@prefix vivo: . @prefix edm: . @prefix ns0: . @prefix dcterms: . @prefix skos: . vivo:departmentOrSchool "Arts, Faculty of"@en, "Psychology, Department of"@en ; edm:dataProvider "DSpace"@en ; ns0:degreeCampus "UBCV"@en ; dcterms:creator "Powell, Joan Anne"@en ; dcterms:issued "2012-03-16T21:28:30Z"@en, "1951"@en ; vivo:relatedDegree "Master of Arts - MA"@en ; ns0:degreeGrantor "University of British Columbia"@en ; dcterms:description "This study was designed to investigate the relationship between the intelligence quotients yielded by two widely used individual tests of intelligence for children, namely, the Stanford-Binet, Form L, (1937 Revision) and the Wechsler Intelligence Scale for Children (WISC). Until recently, the Binet has been used almost exclusively to ascertain the intelligence of the school-age child but, with the publication of the WISC in 1949, there has been an increasing trend toward using the tests either interchangeably or in conjunction with one another. In view of this development, an attempt to discover the relationship between the two scales would seem to be of much practical value. Although the two scales agree in assuming a \"g\" factor of intelligence, they differ as to the nature of their content and construction. The Stanford-Binet does not include any test items designated as measuring a particular skill, whereas the WISC is composed of twelve subtests, each supposed to tap a specific ability, and it yields a separate verbal and performance intelligence quotient. The two scales also differ in the manner of computing an intelligence quotient; the Binet scale depends upon a Mental Age concept of intelligence, whereas the Wechsler Intelligence Scale for Children is a point scale. This study attempted to ascertain to what extent the Stanford-Binet correlates with each of the WISC scales (Verbal, Performance and Full Scale) at three different age levels and three levels of intelligence, i.e., with subjects of Superior, Average and Retarded intelligence. It also attempted to find out what differences, if any, might occur between the Mean intelligence quotients yielded by the two tests in the above age and intellectual categories, and what direction these differences might take. Wechsler has objected to the Stanford-Binet deviations, which vary in size at difference age levels. At 6 years, the Binet standard deviation is unusually small, and at 12 years of age it is unusually large: the WISC standard deviations are the same size at each age level. It was hypothesized, therefore, that at the extremes of the intelligence distribution at ages 6 and 12 years, there should be differences between the Mean intelligence quotients yielded by the two tests in the direction of the size of the Binet standard deviations at these two age levels – a smaller Mean Binet than Mean WISC intelligence quotient at age 6 years, with a higher Mean intelligence quotient on the Binet at age 12 years. Subjects of these two ages, 6 and 12 years, were included in the experimental group in order to test this hypothesis, while the use of subjects of superior and retarded intelligence insured that extreme scores would occur. The sample of subjects of average intelligence, plus a group of 9-year-olds, were included for control and comparison in testing this hypothesis but also for their own research value. The sample was composed of 85 subjects - ten children in each age category of the Superior and Average intelligence groups; and in the Defective group, fourteen 12-year-olds, nine 9-year-olds, and four 6-year-olds. The positive correlations which occurred may be summarized as follows : 1. In the 9-year-old Superior group, the Stanford-Binet IQ correlated significantly with - (a) the WISC Verbal IQ at the 1% level of confidence; (b) the WISC Performance Scale IQ at the 5% level of confidence; (c) the WISC Full Scale IQ at the 1% level of confidence. 2. In the 9-year-old Average group, the Stanford-Binet IQ correlated significantly with - (a) the WISC Verbal Scale IQ at the 1% level of confidence; (b) the WISC Full Scale at the 5% level of confidence. Significant differences between the Mean IQs of the two tests may be summarized as follows : 1. In the group of Superior 9-year-olds, the Stanford-Binet IQs were significantly higher (at the 1% level of confidence) than the WISC Verbal, Performance, and Full Scale IQs. 2. In the group of Superior 12-year-olds, the Stanford-Binet IQs were significantly higher at the 1% level of confidence for the WISC Full and Verbal Scale IQs. 3- In the group of Average 12-year-olds, the Stanford-Binet is significantly higher at the 5% level of confidence than the WISC Verbal IQ. The major conclusions of this study are : 1. The obtained results are in essential agreement with the studies comparing the Wechsler adult scale and the Stanford-Binet. 2. There seems to be a consistent tendency in this study and others reviewed previously for lower correlations between the Stanford-Binet and WISC Performance Scale, than between the Stanford-Binet and WISC Verbal and Performance Scales. 3. There seems to be no support for the hypothesis that the difference between the Mean Stanford-Binet and the Mean WISC IQs at the Superior level will differ in direction according to the size of the Binet standard deviation at the age level in question. 4. The WISC appears to be an unsatisfactory test for measuring the markedly retarded children. Both in terms of construction, and interest value to subjects, the Stanford-Binet seems to be a better scale for the measurement of the lower levels of intelligence. 5. Keeping in mind the limited sample upon which this study was based, the two scales do not seem to be interchangeable. The practical import of this conclusion is that clinicians, social workers, psychiatrists, school teachers, and so on, should be fully aware that the child given both the tests may well yield widely different IQs on the respective tests. Suggestions for future research have been included."@en ; edm:aggregatedCHO "https://circle.library.ubc.ca/rest/handle/2429/41514?expand=metadata"@en ; skos:note "/fa A COMPARISON OF THE STANFCRD-BINET (1937 REVISION, FORM L) AND WECHSLER INTELLIGENCE SCALE FOR CHILDREN AT DIFFERENT AGE AND INTELLECTUAL LEVELS by JOAN ANNE POWELL A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS i n the Department of PHILOSOPHY AND PSYCHOLOGY We accept this thesis as conforming to the standard required from candidates for the degree of MASTER OF ARTS Members of the Department of Philosophy and Psychology. THE UNIVERSITY OF BRITISH COLUMBIA August, 1951 A COMPARISON OF THE STANFORD-BINET (1937 REVISION, FORM L) AND WECHSLER INTELLIGENCE SCALE FOR CHILDREN AT DIFFERENT AGE AND INTELLECTUAL LEVELS Abstract This study was designed to investigate the relationship between the intelligence quotients yielded by two widely used individ-ual tests of intelligence for children, namely, the Stanford-Binet, Form L, (1937 Revision) and the Wechsler Intelligence Scale for Children (WISC). Until recently, the Binet has been used almost exclusively to ascertain the intelligence of the school-age child but, with the publication of the WISC in 1949, there has been an increasing trend toward using the tests either interchangeably or i n conjunction with one another. In view of this development, an attempt to discover the relationship between the two scales would seem to be of much practical value. Although the two scales agree i n assuming a \"g\" factor of intelligence, they differ as to the nature of their content and construction. The Stanford-Binet does not include any test items designated as measuring a particular s k i l l , whereas the WISC i s com-posed of twelve subtests, each supposed to tap a specific a b i l i t y , and i t yields a separate verbal and performance intelligence quotient. The two scales also dif f e r i n the manner of computing an intelligence quotient; the Binet scale depends upon a Mental Age concept of intelligence, whereas the Wechsler Intelligence Scale for Children i s a point scale. This study attempted to ascertain to what extent the Stanford-Binet correlates with each of the WISC scales (Verbal, Performance and F u l l Scale) at three different age levels and three levels of intelligence, i.e., with subjects of Superior, Average and Retarded intelligence. It also attempted to find out what differences, i f any, might occur between the Mean intelligence quotients yielded by the two tests i n the above age and intellectual categories, and what direction these differences might take. Wechsler has objected to the Stanford-Binet deviations, which vary i n size at difference age levels. At 6 years, the Binet standard deviation i s unusually small, and at 12 years of age i t i s unusually large: the WISC standard deviations are the same size at each age le v e l . It was hypothesized, therefore, that at the extremes of the intelligence distribution at ages 6 and 12 years, there should be differences between the Mean intelligence quotients yielded by the two tests in the direction of the size of the Binet standard deviations at these two age levels - a smaller Mean Binet than Mean WISC intelligence quotient at age 6 years, with a higher Mean intelligence quotient on the Binet at age 12 years. Subjects of these two ages, 6 and 12 years, were included i n the experimental group i n order to test this hypothesis, while the use of subjects of superior and retarded intelligence insured that extreme scores would occur. The sample of subjects of average intelligence, plus a group of 9-year-olds, were included for control and comparison i n testing this hypothesis but also for their own research value. The sample was composed of 85 subjects - ten children i n each age category of the Superior and Average intelligence groups; and in the Defective group, fourteen 12-year-olds, nine 9-year-olds, and four 6-year-olds. The positive correlations which occurred may be summarized as follows : 1. In the 9-year-old Superior group, the Stanford-Binet IQ correlated significantly with -(a) the WISC Verbal IQ at the 1% level of confidence; (b) the WISC Performance Scale IQ at the 5% level of confidence; (c) the WISC F u l l Scale IQ at the 1% level of confid-ence. 2. In the 9-year-old Average group, the Stanford-Binet IQ correlated significantly with -(a) the WISC. Verbal Scale IQ at the 1% level of confidence; (b) the WISC Full Scale at the 5% level of confidence. Significant differences between the Mean 10s of the two tests may be summarized as follows : 1. In the group of Superior 9-year-olds, the Stanford-Binet IQs were significantly higher (at the 1$ lev e l of confidence) than the WISC Verbal, Performance, and Fu l l Scale IQs. 2. In the.group of Superior 12-year-olds, the Stanford-Binet IQs were significantly higher at the 1$ level of confidence for the WISC F u l l and Verbal Scale IQs. 3- In the group of Average 12-year-olds, the Stanford-Binet i s significantly higher at the 5$ level of confidence than the WISC Verbal IQ. The major conclusions of this study are : 1. The obtained results are i n essential agreement with the studies comparing the Wechsler adult scale and the Stanford-Binet. 2. There seems to be a consistent tendency in this study and others reviewed previously for lower correlations between the Stanford-Binet and WISC Performance Scale, than between the Stanford-Binet and WISC Verbal and Performance Scales. 3. There seems to be no support for the hypothesis that the difference between the Mean Stanford-Binet and the Mean WISC IQs at the Superior level w i l l d i f f e r i n direction according to the size^ of the Binet standard deviation at the age lev e l i n question. 4. The WISC appears to be an unsatisfactory test for measuring the markedly retarded children. Both i n terms of construction, and interest value to subjects, the Stanford-Binet seems to be a better scale for the measurement of the lower levels of intelligence. 5. Keeping in mind the limited sample upon which this study was based, the two scales do not seem to be interchangeable. The practical import of this conclusion i s that clinicians, social workers, psychiatrists, school teachers, and so on, should be f u l l y aware that the child given both the tests may well yield widely different IQs on the respect-ive tests. Suggestions for future research have been included. C O N T E N T S CHAPTER PAGE I INTRODUCTION 1 Statement of the Problem 1 Historical Background 1 Review of the Literature 3 II PROCEDURE 13 Selection of Subjects 13 Administration of the Intelligence Scales 16 III TREATMENT OF DATA AND RESULTS 22 Treatment of Data 22 Results : -(a) Obtained Correlations 22 between Scales (b) Comparison of Mean IQs 28 (c) The Performance of the 4 1 Defective Group IV DISCUSSION OF RESULTS 4 7 Theoretical Expectations and Obtained 47 Results Obtained Results compared with the 56 Literature Consistent Trends emerging from the Data 58 Suggestions for Future Research 58 V SUMMARY 60 BIBLIOGRAPHY 64 T A B L E S Original Plan for selection of Sample Obtained Sample represented as to Age and Intelligence Level Groups Correlations between Stanford-Binet and WISC Verbal, Performance, and F u l l Scales Correlations between Stanford-Binet and WISC Verbal, Performance, and Full Scales Mean WISC Verbal Scale IQs compared with Mean Stanford-Binet IQs Mean WISC Performance Scale IQs compared with Mean Stanford-Binet IQs Mean WISC F u l l Scale IQs compared with Mean Stanford-Binet IQs Mean WISC Verbal Scale IQs compared with Mean Stanford-Binet IQs Mean WISC Performance Scale IQs compared with Mean Stanford-Binet IQs Mean WISC F u l l Scale IQs compared with Mean Stanford-Binet IQs Performance of 12-year-old Mental Defectives on WISC Verbal, Performance, and F u l l Scales, and on Stanford-Binet Performance of 9-year-old Mental Defectives on WISC Verbal, Performance, and F u l l Scales, and on Stanford-Binet Performance of 6-year-old Mental Defectives on WISC Verbal, Performance, and F u l l Scales, and on Stanford-Binet ACKNOWLEDGEMENTS The author i s most grateful to her Faculty-adviser, Mrs. D. T. Kenny, for her continued encouragement and many helpful suggestions and criticisms. The author i s indebted to Dr. R. F. Sharpe, Acting Assistant Superintendent and Inspector of Secondary Schools, for permission to test i n the Vancouver Elementary Schools. She i s also grate-f u l to Dr. L. L. Sauriol, Superintendent of The Woodlands School, New Westminster, B.C., and to Dr. T. D. Barber, Superintendent of The Rainier State School, Buckley, Washington, U.S.A., for making available the retarded subjects used i n this study. CHAPTER I INTRODUCTION Statement of the Problem The purpose of this study i s to investigate the relat-ionship between the intelligence quotients yielded by two widely used standard intelligence scales, namely the Stanford-Binet Form L (1937 Revision), and the Wechsler Intelligence Scale for Children (WISC). Until recently, aside from group intelligence tests, the Binet scale has been used almost exclusively to ascertain the intelligence of the school-age child. With the emergence of the WISC i n 1949, there has been an increasing shift toward the use of this scale either i n con-junction with, or in place of, the Binet scale. In view of this trend, a comparison of the two scales, such as i s proposed i n this study, i s of much practical value. That i s , i f the tests are to be used inter-changeably, some attempt to demonstrate the relationship between the two scales i s urgently needed. Historical Background Although the authors of both of these scales follow 2 Spearman i n assuming a global factor of intelligence, the two scales are quite different i n the nature of their construction. No attempt i s made i n the Binet scale to group the items in terms of the kinds of s k i l l s involved, whereas the Wechsler Intelligence Scale for Child-ren i s composed of twelve subtests, each of which involves a particu-l a r s k i l l , and the scale yields a separate Verbal and Performance intelligence quotient. In this respect, the Wechsler scale i s some-times thought to lean toward Thurstone's approach to intelligence i n terms of special a b i l i t i e s . Wechsler himself, however, i s careful to point out that he i s primarily interested i n arriving at a general measure of intelligence and that the subtests are not \"...a series of tests that measure primary a b i l i t i e s , \" (p.5, 20). Besides differing as to the type of material and i t s arrangement, the two tests show sharp differences i n the method of arriving at the intelligence quotient. The Binet scale depends on a Mental Age concept of intelligence, whereas the Wechsler scale f a l l s into the category of a point scale. Wechsler has voiced strong objections to the Stanford-Binet' = IQ formula for several reasons, CA to be discussed later i n this paper. Both the adult, and the more recent Wechsler Intelligence Scale for Children, yield an intelligence quotient resulting from a comparison of the subject's performance with that of others of his own age, rather than the equating of the perform-ance with a certain Mental Age leve l . Wechsler j u s t i f i e s this move on several grounds. F i r s t l y , he points out that a Mental Age of % 3 years i s not at a l l the same intellectual capacity i n a 10-year old or a 5-year old as i t i s i n a 7-year old, as i s implied by the construction of the Binet scale. Secondly, he questions the assumption, innate i n the Stanford-Binet, that intellectual growth stops at 16 years. His method of computing the IQ makes any such assumption unnecessary; i t also makes allowances for the slowed-up intellectual development of puberty and the early teens, and for the decrease i n intellectual a b i l i t y which occurs with age. Whether or not this method, which i s more lenient to older subjects, i s better depends upon the use to which the score w i l l be put. Thirdly, Wechsler's chief objection to the Stanford-Binet i s to the size of the standard deviations at different age-levels. The greatest deviations occur at age 12 years (CTLIQ = 20.00), and at 6 years (CTLIQ = 12.50). He points out (p.26, 19) that a child 2 sigma away from the Mean at age 6 years w i l l have an IQ of 75, whereas a child similarly placed at 12 years w i l l have an IQ of 60. Hence the scale yields a rather unreliable estimate of the subject's intelligence. Review of the Literature Early attempts to investigate the relationship between intelligence quotients derived from Mental Ages and those derived from point scales have been based on comparisons of the Stanford-Binet with the Wechsler-Bellevue Adult Intelligence Scales. Since the Wechsler Intelligence Scale for Children, used i n this study, has grown out of 4 and closely parallels the Adult Scale, a review of the above mentioned studies seems i n order at this point. A number of these studies were done with adolescents since i t i s i n this age range that both scales are widely used. Sartain (13) compared IQs obtained on the Stanford-Binet with those obtained on the Wechsler-Bellevue by f i f t y adolescent college freshmen. He found that the Stanford-Binet correlated with the F u l l Scale, .774; with the Verbal Scale, .802; and with the Performance Scale, .510. He estimates that the Stanford-Binet yields an IQ approximately 5 points higher than does the Wechsler-Bellevue; The Mean Binet IQ was 129-44, and the Mean Wechsler IQ was 117.44 with a c r i t i c a l ratio of 5«55-No definite conclusions are stated by the author and the results are ambiguous because one cannot be sure which of the variables, (a) adoles-cence, (b) superior intelligence, or, (c) both, accounts for the results. Goldfarb (6) used a sample of sixty adolescents i n foster homes, with Mean Age equal to 14*6 years. The Mean Verbal IQ was highest for 62$ of the subjects. The F u l l Scale Wechsler correlat-ed with the Stanford-Binet to produce a coefficient of .86; the Verbal, a coefficient of .80; and the Performance, a coefficient of .67. The author claims that bright subjects test higher on the Revised Stanford-Binet than on the Wechsler, while the advantages are reversed for the dull subjects. The author does not qualify what he means by \"bright\" and \" d u l l \" . He further states that younger children get higher 5 Stanford-Binet IQs and older subjects higher Wechsler IQs, but again does not specify as to what these categories encompass. He also maintains that the Wechsler does not discriminate very well amongst groups of superior adolescents. The fact that these subjects were i n foster homes does not appear to have produced results different from those that might have been expected from a comparable sample not having this characteristic. Guertin (7) refers to an unpublished study of his own with a sample of feebleminded subjects with IQs between 51 and 75, and between the ages of 15 and 22 years. He found that the Wechsler-Bellevue EQ for these subjects was 5.33 points above the Binet IQ, although he does not state whether or not this difference was significant. From these studies i t appears that adolescents of superior intelligence achieve higher IQs on the Stanford-Binet, and that du l l adolescents achieve higher IQs on the Wechsler. Younger adoles-cents seem to get higher Binet than Wechsler IQs, and older adolescents seem to get higher Wechsler IQs. Kutash (10) used a sample composed of f i f t y adult mental defectives. He found that (a) the Wechsler-Bellevue yielded higher IQs i n BU% of the cases, (b) the Mean Wechsler IQ was 11 points higher than the Mean Stanford-Binet IQ, the difference significant at the 1$ lev e l of confidence, (c) the size of the difference in IQs varies directly with the chronological age of the subject, and (d) the two 6 scales correlated, r = .77' He concludes that differences i n the IQs obtained on the two scales are due to differences i n norms and principles of standardization. The standardization and norms of the Wechsler take into account the normal deterioration with age which the Binet does not do. Rabin (12) and Guertin (7) point out that the results of this study may not be valid owing to the lack of homogeneity as to age i n the group selected. Halpern (9) conducted a study using one-hundred-and-thirty-three patients at a mental hygiene c l i n i c . The subjects were a l l classified as having d u l l normal intelligence, and were divided into four age groups. For ages 15 to 34 years, the tests yielded similar measures, although Halpern prefers the Wechsler for several technical reasons. I t was found that from ages 10 to 14 years, higher IQs are consistently yielded by the Stanford-Binet. Halpern believes that for subjects who are 13 years and younger, and for 14-year-olds who are of low intelligence, the Binet i s the preferable instrument. He suggests that the Binet IQs are too high and that the Wechsler IQs are too low. When he took a group of subjects and divid-ed them according to intelligence, the highest correlations were at. the extremes. In spite of this fact, the greatest differences occurred with the superior subjects, and the least with the retarded subjects. He would postulate that \"...both scales tap the defective's limited capacity equally well, but that the range of the superior subject's i s reached equally by a l l tools,\" (p.210, 9). Because the Wechsler norms 7 take the deterioration of intellectual a b i l i t y with age into account, he says that differences between the IQs i n those over 34 years are general-l y accountable for i n terms of differences i n test construction. He believes that the introduction of the non-verbal tasks i n the Wechsler-Bellevue makes for further test differences. Balinskyj Israel and Wechsler (2) tested the relative effectiveness of the Wechsler-Bellevue and the Stanford-Binet i n diagnosing mental deficiency. The criterion was psychiatric diagnosis. The best predictions were made i n this order : the Wechsler F u l l Scale, the Wechsler Verbal Scale, the Stanford-Binet, and, last, the Wechsler Performance Scale. Using b i - s e r r i a l r's, the authors computed the forecasting a b i l i t y of the Wechsler to be 4C#, and that of the Binet to be 5%. The sample was of patients i n the Bellevue Hospital. Since the Wechsler-Bellevue was validated c l i n i c a l l y i n the Bellevue Hospital (p.127, 19), and, presumably, according to the standards of that institution, and since this study was carried out i n the same institution, i t i s not surprising that Wechsler scores of a subsequent sample should agree more closely with those standards than scores on a test validated on other c r i t e r i a . In summary, i t would appear that retarded subjects attain higher IQs on the Wechsler-Bellevue than on the Stanford-Binet. This could result from the fact that, as Wechsler points out (p.157, 19), the feebleminded generally do better on the Performance than on the Verbal Scale. Differences would be expected at this level of intelligence, 8 since, of a l l the three Wechsler scales, the Performance Scale correlates lowest with the Binet, and a higher IQ would be expected because a better Performance index would naturally tend to raise the F u l l Scale IQ. Benton, Weider and Bleauvelt (3) studied a sample of sixty subjects described as \"mental cases\" who were cooperative. The subjects ranged i n age from 16 to 59 years, with a Mean Age of 35 years. The correlations were a l l high, although, as usual, the Performance Scale showed the lowest agreement. The coefficients were as follows : Fu l l Scale, .93; Verbal Scale, .92; and Performance Scale, .73. A l -though the Mean scores are similar, the distributions d i f f e r greatly , the standard deviation of the Stanford-Binet far exceeding that of the Wechsler. The authors feel that these disparities do not indicate a real difference i n what the tests measure because widely diverse scores on the Binet and Wechsler may indicate similar positions i n relation to the Means. This becomes obvious when the high correlations are remem-bered. The difference i s , again, due to the Wechsler norms; the authors express the opinion that the scores would be comparable i f con-verted into percentile ranks or standard scores. This study must be regarded as inconclusive since the nature of the sample i s so i l l -defined. Mitchell (11) used a sample of two-hundred-and-sixty-eight subjects, mostly delinquents and chronic alcoholics without psychosis, and cooperative psychotics. He found that a l l three parts of the Wechsler correlated significantly with the Binet; tx>renty-one 9 of his subjects showed a difference of more than 20 IQ points between their scores on the two scales; of these, sixteen showed a higher IQ on the Wechsler. The ages of these subjects (M c 48.2 years) was nearly double that of the five subjects who scored 20 points higher on the Stanford-Binet. Since the sample i s heterogeneous, any characteristics which might have accrued to one c l i n i c a l group may have been obscured by the opposite tendencies i n another group; no just i f i c a t i o n i s given by the author for the sample selected. On the basis of these studies, i t appears that superior subjects do better on the Stanford-Binet than on the Wechsler, and that retarded subjects do better on the Wechsler than on the Binet. Up to the early teens, higher scores are found on the Binet, and, after this age lev e l , higher scores are found on the Wechsler. In a l l studies, the Wechsler Performance Scale correlates lower with the Stanford-Binet than do the Verbal or F u l l Scales. With the exception of those studies involving adolescents, a l l these comparisons are at a disadvan-tage. They are attempting to compare IQs on two tests, one of which was standardized on adults and the other on children. Literature on the Wechsler Intelligence Scale for Child-ren has been scant to date. Of the four articles published on i t , only two bear on this investigation. The f i r s t , by Seashore, Wesman and Doppelt (14), i s concerned chiefly with describing the standardization of the WISC, a matter which i s not of great interest here. The authors do point out, however, that i t i s unlikely that there w i l l be very 10 extreme scores because the range of IQs on a point scale such as the WISC i s quite arbitrary. I f the standard deviation had been set at 20 IQ points rather than the 15 points actually used, extreme scores might be expected. The only other study that i s related to this investigat-ion was that done by Frandsen and Higginson (5), who compared IQs on the WISC with those on the Stanford-Binet. The scales were also tested for v a l i d i t y in predicting school success. The sample included fifty-four Fourth Grade children between the ages of 9-years-and-l-month and 10-years-and-3-monthsj the subjects were of average a b i l i t y and average achievement. The F u l l Scale WISC and Stanford-Binet correlat-ed to yield a coefficient, r = .80, and the Verbal and Performance Scales yielded coefficients of .71 and .76 respectively. The val i d i t y of the Binet for predicting school success as measured by the Standard Achievement Test was found to be .63, and that of the WISC to be .76. Only in the Language Achievement subtest did the Stanford-Binet predict better than the WISC. Unlike the studies reviewed comparing the Stanford-Binet and Wechsler-Bellevue, this study by Frandsen and Higginson answers l i t t l e concerning test differences i n any except the average i n t e l l e c t -ual group. As indicated by the literature on Wechsler-Bellevue and Stanford-Binet studies, large differences do not occur i n the groups of average subjects, but do occur with the superior subjects and with the retarded ones, and with the older subjects and with the younger 11 ones. On the basis of these same studies, for instance, one might expect that retarded subjects would achieve higher WISC than Binet IQs, and that the superior subjects would have lower WISC than Binet IQs. In the present investigation, an attempt w i l l be made to compare the performance of subjects of superior and defective intelligence, as well as those of average intelligence. In the selection of age levels to be used, this study-has an advantage over the studies comparing the Wechsler-Bellevue with the Stanford-Binet, i n that both of the present tests have been standardized on children. The WISC i s a point scale with a set standard deviation, while the size of the Stanford-Binet standard deviation varies, being unusually small at age 6 years and unusually large at age 12 years. As was discussed earlier, Wechsler argues that this characteristic of the Binet produces unreliable results, since subjects at the top and bottom parts of the normal curve show a different IQ at different age levels. At the extremes of the normal curve, i.e., with the retarded and the superior subjects at these ages, therefore, there should be a Mean difference between the IQs produced by the two scales. Since the WISC has a set standard deviation, the same at each age level , the Mean Differences should be in the direction corresponding to the size of the Binet deviation at that l e v e l . These ages, 6 and 12 years, w i l l be included i n the sample, therefore, i n order to test this hypothesis. The inclusion of samples of superior and retarded 12 children w i l l insure that extreme scores (i.e., at either end of the normal curve) w i l l occur. This w i l l give information omitted by Frandsen and Higginson, for, although they report high correlations, they do not say whether the scores are comparable as to size. It w i l l also give information as to whether the high agreement they report also holds for other age levels. Besides the 6 and 12-year-olds, the sample w i l l include a group of 9-year-olds, since the Binet standard deviation at age 9 i s not only near the average for that test ( i t i s equal to 16.4 IQ points and the average i s 16 points) but i t also approximates the WISC deviat-ion of 15 IQ points. The 9-year-olds, plus a sampling of average children, will, be included for their own research value, but mainly for purposes of control and comparison. In brief, this investigation i s planned to find out to what extent the three indices of intelligence yielded by the WISC cn the Fu l l , Verbal and Performance Scales, correlate with those yielded by the Stanford-Binet at ages 6, 9 and 12 years, using superior, average and defective children. It w i l l endeavor to ascer-tain whether the Mean scores i n these groups show real differences i n average IQs yielded by the two tests, and, i f so, i n what direction these differences do occur. CHAPTER II PROCEDURE In order to determine the extent to which intelligence quotients by the Binet scale (1937, Revised edition) and the Wechsler Intelligence Scale for Children are comparable, the following method of investigation was employed. Selection of Subjects The study was planned so that the sample should include t h i r t y children of superior intelligence, t h i r t y of average i n t e l l i -gence, and t h i r t y retarded children. Each group of t h i r t y would include ten children aged 12 years, ten aged 9 years, and ten aged 6 years. Thus, the sample should include t h i r t y children at each age lev e l . In this way, varying intelligence and varying age levels were represented i n the sample used i n this study. The t o t a l sample should equal ninety. The planned selection i s represented i n Table I, (p- 14). The 9 and 12-year-old Superior subjects were obtained from various elementary schools throughout Vancouver. They were 1 4 TABLE I ORIGINAL PLAN FOR SELECTION OF SAMPLE N = 9 0 Age Intellectual Level Superior Average Defective Total 6 years N = 1 0 N = 1 0 N = 1 0 = 3 0 9 years N = 1 0 N = 1 0 N - 1 0 = 3 0 1 2 years N = 1 0 N • 1 0 N = 1 0 = 3 0 Total = 3 0 = 3 0 = 3 0 = 9 0 15 selected on the basis of scores on either the Detroit Fi r s t Grade Exam-ination, or Otis Score, or scores on the National Intelligence Scale, or on a l l of them, depending upon what tests the child had taken. The sample does not include any subject whose academic standing i s at variance with the level of a b i l i t y indicated by the intelligence tests, nor does i t include any subjects for whom the results of different tests were equivocal. Some of the 6-year-old Superior subjects were selected on the basis of teachers' reports: these evaluations proved unreliable, however, and additional subjects had to be tested i n order to obtain the required number of subjects at this l e v e l . The subjects i n the Average category were selected on much the same basis as the Superior group, that i s , scores on previous intelligence tests, grades and marks and class standing. No d i f f i c u l -ty was encountered i n obtaining the required subjects i n this category. The sample of feebleminded subjects was obtained from the Woodlands School, New Westminster, B.C., and the Rainier State School, Buckley, Washington, U.S.A. Besides f u l f i l l i n g the age requirements, these children had to be defectives classified as \"familial\", and an attempt was made to use only the brighter children within the \"moron\" or \"borderline\" classification. However, several problems presented themselves i n the work at this l e v e l . In the f i r s t place, neither of the two schools had a sufficient number of children at these age levels who were classified as \"familial 1', and, i n the second place, of those obtainable few were su f f i c i e n t l y 16 intelligent to meet the requirements of the study. These requirements were set up because Wechsler included i n the retarded sample of his standardization group \"...cases of the required ages who were rated as having IQs under 70 and not below 50,\" (p.8, 20). The th i r d and largest problem occurred with the 6-year-old Retarded group. Except i n unusual instances, such as cases of physical damage and mongolism, children are not admitted to the Woodlands School u n t i l they are 6 years of age. Few 6-year-olds were obtainable, therefore^ and those who were showed very low intelligence. The reason for the great d i f f i c u l t y i n obtaining subjects at this level may be i n the ages selected. In other words, early committal to an institution may, generally speaking, be confined to cases of physical d i s a b i l i t y , mongolism and very great retardation. For these reasons, the sample of Retarded children has been curtailed, especially i n the 6-year-old group. The obtained sample used i n this study i s indicated i n Table II, (p. 17)-Administration of the Intelligence Scales A l l the testing was done by the investigator who had been trained in the administration and scoring of both of the scales used i n this study. Every effort was made to follow the standard procedures recommended in the test manuals. The records were scored f i r s t by the examiner and then the scoring was checked by a trained worker in the f i e l d . 17 TABLE II OBTAINED SAMPLE REPRESENTED AS TO AGE AND INTELLIGENCE LEVEL GROUPS N = 85 Age Intelligence Level Superior Average Defective Total 6 years N = 10 N = 10 N = 4 = 24 9 years N = 10 N = 10 N = 9 =29 12 years N = 10 N = 10 N = 12 =32 Total = 30 = 30 = 25 = 85 18 Half the subjects i n each age-intellectual group were given the Stanford-Binet f i r s t , and the other half were given the WISC f i r s t , i n order to nrinimize and even out the effects of practice. The second test was given within a week, but never immediately after the f i r s t i n order to avoid fatiguing the subject. The two tests were given as nearly as possible under the same conditions, and care was taken to prevent the testing running through any preferred a c t i v i t y of the child, e.g., recess or a school concert, or when the children were i n a state of excitement, for instance, before Sports' Day. In general, a high level of cooperation was obtained from the subjects. The school children, especially, participated most eagerly. In the case of the 6-year-olds, the examiner was introduced to the class and i t was then explained that some of the children were to \"play some games\" with her. In this way, the testing became a most desirable and pleasurable thing to these youngsters. I n i t i a l l y , these children were asked not to t e l l their friends the contents of the tests. However, this seemed to impart to the whole a f f a i r the atmosphere of a \"secret\" which the youngsters found most d i f f i c u l t to keep, so that after a while this practice was dropped. On the whole, whether or not the subjects were asked not to t e l l , very l i t t l e evidence was shown of subjects knowing the test material beforehand. Some of the subjects were asked to name the test they 19 preferred, but, i n general, there did not seem to be a preference for either test. More d i f f i c u l t y was encountered i n establishing rapport with the feebleminded children, especially the very young ones. These children were usually more cooperative, however, when they were brought a second time after the f i r s t attempted interview had been abandoned. Occasionally, i t seemed better to allow the attendant to remain i n the room throughout the testing i f the child were very shy. In these cases, the attendants were asked not to help the child i n any way, and this request was honored without exception. It was the impression of the investigator that, i n the main, the WISC was much more d i f f i c u l t for the Retarded child than was the Binet. Such a distinction can easily be made simply i n terms of establishing rapport and engaging the subject's attention. The l i t t l e toys of the Stanford-Binet are immediately attractive to sub-jects of low mental age, but the WISC equipment does not have a vestige of this attractiveness. The statement - \"These pieces, i f put together correctly, w i l l make a boy. Go ahead and put them together.\" - was, apparently, meaningless to most of the Retarded subjects. Obviously, i f the subject f a i l s to comprehend the nature of the task, he can not successfully complete i t . Differences between the two tests i n the practical value of a score at this level are illustrated very well i n the Picture Completion Test. Instructions to t e l l what was missing only occasionally evinced a suitable response 20 from the subject, but most of these children could and did spontaneously name the pictures. This a b i l i t y would add to the subject's score on the Binet, but, so far as a WISC score i s concerned, i t makes no difference whether the subject did name the pictures or merely sat i n d u l l silence. For these reasons, the Stanford-Binet seems more suit-able for use with Retarded children, and also because i t does yield IQs at a lower level than does the WISC, although i t did f a i l with most of the Retarded 6-year-olds and with some of the 9-year-olds i n this study. In administering the WISC to subjects of $ years and older, not suspected mental defectives, the examiner i s permitted to omit the f i r s t few items i n each subtest and to start at a designated point of d i f f i c u l t y along the scale. If the subject does not achieve a certain number of consecutive correct responses, however, the examiner must work back from the designated point u n t i l the required number of consecutive correct responses has been given. While this technique i s at times undoubtedly time-saving i n testing older and brighter subjects, i t i s also somewhat awkward, especially i f i t i s constantly necessary to move backwards to easier items. Also, some subjects notice the marked change in the l e v e l of d i f f i c u l t y of the items, and, i n this way, become aware of having fa i l e d the more d i f f i c u l t items. Finally, the administration of the easier items i n the Performance scale seems excessively complicated; for instance, different directions and examples are given with each of the f i r s t four items i n the Block Design and Picture Arrangement tests. CHAPTER III TREATMENT OF DATA AND RESULTS Treatment of Data In order to find out to what degree the three indices of intelligence, yielded by the WISC on the Verbal, Performance, and Fu l l Scales, correlate with those yielded by the Stanford-Bine-^ at ages 6, 9 and 12 years, using Superior and Average subjects, Pearson Product Moment correlations were computed i n each age-intellectual group between each Wechsler scale and the Stanford-Binet. An attempt i s made to ascertain the nature of IQ differences yielded by the two tests by computing Mean Differences between the Stanford-Binet and Verbal, Performance, and F u l l Scales of the WISC i n each age-intellect-ual group, and then using the t-test to test for significance of the difference. Results (a) Obtained correlations between scales : Superior Subjects The obtained coefficients of correlations between these two measuring instruments based on the Superior subjects are shown i n 23 Table III, (p. 24). With the 6-year-old Superior subjects, the Verbal Scale correlation with the Binet was the lowest i n this age-intellectual group, and the third lowest i n the entire Superior category; the correlation was not significant. The Performance Scale correlation coefficient i s the highest i n this age-intellectual group, but i s not significant. The F u l l Scale correlation i s s l i g h t l y smaller than that of the Performance Scale; i t i s the median value for this entire intellectual category. In the 9-year-old Superior group, the Verbal Scale correlation i s significant at the 1% l e v e l of confidence, and i s the highest i n the entire study. While the Performance Scale coefficient i s the lowest i n this age-intellectual group, i t i s significant at the % level of confidence. The F u l l Scale correlat-ion i s the second highest i n the whole intellectual category, and i s significant at the 1% level of confidence. The Verbal Scale correlat-ion i s the highest i n the 12-year-old Superior group, and the correlat-ion next to, but lower than, the median value for the entire i n t e l l e c t -ual category; the value, however, i s not significant. The Superior 12-year-old Performance Scale correlation i s the lowest i n the entire study. The F u l l Scale correlation i s the second lowest i n the entire intellectual category; the coefficient i s not significant. Average Subjects The correlations based on Average subjects are found ( TABLE III CORRELATIONS BETWEEN STANFORD-BINET AND WISC VERBAL, PERFORMANCE, AND FULL SCALES Superior Subjects Age Verbal Scale Perform-ance Scale F u l l Scale 6 years .34 .53 •5.2 9 years .93 .67 • 91 12 years .50 .01 .33 .05 = .01 = .602 .735 25 i n Table IV, (p.26). In the 6-year-old Average group, the Verbal Scale corre-lation i s not significant and i s the lowest i n the group, and the third lowest i n the average intellectual group. The Performance Scale shows the highest correlation i n this age-intellectual group, although the coefficient i s not significant. The Full Scale correlation i s slight-l y lower than the Performance Scale correlation; i t i s the median value for the entire Average intelligence group. The Verbal Scale correlation at the average 9-year-old level i s the highest for the entire Average group, and i s significant at the l$level of confidence. The Performance Scale coefficient i s the lowest i n this age-intellectual group, and the second lowest i n the entire Average category; i t i s not significant. The F u l l Scale correlation i s second highest i n this age-intellectual group and i n the entire Average category; i t i s s i g n i f i -cant at the 5$ level of confidence. In the 12-year-old Average group, the Verbal Scale shows the highest correlation; i t i s the same as the Average 6-year-old F u l l Scale coefficient, and i s also the median value for the Average group; the value i s not significant. The Performance Scale correlation i s the lowest for this age-intellectual group and for this entire intellectual category. The F u l l Scale correlation for this age-intellectual group i s the medium one and i s not significant. Summarizing the results of Tables III and IV, i t i s immediately apparent that i n both intellectual groups, the rank order of the correlations i s the same at the same ages. The highest, second TABLE IV CORRELATIONS BETWEEN STANFORD-BINET AND WISC VERBAL, PERFORMANCE, AND FULL SCALES Average Subjects Age Verbal Scale Perform-ance Scale F u l l Scale 6 years .31 .52 .50 9 years • 77 .28 • 73 + 12 years • 50 .24 .46 .05 = .01 = .602 .735 A l l quantities are rounded off to second decimal place. When worked to third decimal place, r = .727« 27 highest, median, third lowest, and lowest correlations for each i n t e l -lectual group occur with the same WISC scales (i.e., Verbal, Perform-ance, or Pull Scales) at the same age levels. In each age-intellectual group, the F u l l Scale correlation i s the middle value, and i n both 6-year groups the rank order of the Verbal and Performance Scales i s the reverse of what i t i s in both of the 9-year and both of the 12-year groups. The rank order of the Verbal correlations i s the same i n both intelligence groups, with the 9-year-olds showing the highest coefficients, the 12-year-olds the next highest, and the 6-year-olds, the lowest. The rank order of the F u l l Scale correlations i s also the same in both intelligence groups, but the order i s not the same as for the Verbal Scale correlations. In this case, the 9-year-olds show the highest coefficient, but the 6-year-olds are second highest i n this case, with the 12-year-olds l a s t . With the exception of the fact that the 12-year-olds both show the lowest coefficients, there i s no rank order agreement between the two intellectual groups for the Performance Scale correlations. Defective Subjects The failure to obtain a sufficient number of the more intelligent defectives (described i n Chapter II) resulted i n only five of this group achieving WISC F u l l Scale intelligence quotients. As a result no correlations were run i n this sample. The actual results 28 w i l l be discussed later, under a separate heading. (b) Comparison of Mean IQs : Superior Subjects The comparison of the Mean intelligence quotients between the two tests for the Superior subjects are shown i n Tables V, VI, and VII, (pp.29, 30 and 31)-In the Superior group, a l l the Mean WISC IQs are smaller than the Mean Binet IQs with which they are being compared. In the Superior 6-year-old group, the Stanford-Binet IQs range from 110 to 154, a difference of 44 IQ points, the largest range in this age-intellectual group; the Binet standard deviation i s slight-l y larger than those of the Wechsler Intelligence Scale for Children i n this age-intellectual group. The WISC Verbal Mean IQ for the Superior 6-year-olds differs by only a small amount from the Binet Mean IQ for the same group and the difference i s not significant; the range of IQs on the Verbal Scale i n this age-intellectual category i s from 105 to 131, a range of 26 IQ points; this scale also shows the smallest standard deviation i n this age-intellectual group. The Mean Perform-ance IQ for the Superior 6-year-olds differs from the Mean Binet IQ by the same amount as does the Verbal Scale; again, the difference i s not significant. The IQs range from 106 to 138, a range of 32 IQ points, the largest range of any WISC scale i n this age-intellectual group; TABLE V MEAN WISC VERBAL SCALE IQs COMPARED WITH MEAN STANFORD-BINET IQs Superior Subjects Age Stanford-Binet WISC Verbal Scale M. IQ S.D. S * E ' M M- I Q S ' D ' S - E « M M- Diff- t + 6 years 122.80 11.73 3.91 119.30 7«87 2.62 3*50 . 90 9 years 137.90 17-00 5-66 125.20 11.86 3.95 12.70 5.16 12 years 131.90 8.75 2.92 113.20 8.07 2.69 18.70 6.65 + .05 = 2.26 .01 = 3-25 TABLE VI MEAN WISC PERFORMANCE SCALE IQs COMPARED WITH MEAN STANFORD-BINET IQs Superior Subjects M. IQ S.D. S.E.M M. IQ S.D. S.E.M M. Diff. t + 6 years 122.80 11.73 3.91 119.30 10.56 3.52 3.50 .97 9 years 137.90 17.00 5.66 119.10 10.62 3.54 18,80 4«47 12 years 131.90 8.75 2.92 123.90 10.67 3-56 8.00 1.75 +.05 = 2.26 .01 = 3.25 TABLE VII MEAN WISC FULL SCALE IQs COMPARED WITH MEAN STANFORD-BINET IQs Superior Subjects Age Stanford-Binet WISC Fu l l Scale M. IQ S.D. S.E.M M. IQ S.D. S.E» M M. Diff. t + 6 years 122.80 11.73 3.91 121.10 9.03 3.01 1.70 .49 9 years 137.90 17-00 5.66 124*40 11.00 3.66 13.50 4.86 12 years 131.90 8.75 2.92 120.20 7-33 2.44 11.70 3-74 + .05 » 2.26 .01 = 3.25 32 the standard deviation i s slig h t l y larger than i n the Verbal Scale. The Mean F u l l Scale IQ for the Superior 6-year-olds differs from the Mean Binet IQ by a much smaller amount than do the Verbal and Perform-ance Scales; the difference i s not significant. The IQs range from 108 to 13<3 points, a difference of 30 IQ points; the standard deviation i s smaller than that of the Performance Scale, and larger than that of the Verbal Scale. A l l the WISC scale standard deviations i n this age-intellectual group are smaller than the Binet standard deviat-ion, but there i s not a great difference between any of them. In the group of Superior 9-year-olds, the Stanford-Binet IQs range from 119 to 168, a range of 49 IQ points, the largest i n the entire study; this group of IQs also shows the largest standard deviation i n the study; i t shows a considerable difference from the next largest standard deviation. The Mean Verbal IQ i n the Superior 9-year-old group shows the fourth greatest difference from the Mean Binet IQ i n the entire study, although i t i s the smallest difference i n this age-intellectual group; the difference i s significant at the 1% l e v e l of confidence. The IQs range from 110 to 148, a difference of 38 points, and the largest WISC range i n this age-intellectual group; i t also shows the largest WISC standard deviation at this age-intellect-ual le v e l . The Mean Performance IQ differs from the Mean Binet IQ by the largest amount i n the study; the difference i s significant at the 1$ level of confidence. The IQs range from 104 to 136, a range of 32 points, the smallest range i n this age-intellectual group; the 33 standard deviation i s also the smallest i n the age-intellectual group. The Mean Fu l l Scale IQ for the Superior 9-year-old group shows the third largest difference from the Mean Binet IQ i n the entire study; the difference i s significant at the 1% level of confidence. The IQs range from 112 to 146, a difference of 34 points. The range of the sizes of the standard deviations i s smaller i n this age-intellectual group than for the Superior 6-year-old group, but the values are a l l larger. In the Superior 12-year-old group, the Stanford-Binet IQs range from 118 to 146, a difference of 28 IQ points, the smallest Binet range i n this intellectual group. The Superior 12-year-old group also shows the smallest Binet standard deviation i n this i n t e l -lectual group. The Mean Verbal Scale IQ sho\\re the second largest difference from the Mean Binet IQ i n the whole study, and the largest in this age-intellectual group; the difference i s significant at the 1 1% level of confidence, showing the largest c r i t i c a l ratio obtained i n the entire study. The IQs range from 104 to 130, a difference of 26 IQ points. The Mean Performance IQ shows the largest non-significant difference from the Mean Binet IQ i n the entire study; i t i s the smallest Mean Difference i n this age-intellectual group. The range of IQs i s from 100 to 140, a difference of 40 IQ points, and the largest range i n this age-intellectual group; the standard deviation i s the largest i n this same group. The Mean Fu l l Scale IQ for the Superior 12-year old subjects differs from the Mean Binet IQ by the smallest 34 significant difference at this intellectual levelj the difference i s significant at the 1$ level of confidence. The IQs range from 108 to 133. a range of 25 IQ points which i s the smallest range for this intellectual group; the standard deviation i s also the smallest for the intellectual group. The 6-year level has the smallest Mean Differences within each WISC scale; i n the Verbal Scale, the 12-year l e v e l shows the largest difference, with the 9-year level coming next. This order i s reversed for the Performance and F u l l Scales. In the Stanford-Binet, the largest standard deviation occurs at the 9-year leve l , the smallest at the 12-year level, and the middle value at 6 years. This order holds also for the WISC F u l l Scale. In the WISC Verbal Scale the largest deviation occurs at the 9-year level, the next largest at 12 years, and the smallest at 6 years; i n the Performance Scale, the-largest deviat-ion i s at 12 years, the next largest at 9 years, and the smallest at 6 years. The range of size for these deviations i s only .11 IQ points. In the Stanford-Binet, the largest Mean IQ occurs at 9 years, the next largest at 12 years, and the smallest at 6 years. In the WISC Fu l l and Verbal Scales, the highest Mean IQ occurs at 9 years, the next largest at 6 years, and the smallest at 12 years. In the Performance Scale, the highest Mean IQ occurs at the 12-year level, the next largest at the 6-year level, and the smallest at 9 years. Average Subjects The comparison of the Mean intelligence quotients 35 between the two tests for the Average subjects are shown i n Tables VIII, IX and X (pp.36, 37 and 38). In the average 6-year-old group, the Stanford-Binet IQs range from 89 to 112, a difference of 23 IQ points, which i s the small-est range i n the entire study; the standard deviation i s smaller than any i n the Superior group, and i s the smallest i n this age-intellectual group. The Mean Verbal Scale IQ for the Average 6-year-olds does not differ significantly from the Mean Binet IQ, but i t i s s t i l l the largest difference at this age-intellectual le v e l . The IQs range from 90 to 109, a difference of 19 IQ points, the smallest range i n the study, and the standard deviation i s the largest i n this age-intellectual group. The Mean Performance Scale IQ differs from the Mean Binet IQ by the smallest amount of any of the WISC scales i n this age-intellectual group; the difference i s not significant. The IQs range from 89 to 114, making a difference of 25 IQ points; the standard deviation i s the middle value deviation for the WISC scales at this age-intellectual level. The Mean F u l l Scale IQ differs from the Mean Binet IQ by a small amount which does not constitute a significant difference. The IQs range from 91 to 113, a difference of 22 IQ points, and the standard deviation i s the smallest WISC deviation i n this age-intllectual group. A l l the WISC Mean intelligence quotients at this level are smaller than the Mean Binet intelligence quotient. At the average 9-year-old leve l , the Stanford-Binet IQs range from 84 to 113, a difference of 29 IQ points, and the standard TABLE VIII MEAN WISC VERBAL SCALE IQs COMPARED WITH MEAN STANFORD-BINET IQs Average Subjects Age Stanford-Binet WISC Verbal Scale M. IQ S.D. S.E.M M. IQ S.D. S.E.M M. Diff. t + 6 years 105.00 6.56 2.19 101.30 7.71 2.57 3*70 1.32 9 years 99.40 8.08 2.69 101.60 6.39 2.13 2.20 1.28 12 years 101.30 8.03 2.68 95*20 6.73 2.24 6.10 2.45 + .05 = 2.26 .01 = 3.25 , TABLE IX MEAN WISC PERFORMANCE SCALE IQs COMPARED WITH MEAN STANFORD-BINET IQs Average Subjects Age Stanford-Binet WISC Performance Scale M. IQ S.D. S.E.M M. IQ S.D. S.E-M M. Diff. t + 6 years 105.00 6.56 2.19 102.00 7*24 2.41 3.00 1.33 9 years 99.40 8.08 2.69 98.20 7-.ll 2.37 1.20 .35 12 years 101-30 8.03 2.68 99-90 7-93 2.64 L 4 0 .43 +.05 = 2.26 .01 = 3.25 TABLE X MEAN WISC FULL SCALE IQs COMPARED WITH MEAN STANFORD-BINET IQs Average Subjects Age Stanford-Binet WISC F u l l Scale M. IQ S.D., S.E.M M. IQ S.D. S.E. M M. Diff. t + 6 years 105.00 6.56 2.19 101.60 6.59 2.20 3.40 1.55 9 years 99.40 8.08 2.69 99.90 5-11 1..70 . 50 . 27 12 years 101.30 8.03 2.68 97.10 6.69 2.23 4.20 1.63 +.05 = 2.26 .01 - 3.25 39 deviation i s the largest i n this intellectual group. The Mean Differ-ence between the Mean Verbal IQ for Average 9-year-olds and the Mean Binet IQ i s not significant, but i t i s the largest i n this age-intellect-ual group. The IQs range from 89 to 113, making a difference of 24 IQ points, the standard deviation i s small. The Mean Performance IQ differs from the Mean Binet IQ by the second largest amount for this age-intellectual group; again, the difference i s not significant. The IQs range from 89 to 110, the smallest range i n this age-intellect-ual group; the standard deviation i s the largest i n this age-intellect-ual group. The Mean Fu l l Scale IQ differs from the Mean Binet IQ by the smallest amount i n the study; the difference i s not significant. The IQs range from 88 to 109, making a difference of 21 IQ points, the same size as the range of the Performance Scale, at this age-intellectual level; the standard deviation i s the smallest i n the entire study. A l l the Mean Differences at this age-intellectual l e v e l are smaller than those at the Average 6-year-old level, and a l l the WISC scales show higher Mean intelligence quotients than does the Stanford-Binet. In the Average 12-year-old group, the IQs range from 82 to 118, a difference of 36 IQ points, the largest range i n this intellectual group; the standard deviation i s only the second largest Binet deviation i n this intellectual group, but i t i s the largest i n the age-intellectual group. The Mean Verbal IQ differs from the Mean Binet IQ by the largest amount i n this intellectual group, although the difference i s not as large as the largest insignificant difference; 40 the difference i s significant at the 5% level of confidence, and i s the only significant difference i n this intellectual group. The IQs range from 85 to 108, a difference of 23 IQ points. The Mean Performance IQ differs by the smallest amount i n this age-intellectual group from the Mean Binet IQ; the difference i s not significant. The IQs range from 89 to 110, a difference of 21 IQ points, the smallest range i n this age-intellectual group; the standard deviation i s the largest for this age-intellectual group. The Mean Pull Scale IQ differs from the Mean Binet IQ by an amount which i s not significant. The IQs range from 86 to 109, a difference of 23 IQ points; the standard deviation i s the smallest for this age intellectual group. A l l the Mean WISC scale IQs are smaller than the Mean Binet IQ at this age-intellectual level. In the Verbal and Fu l l Scales, the largest differences occur at 12 years, then 6 years, then 9 years. In the Performance Scale, the largest difference i s at age 6 years, then 12 years, then 9 years; the smallest difference i s always at the 9-year lev e l . The largest Verbal standard deviation occurred at 6 years, the next largest at 12 years, and the smallest at 9 years; i n the Performance Scale, the largest standard deviation occurred at 12 years, the next largest at 6 years, and the smallest at 9 years; i n the Fu l l Scale, the order was the same. In a l l the WISC scales, the 9-year level shows the smallest standard deviation. In the Stanford-Binet at this i n t e l l e c t -ual le v e l , the 9-year le v e l shows the largest standard deviation; the 41 12-year level, the next largest; and the 6-year level, the smallest. In the Stanford-Binet, the 6-year level shows the highest Mean IQ; the 12-year level, the next highest; and the 9-year level, the lowest. For the Verbal Scale, the highest Mean IQ occurs at 9 years; the next highest, at 6 years; and the lowest at 12 years. For the Performance Scale, the highest Mean IQ occurs at the 6-year level, the next highest at the 12-year level,- and the lowest at the 9-year level. For the F u l l Scale, the highest Mean IQ occurred at the 6-year level, the next highest at the 9-year level, and the lowest at the 12-year l e v e l . (c) The performance of the Defective Group : Of the twelve Retarded 12-year-olds, a l l twelye obtained intelligence quotients on the Stanford-Binet; the IQs ranged from 28 to 68, a difference of 40 IQ points. Of these twelve, only three obtained WISC F u l l Scale IQs; these ranged from 55 to 67, a difference of 12 IQ points. Of the remaining nine, one subject achieved an IQ on both the Verbal and Performance Scales, and two subjects obtained IQs on the Verbal Scale alone. Of the remaining six subjects, one achieved no scaled score at a l l , and the others achieved scaled scores only on the Verbal and F u l l Scales. In the group of Retarded 9-year olds, only five of the nine subjects achieved Stanford-Binet IQs; these IQs ranged from 26 42 to 60, a difference of 34 IQ points. Of these five, the two subjects, who achieved the highest Binet IQs for this group, also achieved Full Scale WISC IQs of 49 and 53• A l l the subjects in this age group achieved scaled scores on the WISC Full, Verbal and Performance Scales, because, due to the system of standard scores, i t is possible for a subject at this, and lower age levels, to achieve scaled scores without having given a single correct response. However, two of the subjects did achieve scaled scores (but no IQs) because they had given some correct responses. Of the group of four 6-year-old defectives, only one achieved a Stanford-Binet IQ (46) and none achieved a WISC Full Scale IQ. The three subjects who did not achieve Binet IQs, a l l achieved Verbal and Performance IQs, and the subject who achieved the Stanford-Binet IQ achieved a Verbal IQ but no Performance IQ, although he did obtain a scaled score on that section of the test. A l l the subjects achieved Full Scale scaled scores. The achievement of Verbal and Performance IQs by three of the subjects is due to the construction of the test, as outlined before, and no correct responses were actually given. In the case of the 12-year-olds, the WISC Full Scale IQs were of a size comparable to those on the Binet, and this is true of a l l the Verbal and Performance IQs obtained at this age level. In the 9-year old group, the WISC Full Scale: IQs are both smaller than the accompanying Binet IQs. 43 The actual scores obtained by these subjects are shown in Tables, XI, XII and XIII, (pp. 44, 45 and 46). 44 TABLE XI PERFORMANCE OF 12-YEAR OLD MENTAL DEFECTIVES ON WISC VERBAL, PERFORMANCE, AND FULL SCALES, AND ON STANFORD-BINET Subject Stanford-Number Binet IQ WISC Verbal Performance F u l l Scale Scaled IQ Scaled IQ Scaled IQ Score Score Score 1 68 25 69 29 71 54 67 2 36 0 - 0 - 0 -3 47 8 47 8 - 16 -4 57 5 0 - 5 -5 37 1 - 0 - 1 -6 32 2 - 0 - 2 -7 41 1 - 0 - 1 -8 28 3 - 0 - 3 -9 46 12 52 8 - 20 -10 52 11 51 27 68 38 55 11 43 9 48 12 47 21 -12 53 11 51 28 69 39 56 TABLE XII PERFORMANCE OF 9-YEAR-OLD MENTAL DEFECTIVES ON WISC VERBAL, PERFORMANCE, AND FULL SCALES, AND ON STANFORD-BINET Subject Stanford-Number Binet IQ WISC Verbal Performance F u l l Scale Scaled IQ Score Scaled IQ Score Scaled IQ Score 1 26 2 1 3 2 + 2 5 7 3 + 2 - 2 4 4 60 14 55 21 60 35 53 5 52 12 52 18 55 30 49 6 + 2 1 3 7 41 3 7 10 8 25 2 1 3 9 + 2 1 3 Impossible to compute an IQ because no basal age could be found. 46 TABLE XIII PERFORMANCE OF 6-YEAR-OLD MENTAL DEFECTIVES ON WISC VERBAL, PERFORMANCE, AND FULL SCALES, AND ON STANFORD-BINET Subject Stanford-Number Binet IQ WISC Verbal Performance Full, Scale Scaled IQ Scaled IQ Scaled IQ Score Score Score 1 + 11 51 12 47 23 2 + 11 51 12 47 23 3 + 11 51 12 47 23 4 46 12 52 9 - 21 + Impossible to compute an IQ because no basal age could be found. CHAPTER IV DISCUSSION OF RESULTS Theoretical Expectations and Obtained Results As has been previously pointed out, since the Stanford-Binet has a standard deviation smaller at age 6 years, larger at 12 years, and approximately the same size at age 9 years, than that of the Wechsler Intelligence Scale for Children, there should be certain d i f -ferences i n the Mean IQs yielded by the two tests when testing i n t e l -ligence at the extremes of the normal curve. At the 6-year level, for instance, the Mean Stanford-Binet IQ for superior subjects should be lower than the Mean WISC IQs; at the 12-year level, the Binet IQ for superior subjects should be higher than the Mean WISC intelligence quotients; and, at age 9 years, the Mean IQs on the two tests should be approximately the same. These expectations are actually f u l f i l l e d at the Superior 12-year level where the Binet Mean IQ i s considerably higher than the WISC Mean IQs, although the Mean Difference between the Binet and Performance Scale i s not significant. The differences are far larger, however, than might be expected i n terms of the sizes of the 48 standard deviations for the two scales. Furthermore, inspection of the differences obtained in the Superior groups at the other two levels tend to n u l l i f y the results of the Superior 12-year-old group of any s i g n i f i -cance as far as the theory about the standard deviations i s concerned. Although none of the differences between the Mean Binet and the WISC Scale IQs at the 6-year level i s significant, a l l the WISC scale Mean IQs are smaller than the Mean Binet IQ with which they are being com-pared- Moreover, i n the Superior 9-year-old group, where the closest agreement should be expected, there occur three of the four largest Mean Differences i n the study, a l l significant at the 1% level of confidence; again, the Mean Binet IQ i s higher than any of the Mean WISC intelligence quotients. In general, these results are f u l l y substantiated by the size of the obtained standard deviations. With one exception, the Stanford-Binet standard deviations are larger than the deviations obtained by the same group on the WISC scales. The exception occurs i n the Superior 12-year-old group where the Performance Scale standard deviation i s larger than that of the Stanford Binet. It w i l l be remembered that although the Mean Difference here was large, i t was not significant. It i s d i f f i c u l t to say, therefore, whether or not Wechsler's objections to the unusual sizes of the Stanford-Binet standard deviations at ages 6 and 12 years are j u s t i f i e d on practical grounds. Within the limits of this study and compared with his own point scale with i t s set deviations, they certainly are not j u s t i f i e d . 4 9 In general, the Average group appears to have achieved similar IQs on both the Stanford-Bine-^ and Wechsler Intelligence Scale for Children, Verbal, Performance and F u l l Scales. There i s one exception to this trend, and that i s when the 12-year-old Mean Binet IQ i s compared with the Mean Verbal IQ. This difference i s s i g n i f i -cant at the 5% lev e l of confidence. The nature of the obtained standard deviations i s not always the same as those which occurred i n the standardization groups of the two scales. It i s rather interesting to note that in neither the Superior nor the Average groups do the Stanford-Binet standard deviations, obtained at the three age levels, rank i n size i n the same order, as do the standard deviations obtained with the original Binet standardization group. According to the latt e r , the 12-year-old group should show the largest standard deviation, the 9-year group the next largest, and the 6-year group the smallest. In the Superior group i n this study, the 9-year group shows the largest standard deviation; the 6-year group the next largest; and the 12-year group the smallest. In the Average group, the 9-year group again shows the largest standard deviation, the 12-year group the next largest, and the 6-year group the smallest. On the other hand, within the results for each WISC scale of the Average group, and i n the Superior group's Performance Scale, the standard deviations at a l l ages show considerable similarity, as would be expected by the fixed nature of the WISC standard deviations. Such similarity does not occur, however, i n the results of the Superior X 50 group for the Verbal and Performance Scales. The dissimilarity of these scales i n this matter must be regarded as mere chance variations, since no explanation can be found i n terms of unusual ranges. In only one instance, which occurs i n the Superior 9-year-old Stanford-Binet results, are the obtained standard deviations as large as those reported by the authors of the two scales. The fact that the size of the obtained Stanford-Binet deviations at the three age levels do not rank i n the same order as do those i n the original standardization group, i s probably due to sampling errors, and could be explained by the differences i n the range of obtain-ed IQs. A sample of ten i s much more subject to influence by a single extraordinary score than i s a larger group. The larger size of the Binet standard deviations as compared with those of the WISC could be explained in terms of the point scale nature of the WISC with i t s set deviations, a matter to be discussed later i n this section. The smallness of the obtained standard deviations com-pared with those on the standardization groups of the two tests can probably be explained by the presence of factors i n t r i n s i c i n the present experimental design. The limited size of the sample would make for a considerable restriction i n the range of talent i n any one group, that is, the homogeneous nature of the samples probably accounts best for the small size of the standard deviations; the ranges of the Average and Superior groups are necessarily smaller than those of the standardization groups which encompass not one, but a l l , intellectual 51 levels. The exception which occurs in the results of the Superior 9-year-old group on the Stanford-Binet, is probably the result of chance influence. The possible range in the Superior group is necessarily less restricted than in the Average group, and it so happens that this group contains two exceptionally high intelligence quotients of 164 and 168. These differences between the two scales with regard to obtained Means and standard deviations are probably best explained in terms of differences in test construction. The study by Seashore, Wesraan and Doppelt (14), described in the introduction, it will be recalled points out that the range of IQs on a point scale is quite arbitrary. Since the range of IQs and standard deviations on the WISC is quite narrow, extreme scores are not to be expected. Thus, the range of IQs, which it is possible for a subject to achieve, is much more restricted. The obtained results, i.e., lower Mean WISC IQs for the Superior subjects, and smaller WISC standard deviations is, there-fore, to be expected on the basis of the construction of the test. The approximate equality of the Mean IQs in the Average group is also to be expected, since a difference in test construction, such as we are concerned with here, manifests itself only at the extremes of the normal curve. Since the WISC does not yield an IQ below 45, only five of the defective populations achieved Full Scale IQs. The apparent equality between the IQs at the 12-year level, with a slight lowering of 52 the WISC IQs at age 9, i s not i n accordance with the verifying Binet standard deviations, or the restricted IQ range of the WISC. The sample i s far too small, however, to permit any safe generalizations. For reasons outlined i n Chapter II and because i t yields an IQ below 45, the Stanford-Binet does appear to be the preferable i n -strument at this l e v e l . In terms of the many objections to the Mental Age con-cept, there is undoubtedly much to be said for the point scale, and a \"conservative\" one, such as the WISC which sets i t s scale so that extreme scores are rare, i s useful because, for example, a Superior score i s f a i r l y certain to indicate superior a b i l i t y . It i s possible, however, that the \"compression\" of the values eliminates certain shadings of quantity, especially at that point between the average and the extremes. These shadings are obtainable on a scale l i k e the Stanford-Binet, where the subject and his performance set the index to a far greater degree. Here the subject's age i s set to the month, rather than within four months, and there i s actually no IQ which he cannot attain. Whether the r e l i a b i l i t y and v a l i d i t y of the Binet, or of any intelligence test, warrants such precision i s another question. One rather confusing characteristic manifested by the point scale nature of the WISC i s the fact that, at the upper age and intelligence levels, a subject may make a perfect performance on a subtest and not achieve a perfect scaled score for i t , or, even more, confusing, at the lower age and intellectual level, the subject may give no correct responses whatever, and s t i l l achieve a 53 scaled score. Of the eighteen correlations computed, only five were significant, and a l l of these occurred at the 9-year l e v e l . It would appear then that, except at one age level, these two tests are not measuring the same thing. There are two major explanations as to why this may appear to be so. The f i r s t one could be i n terms of the motivational and intellectual development of the child during these years. A 6-year old child i n Grade I does not have a l l the desirable attitudes and attentiveness more common in an older child; one encounters, i n lesser degree, a l l the motivational problems of testing a very young child. As Anderson (l) points out, the younger the child, the less reliable the result; variation in motivation could account for the low correlat-ions at this age. With a few exceptions, this problem i s not a serious one i n testing the 9-year-old; by this age, the child has been i n school for three or four years and has, presumably, acquired the desir-able attitudes and motivation. According to Thurstone (p.206, 18), a child of this age has no manifest special a b i l i t i e s due to a maturational factor, thus giving the impression of global intelligence; on the basis of this theory, the high correlations which occur at this age l e v e l would be expected. As the child grows older, he tends to develop specialized a b i l i t i e s ; thus, the 12-year-old correlations are low due, presumably, to the fact that a good Performance Scale IQ does not always mean a good Verbal Scale IQ at this age le v e l . 54 There are several objections to this explanation. Although the correlations at the 9-year l e v e l are higher than at other ages, the Performance Scales i n both intellectual groups correlate lowest at this age - evidence that a high IQ on one WISC scale does not necessarily mean a high IQ on the other. A study by Swineford (16) gives evidence contrary to the results suggested by this present i n -vestigation that there are special a b i l i t i e s at 9 or 12 years of age. Students i n Grades VIII and IX (who would be older than 12 years of age) showed no change i n f a c t o r i a l composition from one year to the next, and Swineford concludes that with increasing maturity, the general factor (presumably, whether factual or a r t i f actual) plays a less important part. The second explanation could be purely i n terms of chance. The influence of chance factors i n a sample of ten i s bound to be much more distortive than i n a larger sample. Presumably, the action of Probable Errors of the IQ within a small and homogeneous group tends to obscure any true relationship which might exist. In a large sample, a few reversals i n rank order would be hidden by the general trend, what-ever i t might be . Whatever the true explanation, keeping i n mind the limited sample upon which this study was based, the two scales do not seem to be interchangeable. It does seem possible that a child given both of these tests might well yield widely different intelligence quotients on the respective tests. 55 The high correlations with large Mean differences within the Superior 9-year-old range are readily explainable. The high correlations indicate that the tests are certainly measuring the same a b i l i t y i n this group, and the differences i n scores are attributable to differences i n test construction. The WISC i s a scale with a set standard deviation, i n which the range has been comparatively r e s t r i c t -ed so that lower scores i n this indicate much higher a b i l i t y than the same score i n the Stanford-Binet; for instance, an IQ of 130 amongst a group of Superior children, such as those encompassed by this study, i s quite usual on the Stanford-Binet, but i s much rarer on the WISC. The dissimilarity of intelligence quotients yielded by the two tests i s especially noticeable i n the Superior 12-year-old group: correlations are a l l low and the Verbal and F u l l Scale IQs show a significant difference from the Mean Binet IQ at this age-intellectual l e v e l . Inspection of the actual scores reveals extreme differences and, with rare exceptions, these differences are i n the direction of a lower WISC score. Such a situation i s bound to be confusing i n some degree to the clinici a n ; he cannot be sure, for instance, when a subject receives a slightly-above-average IQ on the WISC, whether he would have made a comparable showing on the Stanford-Binet, or whether, as happened several times i n this group of Superior 12-year-olds, he might have made a Superior showing i n the Stanford-Binet. This latter does not constitute a fixed criterion by any means, but i t certainly i s a criterion of some sort, i f only through long use, and one which cannot 56 be entirely ignored. It i s interesting to note that i n both intellectual groups, the rank order of the three correlations (Binet with Verbal, Performance, and Full Scales) i s the same for the same age, i.e., the rank order of the correlations i s the same for the Superior 6-year-olds as for the Average 6-year-oldsj the same for the Superior 9-year-olds as for the Average 9-year-olds; and the same for the Superior 12-year-olds as for the Average 12-year-olds. Any similarity i n rank order of these correlations, within each intellectual group, is robbed of i t s s i g n i f i -cance by the fact that both intellectual groups display exactly the same rank order at each age. Furthermore, the highest, second highest, median, third lowest and lowest correlations, i n both intellectual groups, occur i n precisely the same correlations, i.e., with the same WISC scales (Verbal, Performance or Full) and at the same age levels. These results give some suggestion that age level influences much more the degree to which the Stanford-Binet and Wechsler Intelligence Scales for Children measure the same thing, than does intellectual l e v e l . Obtained Results compared with the Literature As has been previously pointed out, these results agree very largely with the predictions made by Seashore, Wesman and Doppelt (14). The findings also agree, i n the main, with the results published by Frandsen and Higginson: the correlations at the 9-year le v e l are high (sometimes s l i g h t l y higher than those reported by Frandsen and 57 Higginson) with the Performance Scale correlating the lowest. This study did not find, however, that such high correlations occur also at other ages as i s implied by the Frandsen and Higginson study, (5)' The results agree with the findings of Sartain (13), on the Stanford-Binet and Wechsler-Bellevue, that Superior subjects score higher on the Binet, although most of the differences found on this study are greater than the 5 IQ points suggested by him; similar findings were made by Goldfarb (6). The results for the Retarded group are simply not adequate to permit a comparison of the findings with those of the literature. In agreement with Wechsler (p.157, 19), however, there does appear to be some trend toward the Performance intelligence quotient being higher than either the Verbal or F u l l Scale IQs. The results of this study agree very closely with the findings of Halpern (9)* She found that the highest correlations between the Stanford-Binet and Wechsler-Bellevue occurred at the extremes of the normal curve, and that, i n spite of this fact, the greatest d i f -ferences occurred with the Superior subjects and the least with the Defectives. In general, this study agrees with the literature of the Stanford-Binet and Wechsler-Bellevue, i n showing lower correlations between the Binet and WISC Performance Scale, than between the Binet and WISC Verbal or Fu l l Scales. 58 Consistent Trends emerging from the Data The Superior 9-year-old w i l l probably score much higher on the Stanford-Binet than on the WISC, but his position relative to the Mean i s probably similar on both scales. The results for the Superior 1 2-year-old are a great deal more confusing; the two tests do not appear to be measuring the same phenomenon at this age-intellect-ual le v e l . Much more research should be done at this, and surrounding, age-intellectual levels to ascertain whether or not the results of this investigation are due merely to chance influences, or whether they are due to some difference i n subject matter i n the two tests, or to the influence of some maturational factor. One should also remember that the WISC does not yield an IQ below 45, and that, therefore, the Binet i s much more useful i n test-ing at this l e v e l of intelligence. The fact that age influences the degree of correlation between the Binet and each of the three WISC scales, much more than does intellectual level, may be i n agreement with some theories of general or special a b i l i t i e s and the influence of maturation. As was pointed out previously, however, the trend i s not clearly towards either specific or general a b i l i t i e s . Suggestions for future Research This study should be repeated to ascertain whether or not 59 the trends discerned i n i t are actual or merely the result of various chance influences; such a study would probably be rendered more reliable by the use of a larger sample. Since the results of this investigation suggest that the use of the WISC at very low levels of intelligence i s not entirely satisfactory, future studies might u t i l i z e a less extreme group of defectives. As outlined i n a previous section, a future investigation might be directed toward ascertaining whether or not the wide d i f f e r -ences i n intelligence quotients, yielded by the two tests at the Superior 12-year-old level, are a manifestation of the development of special a b i l i t i e s . This purpose could, perhaps, be accomplished by testing above and below the 12-year age level, and noting consistent trends toward smaller correlations and wider significant differences with increasing age. SUMMARY The purpose of this study was to investigate the relationship between IQs yielded by the 1937 Revised Stanford-Binet, Form L, and the Wechsler Intelligence Scale for Children (WISC) at different age levels and different levels of intelligence. The investigation endeavored to find out to what extent the Stanford-Binet correlates with each of the WISC Verbal, Performance and F u l l Scales at ages 6, 9 and 12 years, and with superior, average and defective children. It attempted to ascertain whether the Mean scores i n this group show rea l differences i n average IQs yielded by the two tests, and, i f so, i n which direction these differences occur. The WISC i s a point scale with a set standard deviation, while the size of the Stanford-Binet varies, being unusually small at age 6 years, and unusually large at age 12 years. It was hypothesized that at the extremes of the intelligence distribution at these ages, there should be a difference between the Mean IQs of the two scales i n the direction corresponding to the size of the standard deviation at the age i n question. To insure that extreme scores would occur, the sample included superior subjects and mental defectives. Besides the 6 and 12-year-olds necessary to test this hypothesis, the sample also 61 included 9-year-olds, whose Stanford-Binet standard deviation approximates that of the WISC. These 9-year-olds, plus a sampling of average child-ren, were included for purposes of control and comparison but also for their own research value. The sample was composed of eighty-five subjects - ten children i n each age category i n the Superior and Average groups; and i n the group of Defectives, fourteen 12-year-olds, nine 9-year-olds, and four 6-year olds. The Average and Superior subjects were obtained in the elementary schools throughout Greater Vancouver, and were chosen on the basis of past records of a b i l i t y and achievement. The group of Defectives was obtained from the Woodlands School, New Westminster, B.C., and from the Rainier State School, Buckley, Washington, U.S.A.: a l l the feebleminded subjects had been diagnosed as \"familial\" defectives. The positive correlations which occurred may be summarized as follows : 1. In the 9-year-old Superior group, the Stanford-Binet IQ correlated significantly with -(a) the WISC Verbal IQ at the 1% level of confidence; (b) the WISC Performance Scale IQ at the 5$ level of confidence; (c) the WISC F u l l Scale IQ at the 1# level of confidence. 2. In the 9-year-old Average group, the Stanford-Binet IQ correlated significantly with -62 (a) the WISC Verbal Scale IQ at the 1% level of confidence; (b) the WISC F u l l Scale IQ at the 5% level of confidence. Significant differences between the Mean IQs of the two tests may be summarized as follows : 1. In the group of Superior 9-year-olds, the Stanford-Binet IQs were significantly higher (at the 1$ level of confidence) than the WISC Verbal, Performance and F u l l Scale IQs. 2. In the group of Superior 12-year-olds, the Stanford-Binet IQs were significantly higher at the 1$ level of confidence for the WISC Full and Verbal Scale IQs. 3» In the group of Average 12-year-olds, the Stanford-Binet i s significantly higher at the 5% level of confidence than the WISC Verbal Scale IQ. The major conclusions of this study are : 1. The obtained results are i n essential agreement with the studies comparing the Wechsler adult scale and the Stanford-Binet. 2. There seems to be a consistent tendency i n this study and others reviewed previously, toward lower correlations between the Stanford-Binet and WISC Performance Scale, than between the . Stanford-Binet and WISC Verbal and F u l l Scales. 3' There seems to be no support for the hypothesis that the differences between the Mean Binet and 63 Mean WISC IQs at the Superior level w i l l differ i n direction according to the size of the Stanford-Binet at the age level i n question. 4« The WISC appears to be an unsatisfactory test for the markedly retarded children. Both i n terms of construction and interest value to subjects, the Stanford-Binet seems to be a better scale for the measurement of the lower levels of intelligence. 5. Keeping i n mind the limited sample upon which this study was based, the two scales do not seem to be interchangeable. The practical import of this conclusion i s that clinicians, social workers, psychiatrists, school teachers, and so on, should be f u l l y aware that the child given both tests may well yield widely d i f f e r -ent IQs on the respective tests. BIBLIOGRAPHY ANDERSON, John E. The limitations of infant and preschool tests in the measurement of intelligence. J. Psychol., 1939, 8, 351-379-BALINSKY, B., ISRAEL, H., and WECHSLER, D. Relative effective-ness of the Stanford-Binet and Bellevue intelligence scales i n diagnosing mental deficiency. Amer. J. Orthopsychiat., 1939, 9, 798-801. BENTON, Arthur L., WELDER, Arthur, and BLEAUVELT, Jean. Perform-ance of adult patients on the Bellevue intelligence scales and revised Stanford-Binet. Psychiat. Quart., 1941, 15, 802-806. BURT, C y r i l . C r i t i c a l notice of Thurstone's \"Multiple Factor Analysis.\" Brit. J. educ. Psychol., 1947, 17, 163-169. FRANDSEN, Arden N., and HIGGINSON, Jay B. The Stanford-Binet and the Wechsler Intelligence Scale for Children. J. con- sult. Psychol.. 1951, 15, 236-238. GOLDFARB, William. Adolescent performance i n the Wechsler-Bellevue intelligence scale and the revised Stanford-Binet examination, Form L. J. educ Psychol.. 1944, 35, 503-507. GUERTIN, Wilson H. Mental growth i n pseudo-feeblemindedness. J. c l i n . Psychol.. 1949, 5, 414-418. GUILFORD, J. P. Human a b i l i t i e s . Psychol. Rev.. 1940, 47, 367-394-65 9. HALPERN, Florence. A comparison of the revised Stanford-Binet, Form L, and the Bellevue adult intelligence test as c l i n i c -a l instruments. Psychiat. Quart. Suppl. t 1942, 16, 206-211. 10. KUTASH, S. B. A comparison of the Wechsler-Bellevue and the revised Stanford-Binet scales for adult defective delin-quents. Psychiat. Quart., 1945, 19, 677-685-11. MITCHELL, M. Performance of mental hospital patients on the Wechsler-Bellevue and the revised Stanford-Binet, Form L. J. educ. Psychol.. 1942, 33, 538-545-12. RABIN, Albert I., and GUERTIN, Wilson H. Research with the Wechsler-Bellevue test: 1945-50. Psychol. Bull., 1951, 48, 211-248. 13- SARTAIN A. A comparison of the new revised Stanford-Binet, the Bellevue scale and certain group tests of intelligence. J- s o c Psychol.. 1946, 23, 237-239-14- SEASHORE, Harold, WESMAN, Alexander, and DOPPELT, Jerome. The standardization of the Wechsler Intelligence Scale for Children. J. consult. Psychol.. 1950, 14, 99-110. 15- SPEARMAN, C. Theory of general factor. Brit. J. Psychol., 1946, 36, 117-131-16. SWINEFORD, Frances. Growth i n the general and verbal bi-factor from Grade VII to Grade IX. J. educ. Psychol., 1947, 38, 257-272. 17- TERMAN, Lewis M., and MERRILL, Maud A. Measuring Intelligence-London: George G. Harrap and Co. Ltd., 1937-18. THURSTONE, L. L. Current issues i n factor analysis. Psychol. Bull., 1940, 37, 189-237-19. WECHSLER, David. The Measurement of Adult Intelligence. Baltimore: The Williams and Wilkins Company, 1944. 66 2 0 . WECHSLER, David. Wechsler Intelligence Scale for Children - Manual. New York: The Psychological Corporation, 1949. "@en ; edm:hasType "Thesis/Dissertation"@en ; edm:isShownAt "10.14288/1.0106963"@en ; dcterms:language "eng"@en ; ns0:degreeDiscipline "Psychology"@en ; edm:provider "Vancouver : University of British Columbia Library"@en ; dcterms:publisher "University of British Columbia"@en ; dcterms:rights "For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use."@en ; ns0:scholarLevel "Graduate"@en ; dcterms:title "A comparison of the Stanford-Binet (1937 revision, form L) and Wechsler intelligence scale for children at different age and intellectual levels"@en ; dcterms:type "Text"@en ; ns0:identifierURI "http://hdl.handle.net/2429/41514"@en .