duration, relationship of the child
to the perpetrator, time elapsed, and sex of the child were the significant
predictors for Externalizing, overall F(4, 79) = 5.6, p < .001, R2 = .33.
In summary, externalizing behavior decreased with time elapsed and was
directly related to abuse of long duration by an emotionally close perpetrator
with a male child. This was also a direct relationship. Two variables were
significant predictors of Sexual Behavior, overall F(2 , 81) =
13.6,
p <
.0001,
R2
= .36). These were frequency and number of perpetrators. Apparently,
the more frequent the abuse by a greater number of perpetrators, the more
likely the child was to demonstrate greater sexualized behavior. See Table
III for additional information.
The unique contribution, in terms of additional unique variance
predicted, of the significant predictor variable(s) was calculated for each
dependent variable by entering the independent variable last in the regres-
sion statement. The unique contribution of each of the four significant

54 Friedrich, Urquiza,
and
Beilke
Table III. Stepwise Regression Analyses of CBCL Factors
Dependent
variable
Internalizing
Frequency
Sex of child
Perpetrator
Severity
Externalizing
Duration
Perpetrator
Time elapsed
Sex
Sexual behavior
Frequency
No.
of perpe-
trators
Multiple R
.37
.45
.51
.56
.36
.45
.52
•57
.53
.60
R1
.14
.20
.26
.32
.13
.20
.27
.33
.28
.36
Simple r
.37
.15
-.24
.29
.36
-.30
-.18
.10
.52
-.46
F
5.36
5.6
13.6
P
.001
.001
.0001
variables
for
internalizing
was
frequency (7.5%),
sex
(1.3%), perpetrator's
relationship (6.0%),
and
severity (5.5%). Frequency, perpetrator's relation-
ship,
and
severity were significant.
For
externalizing,
the
unique variance
of the four significant predictor variables was duration (.81%), perpetrator's
relationship (5.1%), time elapsed (3.5%),
and
child's sex (2.3%).
All
except
duration contributed significant unique variance.
For
sexual behavior,
fre-
quency contributed
an
additional, unique
8.1%,
and
number
of
perpetrators,
1.9%.
Only
the
former
was
significant.
DISCUSSION
Since
a
random sample
of
sexually abused children
was not
studied,
the generalizability
of
these results
is
limited.
In
addition,
we d id not
have
teacher ratings
of
children's behavior, which might have provided valida-
tion
of
behavior problems. This type
of
additional source
of
data should
be
an
important feature
in
future research.
The fact that
a
much larger than expected number
of
children were
displaying significant behavior problems
{T > 70)
suggests that
for
many
children, the abuse appears to have
an
im pac t reflected in behavior problems.
This
is not
conclusive
by any
means, given that this study cannot answer
whether the problems existed prior to the abuse
or
are a function
of
the
child's
environment.
It
does indicate that our sample had
a
large number
of
children
that appeared
to be in
need
of
mental health services.
An important finding
is
that
the
behavior
of
young sexually abused
children is related
in a
significant manner
to the
characteristics
of
the sexual

Sexual Abase 55
abuse. Of the seven independent variables that were significant predictors
in at least one of the regression statements, five (i.e., severity, perpetrator's
relationship, duration, frequency, and number of perpetrators) could be con-
sidered reflective of the overall severity of the abuse. The more invasive the
sexual act and the longer and more frequently these acts were committed
by someone close to the child constitute greater severity and are related to
corresponding increases in a variety of behavior problems.
The complexity of these findings is illustrated clearly by the results
demonstrating the importance of the "relationship to perpetrator" variable.
Natural parents who offended had children with greater behavioral sequelae.
One reason for this may be that sexual abuse by one's father results in more
betrayal and a loss of trust in an important relationship. Another reason may
be that an incestuous relationship is suggestive of greater family discord prior
to the abuse and, consequently, the behavior problems were at least in part
preexisting. Finally, it may be that following discovery of the sexual abuse,
an incestuous parent is removed from the home with subsequent loss of in-
come, family integrity, and possible scapegoating of the victim for her role
in the breakdown of the family. These additional stressors then contribute
to greater behavior problems. The point of this discussion is to underscore
how complex the findings are and to illuminate the many possible explana-
tions that may exist for any finding.
Also significant was the variable related to time since last incident. In
this study, time elapsed varied from several weeks to 24 months. One would
expect that the simple elapse of
time
would be reflected in a decline in behavior
problems. This was seen for both male (for internalizing r = -
.51;
for ex-
ternalizing r =
—
.37) and female (for internalizing r = - .29; for exter-
nalizing r =
—
.23).
It appears that the frequency of behavior problems
subside with the passage of time. This finding is similar to what has been
noted in studies of adult victims of sexual trauma (Atkeson, Calhoun, Resick,
& Ellis, 1982). It is possible that although the frequency of behavior pro-
blems appears to subside, qualitatively some symptoms may be more en-
trenched (e.g., sexual problems). Also, it is possible that by averaging results
of different children across different time periods, one loses distinct individual
or subgroup patterns of recovery which would have interesting clinical im-
plications. Future research is needed that examines the life cycle of the
behavioral symptoms in these children in order to determine whether a typical
pattern or patterns are evidenced.
The behavior problems exhibited were quite varied. The CBCL has a
number of second-order factors within the two primary factors. The factors
that were elevated most consistently in our sample included aggression,
depression, and social withdrawal for males and a similar constellation for

56 Friedrich, Urquiza, and Beilke
females. Sexual behaviors were also very much more common in this sample
than expected.
Age was not significantly correlated with sexual problems. In normal
circumstances, children learn to inhibit the expression of sexual behavior as
they get older. However, in our sample, the mean raw score (maximum =
12) for the six sex behavior items endorsed for the 3-, 4-, and 5-year olds
was 3.72, whereas the mean raw score items for the 9- through 12-year-olds
was 3.1. These were not significantly different. It did appear that the sex
behaviors more common in the younger sample were masturbation, whereas
it was more common for the older children to be rated as preoccupied with
sexual matters or having other sexual problems.
Sex differences were noted in that the sex of the child was entered in
the stepwise solutions for both internalizing and externalizing. The correla-
tions suggest greater internalizing behavior for girls and greater externaliz-
ing behavior for boys. This sex d iffe renc e is a frequent finding in the literature
on child psychopathology (Erne, 1979).
In summary, this research indicates that the nature of
the
sexual abuse
(e.g., frequency, severity, duration) is related in a significant manner to
behavior problems in the child. A study of this sort cannot establish cause
and effect, and, consequently, it is beyond our ability to report whether there
is a direct causal relationship between the abuse and the behaviors seen. Many
of these children were reported by their parents as having changed dramatical-
ly for the worse after the abuse, but no prospective data are available for
any of these children. Strengths of the study are that a comparatively large
sample was utilized, characteristics of the sexual abuse were carefully
documented by therapists knowledgeable about the exact abuse, and their
relationship to behavior problems in the children was examined empirical-
ly. The clinical literature to date has generally indicated that a greater
behavioral impact is seen when the abuse is more severe, force is used, and
the perpetrator has a closer relationship to the child. The current study cer-
tainly supported severity and relationship of perpetrator to the child but not
force. It is possible that our simple dichotomous rating of force and its low
prevalence in our sample contributed to its lack of significance. However,
there is greater ambiguity in the literature for the variable duration and age
at onset. Although duration was significantly correlated with internalizing,
externalizing, and sexual behavior, neither it nor age at onset contributed
significant unique variance to the regression statements.
Future research must explore other variables that may be related to the
behavior problems exhibited. For example, the affective relationship of the
perpetrator-victim must be measured more carefully than it is in this study.
A perennially absent father may be less "emotionally close" than an involved
stepfather and, consequently, abuse by the latter may result in a greater sense

Sexual Abuse 57
of loss and betrayal on the part of the child. In addition, the degree to which
the child is further victimized by the system's response to the abuse must
be measured (e.g., lack of support by the family, foster care placements,
numerous interviews, having to testify). For some of the children, the sexual
abuse was only one of a number of abuses they experienced in the time since
the abuse was disclosed. Finally, the quality of the mother-child relation-
ship and the larger family environment should also be examined simultaneous-
ly. These latter interpersonal aspects may also be critical in predicting the
degree to which the victim can successfully resolve this abuse. We strongly
believe that a stable family organization, a positive rapport between parent
and child, and involvement in psychotherapy are important and positive
moderator variables that must be studied simultaneously in future research.
REFERENCES
Achenbach, T. M., & Edelbrock, C. (1983). Manual for the Child Behavior Checklist. Burl-
ington: University of Vermont.
Adams-Tucker, C. (1982). Proximate effects of sexual abuse in childhood: A report on 28 children.
American Journal of Psychiatry, 139, 1252-1256.
Atkeson, B. M., Calhoun, K. S., Resick, P. A., & Ellis, E. M. (1982). Victims of rape: Repeated
assessment of depressive symptoms. Journal of Consulting and Clinical Psychology, 50,
96-102.
Burgess, A., Groth, N., Holmstrom, L., & Sgroi, S. M. (1978). Sexual assault of children and
adolescents. Lexington, MA: Lexington Books.
Eme,
R. F. (1979). Sex differences in childhood psychopathology: A review. Psychological
Bulletin, 86, 574-595.
Finkelhor, D. (1979). Sexually victimized children. New York: Free Press.
Krieger, M., Rosenfeld, A., Gordon, A., & Bennett, J. (1980). Problems of psychotherapy of
children with a history of incest. American Journal of Psychotherapy, 34, 81-88.
Mrazek, P. B., & Mrazek, D. A. (1981). The effects of child abuse: Methodological considera-
tions.
In P. B. Mrazek & C. H. Kempe (Eds.), Sexually abused children and their families
(pp.
235-246). New York: Pergamon.
Nie,
N., Hull, C, Jenkins, J., Steinbrenner, K., & Brent, D. (1975). Statistical package for
the social sciences (2nd ed.). New York: McGraw-Hill.
Russell, D. E. H. (1984). Sexual exploitation. Beverly Hills, CA: Sage.
Tsai, M., Feldman-Summers, S., & Edgar, M. (1979). Childhood molestation: Variables related
to differential impacts on psychosexual functioning in adult women. Journal of Abnor-
mal Psychology. 88, 407-417.