Page:Popular Science Monthly Volume 81.djvu/461

for interest and for ability over the whole period from, say, the age of eleven to the age of twenty-one. Thus, in the sample chosen, the combined ranks give:

	Sum of Ranks for Interest, All Three Periods	Sum of Ranks for Ability, All Three Periods
Mathematics	10	7
History	6	5
Literature	6	6
Science	8	12
Music	17	17
Drawing	17	18
Other hand-work	20	19

 

Turning these into positions from 1 to 7, we have:

	General Rank for Interest	General Rank for Ability
Mathematics	4	3
History	1⁠1/2⁠	1
Literature	1⁠1/2⁠	2
Science	3	4
Music	5⁠1/2⁠	5
Drawing	5⁠1/2⁠	6
Other hand-work	7	7

The differences in the order are then 1, ⁠1/2⁠, ⁠1/2⁠, 1, ⁠1/2⁠, ⁠1/2⁠ and 0, their sum being 4.

I have made the calculation for each of the hundred individuals. On the average this sum of differences is approximately 412, and corresponds to a coefficient of correlation of .91. The individual whose interests follow his capacities least closely still shows a substantial resemblance (nearly .5). The correlation between an individual's order of subjects for interest and his order for ability is in fact one of the closest of any that are known. A person's relative interests are an extraordinarily accurate symptom of his relative capacities.

The effect which the errors to which the original reports are subject is on the whole probably to make this obtained degree of resemblance between interest and capacity too low. Errors due to accident, carelessness, and inability to distinguish or to remember slight differences in interest or in capacity, would make the sums of difference in the long run greater—and the degree of resemblance obtained, less—than the true facts, would have given. The only sort of error that could make the obtained resemblance greater than the true fact would be an error whereby either order was falsified to make it more like the other, notably, the possible tendency to rate oneself higher than one should for ability in a subject which one likes. On the whole, the resemblance