made for a control experiment to measure the effect of supervision. After the experiment in New York was well under way, it was decided to repeat parts of it in the Boston evening-school classes, but under slightly changed conditions. The study as a whole, therefore, illustrates a wide range of applications of the science of measurement to educational work.

The phases of work selected for measurement were oral and silent reading, English composition, spelling, and ability to carry on a conversation with a stranger. Tabulations of the results will be found in the Appendix. Since the lesson material had no permanent value, only rough measurements and standardization were attempted. There was no time for the repeated trials and experimental modification essential to the production of true standard tests; on the other hand, the tests actually used should have a suggestive value, as showing the type of work in test construction which is easily within the reach of any school system or teacher.

By far the most important problem in connection with any educational endeavor is the adjustment of work to individual differences in ability. A concrete illustration of such adjustment is the problem of grading. For example, an immigrant entering an evening school is ready to be assigned to a class. If he can neither write, read, nor speak English, the problem is simple; he is sent to a beginner's class. If, however, as is often the case, he has already some command of one or more phases of work in English, the

problem becomes complex. In most schools at present the assignments to grades are made on the basis of the principal's or teacher's estimate of ability after a brief and superficial examination. Accordingly, the first questions the present inquiry attempted to answer were, "Is the present grading of pupils efficient—that is, does it operate to produce groups of pupils fairly homogeneous with respect to ability? If not, what are the effects of poor grading, and can measurement be used to improve the situation?"

STANDARDS FOR GRADING PUPILS

1

A comparison of the results by grades shows that in all the abilities measured by the tests there is improvement in grade averages from grade to grade (Table IX1 and Diagram 6); that is, the tests measure abilities which are being changed by class work, so that it becomes possible to define ability in terms of grade standards. Thus (Table X) 2 an immigrant entering the New York evening schools in March, and measured by the tests used in this investigation, should be assigned to a first-year class in reading if in the silent-reading test he answers less than six questions in two minutes, and if his accuracy for the whole test is less than 68 per cent. For these are the scores halfway between the average scores of the first and second years, and form the demarcation line between the two grades. Similarly, a new pupil would be assigned to a 1Appendix, p. 391.

2 Ibid., p. 392.

second-year class in spelling if his score in the spelling tests fell between 24 per cent and 42 per cent, and to a third-year class in composition if his score fell above 64 per cent. These

DIAGRAM 6.-STANDARD SCORES BY GRADES, IN TESTS OF IMMIGRANT CLASSES IN NEW YORK AND BOSTON

scores may seem low, but it should be remembered that the tests were constructed for testing both second-year and third-year classes on a uniform basis.

It is probable also that in addition to the tests in specific academic work there should be a measurement of general mentality. Tests of this character suitable for foreign born illiterate in their own languages will soon be available, and should contribute much to facilitate assignment of new pupils to classes where the work will be within their grasp and of immediate value to them. In other words, the grading of pupils can be easily and quickly accomplished by means of standard tests.

DEFECTS IN PRESENT GRADING

Two important questions will at once arise in the mind of the reader: (1) How far do the average scores for a grade represent real grade standards—that is, to what extent do all second or intermediate-grade classes make the same scores? and (2) In terms of such grade standards, what is the range of ability in classes at present?

The answer to the first of these questions is indicated in Table X. For instance, it will be seen that the score midway between the average scores of first-year and secondyear classes in New York in rate of reading is 6.5, and the score midway between the average scores of the second-year and thirdyear classes is 8.5. All classes whose scores in rate of reading fall between 6.5 and 8.5 may properly be rated as second-year classes in ability. Of ten second-year classes in New York City, four fell within the grade limits

in rate of reading, four in quality of reading, six in spelling, and five in composition. For ten intermediate classes in Boston, the corresponding figures were: rate of reading, three; quality of reading, three; spelling, four; and composition, three. These results show that different second-year classes vary widely in ability from school to school in both cities. Similar results could be shown with other tests and for other grades. At present, therefore, a given grade in an evening school does not stand for a definite level of ability (Diagram 7).

Objective measurement serves also to reveal deviations from grade standards. Thus in Table X, Class G of the New York group is shown to score well above the second-year limits in most subjects, while Class B of the Boston group falls below the second-year limits. The significance of such facts from the point of view of supervisory control should need no comment; but in view of the many misunderstandings which have arisen in regard to measurement, it is necessary to point out that while tests reveal conditions they do not in any way show causes. Thus, on the basis of the results tabulated for these two classes, it would be wrong to make the inference that the teacher of Class G is a teacher of exceptional ability, and that the teacher of Class B is correspondingly poor. The facts in the case may be just the opposite. The scores are merely a result, and at best can do no more than indicate what the pupils did with the tests under the given conditions. Teaching is but one of many