--::":.

THE—”49

 

This is to certify that the

thesis entitled

A SEARCH FOR HOMOGENEOUS SUBGROUPS OF LEARNING DISABLED
CHILDREN USING PSYCHOLOGICAL TESTS

presented by

Kathryn Scholes Bolt

has been accepted towards fulfillment
of the requirements for

Ph.D. d . Counseling, Educational
cgree ui .
‘PEVEfiGTofiV, and Specual
Education

 

/flélu% fgg/Lwé;

jor professor

Dr. Harvey Clarizio

Date s/z/M

0-7639 MSU is an Affirmative Action/Equal Opportunity Institution

 

 

 

MSU

 

 

RETURNING MATERIQL§r
Place in book drop to

 

 

g: ‘2)”: if»? ~
iwgwaizdk tad?

 

LIBRARJES remove this checkout from

-:-__ your record. FINES will
be charged if book is
returned after the date
stamped below.

 

 

A SEARCH FOR HOMOGENEOUS SUBGROUPS OF LEARNING DISABLED
CHILDREN USING PSYCHOLOGICAL TESTS

By

Kathryn Scholes Bolt

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Counseling. Educational Psychology.
and Special Education

1984

© 1984

KATHRYN SCHOLES BOLT
All Rights Reserved

ABSTRACT
A SEARCH FOR HOMOGENEOUS SUBGROUPS 0F LEARNING DISABLED
CHILDREN USING PSYCHOLOGICAL TESTS
By

Kathryn Scholes Bolt

Four subgroups of learning disabled children were identified on
the basis of a combination of WISC-R Verbal IQrPerformance IQ
discrepancy and WRAT Reading. Spelling. and Arithmetic subtest scores.
The subjects were identified from a group of school-identified learning
disabled students who had received two psychological evaluations over a
2-to-4-year period. A retrospective study was conducted to determine
if changes in WRAT scores at the second evaluation could be predicted
on the basis of subgroup membership. The findings were not significant
and raised questions regarding the potential for using these tests to
identify homogeneous subgroups in the learning disabled population.
WISC-R Verbal IQ. Performance IQ. and Full Scale IQ were all found to
be quite stable for this population. An effort was made to relate the
"ACID" profile to performance on the WRAT Reading subtest. but too few
subjects actually exhibited this profile to make any such inferences.
Finally. the subjects were grouped on the basis of special education
placement to determine if the severity of the ability-achievement

discrepancy was related to the type of special education placement.

Kathryn Scholes Bolt

Severity of the discrepancy appeared unrelated at the initial evalua-
tion. At the re-evaluation. more severe discrepancies were found for
the children placed in resource rooms and self-contained classrooms

than for children placed in teacher consultant programs.

TABLE OF CONTENTS

Page
LIST OF TABLES O O O O O O O O O O O O O O O O O O O O O O O O 0 1V
LIST OF FIGURES O O O O O O O O O O O O O O O O O O O 0 O O O 0 v
Chapter
I. SCOPEANDPURPOSEOFTHESTUDY............. 1
II. THEORY AND RESEARCH O O O O I O O O O O O O O O O O O O 13
Review of the Research Literature . . . . . . . . . . 13
Major Strategies Using the MISC-R for Diagnosis
of Learning Disabilities . . . . . . . . . . . . . 13
Diagnostic Utility of the Verbal IOrPerformance IQ
D1screpancy O O O O O O O O O O O O O O O O O O O 14
Diagnostic Utility of Subtest Scatter . . . . . . . 15
Search for a Characteristic WISC-R Profile . . . . . 16
Subgroups of Learning Disabled Children . . . . . . . 19
Subgroups Based on VIQrPIQ Discrepancy . . . . . . . 19
Subgroups Based on Profile Analysis . . . . . . . . 23
Subgroups Based on WRAT Subtest Scores . . . . . . . 26
Subgroups Based on Multivariate Classification . . . 28
Use of the ACID Profile as a Predictor of School
Performance . . . . . . . . . . . . . . . . . . . . 31
Partial ACID Profile Research Findings . . . . . . 33
Stability of W150 and WISC-R Scores for the
Learning Disabled Population . . . . . . . . . . . . 34
Critical Summary of Previous Research and Implica-
tions for the Present Study . . . . . . . . . . . . 40
III. ”En-[OD O O O O O O O O O O O O O O O O 0 O O O O O O O O 42
Rationa‘e O O O O O O O I O O O O O O O O O O O O O O 42
Hypotheses and Research Questions . . . . . . . . . . 44
Def1n1t1°ns O O O O O I O O O O O O O O O O O 0 O O 0 46
SUbJeCts O O O O O O O O O O I O O O O O O C O O O O O 48

Instruments . . . . . . . . . . . 50

The Wechsler Intelligence Scale for Children-

Revised . . . . . . . . . . . . . . . . .

The Wide Range Achievement Test . . . . . .
Procedures . . . . . . . . . . . . . . . . . .
Analysis of the Data . . . . . . . . . . . . .
Limitations and Assumptions of the Study ..

Iv. RESULTS 0 O O O O O O O O O O O O O O O O O O O

Stu dy one 0 O O I O O O O O I O O O O O O O 0

Analysis of Subgroups Membership Based on WISC-R

and WRAT Scores . . . . . . . . . . . . .
StUdy Two 0 O O O O I O O O O O O O C O O O 0

Analysis of the Frequency of the ACID Test and

Related Patterns . . . . . . . . . . . . .
Analysis of Stability of WISC—R Scores . . .,

Relationship of Special Education and Ability-

Achievement Discrepancy . . . . . . . . .
V. DISCUSSION . . . . . . . . . . . . . . . . . . .

overVTBW O O O O O O O O O O O O O O C O O O 0

Utility of Subgroups of Learning Disabled Children .

Recommendations for Future Research . . . .
Predictions Based on Subgroup Membership . . .
Use of the ACID Test as a Predictor of School

Performance . . . . . . . . . . . . . . . .

Recommendations for Future Research . . . .
Stability of WISC-R IQ Scores for Learning

Disabled Children . . . . . . . . . . . . .

Recommendations .. .. .. .. . .. . ..
Relationship of Special Education Placement and

Ability-Achievement Discrepancy . . . . . . . . . .
Recommendations for Future Research . . . . . . . .

REFERENCES 0 O O O O O C O O O O O O O O C O O O O O C O O O C O

82

82
82

86

87
87

88
9O

9O
92

93

Table

2.1

4.]

4.2

4.3

4.4

4.5
4.6

4.7

4.8

4.9

LIST OF TABLES

Summary of Studies Investigating Subtypes of
Learning Disabled Children . . . . . . . . . . . .

Subgroup Membership With 15- and lZ-Point Standard
Score Differences . . . . . . . . . . . . . . . . .

x_Cell FSIQ Scores and Standard Deviations by
WRAT Grouping . . . . . . . . . . . . . . . . . . .

Cell x WRAT Arithmetic Analysis of Covariance Using
Estimated True Scores . . . . . . . . . . . . . . .

Cell Means and Standard Deviations on the Two
Administrations of the WRAT . . . . . . . . . . . .

Correlation Between WRAT l and WRAT 2 . . . . . . . .
Frequency of ACID Pattern and Related Patterns . . .

IQ Score Means and Standard Deviations on the Two
AdIII‘InlStf'athnS Of the "ISO-R o o o o o o o o o o o

Correlated t—Tests and Mean Differences of IQ Scores

Correlated 1-Tests and Mean Differences of IQ Scores
Based on the Regrouping of the Subjects . . . . . .

Changes in 10 Category at Re-evaluation by VIOrPIQ
Discrepancy Groups . . . . . . . . . . . . . . . .

Test-Retest Reliability Coefficients for the MISC-R IQ
sea] as I O O O O O O O O O O O O O O O O I O O O 0

Means. Standard Deviations. and ANOVAs for Initial
Special Education Placement Based on the MISC-R -
WRAT Discrepancy Scores at the Initial Evaluation .

Means. Standard Deviations. and ANOVAs for Initial
Special Education Placement Based on the WISC-R -
WRAT Discrepancy Scores for Placement at the
at the Re-evaluation ....... .

Page

32

61

63

65

66
68
69

71

72

74

75

76

78

79

Page

4.14 Means. Standard Deviations. and ANOVAs on the WISC-R -
WRAT Discrepancy at the Re-evaluation . . . . . . . . . 80

LIST OF FIGURES

Figure Page
3.1 Subgrouping of Learning Disabled Children . . . . . . . 44

4.1 Interaction Between WRAT Profile and VIQPPIQ
D1screpancy O O O O O O O O O O I O O O O O O O O O O 64

vi

CHAPTER I

SCOPE AND PURPOSE OF THE STUDY

Results of studies over the last 45 years indicate that there is a
group of children who have unusual difficulty learning when exposed to
traditional teaching methods. although they do not exhibit signs of
comumnihandicapping conditions such as sensory impairment or
retardation (Johnson & Morasky. 1977; Kirk. 1972; Strauss & Lehtinen.
1942; Strauss & Werner. 19420. One such subgroup of underachievers is
commonly described as having specific learning disabilities. and these
children are being identified with increasing frequency as a result of
mandatory special education.

The recognized characteristics that previously have been required
for substantiating the diagnosis of a specific learning disability are:
(1) an educationally significant discrepancy between measured
functioning capacity and actual performance in a given achievement area
and (2) a greater amount of scatter in measured learning abilities due
to a difference in information-processing abilities (Farnham-Diggory.
1978; Johnson & Morasky. 1977; Owens. Adams. Forrest. Stolz. & Fisher.
1977).

The amount of scatter has not proved an effective diagnostic tool.

however. ‘The difficulty may lie in the heterogeneity of the group used

to evaluate the scatter. Although learning disabled children as a
group do not differ from other populations on the amount of scatter
(Kaufman. 1979). some subgroups of learning disabled children may
exhibit greater scatter in abilities than others.

In addition to the two generally accepted criteria for identifying
a learning disability. various researchers have reported on other
characteristics they associate with learning disabled children.

It has been mandated that the learning disabled child has average
or above-average intelligence. adequate sensory acuity but is achieving
much less than the chilcfls age. intelligence. and educational
background would predict (Gearhart. 1973). Valett (1969) wrote that
learning disabled children have specific difficulty in acquiring and
using information and skills needed for problem solving. Another
aspect of learning disability has been described as integrity of
emotional. motoric. sensory. and intellectual abilities with failure to
achieve at an appropriate level (Johnson & Myklebust. 1967). Finally.
Kirk (1972) wrote that the children display discrepancies in ability.
have a specific problem not related to other primary handicapping
conditions. and have behavioral deficits.

It should be noted that the above statements focus more on
exclusion of certain characteristics than the presence of specific
characteristics. Clarizio and McCCy (1983) suggested that learning
disabled children exhibit secondary or associated difficulties in one

or more of the following areas: visual-spatial skills. fine and gross

motor skills. language skills. auditory discrimination. social and
emotional skills. and. finally. in cognitive skills.

One problem in interpreting the literature is that there has been
a proliferation of both theories and formulas for the process of
diagnosis. Many researchers have used their own definition of learning
disability in determining what constituted a learning disabled child.
The authors of Pd” 94—142 attempted to remedy this in part by the
establishment of both a single definition and guidelines for diagnosis:

Those children who have a disorder in one or more of the basic
psychological processes involved in understanding or in using
language. spoken or written. which disorder may manifest itself in
imperfect ability to listen. think. speak. read. write. spell. or
do mathematical calculations. Such disorders include such
conditions as perceptual handicaps. brain injury. minimal brain
dysfunction. dyslexia. and developmental aphasia. Such term does
not include children who have learning problems which are
primarily the result of visual. hearing. or motor handicaps. or
mental retardation. of emotional disturbance. or environmental.
cultural. or economic disadvantage. (Section 5(b))

The Michigan Rule as stated in Public Act 451 includes the above
statement but adds the following qualifiers:

Rule 13. (2) The individualized educational planning committee
may determine that a child has a specific learning disability if
the child does not achieve commensurate with his or her age and
ability levels in 1 or more of the areas listed in this subrule.
when provided with learning experiences appropriate for the
child's age and ability levels. and if the multidisciplinary
evaluation team finds that a child has a severe discrepancy
between achievement and intellectual ability in l or more of the
following areas:

(a) Oral expression.

(b) Listening comprehension.
(c) Written expression.

(d) Basic reading skill.

is) Reading comprehension.
(f) Mathematics calculation.
(9) Mathematics reasoning.

(3) The individualized educational planning committee can not
identify a child as having a specific learning disability if the
severe discrepancy between ability and achievement is primarily
the result of any of the following:

(a) A visual. hearing. or motor handicap.

(b) Mental retardation.

(c) Emotional disturbance.

(d) Environmental. cultural. or economic disadvantage.

As can be seen from the above definition and rules. the essential
characteristic for diagnosis is the presence of a significant discrep-
ancy between expected achievement and actual achievement. There con-
tinues to be much room for interpretation of the meaning of significant
discrepancy. however.

Given the broad area of learning disabilities. this paper will
focus on several areas in need of further investigation. These areas
include identification of subgroups of learning disabled children.
Wechsler Intelligence Scale for Children pattern analysis and its
relationship to achievement test performance. and the consistency of
test performance across evaluations.

The desire for consistency has led to numerous efforts to
establish a characteristic profile on the subtests of the Wechsler
Intelligence Scale for Children (W150) and the Wechsler Intelligence
Scale for Children-Revised (WISC-R) which would distinguish learning
disabled children from other populations. and identify their
characteristic strengths and weaknesses (Ackerman. Dykman. & Peters.
1976; Belmont & Birch. 1966; Clarizio & Bernard. 1981; Huelsman. 1970;
Law. Box. & Moracco. 1980; Smith. Coleman. Dokecki. & Davis. 1977aL

These studies have met with mixed results as either the profiles have

fit normal children as well as the identified learning disabled
children (Clarizio & Bernard. 1981). or it has not been possible to
replicate those studies which do suggest a characteristic profile
(Ryckman. 1981; Thompson. 1981; Vance & Singer. 1979). 'Thus. the
studies are consistent in their failure to find a characteristic
profile.

Several studies have suggested that while there may be no one
characteristic profile. certain subtests seem to occur in an associated
pattern of lower scores. These are the Arithmetic (A). Coding (C). and
Digit Span (D) subtests which together are often referred to as the
Freedom from Distractibility Factor (FDF) (Huelsman. 1970: Kaufman.
1975; Paal. Hesterly. & Wepfer. 1979L A.related pattern. referred to
as the ACID pattern as it includes the Information (I) subtest as well
as the Arithmetic. Coding. and Digit Span subtests cited above. has
also been described by several studies (Ackerman. Dykman. & Peters.
1976; Huelsman. 1970; Swartz. 1974).

There is some controversy regarding which tests actually
constitute the POP. Fisk and Rourke (1979) suggested that an AID
pattern may be more common among learning disabled students. while
Paal. Hesterly. and Wepfer (1979) reported that Information was not
found to be low. Tabachnick (1979) indicated that low scores on Coding
alone are significant in diagnosing the presence of a learning
disability. Bernard (1978) found that his group of learning disabled
subjects had their lowest performance on Arithmetic. Coding. and

Information. These subtests were also the most powerful at

discriminating between the learning disabled and the nonimpaired
groups. Rourke (1981) suggested that the ACID profile may be useful in
determining which learning disabled children are most likely to benefit
from remedial reading programs.

Another recent development coming out of the many efforts to find
a representative WISC or WISC-R pattern has been the recommendation
that researchers give up on this search and turn to investigations of
more homogeneous subgroups of learning disabled children in order to
identify such groups. more clearly define the learning problems they
face. and to begin developing appropriate educational programs for them
(Denckla. 1973; Kaufman. 1981; McKinney. 1984; Rugel. 1974; Ryckman.
1981; Vance. Wallbrown. & Blaha. 1978L These authors believed that
learning disabilities research has continued to focus on the character-
istic profile concept beyond a profitable point.

Another problem encountered in interpreting the literature on
learning disabilities is that researchers have tended to treat their
subject population as though they all share a common problem (Cratty.
1974L. A common approach has been to select children who have been
identified as learning disabled by a school psychologist for research
(Clarizio & Bernard. 1981: Koppitz. 1971: Paal. Hesterly. & Wepfer.
1979; Sattler & Ryan. 1981; Smith. 1978: Smith. Coleman. Dokecki. &
Davis. 1977). In those cases where an effort has been made to
differentiate among subgroups. the WISC/WISC—R Verbal Intelligence
Quotient-Performance IntelligencerOuotient discrepancy has been used

for grouping the subjects (Clements & Peters. 1962; Paal. Hesterly. a

Wepfer. 1979; Rourke. Young. & Flewelling. 1971; Wells. 1970). Recent
studies have indicated significant differences between the Verbal and
Performance intelligence quotients are fairly comnmuiin normal students
(Bortner. Hertzig. & Birch. 1972; Kaufman.l976a; Vance & Singer. 1979).
thus giving this comparison little diagnostic utility although it may
be descriptively useful (Bernard. 1978; Veres. 1982).

The question of what happens to learning disabled children over the
years has been explored in several longitudinal studies (Eaves &
Crichton. 1975; Eaves. Kendall. a Crichton. 1974: Fox & Routh. 1983;
Kaufman. 1981; Koppitz. 1971: Smith. 1978: Vance. Blixt. Ellis. &
Debell. 1981; Werner & Smith. 1977; White. Alley. Deshler. Schumaker.
Warner. & Clark. 1982; Levin. Zigmond. & Birch. 1983L. These studies
described the long-term effects in terms of academic achievement.
emotional and social adjustment; All of the above indicate continuing
problems in at least one of these areas.

One limitation of the Werner and Smith and the Eaves and Crichton
studies is that they did not describe the continued problems in terms
of standardized measures such as intelligence or achievement tests.
Rather. they used reports of teachers. parents. and the students
themselves.

The Koppitz study did report longitudinal data on the WISC and the
WRAT; These data showed both gains and losses in Verbal IO. Perform-
ance IO. and Full Scale IO with varying gains on the WRAT from age
group to age group. Again. heterogeneous groups of learning disabled

children were used. which may account for the inconsistent findings.

White. Alley. Deshler. Schumaker. Warner. and Clark (1982) com-
pared a group of normal adults and learning disabled adults in terms of
their adjustment on vocational. social. personal. community. medical.
and educational variables. The learning disabled sample reported less
satisfaction with their education. were less involved in social groups.
and took more prescription medicine. The authors concluded that while
the two groups were generally comparable. the lack of satisfaction of
the LD sample needs further exploration.

Fox and Routh (1983) compared groups of normal readers and
retarded readers in the first grade and again 3 years later. They
found that children who lacked phonemic segmenting ability continued to
exhibit significant difficulties in reading and were found to be
dysphonetic spellers 3 years laten. These subjects were compared
on the Camp-Dolcort version of the Boder reading and spelling test.

Levin. Zigmond. and Birch (1983) followed up on 52 learning dis-
abled students after 4 years. ‘The greatest progress was made in the
first year after identification as learning disabled. Mathematics
showed much less growth than reading. as measured by the Peabody Indi-
vidual Achievement Test.

Kaufman (1981) suggested that Verbal IO scores may go down after
time in learning disabled children. Smith (1978) retested a group of
learning disabled children after 7 months on the WISC-R with the result
of a four-point rise in Performance IQ while Verbal IO remained
unchanged. Vance. Blixt. Ellis. and Debell (1981) retested a group of

children diagnosed as learning disabled and educable retarded after 2

years and reported that the Performance IO went up two points and the
Verbal IO went down two points. These results indicate that no
significant change was observed.

One point that the longitudinal studies have all agreed on is that
children who are diagnosed as learning disabled appeared to have fairly
persistent problems in academic areas. whether or not treatment is
provided. In the Koppitz study. the children were assigned to learning
disabled classrooms where they remained for l tots years. She found
varying amounts of change. ranging from a few months to one grade level
for those who remained in L0 classrooms throughout the 5 years.
recording 3 or 4 months gain for each year in the class. She suggested
that the child's functioning at the time of admission has a direct
bearing on how long the child will remain in special education. In the
Werner and Smith study. the students did not receive any consistent
form of treatment. although four-fifths had contact with at least one
community agency. They reported that improvement was primarily related
to early identification and continual emotional support to maintain
self-esteem. Eaves and Crichton (1975) suggested that their population
may represent a severely involved subject sample due to the procedure
of selecting children who had been admitted to a diagnostic clinic. as
compared to using a random selection procedure. Of the 39 subjects.
half were placed in special education for a portion of their schooling.
11% received remedial help while being mainstreamed. 19% repeated a
grade. and 21% were not receiving supportive services. The disabled

readers in the Fox and Routh study were all held back 1 year. but no

10

other mention is made of supportive services. Limitations associated
with basic academic skills such as reading. coupled with poor ability
to conceptualize. seem to lead to secondary problems in social and
emotional areas (Koppitz. 1971; Werner & Smith. 1977). Once again. it
is not clear what percentage or what subgroups of learning disabled
children continued to have significant problems over time.

The present study was designed in two parts: Study I was designed
to identify homogeneous groups of learning disabled children. Study II
was designed to compare the subgroups identified in Study I: (l) to
investigate whether particular combinations of skills and/or weaknesses
change over time and (2) to determine whether profile analysis is
useful for the'prediction of future performance. The independent
variable in the study was the combination of WISC-R Verbal IO and
Performance IO scores and the Wide Range Achievement Test (WRAT)
subtest scores. 'The dependent variables were changes in the WISC-R
subtest and IQ scores and the WRAT subtest scores at the time of the 3-
year reevaluation. A second independent variable was the presence or
absence of the ACID profile with a dependent variable of the amount of
change on the Reading Subtest of the WRAT. A third independent
variable was the type of special education placement with a dependent
variable of the severity of the ability-aptitude discrepancy.

Five research objectives of Study II were identified. The first
objective of the research study was to determine if there are differ-

ences among homogeneous groups of learning disabled children on the

11

amount of change observed over time on a standardized achievement test
with each child serving as his own control.

A second objective was to determine if it is possible to predict
the degree of change on the achievement subtests based on the initial
pattern of WISC-R Verbal IO-Performance IO discrepancy and the initial
WRAT subtest pattern. Identification of students who make minimal
progress academically as compared to those who make significant
progress was seen as an important first step in determining which
students are at greater risk for continuing school failure.

A third objective was to determine if those students who exhibit
the ACID profile differ in reading achievement over time as compared
to those who do not exhibit the profile. regardless of the pattern
analysis mentioned in the above objective. Again. the ability to
predict which students are more likely to continue to be limited has
implications for school programming. Two related questions are
(1) whether the child who exhibits this pattern at the time of the
initial evaluation continues to exhibit the same pattern at the time of
later evaluations and (2) whether all four of these subtests are useful
predictors of reading achievement. This knowledge would help educators
to determine which learning disabled children have poor prognoses and
are at the greatest risk for continued reading failure.

A fourth objective was to determine the stability of the learning
disabled child's intelligence quotients. This was considered important
as the IQ score is generally considered to be a measure of the child's

basic ability against which one's level of achievement is compared in

12

order to determine if a significant discrepancy exists. ‘This informa-
tion is used both to determine educational placement of learning dis-
abled children and need for continuation in a special education
program.

A fifth objective was to determine if the severity of the
discrepancy between the child's ability and achievement was related to
the type of classroom setting in which the child was placed. It seems
logical that those children having the most severe discrepancy would be
placed in a program with a greater amount of time spent in the special
education placement than those with a less severe discrepancy; however.

this has not been studied previously.

CHAPTER I I

THEORY AND RESEARCH

This chapter focuses on the literature in the field of learning
disabilities as it relates to the objectives of the present study. The

chapter concludes with a critical summary of studies reviewed.

Wm

The review of the literature is divided into six sections. The
first three sections focus on research on the major diagnostic
strategies for identifying learning disabled children using the WISC
and WISC-R as they relate to the objective of Study I. Those studies
designed to investigate the presence of subtypes of learning disabled
children are reviewed in the fourth section as they relate to the
objective of Study I and the first two objectives of Study II. The
last two sections are based on the third and fourth objectives of Study
II. ‘These studies were selected as they provide background and format
for the present study.

WE
19W

As has been described earlier. numerous studies have sought to

clarify and specify the utility of the Wechsler Intelligence Scale for

l3

14

Children (WISC) and the Wechsler Intelligence Scale for Children—
Revised (WISC-R) for distinguishing normally achieving students from
those with a learning disability. 'Three major strategies have been
investigated for this purpose: (1) comparison of Verbal IQ and Per-
formance IQ scores for the presence of a significant difference.

(2) use of scatter analysis. and (3) profile analysis of the subtests.

WWW
W

The use of the Verbal IQ-Performance IQ discrepancy for diagnosing
the presence of dysfunction has been popular since the publication of
the WISC. The research in this area is confusing and contradictory.
It.has been commonly assumed that large differences are indicative of
neurological dysfunction. Anderson. Kaufman. and Kaufman (l976)
compared the average discrepancy found for a group of learning disabled
children to that found for normal children. 'They reported that while
differences were statistically significant. they were not meaningfully
significant. ‘The mean Verbal IQePerformance IQ discrepancy for the
learning disabled children was approximately three points greater than
that of the normal children. Vance. Gaynor. and Coleman (1976)
examined a group of 60 learning disabled children on the WISC-R.
Analysis of the data using a test for differences between Verbal and
Performance IQ scores was not significant. Vance at al. concluded that
the Verbal-Performance IQ.discrepancy is not useful as an index for the
possible diagnosis of learning disabilities. Stevenson (l979)

evaluated 55 children referred to a psychoeducational clinic. While she

15

found a greater proportion of these children had significantly higher
Verbal IQ scores than Performance. she noted that overall. more of the
children had somewhat higher Performance IQ.scores. She suggested that
high Performance-low Verbal IO.scores are a valid clinical sign if one
ignores the question of statistical significance. Paal. Hesterly. and
Wepfer (l979) assessed 40 learning disabled children on the WISC-R. and
their findings also supported the use of Verbal-Performance differ-

ences. Obviously. this continues to be a controversial question.

WWW
Another theory holds that learning disabled children exhibit an

unusual amount of scatter or irregular performance on the subtests of
the WISC-R as compared to normal children (Clements. 1966). Kaufman
(l976b) analyzed the test results of the 2.200 children used in the
WISC-R standardization sample and found that the average difference
between each child's highest and lowest scaled scores was seven points.
with a standard deviation of two points. ‘This indicated a much greater
range of variability should be expected with all children than was
previously thought.

Tabachnick (l979) compared a group of 105 learning disabled
children with Kaufman's normative data. She found a greater amount of
scatter for the learning disabled children in her sample. She reported
that the range was comparable to Kaufman's data on the Verbal subtests
as opposed to the range on the Performance subtests. Those subjects
who were younger than ll years old had greater scatter on the Verbal

subtests than those ll and older.

16

A similar study compared the scatter of 434 children referred to a
reading clinic to the WISC-R standardization sample and replicated
Kaufman's findings (Moore & Wielan. 198l). Tabachnick and Turbey
(198l) reported that a comparison of lOO learning disabled children to
Kaufman's data resulted in no significant differences. Ryckman (1981)
suggested that while it is possible that greater scatter exists for
learning disabled groups of children than normal groups. there is
clearly considerable overlap. This suggests that the demonstration of

greater scatter is of questionable value.

W
The third major approach to the diagnosis of learning disability

has been the search for a characteristic WISC-R profile. Bannatyne
(l97l) regrouped the subtests into three factors labeled Conceptual.
Spatial. and Sequencing. Using this method of reorganizing the
subtests. he suggested that groups of learning disabled children score
high on Spatial. lower on Conceptual. and lowest on Sequencing. He
later revised this to four factors: Spatial. Verbal Conceptual.
Sequencing. and Acquired Knowledge (1974). His theory about the order
of factors was supported by Smith. Coleman. Dokecki. and Davis (l977b)
in a study of 208 learning disabled children.

Vance and Singer (1979) attempted to replicate the Smith et al.
study and found that 39% of their group met the Spatial>Conceptual>
Sequential pattern. They reported that no unique or distinguishable

pattern was evident and found that the use of WISC-R recategorization

17

strategies is a weak method for diagnosis as there is such high indi-
vidual variation within groups.

Thompson (1981) compared a sample of 64 learning disabled to
14 mentally retarded children and El children with psychological
disorders. In his group. comparison of the learning disabled and
mentally impaired children showed a pattern of Conceptual>$patial>
Sequential. while the children with psychological disorders showed the
Spatial>Conceptual>Sequential pattern. Not only did this research fail
to support Bannatyne's hypothesis of a unique profile for groups of
learning disabled children. but in looking at the patterns for the
individual children. Thompson found no significant difference in the
frequency of this pattern among these three groups.

Law. Box. and Moracco (1980) also attempted to replicate the Smith
study without success. 'They indicated that this may be due to the
local nature of their sample and differences in the overall Verbal IQ-
Performance IQ patterns exhibited by the subjects.

Clarizio and Bernard (1981) also attempted to determine if the
three-factor grouping was present in the majority of learning disabled.
and whether it was useful in discriminating them from other groups of
handicapped and nonimpaired children. In a comparison of learning
disabled. educable mentally impaired. emotionally impaired. otherwise
impaired. and nonimpaired children. all of the groups with the
exception of the educable mentally impaired children exhibited the
Spatial>Conceptual>Sequencing pattern. again suggesting that

Bannatyne's recategorization is not a valid procedure for diagnosis and

18

placement. Again. these authors suggested that school-verified
learning disabled children are too heterogeneous a group to expect one
characteristic to be typical of all or even most of them.

Overall. it appears that the search for a characteristic profile
has not been particularly successful in improving the process of
diagnosis. While patterns and trends have been suggested. none have
held up in terms of either describing individual children or discrimi-
nating learning disabled children from other handicapped populations or
the nonhandicapped population. Several of these authors have suggested
that while this type of approach may not provide valid discriminators
for diagnostic purposes. it may be more useful in attempting to iden-
tify and describe homogeneous subgroups within the larger category of
learning disabled (Clarizio & Bernard. l98l; Kaufman. 198l; Tabachnick.
1979; Ryckman. l98lL

Kaufman (l979) presented several conclusions in using group data
such as above. First. the group data tell little about the specific
individuals comprising the group. although they do provide us with
information about the exceptionality in question. Second. group data
can mask two or more subprofiles. Lutey (l977) identified three
different subprofiles for reading disabled children based on her review
of WISC studies. Kaufman also raised the point regarding significant

scatter as mentioned earlier.

19

WW
WW9:

Several studies have identified subgroups of learning disabled
children for a variety of purposes. Reed (1968) compared the reading
achievement of 6-year-old children and 10-year-old children who were
divided into three groups on the basis of Verbal-Performance 10 score
discrepancies. Rank ordering for both age groups indicated that the
high Verbal-low Performance groups achieved the highest reading score.
while the Verbal-equal-to-Performance IQ group performed at the lowest
level. The differences were not significant for the 6-year-old
subjects. Doehring (1968) found similar results. as did Rourke
(1976). These authors suggested that the variables which distinguish
normal readers from retarded readers at younger ages are essentially
the same as those which differentiate the two groups at later ages.
Doehring indicated that visual motor skills may have relatively greater
importance than verbal skills in learning to read at the younger ages.

Rourke and Telegedy (1971) divided 4S 9-to-14-year-old boys into
three groups on the basis of their WISC Verbal IOrPerformance IQ
scores: high Performance-low Verbal. high Verbal-low Performance. and
Verbal equal to Performance. The authors hypothesized that the Verbal
IQrPerformance IQ discrepancies reflect the relative integrity of the
cerebral hemispheres and sought to test this hypothesis by comparing
their performance on a series of neuropsychological tests designed to
test lateralization. 'They predicted that the high Performance-low

Verbal group would perform at a higher level than the high Verbal-low

20

Performance group on psychomotor tests which involve complex visual-
motor coordination regardless of the hand employed. This would be
consistent with the view that these abilities are mediated by the right
hemisphere. They also expected to see better left-hand performance by
the high Performance-low Verbal group and the reverse for the high
Verbal-low Performance group. The Verbal-equa1-to-Performance group
was expected to fall somewhere between the above two groups. ‘The first
hypothesis was supported. However. there was no significant difference
between groups for right-hand or left-hand superiority. The authors
concluded that these results suggest that groups of older children who
exhibit varying Verbal IQrPerformance IQ discrepancies on the WISC also
exhibit patterns of abilities and deficits which are consistent with
theories on the relative specialization and intactness of the cerebral
hemispheres. ‘They also stated the Verbal IQrPerformance IQ relation-
ship is likely to be more important for predicting reading disabilities
than is general level of intelligence.

A similar study by Rourke. Young. and Flewelling (1971) identified
three groups of learning disabled children in the same manner as the
above study using 90 subjects. It was expected that the high Verbal-
low Performance group would do better on those tests which involve
verbal. language. and auditory-perceptual tests while the high
Performance-low Verbal group would do better on those tasks involving
visual perceptual skills. Again the Verbal-equal-to-Performance group
was expected to perform at an intermediate level. These predictions

were. in general. supported. These authors noted that an a posteriori

21

comparison of the Reading. Spelling. and Arithmetic subtests of the WRAT
indicated a striking difference between Reading and Spelling which were
high and Arithmetic which was low for the high Verbal-low Performance
group. These differences did not appear in the other two groups and
suggest that the WRAT Arithmetic subtest involves something more than
"verbal" ability. These results. in combination with the previous
study described. further support the relationship of Verbal IQ-
Performance IO discrepancies to lateralization of cerebral functioning.
Rourke. Dietrich. and Young (1973) established three groups of
learning disabled children in the same manner as above. In this study.
the children were 5 to 8 years old. while in the above two studies.
the children's ages ranged from 9 to 14 years old. This study was
designed as a developmental extension of the above two studies and used
the same measures as described above. A similar pattern as that
demonstrated by the older group was observed on the verbal. auditory-
perceptual. visual-perceptual. and problem-solving tasks. The
performance on the motor and psychomotor tasks did not resemble that
obtained for the older children. In comparison to the older group.
there were few significant differences among groups in the present
study. While the conclusion was made that these results do not support
the view that Verbal-Performance discrepancies reflect the differential
integrity of the cerebral hemispheres in younger children with
learning disabilities. this study did suggest that patterns of
strengths and weaknesses do persist through the two age groups. These

results were in part attributed to a greater variability of performance

22

of the younger children and the fact that other studies with similar
student populations have found lower WISC reliability coefficients than
do normal children.

Rie and Rie (l979)lcompared three groups of learning disabled
children with average Full Soale IQ scores who were divided according
to Verbal IQ-Performance IQ differences. The sample was composed of
56 6-to-lO-year-old children. Reading deficits were assessed by
computing grade equivalents and expected grade equivalents for each
subject on the Iowa Test of Basic Skills. ‘The groups were then
compared using a two-way analysis of variance. The reading deficits
were lowest for the high Verbal-low Performance group and greatest for
the low Verbal-high Performance group. ‘They found no differences based
on age. These results were consistent with the findings of Rourke.
Young. and Flewelling (1971) and suggested that general ability is not
a factor in comparing these groups. ‘This study also demonstrated that
the disparity in Verbal-Performance patterns. as it affects reading
skill. is observed in early to middle elementary-aged children as it is
in older learning disabled students and randomly selected groups.
Third. it suggested that even at early stages of reading in children
with learning disabilities. disparity between Verbal IQ and Performance
ID. in favor of the Verbal IQ. is associated with a significantly less
severe deficit than is a significant discrepancy in favor of the

Performance 10.

23

WW
Using a different approach. Vance. Wallbrown. and Blaha (1978)

used profile analysis in an effort to identify representative profiles
on the WISC-R. The subject population was made up of 128 reading
disabled children between 7 and 12 years of age. having Full Scale I05
greater than 85. They were able to identify five representative pro-
files. Fourteen percent of the population exhibited the distracti-
bility profile which was characterized by low scores on Digit Span and
Arithmetic. somewhat higher scores on Information. relatively high
scores on Similarities. Comprehension and Block Design. with average
scores on the other subtests. The perceptual organization profile was
found in 15% of the subjects. 'This group had low scores on Comprehen-
sion and Picture Arrangement. ‘The language disability-automatic group
was low on Digit Span and Comprehension and somewhat higher on Simi-
larities. Picture Completion. and Coding. ‘This comprised 21% of the
subjects. Language disability-pervasive was characterized by extremely
low scores on Similarities. moderately low Arithmetic and Digit Span.
low Information. and high scores on Mazes and Block Design. ‘This
pattern appeared in 9% of the cases. The fifth pattern. behavioral
comprehension and coding. exhibited very low scores on Coding. Picture
Arrangement. and Comprehension. low scores on Mazes and Information.
and higher scores on Similarities. Picture Completion. and Object
Assembly. This was seen in 16% of the population. Approximately one-
fourth of the subjects did not fit any of the above profiles. These

subjects ranged from having relatively flat profiles to having a

24

significant amount of variation. {As a result. these authors recom-
mended that caution be used when diagnosing children. ‘They suggested
these profiles have greater utility for developing remedial programs
than for identification purposes.

In a second study using 104 of the subjects from above. Wallbrown.
Vance. and Blaha (l979) reanalyzed the data and found that 75% of the
subjects had profiles which clearly fit into one of the above-described
ability patterns. Fourteen percent seemed to split evenly between
ability patterns or showed minimal similarity. ‘The other 24 subjects
of the first study were eliminated from this one as their profiles were
flat or showed minimal dispersion. They again suggested that these
profiles are primarily useful for developing remedial hypotheses.

One portion of Stevenson's 1979 study of 55 children referred for
educational evaluation analyzed the subtest patterns on the WISC-R in
an effort to delineate probable learning process correlates. This
analysis was based on Witkin's Verbal Comprehension. Analytic Field
Approach. and Attention-Concentration Factors (Witkin. Dyk. Paterson.
Goodenough. & Karp. 1962L In comparing the three factors. only the
Attention-Concentration factor was found to be significantly lower. 4A
comparison was also made across grade levels. and again a significant
decrease was noted on the Attention-Concentration factor with
increasing age.

Tabachnick and Turbey (1981) selected a group of 95 children
ranging in age from 6 years 6 months to 16 years 10 months who had been

diagnosed as learning disabled on the basis of difficulty in school.

25

performance on psychodiagnostic tests. and judgment of experienced
psychodiagnosticians. 0n the basis of these data. each child was
assigned to one of nine categories describing the area of primary
disability. Later. however. it appeared that some of the categories
were too rare to be useful while others were difficult to distinguish
between. As a result. the categories were reduced to the following
three categories: (1) a category combining visual-motor and visual-
perceptual disability (Visual) containing 66 of the subjects: (2) a
category combining auditory-perceptual and receptive language
disability (Audlang) containing 18 of the subjects: and (3) Memory.
containing 18 children who showed all types of sequential memory diffi-
culties. These groups were compared on three variables: (1) subtest
scatter. (2) Bannatyne scores. and (3) subtest profiles. 0f the first
two variables. only the total range of scaled scores and the Verbal IOr
Performance IQ discrepancy were found to be significant. Greatest
scatter was found for the visual group and least for the auditory
language group. Classification efforts resulted in only 50% correct
classifications. A separate profile analysis of the 11 subtests
revealed no evidence of significant differences among the three groups.
They concluded that WISC-R subtest scatter is not of diagnostic value
in the discrimination of subtypes of learning disabled children. nor
can the form of the learning disability be determined with recategori-
zation into Bannatyne groups. Profile analysis was also thought to

lack utility in designing treatment programs.

26

MW
Rourke and Finlayson (1978) divided 45 9-to-l4-year-old children

with learning disabilities into three groups on the basis of their
patterns of reading. spelling. and arithmetic as measured by the WRAT.
The WISC IQ range was 86 to 114. The primary question being studied
was whether children who exhibit varying patterns of academic abilities
would exhibit unique. meaningful. and consistent patterns of verbal and
visual-spatial behaviors. A related hypothesis was that if the above
theory was supported. these findings might also shed some light on the
nature of the brain-related abilities that are necessary for perform-
ance on reading. spelling. and arithmetic. It was predicted that
children who had poor arithmetic performance with adequate reading and
spelling skills would be relatively deficient in visual perceptual and
visual spatial abilities. The three groups were: (1) relatively
.deficient in reading. spelling. and arithmetic: (2) relatively defi-
cient in reading and spelling as compared to arithmetic: and (3) rela-
tively deficient in arithmetic as compared to reading and spelling.
The last two groups were actually somewhat matched on arithmetic.
These groups were compared on the subtests of the WISC and several
other neuropsychological tests designed to measure either verbal and
auditory perceptual functioning or visual spatial and visual perceptual
functioning. These hypotheses were supported. Groups 1 and 2
performed in a manner similar to that expected for groups of older
children who exhibit a low Verbal-high Performance 10.pattern on the

WISC. while Group 3 was comparable to those who exhibit a high

27

Verbal-low Performance pattern. Rourke and Finlayson emphasized that
the differential patterns chosen for examination were far more salient
with respect to the prediction of patterns of performance than were the
levels of performance on the control variables. Again the finding that
the younger group performed in a manner similar to the older group
suggests a persistence of the pattern of strengths and weaknesses over
time.

Rourke and Strang (1978) carried out a similar study in which 45
learning disabled children were divided into three groups in the same
manner as above. (However. this study proposed to examine inferences
regarding differential hemispheric integrity by evaluating the patterns
of performance on motor. psychomotor. and tactile-perceptual tasks for
the three groups. ‘This was considered important. as in the previous
study. the two groups who had been equated for deficient arithmetic
performance exhibited quite different performances on verbal and
visual-spatial tasks although their levels of performance were quite
similar. The conclusion was drawn that Group 3 displayed a marked
deficiency relative to Groups 1 and 2 on some psychomotor and tactile
perceptual skills. ‘The authors also suggested that children with low
arithmetic difficulty and high reading and spelling skills are probably
not identified at as early an age as children with other types of
learning disabilities. The relatively stronger reading and spelling
abilities were thought to mask the arithmetic weakness. Also. a

specific arithmetic impairment is often not attributed to brain-related

28

deficiencies in the same manner as reading and spelling problems and is
expected to improve with time rather than remediation.

Nolan. Hammeke. and Barkley (1983) attempted to determine whether
subtypes of learning disabled children would exhibit unique neuropsy-
chological profiles on the Luria-Nebraska Neuropsychological Battery as
have been found in previous studies using other neuropsychological
measures. ‘Their 36 subjects ranged from 7 to 13 years old and had Full
Scale WISC-R 105 of at least 80. Three groups of 12 subjects each
were formed: a normal control group whose WRAT Reading. Spelling. and
Arithmetic scores were 2 the 40th percentile; a reading/spelling dis-
abled group whose Reading and Spelling scores were 5 the 20th per-
centile and Arithmetic score was 2 the 40th percentile: and a math
disabled group whose Arithmetic scores were 3 the 20th percentile and
Reading and Spelling were 3 the 40th percentile. The study found only
partial support for previous research. No Verbal IQ-Performance IQ
discrepancies were found. nor was the math disabled group different
from the normals when equated for IQ level and age. ‘The reading/
spelling disabled group was comparable to previous studies measuring
impairments in linguistic functioning.

Wu
Won

Petrauskas and Rourke (1979) attempted to identify and describe
subtypes of reading disabled children in terms of differential patterns
of performance on a battery of neuropsychological measures including

the WISC. using a multivariate classification procedure. Data from

29

160 subjects between the ages of 7 years 2 months and 8 years 11 months
were collected. One hundred thirty-three were reading disabled. and 27
were normal readers. Using a 0 factor analysis. this resulted in the
identification of three subtypes of retarded readers. Type 1

contained the largest number of subjects (40). The characteristics
considered significant here were clear evidence of disturbance in
auditory-verbal memory and auditory-perceptual skills. ‘They showed the
greatest Verbal IOrPerformance IO discrepancy in favor of Performance.
and their WRAT Reading and Spelling scores were somewhat poorer than
was their Arithmetic score. Type 2 contained 26 subjects and was
characterized by minimal difference between Verbal and Performance 10
scores on the WISC: lowest subtest scores on Information. Arithmetic.
Digit Span. and Coding: and uniformly poor performance on all three of
the WRAT subtests. ‘Type 3 contained 13 subjects who appeared to have a
predominant deficit in conceptual flexibility. especially as it relates
to linguistic coding. This group had a lower Verbal IO than Perform-
ance IO. They also were lowest on Arithmetic. Coding. Information. and
Digit Span. WRAT performance was not reported for this group. Type 4
identified eight normal readers and was not produced reliably. ‘Type 5
also had a lower reliability than the first two. This group was simi-
lar to Type 2 with the exception of the WRAT subtests. Reading and
Spelling were lower than Arithmetic for these children. The authors
stated that this study demonstrates that retarded readers are not a
homogeneous group. and the profiles of these three groups suggest that

these subjects had significant deficits in skills which have been

30

suggested by previous researchers to be ”responsible" for reading
retardation. ‘They also suggested that this knowledge is important for
developing remedial programs which fit the student rather than vice
versa.

Fisk and Rourke (1979) used the 0 factor analysis in an effort to
identify subtypes of learning disabled children rather than strictly
reading disabled. This study had 264 subjects between the ages of 9
and 14.9 years with Full Scale 105 between 86 and 114. and obtained
centile scores < 30 on all three of the WRAT subtests. The subjects
were divided into three age ranges (9-10 years. 11-12 years. 13-14
years) to determine (1) if subtypes would be observed at different age
levels and (2) if some of the subtypes would be more evident at one age
level than another. Subtype A was not described in terms of the WISC
or the WRAT. ‘This group was the strongest of the three subtypes on
auditory-verbal and language-related tasks.

Subtype 8 exhibited poor auditory-verbal processing and psycho-
1inguistic skills. This group had the largest Verbal IOrPerformance IO
discrepancy. and it was in favor of the Performance IO. Type C
appeared to be a subgroup of Type A and was distinguished by an
unusually poor performance on fingertip number writing. Again. WISC
and WRAT performance was not reported. ‘This subtype did not emerge in
the 9-to-10-year-old sample. Some general observations of this entire
sample were reported. WISC Information. Arithmetic. and Digit Span
were below average. while Coding was within one standard deviation of

the mean. suggesting an AID pattern rather than an ACID pattern.

31

Picture Arrangement. Object Assembly. and Block Design were average to
above average. No comments were made about patterns of performance on
the subtests of the WRAT. They concluded that further research is
needed in the area of longitudinal tracking of children of the various
subtypes in order to determine if the data here actually do represent
developmental trends. This. again. is expected to make it possible to
evaluate specific approaches to academic remediation for learning
disabled in a more reliable manner.

Lyon. Stewart. and Freedman (1982) identified five subgroups of
learning disabled readers using multivariate analysis of variance and
discriminant analysis on 10 neuropsychological measures. ‘The subjects
in this study were between 6 years 5 months and 9 years 9 months of
age. The 75 learning disabled readers had a mean WISC-R Full Scale IO
score of 102.9. and the 42 normal readers had a mean Full Scale 10
score of 105.3. Of the 75 learning disabled readers. 11 could not be
located consistently in any subgroup. so their scores were excluded
from further analyses. The findings supported Fisk and Rourke's (1979)
finding that similar subgroups can be found over a range of developmen-

tal levels.

11W
W

A frequent finding in the WISC-R literature is that the
Information. Arithmetic. Digit Span. and Coding scores are often
relatively low (Ackerman. Dykman. & Peters. 1976: McManis. Figley.

Richert. & Fabre. 1978: Robeck. 1971: Rugel. 1974). Kaufman (1981)

32

.muLOun >ucu.e>_:av ouncuu
.¢-um.) :0 women

U

.noLOun e__ucouLon

.vOuLonoc «oz I .¢.ze

a

 

 

Table 2.1

 

 

Em>m Aucm.v
.<.z .<.z .¢.z .<.z .oo.~_voa.~c. -smso m~ .o.¢ .uconuum.wnusu
A_¢.o_voc.sm Ao_.a .m~.s_ .Nn.~..m~..m Am_.n_.~m.¢o. a cw“:

. . . . . . . . c... m .nwm.v
.mm 5 .mN _. .mo a .s. on Ao~ m Vow o. Au. ~_.~o ac. \uc.eoo¢ >o_x.~¢ .
ANo._..m~._~ .mo._..cm.am Aso.o_.m~.o~ on. uAN..o..m~.m.. _.eLo: n_-~ on .a.5 .oxossoz .e..oz
.c~.~ Vms.a. ._o.o .m~.o Aom.m .mo.c. ..m.h Vom.am ._-m_ oa<
.m~.~ .om... .em.m c~°.h Ame.» .sO... A»... cxm.om ~_-__ 00¢ a... .m~a_. .x.=o¢
Aam.o ._o.¢_ Ama.m v.8.a Ame.“ Vam.~. ...-oa .-.m .o_.om o.-a ou¢ -o.a .oN .o.5 . x...

s__>o “mum—v oxcsox
.¢.z .¢.z .m_.o ..~.n. c~.-oo Ama.o .Na.om .>5 mm. .a.¢ . «aunsacuoa
. . . . . . . . . . 5a. 0:...»
z z x z a a 5..-.» x z :_-m a. o 5 A» w oxL=ox
A~a._ Va~.~ Amc._ .m~.. om._ “a. As~.m voo.mm Am. Aa~m_.
Ammn v_mH~ Ros. voo.: “we. “on.“ A_n.s vo~.mm “NV c0n>-.c_m
.mo _ .na ~ uA_c._ .m..n Ac... .-.n s._-oo .~m.~ .n~.~m ... ._-m m. .a.s . oxcsoc
Ao.n_v m.o_ Am.o_. ~.m_ Am.n~. ¢.m~ ...m . n.oa .av
.a.m . ..._ xa.¢ . o.m Ao.n_c _.m. A~.m c 3..» an. .n~a_. mesa.
.~.n_. n.a. A_.e~. n.~n Am.~m. a... an.._v ~..m A~V . .e_o:oouox
Ac.o_. ~.~_ .a.o_. ..~. A..o~. _.a~ o~.-oo As.» V a..a ... ._-o_ o: .a.5 ..cc-» .oxL=O¢
sm>o_ A.am.. >oa.:»
.«.z .¢.z .¢.z mn_-m~ A.¢.z. u9:: -eeso mm .o.s . a.:guon~»
Aom.°_. n.5m as->x
Aa~.m . ".mm . . s
.¢.z .¢.= .¢.z o..-om .~¢.o . n.ma >5-a= o_-¢ on .o.5 .mhm_v o_¢ . o.¢
sass. Am~m_v
.¢.z .¢.z .¢.z o__-mu ..¢.z.uma.om -so_>m mm .o.5 concosu.m
.¢.z .¢.z .¢.z .n.-om Ao.~_. ~._o_ ~_-~ o~. .o.¢ .cthnwnwso¢nwuon
. . . . . . . . . . .msm.v ease» a
x z x z x z o~_-m~ a z o-m «o a 5 .go...o_a .ox.=o¢
Ae._~. ..w~ Aa.~mv a.as Am.~nv m.Om As.o_v m.am a5->= ._~a_.
Mm.m_w n._~ Ma.o~w o.o~ Mm.o~w m.nm M_.°_W «.5» a.> oc__.oxn_a .
e... ~.~_ _.m_ ~.~. s m.~_ .N. m__-m~ _.o «.mm >5-az ;_-m cm .o.5 .mcao» .oxLaox
Ae.°.. o.~m as->: ..~m_.
.o.c_. a.mm a u > >uouo.op
.¢.2 .¢.2 ..x.z m__-mc Am.» . m.mm >5-az ._-m as .o.5 . ox.=ox
mm m mm < mm x coco: o.mc mm o.mc M em< .m nuoqusm
p<¢n new. om.)

 

.cocv__;u vo_aou_v uc_cceo_ co moa>unan mc.umm_umo>c_ mo_v:un co >LQEE:m--...~ o_neh

33

noted that the factor he referred to as Fredom from Distractibility
makes up three-fourths of the ACID profile. He did not find. in his
factor analysis of the standardization data. that Information loaded
with the other tests in the factor. In Huelsman's 1970 review of the
WISC research on characteristic profiles. he found that 100% reported
low Arithmetic. 95% reported low Coding. 80% reported low Information.
and 60% reported low Digit Span. providing some support for this

factor.

W

Cullen. Boersma. and Chapman (1981) compared a group of learning
disabled children to a group of normal children to (a) ascertain the
utility of perceptual motor. verbal cognitive. and affective variables
for discriminating between the two groups of "average" ability
children and (b) to elaborate on the possible influence of affective
variables on school learning for the two groups. They found that
WISC-R Performance 105 were not significantly different for the two
groups. but Verbal 105 were. The differences were largely due to lower
learning disabled scores on the Information. Arithmetic. and Digit Span
subtests. Coding was also lower. but not significantly. Wide Range
Achievement Test (WRAT) scores were also lower on the three subtests.
as were Student's Perception of Ability subtest scores. The Beery
Test of Visual Motor Integration did not differ between the groups.
These authors concluded that remedial procedures should focus on

verbal-cognitive and affective dimensions of learning.

34

Vance. Gaynor. and Coleman (1976) analyzed the WISC-R subtest
scores of 58 learning disabled children. Their data indicated that low
subtest scores on Arithmetic. Coding. and Information were characteris-
tic of this group. Bernard (1978) in a profile analysis of 197 school-
verified learning disabled students also found the Arithmetic. Coding.
Information pattern to distinguish the learning disabled students from
the nonimpaired students.

Tabachnick (1979) suggested that low Coding may be significant but
did not find the other three tests to be consistently low. Fisk and
Rourke (1979) reported that an Arithmetic. Information. Digit Span
pattern may be more commonly observed than the ACID pattern.

Huelsman (1970) and Ackerman. Dykman. and Peters (1976) both
indicated that this profile is not representative of all children who
exhibit learning disabilities. but if one or more of these tests is
significantly depressed. the child is almost certain to have difficulty
in the classroom.

W
W

Koppitz (1971) studied a group of 177 learning disabled public
school children for 5 years. The children ranged in age from 6 to 12
years. and Full Scale 10 scores on the WISC ranged from below 79 to
above 120. Of this sample. 42 returned to regular classes during the
five years; 26 were referred for hospitalization: 38 either withdrew.
transferred. or moved away: and 71 remained in the learning disabled

program for the 5-year period. It appeared that a larger proportion of

35

the younger children (6-8) and the oldest children (ll-12) had

dull normal or borderline intelligence at the beginning of the study.
Koppitz concluded that the younger students often required long-term
special education. and a minimal Full Scale 10 score of 85 at the time
of diagnosis is needed if the child is to return to regular classes.
The initial mean IQ was 98 for those children who eventually returned
to regular classes. while mean IO was 87 for those who remained in the
L0 program. Reading achievement as measured by the Wide Range
Achievement Test was higher initially for the group who returned to
regular education as well. The group who returned to regular classes
spent an average of 3 years in special education.

It was found that the length of a pupil's stay in the learning
disabled program before returning to regular education was related to
his level of functioning but not to his age when entering the program.
Those children who returned after 1 to 2 years were significantly
brighter and had higher achievement on the WRAT Reading. Spelling. and
Arithmetic subtests.

When comparing those who were in special education for 1 to.3
years to those who were in for 4 to 5 years before returning to regular
education. Koppitz found stable IO scores for both groups. Academic
progress. however. showed an inverse relationship to the amount of time
spent in special education. The children who remained for shorter
times showed greater annual gains on the WRAT. The children who spent
1 year in the program gained 1.7 grade levels on the Reading subtest

and‘LO grade levels on the Spelling and Arithmetic subtests. The

36

group who spent 2 years in the program made almost the same progress
but over the 2-year period. The average yearly progress for the
children who remained in the learning disabled program for 1 year was
one grade level: for children who remained for 2 to 3 years. the gain
was three-fourths of a year per school year: and for those who remained
4 to 5 years. the gain was one-half to one-third of a grade level per
year. It was noted that the greatest average gain occurred in the
first year and decreased with each year spent in special education from
there on.

Full Scale. Verbal. and Performance IO scores for the 71 children
who remained in the learning disabled program throughout the study were
quite stable. The initial means were: Verbal 109-87. Performance 102-
89. and Full Scale IO-87: five years later they were 87. 87. and 81.
respectively. On the WRAT it appeared that the students fell behind
1 year over the 5-year period. The average gain for the group was 1.7
grade levels in Reading. 1.4 grade levels in Spelling. and 1.9 grade
levels in Arithmetic.

In terms of age groups. the children in the 8-to-lO-year-old
range made the greatest gains. When the data were reanalyzed for each
age group (6. 7. 8. 9.10. 11. and 12). it appeared that age was not
related to the gains made. As a result. the children were redivided
into groups on the basis of other characteristics.

Eighteen children were classified as dull and at the end of the

study were established as moderately retarded. ‘This group had an

37

average Full Scale IQ of 77 initially and 72 after 5 years. They
gained an average of 1.3 years on the WRAT subtests.

Nine children were described as dull children with good memory.
Their average Full Scale IO scores went from 81 to 80. and they showed
a 3.6-year gain on the WRAT Reading and 2.2- and 2.5-year gains on the
Spelling and Arithmetic. “This was attributed to their strength in
visual recognition and recall and lack of reasoning skills.

The third group was made up of 21 children having specific
learning disabilities. Koppitz described them as experiencing "serious
perceptual malfunction and severe memory deficits." Mean Full Scale IO
went from 91 to 90. and the group gained 1.0 grade levels in Reading
and Spelling and 1.8 in Arithmetic. This group was described as quite
heterogeneous in the types of difficulties exhibited.

The last group. emotional and behavior problems. was composed of
23 students. This group gained 2 points in mean 10 (94 to 96) and 2.4
grade levels in Reading. 2.1 in Spelling. and 2.2 in Arithmetic. The
range of 10 scores was broader for this group. and again the problems
observed were heterogeneous. When comparing changes in IO.scores for
individual students. Koppitz reported that those children who showed
the greatest gains were not fully cooperative at the time of the
initial evaluation. so the results were not accurate. ‘The children who
exhibited a decrease were thought to have reached a plateau in mental
development.

Koppitz concluded that children who can learn to read will learn

to read even without motivation. but arithmetic requires more

38

application and integration of skills. so children will not learn math
without outside help. A second conclusion was that once a child was
working up to ability. it was not possible to speed up the child's
natural rate of progress through special strategies. Finally. she
stated that repeating a grade is rarely helpful. as the learning
disabled child requires a slower pace of learning and greater
individual help than can be provided in a regular classroom.

Smith (1978) tested and retested a group of 161 learning disabled
children from Nashville over a‘7-month period. He reported that the
mean Performance IO was significantly higher than the Verbal 10 at both
testings. Verbal IO remained the same while Performance IO rose by 4
points. This finding was statistically significant but was attributed.
in part. to a practice effect. All of the Performance subtest scores
rose. while the mean Vocabulary score decreased.

Vance. Blixt. Ellis. and Debell (1981) retested a sample of 75
learning disabled and educable mentally retarded children. ranging in
age from 8 to 14 years. after 2 years. No significant differences were
found in any of the three IO scores.

Byrd. Buckhalt. and Byrd (1981) reviewed the WISC-R profiles of
385 children referred for academic difficulties and compared these on
the basis of age. .A decrease with age in the Acquired Knowledge factor
was found for both the low—IO and high-IO groups. which was more marked
for the low-IO group. ‘The subtests composing the Spatial Ability
factor (Picture Completion. Block Design. and Object Assembly) were

relatively high for the high-IO group and showed little change with

39

age. while the rankings were lower for the low-IO group and showed a
positive change with age. The older children did better on the speeded
tests than the younger. with the exception of Coding which became
lower. While the Coding decrease was inconsistent with the findings for
the other speeded tests. it was consistent with previous findings of
relative impairment in Coding performance by learning disabled children
(Kaufman. 1979). The correlation between the subtest ranking of the
youngest and oldest groups did not reach significance. but significant
agreement was found for the 9-to-12-year-old group and the 13-to-l6-
year-old group regardless of high IQ or low IO. These authors concluded
that profile analysis must take age into account as the patterns may
vary considerably within groups according to age.

Frauenheim and Heckerl (1984) completed a follow-up study on 11
males who had been identified as dyslexic in childhood. ‘The mean age
at diagnosis was 10 years 6 months and at follow up was 27 years. All
of the subjects received special reading help in addition to regular
education programming. The subjects were tested on the WISC at the
time of the initial evaluation and the WAIS—R at the time of the
present study. The 10 scores were remarkably stable given the time
interval and the different forms of the test. with the largest change
being a 2-point decrease on the Performance ID. The range of IO.scores

was also quite comparable.

40

Wad
WWW

Many limitations exist in the above literature. The method used
to determine expected grade level in the Rourke studies was based on
the chronological age of the subjects and did not take intellectual
ability into consideration. ‘The method was not reported in the studies
but was explained in correspondence with Dr. Rourke. ‘This method
assumes that all children at a particular age will be at the same
level. although this is contradictory to common knowledge. Also. none
of the studies listed in Table 2.1 indicated what method of determining
expected achievement was used. nor did they report on the reliability
of the ability-achievement discrepancy used.

Another problem is that the research does not quantify the
severity of the learning disability. making it difficult to compare
studies. The reader often must guess at the severity by the setting
from which the subjects are drawn. The most common are public schools.
private schools. or psychoeducational clinics.

The majority of the studies identifying subtypes of learning
disabled children used the WISC rather than the more recently published
WISC-R (Fisk & Rourke. 1979; Petrauskas & Rourke. 1979; Rourke. 1976;
Rourke & Telegedy. 1971; Rourke. Young. & Flewelling. 1971; Rourke.
Dietrich. & Young. 1973; Rourke & Finlayson. 1978; Rourke & Strang.
1978). As a result. the findings may not be generalizable to current
diagnostic practices.

Only one of the longitudinal studies used the WISC and the WRAT as

measures of the change over time of the subjects (Koppitz. 1971). The

41

group was fairly heterogeneous with a number of children who were later
determined to have disorders other than a specific learning disability.
This study also compared groups rather than individuals and is somewhat
dated. The other longitudinal studies described either employed a less
commonly used test (Fox & Routh. 1983) or followed up using interviews
rather than standardized testing (Werner & Smith. 1977; White. Alley.
Deshler. Schumaker. Warner. 8. Clark. 1982). As a result. the outcomes
were more descriptive of vocational. social. and emotional factors than
academic or intellectual factors.

The majority of the research compared groups of children. rather
than individuals. which may mask patterns more unique to individual
children. While this information is helpful in describing the popula-
tion. it cannot be used for the description of a specific child.

A fifth limitation is the emphasis on discovering the cause of the
disability. This information is useful in identifying high-risk
children. but little is known about the academic prognosis for
individual children who are diagnosed as specifically learning
disabled. Having identified these problems. the paper will now focus

on the methodology including how these problems were controlled or

reduced in this study.

CHAPTER III

METHOD

.Bntinnila

Study I was designed to determine if subgroups of learning
disabled children could be identified on the basis of the WISC-R and
the WRAT. One goal of this study was to investigate whether there are
differences between subgroups of learning disabled children relative to
their general cognitive ability and academic achievement at the time of
repeat evaluations. An effort was made to use methods for the identi-
fication of such differences which were similar to current diagnostic
practices in the schools.

Methodological departures from the previous research studies were
made. First. the subjects were identified on the basis of their
performance on both the WISC-R and the WRAT. rather than their
performance on one or the other. This approach is more consistent with
the methods commonly employed for determining eligibility for special
education services. Previous research has used one or the other test
for identifying subgroups in conjunction with other less commonly used
tests. such as the Halstead-Reitan Neuropsychological Test Battery. A
second difference is that the majority of the earlier studies in this
area selected subjects who were originally diagnosed on the basis of

their performance on a neuropsychological test battery. It is unlikely

#2

43

that these tests will become the basis for diagnosis in the schools. so
this study again attempted to use a process which is more applicable to
the manner in which school psychologists actually operate.

The first goal of Study II was to determine if any of the sub-
groups showed greater improvement or deterioration in achievement test
performance over time than others. This question was considered
important for long-term special education programming. This study
differs methodologically from previous studies in that each subject was
compared to himself in addition to comparing group means. ‘This change
allowed the analysis of patterns of change within groups as well as
across groups.

Previous studies have reported that students exhibiting lower
scores on the Arithmetic. Coding. Information. and Digit Span subtests
are likely to demonstrate difficulty in school. Rourke (1981) sug-
gested that this profile may distinguish learning disabled readers who
improve from those who continue to exhibit a severe reading problem.
This study attempted to answer Rourke's question with respect to per-
formance on the Reading subtest on the WRAT.

This study assessed changes in Verbal. Performance. and Full Scale
IO scores over time. It is presently unclear whether the performance
of individual learning disabled children on intelligence tests remains
stable or if this fluctuates over time. This question has a strong
bearing on the recertification process for special education services
and potential placement changes. Again. each child was compared to

himself for a measure of individual change.

44

Finally. this study addressed the question of how the severity of
the discrepancy relates to the type of educational placement. While it
previously seemed logical to place learning disabled children into
different types of educational settings according to ability-
achievement discrepancy. there is no empirical evidence that it is

actually the case.

 

WWW
EBAI
Reading and Reading and Reading =
Spelling > Spelling +15 Arithmetic =
.fllfigzfl Arithmetic +15 < Arithmetic Spelling

Verbal IO >
Performance IO A B 0
+9

 

Verbal IO Within
10 Points of D E F
Performance IO

 

Verbal IO +9
< Performance G H I
IO

 

 

 

 

Figure 3:L--Subgrouping of learning disabled children.

The following research hypotheses were tested in Study I:

l. Homogeneous subgroups of learning disabled children can be
identified utilizing a combination of the WISC-R VIO-PIO
discrepancy and WRAT Reading. Arithmetic. and Spelling
subtest patterns.

 

45

2. The subgroups will not differ significantly on the WISC-R
FSIO scores at the time of the initial evaluation.

The following research hypotheses were tested in Study II:

Hypothesis 1 deals with the first objective of Study II. namely.
to determine if differences exist among homogeneous groups of learning
disabled students on the amount of change observed over time on
standardized tests.

1. Significant differences will be found among the subgroups of
learning disabled children on the measured change in the
subtest scores of the WISC-R and the WRAT at the time of
re-evaluation.

Hypotheses 2 through 5 deal with the second objective of Study II.
namely. is it possible to predict the degree of change on the WRAT
subtests based on the initial pattern of WISC-R Verbal IO-Performance
IO discrepancy and the initial WRAT subtest pattern.

2. The children in cells A. B. and C will show greater improve-
ment on the WRAT Reading and Spelling subtests at the time of
re—evaluation than the children in cells 0 through I.

3. The children in cell 0 will show less improvement on the
Reading and Spelling subtests than those in A. B. and C but
greater improvement than the children in cells E through I.

4. The children in cell H will show greater improvement at the
time of re-evaluation on the WRAT Arithmetic subtest than
the children in the other eight groups.

5. The children in cell F will show less improvement on the
WRAT subtests than the children in the other cells.

Hypothesis 6 deals with the third objective. namely. to determine
if those children who exhibit the ACID profile differ in reading
achievement over time as compared to those who do not. regardless of

the subgroups.

46

6. Those children having Arithmetic. Coding. Information. and
Digit Span as their lowest WISC-R subtest scores will show
significantly less improvement on the WRAT Reading subtest at
the time of re-evaluation than those children not exhibiting
this pattern.

Hypotheses 7 through 9 deal with the fourth objective of the
study. namely. to determine the stability of the learning disabled
child's intelligence quotients.

7. Verbal intelligence scores will show a significant decrease
at the time of re-evaluation.

8. Performance IO will not show a significant change at the time
of re-evaluation.

9. Full Scale intelligence scores will show a significant
decrease at the time of re-evaluation.

Hypothesis 10 deals with the fifth objective of Study II. namely.
to determine if the severity of the achievement-ability discrepancy is
related to the type of special education classroom placement.

10. The severity of the discrepancy between ability and
achievement is not related in a consistent manner to the
type of educational placement.

1191101112115

5peg1£19_1§ann1ng_disab111ty. ‘The Michigan Rule defines specific
learning disability as follows:

Rule 13. (1) “Specific learning disability” means a disorder
in one or more of the basic psychological processes involved in
understanding or in using language. spoken or written. which may
manifest itself in an imperfect ability to listen. think. speak.
read. write. spell. or to do mathematical calculations. ‘The term
includes such conditions as perceptual handicaps. brain injury.
minimal brain dysfunction. dyslexia and developmental aphasia.
The term does not include children who have learning problems
which are primarily the result of visual. hearing. or motor
handicaps. of mental retardation. of emotional disturbance. or of
environmental. cultural. or economic disadvantage.

47

(2) The individualized educational planning committee may
determine that a child has a specific learning disability if the
child does not achieve commensurate with his or her age and
ability levels in one or more of the areas listed in this subrule.
when provided with learning experiences appropriate for the
child's age and ability levels. and if the multidisciplinary
evaluation team finds that a child has a severe discrepancy
between achievement and intellectual ability in one or more of the
following areas:

(a) Oral expression.

(b) Listening comprehension.

(c) Written expression.

(d) Basic reading skill.

(e) Reading comprehension.

(f) Mathematics calculation.

(9) Mathematics reasoning.

(3) The individualized educational planning committee may not
identify a child as having a specific learning disability if the
severe discrepancy between ability and achievement is primarily
the result of any of the following:

(a) A visual. hearing or motor handicap.

(b) Mental retardation.

(c) Emotional disturbance.

(d) Environmental. cultural. or economic disadvantage.

Want. The expected achievement will be stated in

the form of a z score rather than a grade equivalency score in order
to provide comparability to the measured achievement. The expected
achievement will be computed using the formula

(xi - i-)
Zx - o
X

where x1 is the WISC-R Full Scale IO score.
Want. The measured achievement will be stated as
a‘z score. 'The score will be computed for each WRAT subtest using the

formula

2 _(Yi--)

C
y V

where Y. is the WRAT subtest score.

48

.519n111san1_d159nepangy. The discrepancy or difference score will

be computed using the following formula:
Iz-ZI
X Y

/ (l-rxx) + (l-ryy)

 

 

Since the reliabilities of the WISC-R and the WRAT subtests fall
between .90 and .95. a minimum difference of .66 standard deviations
between the two scores is necessary to be certain that a significant

discrepancy exists (Reynolds. 1981).

ms:

The learning disabled population included all children referred to
the Holton. Holland. Lowell. and Ottawa Area Intermediate School
Districts between 1974 and 1981 who were subsequently diagnosed as
having a specific learning disability as described by Michigan Public
Act #51. Rule 13. This diagnosis was based on the findings of the
multidisciplinary evaluation team of the above school districts with
the psychological testing having been administered by an approved
school psychologist. One hundred ninety-six children who met the
following subject criteria were identified out of the above population:

1. Ages at initial evaluation ranging from 7 years 6 months to 13
years 6 months.

2. Initial diagnosis using the WISC-R and WRAT.

3. Re-evaluation on the WISC-R and WRAT 2 to 4 years later.

4. Full Scale IO scores falling between 70 and 115. as children
who fall outside of the "normal" range may differ markedly for reasons

other than a specific learning disability. and this investigation is

11""!

49

meant to apply to average children rather than those who are being
considered for special education placement for either giftedness or
retardation.

5. English as a primary language.

6. No primary diagnosis of emotional impairment and/or the
subject had not been referred to a psychiatrist for evaluation.

7. Regular school attendance since the beginning of
kindergarten.

8. No physical deficits in the areas of vision or hearing.

The subjects were divided into three groups on the basis of their
performance on the initial administration of the WISC-R as follows:

1. Verbal IO score and Performance 10 score within 10 points of
each other.‘

2. Verbal 10 score at least 10 points higher than the Performance
10 score.

3. Performance IO score at least 10 points higher than the Verbal
IO score.

These differences were selected in order to be able to make
comparisons to the subgroups identified in earlier studies by Rourke
and others even though differences of at least 10 points are not
considered diagnostically significant.

Each of these groups was further divided into three subgroups
based on their performance on the initial administration of the WRAT

as follows:

50

1. Reading. Spelling. and Arithmetic standard scores at least
1 standard deviation below the expected grade placement.

2. Reading and Spelling standard scores at least 12 points below
Arithmetic.

3. Arithmetic standard score below Reading and Spelling by at
least 12 points.

Again. the differences were selected in order to make comparisons
to the subgroups identified in earlier studies.

The subjects included 161 males and 35 females. ‘The higher pro-
portion of males (82%) is consistent with the research on sex-related
differences in diagnosis of learning disability. which indicates that
in almost any sample of learning disabled children. 80 percent are

likely to be male (Farnham-Diggory. 1978).

Instnuments

Two standardized assessment instruments were used to evaluate the

change in the students in the study.

lbLWechsJeLIntelligencLScaJe
mummified

The Wechsler Intelligence Scale for Children-Revised (WISC-R) was
administered to all children in the study. The WISC-R was designed to
evaluate general intellectual ability as well as specific strengths and
weaknesses in verbal and visuoperceptual performance in children
through the ages of 6 years 0 months to 16 years 11 months.

The standardization of the WISC-R was based on a national sample

stratified on the variables of age. sex. race. geographic region. and

51

occupation of the head of the household. The total sample comprised
2.200 cases (Wechsler. 1974. p. 17). The average spl it-half
reliability coefficient is 1; = .94 for Verbal IO. .1: = .90 for
Performance IO. and .E = .96 for Full Scale IO (corrected by the
Spearman-Brown formula and computed using Fisher's z transformation).
The average corrected stability coefficient of the Verbal IO score is
.1212 = .92 (computed using Fisher's z transformation). The test and
retest were given 3 to 5 weeks apart. Reliability of the information
subtest ranged from I: = .80 at age 7.5 to .90 at age 15. Coding ranged
from .63 at 7.5 years to .80 at 15.5 years. with an average reliability
score of .72. Arithmetic's lowest reliability score falls at age 8.5.
with an 1: = .69. and the highest reliability of 1: = .81 is obtained at
ages 11.5 and 13.5. On Digit Span. the reliabilities ranged from J: =
.71 at age 10.5 to .84 at age 7.5. Reliability coefficients were not
available for all ages on the Coding and Digit Span subtests. The
reliabilities reported above were clearly acceptable for the purposes
of the research.

A group of 303 children from six age groups in the WISC-R
standardization sample were retested after a 1-month interval to assess
the stability of the WISC-R. The Full Scale 10 score changed about
7 IO points. the Verbal IO 4 IO points. and the Performance IO about 10
10 points. All changes were increases. These results suggest a
definite practice effect. especially on the Performance Scale.
Stability coefficients were .95 for F510. .93 for VIO. and .90 for PIO
(Wechsler. 1974).

 

 

52

One hundred eighteen children in four age groups participated in
the study of the validity of the WISC-R. Each child was administered
the Stanford-Binet Intelligence Scale Form L-M (SBIS. L-M) within 9J5
months of the WISC-R. The average correlation between the $818. L-M
and the Verbal IO was .71. Performance 10 was .60. and the Full Scale

ID was .73.

MW

The second standardization research instrument selected for the
study was the Wide Range Achievement Test (WRAT) (Jastak & Jastak.
1976). This test was designed to assess reading. spelling. and arith-
metic performance.

There are two levels of this test: Level 1 for students younger
than 12. and Level 2 for those 12 and over. Each level was standard-
ized on at least 150 males and 150 females at each of 19 age levels.
resulting in a population of 5.868 persons at Level 1 and 5.933 at
Level 2. Norms were not stratified on the basis of any specific cri-
terion.

Split-half reliability coefficients are reported for each subtest
by grade level. All exceed .90 (Jastak 8 Jastak. 1978). Validity data
are not reported; however. content validity is assumed (Salvia &
Ysseldyke. 1978). A revision of this test was published in 1978.
Breen and Prasse (1982) compared the 1976 and 1978 forms and reported
that they were equivalent if standard scores were used for the

comparison.

53

Emcedunes

The directors of special education in Kent and Ottawa Counties
were contacted by telephone to request permission to review their files
of those students who had been or were presently enrolled in a special
education program. Four of the six directors contacted agreed to
participate in the study. and they selected files which met the
researcher's requirements. from which the data were collected. Data
were gathered on sex. birth date. dates of evaluation. WISC-R scores.
WRAT scores. and type of special education placement. The districts
represented rural and small urban communities in Western Michigan. Two
(one urban and one suburban) school districts refused to participate.
As a result. no children from a major urban community were included.
The participating communities were composed primarily of middle- and
lower-income Caucasian families. The largest minority group was com-

posed of Mexican-Americans.

WW
Study 1. Hypothesis 1 was tested by determining the number of

subjects having WISC-R and WRAT test patterns which meet the require-
ments of the cells. Hypothesis 2 was tested using a one-way ANOVA to
compare the cell FSIO means.

Study II. Hypotheses l. 2. and 3 were not testable due to insuffi-
cient cell sizes of cells A. B. C. D. and G. Hypotheses 4 and 5 were
tested using separate analyses of covariance using estimated true
scores for each WRAT subtest to demonstrate the differential change

among cells E. F. H. and I.

54

Hypothesis 6 was tested using a one-way ANOVA of an index of
response to compare the change in WRAT Reading scores of the two
groups. In addition. a distribution of the various combinations of the
above tests found as the lowest subtests was tabulated.

Hypotheses 7. 8. and 9 were tested using correlated t-tests. The
educational significance was evaluated according to the number of
subjects who change categories. as outlined in the WISC-R manual.

Hypothesis 10 was tested by dividing the subjects into groups
based on their initial placement into self-contained. resource room. or
teacher consultant programs. ‘The discrepancies for each subject were
computed by transforming the WRAT subtest scores and WISC-R Full Scale
10 scores into 2 scores for comparability. ‘The difference between each
WRAT subtest standard score and the Full Scale IO score was computed
using the comparison between 1 scores described earlier. Three one-way
ANOVAs were computed to test for significant differences among the

three groups'cHscrepancies between F510 and each WRAT subtest.

W

The study assumes that only one population of learning disabled
children exists. namely. those children who have been identified and
placed. It is possible. however. that these children may represent a
small portion of a much larger population of learning disabled
children. For example. although the referral rate of males for
specialized services is much higher. Lambert and Sandoval (1980) found

an equivalent rate of incidence for underachievement of males and

55

females in an unreferred sample. If it is true that ability-
achievement discrepancies are evenly distributed across males and
females. then a large number of learning disabled females are not being
identified.

The subjects consisted of a sample of special education students
in small urban and rural school districts who were identified by school
psychologists as learning disabled and placed in special education by
Educational Planning and Placement Committees. The process involved in
identifying these children was not examined in this study. An assump-
tion was made. therefore. that all subjects were appropriately classi-
fied. By this it is meant that it is expected that the psychologists
diagnosed the children according to federal guidelines. Public Law 94-
142. and Michigan PJk 451. and that the Educational Planning and
Placement Committees similarly followed the law in placing the children
in their programs. Ysseldyke. Algozzine. Shinn. and McGue (1979)
compared a sample of school-identified learning disabled children to a
sample of nonlabeled children with similar levels of underachievement.
These researchers found that an average of 96% of the scores were
within a common range. In comparing the characteristics of these
children with the federal definition. they reported that as many as 40%
may be misclassified. Shepard. Smith. and Vojir (1983) also reported a
high rate of misclassification.

Another limitation of the study is based on the use of difference
scores. Two types of difference scores are used in this study. gain

scores and discrepancy scores. A problem related to the use of gain

I. II I}
1.. ‘
Ill! III
I‘l-
I!
II
[I
III-
II.
I
'.

 

56

scores and to the use of a nonrandom sample is that of regression to
the mean or statistical regression. The difficulty in evaluating
change in groups who are selected on the basis of poor or extreme test
performance is that it is statistically predictable that. on a retest.
the average score will have a strong probability of being higher than
the average score on the first test. This result is not related to any
genuine effect of any treatment but rather is the result of the
imperfect correlation between the first and second testings. 'The less
perfect the correlation between the two tests. the greater the
regression to the mean will be. This does not mean that an individual
will necessarily change toward the mean but. simply. that our best
guess is that the direction of any change will be toward the mean. In
other words. the farther a score is from the population mean. whether
it is extremely high or low. the greater the likelihood that the retest
score will be closer to the mean than the first score. Measurement
theory assumes that a more deviant score contains a greater proportion
of measurement error than a less deviant score. Since error is assumed
to be random. it is likely that the measurement error will be less at
the time of the retest. A common flaw in research studies which use
extreme groups and gain scores is interpretation of the data without
acknowledging the issue or effects of regression to the mean (Campbell
& Stanley. 1963). As mentioned earlier. the present study includes the
use of gain scores on the WISC-R and the WRAT. 'The results of these
analyses must be interpreted quite cautiously. given the problem of

regression to the mean.

S7

Gain scores are considered the least reliable of difference
scores. Porter (1973) suggested that a cautious interpretation of the
results can supply useful evaluation information. When the correlation
between the two measures is reasonably high. it tends to reduce the
reliability of the gain score. However. if the correlation between the
two measures is low. the question of validity. or whether the two tests
are measuring the same thing. is raised. Mehrens and Lehman (1975)
reported that they may be reliable enough for making decisions about
groups.

Discrepancy scores are also less reliable as they are affected by
the reliabilities and the intercorrelation of both tests used to com-
pute the discrepancy score. As a result. discrepancy scores must also
be interpreted quite cautiously. Discrepancy scores were used in this
study as the federal and state definitions of learning disability
include the criterion of a "severe discrepancy between achievement and
intellectual ability." As Shepard. Smith and Vojir (1983) pointed out.
the definition of learning disability is vague and provides equivocal
guidelines for educational diagnosis.

Since the sample was not randomly selected. generalization of the
results is limited. Beyond meeting the dual criteria of age (7 to 13
years) and identification as learning disabled. subjects were included
in the study only if they had two complete WISC-R and WRAT protocols
2 to 4 years apart and if the director of special education in that

district agreed to participate in the study.

58

Subgroups in the present study were established using inferential
classification. As a result. the findings pertain only to other
research studies using a similar approach and not to those studies
which have established subgroups on the basis of multivariate
classification procedures. such as Orfactor analysis and hierarchical
cluster analysis.

The WISC-R and WRAT were not administered during the study.
Rather. the data were taken from case records in retrospective review.
Therefore. inaccuracies of test information due to errors in adminis-

tration and scoring could not be controlled.

CHAPTER IV

RESULTS

The primary focus of this research was to determine the potential
for using commonly used psychological tests for establishing
homogeneous subgroups of learning disabled children and predicting
their performance on'academic achievement tests on the basis of group
membership.

In the first part of this chapter. the findings which relate to
establishing subgroup membership are examined. Of major importance to
the questions addressed is the frequency and ease with which learning
disabled students can be described in terms of subgroups membership.
While it may be of theoretical benefit to look at homogeneous sub-
groups. the practical significance is limited if identification is not
easily achieved using standard measurement procedures. Also. if the
proportion of learning disabled students who fit the subtypes is low.
the practical value is again reduced.

Following the analysis of particular WISC-R and WRAT combinations
for establishing the presence of homogeneous subgroups. a more specific
exploration of the data occurs. 'This is to determine whether subgroup
Inembership provides any basis for meaningful predictions of academic

achievement as measured by the WRAT on later evaluations.

S9

60

In the next section. the frequency of the ACID pattern is
described and the utility of using this pattern for prediction of
reading performance is examined. ‘The frequencies of related patterns
are also reported and compared to previous research.

The fourth section addresses the questions related to stability of
WISC-R IO scores in learning disabled children. Comparisons to other
studies of stability of IO in special education and regular education
students are made. ‘The educational significance is evaluated in terms
of the number of children who change intellectual classification
categories as outlined in the WISC-R manual.

The last section of this chapter compares the three types of
special education placement in terms of the severity of the ability-
achievement discrepancy. ‘The discrepancies for the WISC-R and each
WRAT subtest are evaluated to determine if severity of discrepancy is

related to type of special education placement.

MUD:

WW
Wm:

Hypoth§51§_l. Homogeneous subgroups of learning disabled children

can be identified utilizing a combination of the WISC-R VIOrPIO

discrepancy and WRAT Reading. Arithmetic. and Spelling subtest

patterns.

As discussed in a previous chapter. researchers have established
subgroups on the basis of WISC Verbal IOrPerformance IO discrepancies
or WRAT subtest patterns. Group membership in this study was based on

a combination of Verbal IOrPerformance IO discrepancies and WRAT

subtest patterns. 0f the 196 subjects. 26 had VIO > PIO +9. 87 had a

61

difference of less than 10 points. and 83 had VIO < PIO +9. When
subgroup membership required one standard deviation between the standard
scores on the WRAT subtests in addition to the VIOrPIO discrepancy.

only 32 (16%) of the 196 subjects fit into a specific subgroup. When
the requirement was lowered from a difference of 15 standard score
points to 12 standard score points. 33 more subjects were placed in the
subgroups. Table 4.1 shows the frequency of subgroup membership with

the 15- and 12-point differences.

Table 4.l.—-Subgroup membership with 15- and 12-point standard score
differences.

 

Reading & Spelling Reading & Spelling Reading =

 

WISC-R ‘ > Arithmetic + 15 (12) Spelling =
+ 15 (12)a < Arithmetic Arithmetic
VIO > (P+9) A B C
0 (l) l (2) l (1)
No significant 0 E F
difference 1 (4) S (12) 7 (14)
V10 < (P+9) G H I
2 (2) 6 (11) 9 (l8)

 

a( ) refers to total N identified using the lZ-point
difference.

0f significance is the finding that using the stricter
classification procedure. only 16% of the learning disabled population

could be placed in a subgroup and that even when the requirements were

62

reduced. only 33% fit into the categories. This finding suggests that
(1) it may not be possible to identify homogeneous subgroups which will
include the majority of learning disabled students. (2) subgroups may
only be established using a multivariate approach rather than commonly
used test scores. or (3) the approach used in this study and others
(Rourke 8 Finlayson. 1978; Rourke 8 Strang. 1978; Petrauskas 8 Rourke.
1979; Fisk 8 Rourke. 1979; Lyon. Stewart. 8 Freedman. 1982) may define
subgroup membership too narrowly to include most students.

The finding that it was not possible to fill the categories where
lower arithmetic performance and/or lower visual perceptual problem
solving skills were criteria implies that children with problems in the
above areas are less likely to be identified as learning disabled.
Although the federal and state definitions include a broad range of
skills as possible disability areas. the multidisciplinary teanfls focus
may have remained on verbal skills and reading performance as the most
commonly identified problem area. In fact. only 26 of the 196 subjects
met the requirement of VIO > PIO +9. Due to low numbers of subjects in
cells A. B. C. D. and G. these groups were not included in later data
analyses.

Hypothesis 2. The subgroups will not differ significantly on the
WISC-R FSIO scores at the time of the initial evaluation.

A one-way ANOVA was performed on the mean Full Scale 10 scores of
cells E. F. H. and I to determine if these groups were comparable at
'the time of the initial evaluation. Hypothesis 2 was not supported
(E = 3.462. _d_f, = 3. p < .023). indicating an initial difference among

the cells. As can be seen in Table 4.2. cells E (V = P. R85 < A) and

63

H (V < P. R85 < A) were noticeably lower than cells F (V = P. R = S =
A) and I (V < P. R = S = A). The first two cells have in common rela-
tively weaker verbal skills on the WRAT. while the latter two cells
have relatively equal academic skills. although all achievement scores
are depressed. This finding may be related to the fact that all of the
instructions include a large verbal component. ‘The results should be
interpreted cautiously. however. as the cells' Ns are very small. The
initial difference is also likely to have a bearing on the interpreta-

tion of the predictive hypotheses.

Table 4.2-~78 cell FSIO scores and standard deviations by WRAT

 

 

 

grouping.
RS < A3 R = s = A
V = P E 89.58 ( 5.70) F 96.00 (8.30)
V < P H 90.45 (10.71) I 97.89 (7.81)
aR = Reading
S = Spelling
A = Arithmetic
mum

Hypothasls_l. Significant differences will be found among the
subgroups of learning disabled children on the measured change in
the subtest scores of the WISC-R and the WRAT at the time of re-
evaluation.

.flxngtbesis 2. The children in cells A. B. and C will show greater
improvement on the WRAT Reading and Spelling subtests at the time
of re-evaluation than the children in cells 0 through I.

64

.Hypgtnesis,3. The children in cell D will show less improvement
on the Reading and Spelling subtests than those in A. B. and C but
greater improvement than the children in cells E through I.
Hypotheses 1 through 3 were not tested due to the failure to fill
cells A. B. C. D. and G and the small sizes of the other four cells.
hypothesis 4,. The children in cell H will show greater
improvement at the time of re-evaluation on the WRAT Arithmetic
subtest than the children in the other eight groups.
Hypothesis 4 was not supported and. in fact. cell H (VIO.< PIO.
R 8 S < A) was the lowest of all four cells on the Arithmetic retest.
The analysis of covariance produced a significant interaction between
the WRAT profile and the Verbal IOrPerformance IO discrepancy (E =
4.302. d1 = 1.50. .p = .043) for the WRAT Arithmetic subtest. (See
Table 4.3.) The interaction was graphed using the adjusted cell means

of the WRAT Arithmetic subtest from the second evaluation (Figure 4.1).

 

 

35 ..
u
:3
Q)
E
.c
f: 80 ‘r
L
<
'—
5
3
75 1
RS < A R=S=A
0' V a P
‘0 V < P

Figure 4.1.--Interaction between WRAT profile and
VIOrPIO discrepancy.

65

Table 4.3.--Cell X WRAT Arithmetic analysis of covariance using
estimated true scores.

 

 

Source of Sum of g:_ Mean Square E_ Sig. of_§
Variation Squares
Within cells 2681.07 50 53.62
Regression 1935.94 1 1935.94 36.10 0
Constant 127.35 1 127.35 2.38 .130
RSA 565.88 1 565.88 10.55 .002
IO 1.44 l 1.44 .03 .871
RSA X IO 230.72 1 230.72 4.30 .043

 

The pattern of the interaction (RS < A) < (R = S = A) is consis-
tent with the pattern of FSIO scores at the time of the initial evalua-
tion. In looking at the means for each cell from both evaluations.
cell H had a decrease of‘FL36 standard score points. which was the
largest change of all four groups. (See Table 4.4.) The interactions
and the decrease in Arithmetic raise the question of the role of verbal
skills in Arithmetic. as it would be expected that cell H should be
weakest in verbal skills. Although the WRAT Arithmetic subtest is a
purely computational test. teaching methods are typically verbal.

Hypothesjs_5. The children in cell F will show less improvement
on the WRAT subtests than the children in the other cells.

Hypothesis 5 was not supported. The analysis of covariance uSing
estimated true scores for the Reading subtest produced an.E = 1.20.
_d_f = 1.50. p = .29. and the analysis for the Spelling subtest resulted in
E = .14. 511 = 1.50. p = .706. The analysis for the Arithmetic subtest

was reported earlier in Table 4.3. In comparing the cell means. cell F

66

 

 

 

mm a z

 

 

 

 

 

 

 

Ace.m v em.m~ Amm.ev es.MN Amx.mv .m... Aom.mv 5N.mm Amw.av mm.m~ Ase.ev so.m~ _..op
Ae~.m V .e._m 105.55 eo.s~ As~.mv ma... Asm.mv mm.wu Asm.xv -.ma Ams.xv mm.m~ w__uuu
AN_.w v ~m.e~ Amw.hv em... Amw.av N~.ma Ame.hv m_.wm A~8.ev am... 2.9.85 _N.sx P__u.u
xo~.__v km... xm~.Nv mo.- 2.4.mv mm.w~ A_m.mv se.ma ANs.mv Em.ma Aam.mv mo... N_.wun
Amm.~ V ~_.mm xw_.mc Ns.NN Ao_.ev co.- 5.8.85 mm.om 2.8.mv N_._~ Aae.ev mm.s~ w__wou
mm .x. mm w W. .m m M mm. m. H... M
< m m < m m
N #5:: — .523

 

.h<¢3 ecu mo mco_umcum_:_Eom ozu ecu co mco_um_>ou vcmccmum ucm memos __ou-u.:.: o_nmh

67

(V = P. R = S = A) exhibited the largest improvement on WRAT Reading.
second largest improvement on the Spelling subtest. and the smallest
change of any cell on the Arithmetic subtest. On the Arithmetic sub-
test. cell I (V < P. R = S = A) gained and cells E (V = P. R 8 S < A)
and H (V < P. R = S = A) lost: so here. too. cell F is actually in a
stronger position in relation to the other cells. This finding may be
explained by cell F's initially higher FSIO score. Cell F had the
second highest mean FSIO. suggesting that these subjects may have been
in a better position initially for acquisition of new learning than the
subjects in cells E and H. In fact. cells E and H showed the greatest
decreases on both Reading and Arithmetic. which is consistent with the
FSIO patterns.

Cell E showed the lowest correlation between the first and second
administrations of the WRAT. while the other three groups were more
similar to each other. To explain this. three additional two-way
ANOVAs were computed to determine if age or test-retest interval might
be affecting these correlations. None of these analyses was signifi-
cant statistically (Age 1: E = .50. _d_f = 3.51. .p = .683; Age 2: E =
.87. 511 = 3.51. n = .461; Test-Retest Interval: E = 1.55. _d_£ = 3.51.
p = .214). nor is any pattern evident in looking at the actual numbers
to explain this finding. The lower reliability may be related to the
particular combination of strengths and weaknesses these subjects
exhibit. Lack of homogeneity of the sample may be a problem. However.

no clear explanation is available at this point. (See Table 4.5.)

68

Table 4.5.--Correlation between WRAT l and WRAT 2.

 

 

 

Reading Spelling Arithmetic
Cell E .500 .045 .333
Cell F .604 .675 .696
Cell H .711 .603 .634
Cell I .652 .551 .742
W
WEEDS

.Hxnnlhfiiis.§. Those children having Arithmetic. Coding.

Information. and Digit Span as their lowest WISC-R subtest scores

will show significantly less improvement on the WRAT'Reading

subtest at the time of re-evaluation than those children not

exhibiting this pattern.

The aim related to the ACID test was to determine if the
presence or absence of this pattern had predictive value regarding
performance in Reading. It was hypothesized that children who exhibit
this pattern are less likely to improve on the WRAT Reading subtest at
the time of re-evaluation. Only 11 (6%) of the 196 subjects exhibited
a pattern of lower scores on the Arithmetic. Coding. Information. and
Digit Span subtests at the initial evaluation. A one-way ANOVA using
index of response was calculated to test this hypothesis. The hypothe-
sis was not supported (E = .913. d1 = 1.194. p. = .341). The result is
not surprising given the large difference in the number of subjects in

the two groups. Even if the results had suggested a difference between

these two groups. the findings would require a cautious interpretation

69

as a group of 11 subjects is much more affected by one extreme score
than a group of 185 subjects.

A frequency count of the related patterns was completed. as much
controversy exists regarding both the significance of this pattern and
whether all four tests are needed. Table 4.6 provides a summary of

these data.

Table 4.6.--Frequency of ACID pattern and related patterns.

 

ACID ACI AID ACD CID AC AI AD CI CD ID

 

No. of subjects 11 6 l4 l7 7 4 6 6 6 7 12

% of total group 6% 3% 7% 9% 4% 2% 3% 3% 3% 4% 6%

 

None of these patterns accounts for more than 9% of this popula-
tion. and together they only describe 50% of the subjects. Kaufmanhs
(1979) Freedom From Disability Factor (Arithmetic. Coding. and Digit
Span) was the most frequent pattern found in this group. Digit Span
was the most frequently low subtest. followed by Information. Coding.

and Arithmetic. in that order.

AW
W

.Hxnnihfifils.l. Verbal intelligence scores will show a significant
decrease at the time of re—evaluation.

This hypothesis was tested using a one-way correlated t-test which

was not significant in a comparison of the total group across the two

7O

evaluations. Correlated t—tests were also calculated for the three
groups based on VIO-PIO discrepancy. The Verbal > Performance group
and Verbal = Performance group showed significant decreases in VIO.(1 =
3.13. m: = 25. ,p < .005). while the Verbal < Performance group showed a
significant gain (:1; = -3.113. _d_f_ = 82. p < .05). Verbal IO decreased
4.38 points in the Verbal > Performance group and 2.034 points in the
Verbal = Performance group. Verbal IO rose 2.21 points for the Verbal
< Performance group. Hypothesis 7 was not accepted for the total group
but was not rejected for the Verbal > Performance group and the Verbal =
Performance group. ‘The average standard error of measurement (sem) for
the Verbal IO score is 3.60. The changes reported here for the Verbal
= Performance group and the Verbal < Performance group fall within this
range of error. However. the Verbal > Performance group shows a
greater change than would be expected on this basis. ‘The means and
standard deviations for each group are reported in Table 4.7.

flxnnih§§1§.8. Performance IO will not show a significant change
at the time of re-evaluation.

Hypothesis 8 was supported for the total group. the Verbal =
Performance group. and the Verbal < Performance group. In contrast to
expectations. the Verbal > Performance group had a significant gain in
Performance IO (.1: = -3.058. 311 = 25. p, < .01). The Verbal >
Performance group's Performance IO rose by 4.19 points. The
Performance 10 has an average Sem of 4.66. Thus. though the change
reported here is statistically significant. it is not educationally
significant as it falls within the confidence interval established with

the Sem' The change may be due to the effect of regression to the mean

71

 

 

 

 

 

 

 

 

 

 

 

 

 

A~:.o_vmm.wm Am:.~_vmm.mm A_m.o_vqm.mw Am~.m vmm.mw Amm.__vn_.mm A~:.o_vmm.mm .mHOP
A...m Vo~.om A.~.__vsm.mm A__.o_vo_..w Ame.m v.8.mm Ams.o_vm_.oo_ Ame.m VNm._m mm H n
Asm.o_vme.ew A~e.__vsk.om Ame.o_vma.mm Ass.w Vma.fim Am_.m vam.mw Ame.m vam.~m NW N n
Amm.o_v~:.hm As_.o_vwm.mm Am_.~_v_m.om Amm.o_v~m.fim A~o.m V~m._m Amm.__v_m.mm mm H n
.mm .m .mm .m .mm .m .mm .m .mm .m .mw .m
o_ma a_a u_> c_ma a_a a_>
N ¢-um_z _ m-um_z
.mnum_3 0;» mo mco_umcum_c_Ecm ozu ecu co mco_um_>oc vcmvcmum new mcmoe ocoum d.--.m.: o_nmh

72

or to a practice effect. though a practice effect is unlikely given the
2-4 year interval between evaluations.

.Hypgtnesis 9. Full Scale intelligence scores will show a
significant decrease at the time of re—evaluation.

No significant differences were found for any of the groups.
Table 4.8 lists the t-test results. as well as the mean difference
scores by group. As can be seen from the table. although several of
the t—tests show significant statistical differences. the mean
differences are actually quite small and would not be considered

educationally significant.

Table 4.8.--Correlated t-tests and mean differences of IO scores.

 

 

 

Verbal 10 Performance IO Full Scale IO
Total __ t_= 1.038 t_= -1.258 t_= - .221
V > P +9 __ t_= 3.l3** t_= -3.058** t_= 0.00
E = 26 XD 4.38 -4019 000
v = P _ E = 20575* i = - 0889 _t_ = 10134
N_= 87 XD 2.034 - .885 .770
V < P +9 __ t_= -3.113** t_= .220 t.= -l.594
N_= 83 X0 -2.21 .205 -l.036
in < .01.
*82 < .005.

In an effort to assess the effects of statistical regression on

the findings for Hypotheses 7. 8. and 9. the subjects were regrouped on

73

the basis of the Verbal—Performance IO discrepancy at the time of the
second evaluation and the correlated t-tests for the new subject
groupings. If regression to the mean accounts for the observed changes
in IQ scores. the changes should be in the reverse direction of the
changes reported for the initial groupings. The results are reported

in Table 4.9.

Table 4.9.--Corre1ated I—tests and mean differences of 10 scores
based on the regrouping of the subjects.

 

 

 

Verbal 10 Performance IO Full Scale IO
> P __ t_= .474 t_= -1.750* t_= - .337
.N = 16 XD 1.813 -3.750 -1.813
V=P t_=-.243 _t_=-l.26 t_=-.4l8
N_= 88 Yb - .490 -1.034 - .280
V < P _ t = -2.239** t = 4.076**** t = 2.430***

N_= 92 XD -1.500 3.587 1.489

*3 < 05

**E_< .025.
***B < .0].

%***
E.< .001.

The direction of the change for the Verbal > Performance
regrouping was in the same direction for all three 10 scores as the
change observed for the original grouping. The Verbal = Performance

regrouping changed in the opposite direction on the Verbal IO and Full

 

74

Scale 10 scores as compared to the original grouping and in the same
direction on the Performance IO score. The Verbal < Performance
regrouping changed in the opposite direction on the Full Scale IO score
but not the Verbal or Performance IO scores. These findings suggest
that regression toward the mean is a possible explanation for the
differences noted for the Verbal = Performance group on the Verbal and
Full Scale IO scores as well as the Verbal < Performance group on the
Full Scale IO score. While the findings do not support the expectancy
of regression toward the mean for all of the groupings. the findings
also do not clarify any other basis for the observed change.

To investigate more fully the possible impact of IO score changes.
a further analysis was carried out. Each intellectual classification
category (Borderline. Dull Normal. Mildly Impaired. Bright) on the
WISC-R has a 10-point range. other than the Normal category which has a
20-point range. Seventy-two subjects changed categories. No subject
changed by more than one category.

Only three of the subjects showed a change in IQ score of 10
points or more. suggesting that the majority of those who showed a
category change were near the lower or upper limit of the first
category they were placed in. Table 4.10 shows the changes in more
detail. The mean interval between evaluations was 36 months. with a
mean age of 10 years at the first evaluation and 13 years at the
second. These results suggest that. in general. all three IO scores

are stable for the learning disabled population.

7s

 

 

 

N 4. ON m m m. N. .mDOh
o N m o _ N m a v >
N m m N N m m mn>
o m N _ o N N 8A>
uoc_mae_ >_n__z oc__concom .mELOZ __:a .mEcoz u;m_cm .mELoz .mscoz __:o .I
On Ou Ou Cu 0u cu ou z
oc__couc0m .mELoz __:o .mELoz u:m_cm .mELoz .mELOZ __:o oc__conc0m

 

.maaoLm >ucmamLUm_c d_¢nd_> >3 :o_um:_m>oioc um >c0m0umo d_p; mmmcmzuuu.o_.: o_nmh

76

Pearson product-moment correlations were computed to further
evaluate the test-retest reliability of the V10. P10. and F810 scores
across the two evaluations. The measure of stability ranged from a low
of.L==.64 for the Verbal = Performance group for both Verbal and
Performance 105 to a high of L = .84 for the F810 of the Verbal >
Performance group. The reliability coefficients for all groups are
reported in Table 4.11. Salvia and Ysseldyke (1978) recommended that
tests have reliability coefficients of at least:.9 before making place-
ment decisions and of at least.8 before making screening decisions.
They do not state what kind of reliability they are referring to.
Although only two of the coefficients for FSIO meet the .8 criterion.
the interval used in this study is considerably longer than that used
in the majority of evaluations of test stability. Given the mean
test-retest interval of 36 months. the Wechsler IO scores do appear to

be stable for the learning disabled population.

Table 4JLL--Test-retest reliability coefficients for the WISC-R IO

 

 

scales.
VIO PIO FSIO
V>P £=.82 L=.74 L=.84
V = P g_= .64 __= .64 £.= .73
V < P __= .78 §.= .69 _,= .80

_‘
O
d.
D!
.—I
ll
0
\l
w
| '1
ll
0
\l
h
I!
O
\l
\l

 

77

W
W

HNDQIh§$1§.lQ. The severity of the discrepancy between ability
and achievement is not related in a consistent manner to the type
of educational placement.

At the time of the initial evaluation. 42 students were placed in
teacher consultant programs. 106 were in resource rooms. and 13 were in
self-contained classrooms. Thirty-five students were either diagnosed
as learning disabled but continued in regular education. or were in
placements for which there was no available information. ‘The discrep-
ancies were computed between the FSIO and each WRAT subtest for both
evaluations. Analyses of variance were then calculated for each
discrepancy. comparing teacher consultant. resource room. and self-
contained placements. None of the ANOVAs was significant for the first
evaluation. (See Table 4.12.), This finding supports the suggestion
that the factors which are used to determine the least restrictive
environment have little relationship to the degree of academic dis-
ability.

ANOVAs were computed for the same three groups using the discrep-
ancies found at the time of the second evaluation. (See TableimJB.)
Again. none of the statistical tests reached significance. indicating
that the students in the three placements were comparable in terms of
academic disability.

A third set of ANOVAs was calculated using the type of placement
at the second evaluation as the basis for the groups. (See Table
4.140 This comparison was considered important. as the composition of

the three groups changed. At the time of the second evaluation. only

78

.mu_c: ocoumu~ :— voucoaoxm

 

>ucmaocum_o

mmm. mm..~ mam. .m.ov m:.. .m... .m.. .s... m... u..oscu..<
osm. mm..~ moo. ...~. mo.m .~.~. s~.m .m.~v mm.~ >ocmamtu..a
mc___oam

>ucmaocum_o

 

 

 

 

 

mom. mm..~ o~c.. .m.~. No.s ...~. N..m m.».~. m..m wc_emm¢
.m .mm .m .mw .m .mm mp1. .mm .m
<>oz< .m. ".m. .oo. ".mv .Ns ".mv
500m cocwmuc0u1m_om eoom 60.:Omom acmu_:mc0u Locumoh

 

.co_um:_m>o _m_u_c_ ecu um moEOUm >ocmoo.om_v h<m31muum_3 on»
:0 comma ucoEoom_a co_umo:co _m_uoam _m_u_c_ co» m<>oz< ccm .mco_um_>oc ccmccmum .mcmozun.~_.: o_nmh

79

.mu_c: oLOUmaN c. toucoaoxm

 

>ucmaoLUm_a

 

 

 

 

 

omw. wm_.~ m... Asa... mo.~ an... mm.~ Ao.~. ~:.~ u_uoE;u_L<
. . . . . . . >ucmaoc6m_c
a... mm. N .NN. .o m. mm . .N N. no N .N N. mm N ma....am
. . . . . . . >ocmaocom_o
Nam. mm. N m... .N N. mm m .m 0. mm m m.m m. sN s m=.emo¢
m Hm .w. .olm. x .olm. x ..W x
.2 .N. am. 3.: M. .N. aim.
<>o soom noc_mucounm_om eoom 06.30mom acmu.:m:0u cozumok

 

.:o_um:_m>ono. ecu um mocoom >ocmoocum_u P<mzu¢uum.3 ecu co
comma ucoeoom_a co_umo:uo _m_ooam _m_u_c_ Lo. m<>oz< new .mco_um_>ou ucmccmum .mcmozuu.m_.: 0.3m»

80

 

>ucmqocum_o

 

 

 

 

 

:oo. om_.N mmm.m Am... mm.~ Am... mm.~ AN... :4._ u.uoE;u..<
> a
Nooo. om..N N...m .N.N. .m.m .:.N. so.: .N... .o.N ucmcwflwwhm
> a
mmm. om..N .mo. .N.m. m... .a... mm.m .N.N. N... ucmmwnwwwn
<>oz< .m. ".m. .00. ".m. .N: n.m.
800m coc_mucOuuw_om EOOx 00.:Omom Nana—:mcou Locumoh

 

.co_um:_m>oumc ecu um

acosoom_a to» >ocmaocum_u h<x3umuum_3 ecu co m<>oz< cam .m:o_um_>oc ccmccmum .mcmozuu.:_.: o_nmh

81

20 students were in teacher consultant programs. 125 were in resource
rooms. and 8 were in self-contained classrooms. Forty-three students
fell into other categories. This set of comparisons produced signifi-
cant findings for the Spelling discrepancy and the Arithmetic discrep-
ancy. The students in the teacher consultant group appeared to have a
smaller discrepancy on the WRAT Spelling subtest than the other two
groups (E = 9.167. _d_f = 2.150. L < .0002). The same was true for the
WRAT Arithmetic subtest (E = 5.73. .d_f_ = 2.150. p < .002). The three
groups exhibited similar discrepancies for the WRAT Reading subtest.
The Holton and Lowell school districts do not have self-contained
classrooms. so some children from these communities may qualify for a

self-contained room but be placed in a resource room.

CHAPTER V
DISCUSSION

(News:
In this chapter the findings related to homogeneous subgroups. the
ACID test. stability of 10 scores for learning disabled children. and
the relationship of the ability-achievement discrepancy to special
education placement are examined more closely in terms of their
value for practitioners and researchers. Recommendations for future

research and clinical practice are made for each area discussed.

WWW

Although some researchers remain optimistic about the search for
subtypes of learning disabled children (Fisk 8 Rourke. 1979; Lyon.
Stewart. 8 Freedman. 1982; Nolan. Hammeke. 8 Barkley. 1983; Petrauskas
8 Rourke. 1979; Rourke 8 Finlayson. 1978; Rourke 8 Strang. 1978). the
findings in this study raise questions regarding the usefulness of
establishing homogeneous subgroups of learning disabled children. The
subgroups in this study were established using a clinical inferential
approach. so the results reported here can only be compared to other
studies using a similar approach. Little comparison will be possible
to those studies using a multivariate approach. The approach used in

this study attempted to duplicate the diagnostic strategies of

82

83

practicing school psychologists unlike the studies mentioned above.
which used either more extensive psychological testing than is commonly
used in schools. or which used complicated statistical methods. also
not used in the schools.

In looking over the data used in this study. it did appear that
subgroups other than those postulated may exist. However. the net
result would have been 10 to 12 subgroups of three to five subjects in
each. The finding that only 28% of this group could easily be placed
in a subgroup suggests several possibilities. First. it may be that
the subgroups were too narrowly defined to allow inclusion of a
majority of the subjects. ‘The problem in other learning disabled
research. however. has been that categories have been too broad and
thus have not allowed for clear description and prediction based on
those categories. The difficulty. then. is how to establish the
subgroup criterion narrowly enough to allow for description and broadly
enough to include more of the children. One approach could be to
relate the diagnostic categories to remedial strategies to allow for
the possibility of prescriptive teaching. Behavioral subgroups may be
more promising approach than intelligence-achievement test defined
groups. McKinney (1984). for example. utilized a behavioral classifi-
cation method to identify four subgroups of learning disabled children.
suggesting another practical approach to further clarifying the charac-
teristics of this population.

Second. if learning disabled children cannot be easily placed in a

subcategory by practitioners in the field using the practitionerks

84

methods. it seems unlikely that many of the research findings will be
used in actual practice. It would seem that the ultimate goal of
research related to subgroup identification would be to improve educa-
tional placement and planning for this group of children. yet the
methods used for placement are so esoteric that such a generalization
is not credible. McKinney (1984) noted that certain statistical clas-
sification procedures (Orfactor analysis and hierarchical cluster
analysis) are designed to group individuals into clusters. They do not
guarantee that the clusters are meaningful. nor do they have prognostic
value. He also raised questions about the utility of multivariate
methods if the groupings cannot be related to different developmental
outcomes.

A third issue raised by this study has to do with the reality of
the homogeneous subgroups of learning disabled children. Do they
really exist. or are we simply following a hopeful dream as in the
research on characteristic WISC-R profiles. scatter analysis. and
poorly established cerebral dominance? In looking over the research
cited in Chapter II. it was noted that all of the University of Windsor
studies used very large subject pools from which to draw relatively
small sample sizes for the actual studies. The range was 82 subjects
from a pool of 350 (Rourke. Dietrich. 8 Young. 1973) to 264 subjects
from a pool of 2.500 (Fisk 8 Rourke. 1979). The latter (264) was the
largest subject sample used by Rourke and his coresearchers. Several
of the studies used N5 of 45 to 50 (Rourke 8 Telegedy. 1971; Rourke.

Yanni. MacDonald. 8 Young. 1973; Rourke 8 Finlayson. 1978: Rourke 8

85

Strang. 1978). ‘The small sample sizes used suggest that these
subgroups may not be representative of a majority of learning disabled
subject pools either. adding support to the question of the existence
of homogeneous subgroups into which the majority of learning disabled
children can be classified. Again. behavioral subgrouping may provide
another means of exploring this issue.

Related to this is the possibility that the subgroups in previous
research were somewhat artificially established. Very few children
with a stronger Verbal IO or with stronger Reading and Spelling scores
are diagnosed as learning disabled. In general. the literature
(Farnham-Diggory. 1978; Huelsman. 1970; Ackerman. Peters. 8 Dyckman.
1976; Anderson. Kaufman. 8 Kaufman. 1976; Smith. Coleman. Dokecki. 8
Davis. 1977a) has shown a far greater proportion of children having
higher Performance IO. Yet most of the earlier studies did have the
Verbal IO.greater than Performance group. It was not possible to find
such a group in this study. Nolan. Hammeke. 8 Barkley (1983) also had
difficulty establishing a Verbal > Performance group.

The final possibility is that those who are making the diagnosis
of learning disability are not adhering to the definition such that
children with other problems are being categorized as learning
disabled. The federal definition has often been criticized as too
general (Farnham-Diggory. 1978; Oakland 8 Goldwater. 1979: Shepard.
Smith. 8 Vojir. 1983). Shepard et a1. (1983) analyzed a group of 800
children identified as learning disabled and found that "true" learning

disabled children comprised only 43% of this group. She suggested that

86

"the label applied for the purpose of providing services cannot be
assumed valid)‘ If the label cannot be assumed valid. the possibility
of defining subgroups is severely hampered. McKinney (1984) stated
that most of the research on learning disability subtypes has been
confined to reading disability. Thus. it is reasonable to assume that
learning disabled samples defined by the federal guidelines would be

more heterogeneous than the samples in previous studies.

WW

Recommendations for future research in definition of homogeneous
subgroups are: (l) to establish how accurately the federal and state
guidelines are being used. as inaccurate diagnosis will continue to
confound any findings; and (2) to attempt to define the subgroups such
that they can be identified by practitioners in diagnostic settings
within the school districts. McKinney (1984) has provided another
possible route in his work with behavioral subgroups. He has also
suggested that future studies test the utility of alternative interven-
tions for learning disabled children by using trait x treatment

paradigms.

WW1):
The results provided little support for the concept of predicting

future performance of Reading. Spelling. and Arithmetic achievement as
measured by the WRAT based on subgroup membership. 'The findings were

limited by all of the questions raised in the previous section and

87

imply that further research here should be delayed until such time that
more representative and consistent subgroups can be defined.

MW
Wm

The use of the ACID test as a determinant of a subgroup of
learning disabled children found little support in this study. As a
result of the small number of subjects who fell into this group. it was
not possible to make any meaningful prediction of reading problems.

The fact that earlier studies (Kaufman. 1979; Cullen. Boersman. 8
Chapman. 1981: Vance. Gaynor. 8 Coleman. 1976: Bernard. 1978; Tabach-
nick. 1979; Fisk 8 Rourke. 1979) have also not been able to find this
pattern with any regularity suggests that the base rates for this

particular profile may be quite low.

Wan

The low base rates may limit the diagnostic utility of the ACID
test. ‘The prescriptive validity was also tested in this study with
little support. At this point. the construct of the ACID profile has
no demonstrated utility for either diagnosis or prescription. Future
research might focus on matching those learning disabled children who
do exhibit the ACID profile to learning disabled children who do not.
in order to further evaluate the question of differences in

responsivity to remedial reading programs.

88

W
LeamMsabJedflfldLen

The evaluation of the Verbal. Performance. and Full Scale IO
scores indicated that these scores are quite stable for learning
disabled children. The hypothesis that Verbal IO would decrease over
time was not supported. The mean change for all three IO scores was
less than one point over a 3-year period. Statistically significant
changes were noted for the subgroups based on VIOPPIQ discrepancies.
However. these were not educationally meaningful changes as the largest
changes represented a four-point decrease in Verbal IO and a four-point
gain in Performance 10 for the Verbal IO > Performance IO group. ‘These
changes fall within or close to the expected range given the Sem for
these tests. None of the Full Scale IO difference scores were statis-
tically significant. and the Verbal and Performance IO changes tended
to be in a direction which would reduce the size of the discrepancy.
Further analyses to determine what proportion of this effect was due to
regression to the mean did not indicate that regression to the mean was
a large factor in the observed change. The regression effect is always
a problem in studies such as this. where the subjects are chosen on the
basis of poor performance on a test and are not randomly assigned to
groups. The lack of support for regression is somewhat surprising.

The finding of stability is consistent with previous research using
shorter test-retest intervals in both learning disabled and normal
children (Wechsler. 1974; Tuma 8 Applebaum. 1980: Vance. Blixt. Ellis.

8 DeBell. 1981; Smith. 1978L The majority of subjects remained in the

89

same intellectual category (124 or 63%). A lesser number of subjects
changed IO categories (72 or 37%). and none of the changes was over
more than one intellectual category.

Sattler (1982) reported that 105 obtained after the age of 5 or 6
tend to remain fairly stable although individuals may show great
fluctuation. (Muldren with high 105 tend to show greater variability
than children with low 105. McCall (1977) found that 105 obtained by
the age of 5 years correlated highly with adult IQs L50 and higherL
The correlations between IO scores in this study were greater than the
above. suggesting that the stability of learning disabled childrenus
WISC-R IO scores is no less stable than that of normal children.

The finding of stable IO scores for the learning disabled
population was reassuring in a number of areas. First. the stability
indicates that the test commonly used to estimate ability can be relied
upon to give consistent results. If the finding had been of unrelia-
bility. a question would have been raised about the use of this test
for making periodic assessments and evaluations. ‘The practice effect
reported in other studies appears to decrease with the longer interval.
suggesting less of a problem in making erroneous decisions at the time
of re-evaluation. Second. learning disabled children do not appear to
be falling progressively further behind their non-learning-disabled
peers on the measure of general ability. Kaufman (1981) suggested that
Verbal IO.might decrease as learning disabled children would have

progressively more difficulty on Bannatyne's Acquired Knowledge Factor.

90

His question was not addressed directly in this study. but the consis-

tency in test performance would not lend credence to his hypothesis.

Becomendatmns

Kaufman's hypothesis does raise an interesting question for future
research in this area. Are Bannatyne's factors also stable. or does
the performance on individual subtests vary considerably over time?
Future research might focus on the stability of the various factors of

the WISC-R for both normal and learning disabled populations.

W
W

No significant differences were found in the ability-achievement
discrepancy of students placed in teacher consultant. resource room. or
self-contained programs at the time of the initial placement. The lack
of differences suggests that the multidisciplinary education teams are
using criteria other than those spelled out in PL 94-142 for making
placement decisions. According to this law. the severity of the
ability-achievement discrepancy is to be used for both diagnosis and
determining the least restrictive environmemt for each child. Other
researchers have also suggested that placement decisions have little to
do with the actual data collected on the children. Ysseldyke and
Algozzine (1983) found that placement decisions are more closely
related to naturally occurring student characteristics. Ysseldyke
(1983) reviewed the literature on the team decision-making process and
came to the conclusions that the process is inconsistent and that the

teams function largely to endorse teacher-identified problems. 1k;

91

reported that although many data are collected on students. their
bearing on eligibility and placement is largely ignored by the team.
Analysis of the discrepancies as they related to the type of
placement at the time of the re-evaluation indicated differences among
the groups for Spelling and Arithmetic skills. but not Reading.
Students in teacher consultant programs appeared to have less severe
discrepancies than those in resource rooms or self-contained programs.
The finding of significant differences for two academic areas suggests
that the ability-achievement discrepancy may be used as a more
essential factor in determining the placement of older students than of
younger students. The lack of specific academic skills may be more
disturbing in older children. who have less time in which to acquire
them than younger children. .Another hypothesis is that teachers may
focus more on teaching reading as a basic skill than on arithmetic or
spelling. so that the reading discrepancy changes less. A third
hypothesis has to do with the nature of the WRAT Reading subtest. which
is essentially a test of single-word recognition and pronunciation.
While these students may all continue to be able to sound out words
equally well. greater variation might be found using a test of reading
comprehension. Finally. some of the school districts from which these
subjects were drawn did not have self-contained rooms for learning
disabled students. so it may be that the resource room groups contain
students who would be placed in self-contained rooms if one had been

available.

92
W11
Future research in this area might focus on further clarification
of the different placement strategies for older versus younger learning
disabled students. Second. the discrepancy in reading may be more
meaningful if it is evaluated using a test of comprehension rather than

word recognition. A study using such a test might show a discrepancy

at the older age group which the WRAT is not designed to test.

REFERENCES

93

REFERENCES

Ackerman. P.. Dykman. R.. 8 Peters. J. Hierarchical factor patterns
of the WISC as related to areas of learning deficit. .EeLcootual
m M919: £311.15: 1975: 52: 583-615.

Anderson. M.. Kaufman. A.. 8 Kaufman. N. Use of the WISC-R with a
learning disabled population: Some diagnostic implications.

Esxcboloox in the Schools. 1976. .13. (4). 381-386.

Bannatyne. A. Language. Loading. and Joaming disabilities.
Springfield. 111.: Charles C. Thomas. 1971.

Bannatyne. A. Diagnosis: A note on recategorization of the WISC

scaled scores. Journal oi Loathing DJoabJJJiios. 1974.
1. 272-274.

Belmont. L.. 8 Birch. H. Intellectual profile of retarded readers.
Eamopiual and Motor: Skills. 1966. 22. 787-816.

Bernard. R. Adoscoiotionotaoooulationofssboolmuflodjoaming
disablsdobfldnonintbssiatooimsmoanaomsscontainnu;
ables. Unpublished doctoral dissertation. Michigan State
University. 1978.

Bortner. M.. Hertzig. M.. 8 Birch. H. Neurological signs and intel-

ligence in brain injured children. lounnal.oi.§nfislal.Edncation.
1972. Q. 325-333.

Breen. M.. 8 Prasse. D. A comparison of the 1976 and 1978 WRAT:

Implications for the learning disabled. .louLnal.o£.LoaLning
W. 1982. 15(1). 15-16.

Byrd. P.. Buckhalt. J.. 8 Byrd. K. Age differences in WISC-R subtest
ranks of children experiencing academic-difficulties. ‘Esyghologi:
$.11 m: 1981: 5.8,: 599.604.

Campbell. 0.. 8 Stanley. J. Exoeflmontalandouasjflxnonimonm
.d2519n5.19£.£a§aazch. Boston: Houghton-Mifflin Co.. 1963.

Clarizio. H.. 8 Bernard. R. Recategorized WISC-R scores of learning
disabled children and differential diagnosis. .Esxshnlngx.1n.thfl
5.2119215: 19819 1.8.: 5'12.

94

95

Clarizio. H.. 8 McCoy. G. .BehaxioL.d15QLdeL§.1n.gh11dLen (3rd ed.).
New York: Harper 8 Row. 1983.

Clements. 5.. 8 Peters. J. Minimal brain dysfunctions in the school
age child. 8129111195 of Genenal Estonian. 1962. 6. 185-197.

Clements. S. Minimalonaindxstunctioninobiidnemlezminolggxand
idemjjjjgatign::fiha§e one (NINDB Monograph f3. U.S. Public Health
Service Publication No. 1415) Department of Health. Education. and
Welfare. Washington. 0.0.. 1966.

Cratty. B. Motor activities and learning disabilities: A look ahead.

In M. Krasnoff (Ed.). Leacning disabilities: A decade ahead.
Ann Arbor. Mich.: Institute for the Study of Mental Retardation
and Related Disabilities. 1974.

Cullen. J.. Boersma. F.. 8 Chapman. J. Characteristics of third-grade

learning disabled children. ‘LeaLning.Qisahilitx.fluerte£l¥. 1981.
A. 224-230.

Denckla,14. Research needs in learning disabilities. .JouLnal of
Learning Disabilities. 1973. a (7). 441-450.

Doehring. D. Eattenns of impairment in specific reading disability-
Bloomington: Indiana University Press. 1968.

Eaves. L.. 8 Crichton. J. A five year follow-up of children with
minimal brain dysfunction. .Academic.IneLapy. Winter 1974-1975.
JD.(2). 173-179.

Eaves. L.. Kendall. D.. 8 Crichton. J. Early identification of learn-

ing disabilities: A follow-up study. ‘JQuLnal.oI.LeaLning.Dis:
MS: 1974: l (10): 632-638.

Farnham-Diggory. S. .LeaLning disabilities. Cambridge. Mass.:
Harvard University Press. 1978.

Fisk. J.. 8 Rourke. B. Identification of subtypes of learning dis-
abled children at three age levels: A neuropsychological. multi-

variate approach. ‘JouLnal‘oi.Clinioal.Nennonsxshology. 1979.
l.(4). 289-310.

Fox. 8.. 8 Routh. 0. Reading disability. phonemic analysis. and
dysphonetic spelling: A follow-up study. JguLnal of Clinical
.901 d Esxnhologx. 1983. 12 (1). 28-32.

96

Frauenheim. J.. 8 Heckerl. J. Severe dyslexia--A longitudinal study
of psychological (WISC/WAISR) and achievement test performance
in dyslexic adults. Paper presented at the William Beaumont
Hospital Neuro—Educational Center Annual Spring Conference. Royal
Oak. M1Ch.o March 309 1984.

Gearhart. B. teaming disabilities; Educational stnategies.
St. Louis. Mo.: C. V. Mosby. 1973.

Huelsman. C. The WISC subtest syndrome for disabled readers. .Eeneeo:
tual and Motor Skills. 1970. 3.0. 535-550.

Jastak. J.. 8 Jastak. S- Manuali IbeWideBangeAchiemnentlsst.
Revised Ed. Wilmington: Jastak Associates. Inc.. 1978.

Jastak. J.. 8 Jastak. 5. Manual; IbeifldeBangeAcbiexementIest.
Revised Ed. Wilmington: Jastak Associates. Inc.. 1978.

Johnson. D.. 8 Mykelbust. H. .Leanningidisabilities:.Edueational
principles.and.pnaetices. New York: Grune 8 Stratton. 1967.

Johnson. 5.. 8 Morasky. R. .LeaLning,disabilities. Boston: Allyn
8 Bacon. 1977.

Kaufman. A. Factor analysis of the WISC-R at 11 age levels between

6 and 16 years. Journal of Consulting and Clinical Esxcboiogu.
1975. 43.. 135-147.

Kaufman. A. Verbal-performance IO discrepancies on the WISC-R.

JguLnalgtnonsultingandminioalEsxcboiogx. 1976.44.
739-744. (a)

Kaufman. A. A new approach to the interpretation of test scatter on

the WISC-R. 19.0.2031 of Learning Disabilities. 1976. 9..
160-168. (b)

Kaufman. A. Intelligent testing 11th the film. New York:
Wiley 8 Sons. 1979.

Kaufman. A. The WISC-R and learning disabilities assessment: State of

the art. m DI mug W: 1981: M (9):
520-526.

Kirk. S. .Educating exceptional childLen (2nd ed). Boston: Houghton-
Mifflin. 1972.

Koppitz. E. .Qnfldnennitbieanningdisabiiitiessdflxexeatioliow:
up study. New York: Grune 8 Stratton. 1971.

97

Lambert. N.. 8 Sandoval. J. The prevalence of learning disabilities in
a sample of children considered hyperactive. flournal.o£.AbnoLmal
MW: 1980: .8, (1): 33-50.

Law. J.. Box. 0.. 8 Moracco. J. A validation of recategorized WISC-R
scores of learning disabled children. Edusetion. 1980. 10_(2).
195-199. °

Levin. E.. Zigmond. N.. 8 Birch. J. A follow-up study of 52 learning
disabled adolescents. Paper presented at AERA. Montreal. Canada.
1983.

Lutey. C. Indixidualinteiiigenoetestingsnmanualandsoursebook
(2nd ed.). Greeley. Col.: Carol J. Lutey Publishing. 1977.

Lyon. R.. Stewart. N.. 8 Freedman. D. Neuropsychological characteris-
tics of empirically derived subgroups of learning disabled

readers. Journal of Clinical mm: 1982. i (4). 343-
345. ‘

McCall. R. 8. Childhood IOs as predictors of adult educational and
occupational status. .fisience. 1977. 121. 482-483.

McKinney. J. The search for subtypes of specific learning disability.
JournaloiLearningDisabiiities. 1984. 11(1). 43-50.

McManis. D.. Figley. C.. Richert. M.. 8 Fabre. T. Memory for designs.
Bender-Gestalt. Trail Making Test and WISC-R performance of
retarded and adequate readers. .Eercentual.and.MQIC£.§Kills.
1978. 46. 443-450.

Mehrens. W.. 8 Lehmann. I. Measurement and exaluation in eduoation
.and psychology (2nd ed.). New York: Holt. Rinehart 8 Winston.
1975.

Moore. 0.. 8 Wielan. C. Scatter indexes of children referred for

reading diagnosis. Journal of Learning Disabilities. 1981. i4.
(9). 511-514.

Nolan. 0.. Hammeke. T.. 8 Barkley. R. A comparison of the patterns of
the neuropsychological performance in two groups of learning

disabled children. .lnunnal.91.Qlin19al.Child.E&¥ChClngx. 1983.

Oakland. T.. 8 Goldwater. 0. Assessment and interventions for mildly
retarded and learning disabled children. In G. Phye 8 D. Reschly

(Eds.). Sobool osisbolggi: Esrsoestises and issues. New York:
Academic Press. 1979.

98

Owens. F.. Adams. P.. Forrest. T.. Stolz. L.. 8 Fisher. S. Learning
disorders in children: Sibling studies. .MonogLaphs of the
SogietxiorBesearobinCnfldCereJooment. 1971. Serial No. 144.

Paal. N.. Hesterly. 0.. 8 Wepfer. J. Comparability of the WISC and the

WISC-R. Journal of Learning Disabilities. 1979. .12 (5).

Petrauskas. R.. 8 Rourke. B. Identification of subtypes of retarded
readers: A neuropsychological. multivariate approach. JouLnal

.oi.Clinical.Meuronsxobologx. 1979..l (1): 17-37.

Porter. A. Anahsisstrategiesiorsomeoomoneuaiuationoaradigms
(Occasional Paper No. 21). East Lansing: Michigan State University

Office of Research Consultation. 1973.

Reed. J. The ability deficits of good and poor readers. .Jeurnal.oi
Learning Cisabiiities. 1968. 2. 134-139.

Reynolds. C. The fallacy of "Two Years Below Grade Level for Age" as
a diagnostic criterion for reading disorders. .JouLna1.o£.Schoo1
.Esyonology. 1981..19 (4). 350-357.

Rie. E.. 8 Rie. H. Reading deficits and intellectual patterns among
children with neurocognitive dysfunctions. Intelligence. 1979.
3., 383-389 0

Robeck. M. Identifying and preventing reading disabilities. In

J. A. R. Wilson (Ed.). Ciaonosisoiiearningdisaoiiities. New
York: McGraw-Hill. 1971.

Rourke. 8. Reading retardation: Lag or deficit? In R. Knights 8

D. Bakker (Eds.). 1119 neumpsxsbologx of learning disorders;
Ibeoretioal.aooroaobes. Baltimore. Md.: University Park Press.
1976.

Rourke. B. Neuropsychological assessment of children with learning
disabilities. In S. Filskov 8 T. Boll (Eds.)..flendheek.e£
olinical neuroosxcbologx. New York: Wiley 8 Sons. 1981.

Rourke. B.. Dietrich. D.. 8 Young. G. Significance of WISC verbal-
performance discrepancies for younger children with learning

disabilities. Eerceotual and Motor Skills. 1973. 36.
275-282. .

Rourke. B.. 8 Finlayson. M. A. Neuropsychological significance of
variations in patterns of academic performance: Verbal and

visual-spatial abilities. .Journal.ot.Abnormal.Cbild.Esxobology.
1978. §.(1). 121-133.

99

Rourke. B.. 8 Strang. J. Neuropsychological significance of variations
in patterns of academic performance: Motor. psychomotor and

tactile-perceptual abilities. .JouLnal.oi.Eediat£ie.Eexehelegy.
1978. 3. (2). 62-66.

Rourke. B.. 8 Telegedy. G. Lateralizing significance of WISC verbal-
performance discrepancies for older children with learning dis-

abilities. Eerceotual and Motor Skills. 1971. 33. 875-883.

Rourke. B.. Yanni. D.. MacDonald. G.. 8 Young. G. Neuropsychological
significance of lateralized deficits on the Grooved Pegboard Test
for older children with learning disabilities. .JouLnal.o£.Con;
suiting and Clinioal Esyobologx. 1973. ii (1). 128-134.

Rourke. B.. Young. G.. 8 Flewelling. R. The relationships between
WISC verbal-performance discrepancies and selected verbal.
auditory-perceptual. visual-perceptual and problem-solving
abilities in children with learning disabilities. .JguLnal of
Clinical Esygnology. 1971. 21. 474-479.

Rugel. R. WISC subtest scores of disabled readers: A review with

respect to Bannatyne's recategorization. .Journal.oi.Learning
LMsabilities. 1974. Z.(l). 57-64.

Ryckman. 0. Searching for a WISC-R profile for learning disabled

children: An inappropriate task? JguLnal of Learning Disabili:
ties. 1981. 1.4. (9). 508-511.

Salvia. J.. 8 Ysseldyke. J. .Assessment.in.soeeial.and.remedial
.efiueetien. Boston: Houghton-Mifflin Company. 1978.

Sattler. .1. Assessment of oniidrenls inteliigenoe and special
abilities (2nd ed.). Boston: Allyn 8 Bacon. Inc.. 1982.

Sattler. J.. 8 Ryan. J. Relationship between the WISC-R and WRAT in
children referred for learning disabilities. {Esxobologx.in.tbe
55.11.9915: 1981: .18. (3): 290-92.

Shepard. L.. Smith. M. L.. 8 Vojir. 0. Characteristics of pupils

identified as learning disabled. .AmeLigan.Edugatigna1.BeseaLgh
Journal. 1983. 20(3). 309-331.

Smith. M. Stability of WISC-R subtest profiles for learning disabled
children. .Esxobologx.in.tbe.Sobools. 1978. 15.(l). 4-7.

Smith. M.. Coleman. J.. Dokecki. P.. 8 Davis. E. Intellectual char-
acteristics of school labeled learning disabled children.

W Children. 1977: 53.: 352-357. (a)

100

Smith. M.. Coleman. J.. Dokecki. P.. 8 Davis. E. Recategorized WISC-R
scores of learning disabled children. .Jguznal of Learning 21::
abilities. l977. .].Q (7). 437-443. (b)

Stevenson. L. WISC-R analysis: Implications for diagnosis and educa-
tional intervention of LD children. Paper presented at the Annual
International Convention. The Council for Exceptional Children.
Dallas. TX. April 1979.

Strauss. A.. 8 Lehtinen. L. Psychopathology and education of the brain

inJur‘ed child. Jeutnal .ot Meueus and Mental Disease. 1942. a.
259-289.

Strauss. A.. 8 Werner. H. Disorders of conceptual thinking in the

brain injured child. Jeumal ef Memus and Mental Disease.
1942. 9a. 153-172.

Swartz. G. .Ine.language.leanning.s¥stem. New York: Simon 8 Schuster.
1974.

Tabachnick. B. Test scatter on the WISC-R. 'Jgnnnal.91.Leann1ng.flis:
.ahllliies. l979. 12 (9). 626-628.

Tabachnick. B.. 8 Turbey. C. WISC-R scatter and patterns in three
types of learning disabled children. Paper presented at the
Western Psychological Association Conference. Los Angeles. CA.
April 9-l2. l981.

Thompson. R.. Jr. The diagnostic utility of Bannatyne's recategorized
WISC-R scores with children referred to a developmental evaluation

center. .Esxshnlngx.1n.1h§.59h9915. 198l. lfi.(l). 43-47.

Tuma. J.. 8 Applebaum. A. Reliability and practice effects of WISC-R
IQ estimates in a normal population. .Edueatienal.and.Esxehe1991:
gal Measunement. 1980. 411. 671-678.

Valett. R. .Mndlixlng.thldnanl§.hehaxign. Belmont. Cal.: Fearon.
1969.

Vance. H.. Blixt. 5.. Ellis. R.. 8 Debell. S. Stability of the WISC-R
for a sample of exceptional children. .lgunnal of Clinical
.Esxchnlegx. l98l. 31.(2). 397-399.

Vance. H.. Gaynor. P.. 8 Coleman. M. Analysis of cognitive abilities

for learning disabled children. .Esxshelegx.1n.the.§£hnnls. 1976.
ll.(4). 477-483.

Vance. H.. 8 Singer. M. Recategorization of the WISC-R subtest scaled
scores for learning disabled children. ,Jgunnal of Learning
W. l979. .12 (7). 487-49l.

101

Vance. H.. Wallbrown. F.. 8 Blaha. J. Determining WISC-R profiles

for reading-disabled children. .Jgunna1.gI.Leann1ng.Diijigulties.
l978. 11 (lo). 657-661.

Veres. v. E. Anexalgatjeneftbeusegtmsubtestssenestm:

the identjfleatjen ef. mfltflx bandleaeped ehflenen. Unpublished
doctoral dissertation. Michigan State University. 1982.

Wallbrown. F.. Vance. H.. 8 Blaha. J. Developing remedial hypotheses

from ability profiles. Jeutnal at teaming Disabuities. 1979.
.12 (a). 557-56l.

Wechsler. D. Ine Weehslen Intelligence Seals .ng: .QbJJgLenzfleused
.Manual. New York: The Psychological Corporation. 1974.

Wells. C. Ammeammaugxetemjgtengmeegbxtbneeeasje
mmmmmwmmm.
(Doctoral dissertation. University of Northern Colorado) Ann
Arbor. MI: University Microfilms. 1970. No. 71-14. 543.

Werner. E.. 8 Smith. R. .Kauails ghildnen game of age. Honolulu:
The University of Hawaii Press. 1977.

WhltO: W.. A119Y: G.o Deshler. 0.9 Schumaker. Jo: Warner. M.o &
Clark. F. Are there learning disabilities after high school?
Excenilnnal.9hlln£en. l982. 42,(3). 273-274.

Witkin. H.. Dyk. R.. Faterson. H.. Goodenough. D.. 8 Karp. S.
.Esyghglggigal differentiation. New York: John Wiley. 1962.

Ysseldyke. J. Current practices in making psychoeducational decisions

about learning disabled students. .Jgunnal 91 Learning Disabilrz
Ides. 1983..l§ (4). 226-233.

Ysseldyke. J.. Algozzine. B.. Shinn. M.. 8 McGue. M. .fiimilanlties

anddittenengeseetneenungenaQDJexensandstugentslabeled

disablegtldentiealtwinsntbgtfietentmetbens
(Research report No. l3). Minneapolis: University of Minnesota
Institute for Research on Learning Disabilities. l979.