Generally speaking, it is in the cultivation of desirable traits of personality that education can exert its greatest influence upon original nature. The best of training over and above that which is now commonly available in the schools is not capable of adding so very much to a child's intellectual stature. By suitable methods of instruction the dull may be made to learn a little more than they otherwise would, and by the choice of suitable curriculum material they can be equipped for life far better than they usually are, but their ability to carry on complicated intellectual processes remains practically unchanged. And just as dull children are not made bright by good teachers, so bright children are not made dull by poor teachers. In the case of personality traits the situation is very different. There is reason to believe that it is possible to make a child deceitful or truthful, honest or dishonest, fearless or fear-ridden, fair-minded or prejudiced, religious or non-religious, masculine-minded or feminine-minded, almost at will. It is unfortunate that education neglects most the very things it can do best.

Psychometric Methods. Since individual differences are so important for education, and since they cannot properly be taken into account without quantitative methods of measuring them, it is desirable to give the reader a more definite idea of the technique which such measurements employ. There are now hundreds of psychological measuring instruments, commonly known as mental tests, and it is manifestly impossible to describe them all in this brief treatment of the subject. Fortunately most mental tests have certain important principles in common, and these can perhaps be made clear by giving first an account of the methods of measuring intelligence.

When one thinks of the measurement of intelligence, one inevitably thinks of Binet, the great French psychologist who labored so diligently in this field for more than twenty years (1890 to 1911). It is Binet more than anyone else to whom we are indebted for modern methods of intelligence testing. It is true that Galton had preceded him by fifteen years in appreciating the importance of individual differences and in the first crude attempts to measure them by means of tests; also that Kraepelin in Germany and Cattell in America antedated Binet in their entrance into this field. It was the genius of Binet, however, that most clearly envisaged the problem and that gave us the first workable methods of testing intelligence.

The early tests had confined themselves chiefly to the simpler mental functions, such as sensory discrimination, perception, motor coordination, reaction time, and rote memory. Large individual differences were discovered in all these fields, but they were not found to be correlated in any great degree with intelligence in the ordinary sense of that term. Binet, after many explorative researches in which he investigated the relation of these and other traits to intelligence, abandoned tests of the simple mental functions in favor of tests of such complex functions as logical memory, controlled association, comparison of concepts, detection of absurdities, comprehension of difficult situations, definition of abstract terms, problem solving, etc. Such abilities, Binet found, were highly correlated with ability to master the school subjects and with social efficiency. It is in such traits that the recognizably feeble-minded differ most from normal persons, not in the simple sensory, motor, and perceptual functions.

Intelligence testing is nothing more nor less than a sampling of one's intellectual and behavior responses. Galton likened it to the assaying of mineral ores. The effectiveness of the sampling depends upon: (l) whether it is extensive enough to give consistent results, and (2) whether it is a sampling of intelligence or of something else. If a test is extensive enough to give consistent results, it is said to be reliable. If its verdict correlates fairly well with other criteria of intelligence, the test is said to be valid.

The reliability and validity of tests cannot be judged by opinion, either the opinion of the layman or of the psychologist. Both have to be experimentally determined. The results of such determinations are expressed in quantitative terms, usually correlation coefficients. It has been definitely established that the best of the current intelligence tests are reliable in the sense that they give consistent results, and that they are valid in the sense that they discriminate clearly between groups of subjects classified by other reasonable criteria as bright, average, or dull. Whatever misinformed critics may say, the best intelligence tests really do measure with greater or less approximation what we may reasonably call intelligence; that is, ability to make complex adaptations, to think in conceptual terms, and to solve difficult problems.

The limitations of intelligence tests must, however, be kept in mind. These are chiefly of three sorts. In the first place, no intelligence test is a test directly of native ability. It tells us what grade of intelligence the subject has when he is tested, not necessarily what he might have had if his education and cultural background had been different. Any judgment about native intelligence is always a matter of inference. My personal judgment is that the experimental investigations to date do not suggest that the scores on the best intelligence tests are very greatly influenced by ordinary physical defects (such as adenoids or malnutrition) or by moderate differences in regularity of school attendance, in pedagogical skill of teachers, or in the cultural level of parents. At the same time it is freely admitted that extreme differences in regard to such advantages and disadvantages may and probably do, in some degree, invalidate the test result. Ultimately the influence of these and other possible determinants of intelligence will be measured by the test method. In the second place, one must bear in mind that even the best tests are not perfectly consistent, or reliable. Every test score has its probable error. This has been carefully ascertained for all of the better types of tests. In the third place, one must remember that no test samples every aspect of intelligence; one may emphasize ability to deal with abstractions, another the ability to adapt to common sense situations, and another aptitude for working with things.