British Journaf of Clinical Psychology (1992), 31, 336338
Printed in Great Britain
336
0 1992 The British Psychological Society
Factor structure of the Wechsler Adult Intelligence Scale - revised (WAIS-R) : A clinical sample Adrian Burgess* Department of Pychiatry, Charing Cross and Westminster Medical School, Westminster Hospital, Horsefery Road, Westminster, London SWlP Z A P , UK
Janis Flint Department of Clinical Pychology, Charing Cross Hospital, Fulham Palace Road, London W6 8RP
Heidi Adshead Department of Clinical Pycho logy, University of Surrey, Guildford
Factor analysis was performed on a heterogeneous clinical sample of neurological patients. Both a two- and a three-factor model were extracted. The two-factor solution corresponded to Wechsler’s categorization of verbal and performance subtests. The three-factor solution suggested a verbal comprehension factor, a perceptual organization and a third factor with highest loadings on digit span, arithmetic and digit symbol. Both models were consistent with factor models obtained from the standardization sample. As has previously been reported in neurologically impaired samples, the third factor was relatively more prominent than in the standardization sample. This study provides further evidence for the robustness of the WAIS-R factor structure across different populations and gives good support for the use of the WAIS-R in British clinical samples. The WAIS-R is probably the most widely used cognitive test for adults in the UK and yet surprisingly little information has been published on its performance in a British population. Crawford, Allan, Besson & Stephen (1989) have reported on the factor structure of the WAIS-R in a non-clinical British sample and found that the factor structure was closely comparable to that seen in the US standardization sample. However, until now, no factor-analytic study of the WAIS-R has been reported for a British clinical sample. Factor-analytic studies of the WAIS-R standardization sample have suggested a number of different models with between one and four factors (see Leckliter, Matarrazzo & Silverstein, 1986 for a review), though two- and three-factor models have predominated. The two-factor model supports Wechsler’s division of subtests into verbal and performance groupings, whilst the three-factor models consist of a verbal comprehension factor (information, vocabulary, comprehension and similarities), a perceptual organization factor (block design and object assembly) and a third factor sometimes called freedom from distractibility or simply the third factor (digit span and arithmetic). Digit symbol does not reliably load on any factor. Comparable two- and three-factor solutions have been found in a variety of other US
*
Requests for reprints.
Factor structure of WAIS-R
337
samples and the third factor has been found to be particularly prominent in neurologically impaired samples (Bornstein, Drake & Pakalnis, 1988; Ryan & Schneider, 1986). The aim of this study is to determine whether the factor structure of the WAIS-R for a UK clinical structure is comparable to that found in US samples. Subjects. The sample consisted of a retrospective selection of 90 consecutive patients referred for neuropsychological assessment who met the following criteria: (i) had been assessed with the full WAIS-R; (ii) had either (u) an established clinical diagnosis indicating an organic pathology of the central nervous system or (b) showed a clear abnormality on neuropsychological assessment. Neuropsychological impairment was defined as either a discrepancy between their estimated premorbid IQ and their current Full Scale I Q of greater than 12 (5 per cent level of abnormality; Nelson, 1982) or a verbal-performance I Q difference of greater than 22 (5 per cent level of abnormality ; Sattler, 1982). Premorbid IQ, estimated by the NART was adjusted by subtracting 8 from the values in table 2.1 of the NART manual to take into account the average difference between WAIS and WAIS-R IQs. Patients with a current psychiatric diagnosis or a motor dysfunction considered likely to impair their performance on the WAIS-R were excluded. The sample included 62 males and 28 women with a mean age of 43.9 (SD = 18.7). The sample was diagnostically heterogeneous (closed head injuries = 31, cerebro-vascular disorder = 14; dementia = 12; CNS infection = 6; alcohol-related problems = 5; neoplasm = 5; CNS poisoning = 2; severe epilepsy = 2 ; hydrocephaly = 2; miscellaneous = 7) and included four patients with no established pathology at the time of testing, but who were included because of an abnormality on their WAIS-R scores. The estimated premorbid I Q was 97 (SD = 10.7, range 78-118) with a measured full scale I Q of 88 (SD = 14.1, range 52-131), verbal I Q of 91 (SD = 14.3, range 56-126) and performance I Q of 87 (SD = 15.5, range 51-145). With the exception of digit symbol (mean = 6.7, SD = 3.17), the mean age scaled scores for each subtest were between 7.48 and 9.41 with standard deviations in the range 2.81-3.36. Procedure. Principal component analysis with varimax rotation was performed on the scaled scores of the 11 WAIS-R subtests. Coefficients of congruence were computed between the factor loadings from the clinical sample and the factor loadings from a principal component analysis with varimax rotation of the WAIS-R standardization sample (Wechsler, 1981, table 16). The results of the factor analysis are presented in Table 1. Principal component analysis revealed two factors with eigenvalues greater than 1 and a third with an eigenvalue of 0.97. The first unrotated principal component corresponding tog, produced an eigenvalue of 5.85 and accounted for 53.2 per cent of the variance. The second and third components accounted for 12.4 and 8.8 per cent of the variance
Table 1. Principal component factor structure of the WAIS-R in a clinical UK sample Two factor
Three factor
I
I1
I
I1
I11
Information Digit span Vocabulary Arithmetic Comprehension Similarities
.78 .78 .89 .71 .64 .68
.33 .18 .09 .28 .48 .38
.74 .49 .86 .35 .74 .74
.27 .07 .03 .16 .46 .35
.34 .69 .32 .79 .12 .18
Picture completion Picture arrangement Block design Object assembly Digit symbol
.31 .34 2.4 .09 .47
.76 .71 .80 37 .51
.73 .69 .73 .88 .40
28 .15 .49 .04 .75
Coefficient of congruence
,986
,974
.27 .39 .06 .18 .12 .973
,975
,963
338
Adrian Burgess and others
respectively. The first three principal components from the standardization sample accounted for 55.3, 8.7 and 6.8 per cent of the variance suggesting relatively greater prominence for the second and third factors in the clinical sample. Following earlier authors (Leckliter e t al., 1986), both a two- and a three-factor solution were extracted. (i) The two-factor model was consistent with Wechsler’s division of subtests into verbal and performance subtests. In each case, the verbal subtests showed highest loadings on the first factor and the performance subtests showed highest loading on the second factor. Coefficients of congruence with a two-factor principal component model of the WAIS-R standardization sample were .986 and .974 for the first and second factors respectively. (ii) The three-factor model was consistent with previous three-factor models of the WAIS-R. The first factor corresponded to the verbal comprehension factor (highest loadings in order on vocabulary, information, comprehension and similarities), the second corresponded to the perceptual organization factor (highest loadings, in order on object assembly, block design, picture completion and picture arrangement) and the third corresponded to the freedom from distractibility factor (highest loadings in order, arithmetic, digit symbol and digit span). Coefficients of congruence with a three-factor principal component analysis of the WAIS-R standardization sample were .973, .975 and ,963 for the first, second and third factors respectively. The results of both a two- and three-factor model of the WAIS-R from a British clinical sample are in good agreement with comparable models from the WAIS-R standardization data. As has been noted previously, the third factor is more prominent in neurologically impaired samples and this is confirmed in this study. Overall, this analysis confirms the robustness of the WAIS-R factor structure and gives good support for the use of the WAIS-R in British clinical populations.
References Bornstein, R. A., Drake, M. E. & Pakalnis, A. (1988). WAIS-R factor structure in epileptic patients. Epilepsia, 29, 14-18. Crawford, J. R., Allan, K. M., Besson, J. A. 0. & Stephen, D. W. (1989). The Wechsler Adult Intelligence Scale Revised (WAIS-R) : Factor structure in a UK sample. Personality and Individual Differences, 10, 1209-1212. Leckliter, I. N., Matarazzo, J. D. & Silverstein, A. B. (1986). A literature review of factor analytic studies of the WAIS-R. Journal of Clinical Psychology, 42, 332-342. Nelson, H. (1982). National Adult Reading Test ( N A R 7J Test ManHal. Windsor: NFER-NELSON. Ryan, J. J. & Schneider, J. A. (1986). Factor analysis of the WAIS-R in a brain damaged population. Journal of Clinical Psychology, 42, 962-964. Sattler, J. M. (1982). Assessment of Children’s Intelligence and Special Abilities-Student’r Manual, 2nd ed. Boston: Allyn & Bacon. Wechsler, D. (1981). Wechsler Adult Intelligence Scale - Revised. New York: The Psychological Corporation/Harcourt Brace Jovanovich.
Received 8 October 1990; revised version received 19 AxgHsf 1991
Printed in Great Britain
336
0 1992 The British Psychological Society
Factor structure of the Wechsler Adult Intelligence Scale - revised (WAIS-R) : A clinical sample Adrian Burgess* Department of Pychiatry, Charing Cross and Westminster Medical School, Westminster Hospital, Horsefery Road, Westminster, London SWlP Z A P , UK
Janis Flint Department of Clinical Pychology, Charing Cross Hospital, Fulham Palace Road, London W6 8RP
Heidi Adshead Department of Clinical Pycho logy, University of Surrey, Guildford
Factor analysis was performed on a heterogeneous clinical sample of neurological patients. Both a two- and a three-factor model were extracted. The two-factor solution corresponded to Wechsler’s categorization of verbal and performance subtests. The three-factor solution suggested a verbal comprehension factor, a perceptual organization and a third factor with highest loadings on digit span, arithmetic and digit symbol. Both models were consistent with factor models obtained from the standardization sample. As has previously been reported in neurologically impaired samples, the third factor was relatively more prominent than in the standardization sample. This study provides further evidence for the robustness of the WAIS-R factor structure across different populations and gives good support for the use of the WAIS-R in British clinical samples. The WAIS-R is probably the most widely used cognitive test for adults in the UK and yet surprisingly little information has been published on its performance in a British population. Crawford, Allan, Besson & Stephen (1989) have reported on the factor structure of the WAIS-R in a non-clinical British sample and found that the factor structure was closely comparable to that seen in the US standardization sample. However, until now, no factor-analytic study of the WAIS-R has been reported for a British clinical sample. Factor-analytic studies of the WAIS-R standardization sample have suggested a number of different models with between one and four factors (see Leckliter, Matarrazzo & Silverstein, 1986 for a review), though two- and three-factor models have predominated. The two-factor model supports Wechsler’s division of subtests into verbal and performance groupings, whilst the three-factor models consist of a verbal comprehension factor (information, vocabulary, comprehension and similarities), a perceptual organization factor (block design and object assembly) and a third factor sometimes called freedom from distractibility or simply the third factor (digit span and arithmetic). Digit symbol does not reliably load on any factor. Comparable two- and three-factor solutions have been found in a variety of other US
*
Requests for reprints.
Factor structure of WAIS-R
337
samples and the third factor has been found to be particularly prominent in neurologically impaired samples (Bornstein, Drake & Pakalnis, 1988; Ryan & Schneider, 1986). The aim of this study is to determine whether the factor structure of the WAIS-R for a UK clinical structure is comparable to that found in US samples. Subjects. The sample consisted of a retrospective selection of 90 consecutive patients referred for neuropsychological assessment who met the following criteria: (i) had been assessed with the full WAIS-R; (ii) had either (u) an established clinical diagnosis indicating an organic pathology of the central nervous system or (b) showed a clear abnormality on neuropsychological assessment. Neuropsychological impairment was defined as either a discrepancy between their estimated premorbid IQ and their current Full Scale I Q of greater than 12 (5 per cent level of abnormality; Nelson, 1982) or a verbal-performance I Q difference of greater than 22 (5 per cent level of abnormality ; Sattler, 1982). Premorbid IQ, estimated by the NART was adjusted by subtracting 8 from the values in table 2.1 of the NART manual to take into account the average difference between WAIS and WAIS-R IQs. Patients with a current psychiatric diagnosis or a motor dysfunction considered likely to impair their performance on the WAIS-R were excluded. The sample included 62 males and 28 women with a mean age of 43.9 (SD = 18.7). The sample was diagnostically heterogeneous (closed head injuries = 31, cerebro-vascular disorder = 14; dementia = 12; CNS infection = 6; alcohol-related problems = 5; neoplasm = 5; CNS poisoning = 2; severe epilepsy = 2 ; hydrocephaly = 2; miscellaneous = 7) and included four patients with no established pathology at the time of testing, but who were included because of an abnormality on their WAIS-R scores. The estimated premorbid I Q was 97 (SD = 10.7, range 78-118) with a measured full scale I Q of 88 (SD = 14.1, range 52-131), verbal I Q of 91 (SD = 14.3, range 56-126) and performance I Q of 87 (SD = 15.5, range 51-145). With the exception of digit symbol (mean = 6.7, SD = 3.17), the mean age scaled scores for each subtest were between 7.48 and 9.41 with standard deviations in the range 2.81-3.36. Procedure. Principal component analysis with varimax rotation was performed on the scaled scores of the 11 WAIS-R subtests. Coefficients of congruence were computed between the factor loadings from the clinical sample and the factor loadings from a principal component analysis with varimax rotation of the WAIS-R standardization sample (Wechsler, 1981, table 16). The results of the factor analysis are presented in Table 1. Principal component analysis revealed two factors with eigenvalues greater than 1 and a third with an eigenvalue of 0.97. The first unrotated principal component corresponding tog, produced an eigenvalue of 5.85 and accounted for 53.2 per cent of the variance. The second and third components accounted for 12.4 and 8.8 per cent of the variance
Table 1. Principal component factor structure of the WAIS-R in a clinical UK sample Two factor
Three factor
I
I1
I
I1
I11
Information Digit span Vocabulary Arithmetic Comprehension Similarities
.78 .78 .89 .71 .64 .68
.33 .18 .09 .28 .48 .38
.74 .49 .86 .35 .74 .74
.27 .07 .03 .16 .46 .35
.34 .69 .32 .79 .12 .18
Picture completion Picture arrangement Block design Object assembly Digit symbol
.31 .34 2.4 .09 .47
.76 .71 .80 37 .51
.73 .69 .73 .88 .40
28 .15 .49 .04 .75
Coefficient of congruence
,986
,974
.27 .39 .06 .18 .12 .973
,975
,963
338
Adrian Burgess and others
respectively. The first three principal components from the standardization sample accounted for 55.3, 8.7 and 6.8 per cent of the variance suggesting relatively greater prominence for the second and third factors in the clinical sample. Following earlier authors (Leckliter e t al., 1986), both a two- and a three-factor solution were extracted. (i) The two-factor model was consistent with Wechsler’s division of subtests into verbal and performance subtests. In each case, the verbal subtests showed highest loadings on the first factor and the performance subtests showed highest loading on the second factor. Coefficients of congruence with a two-factor principal component model of the WAIS-R standardization sample were .986 and .974 for the first and second factors respectively. (ii) The three-factor model was consistent with previous three-factor models of the WAIS-R. The first factor corresponded to the verbal comprehension factor (highest loadings in order on vocabulary, information, comprehension and similarities), the second corresponded to the perceptual organization factor (highest loadings, in order on object assembly, block design, picture completion and picture arrangement) and the third corresponded to the freedom from distractibility factor (highest loadings in order, arithmetic, digit symbol and digit span). Coefficients of congruence with a three-factor principal component analysis of the WAIS-R standardization sample were .973, .975 and ,963 for the first, second and third factors respectively. The results of both a two- and three-factor model of the WAIS-R from a British clinical sample are in good agreement with comparable models from the WAIS-R standardization data. As has been noted previously, the third factor is more prominent in neurologically impaired samples and this is confirmed in this study. Overall, this analysis confirms the robustness of the WAIS-R factor structure and gives good support for the use of the WAIS-R in British clinical populations.
References Bornstein, R. A., Drake, M. E. & Pakalnis, A. (1988). WAIS-R factor structure in epileptic patients. Epilepsia, 29, 14-18. Crawford, J. R., Allan, K. M., Besson, J. A. 0. & Stephen, D. W. (1989). The Wechsler Adult Intelligence Scale Revised (WAIS-R) : Factor structure in a UK sample. Personality and Individual Differences, 10, 1209-1212. Leckliter, I. N., Matarazzo, J. D. & Silverstein, A. B. (1986). A literature review of factor analytic studies of the WAIS-R. Journal of Clinical Psychology, 42, 332-342. Nelson, H. (1982). National Adult Reading Test ( N A R 7J Test ManHal. Windsor: NFER-NELSON. Ryan, J. J. & Schneider, J. A. (1986). Factor analysis of the WAIS-R in a brain damaged population. Journal of Clinical Psychology, 42, 962-964. Sattler, J. M. (1982). Assessment of Children’s Intelligence and Special Abilities-Student’r Manual, 2nd ed. Boston: Allyn & Bacon. Wechsler, D. (1981). Wechsler Adult Intelligence Scale - Revised. New York: The Psychological Corporation/Harcourt Brace Jovanovich.
Received 8 October 1990; revised version received 19 AxgHsf 1991
