Miscellanea 
547 
This series is very irregular. The rj for the series is -1553, but this is rather a measure of the 
irregularity than of anything else*. If we take the correlation coefficient for a four-fold table 
as follows 
Infants 
I, 11, III 
IV, V, 
VI 
Totals 
■05 to -JfS 
■50 to 1^00 
71 
67 
53 
58 
124 
125 
Totals 
138 
111 
249 
we find r='0580, a number of scarcely any practical significance and shewing that really place 
in class is, as it should be, independent of standard. 
We have then to sum up the following results : 
Intelligence and Age 
''l3= - 
-1830 
Intelligence and Position in Class . 
■'"12 = 
■7246 
Intelligence and Standard 
''14= - 
•0558 (--2123) 
Position in Class and Ago 
'"23= - 
■0484 
Position in Class and Standard 
'-24 = 
•0580 (-1553) 
Age and Standard . . . . 
■''34 = 
■936g 
In the case of two of these correlations for which some doubt might exist, we have placed in 
brackets the higher values found by different processes. The results for both values of the 
partial correlation coefficient have been worked out. 
Correlation of "General Intelligence" and Examinational Place, for constant age and 
constant standard : 
With lower values of and r-,i : pi2 = ■70 + ^02. 
With higher values of and r^^ : pi2 = '99 ± -01. 
We may safely draw from these results the conclusion that the teacher's estimate of general 
intelligence is not a purely idle character, wholly valueless owing to the personal equation of 
the teacher. Whatever else it may be, it is highly correlated with the place which the student 
will take in an examination test and accordingly has at least one quite definite significance. 
The first of these results shews that with the lesser and more probable values of the correla- 
tions we obtain, allowing for standard and age, a high correlation (-70) between the teacher's 
estimate of general intelligence and actual examination measure of capacity. There are, 
however, cross-currents at work in elementary schools ; the one is the selection of the notably 
dull children about 10 who are drafted into "special schools." This causes rather a defect of 
dull children in the II, III and IV standards ; the next is the selection of the more intelligent 
children in the highest standards (this is most obvious in V) to leave the elementary schools. 
The total effect is to make a somewhat high correlation ratio for standard and intelligence. If 
we took all the children in a class there ought to be sensibly no correlation between position 
in class and standard, but we actually find for classes II, III and IV excess of good places 
(averages ^445, ^393, ■388 instead of -456) and for the high classes V and VI excess of bad places 
(averages ^493, ■493). This seems to be the result of the same cross-currents, the selection of the 
* It must be remembered that not every child, but only 30 to 40 in each standard were physically 
examined, and the number so dealt with is not by any means the number in the class. This selection of 
children accounts for the mean place in class being -456 and not •SOO. 
