ABSTRACT
THERE are no units comparable to centimetres or grams in terms of which mental abilities can be measured. Hence the proposal of Binet and Simon to express a child's level of intelligence as a Mental Age represented an important advance in testing technique. Stern's Mental Ratio, and Terman's Intelligence Quotient (I.Q.) appeared to provide an even more convenient numerical index of intelligence, one which remained almost constant whatever the age of the child. Moreover this could be applied to adults, if their Mental Ages were divided, not by their Chronological Ages, but by some constant age such as 15 years, at which intellectual growth was believed to reach its maximum. Nevertheless there are many difficulties in the use of these units, and it is unfortunate that they have become so widely known among teachers, doctors, and other nonpsychologists, who do not understand them properly. In this article I wish to point out the various reasons why an I.Q. is not a straightforward index which is constant for each individual. The correlation between two applications of the same test, or of parallel tests, at a short interval often exceeds 0.090; but if different tests are given several years apart the coefficient may fall below 0.60. (Indeed if the first test is given much before the age of 5, there may be little or no agreement with intelligence in later childhood or adulthood.) The Standard Error of an individual I.Q. may be as low as 3 points in the first case, and the biggest alteration will hardly exceed IO points. In the second case, the Standard Error may reach IO points, and some individuals will show alterations as great as 30 points. Nevertheless, measures of intelligence obtained from about 5 years onwards are fairly reliable; since half of a group of children will usually retain the same I.Q.s to within ± 7 points throughout their school careers, and half will fluctuate more widely.
