Search

Trait analyses of the Luria-Nebraska's Intellectual Processes, Motor Functions, and Memory Scales were performed using multitrait-multimethod procedures and confirmatory factor analysis. Three measures each of intellectual, motor, and memory functions were administered to 161 neuropsychiatric patients. Intellectual scales (viz., Luria-Nebraska and WAIS-R) were found to have convincing discriminant properties, but motor and memory measures demonstrated visibly less such robustness. Confirmatory factor analyses supported a three-factor model of the 9×9 matrix, in which each of the three Luria-Nebraska scales serves as a marker for the factors. Issues related to the neuropsychological constructs, test construction, and construct validity of measurement procedures employed in clinical neuropsychology are discussed.

An exploratory maximum likelihood factor analysis of the inter-correlations among the 11 subtests of the WAIS-R was undertaken for 167 patients who had a primary diagnoses of cerebrovascular accident (right hemisphere, n = 79; left hemisphere, n = 55; diffuse or multifocal, n = 33). On the WAIS-R, this sample performed below normative standards (average scaled score = 7.1), but demonstrated a pattern of variability among subtests similar to the normative groups. Interestingly, VIQ-PIQ discrepancy analyses revealed V > P profiles for patients with lesions in either or both hemispheres. The best fit for the WAIS-R matrix was an oblique two-factor model comprised of the Verbal and Performance subtests. This factor solution, which was moderately correlated (.52), accounted for 58.8% of the total variance. The stability of a two-factor structure in this neurologic impaired population suggests that the WAIS-R is a robust measure, even when used within such a rehabilitation population. Only two of the eleven subtests (i.e., Comprehension and Picture Arrangement), however, were related to lesion site. Rehabilitation settings may need to rely more on neuropsychological procedures that measure the more “fluid” areas of neurocognitive ability, in contrast to the WAIS-R which appears less sensitive to the neurological impairment associated with stroke.

One hundred seventy-nine psychiatric inpatients were administered the WAIS-R and several neuropsychological and academic achievement tests. All subjects were assigned to three groups based on their WAIS-R's VIQ-PIQ discrepancy scores: (i) Low Verbal (Verbal IQ < Performance IQ. by at least 13 points); (ii) Low Performance (Performance IQ < Verbal IQ by at least 13 points); and (iii) Equal (Verbal IQ-Performance IQ within 13 points). The Low Verbal group made significantly more errors on the Speech Sounds Perception Test, demonstrated lower spelling scores, evinced more aphasic signs, and had had more special education placements than did the Equal or Low Performance groups. The Low Performance subjects demonstrated significantly more signs of constructional dyspraxia and performed more poorly on the Grooved Pegboard test. The neuropsychiatric patients who had Verbal-Performance IQ discrepancies of at least one standard deviation may be at risk for specific, subtle neuropsychological deficits.

Five multifactor models, in both orthogonal and oblique versions, and a single-factor model of the WAIS-R's factor pattern were examined by confirmatory maximum likelihood factor analyses of a data matrix constructed from the results for 90 neuropsychiatric patients. None of the models fits the data matrix in an absolute sense, even though all of the models represented an improvement over a null statistical model. For the multifactor models, the best results were obtained by oblique solutions, in which the degree of correlation between the factors varied from .71 to .93. The single-factor model fit nearly as well as, and in some instances better than, many of the multifactor models. The best-fitting model, viz., a three-factor oblique one, was only marginally better in its fit than two of the three two-factor models or the competing three-factor model.

The Mini-Mental State Examination (MMSE) and the Blessed Orientation-Memory-Concentration test (BOMC) were each administered to 110 nursing home residents. The correlation between the MMSE and BOMC's total score was -.79. Maximum likelihood factor analysis revealed a two-factor structure, consisting of memory-attention and verbal-comprehension factors that were highly correlated (.69). The results suggest that the somewhat longer MMSE may be preferred for bedside mental status testing, as it clearly measures an additional dimension of cognitive ability, while the BOMC is primarily a unidimensional test. The theoretical and clinical implications of bedside mental status testing of elderly patients are discussed.

Summary.-The present example illustrates a method for assessing pattern similarity for correlation matrices from populations known to differ in clinjcally important respects. The data underlying previously published confirmatory factor analyses of neuropsychological traits are presented m two methods for assessing whether the matrices are similar enough to justi€y them aggregation into a single composite matrix. The significance of this procedure for research In applied neuropsych~log~ is commented upon in terms of the benefits to model diagnostics. A common problem in applied neuropsychological research is the availability of data from populations that are known to differ from each other in a number of clinically salient respects. The question is then posed whether these data (e.g., correlation matrices) can be aggregated (or "pooled") into a single composite matrix (Steyer & Schmitt, 1990). The challenges to this process involve, among other things, the intrinsic features of the data (urz., the pattern of the correlations, their Iwel, range, and multicollinearity) and the need to preserve (or enhance) the sensitivity and specificity of the trajr measures, thereby improving model diagnostics (Schwager, 1991).

The pattern and level of WISC-R performances were examined for 113 younger (ages 6–12) and 124 older (ages 12–16) epileptic children. Although these youngsters performed more poorly than the norm group on all 10 subtests, as do adult epileptics on the WAIS (Fowler, Richards, & Boll, 1980), only the younger subjects showed a definite tendency to be more variable in their performance. Factor analyses demonstrated the presence of highly similar Verbal Comprehension and Perceptual Organization dimensions for both groups but, in the case of the 3-factor solutions, an additional freedom-from-distractibility dimension was found for only the older subjects. These results suggest that Kaufman's (1975, 1979) 3-factor model of the Wechsler scales is more strongly approximated as the age of the epileptic respondents increases.

The current study examines the factor pattern of the Wechsler Adult Intelligence Scale for a sample of 151 neuropsychiatric inpatients (97 men, 54 women). As expected, these patients performed less well on all 11 of the subtests and ranged from .60 to 1.54 SD below normative standards. These patients were neither more variable nor were the intercorrelations among the subtests lower than those of the normative group. As in previous research, maximum likelihood factor analyses revealed the presence of a Verbal Comprehension dimension and a Perceptual Organization dimension, factors that were moderately correlated (.60). The stability of a two-factor solution for the WAIS in this population and its implications for the inclusion of the WAIS as a component of a neuropsychological assessment paradigm are discussed.

Structural analyses of the dependencies between early displays of behavior problems, e.g. aggression, developmental delay, and later family environmental structure are presented for a sample of 19 girls and 16 boys living in the northeastern U.S. Multiple regression analyses show a moderate order of predictability ( R — .56, p < .01) for the regression of an “interpersonal relationships” index on five behavioral indices but a much lower order of predictability ( R — .36, p < .05) for the regression of “organization-control” on these same indices. Although these cross-age (18-mo. interval) correlations are higher than would be expected on the basis of recent theoretical commentary in developmental psychology (Sameroff, 1975), the continuing need for consistent application of multivariate designs within the context of generative models (Barth, 1966) is conspicuous.

One hundred eight epileptic patients (59 men, 49 women; 55% partial seizures, 45% generalized) were examined on 45 indices comprising a modified Halstead Neuropsychological Test Battery. Factor analyses identified five dimensions: Verbal Comprehension, Perceptual Organization, Selective Attention, Motor Skills, Abstract Reasoning. These factors are similar to those previously reported for neurocognitive abilities in several different populations. Relative deficits on Verbal Comprehension and Abstract Reasoning were related, as predicted, to left hemispheric dysfunction as indexed by EEG localization. Perceptual Organization demonstrated a further, predictable relationship to right hemispheric function. Deficits on Selective Attention were not clearly lateralized, being correlated with a wide variety of electroencephalographic abnormalities. The implications of these findings for the use of neurocognitive profiles in treatment planning for this population are discussed.

Two confirmatory maximum likelihood factor analytic models of the Personality Research Form-E are presented and discussed for the men's and women's standardization data (Jackson, 1974) (N = 215). The two models investigated were: (1) a six-factor model based on Jackson's conceptual framework for the Personality Research Form-E; and (2) the five-factor empirical model advanced by Fowler (1985). Overall, the patterns of correlations within each data matrix are highly similar, but the factor patterns, factor intercorrelations, as well as the error variances, do differ. The five-factor model was found to be more adequate generally, but produced no absolute fit to the data matrices, while the six-factor model invariably produced inaccurate estimates of the parameters. Comparisons of the completely estimated, Varimax-rotated factor loading matrices to that reported by Fowler found, on the whole, stronger similarity between matching factors for the women than for the men. The implications of this finding for the modelling of the PRF-E and its application in clinical contexts are considered at length.

The multivariate structure of the Preschool Behavior Questionnaire was investigated in a normal population of preschool children living in the northeastern U.S. Girls ( N = 349) and boys ( N = 352) attending kindergarten in a middle-size urban community were rated by their teachers on a shortened form of the questionnaire for 28 behavioral characteristics. Separate principal component and maximum likelihood factor analyses for each sex indicated the presence of two dimensions: the first combining traits of aggressiveness, attention span problems, and elevated activity level; the second being defined by traits of anxiety, fearfulness, and emotional lability. Both the principal component and maximum likelihood methods produced comparable results for girls and boys. The divergence of the present results from those reported earlier by Behar (1977) was interpreted to be a function of the classroom context and differing sample characteristics.

An oblique, five-factor model of a modified Halstead-Reitan Neuropsychological Test Battery is presented. One hundred fifty-one neuropsychiatric inpatients (94 men, 57 women) were examined on 44 neuropsychological indices. Five correlated dimensions (r = .25), viz., Verbal Comprehension, Perceptual Organization, Sensory-attention, Primary Motor, and Tactile-spatial abilities, were identified by maximum likelihood factor analyses of this correlation matrix. These findings are quite similar to those reported earlier by Fowler, Richards, Berent, and Boll (1985, 1987) and Royce, Yeudall, and Bock (1976). The factors may be broadly categorized using the distinction made by Lezak (1983) for verbal, nonverbal, and mental activity variables. MANOVAs revealed systematic relationships between simple-weighted factor scores and: (1) the overall level of cognitive functioning (p less than .01), as well as (2) psychiatric diagnosis (p less than .01). Implications of these findings for understanding the impact of neuropsychiatric disorders on the structure of abilities for such patients are developed.

The pattern and level of performance on the WAIS-R and the Luria-Nebraska's Intelligence, Memory, and Motor Scales were examined for 93 neurologically impaired adults. Maximum likelihood factor analyses of the WAIS-R indicated the presence of strongly correlated (.72) Verbal Comprehension and Perceptual Organization dimensions. Comparisons of these factors to those of the standardization group revealed an acceptable level of similarity (.94) for Verbal Comprehension, but not for Perceptual Organization (.53). Canonical correlations between WAIS-R factor scores and the three Luria-Nebraska scaled scores accounted for 82% of the total variance. Each set of procedures evidently is indexing the same theoretical constructs. Implications of these findings for our understanding of the measurement models that underlie clinical neuropsychology are discussed.

The pattern and level of WAIS performance for 118 epileptic adults (65 men and 53 women) were compared to those of the standardization sample. As expected, epileptic subjects performed more poorly on all 11 subtests (from .61 to 1.17 SD below the norm group). Contrary to expectation, epileptic subjects were neither more variable nor were the inter-correlations of the subtests lower than those of the norm group. Overall, strong similarities in the patterns of performance on the tests were found for the two groups. As in previous research, factor analysis revealed the presence of a verbal comprehension factor, a perceptual organization dimension, and, in the instance of the three-factor solutions, an additional freedom-from-distractibility dimension for both groups.

Presented for 64 Ss a replication of the Family Environment Scale's maximum likelihood factor structure for which the two-factor, Varimax-rotated solution was found to be stable when the correlations among the subscales were corrected for the effects of social desirability response bias. As reported previously (Fowler, 1981), the first bipolar dimension, “cohesion vs. conflict,” measures relationship-centered concerns, while the second bipolar axis, “organization-control,” reflects basic system-maintenance properties of family activity. These two factors are comparable to others reported in small-group research, parent-child interaction studies, and human communication investigations. A comprehensive statistical model of the Family Environment Scale can be constructed where the semantic space of the 10 subscales reflects the two supraordinate concepts of “interpersonal cohesion” and “control.” It is not yet known to what extent the modeling of this impression-space can be expanded to reveal dependencies on imminent (familial and individual) cognitive and personality schemata.

The current study examines the sensitivity of the Wechsler Adult Intelligence Scale in predicting performance on a modified Halstead Neuropsychological Test Buttery for lower-functioning neuropsychiatric inpatients (97 men, 54 women). Regression analyses support earlier-reported findings that the WAIS's subtests more accurately predict neuropsychological test performance than do the WAIS IQ's alone. The IQ's and scaled scores respectively predicted 27% and 39% of the variance in the Halstead's measures. The overall relationship between neurocognitive and WAIS indices is somewhat greater than previous reports and is apparently due to the fact that the lower the overall level of cognitive ability, the lower is neuropsychological functioning in general. However, since most of the variance in the modified Halstead Battery was not common with the WAIS's indices, these findings continue to support the practice of including both sets of measures as part of a comprehensive neuropsychological assessment paradigm for such patients. The importance of developing instruments and procedures which more precisely measure “fluid” neurocognitive abilities is underscored.