factors in the situation. The examiner may have given the tests very differently in the two classes, or the classes may have been composed at the outset of pupils of very unlike abilities, or one

DIAGRAM 7.-COMPARISONS OF NEW YORK SECOND-GRADE CLASS SCORES WITH STANDARD SCORES

class may have been tested at the close of a period of direct teaching of the subject matter of the tests, while the other had not such training. No valid interpretation can be attached to the results of tests in the absence of a knowledge of the conditions under which they were given

and scored. But precisely herein lies their value. For it is possible to keep all known conditions under control, and then unexpected variations in class scores will be signals, warnings, that some unusual factor is operating. One of the purposes of measurement is to bring such variations to light. If no steps are taken to make the necessary modifications, the measurement itself is of no value; but in the hands of a competent supervisor, the results from tests supply just the information needed to enable him to perform efficiently his function of investigating such variations, determining their causes, and making the adjustments that may be needed.

A good illustration is furnished by a comparison of the test results for New York and Boston, respectively, given in Table IX. On a simple reading of these, the teaching in New York City would seem to be much more effective than that in Boston. Such a conclusion, however, is entirely unwarranted, for it is known that in New York the tests were given only to a few selected classes, while in Boston all classes were included. Moreover, some of the other conditions under which the tests were given were not the same in the two cities; therefore no legitimate inferences in regard to the relative efficiency of the work in the two cities can be made from a comparison of the figures in the tables. Suppose, however, that the conditions had been made the same in the two cities, and that the evening-school work was under the direction of the same super、

visor: it would then be evident that the work in Boston was not going as successfully as that in New York, and it would be the supervisor's duty to find the reason and take such steps as proved necessary to eliminate the disturbing factors.

The answer to the second of the questions raised above in regard to present conditions of

DIAGRAM 8.-DISTRIBUTION OF INDIVIDUAL SCORES IN SPELLING WITH REFERENCE TO GRADE LIMITS

grading is found by selecting an intermediate class of standard ability and classifying the individual members with reference to the same grade standards (Table XI1 and Diagram 8). 1 Appendix, p. 393.

Thus, 44 per cent of the pupils of a standard second-year class (Class F, Table X) were found to be of first-year grade in spelling, 20 per cent of second-year grade, and 36 per cent of third-year grade. Similar results were obtained for other subjects, for other grades, and for both cities. This wide variation in ability within a single class cannot be said to be a condition peculiar to evening-school work. Survey reports on day-school classes show that a similar condition is a constant feature of all mass instruction. The results mean, simply, that present methods of grading in both types of schools are utterly inadequate. Efficient grading alone cannot solve the problem, however, because even if homogeneous groups are formed by selection after measurement, individuals differ so in rate of progress that in a short time the original condition is reproduced. The remedy is to be sought in new methods of classroom procedure which will permit of greater adaptation of the work of an individual to his measured needs.

QUALITY OF INSTRUCTION

The attempt was made to determine the effect of poor grading-i.e., of including pupils of varying grades of ability within a single class— upon classroom instruction. Two facts are clearly indicated by an analysis of the data secured: first, the membership of evening-school classes is constantly changing; second, those who withdraw are often those whose abilities deviate

most from the average for the class. The data for these conclusions follow.

The problem of maintaining constant enough attendance in evening-school classes to make the results worth while is a serious one. In the absence of state or city compulsion, voluntary attendance is bound to be uncertain and irregular at best. Intellectual work at the close of a long, hard day is not intrinsically inviting, and school exercises have to compete with the various forms of recreation and relaxation which are so plentiful in our big cities. The ties that bind the immigrant to his task have little strength. Ambition and friendship for teacher or classmate are two of the most potent; but dissatisfaction with any of the evening-school conditions, or with the benefits obtained, results in immediate withdrawal. The enrollment and attendance records show that, small as is the number of the foreign born who apply for training as compared with the total number in need of it, the number who actually persist through a considerable interval is smaller still.

For instance, in this investigation tests were given at the beginning and at the end of the inquiry. In New York the interval from the first test to the second was one month, or twelve lesson periods. In Boston the interval was shorter-twenty-one days, or ten lesson periods. Yet in that short interval, if the figures for the two cities are combined, more than 40 per cent of those present for the first test were not present for the second, while of those who took the second