134 PSYCHOLOGICAL METHODS OF TESTING INTELLIGENCE 



exactly with that of the single elementary class Vb 

 that we have accorded special treatment. Hence it 

 appears that when an estimation of intelligence is 

 made with special thoroughness and caution, there 

 exists only a moderate degree of correlation between 

 it and school efficiency. 



TABLE XVII 



Class: U III of a Gymnasium (7th school year). 



Ages of those investigated : 13.5 to 14.5 years. 



Number investigated: 23 (8 others omitted as too old). 



Number of teachers estimating: 3 (Teacher A the principal 

 teacher ) . 



Correlations between estimated intelligence and class-place : 



Teacher A and Class-Place 0.43 0.13 



Teacher B and Class-Place 0.55 -+- 0.12 



Teacher C and Class-Place 0.33 -+- 0.14 



Teachers B and C (combined) and Class-Place 0.49 -+- 0.13 



Teachers A, B and C (combined) and Class-Place 0.45 -+- 0.13 



Intercorrelations of the Estimations : 



Teacher B and Teacher A 0.69 -j- 0.10 



Teacher C and Teacher A 0.65 -j- 0.12 



Teachers B and C (combined) and Teacher A 0.75 -j- 0.10 



The values obtained for all three of the secondary 

 school teachers alike show that it is much easier for 

 the individual teacher in the secondary school to rid 

 himself from the influence of the class arrangement, 

 because this arrangement has not been determined 

 by himself alone. 



The reliability of the result is augmented by the 

 inter cor relation of the series of the teachers. These 

 correlations are, in fact, much higher: highest (0.75) 

 when the estimates of the two supplementary teach- 

 ers are combined and related to the particularly 

 trustworthy estimate of the regular class-teacher. 

 That is to say, then, the estimation of intelligence un- 

 dertaken by the teachers quite independently of one 

 another exhibit a great similarity to one another, 



