2012. Vol.3, No.8, 601-605
Published Online August 2012 in SciRes (
Copyright © 2012 SciRes. 601
Relationships among Child Maltreatment, Picture Completion
Test, and Posttraumatic Symptoms: Two Examinations Using
WISC-III for Japanese Children
Kohske Ogata1,2
1Graduate School for Creative Cities, Osaka City University, Osaka, Japan
2Osaka Prefectural Government Higashiosaka Child-Family Center, Osaka, Japan
Received May 18th, 2012; revised June 20th, 2012; accepted July 13th, 2012
The aim of the present studies was to confirm a relationship among child maltreatment, Picture Comple-
tion (PC) test, and posttraumatic symptoms in Japanese maltreated children. In some previous researches,
it was found that maltreated children scored higher on the PC test within the Wechsler series subtests. It is
hypothesized that the relative higher score of maltreated children may be an indicator of hyperarousal
symptoms in posttraumatic stress disorder (PTSD). In prior studies, there were some weaknesses on
methodology. They had problems with a small sample size and indirect methods to test the hypothesis. In
study I, the prior finding was replicated through the same indirect methodology as previous studies, using
a larger sample size. The participants were 490 Japanese children (170 maltreated and 320 controls) sam-
pled from the records of Child Guidance Centers in Japan. Results showed that a significant difference
was found only for the PC score in all the subtests in the Wechsler Intelligence Scale for Children third
edition (F[1, 487] = 12.6, p < .001), replicating the previous finding using a larger sample. In study II, the
hypothesis was verified directly using another sample set (40 maltreated). Only the correlation coefficient
was significant between the deviated score on the PC test and the subscale of PTSD in Alternate version
of the Trauma Symptom Checklist for Children (r = .33 for all subtests, p < .05; r = .38 for performance
subtests, p < .05). It may be a robust finding that the relative higher performance on PC in maltreated
children relates to hyperarousal symptoms in PTSD. The author concludes that the hypothesis in the prior
researches was demonstrated through the present two studies.
Keywords: WISC-III; Child Maltreatment; Intelligence; Traumatic Stress
Child abuse and neglect are enormous traumatic events for
children and have various traumatic impacts on their develop-
ment. There have been abundant researches, particularly exam-
ining their intellectual development. It is well known that the
psychometric intelligence, defined as intelligence measured
using an intelligence test, of maltreated children is lower than
that of non-maltreated children (Ex. Buckle, Lancaster, Powell,
& Higgins, 2005; Delaney-Black, Covington, Ondersma, Nord-
strom-Klee, Templin, Ager, Janisse, & Sokol, 2002; Prasad,
Kramer, & Ewing-Cobbs, 2005). Findings that show a lower
intelligence level have been established, whereas there is little
information on subtest scatter and/or the intellectual profile of
maltreated children.
Frankel, Boetsch, and Harmon (2000) studied the elevated
Picture Completion (PC) score of 14 maltreated children. The
researchers demonstrated that maltreated preschoolers scored
significantly higher on the PC subtest of Wechsler Preschool
and Primary Scale of Intelligence Revised (WPPSI-R) than the
children’s mean score of performance subtests and of all sub-
tests. Frankel et al. suggested from their results that a higher
score on the PC related to a hyper-vigilance symptom of post-
traumatic stress disorder (PTSD) in maltreated children; how-
ever, their study had defects in the research design regarding
both the small sample size and no comparison group. Ogata
(2011) replicated the finding using a Japanese sample. In the
study, a comparison group (n = 12) was set and differences in
the types of child maltreatment were analyzed; however, this
investigation also had the weakness of a small sample size (12
sexually abused, 12 physically abused, 12 neglected, and 12
comparison). Nolin and Ethier (2007) investigated neglected
and/or physically abused children using neuropsychological
assessments. The unexpected results surprised the researchers
by showing that neglected children (n = 28) scored significantly
higher on problem-solving ability and abstract reasoning than
non-maltreated (n = 53) and neglected with physical abuse (n =
56) groups. These findings suggest that maltreated children
may have a specific intellectual profile, probably influenced by
their abuse and/or neglect experiences. Thus, to clarify the in-
tellectual profile in maltreated children is considered to par-
tially contribute to understanding their traumatic status. How-
ever, some pioneering researches have the problem on method-
The following two studies addressed the problems on meth-
odology in the prior researches. In particularly, the hypothesis
that the PC score may relate to the hyperarousal symptom in
PTSD was examined.
Study I
The primary weakness in the prior researches involves the
small sample size. A larger sample size is indispensable for
generalization of the previous findings. The purpose of study I
was to confirm the robustness of the findings relating to the
intellectual profile of maltreated children. In study I, the prior
finding that the PC scores of the maltreated children were
higher than those of non-maltreated children was replicated
using a larger sample size.
The participants included 490 Japanese children (207 girls
and 283 boys). The samples were retrospectively selected from
the records of child guidance centers (CGC) in Osaka prefec-
ture, in a region where the child maltreatment were reported
most frequently, in Japan. The CGC, like Child Protective Ser-
vices, are required to investigate child maltreatment reports, to
protect children (under 18 years old) from further maltreatment,
and to provide rehabilitative services to children, parents, and
other family members involved. The children that participated
were classified into two groups (170 maltreated and 320 com-
parisons) according to the CGC records, in which maltreatment
was determined through review of the records. Mean age (in
months) at maltreatment was 135.5 (SD = 32.0) and that of
comparisons was 146.9 (SD = 30.5). The maltreated group con-
sisted of 76 physically abused, 15 sexually abused, 26 psycho-
logically abused, and 53 neglected children. The comparison
group had been neither abused nor neglected according to the
CGC records. Physical abuse consisted of bruises, burns, cuts,
scratches, or bone fractures. Sexual abuse was defined in this
study as attempted or actual sexual contact or interaction of any
form between the participant and a caregiver or other responsi-
ble adult for purposes of the adult’s sexual gratification. Sexual
contact experiences ranged from fondling, genital touching, and
masturbation of or by another person to attempted or completed
vaginal intercourse. In this study, all psychologically abused
children were eyewitnesses to domestic violence. Following
Kotch, Lewis, Hussey, English, Thompson, Litrownik et al.
(2008), neglect was composed of two subtypes: the first sub-
type, failure to provide, involves the failure of the caregiver or
responsible adult to meet the minimum physical needs of the
child; the second subtype, lack of supervision, occurs when the
caregiver or responsible adult does not take sufficient, devel-
opmentally appropriate action to ensure the child’s safety inside
and outside the home setting. The comparison group consisted
of children needing foster care, who needed to be admitted to
child welfare institutions, with school maladaptation, and with
personality problems alleged by their parents; however, they
had not experienced any child maltreatment according to the
information on the records obtained from their parents and
themselves. All the CGC reports, whether substantiated or not,
were considered reliable indicators of maltreatment in Japan.
The intellectual functioning of children was assessed by 10
core subtests of the Wechsler Intelligence Scale for Children
third edition (WISC-III). A child’s cognitive development was
measured by the WISC-III Japanese version; the author con-
firmed from the case files that a trained child psychologist had
administered the WISC-III to the child in a CGC. The intelli-
gence test was administered after informed consent by their
parents or caregivers, and/or themselves. The scale has been
standardized on a national sample of 1125 children ranging in
age from 5 through 16 years and 11 months. Psychometric traits
of the WISC-III Japanese version were adequate on reliability
and validity (Wechsler, 1991/1998). Reliability coefficients,
based on split-half correlations, range from .64 to .85 for the 10
core subtests, and .95 for the full-scale intelligence quotient
(FIQ); reliability coefficients, based on test-retest correlations,
range from .54 to .89 for the 10 core subtests, and .93 for the
FIQ. Construct validity was also confirmed using factor analy-
sis; the four factor model was adopted, which is the same as the
original Wechsler model.
The data were analyzed as follows. First, t-tests were used
repeatedly to confirm that there were no differences between
the maltreated and comparison groups on the 10 subtests. Next,
multivariate analysis of covariance (MANCOVA) was used to
examine the difference between the maltreated and comparison
groups as controlled for the FIQ. Overall significance on
MANCOVA was followed by Bonferroni-corrected ANCOVA.
Statistical significance was set at p < .05, except for Bon-
ferroni-corrected ANCOVA (p < .005). Statistical analyses
were performed using IBM SPSS 20 Japanese version.
Results and Discussion
All results of subtest profiles are summarized in Table 1.
There was no difference in the FIQ between the groups (Mal-
treated M = 82.3, SD = 14.3; Comparison M = 83.5, SD = 13.8;
t [488] = .94, p = .35). To examine subtest scatters, it was nec-
essary to compare the subtest profiles to control for the overall
level of the FIQ. Thus, the FIQ was used as a covariate.
MANCOVA revealed a significant difference in the WISC-III
subtest results between the two groups (Wilks λ = .92, F [10,
478] = 3.91, p < .001). The obtained results demonstrated that
the intraindividual variability in the subtest profile of the mal-
treated children differed from that of the comparison group;
therefore, further examinations are needed to identify the origin
of the significant difference (Table 2).
A significant difference by Bonferroni-corrected ANCOVA
(p < .005) was found only for the PC subtest score (F [1, 487] =
12.6, p < .001). The adjusted score, controlled for the FIQ, of
the maltreated (M = 7.9) was significantly higher than that of
the comparison (M = 7.4) children. Differences in the other
subtests failed to reach statistical significance.
The results were consistent with the findings of both Frankel
et al. (2000) and Ogata (2011), showing a robustness of the
Table 1.
Subtest profiles and differences between maltreated and comparison
children in study I.
(n = 170)
(n = 320)
between groups
Subtests M SD M SD t p
Picture completion 7.8 2.4 7.4 2.4 1.91.057
Information 7.0 2.8 7.5 2.8 1.77.077
Coding 7.9 2.9 8.2 2.8 .89 .374
Similarities 7.2 3.0 7.6 2.9 1.42.155
Picture arrangement7.1 3.1 7.6 2.7 1.85a.066
Arithmetic 7.2 2.8 7.7 2.7 1.87.062
Block design 8.0 3.0 8.3 3.4 1.23.218
Vocabulary 7.4 2.6 7.4 2.9 .09 .924
Object assembly 7.4 3.2 7.9 2.7 1.82a.070
Comprehension 8.0 2.7 8.0 2.9 .11 .916
Note: Student’s t-test, df = 488, aWelch’s t-test.
Copyright © 2012 SciRes.
Copyright © 2012 SciRes. 603
Table 2.
ANCOVAs controlled for the FIQ as a covariate in study I.
Equivalent slope across the groups Regression significance Levene test ANCOVA
Subtests F p F p F p F p
Picture Completion 1.30 .255 455.44 .000* 1.34 .248 12.63 .000*
Information .58 .446 448.56 .000* .40 .525 2.39 .123
Coding .00 .977 204.37 .000* .01 .911 .20 .654
Similarity 2.07 .151 467.16 .000* 2.24 .135 1.15 .285
Picture Arrangement 6.32 .012 344.36 .000* .55 .458 2.92 .088
Arithmetic 4.34 .038 528.49 .000* 13.15 .000* 2.96 .086
Block Design 3.24 .073 401.39 .000* 2.27 .132 .66 .419
Vocabulary 6.84 .009 466.25 .000* 1.01 .316 .62 .432
Object Assembly 3.45 .064 315.01 .000* 2.02 .156 2.90 .089
Comprehension 1.46 .228 277.18 .000* .41 .522 .33 .564
Note: Statistical significance level set at .005 (.05/10), representing with the symbol*.
finding that the PC test score of maltreated may be higher than
that of comparison children. An adequate sample size (n = 490)
may strongly support the evidence. The larger sample size in
the present study, approximately tenfold that in previous studies,
would add robustness to the previous finding. In addition to the
adequate sample size, using a comparison group could enhance
the validity of the findings.
Study II
Study I replicated the previous researches. The relative
higher PC score in maltreated children was obtained using a
larger sample size; however, the methodology may have an-
other problem for examining the hypothesis suggested in the
discussion of Frankel et al. (2000). The hypothesis involved the
relationship between a deviated PC score, from the mean score
of all/performance subtests, and hyperarousal symptoms of
PTSD in maltreated children. To elucidate the hypothesis,
bivariate correlation needs to be confirmed between the PC
score and the measure on the PTSD. Therefore, study II did not
examine indirect evidence that maltreated children scored
higher on the PC subtest than the average subtest score, but
direct evidence of the correlation coefficient between the two
Participants included 40 maltreated children aged (in months)
73 - 197 (M = 143, SD = 30), 26 boys and 14 girls who had
experienced physical abuse (n = 16), sexual abuse (n = 2), psy-
chological abuse (n =6), or neglect (n = 16). Data were elicited
from the same CGC as in study I. They were assessed to evalu-
ate cognitive functioning using the WISC-III and to measure
posttraumatic symptoms using an Alternate version of the
Trauma Symptom Checklist for Children (TSCC-A). Following
or preceding the WISC-III, the TSCC-A was administered after
informed consent by their parents or caregivers, and/or them-
The Japanese version of TSCC-A was standardized for 1698
children. Psychometric traits were as follows. On internal con-
sistency, Cronbach’s alpha was .81 for anxiety, .78 for depres-
sion, .86 for anger, .84 for PTSD, and .78 for dissociation. The
validity was examined using 787 children in child welfare in-
stitutions. Correlations between 33 items on the trauma-related
child behavior checklist and TSCC-A were as follows:.14 for
anxiety, .20 for depression, .25 for anger, .15 for PTSD, and .15
for dissociation (Briere, 1996/2009).
Data were analyzed as follows. After deviation score of PC
was calculated, correlation coefficients between deviated PC
score and TSCC-A were computed. Statistical analyses were
performed using IBM SPSS 20 Japanese version.
Results and Discussion
Results of the two assessments of the participants are shown
in Table 3. Their FIQ was 89.5 (SD = 12.5) and performance
IQ was 90.4 (SD = 12.6). Following Frankel et al. (2000), the
deviation of PC scores from all or performance subtests were
calculated. Correlation coefficients between the deviated PC
score and the subscales of TSCC-A were shown in Figure 1.
Of five subscales, only the PTSD measure was correlated with
PC scores deviating from the average of both all and perform-
ance subtests.
Table 3.
Descriptive statistics on the WISC-III and TSCC-A in study II.
Picture Completion 8.3 2.2
Information 7.9 2.3
Coding 9.1 3.2
Similarities 8.4 2.8
Picture Arrangement 8.6 3.3
Arithmetic 8.5 2.9
Block Design 8.7 3.2
Vocabulary 8.7 2.3
Object Assembly 8.4 2.7
Comprehension 9.0 2.7
Anxiety 57.1 9.0
Depression 54.9 10.7
Anger 52.4 9.2
PTSD 57.2 8.7
Dissociation 54.4 9.9
00.1 0.2 0.3 0.4
All subtests
Ang e r
Anx i ety
Figure 1.
Correlation coefficients between deviated PC score from all/performance subtests and sub-
scales of TSCC-A. Dotted line represents statistical significance at p < .05.
The results may directly support the hypothesis suggested in
Frankel et al. (2000) that only the PTSD scale correlated with
the deviated PC score. The present findings could be evidence
directly examining the relationship between the PC subtest and
posttraumatic symptoms. In addition, results showing that the
other subscales did not correlate with the deviated PC score
suggest a discriminative validity for the higher PC score being
an indicator of the hyperarousal symptom.
However, study II may also have a limitation. Significant
correlation coefficients were found, but the PTSD subscale of
TSCC-A involved all aspects of posttraumatic symptoms: in-
trusion, avoidance, and hyperarousal. Thus, it is possible that
this finding suggests a relation among the other PTSD symp-
toms. Future research needs to measure the PTSD symptoms
separately and to confirm that the higher PC score relates to the
hyperarousal symptom rather than other symptoms.
Total Discussion
The present studies provide and add to evidence for the find-
ing that maltreated children scored higher on the PC subtest
because of a relationship between the PC score and the hy-
perarousal symptom. The hypothesis suggested in Frankel et al.
(2000) was verified through the two studies. In study I, a larger
sample size was used to confirm the robustness of the prior
finding and, in study II, correlations were examined directly to
verify the relationship between the deviated PC score and the
measure of PTSD. The author concludes that the present report
contributes to the body of literature on child abuse and neglect,
particularly to understanding the diverse impact of child mal-
treatment on child development.
However, there are some differences from the prior findings.
Ogata (2011) reported different effects of the types of mal-
treatment on the intellectual profile. Only sexually abused chil-
dren scored higher on the PC test in the research. The design of
Ogata’s study adopted matching methodology. Group differ-
ences were readily detected in the matching design because
their correlation coefficients were considered; however, the
sample size of a matched design is likely to be small. The total
participants in Ogata’s study was 48, whereas study I in the
current report included 490 participants; therefore, this study
did not focus on the type of maltreatment but paid attention to
the overall robustness of the finding.
These findings might demonstrate the adverse and deleteri-
ous circumstance where maltreated children live. Because daily
atmosphere in their family should influence on their emotional
status strongly, they may suffer from hypervigilance and/or
hyperarousal symptom.
There are some clinical implications. Firstly, in these analy-
ses, controlling the FIQ was important because raw subtest
scores between groups were not significantly different (Table
1). The results indicated that the intellectual characteristics of
maltreated children were represented not as a raw score but as
an adjusted score in the PC subtest, controlled for the overall
level of the FIQ. Therefore, it may be useful in clinical settings
to calculate the ratio (subtest score/FIQ), and it was recom-
mended to evaluate intraindividual differences to assess chil-
dren who have experienced maltreatment. Secondly, it was
recommended to include the PC subtest in the test battery to
assess cognitive functioning. The WISC-IV excludes the PC
subtest from 10 core subtests in the current revision. As a sup-
plemental subtest, the PC score is unnecessary to compute the
FIQ; however, the author recommends the PC subtest for as-
sessing maltreated children on the basis of the current findings.
The PC test might be able to screen their hyperarousal symp-
toms in PTSD. A maltreated child who scores relative higher on
the PC should be referred to a child psychiatrist for detailed
diagnosis of PTSD. Whether or not maltreated children have
symptoms related to the PTSD, their psychological treatment
may vary according to their mental status. The author provides
some recommendations from the view of the clinical setting.
Finally, the findings are discussed from cultural standpoints.
The current investigations were conducted in Japan. Almost all
previous researches were Western findings. Thus, the present
findings replicating Western studies should be compared with
other non-Western reports. For example, in the Turkish litera-
ture, Bulut and colleagues studied PTSD in children who have
experienced natural disasters (Bulut, 2004, 2006, 2010; Bulut,
Bulut, & Tayli, 2005). Bulut (2010) reported from longitudinal
investigations that overall PTSD rates in children gradually
decreased. The research revealed gender differences that were
detected in intrusion and avoidance symptoms. It is difficult to
directly compare these findings with the current research be-
cause of the difference in the participants in each study.
Briere, J. (2009). Trauma Symptom Checklist for Children (TSCC)
Professional Manual. In S. Nishizawa, & C. Yamamoto (Trans. and
write). Odessa, FL: Psychological Assessment Resources.
Buckle, S. K., Lancaster, S., Powell, M. B., & Higgins, D. J. (2005).
The relationship between child sexual abuse and academic achieve-
ment in a sample of adolescent psychiatric inpatients. Child Abuse &
Neglect, 29, 1031-1047. doi:10.1016/j.chiabu.2004.12.013
Copyright © 2012 SciRes.
Bulut, S. (2004). Factor structure of posttraumatic stress disorder in
children experienced natural disaster. Psychologia: An International
Journal of Psychology in th e O ri en t , 47, 158-168.
Bulut, S. (2006). Comparing the earthquake exposed and non-exposed
Turkish children’s Post Traumatic Stress Reactions. Anales de Psi-
cología, 22, 29-36.
Bulut, S. (2010). Children’s posttraumatic stress reactions and sub-
symptoms: Three years of a longitudinal investigation study after a
direct exposure to the earthquake and school’s collapse. Turk Psiko-
loji Dergisi, 25, 87-98.
Bulut, S., Bulut, S., & Tayli, A. (2005). The dose of exposure and
prevalence rates of post traumatic stress disorder in a sample of
Turkish children eleven months after the 1999 Marmara earthquakes.
School Psychology International, 2 6, 55-70.
Delaney-Black, V., Covington, C., Ondersma, S. J., Nordstrom-Klee,
B., Templin, T., Ager, J., Janisse, J., & Sokol, R. J. (2002). Violence
exposure, trauma, and IQ and/or reading deficits among urban chil-
dren. Archives of Pediatr i c s & Adolescent Medicine, 156, 280-285.
Frankel, K. A., Boetsch, E. A., & Harmon, R. J. (2000). Elevated pic-
ture completion scores: A possible indicator of hypervigilance in
maltreated preschoolers. Child Abuse & Neglect, 24, 63-70.
Kotch, J. B., Lewis, T., Hussey, J. M., English, D., Thompson, R.,
Litrownik, A. J., Runyan, D. K., Bangdiwala, S. I., Margolis, B., &
Dubowitz, H. (2008). Importance of early neglect for childhood ag-
gression. Pediatrics, 121, 725-731. doi:10.1542/peds.2006-3622
Nolin, P., & Ethier, L. (2007). Using neuropsychological profiles to
classify neglected children with or without physical abuse. Child
Abuse & Neglect, 31, 631-643. doi:10.1016/j.chiabu.2006.12.009
Ogata, K. (2011). Intellectual profile of sexually abused children in
Japan: An analysis of WISC-III subtests compared with physically
abused, neglected, and non-maltreated children. Psychology, 2, 169-
172. doi:10.4236/psych.2011.23027
Prasad, M. R., Kramer, L. A., & Ewing-Cobbs, L. (2005). Cognitive
and neuroimaging findings in physically abused preschoolers. Ar-
chives of Diseases in Childhood, 90, 82-85.
Wechsler, D. (1998). Manual for the Wechsler intelligence scale for
children-third editio n. San Antonio: The Psychological Corporation.
Copyright © 2012 SciRes. 605