ESTIMATION AND TESTING OP FINER GRADATIONS 131 



English correlation between estimated intelligence 

 and class-place is doubtless due to the circumstance 

 that the English were content to use a small number 

 of classificatory groups of intelligence, whereas in 

 our lists serial ranking was required. This more dif- 

 ficult task brought out a somewhat higher depend- 

 ence on the school ranking, which was known by the 

 teachers and in fact prepared by them. Hence this 

 very high correlation is not a proper expression of 

 the actual degree of connection between intelligence 

 and school work, as will be seen more clearly upon a 

 closer analysis of the lists. There are certain symp- 

 toms by which one can tell very positively whether 

 the teacher has or has not made the attempt to free 

 himself from the suggestive influence of the school 

 ranking; and the more seriously this attempt was 

 made, the smaller was the correlation. 



Mention must be made in this connection of the 

 Class Vb, whose teacher plainly undertook the work 

 with great independence and with fine psychological 

 comprehension. This teacher settled the numbering 

 for his intelligence series without glancing at his 

 school-work series and sought to explain the cases 

 of special discrepancy between school performance 

 and intelligence by brief remarks ("moved in from 

 the country, ' ' ' l sick a long time, " i i poor home condi- 

 tions, ' ' etc. ) . The result was astonishing a correla- 

 tion of only 0.47. 



This one correlation is, in my opinion, psycholog- 

 ically and methodologically more important than the 

 much higher ones obtained for the other classes, be- 

 cause the lack of higher correlation is certainly due 

 not to any peculiar composition of the Class Vb, but 



