36S ILLINOIS STATE ACADEMY OF SCIENCE 



taming the names and scores on tests G-l and G-2 from 

 the testing program described above of these same first 

 grade pupils. Table 9 indicates the results of this com- 

 parison. It is to be read as follows : Twelve pupils, 

 or 26.7% of the total number of first grade pupils, who 

 were judged by their teachers to belong to the middle 

 fifth of their classes, or in other words who were judged 

 by their teachers to be of average mentality, received 

 scores on test G-l indicating that they belonged to the 

 lowest fifth of the class. In test G-2 there were eight 

 pupils, or 17.8 per cent, who were judged to be average 

 by the teachers, who were measured by the test as in the 

 lowest fifth. In the middle fifth where we might expect 

 a high percentage of agreement, if there was a high de- 

 gree of correlation between these methods of measure- 

 ment, we find a low percentage, 17.8 and 28.8. In other 

 words, as far as this experiment goes, it appears that 

 those pupils, who are judged by the teacher to be of 

 average mentality, have about an equal chance of making 

 a score on these intelligence tests that would place them 

 in any fifth of the class. 



As between these two methods of determining mental 

 ability, there is no correlation in this case. Evidently 

 the teachers were not measuring the same thing that the 

 test was supposed to have measured, or, if either method 

 is an adequate measure of mental ability, it cannot be 

 said with confidence that it functioned in this attempt. 

 There are a number of factors that would easily explain 

 the inadequacy of either the test or the teacher in this 

 case. The tests were new tests and not fully developed, 

 they were given to first grade children, the teachers had 

 had these children only two months, the school system 

 in which the test was made was entering upon a period 

 of reorganization, and the teachers, though experienced, 

 were not well trained. Any one of these factors would 

 be sufficient to discount the results, unless they were 

 under control (which they were not) or were verified by 

 other similar experiments and studies. 



