Vol 50(1)-Winter 2006 12/15/05 10:13 AM Page 7

W I S C – I I I V E R B A L / P E R F O R M A N C E D I S C R E P A N C I E S

WISC–III Verbal/Performance Discrepancies

Among a Sample of Gifted Children

John D. Sweetland, Jacqueline M. Reina, and Anne F. Tatti

Port Washington School District, Port Washington, NY

environment” (Wechsler, 1994, p. 1). In an attempt to

A B S T R A C T measure a person’s general intelligence and to distin-
guish between verbal and nonverbal abilities, the test is
The cognitive profiles of 161 gifted children—those
constructed of subtests that comprise Verbal and
with a Full Scale, Verbal, and/or Performance IQ of
130 or higher on the Wechsler Intelligence Scale for Performance Scales, the former measuring auditory-
Children, Third Edition (WISC–III; Wechsler, vocal intelligence and the latter, visual-perceptual ability
1991)—were examined. It was hypothesized that (Kaufman, 1994).
this population of very intelligent children would
demonstrate much larger Verbal-Performance dis- P U T T I N G T H E R E S E A R C H
crepancies than are reported in the WISC–III man- T O U S E
ual. The hypothesis was confirmed, and a table
reporting Verbal-Performance discrepancies estab- The present study addresses an interpretive anomaly
lished that very large discrepancies are typical for this regarding Verbal-Performance (V-P) discrepancies
population. The implications of these results are dis- for intellectually gifted children. The data indicate
cussed, as well as the limitations of the study, partic- that the amount and frequency of the V-P discrepan-
ularly as it pertains to the limitations of the sample. cies amongst a sample of gifted children is very dif-
ferent than that reported in the Wechsler
Intelligence Scale for Children, Third Edition
Each year, more than one million intelligence tests (WISC–III; Wechsler, 1991). Large V-P discrepan-
are individually administered to school-aged children in cies are not uncommon and in no way indicate cause
the United States (Macmann & Barnett, 1997). The for concern for a sample of gifted children. Caution
Wechsler Intelligence Scale for Children–Third Edition must be used when interpreting both the frequency
(WISC–III; Wechsler, 1991) is the most widely used and magnitude of V-P differences.
psychometric measure of intelligence for children Careful analysis of the data indicates that in
(Donders, 1996), especially with gifted children almost all cases, this sample of gifted children
(Winner, 2000). The WISC–III has recently been demonstrated much higher verbal scores than per-
updated and revised with the introduction of the new formance scores. Apparently, it is common for very
Wechsler Intelligence Scale for Children–Fourth bright children to attain unusually high scores in
Edition (WISC–IV; Wechsler, 2003). Although the only some aspects of cognitive functioning. This
present study examines the WISC–III, the results con- suggests an important and distinct pattern of cogni-
tive strengths. While the verbal scores were typically
tinue to be relevant to working with the WISC–IV, as
higher, our data suggests that in almost all cases, per-
the discrepancy between the Verbal Comprehension
formance scores were at least average or perhaps
(VC) and the Perceptual-Reasoning (PR) indexes can
slightly stronger. This indicates very highly devel-
be utilized in the same manner in which Verbal-
oped skills in some modalities with many other cog-
Performance (V-P) discrepancies were utilized in ana-
nitive skills average or just slightly better. Our data
lyzing the WISC–III. According to Wechsler, suggests that for cognitively gifted children this may
intelligence is an aggregate and global entity defined as be the rule rather than the exception.
the “capacity of the individual to act purposefully, to
think rationally, and to deal effectively with his or her

G I F T E D C H I L D Q U A R T E R LY • W I N T E R 2 0 0 6 • V O L 5 0 N O 1 7
Downloaded from gcq.sagepub.com at MCMASTER UNIV LIBRARY on April 8, 2015
Vol 50(1)-Winter 2006 12/15/05 10:13 AM Page 8

W I S C – I I I V E R B A L / P E R F O R M A N C E D I S C R E P A N C I E S

Methods and models for interpretation of Method

WISC–III profiles vary from a global interpretation of
the Full Scale IQ score to more specific interpretation Participants
of the 12 separate subtests (Kaufman, 1994). Winner
(2000) suggests that the use of a global IQ score often A total of 161 elementary and middle school students
mistakenly leads to the assumption that gifted children were administered the WISC–III as part of a screening to
are gifted in all academic areas. Verbal-Performance dif- be admitted into an enrichment program for children
ferences are thought to yield critical information about suspected of giftedness. Most of the children were in the
children’s abilities, but the level of differences that fall semester of third grade, although there were also a
should be considered significant is a matter of debate. small number of children in grades two through seven, as
Although there are some children who demonstrate well. More specifically, the sample consisted of 153 third
giftedness across domains, referred to as notationally graders, 1 second grader, 2 fourth graders, 1 fifth grader,
gifted, there are many other gifted children with abili- 1 sixth grader, and 3 seventh graders. The sample
ties that are less balanced across domains. The magni- included 78 females and 83 males. All children included
tude of discrepancies is positively correlated with IQ in the study attained at least one IQ score (Full Scale,
(Detterman & Daniel, 1989; Legree, Pifer, & Grafton, Verbal, or Performance) of 130 or higher, with the range
1996), and previous work with gifted individuals con- of Full Scale IQs between 116 and 147 (Mean = 131, SD
firms the finding that V-P differences are typically = 6.11). The mean Verbal IQ was 136 (SD = 7.04) while
larger than reported in the test manual for the standard- the mean Performance IQ was 119 (SD = 10.59). The
ization sample as a whole, where the mean difference is sample of children was drawn from four elementary
10 points (Wechsler, 1991). schools and one middle school in a suburban, upper-mid-
dle-class school district in Long Island, NY. The data
Research conducted by Wilkinson (1993) examined
were collected over a period of approximately 6 years.
the profiles of 456 third-grade students with Full Scale
An initial sample of 672 students was selected based
IQs of 120 or above for Verbal-Performance discrepan-
upon very high performance on group achievement tests
cies using the Wechsler Intelligence Scale for
in mathematics or reading and formal teacher recom-
Children–Revised (WISC–R). The study revealed that
mendations from two different teachers for two consecu-
the mean V-P discrepancy, regardless of direction, was
tive years, which were required for consideration of
found to be 10.87 (SD = 8.64, range = 0–41), with 29%
admittance into a district-wide enrichment program.
of subjects showing a difference of 15 points or more.
When these two criteria were met, the student was given
Saccuzzo, Johnson, and Russell (1992) similarly exam-
an intelligence test, the WISC–III, which was adminis-
ined Verbal and Performance IQ discrepancies in 4,546
tered by a New York State certified school psychologist.
gifted African American, Caucasian, Filipino, and
Students who attained a score of 130 or higher on the
Hispanic children. Within their gifted sample, 35% of Verbal, Performance, or Full Scale IQ comprised the cur-
children with Full Scale IQs greater than or equal to 120 rent sample of 161 students. All children were required to
were found to have V-P discrepancies of more than 12 have parental consent before any testing was initiated.
points. Subjects were assigned a code to protect confidentiality,
The present study was designed to elucidate V-P and then their WISC–III Full Scale, Verbal, and
differences within an additional population of children Performance IQs were recorded.
suspected of giftedness. The study was inspired by a
highly observable phenomenon of large V-P discrep-
ancies found among those children with high Full Results
Scale, Verbal, and/or Performance IQs. It is hypothe-
sized that children with high Full Scale, Verbal, or Observed discrepancy scores ref lecting differences
Performance IQs will manifest large discrepancies between verbal and performance scores were computed
between their Verbal and Performance IQs. Further, it for children who received at least one IQ score (Full
is hypothesized that the V-P discrepancy found within Scale, Verbal, or Performance) of 130 or greater. Table 1
this gifted population will be much greater than the reports the actual frequencies and cumulative percentages
discrepancy and cumulative percentages of the stan- of this gifted sample obtaining various Verbal and
dardization sample. Performance IQ discrepancies.

8 G I F T E D C H I L D Q U A R T E R LY • W I N T E R 2 0 0 6 • V O L 5 0 N O 1
Downloaded from gcq.sagepub.com at MCMASTER UNIV LIBRARY on April 8, 2015
Vol 50(1)-Winter 2006 12/15/05 10:13 AM Page 9

W I S C – I I I V E R B A L / P E R F O R M A N C E D I S C R E P A N C I E S

T a b l e 1 discrepancy observed in our population ranged from 0 to

51 points. The mean V-P discrepancy was 18.5, the
Cumulative Percentages and Frequency median was 19, and the mode was 21.
of the Verbal-Performance IQ Differences According to the WISC–III manual Table B.2, a dis-
for Gifted Sample
crepancy of 13 points or more occurred in 31.7% of the
Amount Frequency Sample standardization sample (Wechsler, 1991, p. 262). A dis-
of discrepancy percentage crepancy of this magnitude is often considered to be stat-
51 1 0.6 ically significant. Results from the present study indicate
43 1 1.2 that a discrepancy of 13 points or more occurred in
38 6 5.0 68.9% of the gifted sample (see Table 1). Further, a dis-
36 1 5.6 crepancy of 18 points or more occurred in 17% of the
35 4 8.1 standardization sample (Wechsler, 1991, p. 262), while a
33 5 11.2 discrepancy of 18 points or more occurred in 54.7% of
32 3 13.0 the gifted sample. A discrepancy of 18.5 points was the
31 1 13.7 average discrepancy for the gifted sample. Lastly, a dis-
30 4 16.1 crepancy of 25 or more points occurred in 4.9% of the
29 5 19.3
standardization sample, whereas a discrepancy of this size
28 6 23.0
occurred in 26.7% of the gifted sample. This discrepancy
27 2 24.2
26 2 25.5 occurred approximately five times more often in our
25 2 26.7 gifted sample as compared to the standardization sample.
24 9 32.3
23 7 36.6
22 5 39.8 Discussion
21 11 46.6
20 2 47.8 The results of this analysis illustrates that the large V-P
19 6 51.6 discrepancies observed in our sample of gifted children are
18 5 54.7 quite different than the magnitude and frequency of dis-
17 2 55.9
crepancies found in the standardization sample. When these
16 5 59.0
profiles are examined through the use of the WISC–III
15 6 62.7
14 5 65.8 manual (Wechsler, 1991), the discrepancies appear to be
13 5 68.9 rare. However, when examined within this unique popula-
12 2 70.2 tion, one can see that large discrepancies are, in fact, quite
11 7 74.5 typical. Both hypotheses were supported in this study.
10 2 75.8 Children in our sample with Full Scale, Verbal, and/or
9 3 77.6 Performance IQs equal to or above 130 manifested large dis-
8 7 82.0 crepancies between their Verbal and Performance IQs.
7 3 83.9 Additionally, these discrepancies were much greater than
6 6 87.6 those reported in the WISC–III manual (Wechsler, 1991).
5 5 90.7
This suggests that one should be careful when inter-
4 2 91.9
preting the V-P difference of gifted children. Perhaps the
3 1 92.5
2 5 95.7 most important point to be gleaned from the present data
1 6 99.4 is to exercise caution and good clinical judgment when
0 1 100 interpreting psychometric data. The use of local norms or
perhaps specialized, subset norms can sometimes lead to
M 18.5
SD 10.6 significantly different conclusions. What may hold true
Mdn 19.0 for the general population of the normative sample may
require more intense scrutiny than is more commonly
Descriptive analysis was conducted in order to employed when working with specific subsamples (i.e.,
explore the various V-P IQ discrepancies within this gifted children). A thorough analysis of observable pat-
population. Results indicated that the amount of V-P terns and trends can often lead to more accurate inter-

G I F T E D C H I L D Q U A R T E R LY • W I N T E R 2 0 0 6 • V O L 5 0 N O 1 9
Downloaded from gcq.sagepub.com at MCMASTER UNIV LIBRARY on April 8, 2015
Vol 50(1)-Winter 2006 12/15/05 10:13 AM Page 10

W I S C – I I I V E R B A L / P E R F O R M A N C E D I S C R E P A N C I E S

pretations. Blind analysis utilizing testing manuals can, at and Interpretive Manual states that, “. . . the VCI and
times, lead to misleading interpretations, although PRI should be substituted for the VIQ and PIQ in clini-
arguably they may be statistically accurate. For this trun- cal decision-making and other situations where the VIQ
cated sample, misinterpretations of the magnitude and and PIQ were previously used or required” (Wechsler,
frequency of V-P discrepancies may include the sugges- 2003, p. 6). The principal author is currently examining
tion of a learning disability, even though weaknesses are the frequencies and degree of V-P discrepancies found
average, high average, or better. Our data indicate that within a similar sample to this study, using the WISC–IV
these aforementioned large discrepancies are, in fact, no (Wechsler, 2003). Future research might also investigate
way indicative of abnormality within this subsample. real-life correlates amongst a sample of gifted children
The causes of the finding that gifted children are with very large V-P discrepancies. Is it merely represen-
more likely to have a large discrepancy between abilities tative of distinct interest and talents? Additionally, are
is an area of interest that this study, as well as those in the there behavioral, affective, and/or personality attributes
past, brings to mind. Kaufman (1992) cited the timed that are related to distinct patterns of cognitive strengths
nature of the performance sections of the WISC–III and weaknesses?
(Wechsler, 1991) as a possible factor. He interpreted pro-
cessing speed as a penalizing factor for gifted children
who are ref lective in nature and hence tend to work at a References
slower pace. Processing speed is not as large of a factor on
the WISC–IV (Wechsler, 2003) and will be a point of Detterman, D. K. (1987). Theoretical notions of intelligence
and mental retardation. American Journal of Mental
interest in future research. Detterman (1987) suggested
Deficiency, 92, 2–11.
that high-IQ individuals show more subtest scatter
Detterman, D. K., & Daniel M. H. (1989). Correlations of
because, unlike their low-IQ counterparts, they are with-
mental tests with each other and with cognitive variables
out deficits in important central processes. They demon- are highest for low IQ groups. Intelligence, 13, 349–359.
strate more variability than low-ability individuals who Donders, J. (1996). Cluster subtypes in the WISC–III standard-
tend to operate on a more uniform, low level of opera- ization sample: Analysis of factor index scores. Psychological
tion. As one’s ability increases, so does the likelihood of Assessment, 8, 312–318.
variability in skills. Kaufman, A. S. (1992) Evaluation of the WISC–III and
The conclusions of the study are limited due to the WPPSI–R for gifted children. Roeper Review, 14, 154–158.
limitations of the sample. The sample only included chil- Kaufman, A. S. (1994). Intelligent testing with the WISC–III.
dren in grades three through six, with the vast majority New York: John Wilely Sons Incorporated.
coming from the third grade. Whether similar results Legree, P. J., Pifer, M. E., & Grafton, F. C. (1996). Correlations
among cognitive abilities are lower for higher ability
would occur at other ages is an empirical question that
groups. Intelligence, 23, 45–57.
the present data does not address; however, there is no
Macmann, G. M., & Barnett, D. W. (1997). A critical appraisal
theoretical reason to suspect that it would not hold up. of intelligence testing with the WISC–III: Introduction to
Given the relatively high test-retest reliability correlations the series. School Psychology Quarterly, 12, 193–196.
for the WISC–III (Wechsler, 1991), it appears more than Saccuzzo, D. P., Johnson, N. E., & Russell, G. (1992). Verbal
likely that these data would remain reliably consistent. versus performance IQs for gifted African-American,
The sample was also not diverse, in that minorities were Caucasian, Filipino, and Hispanic children. Psychological
underrepresented. In addition, IQs of 130 or above put Assessment, 4, 239–244.
participants in the top 2.5% of the population, limiting Wechsler, D. (1991). Manual for the Wechsler Intelligence Scale for
the generalizability of the study to that particular group. Children–Third Edition (WISC–III). San Antonio, TX: The
This study is also limited in that the normative sam- Psychological Corporation.
Wechsler, D. (1994). The measure of adult intelligence (3rd ed.).
ple of the WISC–III (Wechsler, 1991) is now out of date,
Baltimore: Williams & Wilkins.
with the new WISC–IV (Wechsler, 2003) now in use.
Wechsler, D. (2003). Manual for the Wechsler Intelligence Scale for
However, within the Technical and Interpretive Manual
Children–Fourth Edition (WISC–IV). San Antonio, TX:
of the WISC–IV (Wechsler, 2003), the discrepancy The Psychological Corporation.
between the Verbal Comprehension (VC) and the Wilkinson, S. C. (1993). WISC–R profiles of children with supe-
Perceptual-Reasoning (PR) indexes can be utilized in the rior intellectual ability. Gifted Child Quarterly, 37, 84–91.
same manner in which V-P discrepancies were utilized in Winner, E. (2000). The origins and ends of giftedness. American
analyzing the WISC–III. In fact, the WISC–IV Technical Psychologist, 55, 159–169.

1 0 G I F T E D C H I L D Q U A R T E R LY • W I N T E R 2 0 0 6 • V O L 5 0 N O 1
Downloaded from gcq.sagepub.com at MCMASTER UNIV LIBRARY on April 8, 2015