The remainder of this study will show some of the advantages of the percentage definition for fixing the borderlines with a system of tests and the result of applying such an interpretation to the particular problem of delinquency. The advantage in increased definiteness should already be evident. When a person is classed as presumably deficient it will mean that he is in the lowest 0.5% in intellectual development or within the lowest 1.5%, if he is a persistent delinquent.
[5]. Aaron J. Rosanoff. Survey of Mental Disorders in Nassau County, New York. Publication No. 9, National Committee for Mental Hygiene, 1917.
[6]. Emma O. Lundberg. A Social Study of Mental Defectives in New Castle County Delaware. U. S. Dept. of Labor, Children's Bureau, Publication No. 24, 1917, pp. 38.
[7]. This statement in 1906 seems to be the earliest attempt at a quantitative definition of deficiency. As I discovered it after the present monograph was practically completed, it furnishes evidence of the natural tendency of attempts at more exact definition to take the percentage form.
[8]. C. Macfie Campbell. The Sub-Normal Child—A Study of the Children in a Baltimore School District. Mental Hygiene, 1917, I, 96-147.
[9]. Italics mine.
[10]. The report of the Massachusetts Commission on Mental Diseases (Vol. I, p. 198) shows that social agencies systematically using mental tests reported 19.2% as mental cases, while those using examinations only for obvious cases reported 1.3%.
CHAPTER V. ADAPTING THE PERCENTAGE DEFINITION TO THE BINET SCALE
Sufficiently large random groups have not been tested with any development scale to make the determination of the borderline on the scale more than tentative. Such borderlines must be looked upon as temporary descriptions to be used in aiding diagnosis until more data are available. Nevertheless, the percentage method of procedure seems to be an improvement over other plans of stating the borderline. So far as the Binet 1908 scale is concerned, when we supplement Goddard's results with 1500 school children by the data for the lower limits of a random group of 653 15-year-olds which we tested, the limits on the scale for passable intellects defined by the percentage method will be found, I believe, not only more conservative, but more reliable than those in current use. Moreover the intended meaning of such borders becomes clear.