no. 2.] PSYCHOLOGICAL EXAMINING IN THE UNITED STATES ARMY. 



335 



can parentage. In tables 21 and 22 the data are given for 13-year-old and 14-year-old pupils. 

 In each the number includes all the pupils of American parentage of the given age, who were 

 enrolled in the grades 4 to 9 of those schools in which all grades were tested. Certain schools 

 in which not all the grades were tested were excluded from this comparison because it was 

 desired to have the age groups as nearly representative as possible. However, there were 

 probably a few pupils of these ages enrolled in the grades below the third. Such do not enter 

 into this comparison. 



These tables offer striking proof of the validity of examination a as a measure of educa- 

 bility. The pupils of a given age had had equal or approximately equal educational opportu- 

 nity, and their distribution in the grades is due chiefly to differences in native mental ability. 

 The correlations would probably be higher than they are but for certain constant errors in 

 school grading. It is well known, for example, that the dull are usually over promoted and 

 the bright under promoted, a fact which would tend to make the correlations lower than they 

 should be. The data suggest strongly that the high correlation of total score and school grade 

 in the case of soldiers is due, not to the influence of schooling on the test performance, but to 

 the fact that the point at which children are eliminated from school is determined largely by 

 ability. The correlations with grade location for the various ages are as follows: 



