THE INFLUENCE OF SELECTIVE ATTENTION ON THE
PERFORMANCE OF LEARNING DISABLED STUDENTS
DANIEL WALKER BECTON
A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
Daniel Walker Becton
Attention has been compared to the beam of a searchlight moving
about in the dark. It may focus on the world without or on inner
thoughts and fancies. But the act of paying attention is much more
than the focusing of a searchlight. It selects and draws into the
foreground, while initiating neuronal action on the stage of conscious-
ness. Other action may go on off stage. Off-stage action is not
arrested, but it is ignored only what happens on stage is recorded
Wilber-Penfield, 1969, p. 164-5
I wish to express my sincere appreciation to my committee members,
Dr. Robert Jester, Dr. Don Avila, and Dr. William Wolking, for their
support, encouragement, and constructive suggestions over these years.
Special thanks to Dr. Jester for detailed help in many meetings at
various times and places, including his home during his vacation.
I would like to acknowledge the extensive help I received from
Dr. Bill Edenfield in making possible the collection of the data. I
also appreciate the cooperation of Dr. Ron Bobay, Dr. Lee Rowell, Pat
Gardner, Peggy Williams, Lori Hadley, and Sarah Becton, my mother, in
acquiring the data.
I would like to extend my appreciation to Mrs. Carolyn West, Director
of Special Services, for her encouragement. I would also like to thank
the administration of Central Florida Community College for their support.
A very special thank you goes out to Mrs. Bonnie Dingler. As
secretary for Special Services, my friend, and the typist throughout
many drafts, she has been extremely loyal in her tolerance and encourage-
ment of my work.
To my parents, who undoubtedly are breathing a sigh of relief,
I thank them for their sacrifices, support, and encouragement through-
out my educational career. Finally, I thank my fiancee, Patricia, and
the boys, Mike and David, for their interest and patience during this
TABLE OF CONTENTS
ACKNOWLEDGEMENTS ................................ ....... iv
ABSTRACT............... .................................. vii
I INTRODUCTION.............................................. 1
Rationale for the Study .............................. 1
Specification of Terms................................ 3
Statement of the Problem .............................. 4
Significance of the Study ............................. 5
Research Questions................................... 5
-Limitations of the Study.............................. 7
II REVIEW OF THE LITERATURE ... .............................. 9
Selective Attention................................... 9
Precision Teaching................................... 29
Modality Processing................................... 37
Multisensory Teaching ................................ 46
III METHODOLOGY............................................... 48
Hypotheses ........................................... 48
Research Design.................................. ..... 53
Subjects ......................... .................... 61
IV ANALYSIS AND RESULTS ...................................... 78
Section One. ......................................... 78
Section Two ....................................... 94
Section Three ...................................... 112
Summary .............................................. 125
V DISCUSSION, CONCLUSIONS, LIMITATIONS,
IMPLICATIONS & RECOMMENDATIONS.......................... 127
Discussion.. ........................................ 127
Conclusions ........................................ 135
Theoretical Implications............................. 139
Recommendations for Further Research................. 141
A SPECIFIC LEARNING DISABILITIES GUIDELINES FOR
MARION COUNTY............................................. 142
B HAGEN'S CENTRAL INCIDENTAL ATTENTION TASK................. 147
C TEACHER RATING OF STUDENT'S ATTENDING BEHAVIOR............ 153
D META-ATTENTION TASK .................................... 156
E DATA COLLECTION FORM................................... 164
F TEACHER RATING FORM FOR SIMS PROGRAM...................... 166
G CHARACTERISTICS OF THE LD STUDENTS ....................... 169
H RAW DATA: SELECTIVE ATTENTION AND READING................ 175
REFERENCE NOTES. ............................................. 187
REFERENCES .................................. ................... 188
BIOGRAPHICAL SKETCH............................................ 198
Abstract of Dissertation Presented to the Graduate Council
of the University of Florida in Partial Fulfillment of
the Requirements for the Degree of Doctor of Philosophy
THE INFLUENCE OF SELECTIVE ATTENTION ON THE
PERFORMANCE OF LEARNING DISABLED STUDENTS
Daniel Walker Becton
Chairperson: Dr. Robert E. Jester
Major Department: Foundations of Education
The purpose of this study was to investigate the importance of
visual selective attention to learning disabled (LD) students' reading
performance. Another objective was to examine how this was influenced
by modality preference. The final goal was to investigate the reli-
ability and validity of the measurement instrument, Hagen's Central
Incidental Attention Task.
Section One compared a sample of LD students in the Systematic
Instructional Management Strategies (SIMS) precision taught reading
program (N = 19) to LD students in a multisensory program (N = 20).
Selective attention was significantly related to the dependent vari-
able, reading recognition, but the reading group and the interaction
were also significantly related to reading. Consequently, this result
was possibly due to differences existing between the subjects in the
reading groups prior to the study.
Section Two used a different, larger sample of LD students from
the SIMS program group (N = 67). No significant relationship between
selective attention and reading was found. Modality preference was
significantly related to reading achievement only when analyzed with
selective attention defined as group levels. Modality preference was
not significantly related to selective attention scores. Students
with low performance in visual modality subtests had decreasing selec-
tive attention scores paired with increasing intelligence (IQ) scores,
suggesting that modality preference was learned. However, the inter-
action was not significant.
Section Three used LD students (N = 16) from the multisensory
group to study the selective attention task. The results showed low
correlations with other measures of selective attention. A meta-
attention task showed correlations between selective attention and
the rating of "interest" as an important condition for attention.
Age and achievement performance correlated positively with rating
"reward" and negatively with rating "quiet" as important conditions.
The instrument did correlate significantly with teacher ratings of
some classroom attending behaviors.
In summary, the relationship found between selective attention
performance and reading achievement was not significant. Combining
central and incidental scores (C + I) for selective attention did pro-
duce the closest relationships. Modality preference seemed to be a
learned adaptation, not significantly related to selective attention.
The purpose of this study was to investigate the importance of
visual selective attention to learning disabled students' reading
performance. This study also examined how this relationship was
influenced by the students' modality preferences.
Rationale for the Study
Behaviors related to deficits in attention have often been cited
as a common characteristic of children with learning disabilities (LD).
Furthermore, many research studies and theories have asserted the impor-
tance of studying attention for a better understanding of learning dis-
Within the medical literature on learning disabilities, Silver
(Note 1) describes minimal brain dysfunction (MBD) children as dis-
tractible with short attention spans. He even refers to a subgroup
of "attentional deficit" children. Within psychological practice, the
Diagnostic and Statistical Manual of Mental Disorders (American Psychi-
atric Association, 1980) details the Attention Deficit Disorder which
is sometimes used in diagnosing children who have learning disabilities.
In special education, Mercer's (1979) basic textbook on learning dis-
abilities presents the cognitive process interpretation, which focuses
on attention and memory, as one of three major contexts for interpret-
ing the psychological processes involved in learning disabilities.
Researchers studying learning disabled children have documented
LD students' attentional difficulties. Conners, Kramer, and Guerra
(1969), Grassi (1970), Lasky and Tobin (1973), and Swanson (1980) have
investigated the auditory attentional difficulties. Anderson, Halcomb,
and Doyle (1973), Atkinson and Seunath (1973), Hallahan, Kauffman and
Ball (1974), Mercer (1975), and Swanson (1980) have documented the
attentional deficits in the visual domain. There is evidence that this
extends to inattentiveness in the classroom (Bryan & Wheeler, 1972)
and influences visual selective attention strategies in reading (Schworm,
1982). Torgesen (1977) has studied how these problems in selectively
attending lead to general deficiencies in problem solving strategies.
Samuels and Edwall (1981) asserted that "attention emerges as one likely
candidate for intensive investigation in the search for the etiology
of learning disability" (p. 353). Ross (1976) recommended that the
definition of a learning disability should include the inability to
sustain selective attention.
In the literature, no studies were found which examined the re-
lationship of selective attention to the reading performance of LD
students while in a precision teaching program. Hallahan (Note 2)
also reported that this relationship has not been investigated. In-
vestigation of the influence of selective attention on reading is
designed to increase understanding of the implications of the selective
attention literature for applications in classes for learning disabled
Specification of Terms
Learning disabled children. These were children identified as
such by a public school system. The selective criteria included an
intelligence test score not less than two standard deviations below
the mean, academic achievement below "expectancy age" guidelines, and
evidence of a psychological process disorder (see Appendix A).
Attention or selective attention. Selective attention refers
to the ability to attend to relevant features of a stimulus array that
also contains irrelevant proximal distractors. This was operationally
defined as scores obtained on the Hagen Central Incidental Attention
task (Mercer, 1975) as follows (see Appendix B):
1. The central task (C) was the one that the LD students
received instructions about initially.
2. The incidental task (1) was the later, unexpected
3. The selective attention efficiency index (%C %I) was
the percent of the central task correctly completed
minus the percent of the incidental task correctly
completed (Tarver, Hallahan, Kauffman, & Ball, 1976).
4. The combined selective attention measure (C + I) was
simply the sum of both measures, reflecting all of
the interaction, distraction, and capacity.
5. The selective attention levels were defined by com-
bining overall capacity (C + I) with efficiency
(%C %I) or incidental task scores (I).
Reading achievement. This was defined by reading test scores on
either the Wide Range Achievement Test (WRAT) (Jastak & Jastak, 1978),
or the Peabody Individual Achievement Test (PIAT) (Dunn & Markwardt,
Modality preference. This was defined by performance on the Detroit
Tests of Learning Aptitude (Baker & Leland, 1967). Three groups were
identified. They were LD students with only low visual subtests, LD
students with only low auditory subtests, and LD students with a mixed
preference, i.e. any other combination.
Attending behavior. This was defined by a series of questions
about classroom behaviors that reflect attending (see Appendix C).
Meta-attention. The child's awareness or understanding of his/
her own attentional processes, strategies, and capacity. This was
tested using the instrument used by Loper, Hallahan, and lanna (1982)
(see Appendix D).
Statement of the Problem
The purpose of this study was to investigate the relation of visual
selective attention to reading performance of learning disabled children.
The question developed from the extensive literature asserting that
learning disabled children are frequently characterized as exhibiting
attention deficits and because of the literature showing precision
teaching to be a technique effective for teaching reading to learning
disabled children. The major question was: Do the selective attention
processes of learning disabled children underlie or determine their
reading performance in this program?
Significance of the Study
Because of the widespread and expanding use of precision teaching,
defining the relationship between academic performance when using this
technique and selective attention processes became important. The
literature suggested selective attention as the critical characteristic
that produced learning disabilities. This became an important applied
research question because it applied directly to so many LD classrooms
and training programs. This study focused on relating selective atten-
tion research more directly to current strategies, endorsed by many LD
teachers, that were actually being used in LD classrooms. Continuing
this concern for relevance or generalizability, this study used achieve-
ment and process measures currently widely used in LD programs.
Furthermore, a proliferation of research on Cognitive Behavior
Modification (CBM) has cited the selective attention research for support.
This study has some implications for the use of CBM to compensate for
selective attention deficits because it provides achievement perfor-
mance in a precision teaching intervention for comparison, identifying
selective attention characteristics of the population. Thus new
knowledge of how selective attention influences reading performance
coupled with the other research on how it related to CBM should help
provide for effective integration of the techniques in LD classrooms.
This study was designed to answer the following questions regarding
the influence of selective attention on reading performance.
1. Did using the Systematic Instructional Management Strategies
(SIMS) program favorably alter attending behavior so that LD
students experiencing reading problems attributable to selec-
tive attention would be differentially helped compared to
another method for teaching reading? Would the reading
performance in this other reading method and in non-learning
disabled students have been influenced by selective attention?
2. Was this relationship best characterized in LD students as
due to a limited filtering or limited capacity process
(%C %I), where central performance limits incidental
performance? Or was it best characterized as due to the
total processing capacity with distraction (C + I)?
3. Without comparisons to other teaching methods, would grouping
of LD students by level of selective attention performance
show differential reading improvement?
4. Would the effect of selective attention on reading perfor-
mance, after allowing for age differences, have been best
explained by incidental performance limiting central perfor-
mance as in the efficiency measure (%C %I), or by the
total capacity with distraction measure (C + I)?
5. Did low visual selective attention scores correspond to low
visual modality aptitude? Could these results have been due
to visual modality aptitude or a process underlying it?
6. Was selective attention in the LD group more closely related
to underlying modality aptitudes or to general intellectual
7. Were the effects ascribed to selective attention possibly
due to teacher differences?
8. Was this extensively used research instrument, Hagen's
Central Incidental Attention Task, measuring the same
underlying construct, selective attention, as other
instruments and techniques? Did the incidental score
really appear to increase at the expense of the central
The research study was organized around these questions in three
sections. The first section used three small samples of subjects. They
were students using the SIMS program, students using a traditional LD
reading program, the Visual-Auditory-Kinesthetic-Tactile program (VAKT),
and a sample of students referred but not admitted to the LD program
using the regular basal reading program. This section investigated
questions one and two. The second section used a sample of LD students
in the SIMS program larger than many of the selective attention
research samples and investigated questions three through seven. The
final section used students from the first section but compared their
performance on several other tests to their initial performance on
the selective attention task in investigating question 8.
Limitations of the Study
Hagen's Central Incidental Attention Task has been restricted to
research use and, because of the nature of incidental measures, its
use was only theoretically justified and supported. Validity and reli-
ability information were not available and there was some debate about
whether it clearly measured selective attention (Douglas & Peters, 1979).
This use of learning disabled students also placed restrictions
on the generality of these findings. The generality was also bound
by the limits of the instruments used for diagnosis and the public school
LD criteria (see Appendix A).
The variability of LD students also limited generality. These
research questions were based on other research using selective atten-
tion as a general characteristic of the group of students labeled
REVIEW OF THE LITERATURE
An extensive literature asserting that learning disabled children
frequently exhibit attention deficits revealed the importance of this
research study. In addition to those studies already mentioned in
Chapter I which relate attention deficits to learning disabilities, this
chapter focused on research with Hagen's Central Incidental Attention
Task and similar instruments. Secondly, the importance of reading
achievement in the Systematic Instructional Management Strategies (SIMS)
program was established in this chapter by reporting the results of
achievement data on major precision teaching projects. Finally, current
research on the traditional learning disability concepts, modality
preference and multi-sensory teaching, was presented. The result has
been a clear rationale for the importance of investigating the interac-
tion between a concept that was called the "origin" of a learning dis-
ability (LD), selective attention, and one of the most promising tech-
niques for teaching LD students.
Selective attention refers to any method of information processing
that allows the student to respond on the basis of a limited portion
of the available stimulus information. Two types of models explained
selective attention studies. One type, the filter theories, was developed
by Broadbent's (1958) research. These theories led to the development
of Hagen's Central Incidental Attention Task. They postulated mechanisms
that screen certain kinds of input at various stages of information process-
ing. Thus attention capacity was limited and the filtering mechanism
shut out irrelevant stimuli. The other type, the capacity theories,
postulated that the system, based on goals, selected certain inputs
and ignored others because of insufficient resources within the process-
ing system. The greatest success of capacity theories was their ability
to explain contradictory divided attention studies by ascribing different
capacity.requirements to the tasks being combined.
Samuels and Edwall (1981) did a comprehensive review of research
on attention focusing on aspects such as arousal, alertness, capacity,
and selectivity. In several studies it seemed poor readers and LD students
had difficulty focusing attention on a narrow band demanded by the task,
i.e., selectivity. The authors decided that vigilance did not tap the
sustained aspect of attention corresponding to school tasks. Further-
more, they reported that such tasks did not differentiate LD students
from normal students. Instead it differentiated hyperactive students
from other students.
Keogh and Margolis (1976) also reviewed a large number of attention
studies. However, they focused their review on the following three
aspects of attention that they felt related to remediation: (1) coming
to attention, (2) decision making, and (3) maintaining attention.
They felt that such a differentiated approach to any deficit had more
direct implications for remediation.
Harris (1976) reviewed the attention research, also focusing on
educational attention control. But he felt that attention as studied
related too closely to classroom behaving and that "the mechanisms
of attention have no correlation with academic performance" (p. 55).
He reported how techniques of drug therapy, reduced environmental stimula-
tion, and operant conditioning of drug therapy, reduced environmental
stimulation, and operant conditioning effectively produced behaviors
operationally called attention. He also suggested verbal self-direction
may be a future technique for improving attention behaviors. But generally
he felt "attention training should not be expected to improve academic
skills" (p. 108).
Koppell (1979) criticized the attentional deficit explanation
of learning disabilities saying that research must examine the strength
of the association between the extent of the learning disability and
performance in the experimental paradigm. However, if the measure
of the extent of a learning disability is defined by the academic per-
formance, then this is the major goal of the research questions from
Chapter I. Thus, unlike Harris' opinion, the relation between the
extent of LD students' academic deficiencies and the students' deficits
in selective attention seemed to be the critical measure of the impor-
tance or the usefulness of the hypothesis of selective attention deficits.
Several researchers have dealt with other deficits, such as main-
taining attention, as characteristics of LD students. Anderson, Halcomb,
and Doyle (1973) used a visual vigilance task successfully in signifi-
cantly differentiating between LD students and normal students. A
higher crror rate and longer response latency -reflected vigilance atten-
tion deficits. In a further study, Doyle, Anderson, and Halcomb (1976)
used this task with a visual distractor. While the LD subjects showed
evidence of an attentional deficit, hyperactive LD students were more
severe and much of the difference can be attributed to them. Swanson
(1980) also found older nomoactive (CA = 12) LD children made signi-
ficantly fewer correct detections and more false responses than did
normal children on a vigilance task.
Ross (1976), in an extensive review of the literature, reported
how studies of reading or learning disabled students led to his con-
clusion that "learning disabilities may thus be viewed as the result
of delayed development in the capacity to employ and sustain selective
attention" (p. Gl). He showed that studies using measures like motor
responses, heart-rate changes, changes in brain electrical potential,
and auditory messages repeatedly show that selective attention capacity
improved with age. He offered a theoretical model for selective atten-
tion development. The first stage was overexclusive (also overselec-
tive) attention. Aspects of a stimulus capture attention to the ex-
clusion of other aspects in this sage which normally extended from
infancy to preschool. The next stage was overinclusive attention.
This corresponded to a period of maximum incidental learning, cover-
ing preschool and elementary schooling. Finally, the student developed
selective attention, which influenced the decline in incidental learning
seen around age 12.
Ross' (1976) explanation for learning disabilities suggested some
possible problems in investigating the research questions from Chapter
I. The LD selective attention problem was not permanent, but a delay
causing the student to be in earlier, inappropriate stages during his
reading education. Thus the academic deficit developed, continued,
and hindered future learning, even after the selective attention capacity
matured. Thus maturation over time weakened the relationships in research
questions one to four. An analysis by age groups (hypothesis three)
Ross (1976) also suggested that some results reporting a preference
of auditory digits over visual digits may be a result of negative reactions
to the visual stimuli from repeated reading failure. He thus suggested
that such a modality based reaction may show a consequence rather than
a cause of the students' reactions. He also implied that the modality
dimension of stimuli may become less dominating of attention with
maturation. This suggested that a narrowing or broadening of attention
exaggerates, then reduces, LD students' modality preferences. This
will be elaborated on later in this chapter.
Dykman, Ackerman, Clements, and Peters (1971) also offered a com-
prehensive review of research as evidence of a specific learning dis-
ability syndrome with the cardinal symptom of defective attention. They
suggested that this was due to the interaction of the excitation from
the brain-stem reticular formation and the forebrain inhibitory system,
centering on the diencephalon. They used designations similar to Ross',
but only hyperactive LD children were considered to be overattentive
and hypoactive LD students were underattentive.
Schworm (1979, 1982) also reviewed the literature on LD attentional
deficits, but with an emphasis on reading. In one study (Schworm, 1979)
the treatment consisted of cues to get the student to direct attention
to distinctive features and invariant word properties. He suggested
that inadequate readers failed to develop the grapheme-phoneme corre-
spondences because they did not attend selectively to important stimulus
Schworm (1982) recently reported an impressive investigation very
similar to the goals of this research study. He administered a test
of selective attention, the Select-A-Figure-From-Many (SAFFM) test.
He used the Spelling Pattern Tests which he had developed (Schworm,
1979). Finally, he used the Wide Range Achievement Test (Jastak & Jastak,
1978), which was also used in this study, as outlined in Chapter III.
Comparisons between the LD readers and the low, average, and high
achieving readers from regular classrooms showed the LD readers required
significantly more trials on both parts of the selective attention task.
The letter pattern test results of the regular classroom readers corre-
lated r = .72 with their SAFFM results. Significant correlations showed
that both better achieving and older students completed both tests better
than other students.
There were also significant differences between the LD students
and other students on the two letter pattern and three letter pattern
of the Spelling Pattern Test (Schworm, 1982). Finally, the debriefing
of the selective attention task was analyzed for number of cues reported,
types of cues reported, and degree of interaction recorded (help from
experimenter), using multiple regression analysis. They were good pre-
dictors of reading achievement (R2 = .69), two letter spelling pattern
identification (R = .79), and three letter spelling pattern identifi-
cation (R = .69), as well as performance on both parts of the selective
attention test Al (R2 = .70) and B1 (R2 = .62). The present research
study also investigated the relation of selective attention to reading,
but used Hagen's Central Incidental Attention Task.
The Central Incidental Task
Tarver and Hallahan (1974) performed a literature review of 21
experimental studies. They noted how the hypoactive and hyperactive
LD students differed in some studies. They also noted several ways
that subjects with learning disabilities were different from other subjects.
LD students were deficient in their ability to maintain attention for
prolonged periods (vigilance). LD subjects were more impulsive, or
less reflective. LD students' hyperactivity could be situational-specific.
Finally, LD subjects exhibited more distractibility on tasks with embedded
contexts, or tests of incidental versus central learning. Their own
work with Hagen's Central Incidental Attention Task was a major part
of that research.
Maccoby and Hagen (1965) used Broadbent's ideas (1958) about
"focusing of attention" in studying individual responses to information
in the stimulus complex that exceeds their processing limit. Theoret-
ically the young child was handicapped in focusing attention selectively,
lacking previously established discrimination between task-relevant
and task-irrelevant aspects of the stimulus complex. They used a task
arrangement with background color matching as the central task and
picture matching as the incidental task. They also had a "distractor"
or subsidiary task. The subjects had to tap whenever a single bass note
played on a tape of a piano melody of high notes. The results showed
that central recall increases regularly with age. Distraction reduced
efficiency on the central task as errors rose by 26%. However, in-
cidental recall declined with age, and errors under distraction in-
creased by 15%.
Hagen and Kail (1975) reviewed a series of studies involving
central and incidental stimuli in the task, both auditory and visual.
The major pattern was replicated, with central recall increasing with
age while incidental recall did not change significantly until it
declined in older subjects. Though different modifications were tried,
no differences in central task performance occurred due to pairing
conditions. Apparently, just the mere presence of incidental pictures
was distracting, regardless of how they were modified. They suggested
that older students used verbal labeling and other task-relevant
strategies which younger students lacked. Several studies using differ-
ent tasks supported this trend with incidental learning declining
between ages 12 and 14 years. They also cited studies with retardates
that showed that selective attention improved as the MA label increased.
In Section Two, Chapter IV, of this dissertation, selective attention
was compared to increasing IQ scores in LD students.
Using a questionnaire with this attention task, Druker and Hagen
(1969) found that with increasing age there was a progressive rise in
the tendency to rehearse by saying only the task-relevant items. More
of the older subjects indicated they tried to look at only the relevant
item on the card. Another finding of this study with 80 normal subjects
was that developmental changes monitored by the task did not involve
improved visual discrimination.
In another study, Hagen (1967) used a vigilance distractor with
the Central Incidental Attention Task. Distraction significantly affect-
ed task-relevant performance and did not affect incidental performance
except at the oldest grade. He also had a control group using cards
with only the central stimuli. The results indicated that the presence
of a second picture had a deleterious effect at all age levels.
In a similar study, Hagen, Meacham, and Mesibov (1970) found that
with college students (N = 40) there was no deleterious effect shown
between one versus two stimuli. However, a verbal rehearsal condition,
repeating the name of the animal, had a detrimental effect on the college
students, primarily because of a primacy decrement. The verbal rehearsal
condition had a favorable recency effect. This verbal labeling did
not affect children (N = 96) ages 9 to 14. Hagen also reported an earlier
study which found that verbal rehearsal improved performance of children
from ages 6 to 8. A follow-up study (Hagen, Hargrave, & Ross, 1973)
used children of ages 5 to 8 to induce the verbal rehearsing found with
older children. However, in comparing a rehearsal alone to a rehearsal
with prompting, recall only improved at the younger ages, and only with
Wagner (1974) reported a study in Yucatan, Mexico, using Hagen's
Central Incidental Attention Task. They found the same developmental
patterns as other studies in an urban sample, but a different pattern
in a rural sample. The rural sample did not show a marked primacy effect.
An additional study found that unschooled urban adults were like the
rural sample and significantly different from schooled urban adults.
The results suggested that formal education is a major factor in the
ability to use verbal rehearsal or other attention and memory strategies
as measured in this attention task.
In his review of studies of faulty attention in children, Hallahan
(1975) dealt specifically with selective attention, attention span,
hyperactivity, impulsivity, and with distractibility in general. The
same central incidental task that Hagen developed was used in most of
these studies. The ability of normal children to attend to central
information and ignore incidental material increased with age until a
major drop in incidental recall at twelve to thirteen years of age.
With the groups equated by mental age, educable children performed below
normals while those not institutionalized performed similarly to normals.
Another study with educable retarded, spastic, cerebral palsied and
normal children showed no difference between groups on this task when
equated on mental age. Hallahan (1975) also cited a study where insti-
tutionalized retarded children did not differ from a normal group of
children on mental age.
Further studies (Hallahan, 1975) using Hagen's Central Incidental
Attention Task showed that low-achieving sixth grade boys performed
poorly compared to high-achieving boys. Also poor performance corres-
ponded with impulsive performance on Kagan's Matching Familiar Figures
Task. Hallahan (1975) reported that in several studies learning dis-
abled (LD) children had deficient selective attention performance, using
the central incidental task. Other studies using LD children or under-
achieving children and proximal distractors, such as the Stroop Color-
Word Test, the Portable Rod and Frame Test or other embedded designs,
showed less successful attending than normal children when facing dis-
tracting information. Hallahan (1975) suggested that distal distraction
conditions had not found the same negative effect as proximal distraction.
A test very similar to Hagen's Central Incidental Attention Task was
the Fruit-Distraction Test (Santostefano, 1964). It also investigated
proximal distraction and was used to study LD students. The test was
used for studying cognitive style; however, it related to selective
attention research when it dealt with the inability to effectively and
actively select, organize, assimilate and process information in the
context of distracting stimuli or information competing for attention.
Santostefano (1964) used this test for studying constricted-flexible
cognitive style. He found a "brain-damaged" group from a residential-
educational center did more poorly than a regular school group; however,
their IQ score means were 72 and 110 respectively. Santostefano and
Paley (1964) also found non-significant evidence of a developmental
trend from constricted to flexible, i.e., less difficulty in selectively
devoting attention to the central stimulus with age.
Santostefano, Rutledge, and Randall (1965) used 24 boys with a
reading disability and 23 matched control subjects. They found that
poor readers recalled significantly more background figures in the Fruit-
Distraction Test II and took significantly longer to read the distraction
card. Cotugno (1981) replicated these findings with 17 subjects having
reading disabilities and 17 control subjects. He concluded that there
were significant differences between disabled and non-disabled readers
on field articulation tasks requiring attention to relevant information
in the presence of irrelevant or peripheral and contradictory information.
These results are consistent with other studies of proximal distractors,
particularly Hagen's Central Incidental Attention Task.
Hallahan (1975) cited evidence for an adverse effect on LD student
performance from a linguistic distractor condition. With a simultaneous
presentation of auditory and visual material, older LD students seemed
to be deficient relative to normal in recalling the stimuli in pairs
while younger LD students did more poorly than normal students in re-
calling the stimuli in one modality and then the other. Hallahan (1975)
suggested that auditory stimuli are more distracting than visual ones.
Using other measures of attention, Hallahan (1975) cited studies
suggesting a negative relationship between attention skills and hyper-
activity. His own research related Hagen's Central Incidental Attention
Task to behavioral observation measures of attention and hyperactivity
in institutionalized learning disabled children. He stated the following:
There was no relationship between the laboratory
measure of selective attention--Hagen's C-I
task--and the behavioral measures of attention
and hyperactivity. Although at first glance it
perhaps seems appropriate to question the
utility of Hagen's task in predicting or explain-
ing relevant behaviors of learning disabled
children, the task has differentiated learning
disabled from normal children. (Hallahan, 1975,
This was an important consideration for evaluating proposed applications
of research built upon this central incidental task.
Tarver, Hallahan, Kauffman, and Ball (1976) studied selective atten-
tion, using Hagen's Central Incidental Attention Task, in younger and
older learning disabled boys. The study supported the idea that learning
disabled students have a deficit in selective attention which improves
with age. This suggested a developmental lag. They used 18 learning
disabled (LD) students approximately 8 years old. They found that the
central recall was greater for normal students, primarily due to the
lack of a primacy (positions 1 and 2) effect. This was also cited as
evidence for a verbal rehearsal deficit. It has been shown that verbal
rehearsal strategies underlay primacy recall.
Tarver et al. (1976) did a second study which had two matched groups
of older LD students. The average age was about 13 years for two groups
and about 10 years for the other two. This study dealt more directly
with the verbal rehearsal assumptions. While one matched group received
the normal Central Incidental Attention Task, the other group received
directions to verbally label, chunk, and rehearse the items of the task.
In these older students a primacy and recency effect occurred in all
groups, thus negating the influence of the special instructions of this
theoretical measure of verbal rehearsal. However, a light trend to
increased central measures and decreased incidental measures combined
in the efficiency calculation (%C %I) and showed rehearsal to have
a significant favorable effect.
The results supported the idea of constant developmental increases
in central recall with increasing age in LD students. This idea was
later developed fully into a developmental lag hypothesis. Reference
to the model of selective attention development in normals was used to
suggest that a developmental lag of 2 years was evident. The authors
further suggested that the absence of a significant decline in inciden-
tal learning at any of the age levels investigated suggested that develop-
ing verbal rehearsal strategies caused increases in central recall with
age rather than improvements in selective attention. Yet the theoretical
distinction between verbal encoding and selective attention was not
well defined. Furthermore, the relation of the verbal rehearsal treat-
ment to the efficiency measure (%C %I) was not clearly defined and the
results supporting this relation seemed tenuous in theory, magnitude,
and the number of subjects involved.
Swanson (1979) used 15 LD students and 15 normal students on Hagen's
Central Incidental Attention Task. He reported that when matched on
mental age (MA), there were no significant differences between groups
in selective attention scores.
Vrana and Pihl (1980) studied selective attention using both proximal
and distal placements of incidental stimuli. However, there were two
obvious, major faults in the study. The control group was significantly
different from the learning disabled group on intelligence. Also, after
making an incidental measure on the fifth card, on the sixth card an
effort was made to make another incidental measure. After students
were questioned once about an incidental stimulus, they probably did
not respond to the same stimulus as if it was an incidental part of
Vrana and Pihl's (1980) experimental procedure was very similar
to Hagen's task (the proximal condition). It provided insight on using
the technique. They used solid black circles as the central recall
item on large white cards with solid black squares as the incidental
stimuli. The stimuli were either close to each other in the center
of the card or apart from each other at each end of the card. On the
second, third, fourth and sixth cards questions about the circles were
used to calculate central recall. On the fifth and sixth cards, questions
about the squares were used for the incidental recall, along with a
question about the rows on card six.
Besides the problems already mentioned, the results were comparable
with other research, with the advantage in this study that proximal and
distal aspects were compared while minimizing other stimuli differences.
Thus the results, that learning disabled children had significantly
lower recall of central stimuli than normal children when both stimuli
were presented together proximall) but not in distal or incidental con-
ditions, clearly agreed with the suggestions of Hallahan (1975) about
proximal vs. distal distractors.
Tarver, Hallahan, Cohen, and Kauffman (1977) analyzed the previous
research on verbal rehearsal and selective attention and developed a
model for the developmental lag in learning disabled students based
on studies of normal students and 8, 10, and 13-year-old learning disabled
students. They then extended this research with a study of 15-year-
old learning disabled boys (N = 14) to demonstrate the overall pattern.
Their reliance on previous research such as Hagen's studies led
them to omit any control group. Thus the assumption of a direct com-
parison between the trends reported for normal children and this par-
ticular experimental group seemed to run some risk of an unrecognized
systematic difference being the real causal element. Use of the research
literature, primarily their own studies with small sample sizes, did
not seem justified, particularly with a task having limited psychometric
study. Consequently, in terms of experimental design, sample character-
istics and test administration in this study (Tarver, Hallahan, Cohen,
& Kauffman, 1977) the absence of a control group seemed unjustified.
The results of this study (Tarver, Halldhan, Cohen, & Kauffman,
1977) were compatible with the developmental lag hypothesis. There
was clear evidence of increasing selective attention at the older ages.
Central recall did not differ from the 15 and 13-year-olds. But a
steeper primacy effect for the 15-year-olds was used as evidence for
continuing development of verbal rehearsal strategies. Reciprocity
between central and incidental measures again was refuted.
A study by Hallahan, Gajar, Cohen, and Tarver (1978) attempted
to explore aspects underlying selective attention differences. Twenty-
eight LD junior high students completed two locus of control measures
and a group selective attention measure. This measure was a modification
of Hagen's Central Incidental Attention Task. The results showed that
the LD students scored significantly lower than normals on selective
attention central recall. They differed significantly from normals
on both locus of control measures, showing a greater degree of external
control. One surprising finding was that the two locus of control
measures did not correlate highly, leading the authors to speculate
that the LD student's external locus of control is extremely broad.
The general conclusion supported the idea of a quantitative developmental
lag by showing that the LD student is inefficient in using problem-
Hallahan, Tarver, Kauffman, and Graybeal (1978) began the effort
to relate the research on selective attention and the developmental
lag more directly to the teaching techniques mentioned in several pre-
vious articles for making learning disabled children more actively in-
volved in using problem solving strategies. First they developed a
new version of Hagen's task that would allow repeated tests. Three
different classes of stimuli were used, i.e., animals, household objects,
and geometric figures.
Using this instrument they investigated the effects of reinforcement
and response cost. In the reinforcement condition (R) the subject
received 3 for every correct first card and 1% for the second card.
A response cost condition (RC) caused a loss of 3t for every incorrect
response from 60t initially given. The control condition (C) allowed
2t for responses to both cards.
The reinforcement condition resulted in both a recency and a primacy
effect, while the others showed only a recency effect. A measure of
the child's efficiency in focusing on high payoff (first probe) versus
low payoff (second probe) was presented as analogous to selective atten-
tion. Thus the significantly better performance under reinforcement
for recall of first-probed stimuli (7.56) than second-probed stimuli
(3.81) was explained as showing the child giving up attention to the
low payoff stimuli. However this theoretical relationship needed to
be examined more closely.
The same relation between a primacy effect and verbal rehearsal
suggested that LD students in the reinforcement condition were using
a verbal strategy to attend better. This suggested to the authors that
the LD child is able to use an appropriate strategy, and usually has
difficulty because, even knowing mediators, he is unable to produce
them at the appropriate time. The study by Hallahan, Tarver, Kauffman,
and Graybeal (1978) clearly agreed with the earlier research findings
and their implications for programs. The following by Tarver, Hallahan,
Kauffman, and Ball (1976) was still supported:
A gradual, though slower rate of progression through
the normal developmental sequence is indicative of a
quantitative learning deficit, and suggests that those
procedures which have been found to increase rate of
learning in all populations, e.g., repetition and
reinforcement, should be emphasized in the education
of the learning disabled. In contrast, confirmation
of a more basic, qualitatively different mode of
cognitive functioning would imply that teaching
methods should be qualitatively different for the
learning disabled and normal children. (p. 383)
Torgesen (1981) summarized much of his research on attention and
memory by saying that LD students have a less active, organized approach
to memory tasks. These were defined as control process inefficiencies.
This was compatible with the idea of a developmental lag producing a
quantitative problem. Torgesen even suggested intervention strategies
for learning to compensate for control process inefficiencies. He men-
tioned the use of incentives or reinforcement programs, the use of direct
instruction in processing strategies, and the use of orienting tasks,
such as those that require the repetitive manipulation of material.
There have also been studies using relaxation training or desensitiza-
tion training to improve attending, such as the study by Omizo and Michael
(1982) with hyperactive boys.
Argulewicz (1982) also used a direct intervention. Subjects were
trained in attending behaviors. Compared to a group only exposed to
modeling, this group did significantly better on the indexes of selective
attention used in the study. The indexes chosen were the digit span
subtest of the Wechsier Intelligence Scale for Children--Revised (WISC-R) and
the Memory for Sentences II (MSII) subtest of the Stanford-Binet In-
telligence Scale. In Section Three of Chapter IV these measures were
compared to Hagen's Central Incidental Attention Task.
Educational procedures falling under the rubric of cognitive
behavior modification have been suggested for the strategy-deficient
inactivity of learning disabled students. In one example, an LD student
was taught to monitor his on and off-task behavior, using a cue from
a tape recorder to pace his monitoring. On-task behavior in math and
handwriting was dramatically increased. Eventually use of the tape
recorder, and then self recording were gradually phased out (Hallahan,
Lloyd, Kosiewicz, Kauffman, & Graves, 1979).
In another study Hallahan, Marshall, and Lloyd (1981) used self-
recording to increase on-task behavior of three severely LD students
in one reading group. Observer ratings of the on-task behavior showed
a doubling due to self-recording. They also gradually withdrew wrist
counters and then the tape recorder. Lloyd, Hallahan, Kosiewicz, and
Kneedler (1982) did a comparison of self-assessment and self-recording.
Self-assessment consisted of the student asking himself whether he was
paying attention when he heard the tape-recorded cue. Using three sub-
jects in the second part of the study, self-recording was found to be
superior to self-assessment, but neither increased productivity.
A final approach to investigating selective attention dealt with
the student's awareness of his attending, i.e., meta-attention. Loper,
Hallahan, and lanna (1982) developed an instrument to measure meta-
attention in LD students. Performance on the meta-attention task corre-
sponded to academic achievement for normal subjects, but not for the
LD students. They also noted a developmental trend from younger children
being more impressed with rewards to older children placing more value
on interest when asked the circumstances leading to better attention.
Tarver (1981) attempted to integrate this research using the term
underselective to refer to LD attention behaviors. Her article described
the development of explanations based on the results of studies of under-
selective attention. Early explanations of distractibility were refined
to differentiate proximal distractors for measuring selective attention.
Next analysis of the patterns of central recall scores by serial
position led to identification of a primacy effect deficit. This was
interpreted as a verbal rehearsal deficit. Evidence of a failure to
use effective verbal rehearsal strategies, even when known by the subject,
also suggested problems with metacognition. Other research has studied
how linguistic subskills of attention and memory contribute to the LD
information-processing deficit. She also outlined the creative ability
hypothesis. Thus the LD child's underfocused, underselective attention
might have been indicative of high creativity,
Investigations of selective attention have led to a wide range
of related investigations. The most important concern was the distinc-
tive academic deficit that LD students experience. The merit of these
other constructs will ultimately be the degree to which they explained
and helped to remediate the academic problems. The effectiveness of
precision teaching in interacting with the selective attention deficits
to produce reading achievement was an example of these critical basic
The behavioral approach to attention described it as an unobservable,
unmeasurable phenomena, and attending or its actual verification through
observable responses is the relevant concern (Haring, 1968). Thus the
question of attention centered on the teacher's efforts in guiding attend-
ing and responding to relevant dimensions of stimuli. Attending was
a learning behavior. Haring (1968) further asserted that "faulty attend-
ing and inaccurate, slow responding can be corrected through more precision
in the manipulation of instiuctional conditions available to the class-
room" (p. 45). Precision teaching used most of the features of applied
behavior analysis recommended by Haring. Thus it should be particularly
helpful in guiding attending and responding to relevant dimensions
If a history of ineffective patterns of attending was also reflected
in a minute sample of behavior, Hagen's test, then it could be used to
test whether academic performance corresponds to student's attending
patterns even when teaching with a program guiding attending. Precision
teaching should be particularly effective with those students with poor
selective attention performance. One assumption, which is not supported
by a behavioral view, was that poor selective attention performance
corresponded to a history of academically ineffective patterns of attending.
Mercer and Mercer (1981) described applied behavior analysis (ABA)
as one of the newer approaches in special education. They outlined
its features as including (a) measurement systems, (b) precision
teaching, (c) instructional aims, and (d) learning principles.
Thus, precision teaching is characterized by a direct,
continuous, and precise measurement system, and it
provides an accurate measurement of student progress.
The teacher can obtain a record of past performances,
a plan for where the student is going, and an estimate
of when the student will get there. (Mercer & Mercer,
1981, p. 6)
Precision teaching is "one way to plan, use, and analyze any teaching
style, technique, method, or theoretical position--old or new" (Kunzelman,
Cohen, Hutten, Martin, & Mingo, 1980, p. 12).
Bradfield (1971) outlined four components to precision teaching.
The first was the system of recording and charting data, using a chart
such as the six-cycle logarithmic chart developed by Lindsley. The
second component was the need for precise selection and definition of
the target behavior, called "pinpointing." Bradfield (1971) emphasized
two of Lindsley's basic characteristics for "pinpointed behaviors" or
"movements": (1) They must be a complete movement cycle, having a
definable beginning and end; (2) They must pass the "dead man's test,"
i.e., if a dead man can do the behavior, it can not be counted. The
third component Bradfield (1971) used is Lindsley's IS-DOES formula.
This referred to the five basic parts of the learning environment.
In the IS formula, the five parts were parts that only have the potential
to change behavior. They are as follows:
1. Program, environmental setting such as location, time
of day, etc.
2. Antecedent event, factors such as instructions, demon-
strations, materials which might result in the behavior.
3. Movement cycle, the behavior being measured.
4. Arrangement, the numerical ratio between the movement
and the subsequent event.
5. Subsequent events, events which may result from the
movement cycle such as praise, grades, withdrawal of
This DOES part of the formula is reserved for when these have demon-
strated a behavioral function and become the following: (a) disposition
components, (b) stimuli, (c) responses, (d) contingencies, and
(e) consequences. The final, fourth component outlined by Bradfield
is the Behavior Bank. It used a computer to store cases as examples
of procedures for particular behavior problems.
Mercer (1979) presented a very concise description of the teacher's
activities in precision teaching. They include the following:
(a) selecting a pinpoint or target behavior; (b) developing a task
sheet or probe for evaluating pupil progress in daily timings;
(c) graphing these data daily, setting instructional aims, and par-
ticularly teaching; and (d) analyzing the data and making instructional
The critical elements were the measurement, the rate of the behavior,
and the charting of this measurement. This usually meant that frequency,
the count of behaviors divided by the number of minutes, was recorded
on the Standard Behavior Chart. The chart's sensitivity ranged from
1 movement per 1440 minutes (24 hours) to 1000 movements per minute
(Koorland and Martin, 1975). The widespread use of this chart promised
to improve communication between researchers and instructors.
The following were some terms used in precision teaching:
Accuracy Measure--The proportion of correct behaviors to inappropriate
Frequency--How often something happens in a certain unit of time.
Celeration--The measure for summarizing the trend and direction of
day-to-day frequencies on a chart.
Acceleration--A daily increase in the celebration line.
Deceleration --A daily frequency value decrease causing the celebration
line to move down the chart.
Aims --The specification of precise pupil response, the conditions
surrounding it, and the criteria for acceptable performance.
Proficiency --The level of student performance that is automatic enough
to show a desired frequency (rate) representing the performance
of successful performers, reflecting mastery.
The Precision Teaching Project of Great Falls, Montana, was one
of the most successful projects in the country (Beck, Note 3). Six
schools were used in the study, two with 5% of their families' incomes
under $5,000, two with 12%, and two with 20%. One member of each pair
was randomly assigned to the experimental group. The study involved
students in kindergarten and grades one to three who had "learning
deficits." The degree to which this category overlapped with learning
disabilities is not known. The screening for this group involved per-
forming at one-half or less of the median celebration of classroom peers.
Though screened in exactly the same manner, learning deficit students
in the control schools were not identified until the project was com-
pleted. In the experimental schools the remediation consisted of only
20 to 30 minutes per day, utilizing minute drill sheets focusing on
basic tool skills.
The results included the three grades in the three types of schools
with three behaviors: write numbers, hear-write numbers, and see-say
letters. Nineteen of 27 pretest comparisons showed no significant difference
or a significant difference favoring the control groups. Of these 19
comparisons, the posttest results for 14 showed the experimental groups
to be significantly superior. Three comparisons showed no differences.
Only one, the see-write numbers task in grade two of the second type
schools (12%), had no significant difference on the pretest and sig-
nificant difference favoring the experimental group on the posttest.
Analysis of eight cases, five being see-say letters tasks, was not yet
available because the experimental groups were superior on the pretest.
Finally, in rating the educational importance of these studies, the
report noted the following facts (Beck, Note 3):
1. The magnitude of the effects of precision teaching were
2. The procedure was most effective in grade 1.
3. The procedure was not sensitive to economic status.
4. Its continued use was inexpensive.
Next a precision teaching approach to regular education was developed
(Beck, Note 4). Precision teaching was implemented at Sacajawea Elemen-
tary School. This implementation actually focused on the following
1. Screening and Identification
2. Direct and Daily Measurement
4. Data Decisions and Instructional Intervention
5. Support (Materials Bank)
Another school was designated initially as having equivalent or relevant
variables of socio-economic status and achievement test patterns. Over
a previous four year period, there was no difference in the schools.
The Sacajawea first grade precision teaching pupils had significantly
higher reading scores (an eight month gain) on the Wide Range Achievement
Test (WRAT) reading subtest. There developed a large percentile difference
between the schools, favoring the precision teaching subjects. The
Iowa Tests of Basic Skills was used to compare fourth grade students
for four years. During the year with no treatment and the first year
of precision teaching there was little difference between Sacajawea
and the district. During the third and fourth years (1976-77) there
was a dramatic improvement over the district of +24 percentile points
in reading and +32 percentile points in math.
The SIMS Program
The Systematic Instructional Management Strategies (SIMS) Program
was originally designed to provide a comprehensive self-contained edu-
cational program for severely learning disabled students. The Armatage
SIMS Center was established in 1972. Based on its success, the program
was extended to include resource rooms which serve the mild to moderately
learning disabled students (Harvey, Note 5; Wiseman, Note 6).
There were two aspects to this program, the SIMS materials and
the SIMS procedures. The SIMS materials included a highly structured
phonics sequence which encompassed the basic coding skills of reading
and spelling. Coding skills were divided into 53 categories. Each
category had two category word lists, sentence lists, and four stories.
The procedures advocated that in each category a 90-100% performance
criterion be used for accuracy. Proficiency was monitored daily on
word lists with a criteria of 40-60 words correct per minute with two
or less errors for mastery. Daily rates for the stories suggested 80-
100 words correct per minute with two or fewer errors. In this way
the SIMS program combined task analysis, criterion referenced teaching
and testing, contingency management, and precision teaching (Harvey,
note 5; Wiseman, Note 6).
Though the SIMS program used precision teaching and individual
charts, group comparisons were used to assess the program's effective-
ness in the 1976-77 year. Students made statistically significant
gains. The SIMS group was also compared to the students in the Minne-
apolis Special Education Division school based resource program. The
SIMS students were originally chosen from these same resource programs
based upon their failure to make "acceptable gains." The results
showed a 1.05 average gain, with 0.77 being the gain bythe elementary
students in the year prior to being in the SIMS program, and with .84
being the gain of the comparison group. The report also presented
charts of the average student's time in each category and the weekly
celebration rate for each category (Harvey, Note 5).
When the validation report was submitted to HEW's Joint Dissemina-
tion and Review Board (JDRP), WRAT reading standard scores were compared
between the SIMS group (N = 118) and the control group (N = 147) yielding
a significant difference (p<.001). In the evaluation for the 1977-78
year, t-tests between SIMS students' pre-posttest gains were greater
than might be expected by chance (p <.001). A second comparison, prac-
ticed by the United States Office of Education, compared the average
gain to one-third the normed standard deviation of the test, "suggesting
the gain is educationally significant" (Wiseman, Note 6, p. 27).
Thus there was evidence supporting the SIMS as a successful pre-
cision teaching program, though some of the analysis seems weak or in-
appropriate. The use of gains scores has well-known inherent weaknesses.
The extension of this into a predicted gain score using the following
formula seemed unjustified:
Years x Time
Score = years in school at time of entry
Time = time elapsed between pre and posttests
Furthermore, there were reasons to expect that all LD students would
make some improvements in reading in this program. But student per-
formance while in this program could be used to represent reading
achievement in studying the relationship of student's visual selective
attention levels to reading performance.
Selective attention referred to a method of information processing
while modality processing also referred to the way information was pro-
cessed. Thus several research questions asked how selective attention
scores corresponded to modality aptitude scores. They were attempting
to clarify whether one process underlay the other, whether they were
separate processes, or whether they were both expressions or symptoms
of other underlying processes.
This question was very important because the concern for psycho-
linguistic or modality processes was one of the earliest and most dis-
tinctive characteristics of research, identification, and remediation
of learning disabilities. Most definitions of learning disabilities,
except for the behavioral definitions like those used with precision
teaching, still defined it as due to underlying psychological processes.
Psychological testing focusing on cognitive, perceptual, and expressive
abilities of these students had been used to identify, diagnose, and
to suggest special instructional strategies and materials. Modality
preference was one of the more global aspects of these psychological
The modality model referred to LD students as primarily auditory
or visual learners according to relative strengths and weaknesses in
their auditory and visual channels. Matching the modality preference
to particular teaching strategies had been studied as possible aptitude -
treatment interactions (ATI).
Arter and Jenkins (1977) found that with LD teachers in Illinois,
97% were familiar with the modality model, 95% believed research supported
it, 99% felt modality should be a major consideration in educational
prescriptions, and 78% reported using it frequently. Of 700 questionnaires
sent out, this analysis was based on 340 (48.57%). Olson, Mercer, and
Paulson (1981) suggested that testing for a process disorder is con-
tinuing, even when such a determination was not required in the criteria
for identifying a learning disability.
Two testing instruments that were widely used to measure such
abilities in LD children were the Detroit Test of Learning Aptitude
(DTLA) and the Illinois Test of Psycholinguistic Ability (ITPA). Numerous
research studies, often using the ITPA, have failed to establish the
utility of modality preference for instructional strategies. Very little
research has investigated if the DTLA measures psychological processes
related to academic performance.
The study by Sandstedt (1964) used the DTLA with 45 LD students.
She found students were more successful with the visual test of unrelated
objects than with the auditory test of unrelated words. This was a
significant difference. Olson, Mercer, and Paulson (1981) used the
DTLA with 65 adolescents. In correlations with the WRAT and DTLA, the
highest were oral directions correlating .35 with math, verbal opposites
correlating .33 with reading, and auditory attention span for objects
correlating .27 with reading. It was also interesting that visual atten-
tion span for objects correlated -.03 with reading. They concluded
that the DTLA measures are not related to academic achievement.
Braggio, Braggio, Varner, Smathers, and Lanier (1980) studied the
relation of optimal response modes in 34 LD students to methods of task
presentation. The students' optimal response mode was determined by
administering 10 subtests of the DTLA. Items missed from five selected
subtests were then randomly assigned to several methods of task presen-
tation. One possible criticism was that the five subtests were all
heavily visual; however this may have been necessary for several options
in task presentation. In the first condition, the cover response (CR)
condition, the student could make no visual or auditory response. In
the vocal response (VR) condition the student told the examiner what
was required for a correct response. On the last, the visual-manual
response condition (V-MR), the student demonstrated the response by
moving a finger. Then a DTLA posttest allowed a determination of which
method produced the greatest decrease in errors and was thus the optimum
response mode. Then a paired associate task was administered using
a method of task presentation that resembled the optimal response mode.
In the unmatched condition, the presentation was in the least effective
mode. Using a one-way ANOVA, a significant main effect for CR, VR,
and V-MR treatments was found. On the paired associate task, signif-
icantly more paired-associate items were recalled under the matched
condition. They explained their results by saying that "LD children
may have difficulty filtering the demands of the task from irrelevant
cues . and then selecting the optimal response mode" (p. 94).
Lilly and Kelleher (1973) found a significant interaction using
educationally handicapped subjects and simply one modality memory tasks.
The tasks were very similar to the auditory DTLA attention span for
unrelated words and the visual attention span for objects except that
in the visual test words were used instead of pictures. Stories from
Level 2 of the Durrell Analysis of Reading Difficulty were presented
on tape or print. Students scoring highest on the visual test performed
best on the print while students scoring highest on the auditory test
performed best on the tape.
Bursuk (1971) classified 90 readers by their sensory modality learning
preference. She used aural-visual teaching and only visual approaches.
There was not a simple interaction since all subjects in the aural-
visual approach performed significantly better than the other group.
However the aural-visual approach appeared to be more effective with
the auditory learners while the visual approach was more effective with
visual learners. However, this was not clear evidence for a modality-
treatment interaction since the aural-visual treatment involved both
modalities and possibly required more involvement or attention.
Using normal students, Robinson (1972) found only auditory dis-
crimination contributed to reading achievement for both sight and phonics
programs over three years. Ringler and Smith (1973) used modality specific
treatments, though all subjects received pictoral materials and oral
discussion, resulting in no significant differences among groups.
Vandever and Neville (1974) found no differences between groups taught
to modality strength or to weakness. Smith (1971) used culturally dis-
advantaged students and found no significant differences. Foster, Reese,
Schmidt, and Ohrtman (1976) also found significant differences favoring
audible learners when taught by auditory methods, but there was no
significant disordinal interaction. Newcomer and Goodman (1975) failed
to find an ATI interaction with 167 fourth grade subjects. They did
find that the modality variable was most important for low auditory
learners. The visual treatment mode was superior for both modality
preferences. In an attempt to relate auditory or visual psycholinguistic
skills to phonic decoding (more auditory) or sight-word reading,
Richardson, DiBenedetto, Christ, and Press (1980) concluded that subjects
can not be "successfully sorted into auditory and visual learner categories"
Several people have reviewed large numbers of studies looking for
a modality and treatment interaction. Arter and Jenkins (1977) found
only one supporting study out of 15. Haring and Bateman (1977) listed
the results of the first review and mentioned three additional studies
that related reading instruction to traits other than relative modality
patterns. They reported one significant modality interaction in addition
to the Bursuk (1971) study. Silverston and Deichmann (1975) discussed
many of the same studies. They pointed out the lack of investigation
into relationships of tactile and kinesthetic modalities to reading as
well as the need for more investigations of perceptual shifting and
reading. Ysseldyke (1973) reviewed many of these same studies and
pointed out some of the problems in looking for ATI relations. He also
said many problems were due to the lack of ITPA subtest reliability.
Tarver and Dawson (1978) reviewed 15 studies dealing with the inter-
action of perceptional modality preference and method of teaching reading,
including 11 studies covered in the other reviews. The results showed no
interactions between modality preference and the method of teaching reading.
Larrivee (1981) also reviewed studies for the effect of modality
preference on commercial reading programs in schools, specific reading
skills using reading related treatment tasks, and comparisons of auditory
and visual modes as mediational channels. She emphasized that in most
studies it was difficult to identify preferred modality. Many studies
used different measurement instruments. Often classification was seem-
ingly arbitrary and only a small percentage of the subjects showed
a marked preference. Most measurement devices lacked the necessary
reliability. She concluded that differentiating instruction by modality
preferences did not facilitate learning to read.
Hammill and Larsen (1974) provided one of the most comprehensive
reviews, covering 39 studies using the Illinois Test of Psycholinguistic
Ability (ITPA). Of the 280 experimental-control comparison, only 106
supported training of psycholinguistic skills. One interesting point
in their table was that 11 articles dealt with educably mentally re-
tarded subjects, five with trainably mentally retarded subjects, 19
with disadvantaged subjects, and four with "other" subjects. The
category of LD was not presented in the table codes. The conclusion
was that "the efficacy of training psycholinguistic functioning has
not been conclusively demonstrated" (p. 12). Minskoff (1975) provided
a criticism of many aspects of Hammill and Larsen's (1974) review.
Three major problems involved the incompatibility of the nature of
the subjects, the nature of the treatments, and the experimental designs.
She emphasized that many of the subjects did not have learning disabilities.
Lund, Foster, and McCall-Perez (1978) reanalyzed 24 of the same 38
studies. They found that six studies showed positive results. Of
10 studies reported as showing negative results, they reported four
actually showed positive results and were reported inaccurately. They
reported that two others contained insufficient data, and two compared
treatment groups instead of trained versus untrained groups. An addi-
tional three studies were positive but limited in scope. Finally,
the last four studies had miscellaneous problems in the way they were
Hammill and Larsen (1978) replied by focusing on the 10 articles
showing negative results. They agreed that one study reported as nega-
tive was actually positive, but the other three were correctly reported.
They also replied persuasively to the other criticisms.
Larsen and Hammill (1975) also did a review of 60 studies of visual
perceptual abilities and school learning. They found the results to
be mainly contradictory. The major contaminating influence was the
fact that most studies lacked control for age and intelligence.
Kavale and Glass (1981) explained the need for meta-analysis in
exceptional education using the effect size statistic. They cited
the example of the psycholinguistic debate as an example of the need
for this more systematic technique for generalization across studies.
Kavale then did a meta-analysis of visual perceptual skills and read-
ing (Kavale, 1982). Kavale (1981) also performed a meta-analysis on
the literature assessing the efficacy of psycholinguistic training.
He examined the original 38 studies and six additional studies since
1974. Six were not obtained and two were eliminated, so he based the
analysis on 36 studies. He concluded that the literature had demon-
strated the effectiveness of psycholinguistic training.
Larsen, Parker, and Hammill (1982) replied with a series of crit-
icisms of Kavale's study, primarily focusing on the articles chosen
since Kavale (1981) did not use all of the same studies. They also
performed their own meta-analysis including the additional studies.
In no case did the change due to training exceed the standard error
of measurement (SEM). Again the results failed to demonstrate that
psycholinguistic training had value.
Torgesen (1979) pointed out that in addition to the controversy
over training psycholinguistic processes, generalization to academic
skills still lacked solid evidence. He suggested that task-centered
process assessment be substituted for the child-centered process assess-
ment. This would be a type of task analysis to measure the processes
required for the performance of specific academic tasks in specific
A unique approach to the problem was the single subject design
study by Koorland and Wolking (1982). Using behavior analysis, they
investigated the consistency of modality preferences across tasks when
reinforcement contingencies were either consistent or inconsistent
with the preferences. Experimental control was demonstrated rapidly.
When reinforcement contingencies favored visual responding, visual
responses more than doubled the baseline, and exceeded the auditory
responses (two out of three). The surprising result was that, using
reinforcement, performance in the nonpreferred modality approached
or even exceeded performance in the preferred modality.
These results suggested that modality preference is heavily in-
fluenced by the reinforcement the subject receives. Thus it would
seem that modality preference was a measure of learning heavily in-
fluenced by whatever reinforcers the subject had been receiving. This
explanation was compatible with the ambiguity of the research in the
Multisensory approaches to teaching reading and spelling have
been used with reading disabled students since the 1920s. The modality
processing theories about learning disabilities reinforced the use
of these reading approaches. The theories explained program effective-
ness as due to overcoming the limits of weak modality processing by
using several modalities, some stronger, to reinforce each other. Thus
the literature on modality preference, particularly on aptitude-treat-
ment interaction, also applied to multisensory teaching.
One of the most widely known techniques is the Visual-Auditory-
Tactile-Kinesthetic (VAKT) Technique (Fernald, 1943). This technique
was used with the comparison group, in place of the SIMS program, for
the Section One comparisons in this dissertation.
Thorpe, Lampe, Nash, and Chiang (1981) did a single subject design
study comparing the VAKT technique to a visual-auditory technique.
The results showed the VAKT to be the superior procedure. A review
of the literature by Myers (1978) reported 16 studies where the method
was useful or successful, and several studies where the results were
negligible. His conclusion was that there was little research finding
"differences of gain in reading" when using the multisensory approach.
The VAKT technique does represent a widely used method representative
of traditional LD reading remediation.
The purpose of this research was to investigate the importance
of visual selective attention to learning disabled student's reading
performance. Eleven hypotheses were used to investigate the important
aspects of this study. They were designed to especially compensate for
two major limitations in the selective attention literature review, i.e.,
the size'of the learning disabled (LD) student samples studied, and the
relevance of the cognitive concept, selective attention, to the reading
performance of students in precision taught reading programs.
As stated earlier the research questions were investigated in
three sections. The first section used three small samples of subjects.
These subjects were either in the SIMS reading program, in a traditional
LD reading program (the Visual-Auditory-Tactile-Kinesthetic or VAKT
approach) or, for the small sample of students referred but not admitted
to the LD program, in a regular classroom basal reading program. This
section investigated hypotheses 1 and 2.
Hypothesis 1. Precision teaching improves attending and thus
alleviates selective attention's effects on reading performance among
the three reading groups. Therefore, a VAKT taught group's reading
scores will be most influenced by selective attention, a precision
(SIMS) taught LD group's reading scores will be less influenced, and
a non-LD group's scores will be least influenced by the student's selec-
tive attention level.
This first hypothesis investigated the research question which
asked if precision teaching (SIMS) favorably affected attending be-
havior enough that LD students experiencing reading problems attri-
butable to selective attention would be differentially helped compared
to other reading teaching methods. It also revealed how reading per-
formance in all groups would be influenced by selective attention.
Hypothesis 2. Selective attention will relate to reading perfor-
mance, i.e., lower visual selective attention performance is related
to lower reading achievement performance. This will be shown by the
total task performance measure, combined selective attention (C + I),
more strongly than it will be shown when measured by the percentage
selective attention measure (%C %I). This last measure has often
been used as an indication of selective attention efficiency (Tarver,
Hallahan, Kauffman, & Ball, 1976).
This hypothesis investigated whether the relationship between
selective attention and reading was best characterized as due to the
limited filtering or limited capacity process defined in the literature
(%C %I), or to the student's total processing capacity under distraction
(C + I). It used both ways of defining selective attention.
The second section used a sample of LD students in the SIMS program
larger than most of the selective attention research groups, It inves-
tigated hypotheses three though seven.
Hypothesis 3. Among three groups of LD students differing in
selective attention, the most effective selective attention group will
have the highest reading achievement score. The less effective selec-
tive attention group with good ability will have lower reading achieve-
ment scores. The less effective selective attention group with low
memory/attention ability will have the lowest reading achievement scores.
This hypothesis investigated the research question asking whether
grouping LD students by selective attention performance would show
differential reading improvement among the groups. The three groups
of students were identified according to their scores on the Hagen's
Central Incidental Attention Task. Group one had the poor selectors,
i.e., students showing poor selective attention but better memory or
attention. Group two had the poor performers, i.e., students with
both low selective attention and low memory/attention performance.
Group three had the subjects who showed little distraction with good
selective attention performance.
Hypothesis 4. Selective attention will relate to reading perfor-
mance on the Wide Range Achievement Test (WRAT). Low selective atten-
tion performance will be more closely related to low reading achieve-
ment when measured by combined selective attention (C + I), than when
measured by the composite, selective attention efficiency measure (%C -
This hypothesis investigated the question which asked if selective
attention's effect on reading performance would be explained better
by the incidental performance limiting central performance efficiency
measure (%C %I) or by the total capacity measure (C + I). This was
similar to hypothesis 2 in comparing the different reading methods
except that it used the large LD student sample.
Hypothesis 5. The visual selective attention task will relate
to modality test performance so that students low on visual modality
aptitude will score low on visual selective attention, resulting in
a high correlation. This will also help clarify any interaction be-
tween different but related underlying cognitive structures.
Hypothesis 6. The selective attention task theoretically measures
an underlying LD student characteristic. Therefore more of this variance
will be explained by student modality preference (Detroit Test of Learn-
ing Aptitude) than by student IQ scores.
These hypotheses examined how visual modality aptitude, one of
the first "process" constructs used for explaining learning disabilities,
related to selective attention. Low visual selective attention was
expected to relate to low visual modality aptitude, and these would
be related more to the processes underlying the specific learning dis-
ability than to general cognitive processes. Investigating these re-
lations provided information on how the two psychological constructs
related to each other.
Hypothesis 7. Differences between the students' LD teachers will
influence their reading achievement scores more than the influence
of the students' selective attention performance. This hypothesis
was intended to make sure that any effects ascribed to selective atten-
tion were not due to other influences, such as teacher differences.
The third section used students from the first section. It com-
pared their performance on several other tests to their initial per-
formance on the selective attention task.
Hypothesis 8. Scores obtained from Hagen's Central Incidental
Attention Task will correlate significantly with other measures of
selective attention. This used the indices recommended by Argulewicz
Hypothesis 9. Hagen's Central Incidental Attention Task scores
will correlate significantly with meta-attention scores, revealing
a developmental trend from reward orientation to interest orientation.
These two hypotheses were intended to help determine when Hagen's Central
Incidental Attention Task was measuring the same underlying selective
attention construct as other instruments and techniques used to study
Hypothesis 10. Practice effects in a test-retest administration
of Hagen's Central Incidental Attention Task will produce an improve-
ment in incidental task scores at the expense of central task scores.
These data were intended to clarify the filtering or limited capacity
ideas by readministering the task.
Hypothesis 11. Teacher ratings of classroom attending behavior
will correlate significantly with LD student performance on Hagen's
Central Incidental Attention Task.
This hypothesis represented an attempt to clarify the relation
between attending behavior in the classroom and the experimental instru-
ment. The attending behavior in the classroom should have been ade-
quately represented by ratings given by the student's regular classroom
This study was designed to systematically proceed through the
investigation of these 11 hypotheses in three sections. Each hypothesis
will be the subject of a separate analysis. Each section represented
a different group of subjects.
In Section One, the first analysis used three reading groups from
Marion and Brevard Counties, the SIMS program LD group, the VAKT (Visual-
Auditory-Tactile-Kinesthetic) program LD group, and the classroom program
non-LD group. Later, because of the small size of the third group,
only the SIMS and VAKT program groups were used. In Section Two a
different, larger group of LD students in the SIMS program in Marion
County was used. In Section Three a small sample of students from
Section One was used again. These group labels, i.e., SIMS, VAKT,
and Regular, are used in Table 1 to clarify the variable involved in
the testing of each hypothesis. Specific information on the subjects
will be given later.
Variables Examined for Each Hypothesis
VAKT/Brevard LD (C) WISC-R
R. Programs (2)
Age Group (4)
(C + I)
Table 1 continued
Hypo- Subjects Independent Dependent
thesis # Number Group Variables Variables
9. 16 VAKT/Brevard LD (C) Meta- S. Attention
Attention Task Subscores
10. 16 VAKT/Brevard LD (C) Selective S. Attention
Attention Task Task
11. 16 VAKT/Brevard LD Attending S. Attention
11 Non-LD Behavior Task
C = Continuous variables
As shown in Table 1, the independent variables were measures of
student characteristics. They included selective attention perfor-
mance, modality preference group, intelligence (IQ) scores, and
Selective attention. Selective attention performance was ob-
tained from the results of testing with Hagen's Central Incidental
Attention Task (Mercer, 1975). These results, including the subtests,
were reported the following five different ways:
1. Central Scores (C)
2. Incidental Scores (I)
3. Selective Attention Efficiency Index (%C %I)
4. Selective Attention Combined Capacity (C + I)
5. Selective Attention Level
The central task score (C) was expected to reflect primarily atten-
tion and memory processes, like a recall test. The incidental task
score (I) was much more sensitive to attention or selective attention
and was expected to vary more within the subject.
The selective attention efficiency index (%C %I) was used in
several research studies (Tarver, Hallahan, Kauffman, & Ball, 1976).
It produced an overall score that attempted to reflect the idea that
performance onthe incidental measure (I) reduced the capacity for
performance on the central measure (C).
The.combined selective attention measure reflected the total
capacity for attention under the interaction, distraction, and capacity
limitations of the two measures (C + I). Because there are 12 central
items and six incidental items, the score was weighted toward the as-
signed task. Consequently, poor selection emphasizing the incidental
stimuli would lower the central score more than it could have potentially
raised the incidental score.
Two types of assignments to levels based on selective attention
performance were used. In Section One, only high and low selective
attention performance levels were identified:
1. High: performance with a central score greater or equal
to 6, and a selective attention efficiency index greater
2. Low: performance on C or on the efficiency index less
than the high level's criteria.
In Section Two, three performance levels were identified. These
were first differentiated by observation, looking for people who did
well on the central task while showing some attention to the incidental
task. The three levels found were as follows:
1. Level one poor selectors with a central score less
than or equal to 8 and a total score (C + I) greater
or equal to 9.
2. Level two subjects with poor overall performance
having central scores less than or equal to 8 and
a total score (C + I) under 9.
3. Level three subjects with good selective attention
performance having central scores equal to or greater
than 8 and incidental scores under 5.
Groups one and two differed in the overall level of performance.
Group one only showed poor selective attention while still attending
and remembering a lot of information. Group two had a more general
problem, either in attention or memory, and showed both poor selective
attention and also overall poor performance.
Modality preference. This was determined by ratings of LD student's
performance on the Detroit Tests of Learning Aptitude (DTLA) (Baker
& Leland, 1967). The ratings were performed in a global manner, using
low age-equivalent scores previously reported by teachers as evidence
of the LD disorder (Appendix A). These scores were identified by the
teacher as 80% or 70% or less of the expectancy age (EA) calculated as
2 x MA + CA EA
But rather than emphasize this quantitative approach, using very in-
adequate norms, these ratings consisted of recognizing overall patterns
in the low scores reproted as follows:
1. Group 1: low scores in visually oriented subtests only
2. Group 2: low scores in auditorily oriented subtests only
3. Group 3: low scores in both areas or no low scores.
Intelligence scores. These are total IQ scores reported from
individually administered intelligence tests. Most of the scores were
from the Slossen or the Wechsler Intelligence Test for Children Revised
Chronological age. Ages were reported in years and decimal fractions
of years. This was the age at the first testing with Hagen's Central
Incidental Attention Task.
The following types of data were collected:
1. The description of the student diagnosed as learning disabled
(LD) included the student's chronological age, intelligence
test score or mental age, a number indicating different
teachers' classes, and the teacher's judgment as to whether
hyperactive or emotional behaviorvas also exhibited by the
student in addition to his LD learning problems.
2. The achievement test performance included scores on the
instruments at the end of the 1979-80 school year, and
at the end of the 1980-81 school year. The analysis
focused on the oral reading scores (decoding) and com-
prehension scores from the Peabody Individual Achievement
Test (PIAT) for Section One and the Wide Range Achieve-
ment Test (WRAT) for Section Two. Spelling and mathe-
matics test performance was recorded but not analyzed
3. Modality performance was obtained by rating teachers'
reports of low subtest scores on the Detroit Test of
Learning Aptitude as explained above.
4. The scores on Hagen's Central Incidental Attention Task
were recorded as a central score, an incidental score,
an efficiency index (%C %I), and combined capacity
(C + I). It was administered at the end of the 1980-81
The rest of the data was obtained only on certain subgroups.
5. All the elementary level LD teachers in Marion County
(N = 15), including those involved in this study, were
asked to complete a questionnaire giving their opinion
of the SIMS program at the end of the 1980-81 school
year (Appendix F). Seven of this group completed the
same questionnaire at the end of the 1981-82 school
6. Additionally, teacher ratings were obtained on the 31
Brevard County students at the end of the 1980-81
school year. This included the 20 LD students in
the VAKT reading program and the 11 non-LD students
in the regular classroom program. These were ratings of
the LD student's attentional behavior in the classroom
(see Appendix C). The ratings were again collected on 16
subjects remaining from the original group at the end of
the 1981-82 school year.
7. As shown in Table 1, 16 subjects from the Brevard VAKT
group received additional testing at the end of the
1981-82 school year. These results provided infor-
mation for developing correlations with the initial
administration of Hagen's Central Incidental Attention
Task. First the indices of selective attention used
by Argulewicz (1982) were obtained by using the
Wechsler Intelligence Scale for Children Revised
(Wechsler, 1974) and the Stanford-Binet Intelligence
Scale (Termin & Merrill, 1973).
8. Next a meta-attention task devised by Loper, Hallahan
and lanna (1982) was administered to the 16 subjects.
9. Hagen's Central Incidental Attention Task was also
readministered to these 16 subjects, providing a one
year interval between testing.
Data analysis. The general relationship and distribution of the
variables were investigated using procedures from the Statistical Analysis
System (SAS) program package (Barr, Goodnight, Blair, & Chilko, 1979). Bar
graphs and correlations helped clarify the relationships among the
variables (see Appendix G). The univariate procedure was used to
determine if the characteristics of the samples in Section One and
Section Two were normally distributed. Again, a major limitation
was the use of small and unbalanced samples.
As indicated in Table 1, for the first seven hypotheses a re-
gression analysis of the data was done usinq the General Linear
Models (GLM) procedure in the Statistical Analysis System (SAS) program
package. In most cases this allowed for the use of the pretest reading
score as-a covariate. In addition to beinq the best predictor of
posttest scores, the pretest was the ideal measure for including the
influence of history, training, and native ability in one measure.
Though it would be influenced by the learning disability and selective
attention, usinq the pretest as a covariate with the posttest as a
dependent variable was chosen as the best way of representing academic
For hypotheses 8, 9, and 10, Pearson product-moment correlations
were chosen for evaluating the comparisons. The scores on the instru-
ments were continuous, ratio type measures. Because of the rating
scale (Appendix C), hypothesis 11 was analyzed using Spearman rank-
The selective attention testing was done with volunteer teachers,
so there was a definite teacher bias. However, testing was done with
all their LD students, so the student sample was not biased. Testing
of the SIMS students was completed in the spring term of 1981. All
the students in the SIMS program were eligible for, and participating
in, the Learning Disability Program in the Marion County, Florida, public
schools. All these students met the LD criteria of Marion County
(Appendix A). Modality information and recent SIMS reading performance
were obtained on the entire population of the LD program in Marion
County, Florida. Only data from those classes undergoing the selective
attention testing were actually used.
The VAKT program LD subjects were an entire LD class in Brevard
County, Florida. The LD criteria were the same (Appendix A). The non-
learning disabled subjects were referrals to that LD class from the
total school. They were tested and not staffed into the LD program.
Therefore they constituted an excellent comparison group since teachers
considered their behavior similar to that of LD students but they were
found not to have met the eligibility criteria.
The Marion County Public School System is located in the central
part of Florida. It is a rapidly growing area of approximately 100,000
people. A large proportion of the population is in rural areas. The
socioeconomic level of this sample ranges from low to upper middle
class. The Brevard school is in a middle to upper class residential
This study attempted to maximize its implications for LD students
by using three instruments widely used in LD programs. Thurlow and
Ysseldyke (1979) studied the Child Service Demonstration Center LD
models. They found that only five assessment devices were used by
over half of the centers. This study used two of these widely used
measures, the Wide Range Achievement Test (WRAT) used by 59.0% (Jastak
& Jastak, 1978), and the Peabody Individual Achievement Test (PIAT)
used by 53.8% (Dunn & Markwardt, 1970).
The Wide Range Achievement Test
Description. The Wide Range Achievement Test (WRAT) (Jastak
& Jastak, 1978) had norms from age 5-0 to 11-11 for Level I. It
featured three subtests and took between 20 and 30 minutes to administer.
The reading test was administered individually. It required supplying
the subject with the list of words and asking that they be pronounced
aloud. In Level I, a pre-reading task, a vestibule test, was administered
when there were failures in the first line of the reading test. The
student then named 13 letters presented, matched 10 letters, and named
the first two letters of his name.
The WRAT manual (Jastak & Jastak, 1978) described the normative
population. All ethnic groups were included in the norms as they occurred
in the population at large. In Level I the number of subjects in each
population, i.e., 14 age groups at half year intervals from 5-0 to
11-6, ranged from 400 to 600. No attempt was made to obtain a representative
national sampling. Thurlow and Ysseldyke (1968) did a comparison of
the 1976 and 1978 normative data. They found that standard scores
on the two editions of the WRAT were closely comparable. Merwin (1972)
criticized the ambiguity in the identity and nature of the groups.
Validity and reliability. Validity and reliability information
in the manual (Jastak & Jastak, 1978) involved several editions because
all editions of the WRAT were based on identical test items. Standard
scores and percentiles were comparable for all editions, though the
1978 edition featured changes in the scaling techniques for the WRAT
norms. Previous editions used a scaling of arbitrarily assigned grade
ratings. In the 1978 edition, the raw scores were scaled.
Split-half correlation coefficients were r = .98 for Level I.
Correlation coefficients between the two forms (Level I and II) yielded
correlations between .94 and .88. The authors reported from "clinical
experience" that the coefficients varied from .90 and .95. In studying
unique variance of the WRAT subtests with the WAIS and WISC, the authors
decided unique variance represented unreliability and thus developed
reliability coefficients for reading from .92 to .97.
Thurlow and Ysseldyke (1979) rated reliability as technically
adequate. Merwin (1972) referred to these as "questionably high reli-
ability coefficients" (p. 64). Thorndike (1972) referred to the "start-
ling" split-half reliability and suggested this was influenced by the
test being timed and by the reading test being stopped after a specified
number of failures.
The manual's estimates of validity cited a study by Wagner where
WRAT levels and mid-term grades correlated +.88 (Jastak & Jastak, 1978).
They also supported the observation that raw score performance re-
flected growth factors since at no time did an older school group obtain
a smaller raw score than an earlier group.
Thurlow and Ysseldyke (1979) rated the validity of the WRAT as
technically inadequate, based on the manual's report. Merwin (1972)
questioned the degree to which these test tasks measured "achievement."
Thorndike (1972) complained that content validity was not even considered
and questioned whether the vestibule tests were measuring the same attri-
bute as the body of the test. He characterized the validity information
as containing "bizarre conceptions, and...somewhat exotic procedures"
(p. 68). Thorndike (1972) and Merwin (1972) both ended suggesting
the WRAT as primarily a clinical or research tool for a quick estimate
of general level of ability and educational background.
The Peabody Individual Achievement Test
Description. The Peabody Individual Achievement Test (PIAT) (Dunn
& Markwardt, 1970) was an individually administered achievement measure.
The norms were based on 13 levels (K-12). It featured five subtests:
mathematics, reading recognition, reading comprehension, spelling,
and general information. It was an untimed power test (30 to 40 minutes).
The reading recognition subtest had 84 items. Items 1 to 9 involved
matching letters. Items 10 to 18 presented letters to be named. Items
19 to 84 were words which the subject read aloud.
The PIAT manual (Dunn & Markwardt, 1970) reported that the national
test standardization sample used the same distribution as the population
for geography and type of community (urban, suburban, rural, etc.).
It had approximately 200 subjects in each of the 13 grade levels. Ages
ranged from 5 to 12. Thurlow and Ysseldyke (1979) performed a factor
analysis on the data from the standardization sample. For younger
grades both of the reading subtests and the spelling subtest formed
one factor, a verbal comprehension and reasoning factor. For higher
grades, this factor included the mathematics, general information,
and reading comprehension subtests, while reading recognition and spell-
ing formed a separate factor.
There were a few general criticisms of the test. Bannatyne (1974)
called the PIAT a "high quality" test and restricted his criticisms
to the spelling score. He lamented the lack of a rate of reading measure
and the lack of a frustration reading level measure. Proger (1970)
found a few general criticisms. He requested more directions specific
to special populations, i.e., particularly testing time, decried the
"denigration of professionalism" (p. 464), i.e., the manual stated
that professional experience was not needed for administering the PIAT,
and worried over aspects of the basal-ceiling procedures. French (1972)
also commented negatively on using paraprofessional administrators.
Lyman (1972) complained about the smaller standardization samples and
fewer subtests than group tests. Silverstein (1981) and Reynolds and
Gutkin (1980) complained of the practice suggested in the manual of
comparing raw scores on pairs of subtests. They advocated comparing
the standard score on each subtest to the mean of the standard scores
on all tests, and both articles offered tables of differences required
Validity and reliability. The manual reported reliability based
on Pearson product-moment correlations (Dunn & Markwardt, 1970).
Samples of from 50 to 75 subjects at kindergarten, first, third, fifth,
eighth, and twelfth grade levels were tested at one month intervals,
then correlations were calculated on raw scores. Reliability coefficients
varied from .42 for kindergarten subjects in spelling to .94 for third
grade subjects in reading recognition. For the reading recognition
subtest, 60 students in first grade had a reliability coefficient of
.89 and 54 students in third grade had a reliability coefficient of
.94. Finally, 51 students in the fifth grade had a reliability co-
efficient of .89. The standard error of measurement describing the
expected bands of error for these test scores in reading recognition
were 1.74 for grade one, 2.21 for grade three, and 3.90 for grade five.
Lyman (1972) offered one of the few related criticisms when he complained
that the test had lower test-retest reliability coefficients than group
The PIAT manual focused on two aspects of validity, i.e., item
(content) validity and concurrent validity (Dunn & Markwardt, 1970).
Content validity was developed by reducing 300 items per subtest to
84 items using item discrimination and difficulty indices.
Proger (1970) characterized this content validity as "sound enough"
(p. 466) but decried the lack of evidence for construct validity and
for predictive validity. Lyman (1972) made a similar point in asserting
that more research was needed before establishing the PIAT as a "valid
test" (p. 35).
Concurrent validity was the primary validity evidence in the manual
and the literature. The manual (Dunn & Markwardt, 1970) reported a
correlation with the Peabody Picture Vocabulary Test (PPVT) based on
testing the same subjects used for retesting for reliability data.
The overall median correlation was .57. The authors pointed out that
the PPVT generally correlated with achievement tests in the .50s. Most
concurrent validity studies with achievement tests compared the PIAT
to the other individual achievement test, the WRAT.
Comparison of the WRAT and PIAT
PIAT subtests differed from the WRAT subtests in both content
and approach, except for reading recognition. The PIAT manual (Dunn
& Markwardt, 1970) stated that "correlation with the WRAT cannot be
considered completely appropriate or pertinent evidence on the validity
of the PIAT" (p.51). Yet these were the two individually administered,
multiple subtest achievement tests available.
Several studies comparing the PIAT to the WRAT were summarized
in Table 2. The study by Sitlington (cited in Dunn & Markwardt, 1970)
used educable mentally retarded (EMR) adolescents (see Table 1). The
study by Soethe (1972) used normal students (13), reading disabled
students (12), and mentally retarded (MR) students (15). Wetter and
French (1973) used 23 male and 7 female learning disabled pediatric
outpatients (see Table 2).
WRAT and PIAT Correlations
C. Wetter &
D. Bray &
100 LD (by ag
G. Scull & 49 LD
Bray and Estes (1975) also used a group test, the California Achieve-
ment Test (CAT), and teacher ratings of students' functional academic
levels. The WRAT reading correlated .81 with CAT vocabulary, .72 with
CAT comprehension, and .83 with CAT total reading. The WRAT correlated
.90 with teacher ratings of both reading recognition and comprehension.
The PIAT reading recognition correlated .84 with CAT vocabulary, .65
with CAT comprehension, and .81 with CAT total reading. The PIAT
correlated .85 with teacher ratings of reading recognition and .87
with teacher ratings of reading comprehension. The WRAT and PIAT com-
parisons were summarized in Table 2.
Baum (1975) used 25 subjects from each of 4 age groups of subjects
from self-contained LD classes (see Table 2). Harmer and Williams
(1978) used students from a learning disabilities center (see Table
2). They used 62 subjects for the spelling comparison and 66 subjects
for the word recognition and math subtests. In their discussion, they
presented a few examples of students who had dramatically different
scores on the math subtests. They discussed how the differences in
test content, testing procedures, test format, and method of scoring
accounted for these differences. However, Scull & Brand (1980) pointed
out that these differences only reflected the population's interaction
with the test, and that many subjects in the sample were only experi-
encing mild learning disabilities. In their own study Scull and Brand
(1980) used severely learning disabled students. Both tests were ad-
ministered on admission to a treatment center and again, two years
later. They reported correlations for both original and follow-up
testing (see Table 2).
The Detroit Tests of Learning Aptitude
Descriptions. The process tests for this study are from the Detroit
Tests of Learning Aptitude (DTLA) (Baker & Leland, 1967) which is a
collection of 19 subtests. It has been widely used by LD teachers.
Thurlow and Ysseldyke (1979) reported that 20% of the model centers
used this instrument, and 75% of these used it for placement and instruc-
tional programming, with 62% using it for pupil evaluation and 25%
using it for program evaluation. Olson, Mercer, and Paulson (1981)
reported that in many states teachers and diagnosticians were currently
using the DTLA to evaluate process disorders, even when it was not re-
quired in the criteria for learning disability identification.
There are several limitations to the DTLA. Because of its limited
standardization sample, age of the norms, and lack of adequate reli-
ability and validity information, this research study used the DTLA
only as a research instrument for estimating the relative performance
of the students in auditory versus visual tasks. The tasks, parti-
cularly the auditory attention span (subtest no. 13) for related
syllables and the visual attention span for objects (subtest no. 9),
were similar to visual and auditory techniques used in research for
attention and memory estimation. The DTLA was used in this study to
make a very general distinction among subjects.
The subtests available were as follows: (1) pictorial absur-
dities, (2) verbal absurdities, (3) pictorial opposites, (4) verbal
opposites, (5) motor speed and precision, (6) auditory attention
span for unrelated words, (7) oral commissions, (8) social adjustment
A, (9) visual attention span for objects, (10) orientation, (11) free
association, (12) memory for designs, (13) auditory attention span
for related syllables, (14) number ability, (15) social adjustment
B, (16) visual attention span for letters, (17) disarranged pictures,
(18) oral directions, and (19) likenesses and differences. The authors
suggested that there were eight psychological functions that explained
The DTLA manual (Baker & Leland, 1967) reported that the standard-
ization initially involved 50, and later 150 students at every age
level from 3 years to 19 years by 3 month intervals. The initial 50
students ranged in IQ from 90 to 110 as measured by group intelligence
exams that were not specified. The initial standardization data were
developed in 1935. The current norms were developed in 1955 (Buros,
1972). All the norm population were students examined in the Psycho-
logical Clinic of the Detroit Public Schools.
Validity and reliability. Test reliability was reported in the
manual on 48 cases over a five month interval to be .959 (Baker & Leland,
1967). Another group of 792 pupils retested over a two or three year
interval had a correlation of .675. For this group of mentally retarded,
delinquent, and emotionally unstable children, the median IQ (multiplied
by 100) from the DTLA remained almost the same, 70 to 71, and the stan-
dard deviation remained at eight IQ points. The manual also provided
information that the correlations between 16 subtests on 100 children
ranged from .2 to .4 (Baker & Leland, 1967).
The DTLA manual was basically oriented toward computing an IQ
score. This score was inadequate compared to the Wechsler Intelligence
Scale for Children--Revised (WISC-R) and other similar intelligence tests
as well as unrelated to the main usefulness of the DTLA in LD programs.
Silverstein (1978) made the most appropriate criticism by saying that
considering the lack of appropriate reliability and validity after
40 years, one "would think that the situation might have been remedied
long before this" (p. 303). Studies have compared the DTLA to various
intelligence tests (Chiappone, 1968; Sandstedt, 1964) and to achievement
tests (Olson et al., 1981), but the major emphasis here was on whether
the DTLA represented a modality or process pattern with direct impli-
cations for teaching effectiveness.
Chiappone (1968) used the DTLA with educationally mentally retarded
(EMH) students to examine correlations with the Binet Intelligence Test
and the Wechsler Intelligence Scale for Children (WISC). No significant
differences were found between the DTLA and Binet means or the Binet
and WISC means, but the DTLA was significantly different from three
WISC subtests. This showed the DTLA related closely to the verbal
material of both tests, but not as closely to the performance content
of the WISC.
Sandstedt (1964) studied the relationship between a "memory span
test battery" from the DTLA and the WISC. Her subjects were 45 retarded
readers of average general mental ability. The test battery from the
DTLA included 5 subtests: auditory attention span for unrelated words
(No. 6), visual attention span for unrelated objects (No. 9), auditory
attention span for related syllables (No. 13), visual attention span
for letters (No. 16), and oral directions (No. 18).
Sandstedt (1964) found a significant difference for the retarded
readers between their lower performance on the auditory memory span
test of unrelated words and the visual test of unrelated objects. The
correlation between the Detroit total memory span, IQ, and the WISC
full scale IQ was .69. WISC performance scores were more closely
related to total memory span (.66) than were the WISC verbal scores
(.58). The auditory test for unrelated words correlated closer than
the total auditory memory span. This study thus supported the con-
tention that the DTLA really supplies supplementary information on
intellectual aptitude. The relation between the DTLA and academic
performance had been a major research concern as part of the overall
effort to study the relations between modality or process tests and
academic achievement reviewed in Chapter II.
Hagen's Central Incidental Attention Task
Unlike the previous tests, Hagen's Central Incidental Attention
Task (Mercer, 1975) was not commonly used in LD programs. Despite its
limitations, this instrument was chosen because of its frequent use in
the research literature on selective attention in LD students. It
should be pointed out that the Central Incidental Attention Task is
primarily a research instrument. As such, there were no norms for
the test, nor is there validity, or reliability data available. However,
the research literature utilizing this instrument was often used for
comparison and for providing information on construct validity.
The task could not be readministered because once the incidental match-
ing was tested, the student no longer "ignored" these stimuli. Conse-
quently, stability and reliability could not be measured. Another
question concerned the selective attention index of efficiency measure
(%C %I). It was often used and occasionally produced some interest-
ing correlations with other measures. Though it appeared to be most
directly related to the filter theory model where successful selective
attention involves filtering out incidental information, the data had
clearly shown this model to be too simplistic. A detailed explanation
for the continued use of this designation has not been offered. Hallahan
(Note 2) suggested this was a practical measure from the definition
of selective attention. Another measure was the total combined score
of both tasks (C + I), which reflected the influence of both tasks
on each other. This technique was suggested by Mercer (Note 7).
Description. The central task consisted of six 3 x 6 inch stimulus
cards with black line drawings of a household object on the top half
and an animal on the bottom half. Though the drawings were of some-
what poor quality, the same pictures (Mercer, 1975) were used in this
study. Pictures of the deer, dog, and camel were used in the practice
trial so that the subject would be familiar with them. The incidental
task consisted of a 22 x 6 inch piece of cardboard displaying the six
animals and pictures of the six objects. One modification for this
study was that sufficient cards were printed to allow separate packets
of the stimulus cards for each trial. This decreased the time required
Development. One of the earlier tasks that led to developing
this instrument was a series of picture cards with distinctively colored
backgrounds (Maccoby & Hagen, 1965). After a brief exposure, a cue
card with the same color as the background of one of the picture cards
was presented to the subject to match to the covered presentation card.
The incidental learning task consisted of matching a set of solid color
cards to a set of the pictures without the background color. Later
the stimuli were changed to cards with objects and animals paired on
each (Hagen & Kail, 1975). Hallahan, Gajar, Cohen, and Tarver (1978)
further modified Hagen's task for use as a group selective attention
measure. Hallahan, Tarver, Kauffman, and Graybeal (1978) made a more
complicated version of Hagen's task to allow repeated tests. Three
classes of stimuli were used, i.e., animals, household objects, and
Reference groups. There have been several studies outlining the
general performance trends of normal subjects of various ages (Hagen
& Kail, 1975; Maccoby & Hagen, 1965). The effort has begun, with small
samples, to outline the general pattern of performance for various
ages of LD subjects (Tarver, Hallahan, Cohen & Kauffman, 1977).
Procedures. The procedure used for Hagen's Central Incidental
Attention Task began with oral directions to the student: "I will
be showing you some cards. Pay attention to the animal. I will ask
you to point to the correct animal after I show you several cards."
Next the subject was given two practice trials using three stimulus
cards. The examiner showed them to the student, placed them face down,
and then asked the student to point to the camel. This procedure was
then repeated, but the student was asked to point to the dog.
For the central task, the experimenter presented each 3 x 6 stimulus
card face up for two seconds, then placed it face down in front of
the subject in a manner which proceeded from the subject's left to
right. After the last card for each trial, the cue card identical
to one of the trial cards was presented. The subject tried to select
the matching card. On the data sheet, the examiner indicated under
central task response whether the student was correct (+) or not (-).
The number of successful trials was the central task score. There
were 12 trials of 3 to 6 cards in length (see Appendix B).
Hagen's incidental task immediately followed the central task,
trial 12. It only followed the last central task. The student was
presented the 22 x 6 inch piece of cardboard with the six animals
pictured. Next, the student took picture cards (3 x 3 inch) of the
six household objects and matched the household object (incidental
stimulus) with the animal picture (central stimulus). The incidental
score was obtained by recording the number of correct pairings at the
bottom of the data sheet. The combined selective attention measure
was obtained by adding the incidental and central scores (C + I). The
selective attention efficiency measure (%C %I) or index score was
obtained for each subject by subtracting the proportion of correct
incidental scores from the proportion of correct central scores.
ANALYSIS AND RESULTS
The purpose of this reserach was to investigate the importance
of visual selective attention to learning disabled students' reading
performance. The results are divided into three sections corresponding
to those described in Chapter III. The first section reports the
comparison of the influence of selective attention on student perfor-
mance in different reading programs. The second section reports the
performance of students with different levels of selective attention
in the SIMS precision taught reading program. The final section describes
the examination of the instrument, Hagen's Central Incidental Attention
Task (Hagen & Kail, 1975; Mercer, 1975).
This section involved comparing 19 learning disabled (LD) students
in the Systematic Instructional Management Strategies (SIMS) reading
program in Marion County, 20 learning disabled students in a Visual-
Auditory-Tactile-Kinesthetic (VAKT) type reading program in Brevard
County, and 11 non-learning disabled students referred for testing but
not staffed into the Brevard learning disability program. The difference
in counties confounded the program variable. Additionally, due to their
small numbers and severe attrition, the non-learning disabled students
were later omitted from further analysis.
Examinations of student characteristics were performed using the
univariate procedure in the Statistical Analysis System (SAS) program
package (Barr, Goodnight, & Sail, 1979) to determine the characteristics
of the samples. The mean values for three groups on CA, IQ, and pretest
are presented in Table 3. The distributions of chronological age,
intelligence quotients, and pretest scores for the groups are shown
in the histograms in Appendix G.
Means and Standard Deviations of Chronological Age (CA), Intelligence
Quotient (IQ), PIAT Reading Recognition Pretest, PIAT Reading
CA IQ Reading Rec. Reading Comp.
Group Number Mean SD Mean SD Mean SD Mean SD
SIMS Prog. N = 19 8.75 1.32 96.2 11.52 1.81 .827 1.83 1.25
VAKT Prog. N = 23 8.77 1.39 103.0 13.69 3.50 1.17 3.91 1.30
Non-LD N = 11 9.10 .59 109.0 14.17 *4.41 1.47 *4.80 1.56
*Pretest-Posttest scores were only available for 6 of 11
Table 4 presents the results of the univariate procedure explained
in Chapter III. The study of LD students required that reading achieve-
ment scores would not be normally distributed. These subjects constituted
a sample having scores below a cut-off of the whole population of reading
students. The distribution would therefore be positively skewed, begin-
ning at the point wherever the limits (%EA) for serious deficiencies
were defined, in the most dense part of the distribution. Table 4
reveals that only the SIMS program students showed this pattern.
The Shapiro-Wilk W-Statistic and Skewness Statistic
for the Data Values as a Random Sample from a Normal Distribution
W: Normal Statistic
PIAT Pretest Reading
Group Number CA IQ Recog. Comp.
SIMS Prog. N = 19 W: .962 .949 .858 .845
P : .61 .42 .01* .01*
SK: -.40 .02 1.43 -.53
VAKT Prog. N = 20 W: .949 .962 .983 .943
P : .39 .57 .95 .42
SK: .40 -.43 -.27 -.39
Non-LD N = 11/6 W: .891 .877 .940 .821
P : .19 .15 .61 .09
SK: -1.37 -.59 .14 1.41
*p < .05
The small sample sizes available imposed limitations on statistics.
The three groups differed on intelligence and reading. The means re-
vealed that the non-LD group had the highest scores and the SIMS program
group had the lowest scores. The VAKT group's scores were higher than
the SIMS program group's scores in IQ and achievement. This made their
comparison difficult, but the use of a covariate technique helped correct
for some of this difference. Likewise, subject characteristics such as
chronological age (CA) and intelligence scores (IQ) had essentially
normal distributions even though the samples were small. However, the
students in the SIMS program had pretest PIAT reading recognition and
pretest PIAT comprehension scores that were not normally distributed.
Hypothesis 1 assumed that because precision teaching improves
attending it thus will alleviate selective attention's effect on
reading performance. Therefore a VAKT taught LD group's reading scores
will be most influenced by selective attention, a precision taught
(SIMS) LD group's reading scores will be less influenced, and a non-
LD group's reading scores will be the least influenced by the students'
selective attention levels. A regression analysis of the data was done
using the General Linear Models procedure (GLM) in the Statistical
Analysis.System (SAS) program package. The three reading treatment
groups (Table 5) were later reduced to two, SIMS and VAKT groups, to
better balance the sample size (Table 6). It was believed that the
following model would best describe the relationships expected:
Y = B0 + B1X1 + B2X2 + B3X3 + B4X2X3 + e
Y = Reading Posttest Achievement Score
X1 = Reading Pretest Achievement Score
X2 = Selective Attention Group
Level 1 = with C score > = 6, and %C %I > 8;
Level 2 = all others
X3 = Membership in treatment group
SIMS program, VAKT program, or regular basal program
(or later just the first two groups)
B4X2X3 = the interaction term for selective attention vs.
membership in treatment group
There was a strong relationship between model 1 and the dependent
variable due to the covariate. There was a significant relationship
between the three groups and PIAT reading recognition (R2 = .87) and
reading comprehension (R = .70), and between the two groups (VAKT
and SIMS) and reading recognition (R = .85) and reading comprehension
(R2 = .64). The results of the regressions for all three reading treat-
ment groups are given in Table 5. The results for just the LD students
in the SIMS program and the VAKT program are presented in Table 6.
Results of the Statistical Test for the
Analysis of Hypothesis 1: Three Groups
Reading Recognition as Y and X,
Source df SS F P>F
Pretest X1 1 37.81 112.11 .0001*
SA X2 1 .40 1.18 .28
Rdg. Groups X3 2 .57 .84 .43
Interaction X2X3 2 .29 .43 .65
Error 38 12.81
Total 44 100.48
Reading Comprehension as Y and X1
Source df SS F P>F
Pretest X1 1 43.69 24.55 .0001*
SA X2 1 .17 .09 .76
Rdg. Groups X3 2 2.20 .62 .54
Interaction X2X3 2 7.76 2.18 .13
Error 33 58.73
Total 39 198.85
Using PIAT reading recognition (Y) as the dependent variable,
the F value observed for testing for the interaction of selective
attention, X2, and the three reading methods used, X3, was F = .43.
The probability of obtaining that value when assuming a null hypo-
thesis was .65. Thus, there was not enough evidence to conclude that
such an interaction was present. The F value observed for testing for
the main effect of the selective attention group, high or low, was
1.18, yielding a probability of obtaining that value, assuming a null
hypothesis of .28. Once more, there was not enough evidence to con-
clude that an effect was present.
Using PIAT reading comprehension as the dependent variable, the
results were essentially the same. They lacked any evidence to con-
clude that an effect was present.
Results of the Statistical Test for the Analysis
of Hypothesis 1: Two Groups**
Reading Recognition As Y and X1
Source df SS F P>F
Pretest X1 1 30.18 84.99 .0001*
SA X2 1 .05 .14 .71
Rdg. Group X3 1 .62 1.74 .20
Interaction X2X3 1 .03 .08 .78
Error 34 12.07
Total 38 78.87
Reading Comprehension As Y and X,
Source df SS F P>F
Pretest X1 1 24.44 13.49 .001*
SA X2 1 4.12 2.28 .14
Rdg. Group X3 1 2.96 1.64 .21
Interaction X2X3 1 7.23 3.99 .06
Error 29 52.51
Total 33 145.59
*p < .05
** SIMS and VAKT programs
As shown in Table 6, the F value for the interaction was .08,
using Y = PIAT reading recognition score. The probability of obtain-
ing this value, assuming a null hypothesis, was .78. Thus there was
not evidence to conclude that this interaction was present. The F
value observed for testing for the main effect of the selective atten-
tion group (high or low) was .14, yielding a probability of .71.
Once more there was insufficient evidence to conclude that an effect
Using the PIAT reading comprehension as the dependent variable
(Y), the results were essentially the same. There was a lack of
significance in the tests for the interaction and for the main effect,
selective attention group (X2).
Hypothesis 2 stated that selective attention will relate to read-
ing performance. Low visual selective attention performance will be
related to low reading achievement when measured by the total task
performance measure (C + I) more strongly than when measured by the
percentage selective attention measure (%C %I). This last measure
had often been used as an indication of selective attention efficiency
(Tarver, Hallahan, Kauffman, & Ball, 1976). A regression analysis of the
data to investigate Hypothesis 2 was done using the General Linear
Models (GLM) procedure in the Statistical Analysis System (SAS) program
package. The model best describing this was still model 1 with X =
selective attention changing from the categorical group membership of
hypothesis one to the continuous measures, combined capacity (C + I)
or efficiency (%C %I). Again the model containing the covariate
explained a significant amount of the variance of Y, with R2 = .88
for reading recognition and R2 = .60 for reading comprehension. The
F ratio statistic was used to evaluate hypothesis 2. The results
when defining selective attention as the total capacity under dis-
traction (C + I) are presented in Table 7. The results when using
selective attention as the efficiency index (%C %1) are presented
in Table 8.
Results of the Statistical Test for the Analysis of Hypothesis 2:
Selective Attention as Total Capacity (C + I) With Two Groups**
YA = Reading Recognition
SA Total X2
**Rdg. Group X3
YB = Reading Comprehension
Source df SS F P>F
Pretest X1 1 30.85 41.05 .0005*
SA Total X2 1 .00 .00 .99
**Rdg. Group X3 1 1.85 .93 .34
Interaction X2X3 1 3.78 1.90 .18
Error 29 57.82
Total 33 145.59
*p < .05
**SIMS and VAKT groups
Using PIAT reading recognition as the dependent variable, the F
value observed for testing for the interaction of selective attention,
defined as total capacity (C + I), with the two LD reading groups was
4.21. This meant that, assuming a null hypothesis, the probability of
observing data that would yield an F value of 4.21 was .05. This led
to the conclusion that there was evidence that a relationship did
exist between Y, PIAT reading recognition, and the interaction, since
it was improbable that this F ratio occurred only by chance. The
data also yielded an F value of 6.50 for the effect of the reading
groups, with a corresponding P value of .02. This was significant at
the .05 level. The F value for the main effect of the continuous
capacity measure of selective attention (C + I) was 9.09. This had
a corresponding P value showing it was significant at the .05 level.
This led to the conclusion that a relationship did exist between
selective attention, measured as total capacity, and reading perfor-
mance, since it was improbable that the observed F ratio occurred
only by chance. However, because the VAKT group's means were higher
in IQ, pretest and posttest measures, and combined selective atten-
tion (C + I), the results may be emphasizing the difference between
Using the PIAT reading comprehension scores as the dependent
variable (Y), the F value for the interaction was F = 1.90, with a
corresponding probability of .17. There was not enough evidence to
conclude that this interaction was present. For selective attention
capacity (C + I), the F value for the main effect was .35. The
corresponding probability of obtaining that value of F was .56. Thus,
there was not enough evidence to conclude that an effect was present.
The results using the selective attention efficiency measure (%C %I)
are reported in Table 8. Again model 1 containing the covariate
explained a significant amount of the variance of Y, with R = .85
for reading recognition and R2 = .57 for reading comprehension.
Results of the Statistical Test for the Analysis of Hypothesis 2:
Selective Attention as Efficiency Measure (%C %I) With Two Groups**
YA = Reading Recognition
Source df SS F P>F
Pretest X1 1 32.24 95.67 .0001*
SA % X2 1 .48 1.42 .24
**Rdg. Group X3 1 .32 .96 .33
Interaction 1 .27 .83 .37
Error 34 11.45
Total 38 78.87
YB = Reading Comprehension
Source df SS F P>F
Pretest X1 1 39.98 18.53 .0001*
SA % X2 1 .11 .05 .81
**Rdg. Group X3 1 1.00 .47 .50
Interaction 1 .16 .07 .79
Error 29 2.56
Total 33 145.59
*p < .05
**VAKT and SIMS Programs
Using PIAT reading recognition as the dependent variable, the F
value of the interaction yielded a probability of obtaining that value
of .37. Consequently there was not enough evidence to conclude that
an interaction was present. In testing the main effect, selective
attention efficiency, the data yielded an F value of 1.42 which had
a corresponding P value of .24. Again, there was not enough evidence
to conclude that an effect was present.
Using PIAT reading comprehension as the dependent variable, the
F value of .07 for the interaction of selective attention efficiency
(%C %i.) and the reading programs (SIMS or VAKT) yielded a probability
of obtaining that value by chance of .79. Thus there was insufficient
evidence to conclude that such an interaction was present. The F
value observed for testing the main effect, selective attention
efficiency (%C %I), was F = .05, corresponding to a probability of
.81. Thus there was no evidence to conclude that an effect was present.
Table 9 presents the means for these groups on the posttests and
selective attention measure to show the magnitude and direction of these
effects. Gain scores are also included to facilitate comparisons.