The Interpretation of Test Results with Especial Reference to Race Comparisons

Abstract

The comparison of intelligence of different races depends upon two fundamental conditions. In order that such a comparison shall be valid, the sampling or selection of individuals in the several races must be comparable. In the second place, the environmental conditions, and particularly the education of the races, must be comparable. Both conditions are extremely difficult to obtain. There is enormous individual variation in ability within any racial group. It is, therefore, possible to take samplings of any racial group which will vary between themselves to a very large degree. If a sampling were random, it would only be necessary to make it large enough to balance the chance variations against each other. Samplings, however, are very rarely random. The selection is drawn from groups of persons which contain not a representative sampling, but one which is biased in one direction or the other. For example, the sample often consists of school children within a given range of grades. It is well known that these samples are biased. The mere size of the sample may not, therefore, be sufficient to make it representative. The other requirement is also a difficult one to meet. It may be possible to select individuals of different races who are similar in some particular aspect of their environment, such, for example, as schooling, but the environment of these individuals may differ in other important respects. Even in the case of schooling, a superficial resemblance may turn out to be different in reality. For example, although the number of days of attendance at school may be the same, the curriculum, the preparation and skill of the teacher, and the equipment may differ. In the case of home environment, also, superficial similarities may serve to conceal fundamental differences. For example, the; material features of the home and the economic conditions of the parents may be similar, but the social and family tradition may be radically different. The two conditions which have been mentioned must, of course, obtain together in respect to the same sample. That is, the samples must both be representative and must have been subjected to the same environmental influences. A variation in either of these two factors will invalidate the comparison. The necessity of comparable samplings is obvious. An additional word, however, may be said concerning the necessity of equal environmental influences. It might be supposed that the use of intelligence tests makes unnecessary the observance of this sec-