Parham1998 IS Afecta Desemepño Academico

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 23

The Relationship of Sensory

Integrative Development to
Achievement in Elementary
Students: Four-Year Longitudinal
Patterns

L. Diane Parham

Measures ofsensory integration in elementary students are


significantly related to school achievement both concurrently
and predictively over a 4-year period, even when controlling
for intelligence. A particularly strong link between praxis
and arithmetic achievement is evident.

Key words: perceptual-motor • learning disabilities


• child development

Abstract

This longitudinal investigation evaluated the relationship of sensory inte-


grative development to achievement. It was hypothesized that perceptual
abilities would be strongly related to achievement when participants were
6 to 8 years ofage, but not 4 years later. Participants (32 school-identified
learning handicapped children and 35 non-learning handicapped chil-
dren) were administered the Sensory Integration and Praxis tests and the
Kaufman Assessment Battery for Children. As predicted, sensory inte-
grative factors were strongly related to arithmetic achievement at younger
ages, and the strength of the association declined with age. The reverse

L. Diane Parham, PHD, OTR, FAOTA, is Associate Professor in the


Department of Occupational Therapy at the University of Southern
California, Los Angeles, California.
This paper was accepted during the Editorship of Margaret Short-DeGraff.

Summer 1998, Volume 18, Number 3 105

Downloaded from otj.sagepub.com at UNIV OF PENNSYLVANIA on June 27, 2015


pattern was found for reading: sensory integration was not significantly
related to concurrent reading achievement at younger ages, but was relat-
ed to later reading. An unexpected finding was the strength of the rela-
tionship of the sensory integrative factors, particularly Praxis, to arith-
metic achievement.

he value of perceptual and perceptual-motor assessment in

T special education is a controversial issue. From a traditional


perspective, mild perceptual and motor disabilities are hall-
mark features of learning disabilities (Cruikshank, 1985; Frostig,
1972; Kirk & Gallagher, 1983), and therefore deserve careful assess-
ment. However, this view has come under heavy fire by educators
who view perceptual and perceptual-motor assessment as tangen-
tial, irrelevant, or, at worst, detrimental to educational programs for
children with mild learning problems (Coles, 1978; Council for
Learning Disabilities, 1986). As a result, some experts have recom-
mended eliminating perceptual testing from services delivered to
learning-disabled children (Coles, 1978, 1987; Council for Learning
Disabilities, 1986; Larsen, Rogers, & Sowell, 1976).
The negativity toward perceptual and perceptual-motor testing
is often linked to the psychometric shortcomings of commonly used
instruments (Coles, 1978; Council for Learning Disabilities, 1986;
Larsen et al., 1976). A particularly critical validity question is
whether or not perceptual-motor measures are related to academic
achievement.
Age of participants may be a crucial factor in the relationship of
perceptual-motor development to achievement. Satz and colleagues
found that some perceptual measures are very discriminative of
achievement in kindergarten and the early elementary grades, but
not as sensitive at older ages (Satz et al., 1978). These researchers
hypothesized that perceptual measures are strong predictors of
learning problems during the early school years, when perceptual
abilities are in developmental ascendancy, but not after perceptual
development has plateaued in middle childhood. If this is so, per-
ceptual-motor measures may be useful for assessment and early
identification of children with learning problems in the primary
school years or earlier, but not later on. Consistent with this view
are the findings of Kavale (1982), who reported on the basis of an
extensive correlational and meta-analytic study that visual percep-
tual performance is an important factor in reading achievement in
preschool and primary grades, but relatively inconsequential at
intermediate grades.
Clearly, the developmental characteristics of perceptual-motor

106 The Occupational Therapy Journal of Research


Downloaded from otj.sagepub.com at UNIV OF PENNSYLVANIA on June 27, 2015
measures in relation to achievement must be better understood if
their significance for learning disabilities is to be clarified. This
research contributed to knowledge in this area through a longitudi-
nal study of elementary school children. It was designed to assess
the Satz hypothesis that perceptual measures are strongly related to
achievement in the primary grades, but not at older ages (Satz et al.,
1978). In addition, because the perceptual-motor instruments used
in this study are based on the sensory integrative conceptual frame-
work developed by Ayres (1972, 1979, 1985), an evaluation of this
theoretical orientation was permitted.
The sensory integrative conceptual framework uses a neuropsy-
chological approach to explain the development of perceptual and
perceptual-motor problems in children with learning disorders
(Ayres, 1972). Ayres defined sensory integration as the ability to
organize sensory information for use in goal-directed activity (1972)
and asserted that sensory integration is relevant to school success
because it impacts on learning and behavioral organization. In this
view, disorders in sensory integration account for aspects of some
learning and behavior disorders.
Ayres (1972) posited that if sensory integration at a more primi-
tive, body-centered level is not well developed by the time a child
reaches school, then skills needed for success in school will not be
acquired efficiently. This was not to say that higher cognitive skills
are wholly dependent on sensorimotor abilities, or that a child with
sensory integrative immaturity necessarily will do poorly in school;
rather, it was proposed that immature or deficient sensory integra-
tion is likely to interfere with the efficiency, ease, or automaticity
with which a child handles new information and accomplishes
tasks. This, in turn, impedes classroom functioning and creates
stress, making the child vulnerable to achievement or behavioral
difficulties in the school setting.
Empirical studies indicate that scores on the sensory integrative
measures developed by Ayres discriminate between learning-dis-
ordered and normally achieving children (Ayres, 1965, 1989;
Cermak & Ayres, 1984; Freides et aI., 1980; Kinnealey, 1989;
Morrison, 1986; Servello, 1982). However, previous studies have
not directly examined the relationship between sensory integra-
tion and school achievement. The present study was designed to
address this problem.

Research Question and Hypotheses

The research question that guided this study is: Is sensory inte-

Summer 1998, Volume 18, Number 3 107

Downloaded from otj.sagepub.com at UNIV OF PENNSYLVANIA on June 27, 2015


grative performance in the early grades of school related to concur-
rent and later academic performance? According to the conceptual
framework of Ayres, it should be related to academic performance
both in the early and later grades, but this has not been tested
empirically. The perceptual ascendancy theory of Satz et al. (978)
would be consistent with this expectation, and further, would add
that sensory integrative performance in the later grades should not
be concurrently related to academic performance.

Hypotheses

1. Sensory integrative performance at younger ages (6 to 8 years)


is significantly related to concurrent academic achievement.
2. Sensory integrative performance at younger ages is significant-
ly related to later academic achievement; a predictive relation-
ship over time exists.
3. Sensory integrative performance at older ages 00 to 12 years) is
not significantly related to concurrent academic achievement.

Method

Procedures

This research is a multivariable, two-wave longitudinal study.


The first wave of data collection took place in Fresno, a university
community in the agricultural central valley of California, during
the Fall of 1984. The second wave of data collection took place 4
years later, during the Fall of 1988. All but three of the participants
remained in the Fresno area at the time offollow up. Procedures for
both waves of data collection were similar.
Most participants were tested at school during regular school
hours; a few were tested in their homes. The Sensory Integration
and Praxis Tests (SIPT) and the Kaufman Assessment Battery for
Children (K-ABC) were administered to each subject on separate
days, in two 2-hour sessions. Tests were administered in accordance
with standardized procedures by trained examiners.

Measurement of Sensory Integration and Praxis

The Sensory Integration and Praxis Tests (SIPT) provide the mea-

108 The Occupational Therapy Journal of Research

Downloaded from otj.sagepub.com at UNIV OF PENNSYLVANIA on June 27, 2015


sures of sensory integration and praxis in this study (Ayres, 1989).
The SIPT is a battery of 17 tests measuring motor-free visual per-
ception, somatosensory perception, visual-motor coordination,
praxis, and vestibular functions. Extensive information on norma-
tive data, reliability, and validity is presented in the SIPT manual
(Ayres, 1989).
Three SIPT factor scores, described below, were used in the data
analyses. Use of factor scores enabled the 17 SIPT variables to be
collapsed into several composite measures, making interpretation
of results more straightforward. Factor scores used in the present
study were generated from a principal components analysis report-
ed by Ayres (1989). Participants in this principal components analy-
sis consisted of normally achieving and learning-disordered chil-
dren (n=293), including the participants of the present study.

Praxis Factor

Praxis is the ability to conceptualize, plan, and execute novel


motor acts (Ayres, 1985, 1989). Ayres (1989) viewed this construct as
a cognitive function dependent on adequate integration of spa-
tiotemporal inputs from touch, kinesthetic, proprioceptive, vestibu-
lar, and visual sensory receptors.
This factor primarily involves tasks requiring planned body
movements, as well as a complex, spatiotemporal aspect of tactile
perception. The tests that define this factor are Oral Praxis,
Graphesthesia, Bilateral Motor Coordination, and Sequencing
Praxis. Standing and Walking Balance, Praxis on Verbal Command,
and Postural Praxis also load on this factor (Ayres, 1989).

Visual Perception Factor

This factor reflects visual perception and the ability to visually


guide and organize body movements in space. It is defined primar-
ily by the two motor-free visual perception tests (Space
Visualization and Figure Ground Perception), the two construction-
al praxis tests (Design Copying and Constructional Praxis), and the
Motor Accuracy test of eye-hand coordination (Ayres, 1989).

Somatosensory Factor

This factor is thought to reflect simpler, less cognitive functions

Summer 1998, Volume 18, Number 3 109

Downloaded from otj.sagepub.com at UNIV OF PENNSYLVANIA on June 27, 2015


than the other factors. It primarily involves one-point tactile local-
ization and kinesthesis. The tests that define this factor are
Localization of Tactile Stimuli and, to a much lesser degree,
Kinesthesia and Finger Identification (Ayres, 1989).

Exclusion of a Postrotary Nystagmus Factor

A fourth factor was reported in the same principal components


analysis that yielded the three factors described previously (Ayres,
1989). This fourth factor is characterized by a high positive loading
of postrotary nystagmus duration contrasted with a negative load-
ing of Praxis on Verbal Command. A difficulty in interpreting this
factor is posed by the fact that both very low and very high stan-
dard deviation scores on the Postrotary Nystagmus test are associ-
ated with dysfunction in children (Ayres, 1989). The linear statisti-
cal models employed in this study, therefore, are inappropriate for
analyses that center on questions regarding postrotary nystagmus.
Furthermore, this factor probably is not meaningful for children
who are normally developing, as it did not emerge in a principal
components analysis of 1750 children from the normative sample
who were not identified as having achievement or perceptual prob-
lems (Ayres, 1989). Because results involving this factor would not
be interpretable, it was decided to exclude this factor from the pre-
sent study.

Measurement of Intelligence and Achievement

The Kaufman Assessment Battery for Children (K-ABC) is a set of


tests of cognitive processing and achievement for children aged 2%to
12% years with well-documented normative data, reliability, and
validity (Kaufman & Kaufman, 1983). The Mental Processing
Composite (MPC) of the K-ABC was used as a general measure of
cognitive ability. The measure of reading ability was provided by the
K-ABC Reading/Decoding subtest for participants 6 years of age; for
older participants, the K-ABC supplementary norms for a Reading
Composite score (Kamphaus & Reynolds, 1988) were used to mea-
sure reading achievement. The Reading Composite is derived from
the Reading/Decoding subtest and Reading/Understanding subtest
(not administered to 6 year aids) of the Achievement Scale.
Kamphaus and Reynolds (988) have provided evidence of adequate
reliability and validity for this measure. The K-ABC Arithmetic sub-
test of the Achievement Scale was used separately as a measure of

110 The Occupational Therapy Journal of Research

Downloaded from otj.sagepub.com at UNIV OF PENNSYLVANIA on June 27, 2015


arithmetic achievement. It consists entirely of orally presented prob-
lems in the context of an illustrated story; no paper-and-pencil items
are included.

Participants

During the initial data collection effort in 1984, a total of 91 chil-


dren, aged 6, 7, and 8 years, were tested. Of these, 43 were school-
identified as learning handicapped (LH). This is a California special
education classification used with children with learning disabilities,
behavior disorders, or mild mental retardation. LH participants were
originally identified through random selection, within-age strata, of
names from the school district's record of identified LH students;
exclusionary criteria included the presence of a known developmen-
tal disability, physical handicap, or sensory impairment. The LH chil-
dren were all diagnosed by school psychologists as learning disabled
in 1984, most with concomitant educational diagnoses of communi-
cation or behavioral disorder. By 1988, two had been reclassified as
mildly mentally retarded.
The comparison group consisted of 48 children who were placed
in regular classes in the Fresno Unified School District during the first
wave of data collection, and were not receiving any special education
services for academic or behavioral problems. These non-learning-
handicapped (NLH) participants were randomly selected from the
same schools as the LH participants, using the same age stratification.
Throughout this study, designation of participants as LH and NLH
is based on school classification in 1984. A total of 67 children (74% of
the original sample), aged 10 to 12 years, participated in follow-up
data collection in 1988. Of these, 32 were from the LH group, and 35
were from the NLH group. Table 1 summarizes the age, ethnicity,
gender, and socioeconomic characteristics of the LH and NLH sam-
ples at follow up in 1988.
For some participants, school classification had changed during
the intervening years. By 1988, 11 children from the original LH
group were 1'0 longer receiving special education services, although
most of them were still having academic difficulties. The remaining
children from that group had been continuously classified as LH
since 1984. Two NLH children were classified as learning handi-
capped in 1988 by the school district, one for learning disability and
the other for behavioral problems. One other child from the NLH
group had received services for learning disability for 1 year between
1984 and 1988. Two NLH children had been retained for one acade-
mic year or less.

Summer 1998, Volume 18, Number 3 111

Downloaded from otj.sagepub.com at UNIV OF PENNSYLVANIA on June 27, 2015


Table 1
Demographic Characteristics of Participants in 1988
LH NLH Total
(n=32) (n=35) (n=67)

Age II 0/0 II 0/0 n


10 8 250 16 45.7 24
11 10 31.6 10 28.6 20
12 14 43.8 9 25.7 23

Ethnicity
Asian 1 3.1 0 0.0 1
Black 2 6.2 4 11.4 6
Hispanic 10 31.3 11 31.4 21
White IF. 56.3 20 57.1 3F.
Other (Asian Indian) 1 3.1 0 0.0 1
Sex
Male 21 65.6 18 51.4 39
Female 11 34.4 17 48.6 28
Socioeconomic Status
I (Highest) 1 3.1 0 (J.O 1
II 5 15.6 5 14.3 1lI
III 11 34.4 24 6F..6 35
IV 9 2F..l 5 14.3 14
V (Lowest) 6 11'.8 1 2.9 7

Note. Socioeconomi: <tatu« is rated w~ing ihr Holling...:;Jzeud (1957) .<culr

Approximately 15% of the participants were from homes that


were bilingual (predominantly Spanish and English speaking), with
slightly greater frequency in the LH group. Behavior problems, as
identified by T-scores above fiO on the Walker Problem Behavior
Identification Checklist (Walker, 1983), were more prevalent in the
LH group (37% of these participants in 1984 and 43% in 1988), but
were also found to exist among NLH children (21 % in 1984 and 15%
in 1988).
Table 2 summarizes the performance of participants on the K-
ABC Mental Processing Composite (MPC), Reading Composite,
and Arithmetic subtest. As expected, scores reflecting intellectual
ability and achievement were lower on the average in the LH group.
Overall, these scores were quite stable (see Table 5), although mean
reading scores dropped 10 points in the LH group by 1988.

Analysis

Following testing, SIPT scoring sheets were optically computer-

112 The Occupational Therapy Journal of Research

Downloaded from otj.sagepub.com at UNIV OF PENNSYLVANIA on June 27, 2015


Table 2
Intellectual Ability and Achievement of Participants
LH NLH'
(n=32) (n=34)

Mean SO Range Mean SO Range


K-ABC
Mental Processing
1984 86.0 15.4 54-111 107.1 10.3 85-132
1988 86.6 16.7 56-122 107.5 11.9 85-132
K-ABC
Reading Composite
1984 83.3 16.0 54-118 104.4 11.2 83-130
1988 73.1 22.0 45-116 104.3 10.5 78-126
K-ABC
Arithmetic Subtest
1984 78.9 15.5 54-109 107.2 10.5 92-130
1988 80.6 15.5 56-117 106.7 8.8 85-120

Note: LH and NLH groupsarebased on classification in 1984-85. K-ABC scores arestandardscores


(mean = 100, SD = 15).
t K-ABC 1984 data weremissingfor one subject.
K-ABC = Kaufman Assessment Batteryfor Children, LH = learninghandicapped, NLH = non-learning
handicapped.

scannedby the test publisher to generate z-scores for each test.


Although raw scores. were not available, the test publisher supplied
factor scores to the researcher.
Because normative data on the SIPT do not extend beyond the
age of 8 years 11 months, z-scores for 1988 data were obtained by
using the 1984 ages of participants. In other words, each subject's
1984 and 1988 performance was scored in relation to the normative
data for that subject's age group in 1984. The 1984 data thus pro-
vide a baseline against which to judge 1988 performance.
Factor scores were computed through use of a formula that com-
bines the original SIPT z-scores of the participants with weights that
are proportional to the factor loadings of the tests (Kim & Mueller,
1978). The formula provides for standardization of factor scores so
that the mean is zero and the variance is one. These scores are exact
mathematical transformations of the original z-scores (Chatfield &
Collins, 1980;Kim & Mueller, 1978;Manly, 1986);higher scores indi-
cate better performance.
Factor scores for the LH and NLH groups in 1984 and 1988 are
summarized in Table 3. For each factor, LH mean scores were lower
than or equal to NLH mean scores. Two-sample t tests (adjusting for

Summer 1998, Volume 18, Number 3 113

Downloaded from otj.sagepub.com at UNIV OF PENNSYLVANIA on June 27, 2015


Table 3
SIPT Factor Scores of Participants
LH NLH
(n=32) (n=35)

Mean SO Range Mean SO Range


Praxis
1984 -2.0 1.5 -4.8, +1.6 -0.4 1.1 -2.5, +1.5
1988 -1.0 1.3 -4.0, +1.1 +0.6 0.9 -1.5, +2.1
Visual Perception
1984 -0.1 1.3 -2.3, +2.9 +0.4 0.9 -1.3, +2.2
1988 +0.7 1.1 -2.0, +3.0 +1.4 0.8 -0.2, +3.7
Somato--sensory
1984 -0.3 1.5 -2.7, +2.5 +0.1 1.1 -2.2, +2.3
1988 +0.0 1.1 -2.3, +2.2 +0.0 0.9 -2.3, +1.8

Note. Learninghandicapped (UI) and non-learning handicapped (NUl) groupsarebased on classification


in 1984-1985. Factor scores in 1988werederived from z-scores based on subjects'agesin 1984.

unequal variances) indicated that differences between LH and NLH


factor scores were statistically significant for Praxis in 1984 and
1988, and for Visual Perception in 1988 (P<O.008 using Bonferroni
correction for repeated t tests).
Distributions of SIPT factor scores were examined both within
the LH and NLH groups and in aggregate. Tests of normality con-
firmed that scores for each factor approximated a normal distrib-
ution within groups both in 1984 and in 1988. When the LH and
NLH groups were combined, SIPT factor score distributions
approximated a normal distribution in all cases except for Praxis
in 1988, which had a heavy tail at the low end of the distribution
due to two very low outliers.
To initially explore the relationship between sensory integration
and achievement, correlation matrices were generated. Next, a
series of regression analyses were performed to test whether or not
the sensory integrative factors accounted for a significant amount of
variance in reading and arithmetic scores. Because the hypotheses
reflected theoretical relationships that were applicable to all chil-
dren, regardless of the presence of a learning problem, an a priori
decision was made (prior to data collection in 1988) that the data for
the LH and NLH groups would be aggregated for the correlation
and regression analyses.

114 The Occupational Therapy Journal of Research

Downloaded from otj.sagepub.com at UNIV OF PENNSYLVANIA on June 27, 2015


Table 4
Intercorrelations Among 1984 Variables
SES MPC Read Arith Praxis Visual Somat
Age 0.25 -0.07 -0.21 -0.18 -0.19 0.14 0.05
SES 0.01 -0.28 -0.18 -0.09 -0.13 0.20
MPC 0.71:1: 0.77:1: 0.53:1: 0.32' 0.25
Read 0.76:1: 0.50:1: 0.23 0.09
Arith 0.52:1: 0.33' 0.21
Praxis -0.39t -0.26
Visual 0.16

Note. N=66 (onesubjectmissingK-ABC data). SES=Socioeconomic status, MPC=Mental Processing


Composite, Read=Reading Composite, Arith=Arithmetic, Visua/=Visual Perception,
Somat=Somatosensory.
'P<OOl tP<OOOl. tP<O.OOOl.

Results

Correlational Analyses

Three correlation matrices of Pearson's r coefficients were gener-


ated using data from all participants. The first presented relation-
ships among 1984 variables, the second provided correlations of
1984 with 1988 variables, and the third depicted associations among
1988 variables. In all of these analyses, associations among the SIPT
factor scores were low, as expected, because they were derived from
orthogonal factors in a principal components analysis.

Correlations Among Variables at Younger Ages

Table 4 presents the correlation matrix for 1984 variables. The


high correlations of achievement with intellectual measures are not
surprising, because a large body of research attests to the power of
intelligence in predicting achievement. Of the SIPT factors, Praxis
was, by far, the most strongly related to reading and arithmetic. It
was just as highly correlated with Mental Processing (rs of approx-
imately 0.50). Associations between Visual Perception and achieve-
ment variables were much lower, but still significant. Only very
weak associations existed between somatosensory and achieve-
ment scores.

Summer 1998, Volume 18, Number 3 115

Downloaded from otj.sagepub.com at UNIV OF PENNSYLVANIA on June 27, 2015


Table 5
Correlations of 1984 Variables With 1988 Variables
1988 Variables
MPC Read Arith Praxis Visual Somat
1984 MPC 0.8 6t 0. 69t 0.79t 0. 65t 0.42t 0.13
1984 Read 0.64t 0.83t O.64t 0.59t 0.22 0.02
1984 Arith 0.77t 0.81t 0.83t 0. 67t 0.38' 0.10
1984 Praxis 0.52t 0.48t 0.55t 0.59t 0.11 0.08
1984 Visual D.41t 0.34' 0.39' 0.03 0.46t -0.01
1984 Somat 0.22 0.07 0.07 0.11 0.19 0.42t

Note. N=66for 1984 K-ABC variables, 67 for all others MPC=MentalProcessing Composite;
Read=Reading Composite; Arith=Arithmetic subtest;Visual=Visual Perception; Somat=Somatosensory.
·P<O.01 tP<O.oOl. tP<O.OOOl.

Correlations Between 1984 and 1988 Variables

Table 5 presents the correlations between 1984 and 1988 K-ABC


and SIPT variables. Generally, the K-ABC measures were much
more stable than the SIPT measures over the 4-year period. Praxis
maintained its strong relationship with intelligence and achieve-
ment, particularly with Arithmetic, over the 4 years of the study.
Visual Perception showed somewhat stronger associations with
Reading than it had in 1984. The weak link between the
Somatosensory factor and Arithmetic in 1984 was even further
diminished in 1988.

Table 6
Intercorrelations Among 1988 Variables
SES MPC Read Arith Praxis Visual Somat
Age 0.25 0.02 -D.11 0.02 -D.23 -D.ll -D.01
SES -D.06 -D.35· -D.16 -0.08 0.07 0.08
MPC O.7 1t 0.82t 0.58t O.5°t 0.09
Read 0.74t 0.61t 0.26 0.06
Arith O.64t 0.37' -D.02
Praxis -D.01 -D.22
Visual -D.02

Note. N=67. SES=Socioeconomic status;MPC=Mental Processing Composite, Read=Reading


Composite, Arith=Arithmetic subtest, Visual=Visual Perception, Somat=Somatosensory.
·P<O.Ol. tP<O.OOl. tP<O.OOOl.

116 The Occupational Therapy Journal of Research

Downloaded from otj.sagepub.com at UNIV OF PENNSYLVANIA on June 27, 2015


Correlations Among Variables at Older Ages

Interrcorrelations among variables at older ages are shown in


Table 6. The association between reading and socioeconomic status
was somewhat stronger at older ages than younger, probably
reflecting long-term social influences. Contrary to expectation, the
relationships between Praxis and all cognitive and achievement
variables were also amplified. In addition, the associations
between visual perception and achievement measures were slight-
ly higher than they had been at younger ages. The earlier weak
links between somatosensory perception and measures of intelli-
gence and achievement had completely dissipated by 1988.

Regression Analyses

Multiple regression analyses tested whether the sensory integra-


tive measures accounted for a significant amount of variance in
reading and arithmetic achievement concurrently at different ages
and across a 4-year period of time. Stepwise regression methods
were not used, because these are appropriate for exploratory data
analysis prior to the formulation of specific hypotheses (Afifi &
Clark, 1984;Sall, 1981). A more appropriate approach for this study
was the building of multiple regression models that directly test
hypothesized relationships (Lewis-Beck, 1980; SaIl, 1981).
Regression analyses-therefore, followed a process of model build-
ing in which the value of sensory integrative factors was tested in
relation to achievement.
In each regression analysis, alternative models involving SIPT
variables were generated and compared. Selection of the best model
was directed toward obtaining a large R-square coefficient (which
reflects amount of variance explained by the model) with a small
number of independent variables, all of which significantly con-
tributed to the solution. The goal of the selection process was to
identify "the smallest and simplest model that is adequate to
describe the data" (Sall, 1981, pp. 2-3). The following steps accom-
plished this: (l) generate a series of multiple regression models
using all possible combinations of the independent variables of
interest, (2) select models in which all independent variables are sta-
tistically significant and eliminate the others, and (3) identify the
model with the highest R-square as the best model. If two or more
models yielded identical R-squares in step 3, the most parsimonious
model (fewest independent variables) was selected.

Summer 1998, Volume 18, Number 3 117

Downloaded from otj.sagepub.com at UNIV OF PENNSYLVANIA on June 27, 2015


Table 7
Regression Models of SIPT Factors With Reading as the
Dependent Variable
Independent Parameter Standard Partial
Variables Estimate Error r-square Tvalue P
1984 Using Independent Variables in 1984
df = 3, P = 0.000l)
(11 = 66, R-square = 0.50, F = 20.32,
Praxis
in 1984 8.34 1.14 0.47 7.34 0.0001
Visual Perception
in 1984 7.51 1.52 0.28 4.95 0.0001
Soma tosensory
in 1984 2.62 1.24 0.07 2.11 0.0385
Reading in 1988 Using Independent Variables in 1984
(11 = 67, R-square = .56, F = 40.98, df = 2, P = 0.000l)
Praxis
in 1984 1Un 1.36 0.50 8.07 0.0001
Visual Perception
in 1984 12.93 1.86 0.43 6.94 0.000l
Reading in 1988 Using Independent Variables in 1988
(11 = 67,R-square = .49,F =20.47, df= 3,P = O.OOOl)
Praxis
in 1988 11.39 1.58 0.45 7.23 0.0001
Visual Perception
in 1988 6.43 2.09 0.13 3.08 0.0031
Somatosensory
in 1988 4.86 2.06 0.08 2.36 0.0216

Note. Best regression models were selecied usino all possible combinations of the three SIPTti,ctors.
Witlzin euch model, partial r-suuuretor each independent variablecontrolsfor the linear effEcts afall other
independent variables in the model.

Sensory Integration and Praxis Test Factors as Predictors

For each of the dependent variables (reading and arithmetic),


multiple regression models were first generated using only the SIPT
factors as independent variables. The question here was: how well
can sensory integrative factors by themselves account for variation
in reading and arithmetic achievement?
The selected models can be found in Tables 7 and 8. For each
dependent variable, three models are shown representing the best
combination of SIPT factors using data collected at different
points in time: 0) independent and dependent variables concur-
rently in 1984, (2) independent variables in 1984 and dependent
variable in 1988, and (3) independent and dependent variables
concurrently in 1988.

118 The Occupational Therapy Journal of Research

Downloaded from otj.sagepub.com at UNIV OF PENNSYLVANIA on June 27, 2015


Table 8
Regression Models of SIPT Factors With Arithmetic as the
Dependent Variable
Independent Parameter Standard Partial
Variables Estimate Error r-square T value P
Arithmetic in 1984 Using Independent Variables in 1984
(n = 66, R-square = 0.69, F = 46.35,df= 3, P = 0.0001)
Praxis in 1984 10.51 0.99 0.64 10.60 0.0001
Visual Perception
in 1984 10.41 1.32 0.50 7.86 0.0001
Somatosensory
in 1984 4.63 1.08 0.23 4.28 0.0001
Arithmetic in 1988 Using Independent Variables in 1984
(n = 67, R-square = 0.76, F = 66.77, df = 3, P = 0.0001)
Praxis in 1984 10.29 0.81 0.72 12.65 0.0001
Visual Perception
in 1984 11.40 1.09 0.63 10.47 0.0001
Somatosensory
in 1984 2.38 0.89 0.10 2.69 0.0091
Arithmetic in 1988 Using Independent Variables in 1988
(n = 67, R-square = 0.56, F = 40.85, df = 3, P = 0.0001)
Praxis in 1988 8.71 1.11 0.49 7.83 0.0001
Visual Perception
in 1988 6.99 1.51 0.25 4.63 0.0001

Note. Best regression modelswere selected using all possible combinations of the threeSIPT factors.
Within eachmodel, partialr-square for eachindependentvariable controlsfor the lineareffects of all other
independent variables in the model.

Results showed that for reading (see Table 7), all three SIPT fac-
tors were significantly related to concurrent achievement at
younger and at older ages, but only two factors (Praxis and Visual
Perception) at younger ages were related to achievement 4 years
later. For arithmetic (see Table 8), the three factors at younger ages
were significantly related to achievement concurrently and across
the 4-year span of the study. Only two factors (again, Praxis and
Visual Perception) were significant in relation to concurrent arith-
metic achievement at older ages.
The SIPT factors consistently accounted for more variance in
arithmetic than reading scores. For example, the SIPT factors at
younger ages accounted for 76% of the variance in arithmetic at
older ages, whereas they accounted for only 56% of the variance in
reading at older ages.
The results of these analyses supported hypotheses 1 and 2. The
sensory integrative factors significantly accounted for the variance
in achievement both concurrently at younger ages and predictively

Summer 1998, Volume 18, Number 3 119

Downloaded from otj.sagepub.com at UNIV OF PENNSYLVANIA on June 27, 2015


over a 4-year period. Hypothesis 3 was not supported because
Praxis and Visual Perception at older ages continued to be signifi-
cantly related to achievement. However, these results did not take
into consideration the variance that the SIPT factors shared with
other variables, such as intellectual ability.

Sensory Integration and Praxis Test Factors Controlled by Other


Variables

To evaluate the contribution of the sensory integrative factors to


achievement when the interrelated influences of other variables
were considered, competing multiple regression models were gen-
erated using additional independent variables. These included the
K-ABC Mental Processing Composite and socioeconomic status.
Ethnicity and gender were also entered using dummy coding, but
were not statistically significant in any of the models. The question
of concern was: When the effects of other variables are statistical-
ly controlled, do SIPT factors significantly account for the variance
in reading and arithmetic achievement? This question can be
addressed because multiple regression "holds constant" the effects
of independent variables on one another (Lewis-Beck, 1980).
Results for reading achievement are shown in Table 9. Contrary
to expectation, the sensory integrative factors at younger ages were
not significantly related to concurrent reading achievement once
the influences of intellectual ability and socioeconomic status were
partialed out. Hypothesis 1, therefore, was not supported.
However, Praxis and Visual Perception at younger ages were sig-
nificantly related to reading 4 years later when statistically control-
ling for the influences of intellectual ability and socioeconomic sta-
tus. This was in keeping with the prediction of Hypothesis 2.
Hypothesis 3, which stated that SIPT factors at older ages would
not be concurrently related to achievement, was only partially sup-
ported. Praxis was the only SIPT factor significantly related to
reading at older ages, when the influences of socioeconomic status
and intellectual ability were held constant.
Table 10 presents results for arithmetic achievement. The best
model accounting for variance in arithmetic performance at
younger ages consisted of the three SIPT variables. The Mental
Processing Composite, when included with the three SIPT factors,
did not contribute significantly to the equation; however, a close
competitor for optimal solution was the combination of the Mental
Processing Composite with Praxis and Visual Perception (R-square
= 0.67). Thus, Hypothesis 1 was strongly supported with respect to

120 The Occupational Therapy Journal of Research

Downloaded from otj.sagepub.com at UNIV OF PENNSYLVANIA on June 27, 2015


Table 9
Best Regression Models With Reading as the Dependent
Variable
Independent Parameter Standard Partial
Variables Estimate Error r-square T value P
Reading in 1984 Using Independent Variables in 1984
(n = 66, R-square = 0.59, F = 45.19, df= 2, P = 0.0001)
MPC in 1984 0.69 0.08 0.55 8.86 0.0001
Socioeconomic
status -5.54 1.57 0.17 -3.54 0.0008
Reading in 1988 Using Independent Variables in 1984
(n = 67, R-square = 0.67, F = 30.44, df = 4, P = 0.0001)
Socioeconomic
status -7.34 2.03 0.18 -3.61 0.0006
MPC in 1984 0.50 0.16 0.13 3.07 0.0032
Praxis in 1984 5.97 1.96 0.13 3.05 0.0034
Visual Perception
in 1984 6.79 2.40 0.12 2.83 0.0063
Reading in 1988 Using Independent Variables in 1988
(n = 67, R-square = 0.65, F = 39.51, df = 3, P = 0.0001)
MPC in 1988 0.70 0.12 0.35 5.87 0.0001
Socioeconomic
status -7.49 1.93 0.19 -3.88 0.0003
Praxis in 1988 4.77 1.56 0.13 3.05 0.0034

Note. Best regression models were selected using all possible combinations ofthe following independent
variables: Praxis, visual perception, somatosensory, mental processing composite (MPe), socioeconomic
status, ethnicttv, sex. Partial r-square for each independent variable controlsfor the linear effects ofall
other independent variables in the model.

arithmetic achievement. Hypothesis 2 was also supported, in that


Praxis and Visual Perception at younger ages were significantly
related to Arithmetic at older ages, when controlling for shared
variance with the Mental Processing Composite. Hypothesis 3 was
only partially supported. Instead of disappearing completely, one
SIPT factor (Praxis) was retained in the best model as significantly
related to Arithmetic at older ages, given statistical controls for
intellectual ability. Socioeconomic status was not significantly
related to Arithmetic when included in the equations.

Discussion

The Satz hypothesis was supported by the results for arithmetic,


but not reading. In the analyses of arithmetic achievement, the
SIPT factors clearly diminished in importance with age, as expect-

Summer 1998, Volume 18, Number 3 121

Downloaded from otj.sagepub.com at UNIV OF PENNSYLVANIA on June 27, 2015


Table 10
Best Regression Models With Arithmetic as the Dependent
Variable
Independent Parameter Standard Partial
Variables Estimate Error r-square T value P
Arithmetic in 1984 Using Independent Variables in 1984
(11 = 66, R-square = 0.69, F = 46.35, df = 3, P = 0.00(1)
Praxis
in 1984 10.51 .99 0.64 10.60 0.0001
Visual Perception
in 1984 10.41 1.32 0.50 7.86 0.0001
Somatosensory
in 1984 4.63 1.08 0.23 4.28 0.0001
Arithmetic in 1988 Using Independent Variables in 1984
(11 = 67, R-square = 0.77, F = 69.80, df= 3, P = 0.00(1)
Praxis
in 1984 7.08 1.21 0.36 5.86 0.0001
Visual Perception
in 1984 8.51 1.47 0.35 5.80 0.0001
MPC in 1984 0.31 0.10 0.13 3.09 0.0031
Arithmetic in 1988 Using Independent Variables in 1988
(11 = 67, R-square = 0.72, F = 81.58, df= 2, P = 0.00(1)
MPC in 1988 0.69 0.08 0.52 8.32 0.0001
Praxis in 1988 3.36 1.09 0.13 3.08 0.0031

Note. Best regression models were selected using all possible combinations if the following independent
variables: Praxis, visual perception, eomatosensorv, mental processing composite (MPC), socioeconomic
status, eihnicitv, sex. Partial r-squarefor each independent uariuble controls for the linear ejftcts if all
other independent variables in the model.

ed. Such a picture was not evident in the analyses of reading


achievement. The SIPT variables were not related to concurrent
reading achievement at younger ages when statistical controls for
intelligence and socioeconomic status were applied, but did
demonstrate some degree of relationship to reading achievement
later on, given the same statistical controls.
Perhaps the nonverbal sensory integrative abilities are impor-
tant in acquiring basic number and arithmetic concepts, but are rel-
atively inconsequential in the early acquisition of elementary read-
ing skills, which may be more dependent on linguistic develop-
ment. For both reading and arithmetic, there was a strong relation-
ship longitudinally between sensory integrative performance in
the early grades of school and achievement 4 years later, suggest-
ing a cumulative effect over time. The apparent decline in impor-
tance of sensory integrative variables to arithmetic at older ages
may be due to the ascendancy of more complex cognitive opera-

122 The Occupational Therapy Journal of Research

Downloaded from otj.sagepub.com at UNIV OF PENNSYLVANIA on June 27, 2015


tions needed for mathematical skills beyond a basic level.
Results support the premise that sensory integrative functioning
is related to academic achievement. Ayres' perspective was that sen-
sory integration provides a primitive foundation on which the high-
er cognitive and behavioral skills necessary for academic achieve-
ment are scaffolded (1972, 1985, 1989). According to this view, one
would expect sensory integrative measures to be related to both
intelligence and achievement. The results of this study generally
provide support for this perspective. Intercorrelations among mea-
sures of sensory integration, intellectual ability, and achievement
were moderately high. In the multiple regression models, sensory
integrative factors shared a substantial amount of variance with
measures of intellectual ability, yet often were significantly related
to achievement given statistical controls for this shared variance.
A difficulty with this interpretation is that the same results could
be explained from alternative viewpoints. It may be that the senso-
ry integrative measures are correlated with some variable, not mea-
sured in this study, which directly impacts on both intelligence and
achievement. If this were the case, then the sensory integrative mea-
sures were correlated with achievement because they acted as prox-
ies for another variable, not because the constructs they represent
contribute to achievement in their own right. For example, perfor-
mance on the sensory integration tests may be sensitive to use of
effective cognitive strategies, which in turn affects scores on both
intelligence and achievement tests.
An unanticipated finding was the strength of the relationship of
sensory integrative measures to arithmetic achievement. In every
regression analysis of arithmetic, at least one sensory integrative
variable was significant; this was not the case for reading. For each
analysis of arithmetic at younger ages, a combination of the three
sensory integrative factors produced the best fit with the data.
Given the considerable degree of heterogeneity in the participants
of the study, this result was all the more compelling. Had such a
relationship existed only in a small subgroup of participants, it
would have been obscured by the presence of the many other par-
ticipants for whom sensory integrative performance was irrelevant
to skill in arithmetic.
Why sensory integrative variables, Praxis in particular, would be
closely aligned with arithmetic achievement, especially in the pri-
mary grades, is not clear. Perhaps a core characteristic of the sensory
integrative measures is the tapping of spatial or directional concepts
that are also key to the acquisition of elementary arithmetic skills.
The measures of praxis and intelligence shared a substantial
amount of variance, making it difficult to tease out the relative con-

Summer 1998, Volume 18, Number 3 123

Downloaded from otj.sagepub.com at UNIV OF PENNSYLVANIA on June 27, 2015


tributions of each to achievement. This initially may seem surpris-
ing, because the tasks that comprise the Praxis factor do not resem-
ble tasks traditionally used in intelligence or achievement tests.
Ayres (1985), however, conceptualized Praxis as "a kind of 'intelli-
gence'" (p. 7) of action that contributes to human competence. The
moderately high correlations suggest that praxis shares a common
cognitive element with intelligence and achievement, yet at the
same time has distinctive qualities that are unique. Attempts to sep-
arate the effects of intelligence from Praxis may ultimately be
impossible, or inappropriate, because Praxis is conceived of as
essentially cognitive, an intelligence of doing. It may be that Praxis
makes a key contribution to a child's ability to perform well on tests
of general intelligence.
The omission of language-related measures imposed the greatest
limitations on this study. No specific linguistic measures were
included to permit, for example, analyses of phonemic, syntactic, or
pragmatic development. Verbal ability may well have been a critical
variable in predicting arithmetic achievement, especially in view of
the factor analytic studies identifying a verbal factor as a basic com-
ponent of mathematical ability (Barakat, 1951;Wrigley, 1958).
The arithmetic measure itself was limited, as it consisted of oral-
ly administered problems. Had an arithmetic test requiring paper-
and-pencil problems been used, an even stronger relationship
between arithmetic and visual perception might have been found.
Some of the findings of the present study contradict those of pre-
vious research that reported no relationship, or a relatively insignif-
icant one, between perceptual-motor abilities and achievement
(e.g., Larsen et al., 1976). Complexity of the tests used to measure
perceptual-motor functions may be critical in determining research
outcomes. The cognitively complex Finger Localization test used by
Satz and colleagues (Satz et al., 1978), for example, was a strongly
predictive somatosensory test with respect to achievement in ele-
mentary school children, whereas in the present study, the
somatosensory factor tapped very simple functions and was rela-
tively inconsequential as a correlate of achievement.
Sensitivity of scoring systems used in tests is another possibly crit-
ical variable. Most of the SIPTtests utilize quite detailed, fine-grained
scoring systems that maximize the likelihood of detecting differences
in performance across participants. Many other perceptual tests, in
contrast, rely on coarse-grained scoring methods that are faster and
easier for the investigator, but may lack precision in differentiating
among participants. Selection of tests for operationalization of per-
ceptual and perceptual-motor constructs, consequently, should be a
matter of careful deliberation for researchers as well as clinicians.

124 The Occupational Therapy Journal of Research

Downloaded from otj.sagepub.com at UNIV OF PENNSYLVANIA on June 27, 2015


Results of this study indicate that age of participants, type of
achievement, and domain of perceptual-motor ability are important
variables that influence whether or not a relationship is found
between perceptual-motor ability and achievement. Blanket state-
ments regarding the pertinence or irrelevance of perceptual-motor
abilities to achievement, therefore, may lead to premature conclu-
sions that fail to fully take into account the differential develop-
mental characteristics of perceptual-motor constructs.

Acknowledgments

This study was conducted in partial fulfillment of the require-


ments for the Doctor of Philosophy degree at the University of
California, Los Angeles. lt was supported in part by a doctoral fel-
lowship from the American Occupational Therapy Foundation and
by a grant from the Bureau of Maternal and Child Health (Grant #
MCJ 009048). Portions of this manuscript were presented at the 1990
annual conference of the Occupational Therapy Association of
California, and at the 1991 conference of the American Occupational
Therapy Association.
The contributions of my dissertation committee members, partic-
ularly Barbara Hecht, Barbara Keogh, and Paul Satz, are gratefully
acknowledged. Also much appreciated for their assistance are
George Huba and Diana Marr who were at Western Psychological
Services at the time of the study, and Heidi Estep and Colleen
Torgerson of the Fresno Unified School District. A special note of
gratitude is due to the many capable researchers and assistants who
contributed to this work, particularly Florence Clark, Sandra Eaton,
Shan-Pin Fanchiang, Zoe Mailloux, and Carolyn Snyder, and to the
parents and children who participated.

References

Afifi, A.A, & Clark, V. (1984). Computer-aided multivariate analysis.


New York, NY: Van Nostrand Reinhold.
Ayres, AJ. (965). Patterns of perceptual-motor dysfunction in chil-
dren: A factor analytic study. Perceptual and Motor Skills, 20, 335-
368.
Ayres, AJ. (1972). Sensory integration and learning disorders. Los
Angeles, CA: Western Psychological Services.
Ayres, AJ. (979). Sensory integration and the child. Los Angeles, CA:

Summer 1998, Volume 18, Number 3 125

Downloaded from otj.sagepub.com at UNIV OF PENNSYLVANIA on June 27, 2015


Western Psychological Services.
Ayres, AJ. (1985). Developmental dyspraxia and adult-onset apraxia.
Torrance, CA: Sensory Integration International. .
Ayres, AJ. (1989). Sensory integration and Praxis tests manual. Los
Angeles, CA: Western Psychological Services.
Barakat, M.K. (1951). A factorial study of mathematical abilities.
British Journal of Psychology, Statistical Section, 4, 137-156.
Cermak, S.A, & Ayres, A.J. (1984). Crossing the body midline in
learning-disabled and normal children. American Journal of
Occupational Therapy, 38, 35-39.
Chatfield, C, & Collins, AJ. (1980). Introduction to multivariate analy-
sis. London: Chapman & Hall.
Coles, G.5. (1978). The learning-disabilities test battery: Empirical
and social issues. Harvard Educational Review, 48, 313-340.
Coles, G.5. (1987). The learning mystique-A critical look at 'learning
disabilities.' New York, NY: Fawcett Columbine.
Council for Learning Disabilities. (1986). Measurement and training
of perceptual and perceptual-motor functions. A position state-
ment by the Board of Trustees of the Council for Learning
Disabilities. Learning Disability Quarterly, 9, 247.
Cruikshank, WM. (1985). Learning disabilities: A series of chal-
lenges. Learning Disabilities Focus, 1, 5-8.
Freides, D., Barbati, J., Van Kampen-Horowitz, L.H., Sprehn, G.,
Iversen, C, Silver, J.R., & Woodward, K (1980). Blind evaluation
of body reflexes and motor skills in learning disability. Journal of
Autism and Developmental Disorders, 10, 159-171.
Frostig, M. (1972). Visual perception, integrative function, and aca-
demic learning. Journal of Learning Disabilities, S, 1-15.
Hollingshead, A.B. (1957). Ttoo-factor index of social position.
Unpublished manuscript, New Haven, CT: Yale University.
Kamphaus, KW, & Reynolds, CR. (1988). Clinical and research appli-
cations of the K-ABC. Circle Pines, MN: American Guidance
Service.
Kaufman, AS., & Kaufman, N.L. (1983). Kaufman Assessment Battery
for Children: Interpretive manual. Circle Pines, MN: American
Guidance Service.
Kavale, K. (1982). Meta-analysis of the relationship between visual
perceptual skills and reading achievement. Journal of Learning
Disabilities, 1S, 42-51.
Kim, J., & Mueller, C.W (1978). Factor analysis: Statistical methods and

126 The Occupational Therapy Journal of Research

Downloaded from otj.sagepub.com at UNIV OF PENNSYLVANIA on June 27, 2015


practical issues. Sage University Paper Series on Quantitative
Applications in the Social Sciences, 07-014. Beverly Hills, CA:
Sage Publications.
Kinnealey, M. (1989). Tactile functions in learning-disabled and nor-
mal children: Reliability and validity considerations.
Occupational Therapy Journal of Research, 9, 3-15.
Kirk, S.A., & Gallagher, J.]. (1983). Educating exceptional children (4th
ed.). Boston, MA: Houghton Mifflin.
Larsen, S.c., Rogers, D., & Sowell, V. (1976). The use of selected per-
ceptual tests in differentiating between normal and learning dis-
abled children. Journal of Learning Disabilities, 9,85-90.
Lewis-Beck, MS. (1980). Applied regression: An introduction. Sage
University Paper Series on Quantitative Applications in the
Social Sciences, 07-022. Beverly Hills, CA: Sage Publications.
Manly, B.EJ. (1986). Multivariate statistical methods: A primer. London:
Chapman & Hall.
Morrison, D.C. (1986). Neurobehavioral dysfunction and learning
disabilities in children. In S.J. Ceci (Ed.), Handbook of cognitive,
social, and neuropsychological aspects of learning disabilities (Vol. 1,
pp. 475-491). Hillsdale, NJ: Lawrence Erlbaum.
Sall, J. (1981). SAS regression applications. SAS Technical Report A-
102. Cary, NC: SAS Institute.
Satz, P.,Taylor, H.G., Friel, J., & Fletcher, J.M. (1978). Some develop-
mental and predictive precursors of reading disabilities: A six
year followup. In A. L. Benton & D. Pearl (Eds.), Dyslexia: An
appraisal of current knowledge (pp. 315-347). New York, NY: Oxford
University Press.
Servello, M.B.(1982). Vestibular-based functions and behavior prob-
lems of children in special education and regular classes.
Perceptual and Motor Skills, 55, 1289-1290.
Walker, H.M. (1983). Walker Problem Behavior Identification
Checklist-Manual Revised 1983. Los Angeles, CA: Western
Psychological Services.
Wrigley,J. (1958). The factorial nature of ability in elementary math-
ematics. British Journal ofEducational Psychology, 28, 61-78.

Summer 1998, Volume 18, Number 3 127

Downloaded from otj.sagepub.com at UNIV OF PENNSYLVANIA on June 27, 2015

You might also like