336 



MEMOIRS NATIONAL ACADEMY OF SCIENCES. 



[Vol. XV, 



For the 87 13-year boys each test of examination a was correlated with grade location. 

 The results were as follows: 



Teachers' ratings on intelligence for approximately 1,300 California children were corre- 

 lated with examination a. The pupils of each room were rated by the teacher as very superior, 

 superior, average, inferior, or very inferior in intelligence. These ratings were then compared, 

 not with total score of examination a, but with the intelligence quotients. That is, mental age 

 norms were worked out for examination a, and each child's intelligence quotient was ascer- 

 tained by dividing his mental age score by his actual age. Had total score been used instead 

 of intelligence quotient, it would have been necessary to confine each comparison to the children 

 of a single school grade. The use of the intelligence quotient permits the grouping together of 

 children of all ages and grades. The following correlations were computed (Pearson method) : 



School A, grades 3, 4, and 5 combined, r = 0.55. 



School B, grades 3, 4, 5, and 6 combined, r = 0.67 



School C, grades 4 to 8, combined, r = 0.76. 



School D, grade 4, r = 0.57; grade 5, r = 0.62; grade 6, r = 0.61; grade 8, r = 0.47. 



School E, grade 2, r = 0.31; grade 3, r = 0.42; grade 4, r = 0.65; grade 5, r = 0.32; grade 6, 

 r = 0.24; grades 7 and 8 combined, r = 0.29. 



School F, grades 3 and 4, r = 0.50; grades 5 and 6, r = 0.59; grades 9 A and 10 B, r = 0.49. 



School G, grade 4, r = 0.65; grade 5, r = 0.58; grade 6, r = 0.74; grade 7, r = 0.63; grade 8, 

 r=0.38. 



School H, grade 3, r = 0.42; grade 4, r = 0.55; grades 5 and 6, r = 0.51. 



Comparing the agreement with teachers' estimates of intelligence in various school grades 

 it appears that examination a is as valid a measure in the third and fourth grades as in the 

 higher grades. 



Correlations between total score in examination a and teachers' ratings on quality of school 

 work were plotted for approximately forty classes, but were computed only for the following, 

 which were selected as typical: 



Correlations between intelligence quotient for examination a and ratings on school work 

 were computed as follows: Grade 2 (35 cases), r = 0.17; grade 3 (155 cases), r = 0.44; grades 

 3 and 4 combined (120 cases), r = 0.42; grades 4 to 8 (107 cases), r = 0.63. (For the same 107 

 cases, correlation between Stanford-Binet intelligence quotient and teachers' ratings was also 

 0.63). 



For 84 Stanford University students of sophomore to senior grade, each test and total 

 score were correlated with average class mark earned since university entrance. The results 

 were as follows: 



