ESTIMATION AND TESTING OP FINER GRADATIONS 121 



geneity within the class, not only by excluding chil- 

 dren who are plainly abnormal, but also by limiting 

 the estimation to a definite range of ages. If, for in- 

 stance, in the 5th school year, normally corresponding 

 to ages 10-11 years, the class contains some 13-year- 

 old boys, they should not be included in a rank-order 

 of intelligence, for the teacher is not in a position to 

 determine what portion of the intelligence that they 

 exhibit is to be ascribed to their greater age: that 

 would have to be deducted even if they were com- 

 pared to 11-year-old children. Hence, you must first 

 sort out any class within which you would undertake 

 an estimation of intelligence. It is impossible to lay 

 down any hard and fast rule for this: the range of 

 ages that should be included depends on different 

 circumstances ; a greater latitude of age is permissi- 

 ble in maturer than in the younger years. In gen- 

 eral, the calculations that I shall speak of below in- 

 dicate that some 20 to 25 per cent, of the members of 

 every class must be excluded. 



The arrangement of the pupils on the basis of their 

 intelligence can be by groups or in a serial order. The 

 first of these arrangements is far easier for the 

 teacher. In fact, he is accustomed to classify almost 

 all things that are judged in terms such that four to 

 six classes are formed ; and a similar schema can be 

 applied to intelligence as, for example, in the form : 

 I very high intelligence, II good, III medium, IV 

 slight, V very weak intelligence. Thus, Pearson and 

 those of his followers who have made use of estima- 

 tions of intelligence in their statistical investigations 

 of school children have been content with groupings 

 of this sort (74,76,81). 



