The intelligence of school children : how children differ in ability, the use of mental tests in school grading and the proper education of exceptional children

Front matter

^0' •!

VAQ^ r. ^vo.c-i^

4 o

Preface

This book has been written for the rank and file of teachers, school supervisors, and normal-school students. Its purpose is to illustrate the large individual differences in original endowment which exist among school children and to show the practical bearing of these differences upon the everyday problems of classroom management and school administration. It does not treat, except incidentally, the psychological principles underlying intelligence tests. Some of these problems the writer has touched upon elsewhere.* The technique of giving the tests of the revised Binet scale and the general significance of mental tests for education have been set forth in some detail in another volume of this series. The Measurement of Intelligeneey^ which should be read in connection with the present volume.

In the preparation of this volume the writer has drawn heavily upon the data from investigations made by a number of his students at Stanford University. His debt to them is very great, not only for the generous way in which they have placed valuable data at his disposal, but if possible even more for the loyalty

, ^ See Terman, Lewis M., The Stanford Revision and Extension of the Binei-Simon Measuring Scale of Intelligence. Warwick and York, Baltimore, 1917; pp. 179.

2 Houghton Mifflin Company, 1916; pp. 362.

xii PREFACE

and enthusiasm with which they have worked together in carrying through cooperative undertakings of the most laborious nature. What a single individual working alone can accomplish in research with mental tests is well-nigh infinitesimal. Substantial progress can come only from the cooperative work of many on closely allied problems. This volume is in large measure the outcome of studies made by various members of the author's classes in intelligence tests during the years 1916-17 and 1917-18, the central topic for each year being the relation of school success to intelligence. Students who have contributed important data to the various chapters include the following: Virgil E. Dickson (tests of first-grade pupils); W. M. Proctor (tests of high-school pupils) ; Irene Cuneo (tests of kindergarten children) ; Margaret Hopwood Hubbard (tests of superior

children) ; O. S. Hubbard (tests of fifth-grade pupils) ; Isabel Preston (analysis of discrepancies between

mental age and school success) ; J. K. Flanders (tests of Express Company employees); H. E. Knollin (tests of unemployed, prisoners, and

business men) ; Dr. J. Harold Williams (tests of juvenile delinquents) ; Lowery Howard and Virgil Dickson (tests of retarded children in the schools of "X" County);

PREFACE xiii

Various students who cooperated in gathering the data on which the Stanford Revision of the Binet scale was based. Among these were Dr. George Ordahl, Dr. Louise Ellison Ordahl, Grace Lyman, Neva Galbreath, and WiKord Talbert. These studies are but parts of a larger investigation of mental growth and individual differences. Several of them are far from complete at the time of this writing. Hundreds of children who have been tested in the vicinity of Stanford University are being followed up in order to discover the value of mental tests as a means of forecasting the educational achievements possible to children of various degrees of intelligence. The investigation also involves the re-testing of a large number of children in successive or alternate years in order that typical curves of mental growth may be established. The writer believes that studies of this kind should entirely replace the controversial literature on the value of Binet and other mental tests. There is no other foundation for science, whether pure or appUed, than positive, definitely verifiable facts. Psychology is no exception.

Another study should be mentioned in this connection, although circumstances prevent the publication of its results at present. With the assistance of a number of Stanford University students, the group intelligence scale devised for use in the United States Army was given during the school year of 1917-18 to approximately six thousand pupils from the third

XIV PREFACE

grade to the senior year of high school. The purpose of the investigation was to secure data on the reHability of the army tests, and to this end a large amount of supplementary information regarding each pupil was secured for correlation with the tests results. This information included age, grade, years in school, nationality of parents, occupation of father, teachers' ratings of the children on intelligence, quality of school work, and several character traits. Approximately six hundred of the same pupils had been tested with the Stanford Revision of the Binet scale. In every respect the results of this investigation support the data and conclusions presented in the various chapters of this volume. The army tests, which were given to approximately 1,700,000 soldiers, demonstrated beyond question that the methods of mental measurement are capable of making a contribution of great value to military efficiency. That their universal use in the schoolroom is necessary to educational efficiency will doubtless soon be accepted as a matter of course.

The fact that the conclusions here offered are based chiefly upon results secured by the use of the Stanford Revision of the Binet-Simon tests must not be understood to imply that the writer looks with disfavor upon other intelligence scales. To the extent that the conclusions are valid at all, they can be confirmed by any system of tests which affords a reasonably accurate measure of general mental ability. However, it is not

PREFACE XV

the purpose of this volume to summarize the hundreds of interesting and valuable investigations which have utilized either Binet or other tests of school children. For the most part such investigations have been directed toward the improvements of methods. The writer's present aim is the more practical one of showing how the results of mental tests may be put to everyday use in the grade classification and in the educational guidance of school children.

The author is indebted to Professor R. M. Yerkes for reading several chapters of the manuscript and for many helpful criticisms.

Stanford University March 1, 1919

Contents

I. Some Principles of Intelligence Testing . . 1

The Binet tests, a method of assaying inteUigence — Why a mental test is significant — The meaning of mental age — Mental age a basis for school grading — The intelligence quotient — The I Q as a basis for prediction — Effect of environment on the 1 Q — Scales for group testing.

n. Ajmount and Significance of Individual Differences 17

Chapter I Some Principles Of Intelligence Testing*

The Binet tests, a method of assaying inteUigence. In order to find out how much gold is contained in a given vein of quartz it is not necessary to uncover all the ore and extract and weigh every particle of the precious metal. It is sufficient merely to ascertain by borings the linear extent of the lode and to take a small amount of the ore to the laboratory of an assay er, who will make a test and render a verdict of so many ounces of gold per ton of ore.

A half -century ago Francis Galton predicted that it would sometime be possible to obtain a general knowledge of the intellectual capacities of a man by sinking shafts, as it were, at a few critical points. Already Galton's dream is in process of realization, for in the last decade mental testing has become one of the most fruitful branches of psychological science. The credit for pointing the way belongs largely to the French psychologist, Alfred Binet, who, after more than fif-

^ For a more extended discussion of the principles of mental testing, including detailed instructions for the use of the Stanford Revision of the Binet-Simon InteHigeiice Scale, see Terman, Lewis M., The Measurement of Intelligence. Houghton MiflBin Company, 1916; pp. 362.

teen years of patient research, gave to the world in 1908 the system of mental tests now known as the Binet-Simon Intelligence Scale. In various revised forms the method has come into general use in public schools, institutions for defectives, prisons, reform schools, and juvenile courts in the United States and in Europe. Our debt to Binet is very great, for he succeeded in bringing psychology down from the clouds and making it useful to men.

The Binet scale is made up of an extended series of tests in the nature of problems, success in which demands the exercise of the intellectual processes. As left by Binet, the scale consisted of fifty-four tests, ranging in difficulty from tests which are passed by the average child of three years, to tests which are difficult enough for the average adult. The Stanford Revision has increased the number of tests to ninety and has extended the scale far enough to measure the intelligence of superior adults.

The ninety tests in the revised scale constitute an extremely variegated series. This is necessary, since their purpose is to measure the subject's general intelligence, not his special ability in a particular line. They include tests of memory, language comprehension, size of vocabulary, orientation in time and space, eyehand coordinations, knowledge about familiar things, judgment, ability to find likenesses and differences between common objects, arithmetical reasoning, resourcefulness and ingenuity in difficult practical situations, ability to detect absurdities, apperception, the speed and richness of association of ideas, the power to combine the dissected parts of a form board or a group of ideas into a unitary whole, the capacity to generalize from particulars, the ability to deduce a rule from connected facts, etc. Thus the tests give a kind of composite picture of the subject's general mental ability, and since standards of comparison have been established for each of the individual tests by trying it out on hundreds of unselected normal children of all ages, it is possible to express the total result of an examination in terms of "mental age" norms.

Why a mental test is significant. Are we justified in attributing real diagnostic significance to the little intellectual "stunts" called for by an intelhgence scale .^ Some of these may even appear trivial. What does it signify, for example, whether a given ten-year-old subject names forty words or a hundred words in three minutes.'^ Whether he puts together the parts of a form board in thirty seconds or in two minutes? Whether he defines thirty words or sixty words of a hundred-word list? Whether his definitions of words are stated in terms of " use " or in terms " superior to use"? Whether a series of five digits or only a series of three digits can be repeated backwards after a single auditory presentation? Whether there are three, two, one, or no successes in the attempt to draw a diamond-shaped figure from copy?

The secret lies in the standardization of the tests upon normal children of different ages. Without such a standardization the tests would mean nothing. Standardization is coming to play the same role in psychology that it has long played in the various branches of applied science. The architect or bridge engineer plans his structure with constant reference to foot-pounds of strain which various materials will withstand. The physician analyzes a drop of blood and, by comparison of corpuscle count and haemoglobin with the norms for health and disease, is able to render an important diagnosis. The psychologist working with mental tests may be compared with the palaeontologist who finds in a gravel bed of some prehistoric age a skull cap, a fragment of jaw, and a broken humerus. Although the layman might not even recognize the human origin of such remnants, the palaeontologist is able to tell us that the bones are those of a middle-aged male, that the species to which he belonged had not yet learned to stand erect, that he probably did not know the use of fire (worn teeth indicate that he subsisted on uncooked foods), that his intelligence was inferior (cranial contents only two thirds that of modern man), and that he had probably evolved but limited p>ower of speech (diminutive points of attachment for the speech muscles). A little technical acquaintance with the standards of shape, size, and structure of human bones has transformed the meaningless fragments into a "missing link" — Homo neanderthalensis.

Perhaps no two things could be more alike to casual inspection than the balls of two thumbs; yet one who has been taught to read finger prints can ordinarily find from forty to seventy separate and individually sufficient points of identification. Just as many a man has been hanged on the evidence of his finger prints, so many an individual might safely be committed to an institution for the feeble-minded on the evidence of ten or a dozen intelligence tests which have been standardized according to age norms.

The meaning of mental age. Both the individual tests of the Binet scale and the scale as a whole have been standardized on the basis of age norms. The tests themselves are located in age groups in such a way as to bring it about that the average child of eight years will earn by the scale a "mental age" of eight years, the average twelve-year-old a "mental age" of twelve years, etc. Such an arrangement was arrived at empirically by trying out a series of tests upon hundreds of normal children of different ages. The Stanford Revision, for example, was based on tests of 1700 children and 400 adults.

To illustrate the use of the scale, let us suppose we are testing a child of eight years. If our subject passes successfully as far as the average child of eight years, we say that his mental age is eight years, or in this case normal. If he goes as far as the average ten-year-old, we say that he has a mental age of ten years. If he earns no more credit than the average six-year-old, his mental age is six years. Binet merely took a standard of comparison which every one uses (namely, the standard of age) and made it definite by finding out what intellectual performances representative children of different ages are capable of.

It is necessary that the reader should at the outset arrive at a correct understanding of what the term "mental age" is and is not intended to signify. Two misconceptions are to be avoided:

1. That each "mental age" is a separate and qualitatively distinct level of mental attainment, contrasting markedly with both the mental age which precedes it and that which follows it. Such a use of the term is not in harmony with the facts. Mental development is consecutive and gradual. There is probably no mental power, capacity, or function which has a Minerva birth. The "faculty" in question develops first in rudimentary form, then grows gradually stronger and more definite until, by imperceptible stages, it reaches a state of maturity.

2. Another misunderstanding comes from the assumption that those who use the term believe a given mental age is a stage of development which all normal individuals pass through at the corresponding actual age. Such a belief would imply that at the age of ten years, for example, all children who do not belong to some special type (defective, genius, etc.) should be found at the ten-year mental age, eight-year children at the eight-year mental age, etc. It is one of the main purposes of this book to show how widely children of a given age differ in mental age, and how greatly children of adjacent ages overlap each other in mental age.

'' The real meaning of the term is perfectly straightforward and miambiguous. By a given mental age we mean ikat degree of general mental ability which is possessed by the average child of corresponding chronological age.

Mental age a basis for school grading. The significance of mental age for the teacher hes in the fact that it can be used as a basis for grading the pupils so as to secure class groups of homogeneous ability. As will be shown in succeeding chapters, the pupils of given grades, or even the pupils of one grade in a single classroom, are far from equal in general inteUigence or in ability to master the school work. Generally speaking, not far from a fourth of the pupils in any given grade have a mental level too low to make satisfactory work in that grade p>ossible, while another fourth have reached a mental level which would enable them to succeed in a higher grade.

The intelligence quotient. The mental age merely indicates the level of development which a child has reached at a given time. Considered apart from chronological age it does not tell us whether a child is bright, dull, or average. Of three children all testing at the mental age of eight years, one might very well be exceptionally superior, one average, and one feeble-minded. Such would be the ease if their chronological ages were six, eight, and twelve years. In addition to an index of absolute mental level, we need an index of relative brightness. Such is the intelligence quotient (I Q), which is the ratio of mental age to chronological age. The six-year-old of eight-year mental age has an I Q of 8/6 or 133;^ the twelve-year-old with a mental age of eight years, an I Q of 8/12, or 67. In computing the I Q of an adult subject, years of chronological age in excess of sixteen are disregarded, as the development of native intelligence seems practically to cease not far from this age.

An idea of how greatly school children differ in brightness is shown by the analysis of the I Q's of 1000 representative children in which it was found that: ^

The lowest 1% go to 70 or below, the highest 1% reach 130 or above

" " 2% " " 73 " " " " 2% " 128 "

" " 3% " " 76 " 3% " 125 " "

« " 5% " " 78 ' " 5% " 122 " "

The intelligence quotient a basis for prediction. Just as mental age indicates the school grade in which a child normally belongs at a given time, so the I Q is the basis for prediction in regard to the child's later mental development. The possibility of such predic-

^ More correctly, 1.33; but the decimal point is customarily omitted, the quotient being understood as expressed in per cent.

2 Terman, Lewis M., The Measurement of Intelligence. Houghton Mifflin Company, 1916. See chapter v.

tion comes from the fact that the I Q has been fomid in the large majority of cases to remain fairly constant, at least for the ages between three or fom* and fom-- teen or fifteen. For illustration, we will take the case of a four-year-old child who is found to have a mental age of five years, and whose I Q is therefore 125. The probability is that this child will continue to have a mental age not far from 35 per cent above his chronological age, with the consequences which may be expressed as follows:

ironological

Probable

Probable school age ability

Upper kindergarten

6 "

1\ "

Second school grade

8 "

High fourth grade

10 "

Low seventh grade

12 "

First year high school

It would, of course, be absurd to expect the I Q to maintain itself at an absolutely constant figure. Fluctuations occur for at least three reasons: (1) There may be in exceptional cases a certain amount of irregularity in the actual rate of mental development.

(2) The results of a test may be influenced to some extent by the conditions under which it is given, the state of the child's health, his attitude toward the test, fatigue, and other temporary or accidental factors. Re-tests after a brief interval indicate that errors from this source are ordinarily not large.

(3) There is inevitably a certain amount of error in every I Q ratidg, due to imperfections in the scale used. If the scale has been so standardized that it yields mental ages which are too low, the I Q found will be too low; if the scale errs in the direction of being too generous, the resulting I Q will be too high. A scale may err in one direction at one level and in the opposite direction at another level. It was the most serious fault of the original Binet scale that in the lower range of tests it yielded mental ages which were too high, and in the upper range mental ages which were too low. The effect of such errors is greatly to exaggerate the amount of fluctuation to which mental growth is subject. It was the main purpose of the Stanford Revision to reduce these constant errors. Chapter XI shows in detail the degree of constancy which may be expected for the I Q when the Stanford Revision is used. While the law of constancy is subject to minor revisions, few things are more certain than the essential untruth of the widespread belief that mental development knows no regularity, and that the dullard of to-day becomes the genius of tomorrow. The fact is that, apart from minor fluctuations due to temporary factors, and apart from occasional instances of arrest or deterioration due to acquired nervous disease, the feeble-minded remain feeble-minded, the dull remain dull, the average remain average, and the superior remain superior. There is nothing in one's equipment, with the exception of character, which rivals the I Q in importance. Effect of environment on the I Q. The question is always raised whether, in estimating a child's intelligence on the basis of the I Q, it is not necessary to make allowance for the influence of social environment. For example, it is often argued that the child cannot know his age if he has never heard it, cannot read and report the memory passages if he has never ^ attended school, cannot count from 20 to 1 if he has never been taught to count from 1 to 20, cannot name the days of the week or the months of the year unless he has heard others name them, and that therefore the I Q can have Uttle significance except possibly as an index of the subject's social and educational environment.

It is, of course, true that an individual who for his entire life had been entirely deprived of human environment (assuming such a thing to be possible) could not pass a satisfactory Binet test, however normal his original endowment may have been. To use an extreme illustration, a child of ten years who had been reared in a cage, whose wants had been supplied while he was asleep, or by means of ingenious mechanical contrivances, who had never seen a human being, could hardly be expected to make a brilliant showing in defining words in the vocabulary test, detecting absurdities, repeating sentences, reading the Binet passage, answering comprehension questions, or naming sixty words. We may go further and assume that such a subject would be as little successful with the three-year as with the ten-year test.

Needless to say, the Binet scale was not intended for subjects of the type we have just described. Its use in a given case takes for granted that the ordinary and all but inevitable social contacts have been made, that the subject is not deaf or blind, and that he has had reasonable opportunity to learn the language in which the tests are given. Children who have attended school for any considerable time meet all of these requirements, whatever the social status of the home.

As a matter of fact, limited acquaintance with the language employed in the examination does not put the subject at great disadvantage in many of the tests. In some it does, and in testing subjects who are under this handicap the vocabulary test and a few others may very well be omitted. Following are two illustrations which show that the validity of the scale does not hinge entirely upon the subject's knowledge of English:

1. Kohs tested a Belgian refugee child of nine years who had been in America but two years. Although this child's acquaintance with the English language was very limited, the I Q earned on the Stanford- Binet scale was 99. The child was also doing school work of average quality in the fourth grade.

2. Dickson tested a Japanese boy, aged five years, two months, who had never attended school and who had had little opportunity to learn English; yet this boy earned a mental age of seven years and an I Q of 133.

That lack of schooling does not prevent a subject from earning an average or superior score in the test is shown by the cases of S. S. and Gypsy Mary.

S. S. was tested at the age of seven years. He had never been to school, and although his home advantages were excellent, he had had no formal instruction and had never learned to read. The parents beheved, perhaps rightly, that the important needs of childhood, apart from simple moral instruction, are food, fresh air, and freedom for play. Nevertheless, S. earned a mental age of ten years, eight months, and an I Q of 153.

In 1916 a gypsy girl of sixteen years was given the Stanford-Binet test in a clinic in Oakland, California. This girl had been stolen by the gypsies when she was about four years old, had Uved with them continuously until a few days before the test was made, and had never attended a school. The I Q found was approximately 100.^

It is not denied that the cultural status of the home (even apart from heredity) may affect the result of the test to some extent, although the influence has never been accurately determined. If it were considerable,

^ The girl had run away from the gypsies and had told of being kidnaped by them when a child. The gypsies denied her story and stated that she was weak-minded and not responsible for what she said. The mental test was given to determine her competence. As a result of her testimony, she was freed from her gypsy parents and returned to her home State (Montana), where she was placed in school. Within a year she had completed the work of several grades and was ready to enter high school.

we should find a marked rise of I Q in the case of children who had been removed from an inferior to a satisfactory home environment. Our data on this p>oint are not extensive, but of a dozen or more children of this kind whom we have re-tested, not one showed improvement. Two such children, Walter and Frank, have been under observation for several years. Until the ages of five and seven years they lived in an exceptionally poor home. The mother was dull, the father illiterate and a drunkard. Both of the parents died within a year and the boys were adopted by a woman of decidedly more than average ability who treated them as her own sons. At the time of adoption one tested at 73, the other at 82. Four years later the I Q's were 70 and 77. It is a general rule that children of border-line intelligence improve little if at all in I Q as they get older, notwithstanding their increased school experience and the extra attention they receive in special classes.

That the environment of the home affects the result of the test but little is further shown by the fact that occasionally in a very inferior home all of the children except one test low, as would be expected, while that one tests exceptionally high. In one such family (Portuguese) there are three children who test between 76 and 88, while a brother of these tests at 130. The latter is making a very superior record in high school, which he entered at the age of thirteen years. The others have not been able to complete the eighth grade. All have had the same home environment and the same educational opportunities.^

Scales for group testing. To test each year the intelligence of all the children by the Binet method would involve a larger task than the school is Hkely to undertake. There is accordingly a wide field for tests which can be applied to an entire group, or class, at once. The various scales have been devised for this purpose. The group scales are given as written tests and can be applied to an entire class of fifty or more pupils in about an hour. To score the records requires about ten minutes for each pupil, or a total of about five or six hours for a class of average size. This can be done evenings or at odd times. Most group scales have the advantage of requiring little special psychological training either for giving the tests or scoring them. An unfortunate limitation of such scales is that they are not satisfactory in the lower grades where the need for testing is greatest. As measures of intelligence they are probably somewhat less accurate than scales for individual testing, but their obvious advantages make them deserving of wide use with pupils of the upper grades and high school.^

However, no group scale will ever do away with the necessity of individual testing. Rather it makes the need for individual testing more obvious. All the

1 The mental growth curves of two of these children are shown in Fig. 24, p. 156.

2 The Otis Group Tests and the instructions for using them are supplied by The World Book Company, Yonkers, New York.

pupils in the fourth grade and beyond should be given a test by the group method every year, and those whose scores are either very high or very low in the group examination should be given a Binet test. As will be shown later (chapter IV), it is highly desirable that every pupil be given a mental test within the first haK-year of his school life.