Sensory Integration and Learning Disabilities: Ayres' Factor Analyses Reappraised Robert A. Cummins Between 1965 and 1987 Ayres published eight papers that contain among them 10 multivariate analyses that bear upon her conception of sensory integration. These analyses purport to have identified factors that emerge from the scores of children with learning disabilities but not from the scores of academically nondisabled children. A reappraisal of these analyses in combination finds no support for this claim. As a consequence, these data provide no validity for either the diagnostic procedures or the remedial programs for children with learning disabilities that have been derived from this work.
A
yres has operationalized her concept of sensory integration by conducting multivariate studies on the perceptual-motor performance of children with learning disabilities (Cermak, 1988). Of particular relevance is a series of factor analyses from which Ayres claimed to have identified perceptual-motor factors that emerge from the scores of children with learning disabilities but not from the scores of normally learning children. The perceptual-motor content of these factors has been used to provide both the format and the rationale for diagnostic and remedial programs (e.g., Ayres, 1968, 1975, 1979). Of concern, however, is the diversity of opinion among reviewers of Ayres' studies as to the precise nature and nomenclature of the factors she discovered. Although there seems to be a majority opinion in favor of four factors (i.e., Praxis, Form and Space Perception, Tactile Defensiveness, and Bilateral Integration), not all reviewers agree, and several other factor names have been suggested. Some examples are • Tactile, kinesthetic, and visual perception dysfunction (Abbie, 1974) • Auditory language problems (Wilson, 1975) • Praxis and tactile functions (Kimball, 1977) • Vestibular and bilateral integration (Clark & Shuer, 1978) • Left hemisphere dysfunction and right hemisphere dysfunction (Clark & Shuer, 1978) • Unilateral cerebral dysfunction (Grim-
wood & Rutherford, 1980) • Postural and bilateral integration (Ottenbacher & Short, 1985) The most comprehensive review of Ayres' factor analytic work was published by Clark, Mailloux, and Parham (1985). Those authors examined the emergent factors and concluded as follows (P. 379): • Apraxia: "Emerged in nearly all the studies in which dysfunctional children made up the sample." • Deficits inform and space perception: "Were more varied, and the configuration and patterns of the clusters suggested that there were at least two kinds of visual perceptual problems. The first seemed to be related to somatosensory processing, the second to functions associated with the right hemisphere." • Deficits in postural and bilateral integration: "Emerged relatively consistently and in later studies were linked to vestibular dysfunction." • Auditory language dysfunctions: "Seemed to bifurcate in the same way that visual spatial perception had. One type seemed to be related to the left hemisphere and did not seem to be related to sensory integrative processes. The second type was found in association with vestibular dysfunction and poor bilateral integration." • Tactile defensiveness: "In nearly every study a strong association between tactile defensiveness and hyperactivity was found."
From this analysis the authors generated yet another description of typologies for sensory integrative dysfunction derived from Ayres' work: • Vestibular bilateral integration dysfunction • Dyspraxia • Left hemisphere dysfunction • Right hemisphere dysfunction • Generalized dysfunction The reason for this diversity of views is that Ayres has not attempted to crossvalidate her factors. Each analysis has been performed on a new combination of variables and so has produced a different set of factors. One consequence of this is that factor labeling becomes a critical issue. Since none are identical in composition, it might be assumed that all of the factors would have been given unique labels, but that is not so. Whereas some factor labels have certainly been unique to individual studies and have contributed to the diversity of views mentioned previously, other labels have been more commonly employed. In the absence of cross-validated factors between studies, this use of common factor labels is of special interest. It implies that factors with a similar and distinctive variable composition have emerged from different studies. Indeed, the previously cited reviewers of Ayres' studies appear to have made just that assumption when searching for factor typologies. It seems timely, therefore, to reexamine Ayres' data and to question these conclusions, which assume a simple relationship between factor labels and factor composition. This will be achieved by identifying the groups of similarly named factors and examining each group to determine (a) their distinctive, common content, (b) whether their common content distinguishes them from factors that have been differently labeled, and (c) whether their common content distinguishes between scores derived from children with learning disabilities and nondisabled children.
An Overview of Ayres' Studies Between 1965 and 1987 Ayres published eight papers that contain among Journal of Learning Disabilities
160 Downloaded from ldx.sagepub.com at MCMASTER UNIV LIBRARY on March 2, 2015
them 10 multivariate analyses. All are concerned with the relationships between sensory and motor skills, with academic skill variables sometimes being included as well. The broad characteristics of these studies are outlined in Table 1. There are only nine columns because two different analyses were conducted on the scores of the same children in 1972 (Ayres, 1972b). All children were aged between 4-0 and 10-0 years, were of normal intelligence (where this was reported), and comprised a mixture of males and females (where reported). Of the 10 studies, 9 were factor analyses and 1 a multiple regression. The latter study (1971) will be considered separately, as will the most recent factor analysis (Ayres, Mailloux, & Wendler, 1987). The other eight analyses will be considered as a group, with a view to comparing their factor structures. The factor analyses differed from one another on a number of dimensions as follows: 1. Nondisabled/Learning disabled. Three of the studies (1965, 1966a, 1966c) involved nondisabled children. The 1965 sample was "chosen from public and private schools and child care centers on the basis of parental occupation so as to represent proportionately the working population of the United States" (p. 339). In 1966c, all were adopted children living in middle socioeconomic class families. The 1966a children, however, were a bit different. Of the 92 subjects, Ayres reported that "approximately 10 per cent could be considered to have possible central nervous system dysfunction" (p. 68). No further information on these children's specific problems was given, but it seems likely that their disorders were minimal. In addition, three children had mild cerebral palsy. Since such a mixture of children would not be uncommon within any unselected group of regular school children, for the purposes of future comparisons the children from 1966a will also be considered nondisabled. The remaining five studies all involved children with academic disabilities who had been classified as such using a variety of criteria. For future reference, the studies involving children who are nondisabled will be designated (N) and those involving children who are academically disabled will be designated (D).
TABLE 1 Subject Characteristics in Ayres' Multivariate Analyses Studies
1965
1965
1966a
1966c
1969
1971
1972b
1977
1987
Age range
5-10 8-0 71139
5-8 7-11 ?
4-0 8-0 ?
4-0 8-0 ?
6-1 9-10 X 93
5-10 10-0 70132
7-6 10-0 70132
6-0 10-0 ?
4-0 9-11 ?
-
7 93
-
36
148
148
128
172
64
-
IQ Academic Disabled (N) Nondisabled (N) Cerebral Palsy (N) Males (%) Females (%) Variables (N) Loading cutoff for factor definition
100
-
-
50
-
-
-
-
-
-
-
-
-
-
-
69 31 36
70 30 36
3 ? ? 19
? ? 19
81 19 64
74 26 48
74 26 46
80 20 19
65 35 19
0.3
0.3
None used
None used
0.4
-
0.2
0.3
None used
Note. ? = not reported.
2. Raw scores/Ipsative scores. While six of the factor analyses have used the conventional raw scores derived from each test, the 1969 analysis and one of the 1972b analyses used ipsative scores. This process led to the emergence of quite different factor structures, as will be seen from the 1972b study, which performed both raw score and ipsative score analyses on the same set of data. For future reference, the studies involving raw scores will be designated as (R), while those using ipsative scores will be designated (I). 3. Loading cut-off. As shown in Table 1, the studies differed somewhat in the minimum size of the variable loading that was used to define the content of factors. For comparative purposes, all studies will be standardized such that a minimum loading of 0.30 is required for any variable to be considered part of a factor. 4. Variables. Of the 113 different variables used in these studies, only 8 were common to all analyses, while 50 were used on just one occasion each. A description of the variable names and their test protocols can be found in Ayres (1965, 1966a, 1969).
EXAMINATION OF THE FACTOR LABELS Ayres' first two factor analyses (1965) set the conceptual stage for all of the analyses to follow. As shown in Table 1,
this report contained one analysis that involved children with "suspected dysfunction" and another analysis on nondisabled children. From the former group, Ayres located five factors that "were interpretable as hypothesized behavioral correlates of patterns of neurological dysfunction" (p. 345). From the nondisabled children, on the other hand, she found that the emergent factors "on the whole were not comparable to the major factors emerging from the dysfunction group." This apparent difference between factor structures from the two groups led Ayres to hypothesize that in the case of the dysfunctional group "these factors were not due to normal developmental processes but to underlying deficits in specific mechanisms of integration, resulting in symptom complexes" (p. 465). Over the next 12 years Ayres conducted six more analyses that sought to validate the emergence of such "symptom complexes" from the scores of children with learning disabilities. The labels or descriptions that Ayres gave to the factors from all of these analyses are shown in Table 2. Where a simple label has been attached to a factor, such as "Praxis," this label has been used. Where a factor has not been named, but only described, the major variables involved in the description have been used as the title. What is immediately clear from this listing is that relatively few factor titles or descriptors have emerged on more than one occasion. To summarize:
Volume 24, Number 3, March 1991 Downloaded from ldx.sagepub.com at MCMASTER UNIV LIBRARY on March 2, 2015
161
TABLE 2 Factor Labels or Descriptions Labels/ Descriptions Praxis (or apraxia) Form and space perception Tactile defensiveness Bilateral integration Figure-ground discrimination General perceptual-motor and cognitive ability Body balance Visual perception Visual motor Interaction of sides of body with tactile perception Auditory, language, and sequencing Postural and bilateral integration Left-side coordination, posture, and bilateral integration Academic achievement and sensory perception Auditory/visual association and horizontal perception Auditory, language, and intelligence Postural and ocular mechanisms Reading, spelling, and intelligence Eye-hand dominance and left-eye dominance Left-hand vs. righthand coordination Auditory-language problems Kinesthesia Somatosensory Use of one hand in contralateral space Auditory-perception Uninterpreted factor
1965 DR
1965 NR
1966a NR
1966c NR
1969 Dl
X
X X
1972b DR
1972b Dl
1977 DR
X
X
X
X
X
X
X
X
X X
X X
EXAMINATION OF INDIVIDUAL FACTORS
X X X
Using Table 2 as a guide, each similarly labeled group of factors will now be examined to determine the nature of their common content.
X X X
Praxis
X X X X X
X
X
X
X
X X X
X
X X X X X
X
Note. D = Learning disabled; N = Normal; R = Raw scores; I == Ipsative scores.
• One factor label ("Praxis") has emerged five times. • Three factors have emerged three times. • Three factors have emerged on two occasions. • The remaining 19 factor labels have been used by Ayres on only one occasion each. Despite this lack of consistency, in 1972a Ayres summarized her findings as follows: "From the author's statistical analyses . . . five factors emerged with sufficient frequency and similarity of operative
It is interesting to relate these labels to those actually used to describe the emergent factors (see Table 2). With the possible exception of "Praxis," it seems rather extravagant to claim their general emergence. "Auditory language problems" was used as a label on only two occasions, whereas on no occasion had any factor been designated as a disorder of postural, ocular, and bilateral integration.
parameters to be considered functionally related aspects of human behavior. It is hypothesized that they reflect neural systems in which disorder has been found in children with learning problems" (p. 94). The terms Ayres (1972a) used to designate the five factors were: 1. Disorder in postural, ocular, and bilateral integration 2. Apraxia 3. Disorder in form and space perception 4. Auditory-language problems 5. Tactile defensiveness
162
According to Meulders (1983), the concept of praxia derives from apraxia, the latter term being defined by Liepmann (1908) as the impossibility of performing intentionally certain movements, or sequences of movements, in spite of the integrity of the primary motor pathways and the conservation of most of the elementary or automatic motor functions. In the context of sensory integration theory, the term praxia is used to describe the ability to plan motor acts. The deficit is usually referred to as developmental dyspraxia, which Ayres et al. (1987) describe as being caused by disordered sensory integration resulting in poor development of the ability to motor plan or to sense what sequence of movements will be needed in order to accomplish a motor task. According to Vezie (1975) it also implies that while specific motor skills may be learned through repetition, the growth of generalized ability to motor plan is lacking. Ayres has claimed that this factor tends to emerge more readily from the data of children with LD than from nondisabled children—an important finding, if true. Unfortunately, evaluating this claim on the basis of Ayres' data is not easy, for reasons that can be seen in Table 3. There, under "A," are listed the compositions of all five factors Ayres has labeled as "Praxis." (Note that a key immediately follows this table.) It can be seen that of the 32 variables included at one time or another, only 5 have been Journal of Learning Disabilities
Downloaded from ldx.sagepub.com at MCMASTER UNIV LIBRARY on March 2, 2015
included on three or more occasions, 17 have been included on two or more occasions, and the remaining 15 have appeared only once.
Definition of the Praxis Factor
TABLE 3 The Variable Composition off Praxis A Variables
1965 1966a DR NR
B
1972b 1972b> 1977 1965 DR Dl DR Deffn NR
C 1966c 1969 NR Dl
Wire grommet
Even a casual glance at the five facmanipulation 1 X X 7.5 X X X . tors under "A" will indicate their very S.Calif.Mot.Acc. Test X X 2 X 7 3 X . heterogeneous composition. Ayres has Posture duplication 1 1 3.5 2.5 2 1 12 .. Finger identification 3.5 7 7.5 3.5 •• proposed a content definition for these 5 Graphesthesia 5 1 . praxis factors, which will be discussed Single stimulus later. First, the factors will be examined localization 4 6 3.5 8.5 11 .. for their similarity of content. Stereognosis 7 X 3.5 6 1 . Table 3 permits the formulation of an Eye-motor coordination X X X X 8.5 6 X X Two-point discrimination 8.5 X X X X X operational definition based on common 10 6 9 3.5 3 •• elements: Praxis may be defined in terms Face-hand test X X X X X Time and rhythm X 11 of those variables that have loaded 0.30 X X X String winding X X X 12 or higher, on more than one occasion, 4 4 Visual figure-ground 13 13.5 • within a factor named by Ayres as X X X X X X 14 Superimposed figures X X X 15 "Praxis." In terms of the data in Table Eye pursuit X X X X X Vertical perception 16 3 (A), this definition has been used to X X 6 Design copying 17.5 • provide two different definitional struc17.5 9 8.5 Kinesthetic memory • tures: a "hard" definition, consisting of X X 1 X X 15 Space relations only those variables that have appeared X X X X 12 5 Form constancy 2 X 5 2 5 X Bilateral motor control on three or more occasions (marked •• •• Standing balance-eyes under column B), and a "soft" definition, X 5 7.5 10 open • consisting of variables that have been inX 11 2.5 Tactile defensiveness . cluded at least twice (designated • in colX 3 13.5 Position in space • umn B). Each of these two operational X 1 16 Hyperactivity • X 17 Right-left discrimination definitions has then been applied back to Hands crossing body the five praxis factors to gauge the ex4 4 2 X midline . tent of their internal conformity, both in X X X 7.5 X Muscle co-contraction terms of the 5 "hard" and the 17 "soft" Standing b a l a n c e definitional variables indicated under X X 10 eyes closed 11 Ayres space test "B." X X X 5.5 X ITPA sound blending Unfortunately, this process is compliX X X 5.5 X Muscle tone cated by the fact that each of the studies Note. A = those factors labeled by Ayres as "Praxis"; B = hard (••) or soft (•) definitional to be compared failed to include some variables; C = factors not labeled by Ayres as "Praxis," but being the factor in that study of the definitional variables. For examwith a variable structure most conforming to the definitional variables; D = data from disple, the 1965 study did not include the abled children; N = data from nondisabled children; R = raw data; I == ipsatized data variable "Bilateral motor control" and so The variable list comprises all variables loading 0.3 or more on the Praxis factors under it was not available to form part of the "A." It is not a complete list of variables comprising the factors under" C." Each blank space in the table indicates that the variable was used irl the study but did not load 0.30 or above Praxis factor in that study. To compenon the factor. Each "x" in the table indicates that the variable was not used in the parsate for such differences, the extent of ticular study, and therefore could not have formed part of the factor. Numbers in the table definitional conformity has been judged indicate the rank order of variable loading s within each factor. The variable numbered " 1 " only on the basis of the definitional had the highest loading, and equally ranked loadings are indicated as the combined rank variables available to each study. Thus, midpoint. the extent of definitional conformity, for both the hard and soft definitions, is expressed as a percentage calculated as: In other words, definitional conformity tween studies, ranging from 40% to is to be judged only on the basis of the 100% for the hard definition and from % conformity = — x ^ definitional variables available to each 27% to 100% for the soft definition. Instudy. terestingly, the only factor to contain P = Number of definitional variables The results of these calculations are 100% conformity with both hard and included within the named factor. shown in the upper-left-hand quadrant soft definitions was derived from the Q = Number of definitional variables of Table 4. As can be seen, there are con- scores of nondisabled children (1966a). included within the study. siderable differences in conformity beThese discrepancies are highlighted by Volume 24, Number 3, March 1991
163 Downloaded from ldx.sagepub.com at MCMASTER UNIV LIBRARY on March 2, 2015
TABLE 4 Conformity Between Actual and Definitional Factor Structures for Praxis Soft definition
Hard definition P
Q
%
P
Q
%
1965 DR 1966a NR Factors Named 1972b DR "Praxis" 1972b Dl 1977 DR
4 5 3 2 4
4 5 5 5 5
100 100 60 40 80
11 14 8 4 5
16 14 15 15 13
69 100 53 27 38
Factors with Other Names
1 4 2
4 5 5
25 80 40
3 9 4
16 14 14
19 64 29
Study
1965 NR 1966c NR 1969 Dl
Ayres' definition Soft Loading %
%
Loading %
68 87 78 75
70 89 27 18 33
40
40 100
20 67 33
31 66
-
-
47
25 34
-
-
Note. P = Number of definitional variables included within the Praxis factor; Q = Number of definitional variables included within the study. % = P x 100 Q 1 Soft Loading % = Loading contribution of the soft definitional variables (A) to the total loading of variables (0.30 or above) on the factor. Ayres Loading % = Loading contribution of Ayres' definitional variables to the total loading of variables (0.30 or above) on the factor.
a comparison with the three studies wherein Ayres did not name a Praxis factor. In each of these studies the factor has been selected that most closely conforms to the operational definition of Praxis. The names that Ayres had given these factors are as follows: 1965-NR General perceptual-motor and cognitive ability 1966c-NR Interaction of sides of body with tactile perception 1969-DI Postural and bilateral integration contrasted with Praxis The extent to which these factor compositions conform with the hard and soft definition criteria is set out in the lower left quadrant of Table 4. And what is clear from these comparisons is that, despite Ayres' labeling procedures, two of these differently named factors corresponded to the ideal Praxis structure just as well as the 1972b-DI factor that had been labeled as Praxis. In fact, the differently named 1966c-NR factor corresponded to the definitional composition better than three of the named Praxis factors. Further data relevant to these comparisons are presented in the column labeled "Soft Loading %" (see Table 4). The basis of these calculations is the numerical loading of each variable onto
the Praxis factor. Loading is a measure of the extent to which each variable contributes to the factor, a high loading indicating a major contribution. The "Soft Loading %" is the contribution of the soft definitional variables to the total variable loadings on the'factor. For example, in 1965-DR, the Praxis factor was composed of 18 variables in all, and their combined numerical loading amounted to 806. The 11 soft definition variables found within this factor structure had a combined numerical loading of 552. The proportion of the total loading contributed by the definitional variables is then: 552 100 Soft Loading % — x - r = 68% oUo
1
In other words, the soft definitional variables made up 68% of the total variable loadings onto this factor. Two further matters require clarification in these calculations: (1) In the ipsative data comprising 1972b-DI, the numerical loadings were summed regardless of sign. (2) Loading percentage could not be calculated for either 1969-DI or 1977-DR. In the former case the variable loadings were not given, while in the latter study "composite" variables were used in the analysis, thereby rendering the loadings noncomparable with those of previous analyses.
164
On the basis of these calculations it is apparent that the factor labeled by Ayres as "Interaction of sides of body with tactile perception" (1966c) has, in fact, a greater claim to be called "Praxis" than any of the factors so named, since the nine Praxis variables made up 100% of the factor loadings. It must again be concluded that Ayres' factor titles and factor compositions are not in accord. The major question, however, is whether the Praxis factor can be used to distinguish between the scores of academically disabled and nondisabled children. In this regard Ayres has claimed that one of the "clearest differences" between the factors derived from nondisabled and dysfunctional children is "the failure for the dimension of praxis . . . to appear as separate factors in the normal group" (1966c, p. 289). The data presented in Table 4 show quite clearly that this is not so. In terms of either the number of Praxis variables or their relative loadings onto the factors, it can be seen that Praxis may emerge either strongly or weakly from the scores of either group. It is interesting to see, indeed, that the two factors carrying the highest Praxis loadings (1966c-NR and 1966a-NR) are both derived from the scores of nondisabled children.
Ayres' Definitions Ayres has provided a verbal description of the Praxis factor composition on a number of different occasions (1963, 1965, 1966a, 1967, 1971, 1972a, 1972b, 1977). In general these descriptions reflect the heterogeneous factor composition displayed in Table 3, with not one description being totally consistent with any other. The two most recent descriptions are as follows. Ayres (1972a) defined the "Syndrome of developmental apraxia" primarily by a deficit in the variable Imitation of Postures. Other, related deficits were stated to be in the Motor Accuracy Test and Bilateral Motor Coordination, Tactile Tests, Kinesthesia, Extraocular Muscle Control, and Behavior Problems. It is difficult to reconcile this definition with Table 3. Of the seven variable types, only two were included on three or more occasions (Posture Duplication and Bilateral Motor Control), three had been Journal of Learning Disabilities
Downloaded from ldx.sagepub.com at MCMASTER UNIV LIBRARY on March 2, 2015
included on two occasions, and Extraocular Muscle Control (i.e., eye pursuit) had been included only once. In addition, the three other hard definitional variables had been excluded. On the most recent occasion (1977), Ayres labeled a factor as Praxis and commented: "Most apraxic children show the symptoms complex that appears as Factor D in Table 2. These children usually have low scores on IP and tactile tests of the SCSIT [Southern California Sensory Integration Test]. The more severe cases usually score poorly on Kinesthesia and SVCU" (p. 366). That statement translates into the following list of variables: (1) Figure Ground, (2) Position in Space, (3) Ayres Space Test, (4) Right-Left Discrimination, (5) Illinois Test of Psycholinguistic Ability (ITPA) Visual Sequential Memory, (6) Posture Duplication, (7) Stereognosis, (8) Graphesthesia, (9) Single Stimulus Localization, (10) FaceHand Test, (11) Kinesthetic Memory, and (12) Space Visualization Contralateral Use (SVCU). The following points are relevant to this list: • Two variables (5 and 12) have never formed a significant part of a Praxis factor. • Two variables (3 and 4) have formed part of the Praxis factor on only one occasion each. Hence they did not even rate inclusion in the soft definition formulated earlier. • Two other variables (Finger Identification and Bilateral Motor Control) have been omitted, even though they were included in three out of five of the factors labeled by Ayres as Praxis. Despite these shortcomings, the same calculations have been performed using these 12 definitional variables as were reported using the hard and soft definitions. Once again, "%" refers to the proportion of 12 variables available to the study that were included in each Praxis factor. "Loading %" refers to the loading contribution of the available definitional variables to each factor. The results of the calculations are presented on the right side of Table 4. As can be seen, the pattern of results is quite similar to that of the operational definition. Again, the factors labeled as Praxis show no greater conformity to the
definition than factors labeled otherwise. But most importantly, the application of Ayres' definition to these data failed to discriminate the scores derived from children with learning disabilities and nondisabled children. It must be concluded from these data that there is no evidence that a factor of Praxis emerges more readily from the data of children with learning disabilities as compared to those of nondisabled children.
Ayres, Mailloux, and Wendler (1987) After an interval of 10 years, Ayres reported another factor analysis, which confirms that a simple Praxis factor does not reliably emerge from the scores of children with learning disabilities. This study was not included in the previous group analysis because a quite different set of test scores were used. They included six different measures of praxis and a number of scores derived from combinations of the individual tests listed in Table 3. Three factors emerged from the data. One was labeled "Kinesthesia," a title Ayres has used once before, in the 1972bDI study (see Table 2). Another was labeled "Auditory memory," which Ayres has not used before. But the major factor, which included in its composition the praxis measures and a mixture of other scores, was labeled "Visuosomatopraxis function with elements linked by concept formation" (abstract). Although the precise meaning of this factor may be obscure, the conclusion that may be drawn is not. As the authors conclude, this study failed to support the emergence of a unitary Praxis factor from the scores of children with learning disabilities.
FORM AND SPACE PERCEPTION The analyses of this and subsequent factor compositions will omit the tables that show the variable composition for each; these may be obtained from the author on request. The investigation of this factor proceeded in the same manner described for Praxis. A "Form and Perception" factor has been named by Ayres on three occasions (1965, 1972b-DR, 1972b-DI). The hard definition involved four variables
(Position in Space, Ayres* Space Test, Figure-ground Perception, and Design Copying), with the soft definition including three more (Kinesthetic Memory, Stereognosis, and ITPA-Visual closure). Table 5 shows the degree of conformity between the actual number (%) or variable loading (soft loading %) of definitional variables within the factors. In terms of the hard definition, six of eight studies contained a factor that included 100% of the available variables. However, it should be noted that two of these came from the scores of nondisabled children, and the 1977-DR factor had been labeled by Ayres as "Auditory language functions." Using the soft definition, one certainly finds a more appropriate separation. However, the two distributions did overlap and the factor Ayres labeled "Visual Perception" (1966aNR) from nondisabled children exceeded the "Form and Space Perception" factor from 1972b-DI. When the factor loadings are compared (soft loading %), it can again be seen that, though they tend to favor the named factors from groups with learning disabilities, the five top scores are separated by only 15 percentage points, and two of these five come from nondisabled children. In conclusion, these data do not present convincing evidence that a Form and Space Perception factor could be used to reliably separate the scores of children with learning disabilities and nondisabled children.
Ayres' Definition Ayres provided a verbal description of this factor on six occasions (1963, 1965, 1967, 1971, 1972a, 1972b). While these are fairly heterogeneous, she makes it clear that a central component comprises variables measuring form and space perception across the three modalities of visual, tactile, and kinesthetic perception. In 1972a, Ayres defined the factor as consisting of the variables "visual form and space perception, kinesthetic memory, stereognosis, visual figure ground perception and perception in other sensory modalities that involve some kind of form perception" (p. 191). This is consistent with the soft definition that has been investigated.
Volume 24, Number 3, March 1991
165 Downloaded from ldx.sagepub.com at MCMASTER UNIV LIBRARY on March 2, 2015
TACTILE DEFENSIVENESS One apparently unique feature of Ayres' theory is the idea that children with learning disabilities suffer from Tactile Defensiveness. This is described as physical or emotional discomfort when touched by an examiner during tests of tactile perception (see Ayres, 1965, for a description of the testing procedure). Ayres has described this condition in considerable detail (1964, 1966b, 1972a) and uses the work of Head (1920) as her theoretical basis. Head proposed a duality in the cutaneous systems: The "Protopathic system," so named because of its supposedly primitive characteristics, serves to protect and warn the organism against potential harm. The "Epicritic system" is concerned with higher, discriminatory functions. Ayres used this presumed dichotomy to reason that the tactile defensive reactions in children with learning disabilities are caused by the dominance of the protopathic system. Hence, tactile stimuli imposed upon these children warn them of "danger" and this leads to the "over-
alertness of distractibility, the flight-like behavior of hyperactivity, and a tendency toward negative affect (flight)" (1972a, p. 255). It should be noted, however, that Head's dichotomy has not been experimentally validated (Miller, 1978, pp. 318-324). The investigation of this factor proceeded as before, with two minor modifications: first, that no factor has been included from 1977-DR, because the key variables of Tactile Defensiveness and Hyperactivity were not included in the study. Second, due to the small number of shared variables between the three factors Ayres labeled as "Tactile Defensiveness," only the soft form of the operational definition has been produced. The calculations shown in Table 6 have been performed as before. Once again, it is clear that the definitional variables are unable to reliably differentiate either between factors carrying the title of "Tactile defensiveness" or between the scores of children who are academically disabled and children who are nondisabled. This lack of discrimination
TABLE 5 Conformity Between Actual and Definitional Factor Structure for Form and Space Perception Hard definition
Factors Named "Form and Space Perception Factors with Other Names
Soft definition
Soft
Study
P
Q
%
P
Q
%
Loading
1965 DR 1972b DR 1972b Dl
4 4 4
4 4 4
100 100 100
6 7 5
6 7 7
100 100 71
51 62 64
1965 NR 1966a NR 1966c NR 1969 Dl 1977 DR
1 3 3 2 4
4 3 3 3 4
25 100 100 67 100
1 3 3 2 4
6 4 5 6 7
17 75 60 33 57
13 49 58
%
— -
TABLE 6 Conformity Between Actual and Definitional Factor Structures for Tactile Defensiveness Definition
Loading
Study
%
%
Factors Named "Tactile Defensiveness"
1965 DR 1966c NR 1972b Dl
100 60 80
75 81 43
Factors with Other Names
1965 NR 1966a NR 1969 Dl 1972b DR
20 100 40 60
49 29
166
32
Downloaded from ldx.sagepub.com at MCMASTER UNIV LIBRARY on March 2, 2015
does not support the use of tactile defensiveness as a diagnostic device.
BILATERAL INTEGRATION Whereas Ayres described factors by the simple terms "Bilateral integration" (1965-DR) and "Postural and bilateral integration" (1969-DI) on one occasion each, on two other occasions she used terms that seem to implicate Bilateral Integration as a central component. These are: 1966c-NR
"An interaction of function of the two sides of the body with emphasis on tactile perception" (p. 288). 1969-DI(2) "Poorer coordination on the left than on the right side of the body along with deficiencies in postural and bilateral integration" (p. 166).
It should be noted that two of these factors are drawn from the 1969 study and will be referred to as (1) and (2), respectively. It can be seen from Table 7 that once again the operational definition fails to discriminate between the scores of nondisabled children and children with learning disabilities. It can also be seen that Ayres' labeling of factors as "Bilateral Integration" seems to bear little relationship to their variable content.
FIGURE-GROUND DISCRIMINATION Ayres has asserted on a number of occasions (1963, 1975) that Figure-Ground Discrimination is a factor that discriminates between children with learning disabilities and nondisabled children. It is curious to note, therefore, that a factor has only been given this simple designation on two occasions (1965), when it was found to emerge once each from the scores of nondisabled children and children with learning disabilities.
POSTURAL AND OCULAR REACTIONS According to Ayres, "postural mechanisms are among the more important Journal of Learning Disabilities
TABLE 7 Conformity Between Actual and Definitional Factor Structures for Bilateral Integration Definition
Factors Named "Bilateral Integration"
Factors with Other Names
Loading
Study
P
Q
%
%
1965 DR 1966c NR 1969 Dl(1) 1969 Dl(2)
2 6 2 4
6 6 7 7
33 100 29 57
61 67
1965 NR 1966a NR 1972b DR 1972b Dl 1977 DR
2 5 2 2 1
6 5 7 7 7
33 100 29 29 14
76 31 19 35
areas of assessment of children with learning disorders" (1972a, p. 98). Indeed, a very substantial component of her remedial program is directed to this area. Specific therapeutic targets are the inhibition of primitive reflexes, increased functioning of antagonistic muscle contraction, development of muscle tone, improved extraocular muscle control, and vestibular system functioning. While Ayres has labeled factors as "Postural and Ocular Reactions" on three occasions, this factor is unfortunately not amenable to analysis. The variables common to the named factors are not sufficiently represented within the other studies for comparisons to be made.
EVIDENCE FROM MULTIPLE REGRESSION If one desires to validate the hypotheses arising from a factor analysis, then a useful approach is to employ another multivariate technique, called multiple regression. This statistic seeks to "explain" a single known, observed, and measured dependent variable in terms of groups of independent variables. In other words, the factors discovered by factor analysis can be tested for explanatory power against some other performance variable. Ayres has used multiple regression for this purpose on one occasion (1972b). Unfortunately, the result from that study is rendered uninterpretable due to the uncertain composition of the independent variables that were generated according to the following procedure:
—
-
(Ayres 1965, 1969) for identification of patterns of syndromes of perceptual-motor dysfunction, each child was assigned a system or syndrome score indicative of the degree of dysfunction in five hypothesized neural systems. (p. 329, emphasis added) The five "hypothesized neural systems" that were defined by the children's "syndrome scores" were as follows: 1. 2. 3. 4. 5.
Postural and bilateral integration Praxis Functions of the left side of the body Form and space perception Auditory-language functions
In relation to Ayres' claim that the results from the 1965 and 1969 studies could be used to divine "previously factorially determined criteria," the following points can be made:
1. Neither study cited yielded all five factors, and each yielded factors other than the ones listed. 2. The 1969 report contained two analyses, one using raw data and one using ipsative data. The factors derived from each form of analysis differed not only between the two analyses but also differed from the 1965 analysis. 3. Ayres separated the syndrome scores within each factor into three classes, depending on the "frequency and severity of the disorder" indicated by the score. Thus, each factor was subdivided in terms of the "degree of dysfunction" into definite, mild, or none. Unfortunately, this separation further confuses the picture, because no criteria are provided for the allocaBy perusing the scores of each child and using previously factorially determined criteria tion of scores into each category.
Moreover, the interpretation in terms of degrees of dysfunction is highly equivocal, as no pathology has been shown to be linked to the levels of functioning on these variables. These deficiencies render the multiple regression analysis uninterpretable, primarily because the composition of the independent variables (factors and levels of dysfunction) cannot be determined from the available data. As a consequence, this study will not be given further consideration.
SUMMARY Between 1965 and 1977 Ayres published the results of eight factor analyses. The data from these studies had been derived from a variety of perceptual-motor and other tests applied to either children with learning disabilities or nondisabled children. The factors that emerged were claimed to discriminate between the two groups of children and have subsequently formed the basis of an elaborate diagnostic and remedial training protocol for children with learning disabilities. This claim, that the data from children with learning disabilities gives rise to characteristic factor structures, has not been adequately tested. The claim is made solely on the basis that certain factor labels, such as Praxis, have been differentially applied to data derived from such children. Factor labels, however, are fairly arbitrary devices that do not necessarily reflect true factor content. The present paper explored the relationship between factor labels and factor content in Ayres' studies. The investigation of these data proceeded in a multistep process. Initially the eight analyses were scanned for common factor labels. Seven labels were found to have been used on more than one occasion each. Each group of similarly labeled factors was then examined to determine whether their content could be used to reliably distinguish them from other factors. The core of common content within each group was identified as those variables that loaded 0.30 or more onto each factor. These core clusters were then compared against the content of other factors using two approaches. The first involved a simple numerical comparison
Volume 24, Number 3, March 1991
167 Downloaded from ldx.sagepub.com at MCMASTER UNIV LIBRARY on March 2, 2015
of the number of core variables also found in other factors. The second compared the accumulated factor loadings of the core variables with the same variable loadings found in other factors. In each case, no evidence was found for any variable cluster that could reliably distinguish the similarly labeled factors from other factors that had been differently named. A different approach to this discrimination problem used Ayres' definitions to determine the optimal variable content of the similarly named factors. Comparisons using these different sets of defining variables yielded no better result. Again, these variable clusters could not reliably distinguish the similarly named factors from some other factors that carried different names. It was concluded that there is a lack of consistency between Ayres' factor labels and composition. Indeed, no core group of variables could be detected that allowed the similarly named factors to be reliably identified by their content. The final set of comparisons was between data produced from the scores of children with learning disabilities and data from nondisabled children. Ayres has claimed that the factors that emerge from these two sets of data are reliably different. Within each group of similarly labeled factors, content was examined to determine the extent to which the shared core of variables could be used to discriminate the scores from the two types of children. Both numerical comparisons and factor loading comparisons were utilized, as previously described. The result of this examination revealed that not one of the named factors had a content that reliably distinguished between the scores from children with learning disabilities and from nondisabled children. It is concluded that the body of data that has been examined provides no validity for either the diagnostic procedures or the remedial programs for children with LD that have been derived from Ayres' multivariate studies.
168
ABOUT THE
AUTHOR
Robert A. Cummins, PhD, is senior lecturer in psychology at Victoria College-Toorak. In addition he is principal investigator with the Disability Program Evaluation Unit, which has major responsibility for evaluating government services to people with intellectual disabilities in Victoria. His research and writing activities focus on debunking remedial education myths, the effects of deinstitutionalization, and quality of life measurement. Address: Robert A. Cummins, Victoria College-Toorak, PO Box 224, Malvern, Victoria 3144, Australia.
AUTHOR'S
NOTE
An extended version of this paper is available from the author on request.
REFERENCES Abbie, M. (1974). Treatment of "minimal brain dysfunction." Physiotherapy, 60, 203-207. Ayres, A.J. (1963). The development of perceptual motor abilities: A theoretical basis for treatment of dysfunction. American Journal of Occupational Therapy, 17, 221-225. Ayres, A.J. (1964). Tactile functions-Their relation to hyperactive and perceptual motor behavior. American Journal of Occupational Therapy, 18, 6-11. Ayres, A.J. (1965). Patterns of perceptual-motor dysfunction in children: A factor analytic study. Perceptual and Motor Skills, 20, 335-368. Ayres, A.J. (1966a). Interrelationships among perceptual-motor functions in children. American Journal of Occupational Therapy, 20, 68-71. Ayres, A.J. (1966b). Interrelation of perception, function and treatment. Journal of the American Physical Therapy Association, 46, 741-744. Ayres, A.J. (1966c). Interrelations among perceptual-motor abilities in a group of normal children. American Journal of Occupational Therapy, 20, 288-292. Ayres, A.J. (1967). Perceptual motor dysfunction: Neurobehavioral criteria. American Journal of Orthopsychiatry, 37, 405-406. Ayres, A.J. (1968). Reading—A product of the sensory integrative processes. In H.K. Smith (Ed.), Perception and reading (pp. 77-82). Newark, DE: International Reading Association. Ayres, A.J. (1969). Deficits in sensory integration in educationally handicapped children. Journal of Learning Disabilities, 2, 160-168.
Downloaded from ldx.sagepub.com at MCMASTER UNIV LIBRARY on March 2, 2015
Ayres, A.J. (1971). Characteristics of types of sensory integrative dysfunction. American Journal of Occupational Therapy, 25, 329-334. Ayres, A.J. (1972a). Sensory integration and learning disorders. Los Angeles: Western Psychological Services. Ayres, A.J. (1972b). Types of sensory integrative dysfunction among disabled learners. American Journal of Occupational Therapy, 26, 13-18. Ayres, A.J. (1975). Sensorimotor foundations of academic ability. In W.M. Cruickshank & D.P. Hallahan (Eds.), Perceptual and learning disabilities in children (Vol. 2, pp. 301-358). New York: Syracuse University Press. Ayres, A.J. (1977). Cluster analysis of measures of sensory integration. American Journal of Occupational Therapy, 31, 362-366. Ayres, A.J. (1979). Sensory integration and the child. Los Angeles: Western Psychological Services. Ayres, A. J, Mailloux, Z.K., & Wendler, C.L. W. (1987). Developmental dyspraxia: Is it a unitary function? Occupational Therapy Journal of Research, 7, 93-110. Cermak, S.A. (1988). Sensible integration. American Journal on Mental Retardation, 92, 413-414. Clark, F.A., Mailloux, Z., & Parham, D. (1985). Sensory integration and children with learning disabilities. In P.N. Clark & A.G. Allen (Eds.), Occupational therapy for children (pp. 359-405). St. Louis: Mosby. Clark, F.A., & Shuer, J. (1978). A clarification of sensory integrative therapy and its application to programming with retarded people. Mental Retardation, 16, 227-232. Grimwood, L.M., & Rutherford, E.M. (1980). Sensory integrative therapy as an intervention procedure with grade one "at risk" readers: A three year study. Exceptional Child, 27, 52-61. Head, J. (1920). Studies in neurology, Volumes I and II. London: Oxford University Press. Kimball, G.K. (1977). The Southern California Sensory Integration Tests (Ayres) and the Bender Gestalt: A correlative study. American Journal of Occupational Therapy, 31, 294-299. Liepmann, H. (1908). Drei Aufsatze Aus Dem Apraxiegebiet. Berlin: Karger. Meulders, M. (1983). Praxia and language. In M. Monnier & M. Meulders (Eds.), Psycho-neurobiology (Vol. 4, pp. 577-614). New York: Elsevier. Miller, J. (1978). The body in question. New York: Random House. Ottenbacher, K., & Short, M.A. (1985). Sensory integrative dysfunction in children: A review of theory and treatment. Advances in Developmental and Behavioral Pediatrics, 6, 287-329. Vezie, M.B. (1975). Sensory integration: A foundation for learning. Academic Therapy, 10, 345-354. Wilson, E.B. (1975). Sensory integration therapy for children with learning disorders. Rehabilitation in Australia, 12, 27-29.
Journal of Learning Disabilities
A
yres has operationalized her concept of sensory integration by conducting multivariate studies on the perceptual-motor performance of children with learning disabilities (Cermak, 1988). Of particular relevance is a series of factor analyses from which Ayres claimed to have identified perceptual-motor factors that emerge from the scores of children with learning disabilities but not from the scores of normally learning children. The perceptual-motor content of these factors has been used to provide both the format and the rationale for diagnostic and remedial programs (e.g., Ayres, 1968, 1975, 1979). Of concern, however, is the diversity of opinion among reviewers of Ayres' studies as to the precise nature and nomenclature of the factors she discovered. Although there seems to be a majority opinion in favor of four factors (i.e., Praxis, Form and Space Perception, Tactile Defensiveness, and Bilateral Integration), not all reviewers agree, and several other factor names have been suggested. Some examples are • Tactile, kinesthetic, and visual perception dysfunction (Abbie, 1974) • Auditory language problems (Wilson, 1975) • Praxis and tactile functions (Kimball, 1977) • Vestibular and bilateral integration (Clark & Shuer, 1978) • Left hemisphere dysfunction and right hemisphere dysfunction (Clark & Shuer, 1978) • Unilateral cerebral dysfunction (Grim-
wood & Rutherford, 1980) • Postural and bilateral integration (Ottenbacher & Short, 1985) The most comprehensive review of Ayres' factor analytic work was published by Clark, Mailloux, and Parham (1985). Those authors examined the emergent factors and concluded as follows (P. 379): • Apraxia: "Emerged in nearly all the studies in which dysfunctional children made up the sample." • Deficits inform and space perception: "Were more varied, and the configuration and patterns of the clusters suggested that there were at least two kinds of visual perceptual problems. The first seemed to be related to somatosensory processing, the second to functions associated with the right hemisphere." • Deficits in postural and bilateral integration: "Emerged relatively consistently and in later studies were linked to vestibular dysfunction." • Auditory language dysfunctions: "Seemed to bifurcate in the same way that visual spatial perception had. One type seemed to be related to the left hemisphere and did not seem to be related to sensory integrative processes. The second type was found in association with vestibular dysfunction and poor bilateral integration." • Tactile defensiveness: "In nearly every study a strong association between tactile defensiveness and hyperactivity was found."
From this analysis the authors generated yet another description of typologies for sensory integrative dysfunction derived from Ayres' work: • Vestibular bilateral integration dysfunction • Dyspraxia • Left hemisphere dysfunction • Right hemisphere dysfunction • Generalized dysfunction The reason for this diversity of views is that Ayres has not attempted to crossvalidate her factors. Each analysis has been performed on a new combination of variables and so has produced a different set of factors. One consequence of this is that factor labeling becomes a critical issue. Since none are identical in composition, it might be assumed that all of the factors would have been given unique labels, but that is not so. Whereas some factor labels have certainly been unique to individual studies and have contributed to the diversity of views mentioned previously, other labels have been more commonly employed. In the absence of cross-validated factors between studies, this use of common factor labels is of special interest. It implies that factors with a similar and distinctive variable composition have emerged from different studies. Indeed, the previously cited reviewers of Ayres' studies appear to have made just that assumption when searching for factor typologies. It seems timely, therefore, to reexamine Ayres' data and to question these conclusions, which assume a simple relationship between factor labels and factor composition. This will be achieved by identifying the groups of similarly named factors and examining each group to determine (a) their distinctive, common content, (b) whether their common content distinguishes them from factors that have been differently labeled, and (c) whether their common content distinguishes between scores derived from children with learning disabilities and nondisabled children.
An Overview of Ayres' Studies Between 1965 and 1987 Ayres published eight papers that contain among Journal of Learning Disabilities
160 Downloaded from ldx.sagepub.com at MCMASTER UNIV LIBRARY on March 2, 2015
them 10 multivariate analyses. All are concerned with the relationships between sensory and motor skills, with academic skill variables sometimes being included as well. The broad characteristics of these studies are outlined in Table 1. There are only nine columns because two different analyses were conducted on the scores of the same children in 1972 (Ayres, 1972b). All children were aged between 4-0 and 10-0 years, were of normal intelligence (where this was reported), and comprised a mixture of males and females (where reported). Of the 10 studies, 9 were factor analyses and 1 a multiple regression. The latter study (1971) will be considered separately, as will the most recent factor analysis (Ayres, Mailloux, & Wendler, 1987). The other eight analyses will be considered as a group, with a view to comparing their factor structures. The factor analyses differed from one another on a number of dimensions as follows: 1. Nondisabled/Learning disabled. Three of the studies (1965, 1966a, 1966c) involved nondisabled children. The 1965 sample was "chosen from public and private schools and child care centers on the basis of parental occupation so as to represent proportionately the working population of the United States" (p. 339). In 1966c, all were adopted children living in middle socioeconomic class families. The 1966a children, however, were a bit different. Of the 92 subjects, Ayres reported that "approximately 10 per cent could be considered to have possible central nervous system dysfunction" (p. 68). No further information on these children's specific problems was given, but it seems likely that their disorders were minimal. In addition, three children had mild cerebral palsy. Since such a mixture of children would not be uncommon within any unselected group of regular school children, for the purposes of future comparisons the children from 1966a will also be considered nondisabled. The remaining five studies all involved children with academic disabilities who had been classified as such using a variety of criteria. For future reference, the studies involving children who are nondisabled will be designated (N) and those involving children who are academically disabled will be designated (D).
TABLE 1 Subject Characteristics in Ayres' Multivariate Analyses Studies
1965
1965
1966a
1966c
1969
1971
1972b
1977
1987
Age range
5-10 8-0 71139
5-8 7-11 ?
4-0 8-0 ?
4-0 8-0 ?
6-1 9-10 X 93
5-10 10-0 70132
7-6 10-0 70132
6-0 10-0 ?
4-0 9-11 ?
-
7 93
-
36
148
148
128
172
64
-
IQ Academic Disabled (N) Nondisabled (N) Cerebral Palsy (N) Males (%) Females (%) Variables (N) Loading cutoff for factor definition
100
-
-
50
-
-
-
-
-
-
-
-
-
-
-
69 31 36
70 30 36
3 ? ? 19
? ? 19
81 19 64
74 26 48
74 26 46
80 20 19
65 35 19
0.3
0.3
None used
None used
0.4
-
0.2
0.3
None used
Note. ? = not reported.
2. Raw scores/Ipsative scores. While six of the factor analyses have used the conventional raw scores derived from each test, the 1969 analysis and one of the 1972b analyses used ipsative scores. This process led to the emergence of quite different factor structures, as will be seen from the 1972b study, which performed both raw score and ipsative score analyses on the same set of data. For future reference, the studies involving raw scores will be designated as (R), while those using ipsative scores will be designated (I). 3. Loading cut-off. As shown in Table 1, the studies differed somewhat in the minimum size of the variable loading that was used to define the content of factors. For comparative purposes, all studies will be standardized such that a minimum loading of 0.30 is required for any variable to be considered part of a factor. 4. Variables. Of the 113 different variables used in these studies, only 8 were common to all analyses, while 50 were used on just one occasion each. A description of the variable names and their test protocols can be found in Ayres (1965, 1966a, 1969).
EXAMINATION OF THE FACTOR LABELS Ayres' first two factor analyses (1965) set the conceptual stage for all of the analyses to follow. As shown in Table 1,
this report contained one analysis that involved children with "suspected dysfunction" and another analysis on nondisabled children. From the former group, Ayres located five factors that "were interpretable as hypothesized behavioral correlates of patterns of neurological dysfunction" (p. 345). From the nondisabled children, on the other hand, she found that the emergent factors "on the whole were not comparable to the major factors emerging from the dysfunction group." This apparent difference between factor structures from the two groups led Ayres to hypothesize that in the case of the dysfunctional group "these factors were not due to normal developmental processes but to underlying deficits in specific mechanisms of integration, resulting in symptom complexes" (p. 465). Over the next 12 years Ayres conducted six more analyses that sought to validate the emergence of such "symptom complexes" from the scores of children with learning disabilities. The labels or descriptions that Ayres gave to the factors from all of these analyses are shown in Table 2. Where a simple label has been attached to a factor, such as "Praxis," this label has been used. Where a factor has not been named, but only described, the major variables involved in the description have been used as the title. What is immediately clear from this listing is that relatively few factor titles or descriptors have emerged on more than one occasion. To summarize:
Volume 24, Number 3, March 1991 Downloaded from ldx.sagepub.com at MCMASTER UNIV LIBRARY on March 2, 2015
161
TABLE 2 Factor Labels or Descriptions Labels/ Descriptions Praxis (or apraxia) Form and space perception Tactile defensiveness Bilateral integration Figure-ground discrimination General perceptual-motor and cognitive ability Body balance Visual perception Visual motor Interaction of sides of body with tactile perception Auditory, language, and sequencing Postural and bilateral integration Left-side coordination, posture, and bilateral integration Academic achievement and sensory perception Auditory/visual association and horizontal perception Auditory, language, and intelligence Postural and ocular mechanisms Reading, spelling, and intelligence Eye-hand dominance and left-eye dominance Left-hand vs. righthand coordination Auditory-language problems Kinesthesia Somatosensory Use of one hand in contralateral space Auditory-perception Uninterpreted factor
1965 DR
1965 NR
1966a NR
1966c NR
1969 Dl
X
X X
1972b DR
1972b Dl
1977 DR
X
X
X
X
X
X
X
X
X X
X X
EXAMINATION OF INDIVIDUAL FACTORS
X X X
Using Table 2 as a guide, each similarly labeled group of factors will now be examined to determine the nature of their common content.
X X X
Praxis
X X X X X
X
X
X
X
X X X
X
X X X X X
X
Note. D = Learning disabled; N = Normal; R = Raw scores; I == Ipsative scores.
• One factor label ("Praxis") has emerged five times. • Three factors have emerged three times. • Three factors have emerged on two occasions. • The remaining 19 factor labels have been used by Ayres on only one occasion each. Despite this lack of consistency, in 1972a Ayres summarized her findings as follows: "From the author's statistical analyses . . . five factors emerged with sufficient frequency and similarity of operative
It is interesting to relate these labels to those actually used to describe the emergent factors (see Table 2). With the possible exception of "Praxis," it seems rather extravagant to claim their general emergence. "Auditory language problems" was used as a label on only two occasions, whereas on no occasion had any factor been designated as a disorder of postural, ocular, and bilateral integration.
parameters to be considered functionally related aspects of human behavior. It is hypothesized that they reflect neural systems in which disorder has been found in children with learning problems" (p. 94). The terms Ayres (1972a) used to designate the five factors were: 1. Disorder in postural, ocular, and bilateral integration 2. Apraxia 3. Disorder in form and space perception 4. Auditory-language problems 5. Tactile defensiveness
162
According to Meulders (1983), the concept of praxia derives from apraxia, the latter term being defined by Liepmann (1908) as the impossibility of performing intentionally certain movements, or sequences of movements, in spite of the integrity of the primary motor pathways and the conservation of most of the elementary or automatic motor functions. In the context of sensory integration theory, the term praxia is used to describe the ability to plan motor acts. The deficit is usually referred to as developmental dyspraxia, which Ayres et al. (1987) describe as being caused by disordered sensory integration resulting in poor development of the ability to motor plan or to sense what sequence of movements will be needed in order to accomplish a motor task. According to Vezie (1975) it also implies that while specific motor skills may be learned through repetition, the growth of generalized ability to motor plan is lacking. Ayres has claimed that this factor tends to emerge more readily from the data of children with LD than from nondisabled children—an important finding, if true. Unfortunately, evaluating this claim on the basis of Ayres' data is not easy, for reasons that can be seen in Table 3. There, under "A," are listed the compositions of all five factors Ayres has labeled as "Praxis." (Note that a key immediately follows this table.) It can be seen that of the 32 variables included at one time or another, only 5 have been Journal of Learning Disabilities
Downloaded from ldx.sagepub.com at MCMASTER UNIV LIBRARY on March 2, 2015
included on three or more occasions, 17 have been included on two or more occasions, and the remaining 15 have appeared only once.
Definition of the Praxis Factor
TABLE 3 The Variable Composition off Praxis A Variables
1965 1966a DR NR
B
1972b 1972b> 1977 1965 DR Dl DR Deffn NR
C 1966c 1969 NR Dl
Wire grommet
Even a casual glance at the five facmanipulation 1 X X 7.5 X X X . tors under "A" will indicate their very S.Calif.Mot.Acc. Test X X 2 X 7 3 X . heterogeneous composition. Ayres has Posture duplication 1 1 3.5 2.5 2 1 12 .. Finger identification 3.5 7 7.5 3.5 •• proposed a content definition for these 5 Graphesthesia 5 1 . praxis factors, which will be discussed Single stimulus later. First, the factors will be examined localization 4 6 3.5 8.5 11 .. for their similarity of content. Stereognosis 7 X 3.5 6 1 . Table 3 permits the formulation of an Eye-motor coordination X X X X 8.5 6 X X Two-point discrimination 8.5 X X X X X operational definition based on common 10 6 9 3.5 3 •• elements: Praxis may be defined in terms Face-hand test X X X X X Time and rhythm X 11 of those variables that have loaded 0.30 X X X String winding X X X 12 or higher, on more than one occasion, 4 4 Visual figure-ground 13 13.5 • within a factor named by Ayres as X X X X X X 14 Superimposed figures X X X 15 "Praxis." In terms of the data in Table Eye pursuit X X X X X Vertical perception 16 3 (A), this definition has been used to X X 6 Design copying 17.5 • provide two different definitional struc17.5 9 8.5 Kinesthetic memory • tures: a "hard" definition, consisting of X X 1 X X 15 Space relations only those variables that have appeared X X X X 12 5 Form constancy 2 X 5 2 5 X Bilateral motor control on three or more occasions (marked •• •• Standing balance-eyes under column B), and a "soft" definition, X 5 7.5 10 open • consisting of variables that have been inX 11 2.5 Tactile defensiveness . cluded at least twice (designated • in colX 3 13.5 Position in space • umn B). Each of these two operational X 1 16 Hyperactivity • X 17 Right-left discrimination definitions has then been applied back to Hands crossing body the five praxis factors to gauge the ex4 4 2 X midline . tent of their internal conformity, both in X X X 7.5 X Muscle co-contraction terms of the 5 "hard" and the 17 "soft" Standing b a l a n c e definitional variables indicated under X X 10 eyes closed 11 Ayres space test "B." X X X 5.5 X ITPA sound blending Unfortunately, this process is compliX X X 5.5 X Muscle tone cated by the fact that each of the studies Note. A = those factors labeled by Ayres as "Praxis"; B = hard (••) or soft (•) definitional to be compared failed to include some variables; C = factors not labeled by Ayres as "Praxis," but being the factor in that study of the definitional variables. For examwith a variable structure most conforming to the definitional variables; D = data from disple, the 1965 study did not include the abled children; N = data from nondisabled children; R = raw data; I == ipsatized data variable "Bilateral motor control" and so The variable list comprises all variables loading 0.3 or more on the Praxis factors under it was not available to form part of the "A." It is not a complete list of variables comprising the factors under" C." Each blank space in the table indicates that the variable was used irl the study but did not load 0.30 or above Praxis factor in that study. To compenon the factor. Each "x" in the table indicates that the variable was not used in the parsate for such differences, the extent of ticular study, and therefore could not have formed part of the factor. Numbers in the table definitional conformity has been judged indicate the rank order of variable loading s within each factor. The variable numbered " 1 " only on the basis of the definitional had the highest loading, and equally ranked loadings are indicated as the combined rank variables available to each study. Thus, midpoint. the extent of definitional conformity, for both the hard and soft definitions, is expressed as a percentage calculated as: In other words, definitional conformity tween studies, ranging from 40% to is to be judged only on the basis of the 100% for the hard definition and from % conformity = — x ^ definitional variables available to each 27% to 100% for the soft definition. Instudy. terestingly, the only factor to contain P = Number of definitional variables The results of these calculations are 100% conformity with both hard and included within the named factor. shown in the upper-left-hand quadrant soft definitions was derived from the Q = Number of definitional variables of Table 4. As can be seen, there are con- scores of nondisabled children (1966a). included within the study. siderable differences in conformity beThese discrepancies are highlighted by Volume 24, Number 3, March 1991
163 Downloaded from ldx.sagepub.com at MCMASTER UNIV LIBRARY on March 2, 2015
TABLE 4 Conformity Between Actual and Definitional Factor Structures for Praxis Soft definition
Hard definition P
Q
%
P
Q
%
1965 DR 1966a NR Factors Named 1972b DR "Praxis" 1972b Dl 1977 DR
4 5 3 2 4
4 5 5 5 5
100 100 60 40 80
11 14 8 4 5
16 14 15 15 13
69 100 53 27 38
Factors with Other Names
1 4 2
4 5 5
25 80 40
3 9 4
16 14 14
19 64 29
Study
1965 NR 1966c NR 1969 Dl
Ayres' definition Soft Loading %
%
Loading %
68 87 78 75
70 89 27 18 33
40
40 100
20 67 33
31 66
-
-
47
25 34
-
-
Note. P = Number of definitional variables included within the Praxis factor; Q = Number of definitional variables included within the study. % = P x 100 Q 1 Soft Loading % = Loading contribution of the soft definitional variables (A) to the total loading of variables (0.30 or above) on the factor. Ayres Loading % = Loading contribution of Ayres' definitional variables to the total loading of variables (0.30 or above) on the factor.
a comparison with the three studies wherein Ayres did not name a Praxis factor. In each of these studies the factor has been selected that most closely conforms to the operational definition of Praxis. The names that Ayres had given these factors are as follows: 1965-NR General perceptual-motor and cognitive ability 1966c-NR Interaction of sides of body with tactile perception 1969-DI Postural and bilateral integration contrasted with Praxis The extent to which these factor compositions conform with the hard and soft definition criteria is set out in the lower left quadrant of Table 4. And what is clear from these comparisons is that, despite Ayres' labeling procedures, two of these differently named factors corresponded to the ideal Praxis structure just as well as the 1972b-DI factor that had been labeled as Praxis. In fact, the differently named 1966c-NR factor corresponded to the definitional composition better than three of the named Praxis factors. Further data relevant to these comparisons are presented in the column labeled "Soft Loading %" (see Table 4). The basis of these calculations is the numerical loading of each variable onto
the Praxis factor. Loading is a measure of the extent to which each variable contributes to the factor, a high loading indicating a major contribution. The "Soft Loading %" is the contribution of the soft definitional variables to the total variable loadings on the'factor. For example, in 1965-DR, the Praxis factor was composed of 18 variables in all, and their combined numerical loading amounted to 806. The 11 soft definition variables found within this factor structure had a combined numerical loading of 552. The proportion of the total loading contributed by the definitional variables is then: 552 100 Soft Loading % — x - r = 68% oUo
1
In other words, the soft definitional variables made up 68% of the total variable loadings onto this factor. Two further matters require clarification in these calculations: (1) In the ipsative data comprising 1972b-DI, the numerical loadings were summed regardless of sign. (2) Loading percentage could not be calculated for either 1969-DI or 1977-DR. In the former case the variable loadings were not given, while in the latter study "composite" variables were used in the analysis, thereby rendering the loadings noncomparable with those of previous analyses.
164
On the basis of these calculations it is apparent that the factor labeled by Ayres as "Interaction of sides of body with tactile perception" (1966c) has, in fact, a greater claim to be called "Praxis" than any of the factors so named, since the nine Praxis variables made up 100% of the factor loadings. It must again be concluded that Ayres' factor titles and factor compositions are not in accord. The major question, however, is whether the Praxis factor can be used to distinguish between the scores of academically disabled and nondisabled children. In this regard Ayres has claimed that one of the "clearest differences" between the factors derived from nondisabled and dysfunctional children is "the failure for the dimension of praxis . . . to appear as separate factors in the normal group" (1966c, p. 289). The data presented in Table 4 show quite clearly that this is not so. In terms of either the number of Praxis variables or their relative loadings onto the factors, it can be seen that Praxis may emerge either strongly or weakly from the scores of either group. It is interesting to see, indeed, that the two factors carrying the highest Praxis loadings (1966c-NR and 1966a-NR) are both derived from the scores of nondisabled children.
Ayres' Definitions Ayres has provided a verbal description of the Praxis factor composition on a number of different occasions (1963, 1965, 1966a, 1967, 1971, 1972a, 1972b, 1977). In general these descriptions reflect the heterogeneous factor composition displayed in Table 3, with not one description being totally consistent with any other. The two most recent descriptions are as follows. Ayres (1972a) defined the "Syndrome of developmental apraxia" primarily by a deficit in the variable Imitation of Postures. Other, related deficits were stated to be in the Motor Accuracy Test and Bilateral Motor Coordination, Tactile Tests, Kinesthesia, Extraocular Muscle Control, and Behavior Problems. It is difficult to reconcile this definition with Table 3. Of the seven variable types, only two were included on three or more occasions (Posture Duplication and Bilateral Motor Control), three had been Journal of Learning Disabilities
Downloaded from ldx.sagepub.com at MCMASTER UNIV LIBRARY on March 2, 2015
included on two occasions, and Extraocular Muscle Control (i.e., eye pursuit) had been included only once. In addition, the three other hard definitional variables had been excluded. On the most recent occasion (1977), Ayres labeled a factor as Praxis and commented: "Most apraxic children show the symptoms complex that appears as Factor D in Table 2. These children usually have low scores on IP and tactile tests of the SCSIT [Southern California Sensory Integration Test]. The more severe cases usually score poorly on Kinesthesia and SVCU" (p. 366). That statement translates into the following list of variables: (1) Figure Ground, (2) Position in Space, (3) Ayres Space Test, (4) Right-Left Discrimination, (5) Illinois Test of Psycholinguistic Ability (ITPA) Visual Sequential Memory, (6) Posture Duplication, (7) Stereognosis, (8) Graphesthesia, (9) Single Stimulus Localization, (10) FaceHand Test, (11) Kinesthetic Memory, and (12) Space Visualization Contralateral Use (SVCU). The following points are relevant to this list: • Two variables (5 and 12) have never formed a significant part of a Praxis factor. • Two variables (3 and 4) have formed part of the Praxis factor on only one occasion each. Hence they did not even rate inclusion in the soft definition formulated earlier. • Two other variables (Finger Identification and Bilateral Motor Control) have been omitted, even though they were included in three out of five of the factors labeled by Ayres as Praxis. Despite these shortcomings, the same calculations have been performed using these 12 definitional variables as were reported using the hard and soft definitions. Once again, "%" refers to the proportion of 12 variables available to the study that were included in each Praxis factor. "Loading %" refers to the loading contribution of the available definitional variables to each factor. The results of the calculations are presented on the right side of Table 4. As can be seen, the pattern of results is quite similar to that of the operational definition. Again, the factors labeled as Praxis show no greater conformity to the
definition than factors labeled otherwise. But most importantly, the application of Ayres' definition to these data failed to discriminate the scores derived from children with learning disabilities and nondisabled children. It must be concluded from these data that there is no evidence that a factor of Praxis emerges more readily from the data of children with learning disabilities as compared to those of nondisabled children.
Ayres, Mailloux, and Wendler (1987) After an interval of 10 years, Ayres reported another factor analysis, which confirms that a simple Praxis factor does not reliably emerge from the scores of children with learning disabilities. This study was not included in the previous group analysis because a quite different set of test scores were used. They included six different measures of praxis and a number of scores derived from combinations of the individual tests listed in Table 3. Three factors emerged from the data. One was labeled "Kinesthesia," a title Ayres has used once before, in the 1972bDI study (see Table 2). Another was labeled "Auditory memory," which Ayres has not used before. But the major factor, which included in its composition the praxis measures and a mixture of other scores, was labeled "Visuosomatopraxis function with elements linked by concept formation" (abstract). Although the precise meaning of this factor may be obscure, the conclusion that may be drawn is not. As the authors conclude, this study failed to support the emergence of a unitary Praxis factor from the scores of children with learning disabilities.
FORM AND SPACE PERCEPTION The analyses of this and subsequent factor compositions will omit the tables that show the variable composition for each; these may be obtained from the author on request. The investigation of this factor proceeded in the same manner described for Praxis. A "Form and Perception" factor has been named by Ayres on three occasions (1965, 1972b-DR, 1972b-DI). The hard definition involved four variables
(Position in Space, Ayres* Space Test, Figure-ground Perception, and Design Copying), with the soft definition including three more (Kinesthetic Memory, Stereognosis, and ITPA-Visual closure). Table 5 shows the degree of conformity between the actual number (%) or variable loading (soft loading %) of definitional variables within the factors. In terms of the hard definition, six of eight studies contained a factor that included 100% of the available variables. However, it should be noted that two of these came from the scores of nondisabled children, and the 1977-DR factor had been labeled by Ayres as "Auditory language functions." Using the soft definition, one certainly finds a more appropriate separation. However, the two distributions did overlap and the factor Ayres labeled "Visual Perception" (1966aNR) from nondisabled children exceeded the "Form and Space Perception" factor from 1972b-DI. When the factor loadings are compared (soft loading %), it can again be seen that, though they tend to favor the named factors from groups with learning disabilities, the five top scores are separated by only 15 percentage points, and two of these five come from nondisabled children. In conclusion, these data do not present convincing evidence that a Form and Space Perception factor could be used to reliably separate the scores of children with learning disabilities and nondisabled children.
Ayres' Definition Ayres provided a verbal description of this factor on six occasions (1963, 1965, 1967, 1971, 1972a, 1972b). While these are fairly heterogeneous, she makes it clear that a central component comprises variables measuring form and space perception across the three modalities of visual, tactile, and kinesthetic perception. In 1972a, Ayres defined the factor as consisting of the variables "visual form and space perception, kinesthetic memory, stereognosis, visual figure ground perception and perception in other sensory modalities that involve some kind of form perception" (p. 191). This is consistent with the soft definition that has been investigated.
Volume 24, Number 3, March 1991
165 Downloaded from ldx.sagepub.com at MCMASTER UNIV LIBRARY on March 2, 2015
TACTILE DEFENSIVENESS One apparently unique feature of Ayres' theory is the idea that children with learning disabilities suffer from Tactile Defensiveness. This is described as physical or emotional discomfort when touched by an examiner during tests of tactile perception (see Ayres, 1965, for a description of the testing procedure). Ayres has described this condition in considerable detail (1964, 1966b, 1972a) and uses the work of Head (1920) as her theoretical basis. Head proposed a duality in the cutaneous systems: The "Protopathic system," so named because of its supposedly primitive characteristics, serves to protect and warn the organism against potential harm. The "Epicritic system" is concerned with higher, discriminatory functions. Ayres used this presumed dichotomy to reason that the tactile defensive reactions in children with learning disabilities are caused by the dominance of the protopathic system. Hence, tactile stimuli imposed upon these children warn them of "danger" and this leads to the "over-
alertness of distractibility, the flight-like behavior of hyperactivity, and a tendency toward negative affect (flight)" (1972a, p. 255). It should be noted, however, that Head's dichotomy has not been experimentally validated (Miller, 1978, pp. 318-324). The investigation of this factor proceeded as before, with two minor modifications: first, that no factor has been included from 1977-DR, because the key variables of Tactile Defensiveness and Hyperactivity were not included in the study. Second, due to the small number of shared variables between the three factors Ayres labeled as "Tactile Defensiveness," only the soft form of the operational definition has been produced. The calculations shown in Table 6 have been performed as before. Once again, it is clear that the definitional variables are unable to reliably differentiate either between factors carrying the title of "Tactile defensiveness" or between the scores of children who are academically disabled and children who are nondisabled. This lack of discrimination
TABLE 5 Conformity Between Actual and Definitional Factor Structure for Form and Space Perception Hard definition
Factors Named "Form and Space Perception Factors with Other Names
Soft definition
Soft
Study
P
Q
%
P
Q
%
Loading
1965 DR 1972b DR 1972b Dl
4 4 4
4 4 4
100 100 100
6 7 5
6 7 7
100 100 71
51 62 64
1965 NR 1966a NR 1966c NR 1969 Dl 1977 DR
1 3 3 2 4
4 3 3 3 4
25 100 100 67 100
1 3 3 2 4
6 4 5 6 7
17 75 60 33 57
13 49 58
%
— -
TABLE 6 Conformity Between Actual and Definitional Factor Structures for Tactile Defensiveness Definition
Loading
Study
%
%
Factors Named "Tactile Defensiveness"
1965 DR 1966c NR 1972b Dl
100 60 80
75 81 43
Factors with Other Names
1965 NR 1966a NR 1969 Dl 1972b DR
20 100 40 60
49 29
166
32
Downloaded from ldx.sagepub.com at MCMASTER UNIV LIBRARY on March 2, 2015
does not support the use of tactile defensiveness as a diagnostic device.
BILATERAL INTEGRATION Whereas Ayres described factors by the simple terms "Bilateral integration" (1965-DR) and "Postural and bilateral integration" (1969-DI) on one occasion each, on two other occasions she used terms that seem to implicate Bilateral Integration as a central component. These are: 1966c-NR
"An interaction of function of the two sides of the body with emphasis on tactile perception" (p. 288). 1969-DI(2) "Poorer coordination on the left than on the right side of the body along with deficiencies in postural and bilateral integration" (p. 166).
It should be noted that two of these factors are drawn from the 1969 study and will be referred to as (1) and (2), respectively. It can be seen from Table 7 that once again the operational definition fails to discriminate between the scores of nondisabled children and children with learning disabilities. It can also be seen that Ayres' labeling of factors as "Bilateral Integration" seems to bear little relationship to their variable content.
FIGURE-GROUND DISCRIMINATION Ayres has asserted on a number of occasions (1963, 1975) that Figure-Ground Discrimination is a factor that discriminates between children with learning disabilities and nondisabled children. It is curious to note, therefore, that a factor has only been given this simple designation on two occasions (1965), when it was found to emerge once each from the scores of nondisabled children and children with learning disabilities.
POSTURAL AND OCULAR REACTIONS According to Ayres, "postural mechanisms are among the more important Journal of Learning Disabilities
TABLE 7 Conformity Between Actual and Definitional Factor Structures for Bilateral Integration Definition
Factors Named "Bilateral Integration"
Factors with Other Names
Loading
Study
P
Q
%
%
1965 DR 1966c NR 1969 Dl(1) 1969 Dl(2)
2 6 2 4
6 6 7 7
33 100 29 57
61 67
1965 NR 1966a NR 1972b DR 1972b Dl 1977 DR
2 5 2 2 1
6 5 7 7 7
33 100 29 29 14
76 31 19 35
areas of assessment of children with learning disorders" (1972a, p. 98). Indeed, a very substantial component of her remedial program is directed to this area. Specific therapeutic targets are the inhibition of primitive reflexes, increased functioning of antagonistic muscle contraction, development of muscle tone, improved extraocular muscle control, and vestibular system functioning. While Ayres has labeled factors as "Postural and Ocular Reactions" on three occasions, this factor is unfortunately not amenable to analysis. The variables common to the named factors are not sufficiently represented within the other studies for comparisons to be made.
EVIDENCE FROM MULTIPLE REGRESSION If one desires to validate the hypotheses arising from a factor analysis, then a useful approach is to employ another multivariate technique, called multiple regression. This statistic seeks to "explain" a single known, observed, and measured dependent variable in terms of groups of independent variables. In other words, the factors discovered by factor analysis can be tested for explanatory power against some other performance variable. Ayres has used multiple regression for this purpose on one occasion (1972b). Unfortunately, the result from that study is rendered uninterpretable due to the uncertain composition of the independent variables that were generated according to the following procedure:
—
-
(Ayres 1965, 1969) for identification of patterns of syndromes of perceptual-motor dysfunction, each child was assigned a system or syndrome score indicative of the degree of dysfunction in five hypothesized neural systems. (p. 329, emphasis added) The five "hypothesized neural systems" that were defined by the children's "syndrome scores" were as follows: 1. 2. 3. 4. 5.
Postural and bilateral integration Praxis Functions of the left side of the body Form and space perception Auditory-language functions
In relation to Ayres' claim that the results from the 1965 and 1969 studies could be used to divine "previously factorially determined criteria," the following points can be made:
1. Neither study cited yielded all five factors, and each yielded factors other than the ones listed. 2. The 1969 report contained two analyses, one using raw data and one using ipsative data. The factors derived from each form of analysis differed not only between the two analyses but also differed from the 1965 analysis. 3. Ayres separated the syndrome scores within each factor into three classes, depending on the "frequency and severity of the disorder" indicated by the score. Thus, each factor was subdivided in terms of the "degree of dysfunction" into definite, mild, or none. Unfortunately, this separation further confuses the picture, because no criteria are provided for the allocaBy perusing the scores of each child and using previously factorially determined criteria tion of scores into each category.
Moreover, the interpretation in terms of degrees of dysfunction is highly equivocal, as no pathology has been shown to be linked to the levels of functioning on these variables. These deficiencies render the multiple regression analysis uninterpretable, primarily because the composition of the independent variables (factors and levels of dysfunction) cannot be determined from the available data. As a consequence, this study will not be given further consideration.
SUMMARY Between 1965 and 1977 Ayres published the results of eight factor analyses. The data from these studies had been derived from a variety of perceptual-motor and other tests applied to either children with learning disabilities or nondisabled children. The factors that emerged were claimed to discriminate between the two groups of children and have subsequently formed the basis of an elaborate diagnostic and remedial training protocol for children with learning disabilities. This claim, that the data from children with learning disabilities gives rise to characteristic factor structures, has not been adequately tested. The claim is made solely on the basis that certain factor labels, such as Praxis, have been differentially applied to data derived from such children. Factor labels, however, are fairly arbitrary devices that do not necessarily reflect true factor content. The present paper explored the relationship between factor labels and factor content in Ayres' studies. The investigation of these data proceeded in a multistep process. Initially the eight analyses were scanned for common factor labels. Seven labels were found to have been used on more than one occasion each. Each group of similarly labeled factors was then examined to determine whether their content could be used to reliably distinguish them from other factors. The core of common content within each group was identified as those variables that loaded 0.30 or more onto each factor. These core clusters were then compared against the content of other factors using two approaches. The first involved a simple numerical comparison
Volume 24, Number 3, March 1991
167 Downloaded from ldx.sagepub.com at MCMASTER UNIV LIBRARY on March 2, 2015
of the number of core variables also found in other factors. The second compared the accumulated factor loadings of the core variables with the same variable loadings found in other factors. In each case, no evidence was found for any variable cluster that could reliably distinguish the similarly labeled factors from other factors that had been differently named. A different approach to this discrimination problem used Ayres' definitions to determine the optimal variable content of the similarly named factors. Comparisons using these different sets of defining variables yielded no better result. Again, these variable clusters could not reliably distinguish the similarly named factors from some other factors that carried different names. It was concluded that there is a lack of consistency between Ayres' factor labels and composition. Indeed, no core group of variables could be detected that allowed the similarly named factors to be reliably identified by their content. The final set of comparisons was between data produced from the scores of children with learning disabilities and data from nondisabled children. Ayres has claimed that the factors that emerge from these two sets of data are reliably different. Within each group of similarly labeled factors, content was examined to determine the extent to which the shared core of variables could be used to discriminate the scores from the two types of children. Both numerical comparisons and factor loading comparisons were utilized, as previously described. The result of this examination revealed that not one of the named factors had a content that reliably distinguished between the scores from children with learning disabilities and from nondisabled children. It is concluded that the body of data that has been examined provides no validity for either the diagnostic procedures or the remedial programs for children with LD that have been derived from Ayres' multivariate studies.
168
ABOUT THE
AUTHOR
Robert A. Cummins, PhD, is senior lecturer in psychology at Victoria College-Toorak. In addition he is principal investigator with the Disability Program Evaluation Unit, which has major responsibility for evaluating government services to people with intellectual disabilities in Victoria. His research and writing activities focus on debunking remedial education myths, the effects of deinstitutionalization, and quality of life measurement. Address: Robert A. Cummins, Victoria College-Toorak, PO Box 224, Malvern, Victoria 3144, Australia.
AUTHOR'S
NOTE
An extended version of this paper is available from the author on request.
REFERENCES Abbie, M. (1974). Treatment of "minimal brain dysfunction." Physiotherapy, 60, 203-207. Ayres, A.J. (1963). The development of perceptual motor abilities: A theoretical basis for treatment of dysfunction. American Journal of Occupational Therapy, 17, 221-225. Ayres, A.J. (1964). Tactile functions-Their relation to hyperactive and perceptual motor behavior. American Journal of Occupational Therapy, 18, 6-11. Ayres, A.J. (1965). Patterns of perceptual-motor dysfunction in children: A factor analytic study. Perceptual and Motor Skills, 20, 335-368. Ayres, A.J. (1966a). Interrelationships among perceptual-motor functions in children. American Journal of Occupational Therapy, 20, 68-71. Ayres, A.J. (1966b). Interrelation of perception, function and treatment. Journal of the American Physical Therapy Association, 46, 741-744. Ayres, A.J. (1966c). Interrelations among perceptual-motor abilities in a group of normal children. American Journal of Occupational Therapy, 20, 288-292. Ayres, A.J. (1967). Perceptual motor dysfunction: Neurobehavioral criteria. American Journal of Orthopsychiatry, 37, 405-406. Ayres, A.J. (1968). Reading—A product of the sensory integrative processes. In H.K. Smith (Ed.), Perception and reading (pp. 77-82). Newark, DE: International Reading Association. Ayres, A.J. (1969). Deficits in sensory integration in educationally handicapped children. Journal of Learning Disabilities, 2, 160-168.
Downloaded from ldx.sagepub.com at MCMASTER UNIV LIBRARY on March 2, 2015
Ayres, A.J. (1971). Characteristics of types of sensory integrative dysfunction. American Journal of Occupational Therapy, 25, 329-334. Ayres, A.J. (1972a). Sensory integration and learning disorders. Los Angeles: Western Psychological Services. Ayres, A.J. (1972b). Types of sensory integrative dysfunction among disabled learners. American Journal of Occupational Therapy, 26, 13-18. Ayres, A.J. (1975). Sensorimotor foundations of academic ability. In W.M. Cruickshank & D.P. Hallahan (Eds.), Perceptual and learning disabilities in children (Vol. 2, pp. 301-358). New York: Syracuse University Press. Ayres, A.J. (1977). Cluster analysis of measures of sensory integration. American Journal of Occupational Therapy, 31, 362-366. Ayres, A.J. (1979). Sensory integration and the child. Los Angeles: Western Psychological Services. Ayres, A. J, Mailloux, Z.K., & Wendler, C.L. W. (1987). Developmental dyspraxia: Is it a unitary function? Occupational Therapy Journal of Research, 7, 93-110. Cermak, S.A. (1988). Sensible integration. American Journal on Mental Retardation, 92, 413-414. Clark, F.A., Mailloux, Z., & Parham, D. (1985). Sensory integration and children with learning disabilities. In P.N. Clark & A.G. Allen (Eds.), Occupational therapy for children (pp. 359-405). St. Louis: Mosby. Clark, F.A., & Shuer, J. (1978). A clarification of sensory integrative therapy and its application to programming with retarded people. Mental Retardation, 16, 227-232. Grimwood, L.M., & Rutherford, E.M. (1980). Sensory integrative therapy as an intervention procedure with grade one "at risk" readers: A three year study. Exceptional Child, 27, 52-61. Head, J. (1920). Studies in neurology, Volumes I and II. London: Oxford University Press. Kimball, G.K. (1977). The Southern California Sensory Integration Tests (Ayres) and the Bender Gestalt: A correlative study. American Journal of Occupational Therapy, 31, 294-299. Liepmann, H. (1908). Drei Aufsatze Aus Dem Apraxiegebiet. Berlin: Karger. Meulders, M. (1983). Praxia and language. In M. Monnier & M. Meulders (Eds.), Psycho-neurobiology (Vol. 4, pp. 577-614). New York: Elsevier. Miller, J. (1978). The body in question. New York: Random House. Ottenbacher, K., & Short, M.A. (1985). Sensory integrative dysfunction in children: A review of theory and treatment. Advances in Developmental and Behavioral Pediatrics, 6, 287-329. Vezie, M.B. (1975). Sensory integration: A foundation for learning. Academic Therapy, 10, 345-354. Wilson, E.B. (1975). Sensory integration therapy for children with learning disorders. Rehabilitation in Australia, 12, 27-29.
Journal of Learning Disabilities
