Normative Corrections for the Halstead Reitan Neuropsychological Battery

Transcription

1 Archives of Clinical Neuropsychology, Vol. 14, No. 5, pp , National Academy of Neuropsychology Printed in the USA /99 $ see front matter PII S (98) Normative Corrections for the Halstead Reitan Neuropsychological Battery James A. Moses, Jr. Department of Veterans Affairs Medical Center and Stanford University Medical School David A. Pritchard Roosevelt Warm Springs Institute for Rehabilitation Russell L. Adams University of Oklahoma Health Sciences Center This article investigates the adequacy of the Heaton-Grant-Matthews norms in correcting age and education effects in the Halstead Reitan Neuropsychological Battery. Two hundred and ninety neurological and 346 psychiatric patients were administered the Halstead Reitan Neuropsychological Battery. Raw scores were converted to standard scores using the Heaton-Grant- Matthews norms. Ten percent of the variance in standard score profiles could be predicted by patient age and education. The primary effect of age was to decrease the overall level of the profile; the primary effect of education was to increase relative scores on the Aphasia test. When these standard scores were converted to age- and education-corrected T-scores, less than 1% of the variance could be predicted by age and education. When individual tests were examined, the Heaton-Grant-Matthews norms failed to correct 2% of the variance in the Speech Sounds test. Discussion focuses on the usefulness of the Heaton-Grant-Matthews norms in interpreting performance on the Halstead Reitan Neuropsychological Battery National Academy of Neuropsychology. Published by Elsevier Science Ltd The recent publication of age-, education- and gender-corrected norms for the Halstead Reitan Neuropsychological Battery (HRNB) and supplementary tests by Heaton, Grant, and Matthews (1991) has generated a renewed interest in the merits of adjusting neuropsychological test norms for demographic differences. Vega and Parsons (1967) first reported significant correlations between age, educational level, and performance on the HRNB in Since that time, significant relationships have generally been found between age, education, and or gender, on the one hand, and HRNB test scores and summary scores, on the other hand. Address correspondence to James A. Moses, Jr., VA Palo Alto Health Care System (1116B), 3801 Miranda Avenue, Palo Alto CA

2 446 J. A. Moses, Jr., D. A. Pritchard, and R. L. Adams Vega and Parsons (1967) and Heaton et al. (1991) reported significant age correlations with performance on all HRNB tests among normal subjects. Bak and Greene (1980) and Schludermann, Schludermann, Merryman, and Brown (1983) found age-cohort differences among normal subjects for some HRNB tests but not for others. Similarly, Ernst, Warner, Townes, and Peel (1987) reported age differences among normal subjects on HRNB tasks involving problem-solving and motor strength, but not on verbal and perceptual tasks. Cauthen (1978) and Thompson, Heaton, Matthews, and Grant (1987) reported age-related differences among normal subjects on the Tactual Performance test. Similarly, Long and Klein (1990) found age-cohort differences among nonneurological subjects on spatial and complex integrative tests in the HRNB, but not on sensory, motor, and language tasks. However, Elias, Podraza, Pierce, and Robbins (1990) reported age differences among normal subjects on all of the individual tests comprising the HRNB, except Finger Tapping. In contrast, Ruff and Parker (1993) found significant age differences on Finger Tapping in their sample of normal subjects. Karzmark, Heaton, Grant and Matthews (1984) found that age among nonneurological subjects entered into a multiple correlation between demographic variables and the Average Impairment Rating. Elias et al. (1990) also found age effects on the Average Impairment Rating in a normal sample. Elias et al. (1990) and Elias, Robbins, Walter, and Schultz (1993) reported a relationship between age and the Halstead Impairment Index among normal subjects. Reitan and Wolfson (1995) found a significant effect of age on the General Neuropsychological scale among normal subjects. Although Reitan and Wolfson (1995) cautioned that age corrections estimated from normal subjects may not be necessary in nonnormal subjects, similar effects of age on HRNB test and summary scores have been reported for nonnormal subjects. Age effects have been reported on the Trails test among patients in acute rehabilitation (Corrigan & Hinkeldey, 1987) and among medical-surgical patients (Stanton, 1984). Brain-injured patients have shown age effects on the Category test (Corrigan, Agresti, & Hinkeldey, 1987; Vega & Parsons, 1967) and Tactual Performance test scores (Vega & Parsons, 1967). Alcoholic patients have displayed age effects on individual HRNB tests (Blusewicz, Dustman, Schenkenberg, & Beck, 1977). Goldstein and Shelly (1981) reported age effects among psychiatric and medical surgical patients on the Localization Keys of Russell, Neuringer, and Goldstein (1970). Vega and Parsons (1967) found age effects among neurological subjects on a Revised Impairment Index. Bornstein and Suga (1988), Finlayson, Johnson, and Reitan (1977), Vega and Parsons (1967) and Heaton et al. (1991) found overall differences on HRNB tests by educational attainment among normal subjects. Thompson et al. (1987) reported educational differences among normal subjects on the Tactual Performance test. Karzmark et al. (1984) found that education among nonneurological subjects entered into a multiple correlation between demographic variables and the Average Impairment Rating. Reitan and Wolfson (1995) found a significant effect of education on the General Neuropsychological scale among normal subjects. Similar effects of education have been reported for selected HRNB tests among aphasic patients (Ernst, 1988), psychiatric patients (Warner, Ernst, Townes, & Peel, 1987), epileptic patients (Kupke & Lewis, 1989), rehabilitation patients (Corrigan & Hinkeldey, 1987) and medical patients (Bornstein, 1983; Stanton, 1984). Reitan and Wolfson (1995) found a small but significant effect of education on the General Neuropsychological scale among brain-damaged subjects. In contrast to the above studies, Willis, Yeo, Thomas, and Garry (1988) found no age effects on Category test performance in a sample of elderly normal subjects. However, this negative finding may have been due to the restricted age range (65 79 years old) represented in this study. Similarly, Bornstein (1983) found no age effect on HRNB test

3 HRNB Norms 447 performance in a small, elderly sample of medical patients. Reitan and Wolfson (1989) reported no age effect on the Seashore Rhythm test and Boll and Reitan (1973) found no age effect on the Trail Making test in mixed samples of neurological and normal subjects. Reitan and Wolfson (1995) found no significant effect of age on the General Neuropsychological scale among brain-damaged subjects. Vega and Parsons (1967) found no effect of education on Rhythm, Speech Sounds, or Tapping among brain-damaged subjects. In response to the generally consistent finding of age and education effects on HRNB performance among both normal and nonnormal subjects, several researchers have published demographic corrections for some or all of the HRNB tests (Alekoumbides, Charter, Adkins, & Seacat, 1987; Bornstein, 1985; Charter, Adkins, Alekoumbides, & Seacat, 1987; D Elia, Boone, & Mitrushima, 1995; Steinmeyer, 1986; Yeudall, Reddon, Gill, & Stefanyk, 1987). Most recently, Heaton et al. (1991) published age-, education- and gender-corrected norms for HRNB tests and related neuropsychological tests. First, they converted HRNB raw scores for neurologically normal subjects to normalized standard scores. Then they regressed age education gender on each test s standard scores and calculated T-scores corrected for age, education, and gender effects. The effectiveness of these demographic corrections was empirically evaluated in a new sample of neurologically normal subjects. The present study investigates the adequacy of the Heaton et al. (1991) norms in correcting age and education effects in a sample of nonnormal subjects. 1 First, the effects of age and education on HRNB uncorrected and corrected profiles are examined. Second, since age and education may affect individual HRNB test scores regardless of their effects on the overall profile (Stevens, 1992), their effects on uncorrected and corrected scores for each test separately are examined. METHOD Subjects Raw test scores from 678 patients were available for the Category test (errors), Trails A (seconds), Trails B (seconds), Tactual Performance test (minute block, Memory-correct, and Localization-correct), Seashore Rhythm test (correct), Speech Sounds Perception test (errors), Dominant Finger Tapping (number), Nondominant Finger Tapping (number), and the Aphasia Screening test (total errors). All tests were scored according to specifications outlined in Heaton et al. (1991). These patients had all been referred for neuropsychological evaluation to either the Veterans Administration Medical Center in Palo Alto, CA or the Health Sciences Center in Oklahoma City, OK. Forty-two of these patients were excluded because they were younger than 18 years or older than 80 years, which resulted in a final sample of 636 patients. Two hundred and eighty-one patients (44%) were derived from the California sample and 355 patients (56%) came from the Oklahoma sample. The two samples were pooled in order to increase the generalizability of the sample to the population of patients that are typically referred for clinical neuropsychological evaluation. No between-sample comparisons are made in the present study, so the comparability of the two samples on age, education, diagnosis, or other variables is not an issue. The purpose of the present 1 The effectiveness of the Heaton et al. (1991) norms in correcting gender effects could not be evaluated in the present study because females in the sample were underrepresented in some diagnostic groups. For example, there were only four females with alcoholism as a diagnosis, as compared to 41 males, and there were only four females diagnosed with schizophrenia as compared to 52 males.

4 448 J. A. Moses, Jr., D. A. Pritchard, and R. L. Adams study is simply to describe the effectiveness of age and education corrections in the combined sample. To the extent that the sample is representative of patients typically referred for neuropsychological evaluation, the results may be expected to generalize to other clinical settings. Seventy-six percent (n 481) of the sample was male and 24% (n 155) was female. The average age was 40 (SD 14.4); the average last grade completed was 12.8 (SD 3.1). Information on race ethnicity was not available for the patients included in the sample. Table 1 presents the distributions of the 636 patients on age, education, diagnosis, handedness, and neurological status at the time of testing. Neurologic diagnoses were made by neurologists or neurosurgeons who made use of neurological examination findings in addition to results of computerized tomographic brain imaging, nuclear magnetic resonance brain imaging, electroencephalography, cerebral angiography, radionuclide brain scanning, or findings at brain surgery to establish the clinical diagnosis. Psychiatric diagnoses were made according to the Diagnostic and Statistical Manual of Mental Disorders, third edition (DSM-III; American Psychiatric Association, 1980) or the third edition revised (DSM-III-R; American Psychiatric Association, 1987) criteria by treating staff psychiatrists or psychologists. The psychiatric nomenclature that was current at the time the HRNB test data were collected was used to make those psychiatric diagnoses. In some cases, information from the HRNB was also used to establish neurological or psychiatric diagnosis. The broader diagnostic categories listed in Table 1 were determined by combining subtypes of schizophrenic and schizoaffective disorders into one group (Schizophrenic Spectrum), by combining subtypes of affective disorders into one group (Affective) and by combining otherwise unclassified diagnoses into Other Neurological and Other Unknown groups. Method The 11 raw test scores for each patient were converted to standard scores and to age-, education- and gender-corrected T-scores, using the computer program and normative data provided by Heaton et al. (1991). Although gender was considered in deriving corrected T-scores, it was not examined as a separate variable in the analyses of this study (see footnote 1). To examine the effects of age and education on HRNB profiles, canonical correlation analysis was used to evaluate the association between age and education (last grade completed), on the one hand, and HRNB (uncorrected) standard scores, on the other hand. The analysis was then repeated with HRNB profile (corrected) T-scores to evaluate how well the Heaton et al. (1991) conversions removed the effects of age and education on HRNB profiles. 2 RESULTS Age and education correlated with HRNB standard scores ( , p.001). Table 2 (left columns) presents the factor structure coefficients and redundancies for this analysis. It may be noted in Table 2 that 10% (i.e., redundancies: ) of the variance in HRNB standard scores was predictable from age and education. Examination of the factor structure coefficients for the first canonical variate indicates that all of the HRNB standard scores load positively, while age loads negatively ( 0.870). This 2 To examine the effects of age and education on each test separately, multiple regression was used to predict each test s (uncorrected) standard scores from age and education. The analyses were then repeated using (corrected) T-scores.

5 HRNB Norms 449 TABLE 1 Age, Education, Diagnosis, Handedness, and Neurological Status of Subjects n % Age (years) Education (last grade) Diagnosis Head trauma Vascular disorder Degenerative process Seizure disorder Other neurological Substance abuse Schizophrenic spectrum Affective disorder Adjustment disorder Other unrecorded Handedness Right Left Neurological status Normal Marginal Abnormal Unknown demonstrates that all HRNB standard scores were affected negatively by increasing age, that is, as age increases, all HRNB standard scores decrease (in the impaired direction). These results suggest that the primary effect of age on HRNB standard scores is on the overall level of the standard score profile rather than on specific component tests. Examination of the factor structure coefficients for the second canonical variate indicates that the Aphasia standard score and education load negatively ( and 0.931, respectively). This demonstrates that Aphasia standard scores were increased by increasing educational levels, that is, as educational level increases, Aphasia standard scores increase also (in the nonimpaired direction). However, since the overall level of standard score profiles has been removed by the first canonical variate, it may be stated that education and Aphasia standard scores are positively related regardless of the overall level of the HRNB standard score profile. The primary effect of educational level on HRNB standard scores is on profile shape (i.e., increased relative scores on the Aphasia test) rather than on profile level. 3 3 A comparable analysis of the present data (not presented in this report) using HRNB profile parameters (level, scatter, and shape) confirms that age is negatively related to profile level and education is positively related to Aphasia ipsative z-scores.

6 450 J. A. Moses, Jr., D. A. Pritchard, and R. L. Adams TABLE 2 Canonical Correlation of Age/Education with Halstead Reitan Neurospychological Battery (HRNB) Factor Structure Coefficients Standard Scores T-Scores Root 1 Root 2 Root 1 Root 2 HRNB Category test TPT: Time TPT: Memory TPT: Location Rhythm Speech sounds Tapping (Dom) Tapping (NDom) Trails A Trails B Aphasia Variance extracted Redundancy Age/education Age Education Variance extracted Redundancy Canonical correlations Correlation.508 Chi-square Probability Dom Dominant; NDom Nondominant; TP Tactual Performance test. Age and education correlated with HRNB T-scores ( , p.001), indicating that the Heaton et al. (1991) conversions were not completely successful in removing the relationships between age education and HRNB T-scores. Table 2 (right columns) presents the factor structure coefficients and redundancies for this analysis. It may be noted in Table 2 that less than 1% (i.e., redundancies: ) of the variance in HRNB T-scores was predictable from age and education. Furthermore, the array of structure coefficients for T-scores does not reveal any meaningful pattern (i.e., for both canonical roots, the structure coefficients are small and of mixed sign). This demonstrates that, while some relationship still exists between age education and HRNB T-scores, it is for all practical purposes negligible and uninterpretable. 4 DISCUSSION The present results indicate that the primary effect of increasing age is to lower the overall level of HRNB profiles. The primary effect of increasing education is to change 4 Age and education correlated between 0.20 (Tapping-Dominant) and 0.42 (Category Test) with standard scores on individual HRNB tests. Between 4% (Tapping-Dominant) and 17% (Category Test) of the standard score variance was predictable from age and education. Age and education correlated between 0.00 (Trails A) and 0.14 (Speech Sounds) with T-scores corrected for age and education. The only test for which age and education were significantly related to T-scores was the Speech Sounds Test, accounting for 2% of the T-score variance. For all remaining tests, age and education were not significantly (p.05) related to T-scores.

7 HRNB Norms 451 the pattern of the HRNB profile by raising scores on the Aphasia test, regardless of the overall level of the profile. The Heaton et al. (1991) age education corrections for the HRNB are successful in minimizing the effects of age and educational level on HRNB profile level and pattern among nonnormal subjects. Although the correlation between age education and corrected HRNB profiles (0.295) was statistically significant, less than 1% of the variance in HRNB profiles could be predicted from age education in the present sample. Some small relationship between age education and corrected HRNB scores is to be expected among nonnormal subjects, since the Heaton-Grant-Matthews norms adjust only for the effects of normal aging and education and not for age education effects that are related to psychopathology or neuropathology. For example, some relationship between age and corrected HRNB scores is expected among nonnormal subjects because increasing age is related to the prevalence of dementia. Similarly, some relationship between education and corrected HRNB scores is expected among nonnormal subjects because educational attainment is curtailed by the onset of schizophrenia. Individual tests in the HRNB are also significantly affected by age and education among nonnormal subjects (see footnote 4). In the present sample, age and education were able to predict a significant amount of variance in uncorrected test scores. However, the Heaton et al. (1991) corrections for age and education were able to eliminate these relationships for all but one test. On the Speech Sounds test, age and education were able to account for 2% of the variance in corrected T-scores. More importantly, however, this 2% equates to only 2 T-score points, 5 which is of little practical significance in interpreting HRNB scores clinically. Use of Heaton et al. s (1991) standard scores and corrected T-scores is clearly an advance over the interpretation of HRNB raw scores. The transformation of raw scores to standard scores uses all the information in the raw score distributions rather than just scores above or below a raw cutoff value. Heaton et al. s (1991) corrected T-scores allow users to estimate the departure from average of HRNB test performances in a normal (i.e., nonneurological) population without age, educational level, and gender spuriously increasing or decreasing those estimates. Thus a corrected T-score of 30 indicates that the score is 2 standard deviations below that obtained by the average normal person, regardless of the person s age, educational level, or gender. Furthermore, comparisons can be made among the HRNB demographically corrected T-scores regarding deviations from their respective means in a normal population. If one T-score is 20 and another T-score is 40, one can say that the first is twice as deviant among normal subjects as the second. Furthermore, the Heaton et al. (1991) corrections are normalized T-scores and, therefore, can be interpreted directly in terms of their likelihood of occurrence in a normal population and can be compared directly in terms of percentiles. One can say that a T-score of 30 on one test is just as (un)likely to occur in a normal population as a T-score of 30 on another test. Both the amount of deviation from average and the likelihood of occurrence of different T-scores are equivalent across the multiple tests included in the HRNB. 6 5 The standard deviation of T-scores is defined as 10 and the variance of T-scores is thus 100. Two percent of the variance would equal 2 T-score points. 6 Heaton et al. (1991) presented in Table 7 a comparison of the actual percentages of occurrence of various T-score ranges in their sample with the percentages expected from the normal distribution. Although the actual values do not match expected values exactly for all tests, they are probably close enough to permit comparisons of T-scores between tests in terms of likelihood of occurrence in a normal population. For example, a T-score between 35 and 39 on the Aphasia Screening test included 10.5% of their normative group, while the same T-score range on the Seashore Rhythm test included 6.2% of their normative group. This difference of 4.3% in likelihood of occurrence in a normal population is unlikely to lead to erroneous clinical conclusions.

8 452 J. A. Moses, Jr., D. A. Pritchard, and R. L. Adams However, the Heaton et al. (1991) corrected T-scores cannot be interpreted and compared in terms of degrees of neuropsychological impairment. 7 A T-score of 30 does not mean equivalent degrees of impairment on different HRNB tests. Since the Heaton et al. corrections are based only on a sample of normal subjects, the resulting T-scores cannot be used directly to interpret or to compare degrees of impairment on different tests. 8 Russell (1987) described a procedure for developing norms for neuropsychological tests that allows interpretation and comparisons of tests scores in terms of relative degrees of impairment. He proposed that component scores in a neuropsychological battery be scaled in terms of a common reference scale that is based on scores from both normal and impaired subjects. The resulting scaled scores can then be interpreted in terms of deviations on a scale anchored at the low end by nonimpaired subjects and at the high end by the most impaired subjects. Future work on norm development for the HRNB should focus on demographically corrected scores that allow direct interpretation and comparisons in terms of neuropsychological impairment. It hardly seems cost-effective to spend hours collecting neuropsychological test data only to say that the scores are not normal or that the scores are unlikely to come from a normal person. With additional research it should be possible to interpret HRNB test scores directly in terms of both overall and relative degrees of neuropsychological impairment. Furthermore, such impairment scores may be more successful in predicting real-world impairments than scores based only on normal subjects. REFERENCES Alekoumbides, A., Charter, R., Adkins, T., & Seacat, G. (1987). The diagnosis of brain damage by the WAIS, WMS, and Reitan Battery utilizing standardized scores corrected for age and education. International Journal of Clincial Neuropsychology, 9, American Psychiatric Association. (1980). Diagnostic and statistical manual of mental disorders (3rd ed.). Washington, DC: Author. American Psychiatric Association. (1987). Diagnostic and statistical manual of mental disorders (3rd ed., Rev.). Washington, DC: Author. Bak, J., & Greene, R. (1980). Changes in neuropsychological functioning in an aging population. Journal of Consulting and Clinical Psychology, 48, Blusewicz, M., Dustman, R., Schenkenberg, T., & Beck, E. (1977). Neuropsychological correlates of chronic alcoholism and aging. Journal of Mental and Nervous Disease, 165, Boll, T., & Reitan, R. (1973). Effect of age on performance of the Trail Making Test. Perceptual and Motor Skills, 36, Bornstein, R. (1983). Relationship of age and education to neuropsychological performance in patients with symptomatic carotid artery disease. Journal of Clinical Psychology, 39, Heaton et al. (1991) presented in Table 8 a proposed classification of T-scores in terms of degree of impairment. However, they use the term impairment to mean the odds of occurrence in a normal population and not the severity of impaired performance. Thus, their increasing levels of impairment (mild, moderate, severe) signify decreasing odds that the T-score derived from a normal population and do not directly signify increasing levels of impairment in test performance. 8 Degrees of impairment may be inferred indirectly from Heaton-Grant-Matthews T-scores by demonstrating empirically that subjects who differ in levels or patterns of impairment earn differing T-scores. However, such indirect interpretations are less precise and less persuasive than direct scaling of impairment levels. For example, a demonstration that demented subjects earn lower (more impaired) T-scores on the Category Test than epileptic subjects justifies the inference that lower T-scores mean more impairment, but does not justify the inference that a T-score of 20 is twice as impaired as a T-score of 40.

9 HRNB Norms 453 Bornstein, R. (1985). Normative data on selected neuropsychological measures from a nonclinical sample. Journal of Clinical Psychology, 41, Bornstein, R., & Suga, L. (1988). Educational level and neuropsychological performance in health elderly subjects. Developmental Neuropsychology, 4, Cauthen, N. (1978). Normative data for the Tactual Performance Test. Journal of Clinical Psychology, 34, Charter, R., Adkins, T., Alekoumbides, A., & Seacat, G. (1987). Reliability of the WAIS, WMS, and Reitan Battery: Raw scores and standardized scores corrected for age and education. International Journal of Clinical Neuropsychology, 9, Corrigan, J., Agresti, A., & Hinkeldey, N. (1987). Psychometric characteristics of the Category Test: Replication and extension. Journal of Clinical Psychology, 43, Corrigan, J., & Hinkeldey, N. (1987). Relationships between Parts A and B of the Trail Making Test. Journal of Clinical Psychology, 43, D Elia, L., Boone, K., & Mitrushima, A. (1995). Handbook of normative data for neuropsychological assessment. New York: Oxford University Press. Elias, M., Podraza, A., Pierce, T., & Robbins, M. (1990). Determining neuropsychological cut scores for older, healthy adults. Experimental Aging Research, 16, Elias, M., Robbins, M., Walter, L., & Schultz, N. (1993). The influence of gender and age on Halstead-Reitan Neuropsychological Test Performance. Journal of Gerontology, 48, Ernst, J. (1988). Language, grip strength, sensory-perceptual, and receptive skills in a normal elderly sample. The Clinical Neuropsychologist, 2, Ernst, J., Warner, M., Townes, B., & Peel, J. (1987). Age group differences on neuropychological battery performance in a neuropsychiatric population: An international descriptive study with replications. Archives of Clinical Neuropsychology, 2, Finlayson, M., Johnson, K., & Reitan, R. (1977). Relationship of level of education to neuropsychological measures in brain-damaged and non-brain-damaged adults. Journal of Consulting and Clinical Psychology, 45, Goldstein, G., & Shelly, C. (1981). Does the right hemisphere age more rapidly than the left? Journal of Clinical Neuropsychology, 3, Heaton, R., Grant, C., & Matthews, C. (1991). Comprehensive norms for an expanded Halstead-Reitan Battery. Odessa, FL: Psychological Assessment Resources. Karzmark, P., Heaton, R., Grant, I., & Matthews, C. (1984). Use of demographic variables to predict overall level of performance on the Halstead-Reitan Battery. Journal of Consulting and Clinical Psychology, 52, Kupke, T., & Lewis, R. (1989). Relative influence of subject variables and neurological parameters on neuropsychological performance of adult seizure patients. Archives of Clinical Neuropsychology, 4, Long, C., & Klein, K. (1990). Decision strategies in neuropsychology: II. Determination of age effects on neuropsychological performance. Archives of Clinical Neuropsychology, 5, Reitan, R., & Wolfson, D. (1989). The Seashore Rhythm Test and brain functions. The Clinical Neuropsychologist, 3, Reitan, R., & Wolfson, D. (1995). Influence of age and education on neuropsychological test results. The Clinical Neuropsychologist, 9, Ruff, R., & Parker, S. (1993). Gender- and age-specific changes in motor speed and eye-hand coordination in adults: Normative values for the Finger Tapping and Grooved Pegboard tests. Perceptual and Motor Skills, 76, Russell, E. (1987). A reference scale method for constructing neuropsychological test batteries. Journal of Clinical and Experimental Neuropsychology, 9, Russell, E., Neuringer, C., & Goldstein, G. (1970). Assessment of brain damage: A neuropsychological key approach. New York: John Wiley. Schludermann, E., Schludermann, S., Merryman, P., & Brown, B. (1983). Halstead s studies in the neuropsychology of aging. Archives of Gerontology and Geriatrics, 2, Stanton, B. (1984). Age and educational differences on the Trail Making test and Wechsler memory Scales. Perceptual and Motor Skills, 58, Steinmeyer, C. (1986). A meta-analysis of Halstead-Reitan test performances of non-brain damaged subjects. Archives of Clinical Neuropsychology, 1, Stevens, J. (1992). Applied multivariate statistics for the social sciences (2nd ed.). Hillsdale, NJ: Erlbaum. Thompson, L., Heaton, R., Matthews, C., & Grant, I. (1987). Comparison of preferred and nonpreferred hand performance on four neuropsychologial motor tasks. The Clinical Neuropsychologist, 1, Vega, A., & Parsons, O. (1967). Cross-validation of the Halstead-Reitan Tests for brain damage. Journal of Consulting Psychology, 31, Warner, M., Ernst, J., Townes, B., & Peel, J. (1987). Relationships between IQ and neuropsychological mea-

10 454 J. A. Moses, Jr., D. A. Pritchard, and R. L. Adams sures in neuropsychiatric populations: Within-laboratory and cross-cultural replications using WAIS and WAIS-R. Journal of Clinical and Experimental Neuropsychology, 9, Willis, L., Yeo, R., Thomas, P., & Garry, P. (1988). Differential declines in cognitive function with aging: The possible role of health status. Developmental Neuropsychology, 4, Yeudall, L., Reddon, J., Gill, D., & Stefanyk, W. (1987). Normative data for the Halstead-Reitan neuropsychological tests stratified by age and sex. Journal of Clinical Psychology, 43,