4m // ill/WWW! :7 --nt-“ , ' 3 1293 I r- - - - £~-t.-s- - - a ‘ ma. v =— lur— --... I 79:.fl.‘ ,r"¢- l L.’.; {321-..}! W: This is to certify that the dissertation entitled Use of the WISC-R and the K-ABC With Low Achievers, Learning Disabled, and Students Classified Emotionally Impaired presented by Dennis James Becht has been accepted towards fulfillment of the requirements for Ph.D. degreein Counsel ingLEducational Psychology, and Special Education 7 \ J Majoflofessor Date M MS U it an Affirmative Action/Equal Opportunity Institution 042771 2 6219 RETURNING MATERIALS: 1V1531_J Place in book drop to LJBRARJES remove this checkout from #35. your record. FINES Will _,,, ~ be charged if book is returned after the date stamped below. 9m. o o 1293. "par. WW3 USE OF THE WISC-R AND 111E K-ABC WITH LW AQ-lIEVERS. LEARNING DISABLED. AND STUDENTS CLASSIFIED EMOTIONALLY IMPAIRED By Dennis James Becht A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requiranents for the degree of DOCTOR OF PHILOSOPHY Department of Counseling. Educational Psychology. and Special Education 1986 Copyright by DENNIS JAMES BECHT 1986 AB STRACI‘ USE OF THE WISC-R AND THE K-ABC WITH LCM ACHIEVERS. LEARNING DISABLED. AND STUDENTS (LASSIFIED EMOTIONALLY IMPAIRED By Dennis James Becht The ability and achievement subtest scores of 102 students from a number of Michigan school districts were analyzed to determine the validity of diagnostic methods and to test the utility of theorized characteristic group patterns for classification accuracy. The children had all been referred for psychological evaluation and were categorized as learning disabled (51). emotionally impaired (20). or not qualified for special education services but designated as low achievers on the basis of their achievement scores on word recognition (31). In the literature. various diagnostic methods and subtest score patterns have been proposed as useful for distinguishing among diagnostic groups. Ability-achievement discrepancy. Verbal- Performance. and Sequential-Simultaneous ability differences; abnormal subtest scatter; and general ability and achievement performance have been proposed as useful for differentiating among groups. School- identified classifications served as the reference criterion. To the extent methods affirmed school-identified classifications. they were Dennis James Becht considered useful for differential diagnosis. Eighty-five percent agreement was the level accepted in this study as evidence of diagnos- tic utility. None of the methods was supported as useful in the present investigation. A large percentage of students are misclas- sified using these methods. A major assumption of the current investigation was that the classification decisions of Educational Planning and Placement Commit- tees were accurate. However. it did not appear that placement commit- tees adhered to federal guidelines. particularly with LD students. Pew clues emerged as to the decision systems employed by schools in classi- fying students. The results. pertaining to the usefulness of the diagnostic methods. may be confounded by the failure of school commit- tees to consistently apply the guidelines: and the validity of the criterion (school-identified classifications) is questioned for the same reason. A reduction in the validity of the criterion may under- estimate the ability of the diagnostic methods to differentiate mean- ingfully among the groups. The implications of these findings are discussed and related to labeling and placement practices. Recommendations for improving the decision systems of educational planning and placement committees are suggested. ACKNWLEDGMENTS The completion of this dissertation reflects the contribution of many individuals who have provided me with guidance. concern. and support in innumerable ways. Harvey Clarizio. my advisor and dissertation chairman. has served as a model for my professional development since the beginning of my doctoral program. His encouragement and guiding questions contributed immeasurably to the refinement of my research and clinical skills. Andrew Porter helped me to synthesize the seemingly unrelated and abstruse statistical and research design concepts. His patient teaching and precise examples contributed significantly to my understanding and to the successful completion of this study. Walter Hapkiewicz took the time to thoroughly discuss concerns and questions I had not considered. His suggestions directed me to issues that critically impinge on the results of this investigation. Martha Karson served in a dual role as my internship supervisor and as a member of my doctoral committee. In both of these roles she has added significantly to my personal and professional growth. Her clinical insights helped me to work through my concerns surrounding the completion of this project. ‘Her advice and encouragement have contributed to my professional confidence. A friend who contributed his time and advice deserves mention. Richard Coehelo read and discussed the initial drafts. giving me the benefit of his own research experience. His insightful comments helped clarify my research goals. Simply thanking Sue Cooley for her expert typing and editing assistance seems insufficient. Her patience in deciphering my handwritten drafts and her care and timeliness in completing my work significantly reduced the stress in completing this study. I want her to know that I have appreciated her support and the large contribution she has made in helping me complete the final requirement for my degree. My parents. Charles and Evelyn Becht. have contributed in so many ways that to thank them for everything would double the size of this manuscript. Their enduring love. support. and frequent financial assistance sustained me during both happy and trying times. Not only have they helped me to complete my formal education. but they have taught me. by their example. values that I shall strive for throughout my personal and professional life. I hope that in some small way these words convey to my parents the love. respect. and appreciation I have for them. My brother. Barry Becht. and my sister. Donna Baier. contributed through their moral support. Both seemed to know intuitively when to encourage me and‘when to say nothing at all. I wish to thank my in-laws. Lauro and Olivia Teran. who have always bad faith in me. They understood the many competing demands on vi my time and never complained about infrequent or shortened visits home. Their loving support allowed me to remain single-minded in this endeavor. Terri. my wife. has always been in my corner. She. more than anyone else. has made the completion of this dissertation possible. Her willingness to defer her own priorities helped me to complete this 'work. She listened to and discussed the initial drafts and offered suggestions that helped me to fully develop key points and clarify ideas. Her natural curiosity. together with her insightful comments. challenged me to think and write more clearly. Her unselfish giving of her time helped me through.the most difficult times and exemplified the true meaning of love. Despite my protests. she always seemed to know that I would finish. Together. with mutual love and respect. we have completed it for each other. vii LIST OF Chapter I. II. III. TABLE OF CDNTENTS TEL ES 0 O O O O O O O O O O O O O O O O 0 wow u ION O I O O O O O O O O O O O O O The Testing Controversy . . . . . . . . . . . . . . Academic Disabilities . . . . . . . . . . . . . Similarities Across the Mild-Handicapping Conditions Educational Diagnosis . . . . . . . . . The Elusive Definition of Learning Disabilities The Search for Characteristic Test Profiles Learning Disabilities or Low Achievement . . . . Synopsis of the Present Investigation . Organization of the Study . . . . . . . ORGANIZATION AND DESCRIPTION OF THE KAUFMAN ASSESSMENT BATTERY FOR CHILDREN . . . . Overview . . . . . . . . . . . . . . . . Introduction to the Scale . . . . . Global Scales . . .. . .. . .. The Subtests .. . .. . .. . . Standardization Procedures . . . . The Manuals . . . . . . . . . . Administration and Scoring . . . Types of Scores . . . . . . . Reliability. Validity. Standard Error of REViWSOfthEK’ABCesooeooooo LITERATURE REVIEW . . . . . . . . . . . . Measurement. and Brief Comparison With the WISC-R . . . . . . . mania O O O I I O O O O O O I O O O O C O O O C 0 Learning Disabled Versus Low Achievers: Psychometric cmparisona O O O O O O I O O O O O O O O O O O O Ability-Achievement Difference: Discrepancy Model . viii Page xi H UHOQO‘bUNF—J P'P‘ 14 14 15 17 25 27 27 29 30 33 41 41 42 47 IV. V. Problems With the Discrepancy Approach . . . . . . Discrepancy Models and Statistical Reliability . . Discriminative Efficiency of the Discrepancy Model Verbal-Performance Ability Differences . . . . . . . Sequential-Simultaneous Ability Differences Ability Subtest Scatter Comparison . . . . . . . . General Ability and Achievement Characteristics Verbal-Performance Differences and Emotional Impai ment 0 O O O O O O O O O O O O O O O O O I Sequential-Simultaneous Differences and Emotional Impaiment O O O O O O I O O O C O O O O O I C O Emotional Impairment: K-ABC Mental-Processing Subtest Patterns . . . . . . . . . . . . Emotional Impairment: K-ABC Achievement . Emotional Impairment: High Similarities. Low Information 0 O O O O O O O O O O O O O Direction of the Present Study . . . . . . “men 0 I O O O O O O O O O O O O O O O O O Source of Data . . . . . . . . . . . . . . smpl e O O O O O I O O O O O O O O O O O Rationale for Using Reading Decoding to Identify Low Achievers . . . . . . . . . . . . . Establishing Decision Rules . . . . . . . The Utility of Diagnostic Decision Rules Hypotheses . . . . . . . . . . . . . . . . RESULTS 0 C O O O O O O O O O O O O O O O O mania O I O O O O O I O O O O O O O O O Ability-Achievement Discrepancy . . . . . Comparisons of Verbal. Performance. and Full-Scale Scores Across the Four Groups . . . . . . . . . Verbal-Performance Discrepancy . . . . . . r . . Comparisons of K-ABC Sequential. Simultaneous. and Mental Processing Composite Scores Across Groups Sequential-Simultaneous Discrepancy . . . . . . Ability-Test Subtest Scatter Comparisons Across Group83WISC-R................. Ability-Test Subtest Scatter Comparisons: K-ABC Achievement Test Comparisons: Peabody Individual AChievment Test 0 O O O O O O O O O O O O O O O Achievement Test Comparisons: K-ABC Achievement Scale 0 O O O O O O O O O O O O O O O O O O O O 0 High Similarities. Low Information: WISC-R Pattern- Notional Impairment o o o o o o o s s e e e o e 0 ix Page 48 50 58 61 64 69 71 75 77 78 78 79 84 84 85 91 92 94 95 100 100 101 106 109 111 114 116 119 121 123 126 K-ABC Patterns: Emotional Impairment . . . . . . . . . WISC-R Subtests and Classification of Students . . . . K-ABC Subtests and Classification of Students . . . . Classification Accuracy: Comparison of the R-ABC and the‘WISC-R . . . . . . . . . . . . . . . . . . . . Overlapping Effect: Classification of LD Children Using the X—ABC and the WISC-R Together . . . General Findings: Black Students . . . . . . . . VI. DIstSION O O O O O O O O O O O O O O O O O O O 0 Overview . . . . . . . . . . . . . . . . . . . . Summary and Utility Conclusions . . . . . . . . Disagreement Between School Identifications and Federal Identifications: Classification Labels LD: The Advantages of the Label . . . . Recommendations . . . . . . . . . . . . Recommendations for Clinical Practice Public-Policy Recommendations . . . . Research Recommendations . . . . . . . APPENDICES A. WISC-R AND KPABC RELIABILITY TABLES . . . . . . . B. COURSE DESCRIPTION AND OBJECTIVES . . . . . . . . C. EXAMINER QUESTIONNAIRE . . . . . . . . . . . . . . D. DATA FORM . . . . . . . . . . . . . . . . . . . . “mm CBS 0 C O O O O O O O O O O O O O O O O O O O O O O Page 131 134 135 136 138 141 143 143 143 147 154 158 158 160 161 163 174 175 177 179 Table 2.1 2.2 3.1 4.1 4.2 5.1 5.2 5.3 5.4 5.5 5.6 5.7 5.8 LIST OF TABLES Nonverbal Scale Subtests . . . . . . . . . . . . . . . K‘ABC Subtests and IntendEd A828 0 s o o o e e e s e s Salient Characteristics of Nine Procedures for Cbmputing an Ability-Achievement Discrepancy Score . . . . . . Examiner Characteristics . . . . . . . . . . . . . . . Sample Characteristics . . . . . . . . . . . . . . . . Analysis of Variance of Mean Discrepancy Scores by Group: WISC-R PSIQ - PIAT Reading Recognition . . . . Analysis of Variance of Mean Discrepancy Scores by Group: R-ABC (MPC) - K—ABC Reading Decoding . . . . . Numbers and Percentages of Children Classified Using Three Operational Discrepancy Levels . . . . . . . . Numbers of Students Classified as LD and Not LD by the Federal Guidelines in Comparison to School Placement DECiBions O C O O O O O O O O O O O O O O O I O O O O g-Ratios and Probability Levels for Significant Differences Between.WISC-R Ability Scores of LD and hCh 0f the Other Groups 0 O O O O O O O O O O O O 0 Summary of the Use of WISC-R Ability Scores to Classify students 0 O O O O I O O O O O O O O O O O O O O O O Verbal-Performance Discrepancies for LD. EI. LA < 33. and LA < 17 Children When Direction of Difference Is Considered . . . . . . . . . . . . . . . . . . . . ErRatios and Probability Levels for Significant Differences Between R-ABC Ability Scores of LD and Each of the Other Groups . . . . . . . . . . . . . . xi Page 17 19 51 86 89 103 103 105 107 108 109 111 112 5.9 5.10 5.11 5.12 5.13 5.14 5.15 5.16 5.17 5.18 5.19 5.20 5.21 5.22 Summary of the Use of K-ABC Ability Scores to Classify Students . . . . . Sequential-Simultaneous Discrepancies for LD. EI. LA < 33. and LA < 17 Children (Directional- Simultaneous > Sequential) . . . . . . . . . . . . . . Simultaneous-Sequential Discrepancies for LD. EI. LA < 33. and LA < 17 Children (Directional- Simultaneous > Sequential) . . . . . . . . . . . . . . Group Means for NDEV Verbal and Performance . . . . . . Numbers of Students Classified as LD and Not LD by the NDEV Method in Comparison to School Placenent Decisions Using the WISC-R . . . . . . . . . . . . . . Group Means for NDEV-Mental Processing . . . . . . . . Numbers of Students Classified as LD and Not LD by the NDEV Method in Comparison to School Placement Decisions Using the K-ABC E-Ratios and Probability Levels for Differences Between PIAT Achievanent Scores for LD and Each of the Other Groups 0 O O O O O O O O O O O O O O O O O O O 0 Classification of Students Using PIAT Achievanent Scores g-Ratios and Probability Levels for Differences Between K-ABC Achievement Scores of LD. EI. LA <17 Children . . . . LA < 33. and Classification of Students Using K-ABC Achievement Scores 0 O O O O O O O O O O O O O O O O O C O O O O 0 High Similarities-Low Information Pattern: Comparison of LD. EI. LA < 33. and LA < 17 Groups (Dean' a criterion) 0 O O I O O O O O O O O O O O O O O O O O 0 High Similarities-Low Information Pattern: Comparison of LD. EI. LA < 33. and LA < 17 Groups (3-Point Difference-Stringent Criteria) Numbers of Students Classified as 81 and Not EI by Dean's Criteria in Canparison to School Placanent Decisions . Page 113 115 116 117 119 120 121 122 124 125 127 128 130 131 5.23 5.24 5.25 5.26 A.1 A.2 A.3 A.4 A.5 A.6 A.7 A.8 A.9 Patterns for Emotionally Impaired: K-ABC High Number Recall. Low Spatial Memory Pattern . . . . . . . . . . . Classification of LD. EI. LA.< 33. and LA.< 17 Children Using‘WISC-R Subtest Scores . . . . . . . . . . . . . . Classification of LD. EI. LA.< 33. and LA < 17 Children UBiDSK-ABCSUbteataooeeooesooooooosso Summary of the Methods Used to Classify Students: Comparisons of Correct Decisions by KrABC‘Versus “Isc-Rsossooosoeooeeosuccesses. Internal Consistency: Split-Half Reliability Coefficients for the K-ABC Subtests. by Age. for the Standardization Sample. Ages 2-1/2 Through 12-1/2 e o o o o o o o e a 0 Internal Consistency: Reliability Coefficients for the K-ABC Global Scales. by Age. for the Standardization Sample. Ages 2-1/2 Through 12-1/2 . . . . . . . . . . . Test-Retest Reliability Coefficients for the KrABC subtests O O I O I O O O O O O O C I O O O O O O O O O O Test-Retest Reliability Coefficients for the K-ABC G1 Obd seal e8 0 O O O O O O O O C O O O I O O O O O O O WISC-R Reliability Coefficients of the Tests and IQ sca1 es. by Age 0 O O O O O O O O O O O O O O O O O O O 0 Stability Coefficients of the WISC—R Tests and 108 for Three Groups of Children Tested Twice . . . . . . . . . Standard Errors of Measurement for the K-ABC Subtests. by Age. for the Standardization Sample. Ages 2—1/2 Through 12-1, 2 O O O O O O O O O O O O O O O O O O O O 0 Standard Errors of Measurement for the K—ABC Global Scales. by Age. for the Standardization Sample. AgesZ-l/ZThroughlZ-l/z 00000000000000. WISC-R Standard Errors of Measurement of the Scaled scores and IQB. by Age 0 O O O O O O O O O O O O O O O O xiii Page 134 135 136 138 163 165 166 167 168 169 170 172 173 CHAPTER I INTRODUCTION The Testing:Controversy The use of standardized psychological tests to assist in the diagnosis and placement of handicapped students has in recentjyears become a controversial issue of considerable concern to psychologists. educators. and the courts (Bersoff. 1982; Reynolds. 1982). Intelli- gence tests have been singled out as particularly objectionable. Dur- ing the past 25 years. more attention has been directed to their use in the assessment process as examiners and consumers alike have challenged classification decisions made in part on the basis of intelligence test scores. Heated debates between and among professionals and consumers have resulted from differences of opinion about the efficacy of intelligence tests for selection and classification of handicapped children. With increasing frequency. these disagreements have been channeled through the court system for resolution. Legislation regarding assessment practices has come about as an extension of the courts' assuming greater responsibility vis-a-vis the regulation of testing and the adjudication of educational placement decisions (Bersoff. 1982). Notwithstanding the disagreement and con- flicting views of many psychologists and measurement experts. the outcome of various court cases has influenced assessment procedures in general. and the practice of school psychology in particular. No longer solely under the auspices of professional regulating associations such as the American Psychological Association and the National Association of School Psychologists. the testing field has moved into an era of increased governmental control. Federal rules and regulations have been established to monitor the assessment and placement of all handicapped children. Definitional criteria for edu- cational categories have been instituted. and guidelines for multi- disciplinary assessment procedures have been mandated. Academic Disabilities Although characteristics of specific deficits are described in legal documents such as P.L. 94-142. the major responsibility for interpreting intelligence and achievement test scores within the con- text of the law is borne by school psychologists. Selection of the correct educational placement category for severely impaired students is a relatively easy and straightforward procedure because their handicapping conditions are observable even to nonprofessionals. Deafness. blindness. moderate to profound levels of retardation. and psychotic disorders are examples of categories readily identified by their physical and behavioral sequalae even before formal tests are administered. In these referral situations the responsibil- ity of the psychologist is to confirm diagnoses (previously made by physicians or other health care professionals) and to recommend envi- ronmental/behavioral interventions to assist primary caregivers and special education professionals in designing appropriate educational supports. Federal guidelines describing the diagnostic criteria for this small percentage of special students are adequate. However. the vast majority of referrals made to school psychologists involve categories of educational handicaps that are relatively mild compared to those described in the previous paragraph. Educational problems such as suspected learning disabilities. mild emotional impairment. or border- line mental retardation constitute the largest percentage of referrals and are also the handicapping disabilities that are the most confusing to identify because of similar characteristics and ambiguous diagnostic criteria. Similarities Across the Mild—HandicappingTConditions Children with mild academic/educational problems have one common characteristic that brings them to the attention of school psycholo- gists-low academic.achievement (Algozzine & Ysseldyke. 1981). Con- sequently. differential classification of specific disorders that dis- tinguish one child's educational problems from that of another is often a puzzling task. Yet. federal guidelines require that exceptional children receive specific. individualized educational intervention appropriate to their handicapping condition. This implies that chil- dren with academic deficiencies can and must be differentially classi- fied for treatment. The implication is that categorical placement/ treatment is matched to specific measurable disability-trait characteristics. On the premise that special categories of exception- ality (i.e.. learning disabled. emotionally impaired. educable mentally impaired) will benefit students so categorized (Shinn. Ysseldyke. Deno. & Tindal. 1982). the educational planning and placement committee must recommend appropriate placement categories for each child tested. Relatedly. inappropriate categorization (mislabeling a child) is considered a disservice that may result in either denial of services to some students in need of special education. or restriction of freedom (pull out from general education) for others who are mistakenly labeled as handicapped. Classification by handicap and the provision for homogeneous disability groupings are based on theory suggesting that such an arrangement makes it possible better to teach compensatory techniques (Wiseman. 1984). which in turn results in more efficient remediation of academic deficiencies and expeditious return to regular education placement . Educational Diggnosis Deciding that a child is eligible for special education may have important ramifications. both positive and negative. for the child's educational and career future. Relatedly. determining that a child is ineligible is an important negative decision in the sense that special- ized services for academic achievement will be denied the child who does not qualify for a specific disability (Ysseldyke. Algozzine. Shinn. & McGue. 1982). A major assumption of the federal guidelines has been that par- ticular groups of children can be validly classified by disability and that categorical special educational interventions can be implemented to remediate specific deficits. The guidelines outline definitional and diagnostic criteria for each specific disability. Standardized test results have been and continue to be used to identify the putative characteristics that distinguish particular handicapping conditions. Psychologists. because of their specialized training in psychometrics. have the unenviable task of sorting children into disability groupings on the basis of their test scores. Differential diagnosis of specific mildly handicapping disabilities is extremely difficult because overlap in intelligence test scores. ability patterns. and academic achievement make mildly handicapped groups as similar as they are different (Morri- son. MacMillan. & Kavale. 1985). The learning disability category. perhaps more than any other. is the most difficult to diagnose differentially from other mild handi- capping conditions because the definition has been described and opera- tionalized in so many diverse ways (Tucker. Stevens. & Ysseldyke. 1983). Whereas learning disabilities can be differentiated from mental retardation (Clarizio & Bernard. 1981; Frame. Clarizio. Porter. & Vinsonhaler. 1982) and from nondisabled or normal populations (Clarizio 8: Bernard. 1981; Naglieri. 1985). differentiating learning disabilities from low-achieving and emotionally impaired groupings has not been demonstrated in the literature (Algozzine & Ysseldyke. 1982; Mann. Davis. Boyer. Metz. and Wolford. 1983; Shinn et al.. 1982; Wiseman. 1984; Ysseldyke et al.. 1982; Ysseldyke & Algozzine. 1983). The Elusive Definition of Learninflisabilities The concept of "learning disabilities" has been recognized for more than 40 years but was not coined as such until the early 19603 by Kirk. This term was legitimized about the same time by the federal government when it legislated diagnostic rules and regulations to formally establish and recognize "learning disabilities" as a specific handicapping condition. The guidelines maintained Kirk's original conception of the term. which defined learning disabled children as a subset of students with essentially normal intelligence who fail to make adequate progress in school. However. the federal rules also included characteristic standards for diagnosticians to follow in establishing or determining a child to be learning disabled. The rules state that a multidisciplinary team may determine that a child has a specific learning disability if the child exhibits a disorder in pay- chological processes. does not achieve commensurate with age and abil- ity levels in one or more areas of achievement. and has a severe discrepancy between achievement and intellectual ability. These stand- ards continue to be used by educational planning and placement commit- tees to establish the presence of a learning disability. However. identification of a processing disorder is no longer a mandatory requirement (Wiseman. 1984). and criteria for identifying learning disabled (LD) children largely omit it and concentrate on discrepant achievement as the major identification variable (Algozzine. Forgnone. Mercer. & Trifiletti. 1979; Ysseldyke. Algozzine. Shinn. & McGue. 1982) . A problem with the guidelines was and continues to be the fact that the extent or severity of low achievement is not specified. even though "severe discrepancy" is critical to identifying this handicap (Algozzine & Ysseldyke. 1982). The interpretation of "severe discrep~ ancy" was left to the discretion of state and local education agencies when the federal government abdicated its decision-making powers in 1976. Inconsistencies have occurred as different school districts operationalized and emphasized different definitional components for labeling children as LD. This has led to a failure to identify charac- teristics universal and.specific to learning disabilities.*which.in turn contributes to the heterogeneity of the population and results in the bewildering conclusion that there is no consensus concerning a definition in the field (Mann et al.. 1983; Morrison et al.. 1985; Norman & Zigmond. 1980; Tucker et al.. 1983; Ysseldyke 8: Algozzine. 1983). Myklebust (1983) questioned the validity of the concept because of the instability of the learning disabilities category across examiners and across settings. Mann (1983). reacting to the loose definitional characteristics. indicated that learning disabilities should be consid- ered a generic term for "mildly" handicapped because it currently constitutes the single largest handicapping group in the country. Relatedly. Algozzine and Ysseldyke (1982. 1983) referred to learning disabilities as a category of underachievement and indicated that the "hit rate" or correct differential diagnosis between learning disabili- ties and low achievement remains at approximately 501 or chance levels. A problem with the guidelines was and continues to be the fact that the extent or severity of low achievement is not specified. even though "severe discrepancy" is critical to identifying this handicap (Algozzine & Ysseldyke. 1982). The interpretation of "severe discrep- ancy" was left to the discretion of state and local education agencies when the federal government abdicated its decision-making powers in 1976. Inconsistencies have occurred as different school districts operationalized and emphasized different definitional components for labeling children as LD. This has led to a failure to identify charac- teristics universal and‘specific to learning disabilities.'which in turn contributes to the heterogeneity of the population and results in the bewildering conclusion that there is no consensus concerning a definition in the field (Mann et al.. 1983; Morrison et al.. 1985: Norman & Zigmond. 1980; Tucker et al.. 1983; Ysseldyke & Algozzine. 1983). Myklebust (1983) questioned the validity of the concept because of the instability of the learning disabilities category across examiners and across settings. Mann (1983). reacting to the loose definitional characteristics. indicated that learning disabilities should be consid- ered a generic term for "mildly" handicapped because it currently constitutes the single largest handicapping group in the country. Relatedly. Algozzine and Ysseldyke (1982. 1983) referred to learning disabilities as a category of underachievement and indicated that the ”hit rate" or correct differential diagnosis between learning disabili- ties and low achievement remains at approximately 502 or chance levels. rate for her groups of LD. emotionally impaired (EI). educable mentally impaired (EMI). and normal children using the WISC-R. The Search for Characteristic Test Profiles Early ideas in the field of learning disabilities suggested that a group of children who were having difficulty succeeding in school due to a deficiency in cognitive/psychological processes (which was not related to mental retardation or physical handicaps) could readily be identified by intelligence tests (Veres. 1982). The underlying assump- tion was that patterns of relative strengths and weaknesses in test profiles would help distinguish LD children from those with other handicapping conditions. as well as a larger population of children who were doing poorly in school because of low achievement (Wiseman. 1984). Similarities in the patterns of strengths and weaknesses on stand- ardized ability tests. particularly the Wechsler Intelligence Scales (WISC and WISC-R), have been investigated in various studies in hopes of identifying important learning characteristics and test score pat- terns that would facilitate differential diagnosis of educational han- dicaps. Examination of verbal-performance discrepancy. subtest scatter. ability-achievement discrepancy. and recategorized scores (Bannatyne's verbal-conceptual. spatial. sequential. and acquired knowledge scores; Kaufman's factors: verbal. perceptual-organization. and freedom from distractibility) has revealed little credible support that any linear combination of subtest scores will aid in differential diagnosis (Veres. 1982). Although statistically significant group differences have been found to support some score differences between LD and mentally retarded (MR) and between exceptional populations and "normal" groups. the differences were not considered meaningful or practical for individual diagnostic situations. Evidence against the validity of diagnostic patterns has been reported by a number of writ- ers (Clarizio & Bernard. 1981; Gutkin. 1979; Henry & Wittman. 1981; Rychman. 1981; Schooler. Beebe. & Koepke. 1978; Thompson. 1981). Based on the majority of the investigations. the clinical efficiency of WISC- R subtest patterns for differential diagnosis of learning disabilities as well as other educational handicaps has not been supported. Veres (1982) summarized the research literature in this area by stating: "The use of test score patterns for differential diagnosis is. from a practical standpoint. not much better than a random process" (p. 70). Learninggisabilities or Low Achievement Although there have been fairly consistent findings that learning disabilities and other exceptional classifications can be discriminated from "normals" on the basis of test scores. this is not the proper comparison for demonstrating diagnostic utility. According to Meehl and Rosen (1955). if one only wishes to discriminate from normal popu- lations. base rates of LD are better than tests; but if a test is to be used for differential diagnosis. evidence of its efficiency for dis- crimination cannot be established solely on the basis of distinguishing diagnostic groups from normals. To establish diagnostic utility or the clinical efficiency of a test instrument. it is necessary to demon- strate that score patterns differentiate a higher percentage of LD from 10 other diagnostic groups (emotionally impaired. low achieving) than could be correctly classified without any test at all. Assuming that Ysseldyke et a1. (1982) and others have been correct in suggesting that learning disabilities is primarily a category of underachievement. the appropriate diagnostic differentiation to make in clinical practice is between LD and low-achieving students. If Sexton and Street (1985) were correct in their estimates. nearly 401 of the children referred for evaluation fail to qualify for special education. Considering that these children had problems warranting a referral for evaluation. the importance of differentiating the LD from the low- achieving student who has been adjudged not handicapped becomes criti- cal. Indeed. the similarities and differences between LD children and those who are not LD but who are doing poorly in school are important to consider in determining eligibility for learning disability serv- ices. Due to the shared primary characteristic of low achievement asso- ciated with mildly handicapped populations. considerable confusion and misclassification occur in trying to differentiate among LD. EI. and low-achieving children. Brenton and Gilmore (1976) estimated a classi- fication error rate between 252 and 332 in their study of LD and low- achieving groups. Ysseldyke and Algozzine (1979) and Ysseldyke et al. (1982) determined from their studies that diagnostic errors‘were preva- lent. yielding an overall misclassification rate as high as 40% for LD and low-achieving children. The literature on this matter is reviewed in the next chapter. For now. it is sufficient to suggest that current 11 procedures employed to differentiate LD students from other mild handi- capping conditions and from low-achieving students seem to have little diagnostic utility. at least when compared to school-identified classi- fications. Since none of the above studies was successful in identi- fying the decision-making factors employed by placement committees. it could also be suggested that the criterion (school-identified classifi- cations) has little validity. Until placement team decision systems are uniformly organized. fair evaluation of the utility of diagnostic methods may not be possible. Synopsis of the Present Investigation The diagnostic utility of using ability and achievement test scores to differentiate the LD from the EI or from the low-achieving child who has been adjudged not eligible for special education services is not yet established. The empirical evidence currently available from a number of investigations with the WISC—R suggests that differ- ences in test-score performance of mildly handicapped children will not yield information clinically useful for differential diagnosis. Profile analysis using ability-achievement discrepancy and operation- alizations of the federal definition. general ability and achievement test performance. differences in processing styles. (verbal-nonverbal; simultaneous-sequential). and analysis of scatter may yield few differ- ences between groups so that the use of such methods for differential diagnosis is equivocal and perhaps useless. The primary concern of the present investigation was to determine if subtest scores on the WISC-R and the K-ABC could be used for 12 identifying and differentiating LD students from El students without misdiagnosing or assigning low-achieving students to either category. Comparison of educationally handicapped children with a low-achieving sample was considered a logical and necessary test since low academic achievement (as previously discussed) is a shared characteristic across referred children. The practical utility of the conjectured profile patterns for differential diagnosis was the leading focus of the present study. The analyses in the study included calculation of the proportions of students demonstrating the speculated characteristic test patterns/profiles. The probabilities of incorrect diagnostic decisions were estimated for each presumed distinctive test pattern or diagnostic method. School-identified classifications served as the external criterion. A number of studies similar to the present investigation may help determine which methods and operationalizations of the federal defini- tion are most accurate for distinguishing,LD from EI and from a larger group of not handicapped but low-achieving children. The degree to which the methods affirm school-identified diagnoses may provide an indication of the extent to which multidisciplinary teams adhere to state and federal guidelines and use test score information in decision-making processes. The results may help to determine the factors multidisciplinary teams consider important for categorizing children. l3 Qgggnization of the Study In Chapter II. a brief descriptive overview of the Kaufman.Assess- ment Battery for Children is presented. Chapter III contains a review of attempts to use test score patterns and operationalizations of the federal definition to differentially diagnose learning disabilities as a distinct diagnostic entity separate from other mild handicapping conditions (EI). as well as from nonhandicapped children who are low achieving. Described in Chapter IV is the design of the study. In Chapters V and VI. the results and the discussion summary are pre- sented. GIAPTER II ORGANIZATION AND DESCRIPTION OF THE KAUFMAN ASSESSMENT BATTERY FOR CHILDREN Overview The preceding chapter provided a descriptive history of attempts to use‘WISC-R.score patterns and the federal definition to validate learning disabilities as a distinct diagnostic category separate from other mild handicapping conditions. as well as from nonhandicapped children who are low achieving. In addition a proposal to use a new intelligence test the Kaufman Assessment Battery for Children (K-ABCL. in the same manner as described above. was discussed. A brief descrip- tion of this new intelligence test is presented in this chapter. as well as a summary of the literature reviewing the psychometric and theoretical features of this new instrument. Readers who are familiar with the K-ABC may wish to skip this chapter and proceed immediately to Chapter III for a review of the literature associated with attempts to verify the existence of learning disabilities as a separate diagnostic category. Introduction to the Scale The K-ABC. published in 1983. is an individually administered multisubtest battery designed to measure intelligence and achievement 14 15 of preschool and elementary school children between the ages of 2-1/ 2 and 12-1/2. It was standardized on 2.000 children. 100 at each half year of age between 2 years. 6 months. and 12 years. 5 months. The sample was stratified on the variables of sex. parental education. race or ethnic group (white. black. Hispanic. other). geographical region. community size. and educational placement (Kaufman & Kaufman. 1983). Two major divisions of the test are the Mental Processing Scales and the Achievement Scales. which together yield four Global Scales. each having standard scores with a mean of 100 and a standard deviation of 15. The test is described by the authors as being intended for psycho- logical and clinical assessment. psychoeducational evaluation of learn- ing disabled (LD) and other exceptional children. educational planning and placement. minority-group assessment. preschool assessment. neuro- psychological assessment. and research. Global Scales Scores are attained in four global areas: Sequential Processing. Simultaneous Processing. the Mental Processing Composite (combination of Sequential and Simultaneous Processing). and Achievement. These four scales provide a summary of the child's overall test performance. The Sequential Processing Scale includes three subtests. which present problems requiring sequential or serial processing of stimuli for efficient solution of the problem. Conversely. the Simultaneous Processing Scale comprises seven subtests requiring simultaneous. holistic problem-solving skills or the ability to process many stimuli at once for efficient solution of the problem materials. The Mental l6 Processing Composite is a combination of Sequential and Simultaneous Processing and is considered analogous to a measure of total intellec- tual functioning or problem-solving ability. Six subtests are included in the Achievement Scale. which is considered to assess factual knowledge. skills. and language concepts usually previously acquired in school settings and the home-community environment. In addition to the aforementioned environmental consid- erations. successful performance also depends on motivation and other nonintellectual variables. In addition to these four regular scales. there is a special Nonverbal Scale made up of selected K-ABC subtests. which is suggested as providing fair evaluation of hearing-impaired. speech and language- disordered. and non-English-speaking children. It is actually a short- ened form of the Mental Processing Scales which can be administered by the examiner in pantomime and responded to motorically. Reportedly. it correlates well with total test score and yields the same standard score units as are used with the complete battery to facilitate com- parison. The Nonverbal Scale provides an estimate of intellectual functioning for children in the 4- to 12 l/Z-year-old age range who exhibit communication disorders. Because of the smaller number of subtests used. the Nonverbal Scale does not yield separate Sequential and Simultaneous Processing scores. Table 2.1 shows the subtests that constitute the Nonverbal Scale for different age groups. 17 Table 2.1: Nonverbal Scale Subtests Age 4 Years 5 Years 6 to 12 1/2 Years Face Recognition Hand Movements Hand Movements Hand Movements Triangles Triangles Triangles Matrix Analo- Matrix Analogies g1es Spatial Memory Spatial Memory Photo Series Source: Kaufman & Kaufman. 1983. Although the K-ABC test includes 16 subtests (10 on the Mental Processing Scales and 6 on the Achievement). no child is administered more than 13. The breakdown of the number of tests given to children of different ages is as follows: 7 subtests at age 2-1/2. nine at age 3. 11 at ages 4 and 5. 12 at age 6. and 13 between the ages of 7 and 12—1/2. Predictably. the time required to administer the test increases with age. taking approximately 35 minutes at age 2-1/2. 50 to 60 minutes at age 5. and 75 to 80 minutes with children 7 years of age and above (Kaufman 8- Kaufman. 1983). The Subtests Scaled scores on the 10 subtests of the Mental Processing Scales each have a mean of 10 and a standard deviation of 3. whereas the six 18 subtests of the Achievement Scale each have a mean of 100 and a stand- ard deviation of 15. The Achievement Scale was set at the values mentioned above to facilitate ready comparison with the KEABC global processing scales or other conventional IQ scales for aptitude- achievement comparisons. Table 2.2 contains a brief description of the 16 subtests as discussed in the K-ABC Interpretive Manual and the intended age ranges for each subtest. Subtest 1. Magic Window (Simultaneous Processing78cale._ages 2-6 thrgggh 4-11).-This subtest measures a child's ability to identify and name an object whose picture is rotated behind a narrow slit. so that the picture is only partially exposed at any time. Magic Window was discovered to be the best measure of Simultaneous Processing for pre- school children in a factor analysis of the K-ABC data. Kaufman (1983) described this subtest. saying it reinforces the premise that the type of mental processing (simultaneous) determines K-ABC scale placement. not the method of presenting the stimulus (sequential) or the nature of the response (verbal). Good concentration and attention span are felt to facilitate successful performance. whereas impulsivity. distracti- bility. or an inability to respond when uncertain detracts from per- formance. The unique ability of this subtest is suggested as being the integration of sequentially presented visual stimuli. 19 Mm moms noocoucH was mumoucsm omens .mmaH .cmsmsmx a cosmsmx “mousom mcflvcmumuwvno\mcflwmmm .oa acfiwoow0\mcfiummm .ma mmHoUHm .vH oauoscuaum .MH mmomam use mmumm .NH whoasnmUo> o>flmmoumxm .HH unmfim>mfl£um mmfiumm ouosm .OH >MOEmz Hmwummw .m moflmonzm xauumz .m Howuo ones .h mcfimmmooum Hmucoz mmamcmaue Hamomm Homfisz musmoHo uHmummu mucofio>oz can: cofluwcmoomm comm soocflz Damn: .m .m .w .m .N .H mcwmmoooum Hounmz ummunsm "N.N OHQMB 20 Subtest 2. Face Recogition (Simultaneous ProcessMScalg __age_s 2-6 through_4-ll; Nonverbal Scale. eggs 4-0 through 4-11).-Photographs of one or two faces are exposed briefly to a child. who then has the task of correctly selecting the face(s) in a different pose from a group photograph. This subtest measures the child's ability to attend closely to details. This short-term memory task was found to be one of the best measures of simultaneous processing at the preschool level in the national tryout sample. Visual search and scanning strategies. face perception. and face recognition are unique abilities measured by this subtest. Good attention skills are required for successful per- formance. whereas distractibility. anxiety. or impulsive behavior has a negative effect on performance. Subtest 3. Hand Movements (Sequential Processing Scale age_s 2-6 through 12-5; Nonverbal Scale. ages 4-0 through 12-5).--The child's ability to copy a sequence of taps on the table with fist. palm. or side of the hand as performed by the examiner is assessed by this subtest and considered a measure of sequential processing. This sub- test is described as assessing the unique skill of motoric reproduction of a sequence. Successful performance depends on good attention and concentration. and poor performance can be related to anxiety. dis- tractibility. or lack of perseverance. Subtest 4. Gestalt Closure (Simultaneous ProcessinLScaleLagg 2-6 thrOfih 12-5).--Partially completed inkblot drawings are presented to the child. with the task being to mentally "fill in the gaps" and name or describe the drawing. This subtest has been described as an 21 excellent measure of holistic "right-brain" processing (Kaufman. 1979. 1983) and of simultaneous processing across the entire K-ABC age range (Kaufman. Kamphaus. & Naglieri. 1982). 'This subtest is considered to assess the unique abilities of perceptual closure. perceptual infer- ence. and conversion of abstract stimuli into a concrete object. Good performance is felt to be contingent on alertness to the environment and flexibility in perceiving and thinking. Impaired performance can be a result of lack of perseverance and inability to respond when uncertain. Subtest 5. Number Recall (Sequential Processing:§cale. ages 2-6 thrgggh lZ-SLF-A.series of numbers spoken by the examiner is presented to the child. who then responds by repeating the numbers in the exact sequence presented. Only forward span is included in this subtest as factor analysis revealed forward Number Recall to be the best measure of Sequential Processing for the entire KrABC age range. As would be expected. good attention span is required for successful performance. Subtest 6. Triangles (Simultaneous Processigg:gnd Nonverbal ScalesLag_e_s 4-0 through_12—5).-The child's task on this subtest is to assemble several rubber triangles to match a picture of an abstract design. Triangles is the K-ABC analog of the WISC-R Block Design subtest. Factor analysis has revealed Triangles to be a strong measure of simultaneous processing at all age levels. The unique contribution of this subtest is described as nonverbal concept formation. Difficulty working under time pressure or a field-dependent cognitive style may detract from performance. whereas a flexible problem-solving 22 approach and a systematic strategy for analyzing the model into com- ponent parts enhances performance. Subtest 7. Word Order (Sequential ProcessingiScale. gggs 4-0 throgh 12-5).-On this subtest the examiner names a series of objects in a particular order. and the child must point to silhouettes of the objects in the order named. Ability to recall both with and without an interference task is required by school-age youngsters. The interfer- ence task involves a S-second delay in which the child must name a series of colors before proceeding to the pointing recall task. The interference activity is explained by the authors as permitting the assessment of sequential memory in a limit-testing fashion. which pushes normal individuals to their capacity for receiving. processing. and remembering. Consequently. Word Order is described as testing the upper reaches of a child's immediate memory by requiring the recall of a short series of stimuli without permitting much time for rehearsal (Kaufman & Kaufman. 1983). The unique skills measured are considered to be auditory-visual integration. auditory motor memory. and retention without rehearsal. Like many of the previous subtests. anxiety. dis- tractibility.. and a limited attention span may impair performance. Good concentration. flexibility to shift quickly when the demands of the task change. ability to understand and follow directions. and the ability to generate a strategy for recalling stimuli without rehearsal were all delineated by Kaufman as enhancing performance on this task. 23 Subtest 8. Matrix Analgies (Simultaneous Processing_¥and Nonver- bal Scales. ages 5-0 through 12-5).--A 2-by-2 visual analogy is pre- sented to the child. who must select a correct picture or design that best completes the analogy. This task was found to be a very good measure of simultaneous processing for children in this age range. Analogic thinking in visual-motor areas is the unique contribution of this subtest. An impulsive response style may depress scores. Subtest 9. Spatial Memory (Simultaneous Processing:and Nonverbal Scales. _ages 5-0 through 12-5).--A one-page grid with randomly arranged pictures is presented to the child for a specified brief exposure time followed by presentation of another blank grid page on which the child must point to the locations of the previously presented pictures. This subtest is considered to assess immediate recall and spatial localiza- tion. Performance difficulties may be due to poor attention span. anxiety. distractibility or possibly a field-dependent cognitive style. On the other hand. good concentration skills enhance performance. Subtest 10. Photo Series (Simultaneous ProcessinLand Nonverbal Scales. age 6-0 through 12-5).--The examiner presents in random arrangement a series of photographs illustrating an event and requests the child to order the photographs in their proper sequence. Factor analyses revealed this task to be one of the very best measures of simultaneous processing across the 6 to 12-1/2 year age range (Kaufman & Kaufman. 1983). Parallels are drawn between this subtest and that of Wechsler's Picture Arrangement on the WISC—R in that both are excellent measures of simultaneous processing. Unique abilities felt to be 24 assessed by Photo Series include seriation. temporal relationships. time concepts. anticipation of consequences. and understanding of cause-effect relationships. An impulsive cognitive style may lower scores on this subtest. whereas good concentration skills may enhance performance. Subtest ll. Eyressive Vocabulary (Achievement Scale. ages 2-6 througg 4-11).--Objects pictured in photographs are presented to the child. who must then state the correct name for the objects. This subtest assesses the ability to recall verbal labels. Successful performance is very much dependent on early environmental opportunity. Subtest 12. Faces and Places (Achievement Scale. ages 2-6 throggk _1_2_;_5_)_.--Pictures are shown to the child. whose task is to name the fictional character. famous person. or well-known place depicted in the pictures. This subtest is a measure of factual learning. Success is highly dependent on environmental opportunities and the child's alert- ness to his/her environment. Subtest 13. Arithmetic (Achievement Scale. ages 3-0 throggh 12- 5L—Understanding of mathematical concepts and ability to count and compute are skills measured by this subtest. Performance is enhanced by concentration and past learning experiences. whereas distractibil- ity. anxiety. and short attention span may depress scores. Subtest l4. Riddles (Achievement Scale. ages 3-0 through 12-5).-- The examiner presents several characteristics of a concrete or abstract verbal concept. and the child is required to infer the name of the concept from these descriptors. This task is considered to measure 25 general achievement and verbal intelligence and is felt to be similar to the Wechsler or Stanford-Diner Vocabulary Subtests. 4Abilities assessed include logical classification. conceptual inference. and integration of sequentially presented auditory stimuli. Subtest 15. Readigg/Decoding;(Achievement Scaleggggs 5-0 thrgggg Eiw-This task requires the child to identify letters and to read and pronounce words. Letter naming. word-attack skills. word recogni- tion. and pronunciation are skills assessed. As is true for most of the other Achievement tasks. early environmental opportunities and outside reading experiences enhance performance. subtest 16. ReadingZUnderstandigg(Achievement Scale. ages 7-0 through 12-5).--A novel format is used to measure a child's reading comprehension by asking the child to act out printed commands. The authors suggested that their method.focuses directly on the child’s ability to derive meaning from the written word or statement. Reading comprehension and gestural communication are measured. and performance again depends to some extent on environmental and cultural opportuni- ties. iHowever. the authors also cautioned that performance can be negatively affected if a child is extremely shy or withdrawn and hesi- tant to act out commands. Standardization Procedures The K-ABC was standardized between April and September 1981 on a stratified sample of 2.000 children ages 2-6 through 12-5. These children were tested at 34 test sites in 24 states and randomly selected for participation by computer from a large pool of parental 26 permission slips. Stratification variables within each age group of the sample included gender. geographic region. socioeconomic status (highest educational attainment of parents). race or ethnic group. community size. and educational placement of the child (normal or special class placement). . Educational-attainment categories for parents were divided into four levels according to the number of school years completed: less than high school (1 to 11 years). high school graduate (12 years. including GED). 1 year or more of college or technical school (13 to 15 years). and college graduate (16 years or moreD. The educational level of the parent with the highest number of years completed was used for classification. There were equal numbers of males and females in each age group. representing the four geographic regions of the 1980 U.S. Census data. which included Northeast. North Central. South. and West. Community was stratified by three levels of size: city (population of 50.000 or greater). suburb or small town (2.500 to 49.999). or rural area (popu- lation of 2.499 or less). IRacial-group categories of white. black. Hispanic. and "other" were included in the sample in percentages according to Census proportions across the K-ABC age range. The "other" category comprised Native Americans. Alaskan Natives. Asians. and Pacific Islanders. The educational-placement level of the sample was divided between normal and exceptional placements. Based on statistics from 1980 U.S. Department of Education data. children were sampled from the following 27 six special education categories: speech impaired. learning disabled. mentally retarded. emotionally disturbed. physically or otherwise impaired (health impaired. orthopedically impaired. multihandicapped. and hard of hearing). and gifted and talented. Children in each cate- gory were selected for sampling in proportion to U.S. government fig- ures. Accordingly. 72 of the total sample of children were enrolled in special education. and 932 were in regular classes. A special sociocultural norming program was conducted between November 1981 and March 1982. This procedure involved testing an additional 496 black children and 119 white children to develop socio- cultural norms by race and parental education. This additional group of children was added to the national standardization group. The Manuals Two manuals accompany the test kit. a separate Administration and ScoringManual and the Interpretive Manual. As the name implies. administration and scoring rules. procedures. guidelines. and normative data are discussed in the first manual. In the latter one. standardi- zation procedures. reliability. and validity are discussed. In addi- tion. a systematic approach to profile interpretation is presented. along with an overview of the theory and research underlying the instrument . Administration and Scoring The subtests are contained in three easels and ordered for ease of administration. The first two easels contain the Mental Processing 28 Scales. and the last easel contains the subtests for the Achievement Scale. Instructions for administration. starting and stopping points. and scoring are printed in the easels as well as on the individual test records. Items for each subtest are grouped into 23135. which are readily identified on the test record. The items are grouped into units to facilitate the identification of starting points. stopping points. and rules for discontinuing a particular test. Starting points are designated according to the child's chrono- logical age and correspond to the beginning items of different units. Testing continues to advance if the child passes at least one item in the first unit administered. If the child fails all items in the unit. the examiner has two alternatives: (3) if the starting point was item 1. testing proceeds with the next subtest; but (b) if the starting point was beyond item 1. the examiner returns to item 1 and continues testing up to the original starting item or until the discontinue criteria are met. whichever occurs first. Stopping points are always the last item in a unit and are well marked in the easel and on the test record. However. whenever a child passes all items in the last unit designated for the child's chrono- logical age. testing continues until one item beyond the stopping point is failed or until the subtest is completed. A discontinue rule was developed to prevent undue frustration for the child who fails several items in a row. The rule is to discontinue 29 testing if a child fails every item in one unit before reaching his/her designated stopping point. Scoring a child's responses is easy because all K-ABC items are dichotomous. with correct responses scored 1 and incorrect responses scored 0. Scoring is further simplified by the fact that no partial credit or bonus points for quick performance are given. The authors discussed any variations or irregularities in scoring and differences in criteria for mental-processing subtests versus achievement subtests. Types of Scores After computing raw scores. a conversion table is used to convert these scores to scaled or standard scores. The scaled scores are then summed to obtain a Sequential Processing sum. a Simultaneous Processing sum. a Mental Processing Composite sum. and an Achievement sum. Global Scale standard scores for each of these sums are then obtained by consulting a second conversion table. The standard scores of the four Global Scales have a mean of 100 and a standard deviation of 15. Bands of error or confidence intervals are derived for the Global Scale standard scores by consulting a third table. The authors sug- gested using the 90% confidence level. believing that it provides an appropriate degree of confidence. National percentile ranks are obtained for each of the Mental Processing scaled scores. Achievement scaled scores. and the four Global Scales by consulting a fourth table. If desired. sociocultural percentile ranks may be obtained from a fifth table for the Achievement standard scores and the Global Scale 30 standard scores. These percentiles are considered optional and are used when minority-group children are tested to allow comparison to a reference group more similar to the child's sociocultural background. However. these percentiles are not used for educational placement purposes. The usefulness of these norms has not been investigated. and their purpose is not clearly stated in the manual. The authors sug- gested that a child's potential for learning may be estimated on the basis of these percentiles. However. beyond this suggestion. they offered no guidelines for using this information in planning educa- tional interventions. Although not considered necessary by the authors. they provide supplementary tables for computing optional scores. including stanines. age equivalents. and grade equivalents for the arithmetic. reading/ decoding. and reading/understanding subtests. Descriptive/qualitative categories are provided as well. Reliability. Validitl. Standard Error of Measurement! and Brief Comparison With the WISO-R The following is a discussion of the statistical properties of both the K-ABC and the WISC-R as presented in their respective manuals. The WISC—R subtests are not described or discussed separately because the battery is such a well-known test and numerous published reports and articles are available for review of its subtests. Sattler (1982) is an excellent reference for the reader who wishes a brief review of the W ISO-R. 31 For the K-ABC. average split-half reliability coefficients for the K—ABC subtests across the 11 age groups in the standardization sample are reported. as are age-by-age split-half reliability coefficients for the four Global Scales. Mean subtest reliabilities for school-age children range from .71 to .82 on the Mental Processing subtests and from .84 to .91 on the Achievement subtests. Overall. average split- half reliability coefficients across the entire age range are .80 and above for 12 of the 16 subtests. On the Global Scales the mean coeffi- cients are .89 (Sequential). .93 (Simultaneous). .94 (Mental Processing Composite). .97 (Achievement). and .93 (Nonverbal) for school-age chil- dren. In addition. test-retest reliability was computed for three of the age groups to present stability estimates of the K-ABC Global Scale test data. For the age range 9-0 to 12-5. the reliability coefficients are reported in the manual as .88 (Sequential). .91 (Simultaneous). .93 (Mental Processing Composite). .97 (Achievement). and .87 (Nonverbal). The reliabilities for ages 5—0 to 8-11 range from .82 (Sequential) to .82 (Nonverbal). and for ages 2-6 to 4-11 they range from .77 to .81. The WISC-R manual also reports split-half reliabilities for each of the 11 age groups in its standardization sample and presents test-retest reliability for three age groups. Verbal. Performance. and Full-Scale 10s have average reliabilities of .94. .90. and .96. respectively. across the entire age range. The reliability coefficients for the K- ABC and the WISC—R are presented in Tables A.l through A.6. Appendix A. 32 The standard errors of measurement for the K-ABC and the WISC-R are reported for all age groups in their respective standardization samples by subtests and global or total scores. The mean standard errors of measurement on the K-ABC for school-age children are reported as: 5.0 (Sequential). 4.0 (Simultaneous). 3.5 (Mental Processing Composite). 2.7 (Achievement). and 3.8 (Nonverbal). On the WISC—R the average standard errors of measurement across .the age range are: 3.6 (Verbal). 4.6 (Performance). and 3.2 (Full Scale). The standard errors of measurement for the K-ABC are presented in Tables A.7 and A.8. and for the WISC-R in Table A.9. Appendix A. Validity data are presented for both the K-ABC and the WISC-R in their respective manuals. Construct. predictive. and concurrent valid- ity studies have been conducted with the K-ABC and according to the authors offer supportive evidence of a good degree of validity in these areas. The WISC-R manual presents correlational data comparing the instrument with other measures and evidences a good degree of concur rent validity support. As predictors of achievement (scores on the Peabody Individual Achievement Test). Sequential and Simultaneous Pro- cessing standard scores were reported to be about equally effective. with validity coefficients ranging between .30 and .50 for school-age children. When the Achievement subtests of the K-ABC were used as predictors to PIAT Total scores. coefficients ranged from .69 to .89. Median correlations between the WISC-R and achievement tests and grades were reported to range from the upper .30s to low .80s. The Full-Scale score correlates .60 with PIAT Total scores. 33 Both the K-ABC and the WISC—R manuals provide the average inter- correlation of the subtests for the 11 age groups in their standardiza- tion samples. In addition. the K-ABC presents information regarding intercorrelations of the Global Scale standard scores by age for the standardization sample. Reviews of the K-ABC In a special issue of the Journal of Special Education (1984). the K—ABC was the subject of a comprehensive review and analysis by a number of respected measurement and evaluation experts. The instrument was considered from a number of perspectives: psychometric and statis- tical features. theoretical underpinnings. and research and practical applications. As might be expected. the "jury" is still out with respect to the clinical usefulness of this new intelligence instrument. By and large the reviews were mixed. so that no conclusive statements can be made regarding the diagnostic utility of the K-ABC. Consequently. the current discussion attempts only to highlight the major issues and concerns of these early reviews. Just summaries of some of the review findings are presented because many pages would be required to adequately address all aspects of the issues and concerns. In his article. Mehrens (1984) provided a thorough review of the psychometric properties of the K—ABC. He described the K—ABC as a good test. noting that it has respectable psychometric properties expected of newly constructed instruments. The reliability data are considered adequate. but he presented the concern of a lack of stability data for longer than 18 days. Regarding validity. a concern was raised about 34 the fact that predictive validity studies (to achievement data col- lected after the K-ABC data were gathered) were not conducted beyond 1 year and that there was no stated intention by the test authors to conduct long-range predictive studies. Other concerns included a lack of a definition of "bias." together with an implied inference of the test authors that bias is considered a mean difference across ethnic groups. Relatedly. the intention of the sociocultural norms was ques- tioned because of a failure to provide clear instructions for their use and interpretation. Finally. Mehrens was critical of the fact that reported educational implications of the sequential-simultaneous pro- cessing dichotomy are based on only two studies with very small sample sizes. The positive features of the K-ABC were described by Mehrens as outweighing the limitations. In particular. he referred to the follow- ing as being positive features: the theoretical basis. the complete- ness of the manuals. the discussion of differences between statistical significance and abnormality. and the discussions of difference scores and multiple comparisons of profile scores. Similarly. Anastasi (1984) considered the K-ABC a well-designed test as she indicated that it reflects sophisticated application of current test methodology. However. she challenged the Kaufmans' dis- tinction between aptitude and achievement. She pointed out that the global mental processing scores correlate more highly with the achieve- ment composite than with each other and that the suggested distinction between K-ABC Achievement versus Mental Processing is false. To 35 support her position. she defined an achievement test as one that is closely tied to specific. identifiable instructional content. to which examinees presumably have been exposed. Regarding the K-ABC Achieve- ment scale. she indicated that the Kaufmans took special efforts to dissociate tests (separate mental-processing ability from acquired factual information and minimize the role of language ability to pre- vent contamination of problem-solving ability with verbal fluency) from specific information acquired in the classroom. This comment appears to have been taken out of context as the achievement section of the manual does not reflect such a statement. Perhaps what she was allud- ing to is the Kaufmans' emphasis on the processing of visual stimuli rather than the verbal question-answer format of most achievement instruments. In this respect. the K-ABC achievement scale format is not closely linked to traditional classroom teaching methods and would not qualify as an achievement test by the above definition. She indi- cated there is little justification for the ”achievement" label and suggested the Kaufmans' erroneous mental processing-achievement dis- tinction encourages misconceptions and misuse. Although she applauded the multiple-score format and profile analysis of the K-ABC as means for enhancing diagnostic value. she felt the testing community should be apprised of the possibility of misinterpretation stemming from the Kaufmans' ”achievement" terminology. In their review of the K-ABC. Keith and Dunbar (1984) focused on the factor structure underlying the test. They indicated that factor- analysis studies have generally supported the validity of the 36 mental-processing scales (simultaneous and Sequential). However. in those analyses involving three-factor solutions (inclusion of the Achievement Scale). there has been considerably less support. revealing that an alternative structure or model may fit the data. Based on their findings. the K-ABC might alternately be interpreted as measuring verbal reasoning. nonverbal reasoning. and verbal memory rather than the intended design to measure achievement. simultaneous. and sequen- tial processes. respectively. Because of the discovery that the verbal-reasoning factor (Kaufmans' Achievement factor) was found to be identical with g. Keith and Dunbar questioned the viability of the aptitude/achievement distinction and warned examiners to be cautious in interpretations of K-ABC scores. particularly those on the Achievement scale. Sternberg (1984) soundly rejected the K-ABC as a good test on a number of points. He indicated that under any circumstance he could not advocate the use of this test over the WISC-R or the Stanford- Binet. He enumerated the major negative features of the test as (a) misrepresentation of support for the theory underlying the test. (b) noncorrespondence between definition and measurement of intelligence. (c) inadequate aptitude-achievement distinction. (d) overemphasis on rote learning. and (e) questionable empirical support in studies directly testing the construct validity of the test. In addition. he pointed out the positive features as being (a) an attempt to provide a theoretical basis for intelligence testing. (b) assessment of the ability to deal with novelty. (c) an attempt to integrate psychometric 37 and information-processing paradigms. (d) an attempt to achieve culture fairness and norm representation. and (e) attempts to ensure examinees' comprehension of tasks (p. 251). He presented cogent arguments for all of the positive and negative points but felt that the undesirable features of the K-ABC outweigh the positive ones. His conclusion was that the test is based on an inadequate conception of intelligence and as a result is not a reliable measure of intelligence (p. 259). Goetz and Hall (1984) evaluated the K-ABC to assess how well the instrument adheres to an information-processing theory or model. They reminded the reader that the incorporation of information-processing theory and research was a major goal of the test developers. They examined the K-ABC from this perspective. with an emphasis on four areas: theory. tasks. scores and scales. and educational applications. Regarding the dichotomies theorized to underlie the K-ABC (Mental Processing and Achievement. Simultaneous and Sequential). they indicated that the information-processing literature has suggested that knowledge structures or achievement products are a part of mental processing in any cognitive task. Consequently. they believed that the separation of processing and achievement conflicts with the theory. Considering the Simultaneous-Sequential dichotomy. they pointed out that there is no consensus in the research literature that cognitive processing can be neatly conceptualized by these two modes. Regarding the tasks on the K-ABC. they indicated that the items were chosen to reflect either simultaneous or sequential processing. However. they pointed out that when analyzed there are important 38 sequential components in many of the simultaneous tasks. and vice versa. Consequently. the tasks do not reflect clear distinctions between the two types of processes. Moreover. they indicated that both "simultaneous" and "sequential" tasks are dependent on cognitive struc- tures. strategies. and executive functions not adequately characterized or taken into account in the K-ABC. In terms of the scores and scales provided by the K-ABC. Goetz and Hall believed these do not allow a sufficient understanding of an individual's performance. As per an information-processing model. they disclosed that the intention of the model is to understand the nature of skilled performance and to identify an individual's weaknesses. They indicated that quantification of correct responses related to performance norms does not clarify an individual's cognitive capabilities in other contexts. Referring to the model once more. they indicated that time and error analysis is an important information- processing method for understanding patterns of cognitive functioning. For example. they pointed out that an information-processing analysis includes collection of reaction-time data along with response data. While the current explanation oversimplifies this type of analysis. the amount of time spent on an item has been used as a measure of input activity. Analysis of the nature of errors and examination of the patterns of errors has been used to determine an individual's approach to a task. According to the authors. such an analysis often reveals that a child's approach to problem solving is not unreasonable. but perhaps incomplete. The time and error data are then used to develop a 39 model of how a child solves problems. The task analysis is used as a basis for instructional remediation in that the sequence of cognitive steps necessary to solve»a particular type of problem or task.can'be taught to the child. Knowledge of what steps are missing in the child's approach (through task analysis) allows an instructor to pro- vide specific instructional methods to supplement a child's "flawed" problem-solving strategy. .As organized. the K-ABC does not conform to memmfim. Finally. considering educational recommendations. Goetz and Hall pointed out that suggestions to present instruction simultaneously or sequentially to match a child's assessed strength belie the complexity of the information-processing approach to analyzing and developing effective instructional methods. They concluded that although the attempt to develop a test instrument based on information-processing theory and research is praiseworthy. the K-ABC falls short of that goal. Several other reviews have appeared on this issue but are not reported here. These other reviews have mirrored many of the same positive and negative points previously noted (Das. 1984; Dean. 1984; Kaufman. 1984; Majovski. 1984; Salvia & Hritcko. 1984). However. a more recent review by Keith (1986) is worthy of note. This study appears to be an extension of the 1984 investigation by Keith and Dunbar. It differs from the earlier one in that the K-ABC scores of 585 referred school children were factor analyzed. The use of children referred for psychoeducational evaluation. instead of reliance on the standardization sample. is more representative of the sample of children evaluated by school psychologists. The findings revealed the factor structure for this group of exceptional children to be consistent with that found for the standardization sample. Although the factors appeared stable across normal and exceptional groups. Keith cautioned that the meaningiand names of the factors remain open to interpretation. Again. as in the previous study. achievement was found to be a better measure of general intelligence than the mental- processing scales. In addition. the results:of the factor analysis revealed that the alternative constructs of general ability or reasoning. verbal memory. and reading achievement/verbal reasoning could be substituted for simultaneous processing. sequential processing. and achievement. respectively. as an explanation of the factor structure of the K-ABC. They concluded that it is not clear which of the above constructs or factors are measured by the K-ABC. Considering the fact that the K-ABC was published in 1983. a comprehensive review in 1984 was premature. A series of investigations needs to be completed before this test is fully assessed. For now. the most that can be said is that the K-ABC has generated a considerable amount of research that will add to the knowledge base in the assess- ment and special education fields. Ultimately. this information should benefit the children served by special education. CHAPTER III LITERATURE REVIW Overview The preceding chapter presented a brief descriptive overview of the KrABC subtests. standardization procedures. and administration and scoring criteria. In addition. comparative psychometric properties of the K-ABC and the WISC-R were outlined. Overall. both tests appear comparable with respect to measurement sophistication. This chapter contains a review of the literature associated with attempts to substantiate the existence of learning disabilities as a distinct diagnostic category separate from other handicapping condi- tions (EI). and from low-achieving (LA) nonhandicapped children. The sections of the chapter contain reviews of research in the following areas: (a) psychometric similarities and differences between students labeled LD and LA. (b) ability-achievement differences and the use of the discrepancy model. (c) verbal-performance and sequential- simultaneous ability differences. (d) ability subtest scatter compari- sons. (e) general ability and achievement characteristics. (f) verbal- performance discrepancies and sequential-simultaneous discrepancies: emotional impairment. (g) emotional impairment: high similarities. low information WISC-R subtest pattern. and (h) emotional impairment: KrABC number recall. spatial memory pattern. The final section of the chapter describes the direction of the present investigation. 41 42 Learnipg_Disabled Versus Low Achievers: Payphometric Comparisons Historically. the search for characteristic test profiles to serve as diagnostic indicators has been fueled by a belief that handicapped groups process information differently than do nonhandicapped. Indeed. observations of differences in abilities and academic skills among various exceptional groups have suggested that tests purported to measure such skills would reflect such differences and be useful for classification purposes. Since IQ tests were found to be highly predictive of academic achievement (high correlation between IQ tests and achievement test scores). the belief in the utility of test scores for psychoeducational diagnosis was not unreasonable. The presumption that standardized tests can reliably differen- tiate between diagnostic groups continues to hold considerable appeal to diagnosticians and researchers alike. The lure of the concept seems to encourage a belief system impervious to mounting evidence that test- score patterns provide little useful information for the diagnosis of specific handicapping conditions. This is evident from a perusal of the research in which there continue to he attempts to define/describe characteristic test profiles for specific diagnostic groups. The search for characteristic patterns has occurred most notably with the LD classification. Observed discrepancies between verbal and perform- ance IQs on the WISC—R have spurred the search for and raised the hope that reliable score patterns for diagnosis will be empirically vali- dated. 43 "Looking for LD in all the wrong places“ might be a more approp- riate subtitle for this section because research to date has not revealed characteristic test-score patterns useful for diagnosing learning disabilities. In fact. a considerable body of research has revealed equivocal results with respect to the usefulness of standard- ized test results for classification purposes. To paraphrase Veres (1982). there is no convincing evidence that subtest-score distribu- tions can be used reliably to diagnose anything other than a child's general level of intellectual ability. Veres's admonition seems to be borne out by a 1985 study conducted by Sexton and Street. They investigated the similarity in patterns of strengths and weaknesses of subtest scores on the WISC—R between students evaluated and placed in special education programs and those students evaluated but not placed. This study did not examine specific diagnostic groups; however. all students were referred because of learning and/ or behavioral problems serious enough to interfere with academic adjustment. Consequently. although not specifically related to LD children. this investigation has merit in that it addresses the global decision concerning which children (with what types of score patterns) are placed into special education. At a general level. this is the type of decision that psychologists and placement committees must face with each new referral. The findings revealed the two groups (placed and not placed) to be very similar in variation of their rela- tive strengths and weaknesses. They differed only in degree. revealing that the scores of those qualifying for special education were 44 significantly lower than scores for those not placed. Although the results provide information about group-level differences of special education populations in general. no useful decision rules can be derived for individual diagnostic decisions. However. it does suggest that IQ tests continue to be critical determinants of who does and who does not receive special assistance for learning problems. Moreover. the results mirror the finding that score distributions merely diagnose a child's general level of intellectual ability. This finding is relevant to the present study in that of those children scoring lower on IQ tests. separation of those who are handicapped from those who are merely LA remains problematic. Algozzine and Ysseldyke (1982) pointed out that low achievement of some form characterizes both LD and LA children. This statement could be extended to emotionally impaired (31) as well. The federal and state guidelines indicate that behavioral/emotional problems adversely affect the child's education to the extent that the BI child cannot profit from regular learning expe- riences. Academic/performance-adjustment problems exhibited by such children often alert teachers to the need for psychoeducational evalua- tion. Consequently. Sexton and Street's findings have particular rele- vance for the present investigation examining the discrimination of LD from 81 and from a larger group of LA children who are referred but not placed. Studies examining similarities and differences between LD and LA students have appeared in recent years. Attempts to establish promi- nent characteristics for distinguishing LD test profiles have employed 4.5 a number of comparisons. Brenton and Gilmore (1976) applied an opera- tional index of discrepancy to school-identified LD children and dis- covered that between 252 and 33% of these children would have been misclassified using an operational definition. Warner (1980) found no differences in ability-achievement discrepancy levels for adolescent- age LD and LA populations. Wiseman (1984). in comparing psychometric characteristics on the WISC-R across LD and LA populations. concluded that methods calculating g-score discrepancy levels. ability-test subtest scatter. verbal- performance ability differences. and general ability and achievement characteristics were not practical in individual assessment situations for differential diagnosis. Several investigators (Algozzine & Yssel- dyke. 1982; Shinn et al.. 1982; Ysseldyke. et al.. 1982) have compared LD and LA populations in a number of ways.inc1uding general comparison of ability subtest scores. ability-achievement discrepancy. processing differences. general achievement characteristics. and strengths and weaknesses as reflected by verbal-performance IQ scores. The results of these studies have been consistent in showing considerable overlap between populations and in failing to identify consistent characteris- tics specific to LD or useful for differential diagnosis. A principal study. and perhaps the seminal investigation of psy- chometric similarities between LD and LA groups. was conducted by Ysseldyke. Algozzine. Shinn. and McGue (1982). They administered a battery of tests to school-identified LD students and LA students (those students not identified as LD who scored at or below the 25th 46 percentile on the Iowa Tests of Basic Skills) and compared their per- formance on all measures. The battery consisted of the Wechsler Intel- ligence Scale for Children-~Revised (WISC—R). Peabody Individual Achievement Test. selected subtests of the Stanford Achievement Test and the Woodcock-Johnson Psycho-Educational Battery. two visual motor tests. a self-concept scale. and a behavior problem checklist. In all. there were 49 subtests for comparison. Methods of comparison included overlap analysis. counting both the number of exact pairs of scores and the percentage of scores within a common range for both groups; and discrepancy between ability and achievement using three operationaliza- tions of the federal definition (1 standard deviation. 1.5 standard deviations. and 2.0 standard deviations). In addition. general per- formance patterns for both groups were compared in the areas of cogni- tion. achievement. and perceptual-motor functioning. The results of the study revealed that an average of 962 of the scores for the two distributions were within a common range and that the average number of pairs of identical scores was greater than 25. indicating that more than half the scores in the two groups were identical. This finding meant that. given that an LD student earned a particular score. more than 50% of the time there was a student not receiving services who earned an identical score. Comparison of both groups to the three operationalizations of the federal definition for ”severe discrepancy" revealed that. when matched to the school's classifications. there was considerable misclassification. The investigators concluded that they 47 could not identify psychometric differences of practical utility to differentiate the two groups. In summary. the evidence suggests that LD and LA children are more similar than they are different on commonly used psychometric measures. The results of Ysseldyke et a1. (1982) suggest that on an individual basis it would be almost impossible to differentially diagnose the two groups. This study seems to support Algozzine and Ysseldyke's (1982) contention that LD is a category of underachievement. A_b_i;ity-Achievement Difference: Digcrejancy Model The focus on test score patterns is an example of the search for general psychometric differences to characterize the profiles of LD children. As previously reviewed. there is virtually no information in test scores that supports the category of LD as separate from a larger category of LA children. A more specific procedural strategy. and the one most commonly employed by school districts for identifying children with learning disabilities. is the search for a discrepancy between ability and achievement. Determination of ability-achievement dis- crepancy is not just a popular idea but is stipulated in the federal definition. which requires a LD child to have a "severe discrepancy" between achievement and intellectual ability. The federal guidelines specify two focal criteria to identify a child as LD: (a) low achievement not commensurate with a child‘s age or ability levels in one or more of seven areas. despite being provided with learning expe- riences appropriate for the child's age and ability levels; and (b) demonstration of a "severe discrepancy” between achievement and 48 intellectual ability in one or more of the seven specified academic curriculum areas (demonstration of #b establishes the first crite- rion). In other words. the regulations have implied that test proto- cols of LD have more "severe" deficits between ability and achievement than would be expected in the non-LD population. If this legal defini- tion is correct. significant ability-achievement discrepancy might provide a decision rule for the differential diagnosis of learning disabilities. It remains to be empirically demonstrated that a dis- crepancy score can discriminate between LD and LA populations. Problems With the DiscrepancLApproach The use of the discrepancy model is beset by a number of diffi- culties. the most obvious being that "severe" is not defined. Algozzine and Ysseldyke (1981) indicated that the federal rules and regulations do not specify the magnitude of a discrepancy needed to be considered "severe" so that considerable variation exists across school districts in terms of approaches used to identify "severe discrepan- cies." Wiseman (1984) reflected this concern. saying that school districts' policies for establishing special program cutoff scores or points add to the variation in the use of the discrepancy model. Relatedly. Berk (1984) stated. "the need to determine a 'severe' dis- crepancy has prompted the development of more than a dozen procedures for assessing the magnitude or statistical significance of ability- achievement discrepancies" (p. 262). Considering the various conceptualizations of "severe discrepancies." it is not surprising that 49 some researchers have considered learning disabilities to be an opinion rather than a reliable diagnosis. In this regard. Epps. Ysseldyke. & McGue (1982) attributed the confusion in separating LD students from other handicapped students to the substantial lack of agreement and uniformity in applying the federal model. Another problem involved with the use of the discrepancy model has to do with the reliability of difference scores. Establishing the reliability of a discrepancy score is a prerequisite to determining validity for diagnostic utility. Reliability is a method for evaluat- ing whether a discrepancy is due to chance or errors of measurement rather than a true difference. Reliability allows a level of statisti- cal confidence to be applied to statements concerning the magnitude of an ability-achievement difference. This is important if differences are to be considered "significant" and indicative or predictive of membership in a particular classification-win this case. learning dis- abilities. For example. if a discrepancy score is demonstrated to be significant. one can infer that the difference is due to sampling error or chance 5 times or less out of 100. depending on the significance level used (Berk. 1984). This probability level describes the degree of confidence one assumes in stating that the observed difference is not due to the caprice of sampling. A probability level does not suggest that a _r_egl_ difference exists. but reflects a degree of confi- dence that a real difference may exist (Ferguson. 1971). The chances for misclassification can therefore be minimized by regulating the size of the confidence interval. The problem of mislabeling is discussed 50 later when studies involving error rates for differential diagnosis of LD and LA students are reviewed. For now. it is of more immediate concern to examine and evaluate the statistical reliability of a number of the more common discrepancy models (formulas) used for differential diagnosis of learning disabilities. Discrepancy Models and Statistical Reliability In 1979. Ysseldyke expressed strong sentiment about the use of the discrepancy model when he suggested the use of deficit scores to be a very dangerous and misleading practice. A study by Algozzine and Ysseldyke (1982) seems to support this caveat. They compared 16 dis- crepancy models for differentiating LD students from other LA students and found that no single model (formula) was able to differentially diagnose LD students. In fact. they discovered that 92% of the LA students met LD classification criteria on at least one of the for- mulas. Similarly. Berk (1984) examined nine procedures (Table 3.1) for computing an ability-achievement discrepancy score. The procedures were selected for evaluation on the basis of their being cited most frequently in the learning-disability literature. He ascertained that all of the formulas had one or more statistical problems but found the major weakness of all the procedures to be no provision for determining how errors of measurement or chance factors may account for the dis- crepancy between two scores. Berk revealed that the score indexes produced by these formulas provided the magnitude but not the 51 Am>ono no o.- I £0 chess. v o.m I <0 + cmn .o.- 0» o.o~ I <0 ohms). m o.m I <0 + (ZN Am.m 0» o.w I <0 Ohms). m o.m I (UN + (In Amm-oo .am .vmmn . ou . sum 3 ..o.m-.a.m mono: mum .m m o o n <0 : . .msao< a .~o>o~ Houcoeao~o>ov new unsouom N communes. >~ouousuos uoc on mmuoom <2 cm: mozcoxaumas oh wow Joann son accumum mu< ~cwxsumqs OH .~o>o~ mmocuosouaocmunou Amanda .mm Cu «snob convawno Has moEomm< OOH .homHV magmas .couuauno OHIIOH moduuucoouuo>o omz mu.0~ o.~ + .mr x am. one vcom cod .050“. .xolm:.~.m mono: com um: moo« doucofino~o>oo new unsouom m AOOmImQN .mm >~oumunoou DO: 00 mououm 4: omz modawom mo< 40.4! N.m u <0 + (IN .Aboa. muuuoz .xo.m-.~.m mono: oom .~o>o~ asucmamo~s>ov so» Deacons aaououooos uoc on museum 4: AOMAImNA .mm .couodugu o~-noag nuauaucoofiuo>o omz mo<-mom mox a: o.m . «: .domac Nahum: accuusuafiuq a; now >ucomouomao ucwcomsou wouoomncou ogcassuom >ocsuooaxmv ououm n ceaumufiuu . ucmsw>mazuc moandwum> acocomeou >uuaun< ocumaouOCH mo Hoouo :« voumuqv ouoom >ocomouomuo acofio>o«:Uumxo~aeoo Macaquausmfiou ocaz no mowumqumuoouszu acouuom ”H.n snags 522 .ucofio>m«zom use >ufiaflnm :mo3uon noduoauuuou uumuuomad souu madcasmuu save any cu sawamouvou new unsoous Du uwsu mama :« uasmmu mummy coumnouuoououca aazvflzv ..nom~ .conxwze oonum sum sou m.~ Cu ovens and now m. scum ocwvcou coduouuuu oeucuau s on Has) an m. as some .oo«ammm soon o>sz s«umu«uu mucuuo> .uo>oxoz .vvquuosmm uozu .ucoeo>oa£oo asaoa>uvcu mmoumxo o» .mo4. mugs: mucom aco~o>usvouovouw on: «unavOOOun co>oo umuub .0. .>u:smou0mww ouo>0m a new acquouauu coauaum> >-momuwm§b so >uuosmm o» «gnu .aocomsuumav on» we “oucoufiuusoum noofiumwumum. auuawnowamu any snaauoumv 0» many avenue. HH‘ an. avozuofl ~H¢ Av. avozuofl dd< An. .uuou zoom no auadwnnadoucs can sou gaseous ou aqua accrua- -< .u. ..m0uoom accomouoaqo sundunou .mouoom acoso>muzoo was >ua~ans cooxuon nodumauuuoo osu new unsouoe Cu «Mum noonuua «H4 an. "osHSELOu no unequouulwa n oncom ucoas>uauo umucouco: o.~ on» songs noduoauumcw ocuooou soEROuco: no muoo> swm I o.m Amuoum ucoflo>wsuouoosuo. u:0§o>o«:us cuuoacuuus I << uocooucoc o as swoon ucodo>asvuooosuo madness I o.~ .o.~ u no. noonou cw munch I m» .ououu cuoncoum me «soon sou. ucoso>ouzuo uo ~o>o~ omuoomxo I Am acoEouoda ocean I no “UHOUm ousccsum no ouoos tony acofio>odnoo mo Hm>m~ Hmsuom I 44 On noduma>oo uo assauosv mucooAAAmusw I on ouoom acmuo>asuo ocouo vouuomxo I mum soc nonwoo~ocouzo I <0 ouoom uns~s>a=co mucus Hugues I no: aououo u=o~s>wsvouooo. ova deacon I s! "moussuOu uo mucocoasoo mascuuusauwns .Avmmav xuom Iona coumovc ”ouusom x4< :0 0H .Amvmuxavm nouo: mom .mou: Ho>oa coouu uo :omemuovu umoos ucoao>ouzuo can >uadfins zoom on scausu>wc sauu Hosvamou. .omma. coauocuoo wean: moumuo>c< umousouo zuuz 00H A< u Am 44 OH OH ousmozm mmwocm Imouomqv o>auouoc .BhOu ououmLmv .mhouv .xm.n-x~.n mono: com unwound nu“: ood 44 u om 44 an an comxu«um .onmnaavm mouo: mom .~o>o~ aoucolao~o>uv new ucsoUuo ooH x mum n AmmmAV 333:8: uo: oo 3.38 5.. om 5.5 $3 woe uu< ac.¢o.§ no + <0 + 5.. $233.»: .onmuaavm nova: com 49:: asucgo~o>ov uOu unsound mom n Ammo: >H0usuouuo use on mouoom m: om coca mama cod x mu< mom <4.¢u.‘z o.m u << + <0 + (a cameo: DA new unocomeoo vauovwncou .QHDEuOm >ocsuuoaxm. sea a ouou A «a uuaqq nodusuuuu >ucoaouumuo ucwEo>muzu< moqnswus> a acocomeou >uu~un< m gaudy—.00 «H3. wanna. 53 statistical significance of the deficit between ability and achievement performance levels. Federal and state laws require that a difference is significant as well as "severe" or substantial. In other words. one must be relatively certain that a difference (regardless of magnitude) is not due to chance or sampling error. He focused his attention on a number of the more popular expectancy-discrepancy formulas and limited his comments to the most serious statistical deficiencies. For example. he pointed out that the Harris (1961). Horn (Torgenson & Adams. 1954). and Monroe (1932) formulas (which determine expected grade equivalent by subtracting 5.0 years from a child's mental age) are all subject to regression toward the mean effect. These formulas have a tendency to overidentify high-IQ children and to underidentify low-IQ children. Conversely. the Bond and Tinker (1967) method (IQ x years in school + 1.0). which treats the IQ as a ratio scale of meas- urement. has the same problem as grade-equivalent scores. which do not accurately measure developmental level. As a consequence. this formula tends to overidentify low-IQ children and to underidentify high-IQ children. The other formulas in the table. with the exception of the Erickson (1975) and Shepard (1980) methods. are all considered varia- tions of the Harris and Bond and Tinker formulas. which use age equiva- lents (MA's) derived from intelligence tests. Consequently. these other formulas also fail to accurately measure changes in the distribu— tion of age/grade equivalent scores as grade placement increases (Reynolds. 1981). These types of scores differ from standard scores in 54 that they do not account for dispersion about the mean and are not based on age-level means. Although the Erickson and the Shepard methods attempted to improve on the above-mentioned inadequacies by using deviation standard IQs (which have the advantage of comparability of score interpretation across age—same percentile rank across age). they still. according to Berk. do not account for regression to the mean effect. Because of this inadequacy. he suggested that the use of these formulas. as well as the other seven. would tend to result in more false-positive than false-negative errors. Berk defined false positive as labeling a “normal” child LD and false negative as the reverse. However. in both research and practice. one never really knows whether one's decision is correct or due to error. Perhaps Mann (1983) was alluding to the possibility that many children are mislabeled LD when he stated that LD constitutes the largest handicapped group in the country. Erickson's _Z_-score method fails. according to Berk. in not accounting for the correlation between ability and achievement test scores. Apparently. Berk was referring to the intercorrelation between ability and achievement measures and the problem that the reliability of a difference score is reduced when two measures are highly related. A difference score may be spuriously large because the error variance of the difference score is equal to the sum of the error variances of the two measures (OZED= 023x +02EY ). In addition. be faulted Erick- son's method because there is no definition or empirically validated criterion for severe discrepancy. However. this latter problem is not 55 unique to Erickson's Z-score method. None of the discrepancy proce- dures specifies a severe discrepancy criterion. Moreover. as discussed in an earlier section. a major shortcoming of the federal guidelines is the failure to define "severe." Shepard's (1980) regression discrepancy method. like Erickson's formula. fails to specify an empirically verified severe discrepancy criterion. However. the biggest drawback to using her formula is a practical issue. She advocated using only ability and achievement measures that have been co-normed. She indicated that. by doing so. errors related to differences in normative procedures can be avoided. As Wiseman (1984) pointed out. the practicality of following this procedure is questionable because none of the most frequently used diagnostic tests would meet this criterion. A 1981 study by Algozzine and Ysseldyke suggested that Shepard's precaution may be unnecessarily conservative and perhaps even unfounded. In their study. the authors compared the reliabilities of differences between tests normed on the same population. as well as between tests normed on different populations. Their findings revealed the reliabilities were comparable. These results have particular rele- vance for the present investigation comparing the WISC—R and the K-ABC. Kaufman and Kaufman (1983b) indicated that their test battery is par- ticularly suited to evaluating LD and other exceptional populations because the same norm group was used to standardize the ability and achievement items in the battery. They maintained that. unlike other tests (WISC-R and PIAT). the K-ABC ability and achievement comparison 56 will not be contaminated due to differences in normative procedures for two different instruments. Considering the results of the Algozzine and Ysseldyke study. it appears that comparison of the HISC—R with the PIAT Achievement Test will not result in more contamination than would be expected from a comparison of the K-ABC ability and achievement scores. Moreover. comparison of the WISC-R and the X-ABC will not be confounded by errors related to differences in normative procedures. In summarizing the results of his evaluation. Berk (1984) indi- cated that the unreliability of the nine discrepancy procedures be reviewed would lead to a high number (not specified) of incorrect diagnostic decisions. and therefore he could not recommend any of the procedures for identifying LD children. However. he referred to Reynolds' (1981. 1985) method. which includes determining the reliabil- ity of the discrepancy between two scores. as having merit for applica- tion to the problem of LD identification. He suggested that determining the reliability of a discrepancy score is an important first step to improving the usefulness of ability-achievement discrep- ancy procedures. The second step. after a score is found to be reli- able. is to determine whether such a discrepancy characterizes only LD children. This second step. determining the validity of using discrep- ancy scores for identification of LD. has not yet been demonstrated in the research literature. Reynolds offered a _Z_-score method that allows a correction for differences in test reliabilities. thereby controlling for errors due to chance and measurement. as well as those related to differences in 57 normative procedures. He indicated that the correction. together with the use of calculated _Z_-scores. reveals real or £322 differences in performance. He advocated this method because the use of standard scores. unlike grade and age equivalents. controls for the mean at a given age level and the distribution of scores about the mean. .Accord- ingly. he stated that this method provides for constancy of aptitude- achievement discrepancies across age. His formula for the difference between scores results in a _Z_-score discrepancy that can be compared to the normal curve to determine statistical significance. Rules of thumb indicating the required reliabilities of each test and the related amount of deviation required for significance (probability inference that an observed difference between two scores is not due to chance or sampling error) are provided to allow examiners to ascertain signifi- cant differences without actually having to solve the formula for each new test given. Because test scores must be expressed on a common scale. a conversion chart for equating standard scores is also pro- vided. Computation of the method involves solution of three equations. Although the calculations are complicated. they are useful in terms of providing reliable discrepancies (Berk. 1984). The three-step formula is as follows: (x. - 3?) (1) _ 1 zx — Ox Y. - 37 (2) Z = ( 1 ) 58 D (3) Z = 1 2 L v/(l-rxx) + (l-ryv) O Wiseman (1984) provided an easily understood interpretation of this formula by giving the following written instructions: Absolute difference between firscore (ability) ~ gyscore (achievement) = gfscore /(1-re1iability) + (l-reliability) This method is considered to be statistically reliable and practical (Wiseman. 1984). Although calculating the formula is time consuming. the dividends resulting from the analysis are justifiable in terms of applicability to LD identification criteria (Berk. 1984). As a result. Reynolds's method is endorsed for the present investigation comparing LD and LA students. Discriminative Efficiency of the Discrepancy Model Although practitioners continue to look for ways to use discrep- ancy data to define learning disabilities. research remains to be conducted to determine the prevalence of significant ability- achievement discrepancies in the nonhandicapped population. The need to establish the prevalence of "average" ability-achievement discrepan- cies in the general population is an issue of considerable import. Without knowledge of base rate data. it is impossible to decide if a significant discrepancy is valid for classifying LD children (Algozzine 59 8: Ysseldyke. 1982; Berk. 1984: Salvia & Clark. 1973). Once a discrepancy score is found to be reliable (statistically significant). a discriminative efficiency study is needed to determine estimates of classification accuracy for identifying only LD children (Berk. 1984). Returning briefly to the study conducted by Ysseldyke et al. (1982) (reviewed earlier in the discussion of psychometric comparisons between LD and LA populations). the use of the discrepancy notion yielded a 402 misclassification rate. It will be recalled that. in their study. the investigators compared 50 school-identified LD chil- dren and 49 LA children (those who scored below the 25th percentile on the ITBS). using ability-achievement discrepancy criteria of the fed- eral definition for l. 1.5. and 2 standard deviations. Their results revealed that no children in either group were achieving 2 standard deviations below ability. When a 1 standard deviation deficit was employed. 37 LD and 35 LA students would have been classified as LD. Comparison to the school's classifications. (students labeled LD) yielded 51.5% agreement. Using the 1.5 standard deviation discrepancy. 24 LD and 17 LA students would have been diagnosed LD. matching the school's identification at a rate of 56.52. Whereas discriminant function analysis revealed 78.42 overall correct classification. the analysis of results revealed that. on an individual versus group- classification basis. the discrepancy notion would be inefficient with regard to differential classification of LD and LA students. Similarly. Algozzine and Ysseldyke (1982). performing the same comparisons with 40 school identified LD students and 40 LA students 60 (those who scored below the 25th percentile on a group achievement test). found that a large proportion of school- identified LD children failed to meet operationalizations of the federal guidelines. and many LA children would have been LD according to at least one of the opera- tionalizations. These inconsistent results did not provide decision rules for individual diagnosis. Algozzine and Ysseldyke (1983). referring to an earlier study (Algozzine. Ysseldyke & Epps. 1982). revealed that when commonly used definitions of LD were applied to "normal" students. more than 752 could have been labeled LD. and 252 of the school-identified LD stu- dents could not be classified by the same criteria. Wiseman (1984) achieved similar results with his two groups of LD and LA students. His overall error rate yielded incorrect classifica- tions for 352 of the LD and 462 of the LA. Kaufman and Kaufman (1983). in discussing performance characteris- tics of LD children. indicated a pattern in which the simultaneous processing standard score of the mental ability portion of their bat- tery exceeded the standard scores on the Achievement Scale by about half of a standard deviation. They indicated that these discrepancies were consistent with the nature of the diagnosis and reaffirmed the importance of having separate ability and achievement measures within the same battery. Naglieri (1985) investigated the pattern of scores for LD children using both the WISC—R and the K-ABC. His findings did not support Kauman and Kaufman's description of a characteristic test profile. In fact. he found that Mental Processing-Achievement means 61 revealed no significant differences in comparison to normative values. As with the previous studies. this investigation did not support the diagnostic utility of an ability-achievement discrepancy. However. Naglieri did not include the proper comparison to a nonhandicapped LA group. so no conclusive statements can be made regarding differential diagnosis of LD and LA populations using the K-ABC. On the basis of these studies. it would be rash to suggest unequivocally that significant ability-achievement discrepancy does not distinguish LD from LA groups. There is evidence from at least one of the studies that correct classification exceeds chance levels and that more LD than LA children meet definitional criteria. On the other hand. a large proportion of LA students have ability-achievement dis- crepancies that meet federal guidelines and match or even exceed dis- crepancy points attained by LD students. Verbal-Performance Ability_Differences At least as widespread. and perhaps even more alluring than the attempt to link ability-achievement discrepancy to LD. has been the attempt to use verbal-performance discrepancies on the WISC and WISC-R as diagnostic indicators for identifying LD children. Much has been written on this topic. despite the paucity of support for the use of such differences to separate LD from normal and other exceptional populations. Reviews have been completed on this issue (Veres. 1982; and Wiseman. 1984). Consequently. the current discussion attempts only to highlight the major studies and present the current status of the concept. However. following this discussion. a more in-depth review of 62 the analogous sequential-simultaneous discrepancy on the Kaufman Assessment Battery for Children will be presented. As with many early attempts to associate specific WISC-R patterns with the diagnosis of LD. observations of differences between verbal and performance IQ scores of LD children provided suggestive evidence that particular score patterns would be useful for differential diag- nosis (Veres. 1982). Ackerman. Peters. and Dykman (1971) indicated that the LD children in their research evidenced significant verbal- performance discrepancies of 15 points or'more in favor of the perform- ance IQ. The thought that such patterns might typify the profiles of LD provided the impetus for the prodigious number of investigations that followed. However. as later determined by their research in 1976. Anderson. Kaufman. and Kaufman acknowledged statistically significant verbal-performance differences (regardless of direction) for LD chil- dren but cautioned that. despite the statistical significance. the majority of their LD sample did not have unusually large discrepancies. This finding was determined in part from an earlier investigation in which Kaufman (1976a) examined V-P differences of the standardization group of the WISC-R. He found that 482 of the norm group averaged V-P discrepancies of 9 or more points. 342 averaged differences of 12 or more points. and 251 had differences of 15 points or greater. Support for base-rate data in the normal population was gathered by others as well (Gutkin. 1979: Schooler. Beebe. & Koepke. 1978; Thompson. 1980). Before that time. research had been conducted without benefit of knowledge regarding the prevalence and direction of V—P 63 discrepancies in the general population. These new studies illuminated the fact that the normal population was not characterized by flat WISC-R profiles. a belief that had made V—P discrepancies so enticing to researchers and practitioners alike. Gutkin (1979) demonstrated considerable overlap in distributions of normal and exceptional populations with respect to V-P differences. Schooler. Beebe. and Koepke (1978) were unable to separate LD from other diagnostic groups on the basis of V-P discrepancies. either in terms of size of discrepancy or direction. Similarly. Thompson (1980) reported that employing V-P discrepancies was not useful for discrimi- nating LD from normal or any other diagnostic groups. As Veres (1982) pointed out. "a reliable finding of higher per- formance IQ among the learning disabled in the absence of a similar pattern among the non-learning disabled could have provided a useful diagnostic indicator” (p. 6). Her investigation. like those of Yssel- dyke et al. (1982) and others (Clarizio & Bernard. 1981; Henry & Wittman. 1981: Piotrowski. 1978; Ryckman. 1981). failed to find psycho- metric differences of practical utility for differential diagnosis. In summarizing the majority of findings in this area. there is no defensible base supporting the use of V-P differences for differentiat- ing LD from other handicapping conditions or from a larger group of referred but not placed children. 64 Sequential-Simultaneous Ability Differences It will be recalled from the overview of the K-ABC presented in Chapter II that the sequential and simultaneous scales were described as two types of mental processing. Kaufman and Kaufman (1983b) indi- cated that these two modes of information processing have strong theo- retical foundations in the fields of neuropsychology and cognitive psychology. Sequential processing emphasizes the serial or temporal order of stimuli when solving problems. and simultaneous processing emphasizes gestalt-like. frequently spatial. integration of stimuli. They contrasted these styles of processing with that of the verbal- performance organization of the WISC-R by indicating that the WISC-R is defined by the content of its stimuli (verbal or nonverbal). and the K-ABC intelligence scales are defined by the processes used for problem solving. That is. the focus is on whether the stimuli are manipulated one at a time or simultaneously. regardless of item content. In further analyzing the two processing scales of the K-ABC. Kaufman and Kaufman indicated that sequential skills are closely related to a number of school-related skills. For instance. they indicated that this type of processing is involved in memorizing number facts. lists of spelling words. and associations between letters and their sounds. Simultaneous-processing skills are described as being related to the ability to form gestalts to facilitate performance of perceptual tasks. such as learning shapes of letters and numbers or deriving meaning from pictures and other visual stimuli. Moreover. 65 they suggested that this type of processing is related to higher-level intellectual functions because of the implied integration components. Regarding the organization of the K-ABC into its sequential- simultaneous scales. Kaufman and Kaufman revealed that. in part. the rationale for the organization of their test was related to research on the WISC-R involving the LD. They pointed out that the V-P dichotomy does not sufficiently explain the performance characteristics of LD children. One can assume from the rest of their discussion that they considered the sequential-simultaneous dichotomy to be superior both as an explanatory construct and as a diagnostic aid. In their review of diagnostic implications of the dichotomy and their comparison with the WISC-R. Kaufman and Kaufman indicated that the V—P discrepancy of the WISC-R did not adequately explain character- istic profiles found for LD children. (Their use of "characteristic" here is surprising in view of the senior author's 1976a investigation. which would have suggested "characteristic" LD profiles would not be found.) They pointed out that LD children seem to have a strength in simultaneous processing and weaknesses in both sequential processing and achievement when Bannatyne's (1971. 1974) groupings are applied to Wechsler profiles. Although they indicated that the simultaneous- sequential discrepancy was less sensitive than what they had antici- pated as a potential diagnostic indicator of LD. they went on to cite studies of LD children. LD referrals. and dyslexics. stating that the subtest profiles of these children revealed patterns that would be useful for assessment purposes. In further elaborating on the 66 usefulness of the sequential-simultaneous approach. Kaufman and Kaufman indicated that one investigation revealed that sequential-processing subtests were found to be excellent discriminators of normal and dys- lexic children and that high simultaneous-low sequential patterns were consistently found to characterize the profiles of LD children. This study. as well as others. is examined later in this section. The implications to be drawn from their comments would suggest that the sequential-simultaneous dichotomy has diagnostic utility for identi- fying LD children. The authors' reference to Bannatyne's groupings is astonishing in view of the considerable amount of empirical evidence that does not support the diagnostic utility of these composites. As Veres (1982) so aptly stated. “in the few studies of recategorized scores that did have appropriate controls there was no convincing support for the use of these composites for differential diagnosis (Schooler. et. al.. 1978; Ryckman. 1981: Clarizio & Bernard. 1981)" (p. 16). Apparently. the weight of the accumulated evidence has not influenced the K-ABC authors' belief about the applicability of these groupings. Kaufman and Kaufman referred to the Hooper and Hynd (1982) study as promoting optimism regarding the use of the K-ABC for diagnosis of dyslexia. In this study. the investigators compared the K-ABC test patterns of 30 normal readers (selected from the standardization popu- lation of the K-ABC) with those of 87 reading-disabled students (selected from a larger group of students involved in a reading- disability program). The reading-disabled group was further subdivided 67 into Boder's categories (nonspecific. dysphonetic. dyseidetic. and alexic). The results of the study indicated that the sequential- simultaneous discrepancy was not useful for differential diagnosis and that no consistent. significant high similarity-low sequential pattern was observed. However. the normal group did score significantly higher on sequential subtests in comparison to the four reading-disabled groups. Discriminant function analysis produced an overall hit rate of 35% across the five groups. The highest rates of correct classifica- tion occurred for normal and alexic groups. for whom hit rates were 56.72 and 55.01. respectively. The investigators concluded that the 1(- ABC in general. and the sequential factor in particular. provided a useful method for discriminating between normal and dyslexic readers. This conclusion. considering the percentage rates. was. at the very least. overdone. Methodological flaws aside (in particular. the absence of a low-achieving comparison group). there is no compelling evidence to support either the investigator's claims or Kaufman and Kaufman's enthusiasm. Moreover. the practicability of the findings of this study in terms of individual diagnostic situations is question- able. Klanderman. Perney. and Kroeschell (1983) compared the WISC-R and K-ABC scores of 44 LD children in self-contained primary and interme- diate classes. They apparently compared these children's performance with that of the normative sample of the K-ABC (the comparison group was not specifically described). They concluded that the two tests were substantially correlated and that LD children used processes more 68 characteristic of the mentally retarded than did the standardization group. What the investigators actually meant by this statement was not clear from their article: however they suggested normative groups used more efficient processes. In addition. they indicated that the LD children tended to use a combined processing method with apparently slightly more simultaneous than sequential processing. The reported means indicated nonsignificant differences between sequential- simultaneous processes. Although the study suffered from many methodo- logical flaws. the lack of a nonhandicapped-LA comparison group was paramount. These results. like those of the previous study. did not support the utility of the sequential-simultaneous comparison for diag- nosis of LD. In another study. Haddad (1983) compared test scores on the WISC—R and K-ABC for 33 initially referred children found eligible for LD services. This study suffered from the same primary methodological flaw (no comparison to a LA group). The results failed to supply evidence for the usefulness of the simultaneous-sequential discrepancy in the diagnosis of LD children. A final study conducted by Naglieri (1985) compared the perform- ance of LD. borderline mentally retarded. and normals on the WISC-R and K—ABC. The results indicated that the LD children did not have either a distinct WISC-R pattern or a K-ABC pattern. The three groups earned essentially equal amounts of WISC-R Verbal-Performance. K-ABC Simultaneous-Sequential. and Mental Processing-Achievement discrepancy scores in comparison to national norm groups. The investigator 69 concluded that a Sequential-Simultaneous Scale disparity may not typify LD children under current state and federal guidelines. As noted in the previous studies. this investigation also lacked a comparison nonhandicapped but LA population. Base rates from Kaufman and Kaufman's (1983) standardization group revealed that the average child had a difference between Sequential and Simultaneous Processing standard scores of 12.3 points. the level required for significance at the .05 level. For abnormal scatter of at least 1 standard deviation above the normative mean. a discrepancy of 22 points would be required. In summary. as with the WISC-R research. there is no supporting evidence that LD can be differentiated from other groups on the basis of the K-ABC Sequential-Simultaneous discrepancy. Ability Subtest Scatter Comparison The preceding section examined specific patterns thought to char- acterize LD children. Another scheme frequently employed to aid in diagnosis has been to search for abnormally variant intratest scatter. It has been hypothesized that the LD exhibit more scatter than the non- LD. Veres (1982) examined the usefulness of scatter indices on the WISC-R and found no meaningful significant differences between clinical and normal groups. Others have found similar results in terms of , considerable overlap of WISC-R subtest scatter for LD and normative samples (Algozzine 6. Ysseldyke. 1982: Clarizio & Bernard. 1981; Cutkin. 1979: Ryckman. 1981; Ysseldyke et al.. 1982). 70 Using the WISC-R standardization sample. Kaufman (1976a) found that. on the average. normals had 7 j; 2 points between their lowest and highest scaled scores and had an average of 1.7 1 .02 significant deviations between subtest scores. (Significant deviations are defined as being equal to 3 or more points different from a child's own subtest mean.) The results of numerous studies using the WISC-R with various special education populations have been confusing and inconsistent. Although a number of special education populations have demonstrated significantly more scatter than control groups of normals. large num- bers of children in the normal samples in the same studies have often had the same or even more scatter (Gutkin. 1979: Thompson. 1980). Consequently. significant scatter has not been useful for purposes of differential diagnosis. Similarly. Chatman. Reynolds. and Wilson (1983) examined base rates of both interscale and intrascale scatter (within-scale subtest variability on the Sequential. Simultaneous. or Achievement scales) on the K-ABC. Their results paralleled that of the WISC-R in revealing that flat profiles did not characterize normal groups. They found that almost 601 of the children had an overall range of 6 or more points for the entire battery of Mental Processing Composite subtests. 0n the individual scales. the mean scatter indices were 5 or more points on simultaneous processing and 3 or more points on sequential subtests. The median number of significant differences on the Mental Processing Scales for the normative group was at least one significant difference. 71 In his study. Naglieri (1985) found that the two exceptional groups (LD and borderline) had very similar subtest profiles. In addition. he found that these two populations did not evidence greater scale discrepancies than their normative comparison. In summary. there does not appear to be a significant relationship between WISC-R or K-ABC subtest scatter and learning disabilities. However. not as many studies have been conducted with the K-ABC as have been done with the WISC-R. In addition. comparison of LD and LA populations on the K—ABC has not been examined. One purpose of the present effort is to make that comparison. General Ability and Achievement Characteristics Based on the related literature. very little is clear with regard to typical characteristics of LD students. A myriad of investigations with the WISC-R have offered little evidence that any particular array of subtest scores provides consistent defining characteristics of LD. When compared to federal guidelines. as many as 502 of school- identified LD children fail to exhibit test score characteristics consistent with the definition (Norman 5: Zigmond. 1980). However. to say that there have been no instances in which comparison groups have differed significantly would not be true. Some investigations have noted statistically significant differences in specific abilities and achievements with certain groups. This section reviews some of these differences and discusses the implications with respect to standardized tests and the diagnostic process. 72 Warner et a1. (1980) compared the test scores of junior and senior high school LD and LA students. In achievement areas. as measured by the Woodcock Johnson Psychoeducational Battery. significant differences were noted between the two groups. Across both age levels. LD students performed at significantly lower levels than did LA students on the reading. written language. and mathematics subtests. Kaufman and Kaufman (1983b) summarized the results of a number of studies of LD children completed during the standardization of their test battery. They maintained that. as a group. LD students. LD refer- rals. and dyslexics had Simultaneous Processing Standard scores that averaged 2 to 5 points higher than Sequential Processing Standard scores. Regarding subtest performance. they indicated that these 'chil- dren consistently earned their highest scores on Gestalt Closure. Triangles. and Riddles. Poorest performances were noted on Hand Move- ments. Word Order. Paces and Places. Arithmetic. Reading/Decoding. and Reading/Understanding. In addition. the investigators indicated that Simultaneous Processing standard scores exceeded standard scores on the Achievement Scale and Reading subtests by about half a standard devia- tion (7 to 8 points). Moreover. they indicated "these patterns are consistent with a bulk of research on the WISC and WISC-R that has shown reading- and learning-disabled children to perform well on Bannatyne's (1971. 1974) Spatial category. and poorly on his Sequential and Acquired Knowledge groups" (pp. 141-142). Kaufman and Kaufman did not seem to be overly concerned that these samples were not systematically matched as a result of their combining 73 the findings from several separate investigations. As discussed earlier. the findings of research on recategorized scores (Bannatyne's regroupings) apparently have had little effect on the authors' notions about learning disabilities. It remains to be seen if these patterns on the K-ABC are useful for separating LD children from other mildly handicapped children and from nonhandicapped LA children. If these patterns are found to have diagnostic utility in the absence of similar findings with the WISC-R. the K-ABC may indeed be a major advance in intelligence testing (K-ABC Information Exchange. March 1983). The present investigator explored these possibilities. Perhaps as Shinn et al. (1982) suggested. many of the commonly used standardized tests are indirect measures of academic performance skills and are unable to detect differences that exist between handi- capped and nonhandicapped populations. Using the same samples of LD and LA students as Ysseldyke et al. (1979). these investigators found what they considered to be true differences between the two popula- tions. Instead of traditional tests. they administered weekly competency-based functional measures of classroom performance in read- ing (word lists). spelling (word lists). and written expression (vocabulary words) during a 5-week period. The reliabilities for these measures were reported as .90. .85. and .59 for reading. spelling. and written expression. respectively. The results indicated consistent differences between the two groups in reading and spelling in terms of number of words read and number of words spelled correctly. Group- level differences revealed the LD students read at a rate approximately 74 502 less fluent than the LA.group. In spelling. the samerrelationship existed. Conflicting findings in written expression made it difficult to interpret the results across the two groups. Considering the mod- erate reliability of the written-expression measure. the need for more thorough investigations of writing measures was suggested. The investigators discussed their results in terms of differences in the two measurement systems (standardized, tests versus nontradi- tional competency-based measures) and the implications for classifica- tion decisions. They noted that. as a group. those receiving special services (LD) perfonmed at lower levels than those not receiving serv- ices (LA). It was speculated that students may be receiving LD serv- ices on the basis of having the lowest academic- achievement levels (certainly the results of Sexton & Street. 1985. would support this hypothesis). Because no differences between LD and LA students were noted on standardized psychological tests (Ysseldyke et al.. 1979) but were observed on competency-based measures. the researchers raised the issue that perhaps schools did adequately differentiate between.the two populations. Issues of the overall diagnostic process were discussed vis-a-vis the finding that actual differences in ability and skills may only be observable in terms of classroom performance and may not be reflected on standardized test instruments. This suggested that the placement team may substantiate the teacher's opinion on the basis of factors unrelated to the comprehensive standardized test data. One of the purposes of the present investigation was to determine whether any information in standardized test data can be used to 75 diagnose learning disabilities. Veres (1982) asked this same question when she examined the factorial composition of the WISC-R subtests with respect to LD diagnosis. This question needs to be repeated with the K—ABC. If the answer is negative. perhaps we need to address the methods and team process used for diagnostic decision making rather than the definitions and operationalizations we apply to current test procedures. Verbal-Performance Differences and Enotional Impairment Test score patterns for identifying emotionally impaired (EI) children have not received as much attention in the literature as those used to diagnose LD youngsters. However. one only needs to scan the literature to realize that suggested test patterns for emotional impairment are closely modeled after LD profiles (Veres. 1982). In particular. speculation about a link between Verbal-Performance IQ discrepancies on the WISC-R and emotional impairment corresponds closely to the work in the area of learning disabilities (Veres. 1982). Bren the direction of the Verbal-Performance differences parallels that described for learning disabilities. As a group. EI children have been described as exhibiting a depression in verbal functions on the WISC-R ' (Dean. 1978). This pattern. illustrated by Performance IQs that sig- nificantly exceed Verbal IQs. has been reported in a number of studies (Dean. 1978; Lewandowski. Saccuzzo. & Lewandowski. 1977; Saccuzzo & Lewandowski. 1976). As a result. some researchers have considered this pattern to be a useful sign for diagnosing EI children. 76 A few of the investigations reporting this pattern have been selected for discussion because they typify the quality and genre of research in the area (before the better-controlled study of Clarizio & Veres in 1983) and because they have been largely responsible for the promulgation of the putative EI test pattern. Of particular concern are the investigations conducted by Dean (1978); Lewandowski. Saccuzzo. and Lewandowski (1977); and Saccuzzo and Lewandowski (1976). The samples used in these studies were limited to conduct-disordered adolescents. incarcerated juvenile offenders. and court-referred juvenile offenders. respectively. Pull-Scale IQs ranged from 70 to 89. All of the studies were consistent in the finding that Performance IQ was greater than Verbal IQ for the majority of the subjects. None of the studies included a normal control group or another handicapped/diagnostic category for comparison (EI without conduct disorders. LD. or a LA control). Referring to the above studies. Clarizio and Veres (1983) indi- cated that the lack of a normal control group for comparison made it impossible to ascertain if a Performance IQ greater than Verbal IQ was characteristic of E1 children and. more important. if such a pattern facilitated differential diagnosis. In their own study. they made direct comparisons of E1 and normal children to assess the diagnostic utility of this supposed EI pattern. For the purposes of their inves- tigation. they decided to accept the pattern as useful if it led to correct diagnostic decisions 85% of the time. Their results revealed that the pattern did not distinguish between the normal control group 77 and the E1 group. Moreover. they found that even when they doubled the allowable error rate (from 15% to 30%). the pattern was still not useful for differential diagnosis. Overall. a discriminant-function analysis revealed that any lZ-point difference (regardless of direc- tion) correctly identified only 632 of the E1 children and that Per- formance greater than Verbal (by at least 12 points) successfully identified EI students 66% of the time. The present study extended the work of Clarizio and Veres by including a LA comparison group and an LD group. The discriminative efficiency of this pattern for separating EI children from LD children. as well as from a larger group of LA but not handicapped children. remains to be assessed. Sequential-Simultaneous Differences and Emotional Impairment A perusal of the K-ABC literature revealed a paucity of studies regarding the use of this new instrument for EI populations. The manual presents only one study with a group of 44 behaviorally dis- ordered children. This study was completed by Nelson shortly before publication of the K-ABC. probably in 1983 (no date was given). A computer search of the literature did not uncover a published account of this investigation. Consequently. comments in this section are based solely on Kaufman and Kaufman's interpretation of this investi- gation. As described in the manual. the behaviorally disordered group of children achieved scores on the Mental Processing Composite and 78 Sequential and Simultaneous Processing Scales that were virtually iden- tical to those of the LD group. That is. Simultaneous Processing standard scores averaged 2 to 5 points higher than Sequential Process- ing standard scores. Emotional Impairment: KPABC Mental-Processing Subtest Patterns The highest subtest scores for both groups (EI and LD) were achieved on Gestalt Closure and Triangles. Relatively poorer scores were achieved on Word Order. Hand Movements. and Matrix analogies. The only differences reported between the two groups with respect to Mental Processing were on Number Recall and Spatial Memory. (Apparently. Number Recall was the third highest subtest for behaviorally disordered children. and Spatial Memory was a "striking weakness" for them as well. The ranking of Number Recall was not given for the LD group. and "striking weakness" was not clarified. Emotional Impairment: K-ABC Achievement The primary differences between the two groups were found on the Achievement Scale. The LD group reportedly scored 5 points lower on the Achievement Scale than on the Mental Processing Composite. and the behaviorally disordered group scored 2 points higher on the Achieve- ment Scale. In terms of individual subtest performance on this scale. the behaviorally disordered group apparently scored higher than the LD group by 3 points on Arithmetic and by 7 or 8 points on the other four subtests. Riddles served as the highest achievement score earned by both groups. Kaufman and Kaufman indicated that no conclusions should 79 be drawn from this study because of the lack of additional studies of behaviorally disordered populations. The question of whether the aforementioned profile is characteristic of EI/behaviorally disordered children needs to be addressed. The current study included direct comparisons of EI. LD. and a LA comparison group to examine the utility of this profile for differential diagnosis. Emotional Impairment: High Similarities.4Low Information Based on his studies using the WISC-R. as well as a review of the research. Dean (1977) concluded that EI children earned their highest or second highest Verbal subtest score on Similarities and their lowest or second lowest Verbal subtest score on Information. Dean was con- vinced that this pattern characterized EI children. despite the fact that be included no comparison groups in his study. In his review of other investigations (primarily Saccuzzo & Lewandowski. 1976). he reported that in approximately half of the cases. the high Similarities-low Information pattern was present in the profiles of E1 children. Saccuzzo and Lewandowski also failed to include a control group. Clarizio and Veres (1983) examined this pattern of scores to determine if such a pattern is useful for differentiating EI students from normal controls. They compared the rate of occurrence of the pattern for the E1 group versus the normal group. The pattern was analyzed in the following two ways: (a) a pattern was present if not 80 more than one Verbal subtest was more extreme than the subtest of interest. and (b) a pattern was present if not more than one Verbal subtest score was more extreme than the score of interest. and the score of interest was at least 3 points different from the mean of the remaining four Verbal subtest scores. In either analysis. four out- comes were possible: both patterns could be present. neither pattern could be present. only the Similarities pattern was present. or only the Information pattern was present. The results of Clarizio and Veres's investigation revealed that neither of the above-mentioned analyses supported this pattern of sub- test scores for E1 children and that not even 50% of the children could be classified using these patterns. The present investigation assessed the clinical utility in the same manner as described by Clarizio and Veres but included the rate of occurrence of the pattern across LD. EI. and LA groups. Direction of the Present Study The literature has suggested that there is little supporting evidence for the notion that intelligence tests are useful for differ- ential diagnosis of educational handicaps. Research with the WISC-R has indicated some evidence of significant subtest score differences between groups. but these group-level differences have not been useful at the individual-case level. In addition. some investigations have revealed significant differences in specific abilities and achievements that may distinguish LD students from other nonhandicapped LA stu- dents. but the practical relationship of these findings for individual 81 diagnosis has not been established. As discussed throughout this chapter. attempts to use the WISC-R to verify school-identified classifications have resulted in little agreement and considerable misidentification when operationalizations of the federal definition have been applied. A new test. published in 1983. the Kaufman Assessment Battery for Children. has been touted by the publisher as a major advance in the field. The test authors described it as useful for psychoeducational evaluation of LD and other exceptional groups. Formal announcements introducing this instrument implied favorable differences between the WISC-R and the K-ABC. The organization of the instrument (separating ability from achievement) has been promoted as offering fairer and more accurate assessment of specific groups of children. particularly the LD. An important question is whether using this test will improve assessment and placement services to exceptional children. In particular. there is a need to determine whether distinct score profiles on the K-ABC are characteristic of specific educational handi- caps. Ascertaining the practical utility of using subtest score pat- terns for differentiating among LD. EI. and LA children (referred but not handicapped) became the major objective of the present investiga- tion. Although some studies with the K-ABC have addressed questions of diagnostic utility. none has employed an LA comparison group. It will be recalled from an earlier discussion that low academic achievement is a primary characteristic shared by children with mild educational handicaps. Therefore. the primary concern of the present investigation 82 was to evaluate the usefulness of subtest arrangements for differential diagnosis by comparing educationally handicapped children with an LA sample. The validity of differentiating among the groups in the study was assessed by computing the probabilities of false decisions for the presumed characteristic test patterns and the various diagnostic methods suggested in the literature. The school classifications served as the external criterion for comparison of classifications. By also comparing the proportions of correctly identified students (those matching or verifying school-identified classifications) by WISC-R versus K-ABC. the present investigation compared the diagnostic or discriminative efficiency of the two instruments. Determination of whether the WISC-R or the K-ABC is more effective in correctly identifying LD students was not the primary objective of the present study. 'A number of investigations with considerably larger sample sizes would be needed to begin to address that concern. How- ever. the current comparison of correct classification rates by test may help to formulate hypotheses for future research designed to evalu- ate the classification effectiveness of these two instruments given current federal and state guidelines. The major hypothesis of the present investigation was that: The test profiles of LD. EI. and.LA students will not yield decision rules useful for differential diagnosis because overlap in test performance will result in disturbingly high percentages of incorrect diagnostic decisions. The specific applications of this hypothesis include 83 comparisons to ability-achievement discrepancies. Verbal-Performance and Sequential-Simultaneous IQ differences. subtest scatter. and the general form of ability and achievement profiles. CHAPTER IV METHOD Source of Data The data used in the study were collected during the 1983-1984 school year by the School Psychology Program of Michigan State University in cooperation with the 1983 Fall Institute of the Michigan Association of School Psychology. The Department of Counseling and Educational Psychology sponsored a continuing education course entitled "Kaufman Assessment Battery for Children.” The course was offered to practicing school psychologists or school-psychology interns with prior courses in tests and measurements and individual mental assessment. The objectives of the course were to provide didactic training in the theoretical rationale underlying the K-ABC and to provide field expe- rience in the administration. scoring. and interpretation of the bat- tery. 0f the 32 examiners enrolled in the course. 30*were practicing school psychologists and were included in the project. The other two examiners were interns. and the data from their evaluations were excluded from the study in order to avoid errors due to inexperience. Each examiner was asked to submit at least five cases each. all of which were initial referral s. All of the children were referred for evaluation because of academic. behavioral. or physical problems that 84 85 interfered with classroom performance. To fulfill requirements for the course. each examiner was asked to submit K-ABC. WISC-R. and PIAT test- evaluation protocols along with data forms for each child. containing reason for referral. parents' socioeconomic status (SES). child's sex. age and grade placement. and special education diagnosis and placement. In addition. each examiner was asked to provide information regarding examiner's sex. highest degree earned. number of years in professional practice. and school district size (rural. urban. suburban). Examiners were instructed to alternate the order in which the K-ABC and WISC-R were given from case to case. Female psychologists outnumbered males. comprising 60% of the participants. There were no Hispanic or black psychologists. The number of years of professional experience ranged from less than 1 year to 22 years. with an average of 8 years. All were certified school psychologists. which means that all had at least an M.S. degree. Examiner characteristics are presented in Table 4.1. Sample The children were assigned by the Educational Placement and Planning Committees (EPPCs) to five classification categories: learn- ing disabled (3 = 51). emotionally impaired (33 = 20). educable mentally impaired (n = 14). physically or otherwise handicapped (2 = l). and not eligible for placement (2 = 61). (One examiner submitted only two cases. so that there were 147 children for whom data were available.) 86 Table 4.1: Examiner Characteristics N, 1 Total number 30 100 Sex Male 12 40 Pemal e 18 60 Race White 30 100 Black 0 0 Hispanic 0 0 Years of experience 1 through 4 12 40 5 through 8 5 l7 9 through 12 6 20 13 through 16 5 l7 17 through 22 2 6 Highest degree M.S. 12 40.0 Ed.S. 10 33.3 Ph.D. 1 3.3 Not reported 7 23.3 Note: Totals may vary because of missing data. Percentages are of the available data. It was assumed that the EPPCs followed mandated state placement guidelines and that classification decisions were accurate in terms of reflecting federal and state rules and regulations. However. not all of the factors influencing the EPPC decision-making process were known for the present sample. Therefore. the validity of the aforementioned decisions could be questioned. Nevertheless. the fact remains that the children were assigned to placements as indicated above and were 87 provided specialized services accordingly. Based on the major assumption.that the school-identified classifications were accurate. it was also assumed that the efficacy of the diagnostic methods used in the present investigation to corroborate EPPC decisions could be assessed. Children characterized as educable mentally impaired (EMI) or physically or otherwise health impaired (POHI) were excluded from the sample. The EMI children were excluded from the study because a primary objective of the present investigation was to determine if learning disabled (LD) youngsters could be differentiated from emotionally impaired (E1) and from a larger group of low-achieving (LA) but not handicapped children. State and federal guidelines mandate that the EPPC may not identify a child as having a specific learning disability if the severe discrepancy between.ability and achievement is primarily the result of mental retardation. Consequently. children attaining IQ scores below 70‘were systematically excluded from the analysis. Children classified as POHI (only one subject in the present sample) were considered at the outset of the study to be ineligible for participation because determination of their impairment is not based solely on psychological-test results. Identification guidelines require an evaluation by an orthopedic surgeon. internist. neurologist. pediatrician. or osteopathic internist. in addition to a psychological test battery. 88 Four additional cases (four not eligible) were eliminated from the sample because of incomplete data. Students not eligible for special education were designated as low achievers if they scored below the 33rd percentile on the PIAT Reading Recognition subtest. Of the 57 remaining "not eligible" students. a total of 31 youngsters met this criterion. These 31 were further subdivided into two levels of low achievement based on their test scores: LA < 33rd percentile and LA < 17th percentile. Sixteen stu- dents scored between the 17th percentile and 32nd percentile. and 15 students scored below the 17th percentile. Following the subject- elimination process. a total of 102 cases were available for the analy— sis in this study: 51 LD. 20 EI. 16 LA < 33rd. and 15 LA < 17th. Males accounted for 76.5% of the children in the sample. Ages ranged from 5 to 12-1/2 years of age. with a mean of 8.4 years. The grade placements ranged from kindergarten to seventh. with a mean grade level of 2.9 or approximately third grade. Suburban and rural school districts accounted for 87.2% of the students. and the remaining 12.8% were from urban areas. The SES of the children. as determined from parent's educational level (of the parent with the highest educational attainment). indicated 29% of the families had less than a high school education. 502 were high school graduates. 132 had some college. and 82 were college graduates. This suggests that approximately 63% of the sample were from middle-income families. 29% from low-income families. and 8% at high-income levels. Subject characteristics are outlined in Tabl e 4.2. 89 Table 4.2: Sample Characteristics Group LD EI LA < 33 LA < 17 N Z N Z N Z N 2 Total number 51 50.0 20 19.6 16 15.7 15 14.7 Sex Male 41 80.4 19 95.0 9 56.2 9 60.0 Panale 10 19.6 1 5.0 7 43 8 6 40.0 Race White 46 90.2 18 90.0 13 81.3 11 73.3 Black 5 9.8 l 5.0 2 12.5 2 13.3 Hispanic 0 0 1 5.0 l 6.2 2 13.3 Type of Referral Academic 45 88.2 13 65.0 12 75.0 14 93.3 Behavioral 4 7.8 7 35.0 3 19.0 1 6.6 Physical 0 0 0 0 0 0 0 Age Ranges 5- 7 18 35.3 4 20.0 6 37.5 6 40.0 8-10 29 56.9 10 50.0 8 50.0 5 33.3 11-12 4 7.8 6 30.0 2 12.5 4 26.7 Mean Age (§2) 8.1 (1.4) 9.1 (1.9) 8.4 (1.9) 8.6 (2.3) Grade Ranges 0-2 29 56.9 7 35.0 8 50.0 7 46.7 3-5 21 41.2 10 50.0 7 43.8 5 33.3 6-7 1 1.9 3 15.0 1 6.3 3 20.0 Mean Grade (SD) 2.6 (1.3) 3.5 (1.9) 2.8 (1.6) 2.8 (2.3) School District Urban 4 7.8 2 10.0 3 18.8 4 26.7 Rural 19 37.3 35.0 6 37.5 7 46.7 Suburban 28 54.9 11 55.0 7 43.8 4 26.7 SES (Parent Educ.) 0-11 years 11 21.6 6 30.0 4 25.0 8 53.3 12 years 28 54.9 9 45.0 7 43.8 6 40.0 13-15 years 6 11.8 4 20.0 3 18.8 0 0 16+ years 6 11.8 1 5.0 l 6.3 0 0 Table 4.2: Continued LD LA < 33 WISC-R-FSIQ IQ Level 70-79 80-89 90-110 111-130 131-150 Mean IQ (§2) K-ABC-MPC IQ Level 70-79 80-89 90-110 111-130 131-150 Mean IQ (g2) 3 5.9 11 21.6 30 58.8 7 13.7 0 0 96.9 (12.2) 97.4 (14.1) cocoa-J.- UINN OOOUIUI O 95.6 (14.1) 4 5 coco-H- 97.3 (13.8) 0 000 88.6 (10.1) 6 3 0. 0 0 con» (9.7) Note: Totals may vary because of missing data. available data. Percentages are of the b OOO‘U’O‘ 0"wa 90.5 (14.2) 91 Rationale for UsingL Readinjiecoding to Identify Low Achievers Although the federal guidelines for identifying learning disabilities indicate that a child may have a severe deficit in any of seven areas (oral expression. listening comprehension. written expres- sion. basic reading skill. reading comprehension. math calculation. and math reasoning). the current investigation focused on basic reading skills because reading problems are often the most frequent learning- referral problems encountered by school psychologists. Reid and Hresko (1981) reported that children with reading problems were viewed as naturally occurring at the lower end of the normal curve (p. 262). Consequently. it would be important to discrimi- nate between low-achieving readers and children with learning problems that interfered with reading ability. They suggested that reading problems are related to maturational lag and can easily account for other academic difficulties (writing. arithmetic). In terms of total test score on the PIAT. they indicated that Reading Decoding correlated .78-.88 with total test for early elementary grades. In their review. Rourke and Strang (1978) indicated that children with low arithmetic and stronger reading and spelling skills were not identified as having learning problems at early ages because language abilities masked weaknesses in arithmetic. Similarly. Nolan. Hammeke. and Barclay (1983). in comparing achievement patterns. found that math-disabled children did not differ from normals when equated for age and IQ level. 92 In her study. Bolt (1984) investigated subtypes of LD and found fewer numbers of children with low arithmetic performance. She suggested that children with lower arithmetic achievement were less likely to be identified as having learning problems. Relatedly. she found that very few children with stronger reading and spelling skills or‘with.higher'verbal than perfonmance IQ*were diagnosed as LD. In addition. she found reading scores to be relatively stable and that they changed less than other achievement areas in subsequent reevaluations of LD children. Based on these studies. it appears that reading,may be the primary focus of diagnosticians and EPPCs for identifying children as LD. Moreover. several researchers have suggested reading is an integral skill related to many other achievement areas. Consequently. reading achievement was suspected to have the most influence on the diagnostic decisions in the current investigation. Establishing_Decision Rules Particularly relevant to a study concerned with differential diagnosis is the general issue of the effects of labeling and the consequences correct and incorrect educational placements have on a student's future educational and vocational opportunities. Although the analyses of the present investigation did not specifically address these issues. they represent important considerations within the con- text of interpreting the results and speculating,about the possible factors that influence multidisciplinary teams to make decisions that do not conform to actual test data or classification guidelines. More 93 will be said about these issues in the discussion chapter. For now. it is important to consider the use of diagnostic decision rules to aid in differential diagnosis and to increase appropriate educational deci- sions. Selection procedures or decision rules are usually based on eco- nomic considerations in industrial/organizational settings. However. as discussed by Veres (1982). it would be irresponsible to make educa- tional placement decisions on this basis alone. Although there may be dollar costs associated with incorrect educational decisions. the social-psychological costs of mislabeling are of primary concern because of their potential negative influence on a child's later development. Unfortunately. as Veres (1982) pointed out. "the psycho- logical costs of mislabeling and misplacement of a child as well as the costs associated with not providing children with needed special pro- grams are impossible to assess with our present methodologies” (p. 26). Recognizing the difficulty in specifying costs. Ysseldyke et al. (1982) suggested that. at the very least. students who are denied special education services are deprived of opportunities for academic- achievement activities. Similarly. as implied by the indistinctness of this statement. even the benefits associated with correct diagnosis and placement remain vague. Consequently. as discussed by Veres (1982) and Wiseman (1984). since specific values (in terms of costs and benefits) cannot be assigned to correct and incorrect placement decisions. the establishment of educational-diagnostic decision rules and the assess- ment of their utility remain somewhat arbitrary. 94 The Utility_9f Diaggpstic Decision Rules ‘A major assumption of state and federal guidelines is that specific classes of handicapped children can be identified.and that specific educational interventions can be implemented to remediate categorical deficits. Characteristic sheets (outlining identification criteria) for the various handicapping conditions suggest that federal rules and state guidelines allow no errors in identification of handicaps (Veres. 1982; Wiseman. 1984). However. given that the most commonly used psychoeducational diagnostic methods are not 100% accurate. it is unreasonable to expect faultless classification. Nevertheless. a useful rule or method should lead to correct decisions at a level that exceeds chance or base rates (Berk. 1984; Meehl & Rosen. 1955; Veres. 1982; Wiseman. 1984). The arbitrariness of diag- nostic decision rules (referred to earlier) has to do with operation- ally defining the percentage of correct decisions above chance levels that will be accepted as the criterion for judging a rule useful. That is. a rule is considered useful if incorrect decisions do not exceed a designated error level. Consequently. while statistically significant differences may be obtained in some of the analyses of the current study. they would not offer direct evidence of the diagnostic utility of a decision rule if incorrect decisions exceeded the designated error rate. There are no specific guidelines to consult in setting decision rates. The homogeneity/heterogeneity of the groups. the unreliability/ 95 reliability of the criterion and of the test instruments used. and the costs related to incorrect diagnostic decisions should be considered in establishing an error level. In addition. a more practical concern to consider is the selection of an acceptance rule that would be agreeable to most school psychologists. The 852 correct decision rule used in earlier studies by Veres (1982) and Wiseman (1984) is somewhat high if one considers the unreliability of the criterion--EPPC classifications and the heterogeneity of the groups. However. a lower acceptance rule would be unacceptable to most psychologists who must routinely use test data and diagnostic methods to make recommendations that affect a child's educational career. Consequently. the current investigator operationally defined a rule as useful if it led to a correct decision 85% of the time. Conversely. any decision rule resulting in incorrect classifications at a rate exceeding 152 was considered not useful. In all analyses. the exact correct decision rates are reported. As described earlier by Veres (1982). the presentation of exact rates allows the reader to apply a different acceptance rule if desired. Hypotheses The primary focus of this research was to compare diagnostic methods for differential diagnosis of LD. EI. and LA students. The hypotheses for the present study were: Hypothesis la: WISC-R ability-PIAT achievement discrepancy scores do not distinguish among LD. LA < 33. LA < 17. and EI groups. Hypothesis lb: K-ABC Mental Processing Composite-K-ABC achieve- ment discrepancy scores do not distinguish among LD. LA < 33. LA< l7. and EI groups. 96 The test of this hypothesis was done in the following four ways: 1. A one-way ANOVA comparing groups on mean discrepancy levels was completed to determine if there were significant differences. 2. For each group. the proportion of students having a standard deviation of 1.0 or more was calculated and tested for significant differences. 3. For each group. the proportion of students having a standard deviation of 1.5 or more was calculated and tested for significant differences. 4. For each group. the proportion of students having a standard deviation of 2&)or more was calculated and tested for significant differences. Hypothesis 2a: WISC-R Verbal. Performance. and Pull-Scale scores do not differentiate among the four groups. A repeated measures ANOVA.by‘Verbal versus Performance was done to see if groups differed on Verbal. Performance. and Full-Scale scores and if Verbal-Performance discrepancy was larger for one group than another. Hypothesis 2b: K-ABC Sequential. Simultaneous. and Mental Pro- cessing Composite scores do not differentiate among the four groups. A repeated measures ANOVA by Sequential versus Simultaneous was completed to see if groups differed on Sequential. Simultaneous. and Mental Processing Composite scores and if Sequential-Simultaneous discrepancy was larger for one group than another. 97 Hypothesis 3a: WISC-R Verbal subscore deviations or Performance subscore deviations of 3 points or more from the Verbal and Performance means. respectively. do not distinguish among LD. LA < 33. LA < 17. and EI groups. Hypothesis 3b: K-ABC subscore deviations of 3 points or more from the mean of all the Mental Processing subtests do not distinguish' among LD. LA < 33. LA < 17. and EI groups. This hypothesis was tested in the following two ways: 1. A one-way ANOVA comparing groups on the number of deviations was completed to determine if there were significant differences. 2. For each group. the proportion of students having more than one such deviation was calculated and tested for significant differ- ences. Hypothesis 4a: LD. LA < 33. LA‘< l7. and EI groups do not differ on PIAT achievement. 1. A multivariate ANOVA comparing groups on the five subtests of the PIAT was completed to determine if there were significant differ- ences. 2. If significant differences were found to exist. repeated contrasts were conducted to examine the differences between means. Hypothesis 4b: LD. LA.< 33. LA:< l7. and EI groups do not differ on K-ABC achievement. 1. A multivariate ANOVA comparing groups on the five subtests of the K-ABC was completed to determine if there were significant differ- ences. 2. If significant differences were found to exist. repeated contrasts were conducted to examine the differences between means. Hypothesis 5a: The high-Similarities. low-Information WISC-R pat- tern will not lead to a useful decision rule for differentiating the BI group from LD. LA <33. and LA< 17 groups. 98 This hypothesis was tested in the following four ways (Veres. 1982): 1. For each group. the proportion of students having both patterns was calculated and tested for significant differences. 2. For each group. the proportion of students having only the Information pattern was calculated and tested for significant differ- ences. 3. For each group. the proportion of students having only the Similarities pattern was calculated and tested for significant differ- ences. 4. For each group. the proportion of students having neither pattern was calculated and tested for significant differences. Two levels of analysis were used: The less stringent method identified a pattern as present if not more than one other Verbal subtest was more extreme than the subtest of interest: the second method included the above criteria but added the requirement that the score of interest differed by at least three points from the mean of the remaining four Verbal subtest scores (Veres. 1982L typothesis 5b: The Number Recall. Spatial Memory K-ABC pattern will not lead to a useful decision rule for differentiating the BI group from LD. LA < 33. and.LA.< 17 groups. This hypothesis was tested in the same ways as Hypothesis 5a with one exception: only the stringent level of analysis (3-point rule) was used. The method described in the manual by Kaufman (1983) for examining strengths and weaknesses in individual subtest performance led to the decision to use only the 3-point rule. Based on the 99 performance of the children included in the norms. Kaufman found that an average of 3 scaled-score points was required for significance at the .05 level. Hypothesis 6: No linear combination of WISC-R subtest scores will provide a useful decision rule that differentiates among the four groups. Hypothesis 7: No linear combination of K-ABC subtest scores will lead to a useful diagnostic rule in discriminating among the four groups. CHAPTER V RESULTS Overview The principal focus of this investigation was to analyze the value of ability and achievement test patterns for identifying and differen- tiating children as learning disabled (LD). emotionalby impaired (EI). or low achieving (LA). 'Test scores of LD. EI. and LA groups were compared to determine if decision rules for differential diagnosis could be supported. In the first section of this chapter. the findings that relate to the utility of the particular diagnostic methods for identification of learning disabilities and emotional impairment are examined. Of chief importance is the frequency and accuracy with which LD students can be identified and differentiated from EI students without misdiagnosing or assigning LA students to either category. While a specific conjectured pattern may yield statistically reliable group differences. practical utility is limited if the proportion of students fitting the pattern is small. or if application of the rule pattern results in a high rate of misidentification or placements of children in incorrect categories. Following the analysis of particular diagnostic methods and test patterns for establishing diagnostic groups. more specific inspection of the data occurs. Categorical comparisons of LD students with each 100 101 of the other groups (EI. LA.< 33. and LA < 17) are completed to determine if test performance characteristics provide some basis for meaningful distinctions. In the next section. EI children are considered separately to determine if proposed subscale patterns on the WISC-R and K-ABC allow for efficient distinction. A fourth section of the chapter is a more general exploration of the results to determine if any information in WISC-R subtests or K-ABC subtests is useful for differentiatingyany of the diagnostic groups from the other categories. The last section of this chapter compares the frequency and proportion of overall correct classification by WISC-R versus K-ABC. This comparison is an attempt to assess which of these two instruments holds the most promise for clinical usefulness given the current federal and state guidelines for psychoeducational classification. Ability3Achievement Discrepancy Discrepant achievement remains the major identification variable that multidisciplinary teams use to determine that a child has a specific learning disability. However. as noted by Algozzine and Ysseldyke (1982). the extent or severity of a discrepancy between achievement and intellectual ability is not specified in the federal guidelines. In the present investigation. the guidelines were operationalized in four ways: 102 l. A student was considered to be LD if the discrepancy between WISC-R Full Scale IQ and PIAT Reading Recognition was greater than or equal to l §2 (15 points). The same standard was applied in comparing the discrepancy between the K-ABC Mental Processing Component score and the K-ABC Reading Decoding score. 2. If the discrepancy between ability and achievement was greater than or equal to 1.5 SE (23 points). a student was classified LD. This standard was again applied to both WISC-R and K-ABC discrepancy scores. 3. If the difference between WISC-R and achievement or K-ABC and achievement was greater than or equal to 2 S2 (30 points). the classi- fication was again considered to be LD. 4. If a LA or an EI child exhibited a.l.0 or greater discrepancy. he/she was judged to be incorrectly classified. As a preliminary question. it was of interest to ask: Are there mean discrepancy score differences among or between groups that would be useful for differential diagnosis? To examine this question. one- way analyses of variance (ANOVAS) were conducted comparing groups by test. The results of these ANOVAS are reviewed in Tables 5.1 and 5.2. Statistically significant variation between groupS‘was not found when WISC-R - PIAT discrepancy scores were used. However. some inter- esting relationships are worthy of note. The lower overall ability level of the LA group (Full Scale IQ) in comparison with their reading achievement score resulted in a negative discrepancy value. which created the initial impression that LA children achieved at higher 103 levels than LD children. In fact. they scored a point below the LD children on Reading Recognition. Table 5.1: Analysis of Variance of Mean Discrepancy Scores by Group: WISC-R FSIQ - PIAT Reading Recognition Source of Variation df Mean Square E-Ratio pp Between groups 3 317.2372 2.405 .07 Within groups 98 131.9234 Group Means LD = 6.27 EI .90 LA < 33 = -1.07 LA < 17 = 6.40 Table 5.2: Analysis of Variance of Mean Discrepancy Scores by Group: K-ABC (MPC) - K-ABC Reading Decoding Source of Variation df Mean Square E-Ratio p Between groups 3 651.0859 3.572 .0168 Within groups 98 182.2801 Group Means LD = 14.83 E1 = 5.45 LA < 33 = 5.94 LA < 17 = 14.93 A second interesting relationship existed between the LD and'LA < 17 groups. Based on discrepancy score means. both groups appeared to be equivalent with respect to degree of underachievement. However. the LD group actually outperformed the LA group by almost 12 points on both the IQ test and the Achievement subtest. This would suggest that 104 multidisciplinary teams may have focused on overall ability and achievement levels as primary identification variables. LA children may have been judged as achieving commensurate with their lower ability levels. Significant variation between groups was found to exist when K-ABC Ability - K-ABC Achievement discrepancy scores were calculated. IQ scores did not differ significantly: however. a wide range in achieve- ment performance levels resulted in significant variance to discrepancy mean scores. The EI group achieved a higher score on the reading decoding subtest in comparison to the other three groups. (Table 5.17. in a later section of this chapter. presents the probability level of these differences.) Conversely. the LA < 17 group earned the lowest score on this subtest. thus achieving at a lower level than the other three groups. The LD and LA < 33 groups scored within .31 of a point of each other on this subtest. but the lower ability score of the latter group resulted in a decreased discrepancy score. Surprisingly. the LA < 33 group also scored lower'by 1(72 points than the‘LA‘< 17 group on total ability score. While this difference is not signifi- cant. it was an unexpected finding. and to a small degree added to the differential discrepancy levels of the two groups. If magnitude of discrepancy score was the only identification method used for determining LD.it would be impossible to separate LD and LA < 17 groups. Using K-ABC scores. one would expect to find more LA < 17 children misclassified as LD than either of the remaining two groups. 105 The results of the ANOVA tables do not provide enough information to determine the practical utility of mean discrepancy score differences across groups. The usefulness of these differences for differential diagnosis is examined in the following sections. Table 5.3 contains a cross-tabulation of LD. EI. and both groups of LA students by operational discrepancy levels. As can be observed from this table. using WISC-R - PIAT discrepancy scores and the opera- tional guidelines. 10 LD and 8 of the children in the other three groups would be classified as LD. Using the K-ABC discrepancy scores and the operational guidelines. 28 LD and 13 not LD (children not identified as LD by the schools) would be classified as LD. Table 5.3: Numbers and Percentages of Children Classified Using Three Operational Discrepancy Levels Group Discrepancy Level LD EI LA < 33 LA < 17 (Q) N Z ‘N Z ‘N Z N Z WISC-R - PIAT 1.0 SD 6 12 3 15 l 6 3 20.0 1.5 §Q 2 4 1 5 O 0 2.0 SD 2 4 0 0 0 K-ABC (MPC) - 1.0 SD 14 28 2 10 2 13 2 l3 K-ABC (Ach) 1.5 g 8 16 1 5 o 3 20 2.0 SD 6 12 0 l 6 2 13 Note: Percentages are based on individual group Na and are rounded to the nearest whole number. 106 The extent to which classification decisions (based on opera- tional discrepancy-score guidelines) agreed with school placement deci- sions was determined with the use of a 2 x 2 comparison table. The groups were subsumed under two broad categories: LD and not LD. This manipulation of the data facilitated the computation of overall agree- ment between legally defined classifications and school classifica- tions. As this simplification of the data resulted in changing the composition of the current sample to 50% LD students and 50% not LD students. random guessing would lead to correct classification one-half of the time. The guidelines did not improve on chance. as can be observed from the results in Table 5.4. The largest gain in accuracy occurred using a l S_D discrepancy with the K-ABC. However. only 8% or approximately eight more students were identified. A rule that results in incorrect classifications 422 of the time does not lead to a sound basis for decision making. The number of misclassified students in the sample ranged from 43 to 52. yielding a percentage of error ranging from 422 to 51%. Failure to provide correct classification for at least 852 of the cases supported Hypothesis 1 (a and b). Comparisons of Verbal. Performance. and Full-Scale Scores Across the Four Groups A comparison of WISC-R scores was conducted to determine if groups differed. on Verbal. Performance. and Full-Scale scores. The results of the ANOVAS indicated significant differences between groups on Verbal IQ mean (2 = 5.065. §_f = 3.98. p = .003) and on Full-Scale IQ mean (2 = 4.776. g; = 3.98. 2 = .004). 107 Table 5.4: Numbers of Students Classified as LD and Not LD by the Federal Guidelines in Comparison to School Placement Decisions Federal Definition School Definition 1.0 1.5 2.0 LD Not LD LD Not LD LD Not LD LD 6 45 2 49 2 49 WISC-R — Not LD 7 44 1 50 0 51 PIAT Overall correct rate 49% 51% 52% LB 14 37 8 43 6 45 K-ABC (MPC) - Not LD 6 45 4 47 3 48 K-ABC (Ach) Overall correct rate 582 54% 53 % In Table 515 the means. standard deviations. and significance levels for differences between means are reported for each group. Comparisons are shown between the LD group and each of the other groups. As can be seen. significant differences occurred between the LD group and both of the LA groups on all three IQ scores. The LD and EI groups attained scores that were only minimally different. As the differences between LD and EI children were not substantial. it is dOubtful such differences would be of practical value. Since the main concern in the present research was correct classi- fication. it was important to determine if differences in intelligence test scores would lead to a reasonable decision rule for differential diagnosis. Table 5.6 contains classification results. No more than 108 v0. H9.v Hw.0~ Om.om vm.n~ Qh.m¢ no. 05.. Ow.un 0m.OO Gm.nn mh.Qm On. 6'. NM.VM On.w¢ vm.n~ mb.mm Own KIUWMI ~00. v.nu wm.N~ 09.Nm Oh.- mo.mm no. 56.? mm.N~ OO.QQ Oh.N~ mm.m@ 00. 0?. wa.m~ oo.mw Ob.Nn m0.m0 Ou> KIUWHI moo. n.~u mo.m Ov.mm ON.NH N0.wo NO. hd.w H~.Od Wm.m0 ON.N~ ~m.ww aw. ha. dH.Q~ mm.mm ON.N~ Na.@m Oumh fliuwnl mm cum: Mm can: mw :mo: Mm com: mm sum: .2. m j Ii .2. n .3... .|s- .2. M nlzI a fig v (A 0:0 DJ mm v (A flflfl DA. mu BEN 04 mascuu mango one no room «Eu. 3 mo mauoum 3:3: xnumux coasuom .025:qu accouuucgw new 326..— .»uuzmoaoz ecu mouuaxuM 5.... 03m... 109 36% of the students would be correctly classified if ability scores were used to discriminate among the four groups. The 85% requirement for correct classification was not met. Table 5.6: Summary of the Use of WISC-R Ability Scores to Classify Students Predicted Group Actual Group LD EI LA < 33 LA < 17 Use of WISC-R Ability Scores to Classify Students--VIQ and PIQ LD 20* 11 7 13 BI 5 7* 2 6 LA < 33 1 3 4* 8 LA < 17 4 0 S 6* 362 correct classification Use of WISC-R FSIQ to Classify Students LD 22* 7 ll 11 E1 8 2* 5 5 LA < 33 4 0 6* 6 LA < 17 3 O 5 7* 362 correct classification *Correct classification. Verbal-Performance Discrepangy As discussed in the research chapter. differences between WISC-R Verbal-Performance scores have been considered to be diagnostic indi- cators for LD children as well as EI children. While a considerable 110 body of literature exists to suggest such differences do not have diagnostic utility for separating these groups from each other or from other handicapped and not-handicapped groups. only a relatively few studies have included LA comparison groups. Wiseman (1984) included such a comparison but did not include an El group. Veres (1982) included an El group but no comparison LA group. In the present inves- tigation. all groups were compared to determine if Verbal-Performance discrepancy was larger for one group than another. The ANOVA (E = .904. £1; = 3.98. p = .44) indicated no significant differences among the four groups. A lZ-point difference was determined to be significant at the .05 level by Kaufman (1976. 1979). Table 5.7 contains the numbers and percentages of children in each of the four groups who attained a difference of 12 or more points in either direction. As can be judged from the chi-square statistic for this table (II2 = 4.031, 2; = 6. 2 = .57). there did not appear to be a disproportionate number of students in any group that could be distinguished on the basis of this method. Even when the direction of the difference was considered. no group was overrepresented. The expected Performance-greater-than-Verbal discrepancy was not supported for either LD or E1 groups as theorized. In addition. both groups had a considerable percentage of students scoring at the theoretically unexpected end (V > P). The data in this table do not offer evidence that the Verbal-Performance method is useful for differential diagnosis. Neither absolute difference nor the theorized directional difference (P > V) resulted in correct lll classification for more than 35% of the students in the LB or E1 group. Clearly. the 85% correct criterion was not met. INeither ability scores nor V-P discrepancy provided a reliable basis for differential diag- nosis. Consequently. Hypothesis 2a was not rejected. Table 5.7: Verbal-Performance Discrepancies for LD. EI. LA < 33. and LA < 17 Children When Direction of Difference Is Considered Group LD EI LA < 33 LA < 17 N Z ‘3 X H 1 ‘N Z P 3 (V + 12) 18 35.3 5 25.0 6 37.5 5 33.3 No significant difference 24 47.1 9 45.0 8 50.0 9 60.0 V i (P + 12) 9 17.6 6 30.0 2 12.5 1 6.7 Comparisons of K-ABC Sequential. Simultaneous. and Mental Processing Composite Scores Across Groups Similar to the WISC-R analysis. a comparison of K-ABC Mental Processing scores was completed to see if groups differed on Sequen- tial. Simultaneous. and Mental Processing Composite scores. The results revealed significant mean score differences on the Simultaneous scale (3 = 2.742. if = 3.98. p = .05). The means. standard deviations. and probabilities are reported for all groups in Table 5.8. The data in the table reveal some significant differences in mean scores between LD and both of the LA groups. Do these differences in Mental Process- ing scores help to discriminate between groups? Comparison to the 112 HO. o.m mo.- ~n.~m ma.un v¢.mo dc. ON.0 fim.0~ no.0m ma.uu vm.mm #0. ON. Ofi.mH ON.hm m@.~n va.m0 Eum UQuuamn< umtux :mmzuom noncommuuun accouuucuum uOu mum>wd >uu~unmn0ua mm=OuU muzuo can modumznh um.m Oman? 113 criterion (school-identified classifications) was used to determine the classification accuracy (rate of agreement with school decisions). Table 5.9 may be consulted to determine the practical usefulness of these group-level differences for differential diagnosis. Table 5.9: Summary of the Use of K-ABC Ability Scores to Classify Students Predicted Group Actual Group LD EI LA < 33 LA < 17 Use of K-ABC Ability Scores to Classify Students-- Mental Processing Composite LD 29* 2 l8 2 El 10 1* 7 2 LA < 33 3 3 8* 2 LA < 17 5 0 7 3* 402 correct classification Use of K-ABC Ability Scores to Classify Students- Sequential and Simultaneous Scales LD 17* 17 13 4 El 6 6* 3 5 LA < 33 5 2 6* 3 LA < 17 2 4 5 4* 322 correct classification *Correct classification. 114 If K-ABC ability scores were the only identification variable employed. successful differentiation of LD students would occur no more than 402 of the time. The percentage of correct identification fell short of the 85% correct classification requirement of the present investigation. Sequential-Simultaneous Discrepancy Similar to the WISC-R research. a difference between K-ABC Sequential and Simultaneous scales has been postulated for both LD and EI groups. Analysis of variance results revealed that there was no significant variance in mean discrepancy levels across groups (E = .677. g; = 3.98. p = .5683). A cross-tabulation of groups by significant Sequential- Simultaneous differences of 12 points or more highlights the results of the ANOVA. The chi-square statistic for Table 5.10 (X2 = 7.775. £1; = 6. p = .26) provides evidence that no group had a disproportionate number of children with a significant Sequential-Simultaneous discrep- ancy. When the direction of difference was considered. the theorized Simultaneous-greater-than-Sequential pattern was not supported for either LB or EI groups. Of interest is the fact that 10% of the LD and 252 of the BI were also represented at the theoretically unexpected end (Sequential > Simultaneous). 115 Table 5.10: Sequential-Simultaneous Discrepancies for LD. EI. LA< 33. and LA<17 Children (Directional—Simultaneous> Sequential) Group LD EI LA< 33 LA< 17 N Z _N Z _N Z _N 1 Sim: (Seq + 12) 14 27.5 3 15.0 1 6.3 4 26.7 No significant difference 32 62.7 12 60.0 14 87.5 9 60.0 Seq : (Sim + 12) 5 9.8 5 25.0 1 6.3 2 13.3 As can be judged from the percentages in Table 5.10. statistically significant differences (12 or more points) do not provide practical decision rules for differential diagnosis. However. this result was not surprising as Kaufman and Kaufman (1983) discovered that the average child had Sequential-Simultaneous differences of this magnitude. A discrepancy of this size was found to significantly discriminate the groups at the .05 level. but was not large enough to be of any practical consequence (p. 193). Using a lZ-point difference as a reference point for typical performance of normal children. Kaufman and Kaufman suggested that l _S_D above that mean represented a reasonable criterion for determining abnormality. Consequently. a difference of at least 22 points was considered to denote unusual and marked scatter. Table 5.11 presents the numbers and percentages of students achieving at least a 22-point difference in the predicted direction. 116 These results mirror those presented.in previous tables. as*well as those found using WISC-R Verbal-Performance differences. Table 5J1: Simultaneous-Sequential Discrepancies for LD. EI. LA < 33. and LAn< 17 Children (Directional-Simultaneous > Sequential) Group LD EI LA< 33 LA< 17 N Z N Z N Z N 2 Sim: (Seq + 22) 5 9.8 2 10.0 0 0 No unusual difference (< 22 points) 45 88.2 16 80.0 16 100.0 13 86.7 Seq Z_(Sim + 22) l 2.0 2 10.0 0 2 13.3 x2 = 8.3860 g; = 6 2 = .2112 Neither ability scores nor Sequential-Simultaneous scores appear helpful for differential analysis. As the overall correct classifica- tion rate did not meet the designated 85% criterion of this study. Hypothesis 2b was retained. Ability Test Subtest Scatter quparisons Across Groups: WISC-R Another method or indicator discussed in the literature as useful for differentiating LD from other groups concerns the amount of scatter in subtest scores. The belief has been that LD children exhibit more scatter in subtest scores in comparison to other exceptional groups. 117 Kaufman's NDEV measure was calculated and used to compare groups. This measure of deviation is defined as variance of 3 or more points from a child's own scaled-score mean. The average number of deviations in the WISC—R standardization sample was found to be 1.7 1 .02 subtest scores by Kaufman (1976). Table 5.12 contains the group means for the number of deviations from an individual's own mean (NDEV). The table is divided with Verbal scale deviations in the first column and Performance deviations in the second column. Kaufman advocated separate calculations of NDEV-Verbal based on the Verbal mean and NDEV-Performance based on the Performance mean because of the factor structure of the WISC-R. In school- psychology practice. a child's strengths and weaknesses are determined in this manner. Table 5.12: Group Means for NDEV Verbal and Performance NDEV Group Verbal Performance LD 1.22 .82 E1 .50 1.55 LA < 33 .88 .63 LA < 17 .20 1.20 Analysis of variance on these data revealed significant group- related differences on NDEV-Verbal (§_[3.98] = 4.730. p = .004) and NDEV-Performance (E [3.98] = 4-543. p = .004)- 118 Inspection of the table reveals that the LD group exhibited the most scatter on the Verbal subtests and. as a group. varied the most from the LA < 17 group. Conversely. the E1 group had the most scatter on the Performance subtests and was most different from the LA < 33 group. Are these group-level differences in scatter useful in formulating decision rules for the classification of LD children? To determine the answer to this question. school-identified classifications were compared to classifications based on the NDEV method. By this method a student was correctly identified as LD if he/she had two or more deviations of 3 or more points. Students not labeled LD by the schools were correctly classified if they had no more than one deviation of 3 or more points. The results of this comparison are presented in Table 5.13. As can be determined from the table. the use of the NDEV measure of scatter was successful 64% of the time when Verbal scale deviations were used and 45% of the time when Performance scatter was used. The use of this scatter index for classification of LD students would lead to misclassification from 36% to 552 of the time. Based on the high percentage of error. Hypothesis 3a was supported. 119 Table 5.13: Numbers of Students Classified as LD and Not LD by the NDEV Method in Comparison to School Placement Decisions Using the WISC-R NDEV Definition School Classification LD Not LD LD 17 34 Verbal Not LD 3 48 Overall correct rate 64% LD 12 39 Performance Not LD 17 34 Overall correct rate 452 Ability-Test Subtest Scatter Comparisons: K-ABC In the K-ABC Interpretive Manual several pages are devoted to the topic of scatter. Thermajor portion of the discussion is concerned with global scale scatter. However. reference is made to earlier research regarding WISC-R subtest scale variability and findings that normal populations exhibit considerable amounts of scatter. Parallels are drawn between the WISC-R and the K-ABC. Analyses were conducted with the K-ABC subtest scores similar to those conducted in the previous section of this chapter using the WISC-R. Kaufman's NDEV measure was calculated and used to compare groups. ‘Like the WISC-R. a 3-point difference between subtest scores is required for significance at the .05 level. A major difference in procedure is that deviations are calculated by comparing individual scale scores to the mean of all of the subtests of the Mental Process- ing scale rather than to separate comparisons of the Sequential mean 120 and the Simultaneous mean. One reason for this as described by Kaufman (1983) relates to the fact that the Sequential scale includes only three subtests. Means based on such a small number are considered less stable than those based on five or more subtests. Table 5.14 contains the group means for the number of deviations from an individual's own mean. Analysis of variance on these data revealed nonsignificant group-related differences in terms of the mean number of deviations (2(3.98] = .495. p = .687). As can be seen in the table. the differences in values across groups would have little practical value in clinical situations for differentiating groups. It would be anticipated that if the diagnosis of LD rested solely on NDEV scatter measures. considerable misclassification would occur. Table 5.14: Group Means for NDEV--Mental Processing Group NDEV LD 1.6 EI 1.9 LA < 33 1.3 LA < 17 1.5 Table 5.15 provides a summary table of the use of NDEV scatter to classify LD and not LD groups. LD students were considered correctly classified with two or more NDEVs. while all other (not LD) groups were considered correctly classified if they had no more than one NDEV. The results indicated that correct classification would occur no more 121 than 50% of the time if the NDEV scatter index was the only identification variable for diagnosing LD. Hypothesis 3b was supported. Table 5.15: Numbers of Students Classified as LD and Not LD by the NDEV Method in Comparison to School Placement Decisions Using the K-ABC NDEV Definition School Classification LD Not LD LD 22 29 Not LD 22 29 Overall correct rate 502 Achievement Test Comparisons: Peabody Individual Achievement Test In an earlier chapter. the LD classification was described as a subset of low achievement. Several researchers have advocated the inclusion of a low-achieving comparison group in research designed to evaluate the discriminative efficiency of methods for identification of LD students. Of particular interest in the present study was the concern of'whether or not.achievement test performance differences between LD and other handicapped and not-handicapped populations would lead to a useful decision rule for differential diagnosis. The means. standard deviations. g-ratios. and probability levels comparing group means are presented in Table 5.16. Inspection of the table suggests that the use of PIAT achievement scores might lead to a 122 "so. am.m~ om.m he.mh m~.o~ no.ao Au. o~m.~ «o.o n~.eo m~.o~ ”c.6c no. ono.v am.~a om.mo m~.c~ ~0.ao asses 0° C sec. on.- em.a no.6» Nm.- mo.am do. «so.» an.v~ n—.oo ~m.- ma.sa as. can. vm.- mn.mm ~m.a~ ma.sa sauouww ace. oq.o~ os.m m~.ss om.s ec.om co. ass. oa.o ea.sm co.s qc.oo so. non.» o~.o~ an.sa on.» vo.ao ocazaoaw .950 no. ”No. sv.m as.om so.s~ mm.am Na. and. o~.m~ «6.4m mo.s~ mo.~o as. omo.~ oc.o~ se.~m so.a~ oa.~m schemes IUU moo. sm.- v~.m oo.os oo.m mo.oo 66. mos. cm.~ mm.oo oa.m mo.co ca. ma~.~ om.- mo.vm so.» mo.co ocacmuu so. mmo.s mm.» om.vm am.ca oc.no 6". 653.6 -.o~ om.oo oo.o~ oo.ns so. nes.n om.- om.mm sm.oa oo.na can: mw cmmx Mm cow: .mm moo: . mw cam: om snot mw smut . on I . o I . . On I an m huvj Q.— num .— nnvs s an m an 3 s3 6 es 6:. as an v as on. as as can on mascuo map—mo 9: ac comm can 3 mo mmuouw acme—0:01."; Btu.— comtfimm mmocmmwuuqo new m~w>3 .auzunmnohm was manuS—IM ”wufi 22mm. 123 reasonable decision rule for the differentiation of LD and LA < 17 groups. These two groups differed significantly on all but one of the achievement tests. revealing overall lower achievement for the LA < 17 group. While there were two significant differences in PIAT mean scores between the LD and the El groups and one between the LD and the LA < 33 groups. it is unlikely that these differences would have diag- nostic utility in clinical practice. The usefulness of PIAT achievement scores for differential diagnosis may be evaluated by referring to Table 5.17. where a breakdown of correct classifications is presented. The results shown in the table provide support for Hypothesis 4a. Although PIAT achievement scores resulted in considerable differentiation between LD and LA < 17 groups. the 222 error rate exceeded the acceptable level of error (15%) designated in the present study. The use of achievement scores resulted in the least amount of error (disagreement with school classifications) in comparison to previously discussed methods. These results provide suggestive evidence that school placement teams focus on achievement functioning as a primary decision factor. Achievement Test Comparisons: K-ABC Achievement Scale The means. standard deviations. g-ratios. and probability levels comparing group means of all four groups are presented in Table 5.18. As can be observed in the table. the K-ABC achievement scores suggest that EI children scored significantly higher than LD children on all but the Riddles subtest. The LD group and the LA < 33 group did not appear very different on achievement test performance on the K-ABC and 124 Table 5417: Classification of Students Using PIAT Achievement Scores Actual Group Predicted Group LD EI LA < 33 LA < 17 L0 11* 14 ll 10 El 1 11* 3 2 LA.< 33 4 4 6* 0 LA < 17 l 0 0 10* 432 correct classification LD EI LD 33* 13 BI 6 11* 69% correct classification LD LAK33 LD 29* 17 LA.< 33 7 7 60% correct classification LD LA<17 LD 36* 10 LA < 17 2 9* 782 correct classification *Correct classification. 1125 do. 00.h m0.0 d~.mh hm.N~ OQ.FQ hm. QdQ. On.m 0m.'0 hm.N~ aqohO NO. NN.¢ Od.v~ mO.®G hm.Nu afi.fih HmuOF . . . . . . .vsmumuocca ma. FOO. Nh.- nw.mh GF.NN wN.Gh 00. 60°. db NN OM Oh Oh NN 0N Oh do mn.n mh.bN 0m 0% $F.NN 0N.Qh nonwflmoz . O O O I . . C . . . O . . . O C . 2‘808 m0 NG M NH m cm mh n6 Nu hm NO n0 GOO “5 Q 00 N0 n0 NH hm NO AD ma 0 dfl CH 00 Aw Mm Nd hm NO IvCuwaz H00. 0.0d vn.m Oc.ho wm.d~ Ga.h¢ Nu. mQ.N Od.°u CO.NO mm.- 0O.h¢ ha. Oc.n 'Q.Au ON.ND~ Om.~d OG.BQ uvnflfimz ma. no.5 hm.u~ O0.0Q HQ.H~ 0v.am RN. MN.~ 06.0" Vu.m0 dv.dd aQ.$O CO. mv.v O¢.hn m0.¢$ Av.~d OQ.OG Odavfiludu¢ mucous NH. Nm.N Oo.v~ nn.nm vn.Nu hn.OQ MO. N~.m AM.N~ On.~0 ‘M.Nn hn.00 doc. mcoh Gn.0 om.hm QM.N~ fin.00 a muoam Mm coo: Mm cow: .mm moot Mm new: Mm sow: mum. smut . u I . u I . I £0 A h ha v (A DJ £0 A h HM v (A DJ DONE h H” DA A“ v as can as an v :4 one as us was on :Ohvudzu ea v es was .mn v an .au .oa mo sosoom accessoanug omens coozuon ssucssouuuo sou canton uuaasnsAOCs out aoamnsum .o~.m ugnmfi 126 probably could not be distinguished from each other if achievement scores were the only identification variable employed. The LA < 17 group achieved at significantly lower levels on all measured achieve- ment areas except two when compared with the LD group. Discriminant function analyses were completed to determine the diagnostic utility of K-ABC achievement test scores. Table 5.19 pro- vides a summary of the usefulness of these scores. Hypothesis 4b was supported. The use of K-ABC achievement scores was correct for no more than 68% of the cases. Even if the error rate designated in this study was doubled. the number of correct classifications would still fall short of the criterion. High Similarities. Low Information: WISC-R Pattern-- Emotional Impairment In the literature review chapter. it was noted that test patterns proposed as suggestive of emotional impairment have generally been more vague than those proposed for learning disabilities. However. Dean (1976) was specific in his suggestion that two subtests on the Verbal scale are particularly useful for the diagnosis of emotional impair- ment. He indicated that a number of investigations were consistent in findings that the Information subtest is usually one of the two lowest Verbal subtests and Similarities is one of the two highest for El children. Veres (1982) investigated this pattern and concluded that it had little diagnostic utility for distinguishing between EI and normal groups. However. she included no comparison LA group. This pattern Table 5.19: 127 Classification of Students Using K-ABC Achievement Scores Actual Group Predicted Group LD EI LA < 33 LA < 17 LD 17* ll 8 11 El 3 11* 2 2 LA < 33 3 1 6* 6 LA < 17 3 0 1 8* 45% correct classification LD 81 LD 33* 14 RI 7 11* 682 correct classification LD LA < 33 LD 33* 14 LA < 33 6 10* , 681 correct classification LD LA < 17 LD 31* 16 LA < 17 4 3* 662 correct classification *Correct classification. 128 was tested again in the present investigation to determine if it has any diagnostic relevance for separating EI. LD. and the two LA groups. If the pattern fails to distinguish among these groups. then Veres's original assertion. that the pattern does not provide a meaningful distinction between groups. will have received more support. The analyses of this pattern replicated the methods used by Veres. In the first test of the pattern. the Information pattern was present if not more than one other Verbal subtest was lower than Infor- mation. and the Similarities pattern was indicated if not more than one other Verbal subtest was higher than Similarities. (This analysis conforms exactly to Dean's criteria.) Four outcomes were possible for an individual child. Only the Information pattern or the Similarities pattern might occur. both patterns might be present. or neither pattern might be present. Table 5.20 presents the results of this first analy- sis of the pattern. Table 5.20: High Similarities-Low Information Pattern: Comparison of LD. EI. LA.< 33. and LA < 17 Groups (Dean's Criterion) Group LD EI LA < 33 LA < 17 N Z N Z N 2 N Z Information pattern 16 31.4 7 35.0 7 43.6 8 53.3 Similarities pattern 11 21.6 5 25.0 3 18.8 3 20.0 Both patterns 17 33.3 3 15.0 3 18.8 0 Neither patterns 7 13.7 5 25.0 3 18.8 4 26.7 129 Inspection of Table 5.20 leads to the conclusion that this first analysis of the two patterns did not result in any practical dis- tinction among the groups. An interesting finding was that the LD group. rather than the El group. had the highest proportion of students exhibiting both patterns. A more stringent test of these patterns was proposed by Veres after she obtained results similar to the above. She proposed that a more strict criterion would result in greater attenuation of the pat- terns for normal children rather than EI children. The more stringent analysis was conducted in the present investigation to determine if the proposed pattern offered evidence of meaningful distinction. In this analysis. the Information pattern was present if not more than one Verbal subtest score was lower than Information. and Informa- tion was at least 3 points lower than the mean of the remaining Verbal subtests. Similarly. the Similarities pattern was present if not more than one Verbal subtest score was higher than Similarities. and Simi- larities was at least 3 points higher than the mean of the remaining Verbal subtests. As in the previous analysis. each child could have one of four outcomes. Table 5.21 presents the results of this analy- sis. As in the previous table. the LD group had a higher proportion of patterns than the BI group. In no case did the BI group have both patterns present when the more stringent criteria were applied. Based on the data in Table 5.21. the inescapable conclusion is that Dean's pattern did not differentiate among groups. The EI group was not 130 overrepresented on this pattern. In fact. the BI group ranked third in terms of the proportion of students exhibiting either pattern. A decision rule based on this pattern of signs in either analysis would provide correct classification for no more than 552 of the students. Table 5.21: High Similarities-Low Information Pattern: Comparison of LD. EI. LA.< 33. and LA < 17 Groups (3-Point Difference-- Stringent Criteria) Group LD EI LA.< 33 LA < 17 N Z ‘N 1 IN I ‘N Z Information pattern 6 11.8 4 20.0 5 31.3 2 13.3 Similarities pattern 8 15.7 1 5.0 0 l 6.7 Both patterns 7 13.7 0 2 12.5 0 Neither pattern 30 58.8 15 75.0 9 56.0 12 80.0 Table 5.22 provides the number of students and the percentage of correct decisions when the Low Information-High Similarities pattern is used to classify EI students. IBased on these values. it is obvious that the number of errors exceeded the 15% error level set in this study. Consequently. Hypothesis 5a was supported. 131 Table 5.22: Numbers of Students Classified as EI and Not EI by Dean's Criteria in Comparison to School Placement Decisions Dean's Pattern School Classification EI Not 81 Dean's (less EI 15 5 stringent Not EI 68 14 criteria) Overall correct rate 28% 3-point EI 5 15 criteria Not EI 31 51 Overall correct rate 55% K-ABC Patterns: Emotional Impairment Only one research study about the test performance of EI children was reported in the Interpretive Manual. ‘While the K-ABC test authors recognized the need for more research with El populations. they outlined preliminary findings regarding achievement-score differences considered useful for distinguishing between.EI and LD groups. No specific hypotheses were formulated about these achievement-score differences. However. the following paragraphs suggest some support in the present sample for these score patterns. Reportedly. the standardization sample ofIJJchildren scored 5 points lower on the achievement scale than on the Mental Processing composite. while the behaviorally disordered group scored 2 points higher on the achievement scale. On individual achievement subtests. the behaviorally disordered group outperformed the LD group by 3 points on the Arithmetic subtest and by 7 or 8 points on the other four 132 subtests. Both groups attained their highest scores on the Riddles subtest. The analysis of achievement test performance discussed in an earlier section supported these predicted achievement test differences between EI and LD groups. Not supported was the suggestion by Kaufman and Kaufman (1983) that the El group attained an Achievement Composite score that exceeded their MPC score by 2 or more points. In the present sample the EI group attained MPC scores which on the average exceeded their achievement composite by 1.25 points. However. the difference between observed scores and Kaufman's predicted relationship could be due to sampling error. Consequently. no firm conclusion could be derived from current results. While the trend observed in achievement test performance for LD and EI groups conformed to that predicted by Kaufman and Kaufman. it will be recalled that only 68% of the LD and EI children would have been correctly classified if K-ABC achievement scores were used as the primary method for classification. Practical usefulness of these differences for differential diagnosis‘was not found.in the present results. Regarding Mental Processing score patterns. Kaufman and Kaufman reported that Number Recall was the third highest subtest for behaviorally disordered children. while Spatial Memory was a."striking weakness" (not defined). If Kaufman and Kaufman's observations are correct. it would be expected that a large percentage of behaviorally disordered (EI) 133 children would share this pattern. To be diagnostically useful it must be demonstrated that this particular pattern of scores occurs more frequently among EI groups than it does in other comparison groups. Kaufman and Kaufman did not provide guidelines for evaluating this pattern. Based on a reading of their chapter. it is not known how they defined "striking weakness." In addition. there was no discussion of whether they used the customary 3-point difference between subtest scores to determine significant differences. To provide an evaluation of this pattern. a 3-point difference between scores was used. In school psychology practice. this conventional 3-point rule would be applied to determine significant differences between subtest scores. The analysis performed to test this pattern was similar to the procedure for evaluating Dean's patterns for emotional impairment on the WISC-R. A pattern was present in a child's profile if Number Recall was the third highest subtest and if it was at least 3 points different from the mean of the remaining Mental Processing scores. The Spatial Memory pattern was present if it was at least 3 points lower than the mean of the remaining Mental Processing scores. With this analysis. one of four outcomes was possible: both patterns might be present. only the Number Recall pattern might be present. only the Spatial Memory pattern might be present. or neither pattern might be present. Table 5.23 contains the results of this analysis. 134 Table 5.23: Patterns for Emotionally Impaired: K-ABC High Number Recall. Low Spatial Menory Pattern Group LD EI LA < 33 LA < 17 N Z N Z ‘N Z N 2 Both patterns 1 2 2 10 1 6 1 7 Number Recall pattern 18 35 9 45 6 38 9 60 Spatial Memory pattern 3 6 1 5 0 0 Neither pattern 29 57 8 40 9 56 5 33 It is evident from Table 5.23 that the suggested EI patterns did not provide meaningful discrimination among the four groups. The analysis did not support Kaufman and Kaufman's pattern for El children. Correct classification of El children using any of the signs occurred no more than 452 of the time. Thus. Hypothesis 5b was supported. WISC-R Subtests and Classification of Students Discouraging results were obtained when proposed WISC-R patterns were evaluated for diagnostic utility. None of the methods or test patterns achieved 85% accuracy in discriminating school-identified LD. EI. or LA children. Therefore. it was of interest to determine whether WISC-R subtest scores were useful for identifying any of these groups of children. A discriminant function analysis was conducted to determine if any combination of WISC-R subtest scores provided a useful decision rule that differentiated among the four groups. 135 Table 5.24 contains the classification results based on discrimi- nant function analysis. None of the functions was significant. The 482 correct classification failed to meet the 852 correct criterion of the study. Considering the 52% error rate. it appears that the subtest profiles of the groups were relatively indistinguishable from one another. A practical decision rule for differential diagnosis was not produced. Hypothesis 6 was supported.— Table 5.24: Classification of LD. EI. LA < 33. and LA < 17 Children Using WISC-R Subtest Scores Predicted Group Actual Group LD EI LA < 33 LA < 17 LD 16* ll 7 4 El 4 7* 2 3 LA < 33 2 3 9* 0 LA < 17 l 1 2 5* 482 correct classification *Correct classification. K-ABC Subtests and Classification of Students A discriminant function analysis was conducted using the K-ABC subtests to determine if any combination of subtest scores would be useful for discriminating between groups of children. (See Table 5.25J The use of subtest scores of the children from each group resulted in one significant function (I!2 = 69.20. £l_f = 39. p = .002). However. the function succeeded in only 60% correct classification. 136 Since this fell short of the 85% correct requirement. no practical decision rule was produced for discriminating among the four groups in the study. Hypothesis 7 was retained. Table 5.25: Classification of LD. EI. LA < 33. and LA < 17 Children Using K-ABC Subtests Predicted Group Actual Group LD EI LA < 33 LA < 17 LD 22* 7 8 10 E1 3 14* 0 1 LA < 33 2 2 11* 1 LA < 17 1 O 2 9* 601 correct classification *Correct classification. Classification Accuracy: Comparison of the K-ABC and the WISC-R In the current investigation the practical utility of proposed test score patterns for differential diagnosis was assessed in terms of the number and percentage of students correctly identified by these methods. The proportion of correctly identified students served as an index of the discriminative efficiency of the procedures. It is of interest to compare the WISC-R and the K-ABC to determine if one or the other of the tests is more accurate in classifying children when current federal regulations and state guidelines are used. Determining which of the two instruments yields the highest 137 percentage of correct classifications has practical relevance for the psychologist wishing to make an informed choice about which of these two intelligence scales to use. Table 5.26 provides a summary of correct decisions by test and by method. Correct decisions are either LD or E1 children. depending on the method. It will be recalled that two patterns/methods were spe- cific to the identification of E1 children. and all of the others were theorized to be characteristic of LD students. The table is not help- ful for deciding which test is more accurate. The K-ABC seemed to have an edge in four of the comparisons. while the WISC-R provided more accuracy on three of the other methods. Both tests correctly identi- fied an equal number of students in another comparison. The E-test for the significance of the difference between the mean totals for number of correct decisions was not significant (p = -0.25. if = 7. ns). indicating that both tests provided approximately the same level of accuracy given the current guidelines. As neither test pro- vided more than 651 correct decisions. both could be considered equally ineffective in discriminating among diagnostic groups. In short. there seems to be no quantifiable or objective basis for deciding which of the two instruments to use in assessment situations. The selection of one or the other of these instruments will most likely be governed by individual preference. or in the worst case. by political concerns or pressures. For instance. if the LD label is considered less stigma- tizing than other labels. the K-ABC may be the instrument of choice. 138 Table 5.26: Summary of the Methods Used to Classify Students: Comparisons of Correct Decisions by K-ABC Versus WISC-R WISC-R K-ABC Method Correct Decisions Correct Decisions N Z N Z Ability-Achievement discrep. 10 20 28 55 Total Ability score 28 55 30 59 Seq IQ & Sim IQ/VIQ & PIQ 31 61 31 61 Sim-Seq discrepancy/ V-P discrepancy 23 45 7 l4 NDEV 29 57 22 43 Achievement 33 65 32 63 EI patterns (3-point rule) 5 25 10 50 Subtests 16 31 22 43 Overlapping Effect: Classification of LD Children Using:the K-ABC and the WISC-R Together The correlation coefficient between the K-ABCnand the WISC-R total composite scores (Full Scale) was .70. This indicates a shared variance of 492 and reflects a substantial amount of redundancy between the two instruments. This redundancy in classification decisions was not accounted for in the previous analyses. As a result. it was not possible to distinguish the subset of students who were classified by both instruments from those who were only identified by the WISC-R or by the K-ABC. So as not to preclude the identification of the most 139 inclusive discriminating variables. a supplementary analysis using the two tests together was conducted. It was of interest to ask: If use of the K-ABC subtests resulted in a 472 correct classification rate of just LD children. and the WISC-R subtests resulted in 422 correct classifications of just LD children. how many LD children would be identified if the two tests were used together? An answer to this question facilitates the deter- mination of whether either one of the-tests used alone is as useful as when the two are used together. In clinical practice. the majority of psychologists would not use the two tests together because of the redundancy and because of practical time limitations. However. for research purposes and potential future clinical application. there is practical value in determining if a shorter. more efficient battery could be developed from a subset of the items in both instruments. Three analyses were used in this comparison. First. a backward stepwise discriminant selection procedure including all subtests from both instruments was used to identify the most discriminating vari- ables. In this method. variables passing the tolerance test are entered as a group and are then removed from the equation one at a time. Variables with the smallest g values are removed until a final set of the most discriminating variables is all that remains. The second part of the analysis used only the group of six variables identified in the first analysis. These variables were entered into the discriminant equation by a default procedure. 140 However. after the first two of these variables were entered (WISC-R Similarities and K-ABC Triangles). the others could not be entered. suggesting that these two subtests were the most clinically useful and accounted for the greatest discriminating power as well as most of the variance in the remaining four subtests. The third analysis used a discriminant procedure that forced all six of the variables into the equation to determine if there were any differences in number and percentage of correct classifications. The results of the analyses produced some very interesting findings. Out of the 19 subtests of the K-ABC and the WISC-R. six subtests were identified as having the most discriminating power for differentiating LD and LA students. These six subtests included five from the WISC-R and one from the K-ABC. The subtests identified by this procedure included: WISC-R—Information. Similarities. Arithme- tic. Block Design. Object Assembly-and K-ABC Triangles. The increased number of WISC-R subtests included in the equation provides suggestive evidence that the WISC-R may be the more efficient of the two instru- ments for discriminating between LD and LA populations. More significant evidence for this possibility was suggested by results from the second analysis. The two subtests having the most discriminating power included the WISC-R Similarities subtest and the K-ABC Triangles subtest. Similarities is a measure of verbal concept formation. while Triangles is primarily a measure of nonverbal concept formation and visual-motor coordination. The factor structure and the dichotomy of WISC-R abilities is well represented by these two 141 subtests. While a number of studies would be needed to provide credible support. these preliminary results suggest that the WISC-R may be the more clinically useful of the two instruments. Regarding percentage of correct classifications. when all six subtests were used in the analysis 66.72 of the LD children were correctly classified. and an overall correct rate of 70.9% was obtained. When just the WISC-R Similarities subtest and K-ABC Triangles subtest were used. 72.52 of all LD children were correctly identified and the overall correct rate for all cases was 70.71. Fewer subtests using a combination of the two instruments was more successful than when either of the complete tests was used singly. While diagnostic utility as designated in this study was not achieved. the results promote optimism about the usefulness of the WISC-R as well as the possibility of combining the best information from both tests to develop a shortened. more efficient battery. General Findigg: Black Students No specific hypotheses were included regarding black children because only 10 were included in the study. Of these 10 children. only 20% were from low-SE8 backgrounds. Seventy percent of the sample were male. The average age of the sample was 7 years of age. and 602 of the black sample lived in suburban areas. It was of interest to determine if test performance trends and classifications of black children were consistent with the literature. Previous research has revealed that. contrary to popular belief. black children are not more likely than white children to be placed in 142 special education. In fact. if there is any bias. it is in the reverse direction. That is. the primary differences found were that lower- class black children were less likely to be recommended for special education than either lower-class whites or upper-class black children (Reynolds. 1982). The results of the present investigation suggest a trend in classification decisions that would support the literature. Eighty percent of the black children were from middle- to upper-income families. Forty percent of the children were evaluated but not placed (LA). 50% were classified LD. and only 10% were identified as EI. Because of the sample size. no conclusive statements can be made. but this trend in classification decisions appears to support the litera- ture. CHAPTER VI DISCUSSION Overview In this chapter. conclusions regarding the test results and their relationship to methods proposed in the literature for differential diagnosis are examined more closely in terms of their practical utility for clinicians and researchers. The implications of the present results for developing useful decision-making rules for classification and placement of children are discussed. Failure of test scores to substantiate school-identified classifications are discussed in terms of possible factors influencing EPPC decisions. Recommendations for clinical practice. public policy. and research are offered in the final section of the chapter. Summary and Utility Conclusions A higher proportion of LD children in comparison to the other groups of children exhibited ability-achievement discrepancy scores on both.the‘WISC-R.and the K-ABC within the operational discrepancy ranges designated in the study. However. the use of these definitional dis- crepancies (1.0. 1.5. and 2.0 _SQ) resulted in considerable misclassifi- cation. Comparison of the schools' classification of students to applications of the federal definitions of LD resulted in an error rate 143 144 ranging from 42% to 51%. Using ability-achievement discrepancies. as many as 52 of the 102 students were misclassified. These results» do not improve on chance. Consistent with the results of Ysseldyke. Algozzine. Shinn. and McGue (1982). it appears that. for every student meeting federal guidelines who is receiving special education services. there is a similar student "out there" who exhibits the same test characteristics but who is not getting appropriate service (p. 81). and vice versa. The composite ability score results. V-P discrepancies of the WISC-R. and Simultaneous-Sequential discrepancies of the K-ABC were no more encouraging than the ability-achievement results. No useful diag- nostic rules were supported. Full-Scale scores on the WISC-R resulted in 64% misclassification when all four groups of children were com- pared. Comparison of Verbal and Performance 103 resulted in the same percentage of errors. K-ABC ability scores were no more promising. MPC scores resulted in 60% misclassification rates when all four groups of children were compared. When Sequential and Simultaneous composite scores were used. the errors increased to 682. V-P discrepancies placed children in the correct groups no more than 351 of the time. Simultaneous-Sequential discrepancies on the K-ABC were successful in classifying children no more than 282 of the time when a lZ-point difference was used and 101 of the time when a 22- point difference was used. Theoretical expectations of Performance- greater-than-Verbal and Simultaneous-greaterthan-Sequential were not 145 supported for either LD children or E1 children. That is. in the current results. these children were about equally represented in either direction. The Ability-scatter results revealed that on the WISC-R. LD children had the most verbal scatter. and the El group exhibited significantby more scatter on the Performance subtests. However. no decision rules of practical value were formulated because misclassifi- cation errors ranged from 362 to 55%. As equally discouraging. the use of K-ABC scatter as a diagnostic aid is not warranted. No more than 502 of the students were correctly classified. No useful LD decision rule was supported. Comparison of PIAT Achievement scores revealed statistically significant variation between the LD and LA < 17 groups. While a high rate of correct classification (78%)‘was achieved for these two groups. the 222 error rate exceeded.the acceptable error level designated in this study. An overall correct classification rate of 43% for all four groups did not lead to a useful LD decision rule. The K-ABC Achievement score results were no more successful than those of the PIAT. In no case was a diagnostic rule based on achievement correct for more than 68% of the cases. An overall correct rate of 452 for all four groups did not lead to a useful decision rule. Dean's WISC-R pattern for identifying EI children was correct in no more than 55% of the cases. Kaufmanls pattern for identification of EI children using the K-ABC was successful for 451 of the cases. No useful decision rules 146 for emotional impairment were derived from analysis of either the WISC-R or the K-ABC. No linear combination of WISC-R subtests resulted in a useful LD decision rule. When all four groups were compared. no more than 48% of all cases were correctly classified using subtest information. No useful decision rules concerning any of the groups could be generated. In the discriminant analysis using all K-ABC subtests and all four groups. only 601 of the children were classified correctly. No linear combination of subtest scores led to a useful diagnostic rule. What general conclusion results from the analyses in this investi- gation? The ineluctable conclusion is that none of the diagnostic methods examined achieved the level of utility designated as useful in this study when evaluated in terms of agreement with school-identified classification. The overall results of the present study suggest that. to the degree that the sample of children in this study represented the popu- lation of children referred for psychoeducational evaluation. the use of the various suggested diagnostic methods (with either the ‘WISC-R or the K-ABC) for differential classification may not be trustworthy. No conclusive evidence exists regarding the usefulness of the methods because of the low validity of the criterion (school placement deci- sions). The implications (for psychologists and educational administra- tors) are. quite simply. that the results of this investigation heighten existing concerns regarding the reliability of EPPC decision 147 systems as well as the use of standardized test scores for differential classification purposes. The diagnostic methods examined in the pres- ent study seem to have little practical usefulness for'differential diagnosis and placement of students. Consistent with findings of earlier investigations. the current results do not support the classi- fication of students by these methods. Also consistent with past research. it appears that placement-committee decisions are based on factors that are not readily identifiable. However. it seems reason- ably certain that state and federal guidelines are not primary decision factors used by EPPCs. Dispggeement Between School Identifications and Federal Identifications: Classification Labels A major assumption of the current investigation was that the classification decisions of EPPCs were accurate. However. as revealed by the results of this study. considerable misclassification occurred. Some students who met federal guidelines for LD were not classified by the schools as LD and the reverse. Based on a review of the data used in this study. few clues emerged as to the decision systems employed by schools in classifying students. It did not appear that placement committees adhered to federal guidelines. particularly with LD students. There was a lack of correspondence between discrepancy levels and whether a student was labeled LD or not LD. No psychometric differences of practical diag- nostic value were observed among the four groups; and there was failure to provide evidence of diagnostic utility when the theoretical methods 148 of diagnosis examined in this study were assessed for differential classification accuracy. Confusion in differentiating among the groups in this study raises some important concerns. First. the failure to identify consistent EPPC labeling-decision systems suggests the possibility that subjective processes may have replaced objective decisions-making steps. Such a process could explain lack of correspondence between discrepancy levels where some children with negative discrepancy scores or with nonsignificant discrepancy scores were labeled LD. while others with very significant ability-achievement discrepancies were not labeled LD. Evidence for subjectivity in EPPC decision-making processes was noted by Ysseldyke. Algozzine. Richey. and Graden (1982). They videotaped 20 placement meetings and concluded that 83% of the comments made during these meetings were irrelevant in that they did not refer to classroom or test performance characteristics of students. Regarding the actual classification decisions. they observed that naturally occurring pupil characteristics. such as physical appearance. SES. and sex. functioned as more important determinants of the label than actual pupil perform- ance information. A related concern has to do with the placement committees' propensity to label children LD. The majority of students in this sample were labeled LD. If test scores. diagnostic methods. and federal guidelines have little diagnostic utility. or at least are not being consistently included in decision-making processes. on what basis are EPPCs making their decisions? The answer to this question is of 149 criticalrbmportance. considering that some students are denied services while others (who are very similar in terms of test scores and overall performance characteristics) are placed in special education programs. National prevalence estimates for LD range between 252 and 621 in the referred population. so it seems that the current sample of 50% LD is not unique or biased. Ysseldyke et al. (1982) noted that 70% or 14 of the 20 placement meetings observed resulted in the LD classifica- tion. Lazarus (1985) suggested there was a strong link between the specific children referred and a teacher‘s perception of district services. She observed that teachers referred only those students they were sure would qualify for existing services. Relatedly. EPPC labels often confirmed the teachers' perceptions of a student's problems. It is of interest that referral-to-placement estimates suggest that 921 of the children referred by teachers are tested. and 731 of these are placed in special education. As pointed out by Algozzine et a1. (19823.2EPPCs seem to function in.a manner that validates teachers' conceptions of a student's learning problems. It would seem. then. the assignment of the LD label may be more an artifact of the referral-placement process than the result of a rational data-based assessment of student characteristics. Mehan (1984). referring to EPPC labeling procedures. suggested that a reduction in the range of possible classification alternatives occurs before a case is reviewed by the placement committee. As dis- cussed in an earlier section. children are often referred for testing 150 with verbal or written reports suggesting one or two suspected handi- caps. By the time the case is presented to the placement committee. most evaluation team members have informally agreed on one of the labels. As a result. the EPPC decision-making process has been reduced to a routine "seal-of-approval" ritual rather than a rational step-by- step decision-making procedure. Ysseldyke et al. (1982) observed this process. noting that a high correlation exists between the amount of information presented at a placement meeting and the identification of a child as LD. The more information (or perhaps teacher comments and opinions). the higher the likelihood of an LD classification. If the label is assigned in the manner described. it is not surprising that many of the children in this sample (and in the literature as well) fail to meet federally defined guidelines. Based on the foregoing discussion. it appears that the LD label is frequently assigned by placement committees. In special education. the LD classification accounts for approximately 25-602 of the population of children receiving services. This suggests that those who are making the diagnosis may feel that the label has more beneficial conse- quences associated with it. or makes a difference in terms of the quality of treatment a child receives. However. the research of Mehan (1983) would tend to negate the latter possibility. He found that EPPCs often based decisions for placement on the availability of space in certain classrooms rather than on finding or developing programs to meet the unique learning and behavioral needs of a child. Such 151 placement methods were apparently followed regardless of the label- identification of a child. Bolt (1984) observed similar patterns in her investigation. She noted a lack of correspondence between the severity of a child's dis- crepancy score and whether the child received teacher-consultant resource help or self-contained programming. She suggested that EPPCs were using criteria for placing children in programs other than those specified in federal and state guidelines. It would be safe to assume that matching category label to treatment intervention is not a major consideration in EPPCs' decision-making process. The identification of children as LD does not appear to be linked to ideas that LD special education treatment services per as are qualitatively superior to other special education services. In the current sample. how the schools operated in making classification and placement decisions is open to speculation. The investigator. based on his own consulting experiences in the schools. suggests the following possible factors that may influence placement- team decisions. Teacher characteristics of the receiving special education teacher may be an important determinant of the placement decision. Frequently. there is discussion as to the benefits of placing a particular child with one teacher or another because of the teacher's style of teaching. classroom structure. or the teacher's positive interaction approach to students. Referring teachers at times may even specify a particular teacher as being the best match for the referred child's 152 social-academic characteristics. In the extreme case. the child may have one label such as EI. and the "good" teacher may be the LD teacher. Relatedly. the student characteristics of the children already in a special education program sometimes influence the team in deciding on the "best" placement for an individual child. For example. in some cases discussion revolves around the fact that a particular special education class has too many children who "act out" or exhibit a number of social-emotional-behavioral disturbances. The team may decide that a "calmer" class would be most beneficial for the referred child. In some situations more practical considerations enter into the placement decision. If a self-contained program already has its quota of students. rather than busing a child to another program. the team may decide to put the child in a resource room or on teacher consultant caseload. Parents often seem to have some influence on team decisions. In this investigator's own experience. an outspoken. well-informed parent can often sway the team to modify decisions in some respects. For example. if a parent is adamant in his/her refusal to have the child labeled or placed. the team may decide to have a teacher consultant work with the general education teacher for the remainder of the year and assess student progress. An IEPC may be rescheduled for the end of the year or the beginning of the next year to reconsider the student for special education. The reverse situation has been observed. as well. That is. the team feels that general education intervention 153 would benefit the child. The parent presents a list of "unsuccessful" general education interventions already attempted. The general education teacher often concurs with the parent. perhaps because not to do so would be perceived by the teacher as a personal-professional failure. As a result. the team may decide to place a child in special education. What is the role of the psychologist in placement options? By way of background information. although school psychologists are considered important members of the Multidisciplinary Evaluation Team (MET). their attendance at all Individual Planning and Placement Committee meetings is not mandatory. Psychologists serve as the MET representative at IEPC meetings when a child has a suspected handicap of mental retardation. The teacher consultant assumes this role when the suspected handicap is learning disabilities. and the social worker is the representative when the child is suspected E1. The psychological report contains a recommendation as to the classification category that best meets the child's individual needs and which conforms to state and federal guidelines. The recommendation is that the team consider the child's eligibility for a classification label. The actual placement decision is made through team process. How do teams justify their decisions when placements are not directly related to the label-identification of a child? Reliance on the Individualized Educational Plan--learning and behavioral objectives--seems to be the rationalization used. That is. while a child actually meets eligibility requirements for classification "X" 154 (which may or may not be the label assigned on special education forms). his/her placement in program ”Y" is still considered appropriate because actual teaching objectives are considered to be individualized to meet the child's special needs. LD: The Advanpgggs of the Label If treatment makes little difference. then perhaps members of placement committees believe that the LD label itself has certain benefits associated with it that make it a more acceptable label. In fact. there are some benefits associated with the LD label that may have a significant positive impact on a child's educational and voca- tional opportunities. There seems to be an accrual of benefits with age. That is. because of recent research emphasis on the adolescent with LD. there has been an increasing commitment by the federal govern- ment to fund educational and vocational training programs for LD ado- lescents and young adults (Gerber. 1986). The LD label may actually provide opportunities to access subsidized educational and training benefits not available to nonhandicapped children and children with other handicapping conditions. It appears that once a child is labeled LD. there is considerable advantage to maintaining that label through- out one's educational career. Actual special education interventions offered by LD teachers take on more of an ancillary role as the LD student approaches the middle to end of his/her secondary education. At that time. other forms of assistance preparatory to postsecondary education or vocational train- ing are offered. Being LD allows the privilege of modified testing 155 procedures for those wishing to take college entrance examinations. Test such as the SAT and ACT may be taken orally with a third person writing the responses. In addition. the test is administered untimed. Most LD students who elect to continue their education enter open- admission colleges or bypass competitive admissions by ”transferring" into a four-year college after successfully completing a semester or two in a two-year program (Vogel. 1986). However. because some stu- dents require special advisory and continued academic support. the Association for Children with Learning Disabilities (ACLD) has compiled a list of colleges and universities offering special LD services. This publication provides parents. students. and school counselors with a summary of admission criteria as well as special advising services offered at various universities. In addition to tutorial services. many of the universities now offer waivers or substitutions for specific course requirements such as foreign-language coursework. Modified exam procedures.*word-processing services. and other special accommodations to help the LD student are offered by several universities listed in the ACLD publication. Descriptions of alternative admission criteria used by different institutions suggest increasing recognition of special needs of LD adults. However. considerable variation in requirements exists. indi- cating that there is little agreement about the intellectual function- ing and needs of LD college students. For instance. Wright State University requires high average or above Full Scale IQ on the'Wechsler Adult Intelligence Scale for LD students entering their program. 156 Considerably less stringent are the requirements at Pennsylvania State University. LD students attaining at least 100 points on the Verbal scale are admitted. as admission officials feel that average verbal ability reflects sufficient academic aptitude (Vogel. 1986). While there are more opportunities now available for LD students pursuing postsecondary education. the most direct government support and commitment has been in the area of vocational rehabilitation services. The Rehabilitation Services Administration has recently enacted legislation to include the provision of training services to LD adolescents and adults (Gerber. 1986). The Vocational Rehabilitation Act of 1973 was the first of several laws to allow LD individuals to participate in job-training programs. Subsequent legislation such as the Carl Perkins Vocational Education Act of 1984 extended training benefits to LD individuals through the age of 24. Similarly. the Job Training Partnership Act of 1984 allowed for competitive employment training programs to facilitate procurement of skilled-trade positions in both private and public sectors. Currently. according to Gerber (1986). there is even a greater focus on improving the LD students' transfer from school to work and/or school to postsecondary training. He reported that a special confer- ence in 1983. sponsored by the National Institute for Handicapped Research. set high on their research agenda the special needs of LD adolescents and adults. Particular areas of concern were listed as vocational adjustment. community adjustment. and postsecondary train- ing. He suggested that priorities reflected increased recognition and 157 concern about the persistent effects of the disability throughout the life span. The Social Security Administration has also recognized the persistent effects of learning disabilities. Disability determinations may now include LD individuals who exhibit particular deficits. Simple illiteracy would not qualify an individual as eligible for supplemental income. However. if it is determined that an individual's learning difficulties interfere with job performance. he/she may be entitled to supplemental security income (SSI) benefits. Considering the various benefits. it is reasonable to assume that EPPCs may err on the side of lenience when choosing labels. particularly if there is some uncertainty as to the diagnostic category that is most appropriate for an individual child. Before turning to the final section. some comments regarding the stigma of being labeled should be considered. Even in our psychologi- cally sophisticated society. the E1 label continues to have more negative connotations associated with it than is the case with the LD label. When a child is suspected to be E1. the school social worker completes a child study. This includes extensive interviews with both the teacher and the child's family. While the interview questions are not intended to provoke guilt feelings. the parents and the teacher begin to recount and reflect on their interactions with the child. At the very least. this process leads to self-questioning and at worst sel f-incriminat ion. 158 This is not the case with the LD label. Neither the teacher nor the parent has to feel guilty. The condition exists within the child and is not considered to be the result of social-educational factors. parenting factors. or the child's intellectual and personality make-up. In addition. the child is not viewed as engaging in intentional behaviors that perpetuate the problem. Instead. the child may be perceived as a victim of a little-understood condition that involves brain functioning to some unspecified degree. Thus. an intangible benefit may be that the label itself has less stigma for all concerned. and compared to other special education labels is relatively innocuous. To label some low-achieving students LD and others not is an invidious practice based on criteria that cannot be precisely identified. Borrowing from Ysseldyke et al. (1982): We must begin to evaluate very carefully the purposes and needs being served by identifying certain students as LD while not identifying others (who are very much their twins) (p. 84). The benefits of "appropriate" treatment cannot be specified as there is a paucity of research in this area. However. at the very least. those who are denied services (or at least the label) lose the opportunity to receive vocational and educa- tional support. Recommendations Recommendations for Clinical Practice The efficacy of the federal and state guidelines. as well as the methods proposed for differential diagnosis of educational handicaps. has not been established. It is unclear if these methods are 159 ill-advised for classification purposes because the criterion to which they are compared (school-identified classification) is not reliable. Consistent decision systems used by placement committees could not be identified either in this investigation or in the literature. WISC-R subtest data and K-ABC subtest data have little practical value for differentiating among diagnostic groups (previously classified by schools) and appear equally ineffective to this end. There is no reasonable support either in this study or in the literature to recommend these procedures in clinical practice for the purpose of sorting among educational handicaps. However. these findings may be confounded due to the unreliability of EPPC decisions. Moreover. intelligence tests were never meant for diagnosis. Diagnosis entails evaluation of current status. etiology or causative factors. prognosis. and a recommendation for a specific treatment approach. The original purpose of intelligence tests was to distinguish between normal and mentally retarded or subnormal children. Intelligence tests continue to be valid for this purpose. Unless it can be established that federal and state guidelines are being applied consistently and accurately. there will be no way to determine reliably if the methods are inadequate. or if there is simply no useful information in subtest scores to aid in classification. Any findings regarding classification rules will be confounded by current inconsistent usage of the guidelines by school placement committees. 160 Public-Policy_Recommendations There is considerable confusion with regard to differentiation and identification of LD. EI. and LA (but not handicapped) children. These groups of children are relatively indistinguishable in their low- achievement patterns. The evidence suggests that current categories are not separate. distinct educational handicaps as currently defined. In particular. the LD category is a label of questionable validity because of inability to distinguish from nonhandicapped poor-achieving peers. It seems evident that there will be little progress in under- standing the concept of learning disabilities. or improving and assess- ing intervention and remediation techniques. until classification and diagnostic systems are more clearly structured. Further evaluations of diagnostic utility will be pointless and the results confounded until classification criteria are uniformly organized. defined. and struc- tured into logical decision-making steps. Only then will diagnostic labels be useful for educational planning. School psychologists and educational researchers. because of their training as well as their experience with special education populations. have the knowledge and expertise to work with policy makers to refine definitional guidelines for diagnostic categories. Both professional groups share a common interest in developing decision systems that are both equitable and practical for determining that an individual is eligible for services and benefits. 161 Research Recommendat ions Future research must begin to address the problem of the different classification and placement methods used by EPPCs. It would be of interest to identify precisely the variables placement committees consider important in determining eligibility for programs. The weight given to demographic and performance characteristics of a child and the ordering of that information in labeling and placing children may provide a basis for determining the structure and adequacy of decision systems. A measure of the congruence between the definitional guidelines of educational handicaps and the factors judged important by EPPCs may make it possible to determine the reasons incorrect classification decisions go unnoticed by team members. If preliminary results suggest that poor decisions are associated with confusion regarding definitions of handicaps. then perhaps EPPCs must be provided formal training and practice in applying guidelines to mock cases. Interrater-agreement measures could be used to assess both clarity of definitions and the efficiency and skill level of the decision makers. If the definitions are inadequate. then policy makers. psychologists. and educational researchers must combine expertise and empirical research to refine definitions for practical usefulness. If it is the case that decision makers are failing to adhere to the definitions and guidelines. perhaps increased federal and state auditing is needed to ensure canpliance with standards. A second line of research would address the issue of treatment. Currently. there is little information regarding the extent to which 162 educational handicaps and intervention approaches are matched. or whether consequent interventions really make a difference. The data that do exist are discouraging because they suggest an inverse rela- tionship between academic progress and length of stay in special educa- tion. Clarification of whether particular interventions and strategies make a difference is important for examining the as-yet-vague costs and benefits associated with labeling and placement. There is no point in differentiating among special education students and labeling some LD versus EI versus LA unless there are differential treatments for each of these categories. Furthermore. not only must there be differential treatment. the treatment interventions need to be evaluated and identified as beneficial. Until questions about diagnostic procedures and the costs and benefits of labeling and placement are answered. a classification system that labels all students with educational problems as LD may be an alternative to the present state of confusion in educational diag- nosis and placement. Whether or not school interventions are found to be helpful. at least the student who is designated as LD will have access to the previously discussed postsecondary educational supports and vocational-training benefits. APPENDICES APPENDIX A WISC-R AND K-ABC RELIABILITY TABLES 1(53 nsssoasso mm. om. mm. ~m. so. am. as. on. oom.a oossaoogos new new: ncmuoawso so. mm. mm. ms. ms. as. ms. com Hoocossso mow one: am. mm. mm. as. on. no. so. ow. ooH mnwn ou onwa Hm. no. so. me. vm. be. me. on. com «anaa 0» onHH om. Ho. so. No. No. or. on. $5. oom fiance Ou unoH co. «m. hm. mm. mm. mm. me. me. oom aanm cu one mm. mm. mm. mm. em. om. an. on. oo~ Mano 0» onm we. do. vm. om. mm. on. No. me. oom Hanh Cu one No. on. do. no. mm. mm. on. me. oo~ Mano Cu one on. mm. om. mo. em. on. Am. oom Hanm Ou onm em. am. no. no. he. on. on. com Hanv Cu One mm. me. he. on. we. con aanm Cu onm mm. mm. .om. as. an. ace «H-~ 00 ou~ moauom >u0§ox moaooHcc< macho no ocean Hacoom mannedu maceEo>oz casuacooOom zones: Ouocm Acwucam xenon: one: a . a noneaz uaoumoo occ: some Deon: I z moose oo< mumouaom unannoOOnanHoucmz «NH zoooune mm mood .mamsom sewumuwoumocoum on» new .oo< >n .mumounom UmDAAfinowaom uHMInUMHQm ">ocmumwmcou «newsman ”A.4 manna 164 .coMumEMOumcouu m.m.uo£mam moans >9 oocwouno cum: mucowowuuooo can: a .maoeu0m csoumncmsuoomm or» >n :uocoa amounuamn you oomuouuoua couoawco do. No. om. hm. om. oom.~ o mnHOOcom uOu coo: acouoaaco mm. hm. hm. mm. oom Hoonumoum new coo: om. Ho. om. om. om. mnNH 0» on- mm. mm. mm. mm. mm. oom Hanna Du onHH om. om. no. mm. mm. oom Hanoa 0» onoH we. do. mm. mm. om. oom Hana 0» onm mm. no. em. mm. mm. oom Hana 0» onm mm. mm. om. om. on. com Hfinh 0» onh no. mm. mm. Hm. oow Hana 0» one vo. mm. mm. Hm. oom Hanm 0» onm mm. mm. mm. mm. oom Hanv Cu one mm. mm. on. om. oom Hunm ou onm on. om. ooa Hana 0» onm oesocmumuocco ocaooooo moccaa a >um~onmoo> \ocaomom \oc«ooom moaooam oaumecuwud moomm o>ammouaxm .m macho oo< mumounsm ucoEo>oaco< .ooscwucou ”H.< ounce 165 .coHuMEHOchsuu.m m.uecmum moans ha oecwmuco eue3 museuoummeoo seezn .mHSEHom ecu oucw veueuce mes eaeoe Heaaewuuem e mu umeucsm come now ucewoumueoo muwdunefiaeu maecnuwamm ecB .eufimomsoo e no >uuauceuaeu ecu How cHoEHOM Anon .m .vmmav m.ou0mau:0 moan: oeusmfioo eue3 muceauammeoo emecam newscasts mm. so. so. mm. mm. oom.u wasusoocom How one: assesses so. mm. as. em. om. com woosomoso now see: mm. as. so. mm. so. ooH mums 6» cums so. as. us. mm. om. ooN Hausa 6» ouHH «s. .6. mm. vs. mm. cos dance ou ones em. as. mm. vs. mm. com Hana 6» oum em. as. mm. mm. om. oom Hana on one No. as. . mm. as. mm. oom Hans op on» mm. mm. mm. mm. . as. com Hans 6» one mm. om. mm. mm. «a. cow Hanm ou oum am. mm. mm. mm. as. com Hans 6» one mm. am. so. as. com He's 6» o-m Hm. om. em. mm. ooa Hans ou ens 22...... .22...- .2.wwm.w22 .fiwmfi. wwwfi . 2... 2. mud cosouce MN meoc .eHQEem coHuonoueosmum ecu How .eoc >c .meamom HocoHU undrx ecu How omuceweummeou >uwaflcewaem "acceuewmcou HmcueucH "N.< eacea 166 .mHsEHom Ammm .m .«mmav m.ouomaflao wean: .uaflumwu umuwm mg» no cmcflnuno cofluMa>w© cumvcmum on» no nmmmn .Amumwunsm unwam>mfisom Mom ma u.mm .mummundm @cfimmwooum Hmucwz How m ".mmv anoum mane: mnu mo hufiawnmaum> ms» Mom owuowuuoo oum3 mucmwowmmmoo unflawnmwamu Had a .ucmvsum £00m How mcfiumwu vacuum a mhmzam no: mm: mumsu ow .cowumuumwcaEUm umnuo may no Hw>ma mummm 0:» new newumuumwcwfium mac :0 ommnu m5» m0 Hm>ma vlmmm wnu cw>wm mum3 mmmmo ammuwulummu meow Anv can .mmmcmu mom wmmnu cfinuw3 mmmm cwmuumo 0» same :w>au mum mummunsm 026m any uncommounozu uom m smacks» m can v nmaounu mm mwwm um ammunsm ou ammundm scum >um> mwuam wamfimmm mm. on mm. mm agave-umumocoxmcficmmm .oH «a. on mm. mm unaccomnxmcficmmm .ma cm. on um. am up. as mwacofim .vH Hm. ou mm. «a nm. as cap-snufiua .MH mm. on mm. mm mm. vm moo-Hm on. mmomm .NH mm. Nb >Hmasnwoo> m>fimmmumxm .HH mum wugm #Cwfimsrw .mSOd om. on an. hm mwflumm ouonm .OH ms. or hm. mm muoEm: Hmflummm .m om. on mp. mm mmflmodmc< xfluum: .m ms. or on. mm mm. 0v “mono vno3 .5 mm. on on. mm .5. o. mmflmcmflua .o am. on vm. mm mm. vm Hamowm anEDz .m cm. on vm. mm m5. vm oudmoHU uHmummU .w mm. on am. mm mm. vm mucmam>oz ocmm .m N@. Nb :Ofiuflcvoowm 00mm .N m5. Nb zoocfis 0.0.: .H mummunsm mcflmmmooumuamucmz nu mz an m2 an mz mINH cusouna cum mmmg Haum nwsoune oum mwmc Hfluv nmsouna mum mmmc ammunsm umonm cam v mmwm um >Hno wwumumficflsvm mfi mamom Hmnuw>coz may mmamomn 0v u.%o .mcaummu umuwm mzcfifi mcflummu vcoomm u whoom came a .maneuOM “Nam .m .vmmflv m.ouomaflzu mafia: .mcflumou umuflm may no nuafimuno coflpmfi>mo magnum». on» no ummmn .AmH u.mmv msoum mane: mnu mo muwflflnmflum> may you cmuomuuoo mums mucmfiofimmooo aufiafinmflamu Hag. m.v m.na n.00a n.ma ~.~oa um. Hanuo>coz o.~ H.0H o.moH m.mH o.aoa hm. unmem>mfinu¢ m.v n.5H m.moa b.oa ¢.Hoa mm. mufimomsoo mcflmmmooumuamucmz b.m o.>a 0.50. m.oH m.aoa Hm. mcflmmmooum mac-cmuHsEAm m.~ m.>H H.~oa v.0a m.mm mm. acfimmmooum Hague-swmm .0» u z. muma on o-m mum. ~.¢ n.5H o.moa m.oH m.aoa mm. Hmnum>coz m.H H.ma m.woa m.vH b.~oH mm. u:w&m>m«:o¢ m.v m.bH ~.hoa m.ma «.moa mm. muamomsoo mcflmmoooumuamucmz «.0 v.5H v.moH m.mH o.m0H mm. mqflmmmooum mac-cmuassfim o.H m.mH m.moa N.ma N.NOH mm. ocwmmmooum Hmflucmawmm Amm n z. Haum op oum mean m.m m..H ~.m0H m.~H ”.mm Hm. canny-2:02 o.m v..H o.moa m.NH o.moa mm. ucmsm>m«:04 H.m m.va m.ooa m.mH v.Hoa mm. mufimomeou magma-ooumnamucmz ~.> m.mH >.m0H m.vH m.moa up. mcfimmmooum msomcmudseflm o.H N.VH >.ooa o.ma “.mm up. mcwmmmooum Hmflucmsvmm Ava u zv Hale 0» mum mmm. muoom .mw cum: .mw cmwz I. .H wmo l anamw mafiumwe vcoomm mcwumwe umuflm m H m 0mm M mmflmom Hmnoflo unarx mg» non mus-floflmmmoo aufiafinm.amm amouomuumma ”v.4 GHQMB 168 .mm wow you mmmdm umwn 0:» ma «5 mom um swam gamma How vmcwmuno vm. may ..m.w «Hm>wa mmm unmomnvm ms» um omcwmuno wsam> may ma meow uo: was unfiummumu mumn3 Hm>mH mum am pm mcwcou uo swam uwmwo mo ucmfioflmmwoo mpflHfiQMfiHmu ms» mo mumswumm umwn may a .GOfiumshommcmnu.m m.umnmwm ucwmz >3 vmusmfioo mm3.m mmmuw>m mnam mm. mm. mm. mm. mm. mm. mm. mm. mm. mm. mm. mm. 0H mfimomufiasm om. Hm. om. mm. om. am. mm. mm. am. am. om. Hm. 0H mocmsuomuom «m. mm. «m. mm. mm. mm. mm. mm. em. mm. mm. mm. 0H Hanum> mp. pm. mm. mp. mm. mm. mp. mm. Hp. pp. Hm. mm. mmumz mp. n om. mm. a n mp. mp. a 3 mm. mm. mcmcou 0p. Hp. mm. mp. mp. mm. mp. «m. op. mm. mp. mp. pane-mm. uompno mm. om. mm. .m. mm. mm. mm. mm. om. mm. mm. om. comm-o xooam mp. 0p. mp. vp. mp. mp. mp. mp. mp. mm. mp. pp. acmEmmcmuu¢ musuomm pp. mp. mm. Np. mp. mp. om. mm. mp. mm. mm. vm. nomumamaou muauomm mp. n mp. mp. n n mp. Hp. n a 2m. mp. cmmm ummmo pp. mp. Np. mm. am. pm. mm. Hp. mp. mp. cp. mm. comma-smumsoo mm. mm. om. Hm. mm. mm. mm. em. mm. mm. Op. «p. pumflsamoo> pp. mp. om. mp. Hm. om. Hm. mp. om. mm. mp. mp. emu-en»... Hm. mm. «p. Hm. mp. vm. Hm. mp. mp. mp. mm. pm. mmmumumamafim mm. mm. om. mm. pm. pm. mm. mm. mm. om. om. pm. nomumeuomca «Ham mma «ma mvfl mma mud mHH med mm mm mp mm wmmuw>< macho mod umma Mom com u.mv .04 an .mmamom Amsouw mmfl 30mm OH on. mum-a ocu mo mus-momummoo pumHmAMmHmm muomHs .m.< xflccmmm< 1659 .cowuchOumcmuu M a.uozmpm ocuma >9 usuaaeoo no: ~HM nuuuoquU coouo>u «:9 n .Q:Ouo memo: wzu no >uu~a£¢uh~> on» new vmuoouuou «no: wucwuu«uuooo puuuwnofiaou 0:96 .couvauzu «no puuqu: new nxuo) m Ou n scum vovcou nasaumuu ocouwa 0:6 umuwu 0:» :vozaon aa>uou:« cad» och .non oguumnauzm can .oucthOuuom .sunuo> use uOu nuns: 0H :« can «you» on» uOu nude: wuoomnvOAQUn cu uuo anewuufi>ou unoccaun can accox “vuoz no. om. no. mo. Hp. up. no. mp. mp. pp. am. mm. ow. am. am. gnaw vouuuuuou .o>< ma. om. vo. Om. ~p. mp. mp. @p. mp. mp. om. mm. mm. mm. mm. nNfim vauocuuou mm. am. mo. om. Qp. Vp. om. cp. «m. Am. pm. 0o. em. mm. Hm. Nam o.m~ ~.m~ ~.p~ o.n o.m m.m ~.m o.m p.n ~.m ~.m ~.n o.n m.m o.m mm o.mo~ v.mo~ o.oo~ ~.o~ ~.- p.o~ H.o~ m.- m.oH m.o~ ~.o~ m.¢ m.on ~.o~ o.o~ cue: Oswumoh vacuum v.m~ v.v~ a.o~ n.n «.n m.n o.n p.~ v.m m.n o.m ~.n o.m n.n m.~ mm ~.om ~.oa m.oa @.a n.o~ v.m ~.o ~.¢ n.m m.o m.o 0.0 p.@ m.m o.m coo: ocwunop uuuum .vodnz. «maumva nom< mo. om. mo. mp. pp. pp. mm. mm. ~o. mp. go. mm. mm. mm. mm. o-w oouuouuou mm. mm. no. ~p. pp. op. mm. mp. om. op. Ho. om. ~o. Ho. do. Ndm H.m~ n.0a m.ma ~.m ~.n H.n ~.m n.n H.n o.m w.~ m.~ m.~ N.m o.~ mm ~.oo~ n.mo~ v.~c~ v.- p.HH m.- m.o~ o.- m.o~ o.- m.o~ m.o~ o.o~ m.o~ 0.0" :uwt atmuawh vacuum p.m~ m.n~ v.- ~.m o.m a.~ o.n ~.m H.m ~.n m.~ ¢.N o.~ m.~ v.~ mm o.ma m.mm o.ma a.m N.oH p.@ o.a a.@ p.¢ p.o~ ~.oH o.o~ m.o m.¢ o.m cow! acquwup on... .poauz. m--mo~ numo vm. om. oa. we. no. mp. mp. Op. me. om. cp. mo. Op. vp. vm. GNAM VOuuouuou mo. om. pm. mm. mm. op. op. pm. mo. do. Np. pm. No. Hp. mo. mom o.Va o.m~ p.- m.m m.~ o.n ¢.~ v.n m.~ m.~ m.~ m.n m.~ m.m m.m mm m.vo~ m.oo~ ~.~o~ ~.- m.o~ m.o~ p.o~ o.- H.~a m.o~ o.o~ m.o~ v.o~ N.o~ m.o~ cow! ocwumoh ucouom ~.n~ m.v~ n.- ~.n m.~ ~.n ~.n H.m ~.n ~.m m.~ o.~ m.~ ~.m m.~ mm o.pm o.pm n.mw w.a m.m m.m w.m 6.0 0.0 H.0A m.¢ o.o~ o.w m.o m.¢ :00: vamunuh ununm .pmuz. «p-4o mom< ..mw. .w“. ..m“.. ..... ...... .HMHMMM. .NWMH“ -.M”M”.. ..M“Mw. .mmwm WMwMHMM -“Mww” -“wwwu ..MNDHM. ”WMwmw ..... I-sm neuouuum . «unavan ousuuwm . . . . ooo3p coumop couodoco uo monouo ocucp no. mom on. mom-p ¢-omH: 2:» oo auoofiufioooou pomdfinaum .m.< o~n¢e 1f70 cuuoafioo ~.~ m.H o.~ ~.H ~.H m.d m.H m.H oom.a oomuaoonom hem coo: :muodmgo ~.H o.H o.a m.~ 2.. o.~ m.~ oom floonumouo How cum: H.H v.H o.” m.. v.H N.H p.H m.H ooH mupa 0» o-~H m.H ~.H H.H m.H ~.H v.H p.H m.H oom HH-HH ou o-HH H.H m.a ~.H m.H m.H m.H m.~ o.~ ooN Hauoa ca o-o~ ~.H n.H H.H m.H ~.H m.~ m.H m.H oom Haum cu o-o m.H ~.~ H.H o.H ~.H m.~ m.H v.H oom Hana 0» ouo m.~ m.o ~.H n.H ~.H o.H m.H m.H oo~ HHIp ou oup n.H m.~ m.H H.H o.H ~.H v.H m.~ oom ngm 0» o-m v.H m.H o.~ o.~ H.H v.H m.~ oom Hfi-m 0» o-m ~.H o.. A.“ m.. o.H m.~ m.~ oom Aduo ou one o.. m.H v.H o.~ m.H oom HH-n 0» oum o.H p.H m.~ o.~ m.H ooH --~ ou m-~ mmwumm puOEw: mmwmonc< uocuo mm ocmaufi Hamucm musmomu mucmEv>oz newuwcvoowm soccfi3 ouoza “afloaom x.uomz one: H . .onssz admomwo can: mum. ado-z _m ozouo oo< mumwunsm ocfiwmeOuauHmucoz and £030u59 mm mwo< .maafidm sewumnwcumccmum mnu uOu .mv< >3 .mumoundm um .l \ocwomwm \oc.o.om . . mmomm o>mmmouoxm z osouo moc mumoundm ucwEw>mflzu< .cmSCMucou "p.< wanme 1371 couoflfizo v.o o.v o.m ¢.m o.m oom.H moaufloosum ham cow: :wuoamzo H.m o.m ~.p o.m oom Hoogomuuo now now: o.m m.o o.m p.v o.m ooa muma ou o-~H ~.m o.m o.m o.m o.m oom HH-HH 0» ouafl p.v p.o e.m o.m o.m oom Hauoa ou ouoH ~.v m.o o.m ~.m o.m oom Haum 0» o-m o.m o.v o.m o.m o.m oom Ho-o ou o-m o.m o.m o.m p.m o.m oom HH-p ou oup m.~ o.m o.m m.o oom HHIm ou o-m p.n o.m o.m m.m oom Haum 0» o-m o.m o.m o.m o.m oop Hang on one o.m o.m ~.m p.m oom HH-m 0» o-m o.m o.m ooH Hauw 0» mum ocflwmmwmuwcca ocficoomo moanuflm odomenuflu< mmomao a pumasnmoo> I. msou mm .cmmm \@:wvmwm mmumm 0>Hmmmumxm z o d mumOunzm ucmsm>mwzu¢ .cQDCMuCOU up.< wands 172 :muodano m.m p.m m.m o.v o.m oom.a mmmuaoonom How cum: :muoawnu o.m o.m o.o p.m o.v oom Hoooumoum Mom cums o.v o.m p.m o.v v.m ooH mINH cu OINH p.m o.m p.m o.v o.m oom HHIHH on OIHH p.m o.m ¢.m p.m o.m oom Hated cu OIOH p.m o.m v.m p.m ~.m oom Hana cu onm p.m o.m v.m o.v p.v oom Anna on cum ~.v o.m o.v m.v N.m oom HHIp 0» cup o.v o.m v.m o.v m.v oom Hate on one o.v o.m o.m o.v N.v oom HHIm 0» onm v.m v.m o.v o.m m.v oom HHIv 0» ouv o.v o.m o.m o.m oom Mann on elm m.v p.v o.m o.m ooa Haum 0» mam muwmomeou ocpmmmooum vcwmmmooum .I Hmnum>coz ucmEm>mfino¢ mapmmmooum Hmucmz msowcmuHsEHm Howucmsvmm z macaw oo¢ MNH nmzouna Mm mwmfl .mHmEmm coflumu awoumocmum 0:» new .mm¢ >9 .mmamom Hmnon om¢nx may you unmemuammmz mo muouum cumncmum "m.¢ wands 173 oaocfim m£B .m.¢ manna no a muocuOOM CH wwcflmamxw mm .wm mom um OGHUOU Mom mp cofiumwuxm .Hm>mH mwm uswomflom way an Umcfimono 05Hm> may mp meow uon mm3 mcfiummumu $9053 Hw>mH mom am pm mcflcoo no cmmm uwmwo mo ucwEmHSmmmE mo uounm cumwcmum m3» m0 mumEpumm umwn mnem oH.m mH.m oH.m mH.m m~.m mo.~ mo.~ H~.m oH.m m~.m om.m Ho.m 0H mfimom Hash om.o om.o om.o op.o mo.o om.o mm.o mm.. m... mv.o oo.o mp.v 0H mocmeuomuom om.m pm.m «o.m oo.m «o.m ma.m om.m mm.m om.m mo.m ~o.o oo.o 0H Hmnum> op.H mo.m mm.H mm.a mo.H po.H mm.H Ho.H mm.H mo.H mm.H oN.H mmumz mm.a m «o.m op.H . . om.H m... a m «o.m om.H osmooo op.H op.H No.H mm.a Hp.H om.H pm.H Hp.H om.H op.H mm.H mm.H pans-mm. pompoo pH.H om. oo.H oa.a oa.a mH.H mo.H ~H.H Hm.H om.H p~.H mm.H ooflmoo xuoflm pm.H Hm.H om.H om.H om.H Hm.H om.H mm.H om.H Hm.H mp.H mm.H acme-ocmuua musuomm mo.H om.H mo.H om.H Hm.a om.H pm.H om.H mm.H mH.H o~.H mm.H nomumaoaoo muauoflm oo.a m Ho.a Ho.a m m mv.a mm.a m m mH.H mm.H ammo umofio om.H Hm.H ~m.a -.H p~.H mo.H H~.H m... oo.H mo.H Hm.H mm.H cofimamnmuosoo mH.H pm. om. mm. oo.H mo.H mH.H oH.H H~.H H~.H mm.H mm.H pumasomoo> mm.H om.H m~.H mo.H om.a om.H om.H om.H m~.a o... H..H mm.H afiumszufiua om.H m~.H om.H om.H om.a oH.H pm.H mo.H mo.H Ho.a om.H oa.a mmmumumamamm oH.H NH.H mo. mo.H oH.H mo.H oo.H oo.a om.H mm.H mm.H po.H comp-anomoH mwn. - . mm «ma mmH wwa Mma «NH mad mom mm mm Mp «m monum>¢ umme macho wmd wwlu “macaw mm¢ comm you oom u zv mod >0 .mOH can mmuoom Umamow m5» MD A va ucmEmHSmmmz mo muouum Unaccmum mIUmHz "m.< manna APPENDIX B COURSE DESCRIPTION AND OBJECTIVES 174 CE? 822 Kaufman Assessment Battery for Children Following completion of this course the student should be able to (1) administer and score the K-ABC, (2) to discuss the theoretical rationale underlying the K-ABC, (3) to interpret the K-ABC from an empirical or psychometric standpoint, (4) to interpret the K-ABC from a clinical standpoint, (5) conduct via the K-ABC a psychological assessment of minority youth and a wide variety of exceptional children, and (6) be better able to compare the K-ABC vis-a-vis the WISC-R. COURSE PREREQUISITE: All students must be either practicing school psychologists or school psychology interns with prior courses in tests and measurements and individ- ual mental assessment. COURSE REQUIREMENTS: The course requirements are twofold: (1) Required Readings: The K-ABC Administration and Scoring Manual; The K-ABC Interpretive Manual (2) Field Experience: Each participant is to administer and interpret S KPABC's. To insure that experience is acquired with a variety of children, one KPABC and the WISC-R should be given to each of the following, if at all possible: a.) a learning disabled child b.) an emotionally impaired youngster ) a black child ) an EMI student ) and a student who was referred for psychological evaluation but who was found not eligible for special education. Each case should be an initial referral. The order in which the KPABC and WISC-R.are given should be alternated_from case to case. For example, with your first case, if you give the WISC-R first and then the K-ABC, in your second case you should give the K-ABC first and then the WISC-R. The first 2 tests should be observed by a colleague in the course or by someone trained in the K-ABC. Use school-age children between the ages of 5-0 and 12-5. Do not prorate. Please use the PIAT and the Teacher Rating Scale with each case. These measures will give you an idea of how'well the KPABC relates to school success. For the field based experience portion of the course, you should complete: a.) the data form for each child (delete student's name) b.) the clinical rating sheet c.) the test evaluation protocols (K-ABC, WISC-R and PIAT) d.) and the teacher rating scale. Both the K-ABC and WISC-R test booklets are to be turned in by December 1. The tests will be returned to you when we meet for our final session on December 6, 7-10 p.m. on the MSU campus in Room 0103 Wells Hall. During the first part of the final session, you will observe a video tape of a K-ABC administration and you will be asked to score it. So bring a K-ABC record booklet and manual. During the second portion of our final meeting, we will discuss your observations and judgments regarding the clinical use- fulness of this scale. GRADING is on a Pass/Fail basis APPENDIX C EXAMINER QUESTIONNAIRE 175 CE? 822 EXAMINER QUESTIONNAIRE Please answer the following questions and base your responses on the 5 evaluations you submitted as a requirement of this course. (1) What were the dispositions of the cases that you completed for this course? ( You may refer to your protocols) Please enter the case number on the line preceding the appropriate placement category. Educable Mentally Impaired (EMI) Learning Disabled (LD) Emotionally Impaired (EI) Regular Other (please specify) a Teacher recommendations/consultation only Referral to outside agency (please specify) Comments: (2) How large would the Full Scale IQ score difference between WISC-R and heABC in each of the following F.S. IQ ranges have to be before they would affect your decisions regarding a particular case? I. I9 range 60-25 II. ran e 0 III. IQ range 20-110 __ 1-2 points ___1-2 points _1-2 points _3-5 points _3-5 points :3~5 points :6-8 points :6~8 points :6—8 points _9-1.1 points 9-11 points :9-11 points __'over 11 points __,over 11 points _over 11 points Comments: (3) If you had usedo nly the KeABC scores would some of your case dispositions have changed? yes no If you answered yes please specify the change by entering the case number on the appropriate line. Educable Mentally Impaired (EMI) Learning Disabled (LD) Emotionally Impaired (EI) Regular Other (please specify) Teacher recommendations]consultation only Referral to outside agency (please specify) Comments: 176 (4) Based on your recent experience comparing the WISC—R and the K-ABC, do you think the Kathe will replace the WISC-R? yes no I Please comment: (5) Please indicate the approximate number of psychologicals you administered last year. (6) How many psychologicals do you expect to administer this year? (7) Name: Highest degree earned: Years of training as a School Psychologist including internship: Years of experience as a practicing School Psychologist: Additional Comments: A summary of the combined test results comparing the WISC-R and KaABC will be sent to you as soon as all of the data have been analyzed. APPENDIX D DATA FORM ' CEP Data Form 177' Subject # Sex: Date of Birth: CA: Race: Grade: (White) Black, Hispanic, Other) SE8 (Parental Education)(/): One or More Years High School Grad College or Less than High School or GED Technical School College Graduate (0-11 years) (12 years) (13-15 years) -- (16 years +) Reason for Referral: ~ Order of'Administration: (J) x—asc first wxsc-a first Examiner: . Race: *”___ Sex: __‘- Examiner's School Setting: Urban Rural Suburban Years of Experience as a School Psychologist: Please rate every item on the following scale: very high high average low very low 1 2 3 4 5 K-ABC interest level of the materials _Comments: clarity of instructions and teaching examples Comments: emphasis on language ability required by the_tasks Comments: precision of floor levels of the various subtests for different age groups , Comments: clarity of scoring procedures Comments: WISC-R 10. 11. 12. 178 K-ABC ' WISC-R objectivity of scoring Comments: extent to which Mental Processing (Simultaneous a Sequential) scores can be interpreted to and understood by.teachers and‘ parents Comments: ease of translating test performance to practical suggestions for educational planning and curriculum interventions Comments: sensitivity of Mental Processing Scales (Simultaneous vs. Sequential) for differential diagnosis Comments: extent to which the test assesses needs of minority children Comments: the value of the sociocultural norms (parent educational level) Please rate the final question using the following scale: valid . questionable underestimate O 1 3 this child's score is most likely Comments : REFERENCES REFERENCES Ackerman. P.. Peters. J.. 8: Dyckman. R. (1971). Children with spe- cific learning disabilities: WISC profiles. Journal 2.: Learning Disabilities. £(3). 33-49. Algozzine. B.. Forgnone. C.. Mercer. C. D.. & Trifiletti. J. J. (1979). Toward defining discrepancies for specific learning dis- abilities: An analysis and alternatives. Learning Disability Qgrterly. 3. 25-3 1. Algozzine. R.. & Sutherland. J. (1977). Non-psychoeducational founda- tions of learning disabilities. Journal 2; Special Education. ;1_ (1). 91-980 Algozzine. B.. & Ysseldyke. J. (1981). An analysis of difference score reliabilities on three measures with a sample of low achiev- ing youngsters. Psychology 3 the Schools. _13. 133-138. Algozzine. B.. & Ysseldyke. T. (1982). Learning disabilities as a subset of school failure: The oversophistication of a concept. Institute for Research % Learning Disabilities. University 2; Minnesota. Q. 1-18. Algozzine. B.. Ysseldyke. J.. 8: Christenson. S. (1983). An analysis of the incidence of special class placement: The masses are burgeoning. Journal o_f Special Education. 3. 141-147. Anastasi. A. (1984). The K-ABC in historical and contemporary perspec- tive. Journal 9; Special Education. .l_8(5). Anderson. M.. Kaufman. A. S.. 5: Kaufman. N. L. (1976). The use of the WISC-R with a learning disabled population: Some diagnostic implications. ngchology jg the Schools. _l_3. 381-386. Berk. R. A. (1984). An evaluation of procedures for computing an ability-achievement discrepancy score. Journal 3; Learning Dis- abilities. _1_7_(5). 262-266. Bersoff. D. N. (1981). Testing and the law. American Psychologis . .3_6_(10). 1047-1056. 179 180 Bersoff. D. N. (1982). The legal regulation of school psychology. In C. R. Reynolds & T. B. Gutkin (Eds.). The handbook g_f_ school Egychology (pp. 1043-1074). New York: John Wiley 5 Sons. Bolt. K. S. (1984). _A search for homgeneous subgroups 2; learning disabled children usi_ng psychological tests. Unpublished doctoral dissertation. Michigan State University. Brenton. B. W. & Gilmore. D. (1976). An operational definition of learning disabilities (cognitive domain) using WISC full scale IQ and Peabody Individual Achievement Test Scores. Psychology 112 £113 Schools. 13(4). 427-432. Chatman. S. P.. Reynolds. C. R.. & Willson. V. L. (1983) Multiple indexes .o_f test scatter 93 the Kaufman Assessment Battery for children. Paper presented at APA. Anaheim. California. Clarizio. H. & Bernard. R. (1981). Recategorized WISC-R scores of learning disabled children and differential diagnosis. ngchology i_n the Schools. _1_§. 5-12. Clarizio. H. P.. & Veres. V. (1983). WISC-R patterns of emotionally impaired and diagnostic utility. ngchology _i_r_: the Schools. _2_0_. 409-414. Cone. T. E. 8 Wilson. L. R. (1981). Quantifying a severe discrepancy: A critical analysis. Learning Disability Quarterly. 4. 359-371. Das. J. P. (1984). Simultaneous and successive process and K-ABC. Journal if. Special Education. 229-238. Das. J. P.. Leong. C. R.. & Williams. N. H. (1978). The relationship between learning disability and simultaneous-successive process- ing. Journal 9; Learning Disabilities. .1_1(10). 618-625. Dean. R. S. (1977a). Distinguishing learning-disabled and emotionally disturbed children on the WISC-R. Journal 9; Consultifl and Clinical Psychology. 59(2). 381-382. Dean. R. S. (1977b). Patterns of emotional disturbance on the WISC-R. Journal 2; Clinical Psychology. QM). 486-490. Epps. S.. Ysseldyke. J. B.. & McGue. M. (1982). Differentiating LD and non-LD students: "I know one when I see one." Learning Disabil- i_t_:1 Quarterly. 3-13. Feldt. L. A. (1967). Reliability of difference scores. American Edu- cational Tech Journal. .4. 139-145. 181 Fisk. J. L.. & Rourke. B. P. (1979). Identification of subtypes of learning-disabled children at three age levels: A neuropsychologi- cal. multivariate approach. Journal o_f Clinical NeuropsygholoH. 1(4). 289-310. Flaugher. R. L. (1978). The many definitions of test bias. American Psychologist. .32. 671-679. Frame. R.. Clarizio. H.. 8 Porter. A. (1984). Diagnostic and pre- scriptive bias in school psychologists' reports of a learning disabled child. Journal _o_f Learnifl Disabilities. l_7_(l). 12-15. Frankenberger. W.. & Harper. J. (1986). Participation of psycholo- gists in multidisciplinary team evaluations. ngchology _i_n th_e Furlong. M.. & Yanagida. E. (1985). Psychometric factors affecting multidisciplinary team identification of learning disabled chil- dren. Learning Disability Quarterly. _8. 37-44. Gerber. P. (1986). Learning disabled adult nexus: Emerging American issues and European perspectives. Journal _o_f Learning Disabili- ties. .1_9(1). 2-40 Glaser. R.. & Bond. L. (1981). Testing: Concepts. policy. practice. and research. American Psycholflist. 3_6_(10). 997-1000. Goetz. E. T. & Hall. R. J. (1984). Evaluation of the Kaufman assess- ment battery for children from an information-processing perspec- tive. Journal 2; Special Education. 262-277. Gutkin. T. B. (1979). WISC-R scatter indices: Useful information for differential diagnosis? Journal 2; School Psychology. 11. 368- 371. Haddad. F. A. (1983. March). A validity study 93 the K-ABC with learn- igg disabled children. Paper presented at the convention of the National Association of School Psychologists. Detroit. Michigan. Harvey. R. R. (1980). Learning ability/achievement discrepancy. In Woody. R. H. (Ed.). Encyclopedia 2; Classical Assessment (Vol. II. pp. 790-802). San Francisco: Jossey-Bass. Henry. S.. & Wittman. R. (1981). Diagnostic implications for Banna- tyne's recategorized WISC-R scores for identifying learning dis- abled children. Journal _o_f_ Learning Disabilities. .l_4_. 517-520. 182 Hooper. S.. 8: Hynd. G. (1986). Performance of normal and dyslexic readers on the Kaufman Assessment Battery for Children (K-ABC): A discriminant analysis. Journal 9; Learning Disabilities. 12(4). 206-210. Hopkins. K. D.. & Hodge. S. E. (1984). Review of the Kaufman Assess- ment Battery (K-ABC) for Children. Journal _o_f Counseling and Development. 63. 105-107. Inglis. J.. 8 Lawson. J. (1986). A principal components analysis of the Kaufman Assessment Battery for Children (K-ABC): Implications for the test results of children with learning disabilities. Journal 9_f_ Learning Disabilities. 12(2). 80-85. Johnson. V. (1980). Analysis of factors influencing special education placement decisions. Journal 93 School Psycholgy. 18. 191-202. Kamphaus. R. W.. Kaufman. A. S.. & Kaufman. N. L. (1982). A Cross- Validation Study 93 Sequential-Simultaneous Processifl 93 Ages 3 l_/_2_ :_ 12 1/2 Using the Kaufman Assessment Battery for Children (K-ABC). Paper presented at APA. Washington. D.C. Kamphaus. R. W.. & Reynolds. C. R. (1984). Development and structure of the Kaufman assessment battery for children. Journal 9; Special Education. 213-228. Kaufman. A. (1976a). A new approach to the interpretation of test scatter on the WISC-R. Journal 2; Learnitg Disabilities. 2. 160- 168. Kaufman. A. (1976b). Verbal performance IQ discrepancies on the WISC-R. Journal _9_f_ Consulting and Clinical Psycholgy. fl. 739-744. Kaufman. A. S. (1979). Intelligent testing with the WISC-R. New York: John Wiley 8: Sons. Kaufman. A. S. (1981). The WISC-R and learning disabilities assess- ment: State of the art. Journal 9; Learning Disabilities. .1_4_ Kaufman. A. S. (1982). The impact of WISC-R research for school psychologists. In C. R. Reynolds 6. T. B. Gutkin (Eds.). 1133 Handbook 2; School Psycholgy (pp. 156-177). New York: John Wiley & Sons. Kaufman. A. S.. & Kaufman. N. L. (1983a). Kaufman Assessment Battery for Children Administration and Scorig Manual. Circle Pines. MN: American Guidance Service. 183 Kaufman. A. S.. 8: Kaufman. N. L. (1983b). Kaufman Assessment Battery for Children Interpretive Manual. Circle Pines. MN: American Guidance Service. Kaufman. A. 8.. & Kamphaus. R. W. (1984). Factor analysis of the Kaufman assessment battery for children for ages 2 1/2 through 12 1/2 years. Journal 92 Educational Psycholcm. EM). 623-637. Kaufman. A. S.. Kaufman. N. L.. Kamphaus. R. W.. & Naglieri. J. (1982). Sequential and simultaneous factors at ages 3 - 12 1/2: Developmental changes in neuropsychological dimensions. Clinical Neuropsycholfiy. 2(2). 74-81. Kaufman. A.. & McLean. J. (1986). K-ABC/WISC-R factor analysis for a learning disabled population. Journal 92 Learnipg Disabilities. Kavale. K.. & Andreassen. E. (1984). Factors in diagnosing the LD: Analysis of judgmental policies. Journal 92 Learang Disabili- Keith. T. Z. (1986). Factor structure of the K-ABC for referred school children. Psychology it; the Schools. Q. 241-246. Keith. T.. & Dunbar. S. (1984). Hierarchical factor analysis of the K-ABC: Testing alternate models. Journal _o_f Special Education. 1_8.(5). 367-375. Kirby. J. R.. 5 Das. J. P. (1977). Reading achievement. IQ. and simultaneous-successive processing. Journal _o_f Educational Psy- chology. _62(5). 564-570. Kirk. S. A.. & Elkins. J. (1975). Learning disabilities: Character- istics of children enrolled in the child service demonstration centers. Journal p_f_ Learnipg Disabilities. Q. 630-637. Klanderman. J.. Perney. J.. & Kroeschell. Z. (1983). Comparisons of the K-ABC and WISC-R for LD children. Personal Communication. Knopf. I. J. (1984). Childhood psychopathology; A developmental approach (2nd ed.). Englewood. Cliffs. NJ: Prentice-Hall. Lazarus. B. B. (1985). Practical reasoning and observation: 5 second- grade teacher refers children for special education services. Unpublished doctoral dissertation. Michigan State University. Lewandowski. N.. Saccuzzo. D.. & Lewandowski. D. (1977). The WISC as a measure of personality types. Journal o_f Clinical Psychology. §_§_(l). 285-291. 184 Lyon. R.. Stewart. N.. & Freedman. D. (1982). Neuropsychological characteristics of empirically derived subgroups of learning dis- abled readers. Journal g Clinical Neuropsycholggy. 3(4). 343- 365. Majovski. L. V. (1984). The K-ABC: Theory and applications for child neuropsychological assessment and research. Journal 92 Special Education. 239-250. Mann. L.. Davis. C.. Boyer. C.. Metz. C.. & Wolford. B. (1983). LB or not LD. that was the question: A retrospective analysis of child service demonstration centers' compliance with the federal defini- tion of learning disabilities. Journal o_f Learning Disabilities. 2(1). 14‘17. Matuszek. P.. 8: Oakland. T. (1979). Factors influencing teachers' and psychologists' recommendations regarding special class placement. Journal if. School Psycholggy. 11(2). 116-125. McCue. M.. Shelly. C.. & Goldstein. G. (1986). Intellectual. academic and neuropsychological performance levels in learning disabled adults. Journal 22 Learning Disabilities. l_9(4). 233-236. McKinney. J. (1984). The search for subtypes of specific learning disability. Journal _o_f_ Learnipg Disabilities. 21(1). 43-50. McNutt. G. (1986). The status of learning disabilities in the states: Consensus or controversy? Journal 22 Learning Disabilities. Meehl. P. E.. G: Rosen. A. (1955). Antecedent probability and the efficiency of psychometric signs. patterns. and cutting scores. Psychological Bulletin. 2. 194-216. Mehan. H. (1984). Institutional decision-making. In B. Rogoff & J. Lave (Eds.). Evegyday Cognition: Its development in social context (pp. 641-666). Cambridge: Harvard University Press. Mehrens. W. (1984). A critical analysis of the psychometric proper- ties of the K-ABC. Journal E. Special Education. 12(5). Mercer. C.. Forgnone. C.. & Wolking. W. (1976). Definitions of learn- ing disabilities used in the United States. Journal o_f Learning Disabilities. _9_. 376-386. Miller. M. (1980). On the attempt to find WISC-R profiles for learn- ing and reading disabilities (a response to Vance. Wallbrown. & Blaha). Journal p_f_ Learnig Disabilities. 12. 338-340. 185 Morrison. G.. MacMillan. D.. & Kavale. K. (1985). System identifica- tion of learning disabled children: Implications for research sampling. Learnipg Disability Quarterly. 2. 2-10. Morrow. H.. Powell. G. & Ely. D. (1976). Placement or placebo: Does additional information change special education placement deci- sions? Journal o_f School Psycholggy. l_4. 186-191. Naglieri. J. (1985). Use of the WISC-R and K-ABC with learning dis- abled. borderline mentally retarded and normal children. Psy- chology _i_n the Schools. 2_2_. 133-141. Naglieri. J. A.. Kamphaus. R. W.. & Kaufman. A. S. (1983). The Luria Das simultaneous-successive model applied to the WISC-R. Journal g Lsychoeducational Assessment. 1. 25-35. Nolan. D.. Hammeke. T.. G: Barkley. R. (1983). A comparison of the patterns of the neuropsychological performance in two groups of learning disabled children. Journal 92 Clinical Child Psycholgy. 22(1). 22-27. Norman. C. A.. & Zigmond. N. (1980). Characteristics of children labeled and served as learning disabled in school systems affili- ated with child service demonstration centers. Journal pf Learn- ipg Disabilities. _l_3. 542-547. Novick. M. R. (1981). Federal guidelines standards. American Psy- chologist. 29(10). 1035-1046. Obrzut. A.. Obrzut. J.. & Shaw. D. (1984). Construct validity of the Kaufman assessment battery for children with learning disabled and mentally retarded. chhology Q the Schools. 21. 417-424. Reid. D. R.. 5 Hresko. W. P. (1981). A. cognitive approach_t_o learning disabilities. New York: McGraw-Hill Book Company. Reschly. D. J. (1981). Psychological testing in educational classi- fication and placement. American Psycholoijst. §_6(lO). 1094-1102. Reynolds. C. (1981). A note on determining significant discrepancies among category scores on Bannatyne's regrouping of WISC-R sub- tests. Journal _c_>_§ LearnLEg Disabilities. l_4. 468-469. Reynolds. C. R. (1981). The fallacy of "two years below grade level for age" as a diagnostic criterion for reading disorders. Journal _02 School Psycholgy. _12(4). 350-358. Reynolds. C. R. (1982). The problem of bias in psychological assess- ment. In C. R. Reynolds 8 T. B. Gutkin (Eds.). The Handbook 92 School Pflcholgy (pp. 178-208). New York: John Wiley 8 Sons. 186 Rourke. B.. 8 Strang. J. (1978). Neuropsychological significance of variations in patterns of academic performance: Motor. psychomotor and tactile-perceptual abilities. Journal o_f_ Pediatric Psychology. 2(2). 62-66. Rudel. R. G. (1980). Learning disability diagnosis by exclusion and discrepancy. Journal 92 the American Academy -o_f_ Child Psychiatry. _}_9_. 547-569. Rugel. R. (1974). WISC subtest scores of disabled readers: A review with respect to Bannatyne's recategorization. Journal 92 Learning Disabilities. 1. 48-55. Ryckman. D. (1981). Searching for a WISC-R profile for learning disabled children: An inappropriate task? Journal pf Learning Disabilities. _l_4. 508-511. Saccuzzo. D.. 8 Lewandowski. D. (1976). The WISC-R as a diagnostic tool. Journal 92 Clinical Psychology. 22(1). 115-124. Salvia. J.. 8: Clark. G. (1973). The use of deficit scores to iden- tify learning disabled children. Exceptional Children. 3_7_. 305- 308. Sabatino. D. (1972). School psychology--special education: To ack- nowledge a relationship. Journal 2; School Psycholgy. ,1_0_. 99- 108. Sattler. J. M. (1982). Assessment pf children's intelligence and ppecial abilities (2nd ed. ). Boston: Allyn and Bacon. Sattler. J. M. (1980). Learning disabled children do not have a perceptual organization deficit: Comments on Dean WISC-R analy- sis. Journal 22 Consulting and Clinical ngchology. fl. 254-255. Schere. R. 11.. Richardson. E.. & Bialer. I. (1980). Toward operation- alizing a psychoeducational definition of learning disabilities. Journal _o_f_ Abnormal Child Psycholog. _8_. 5-20. Schooler. D.. Beebe. M.. & Koepke. T. (1978). Factor analysis of WISC-R scores for children identified as learning disabled. edu- cable mentally impaired and emotionally impaired. ngchology g: the School. _12. 478-485. Sexton. L.. & Street. S. (1985). A comparison of WISC-R score pro- files of students placed in special education and those referred but not placed. Psychology E the Schools. Q. 142-145. 187 Shepard. L. A.. Smith. M. L.. & Vojir. C. P. (1983). Characteristics of pupils identified as learning disabled. American Educational Research Journal. 22(3). 309-331. Shinn. M. R. (1986). Does anyone care what happens after the refer- test-place sequence: The systematic evaluation of special education program effectiveness. School Psychology Review. _l_5_(l). 49-580 Shinn. M. R.. Ysseldyke. J.. Deno. S.. 8: Tindal. G. (1982). A com- parison of psychometric and functional differences between stu- dents labeled LD and low achieving. Institute For Research 9p Learnig Disabilitiesy University 92 Minnesota. 7_1_. 1-35. Silverstein. A. B. (1982). Pattern analysis as simultaneous statis- tical inference. Journal 92 Consulting 2 Clinical Psycholgy. 22. 234-240. Smith. C.. 8 Knoff. F. (1981). School psychology and special educa- tion students' placement decisions: IQ still tips the scale. Journal 92 Special Education. g. 56-64. Smith. M.. Coleman. M.. Dokecki. P.. 8 Davis. E. (1977). Recate- gorized WISC-R scores of learning disabled children. Journal 92 Learnig Disabilities. 22(7). 48-54. Smith. R.. Osborne. L.. Crim. D.. & Rhw. A. (1986). Labeling theory as applied to learning disabilities: Survey findings and policy suggestions. Journal o_f Learning Disabilities. _l2(4). 195-202. Snyder. T. J.. Leark. R. A.. Golden. C. J.. Grove. T.. & Allison. R. (1983. March). Correlation if; the K-ABC. WISC-R and Luria- Nebraska Children's Battegy for exceptional children. Paper pre- sented at the annual meeting of the National Association of School Psychologists. Detroit. Michigan. Spreen. 0.. & Hoof. R. (1986). Empirically derived learning disabil- ity subtypes: A replication attempt and longitudinal patterns over 15 years. Journal 92 Learni_pg Disabilities. 22(3). 170-180. Sternberg. R. J. (1981). Nothing fails like success: The search for intelligent paradigm for studying intelligence. Journal 92 Educa- tional chhology. _7_3_(2). 142-155. Sternberg. R. J. (1984). The Kaufman assessment battery for children: An information-processing analysis and critique. Journal o_f Special Education. 251-261. Sternberg. R. J. (1981). Testing and cognitive psychology. American Psychologis . ‘3_6(10). 1181-1189. 188 Taylor. H. G.. Satz. P.. & Friel. J. (1979). Developmental dyslexia in relation to other childhood reading disorders: Significance and clinical utility. Readigg Research Quarterly. _l_5_(1). 84-101. Thompson. R. (1981). The diagnostic utility of Bannatyne's recate- gorized WISC-R scores with children referred to a developmental evaluation center. Psychology 2.2 the Schools. 22. 43-47. Thurlow. M. L.. & Ysseldyke. J. E. (1979). Current assessment and decision-making practices in model LD Programs. Learning Dis- abilities Quarterly. 2. 15-24. Tucker. T. (1980). Ethnic proportions in classes for the learning disabled. Journal 92 Smial Education. 22(1). 93-105. Tucker. J.. Stevens. L.. & Ysseldyke. J. (1983). Learning disabili- ties: The experts speak out. Journal 22 Learning Disabilities. 19(1). 6-300 Vance. H.. Wallbrown. P.. & Blaha. J. (1978). Determining WISC-R profiles for reading disabled children. Journal 92 Learning Dis- abilities. 22. 657-661. Veres. E. (1982). Q evaluation 92 the p_s_e o_f the WISC-R Subtest scores for the identification 92 mild2y handicapped students. Ph.D. dissertation. Michigan State University. Vogel. S. (1986). Levels and patterns of intellectual functioning among LD college students: Clinical and educational implications. Journal 92 Learning Disabilities. 12(2). 71-79. Warner. M. M.. Schumaker. J. B.. Alley. G. B.. & Deshler. D. D. (1980). Learning disabled adolescents in the public schools: Are they different from other low achievers? Excgptional Educatm Qgrterly. 2. 27-36. Wechsler. D. (1974). Wechsler Intelligence Scale for Children-- Revised. New York: The Psychological Corporation. Wiseman. W. (1984). Differences between learnipg disabled and low- achievirg students _op standardized test performance. Unpublished doctoral dissertation. Michigan State University. Ysseldyke. J. E.. & Algozzine. B. (1979). Perspectives on assessment of learning disabled students. Learning Disability Quarterly. 2. 3-130 189 Ysseldyke. J.. Algozzine. B.. Richey. L.. 5: Graden. J. (1982). Declaring students eligible for learning disability services: Why bother with the data? Learnipg Disability Quarterly. 2. 37-43. Ysseldyke. J. E.. Algozzine. B.. Shinn. M. R.. & McGue. M. (1982). Similarities and differences between low achievers & students classified L.D.. Journal 22 Special Education. _l_6(l). 73-85. Ysseldyke. J. E.. & Foster. G. G. (1978. May). Bias in teachers' observations of emotionally disturbed and learning disabled chil- dren. Exceptional Children. 613-615. Zins. J. E.. & Barnett. D. W. (1983). The Kaufman Assessment Battery for children and school achievement: a validity study. Journal 92 gychoeducational Assessment. 2.- 235-241.