CRITICISMS OF METHODS 375

individuals to another group, on the basis of their resem-
blances and differences in a number of specifiable traits.
Unfortunately, among human beings (as contrasted, for
example, with the fruit-flies and the primroses which have
been so successfully employed for the study of genetic
types) the relevant resemblances and differences are far
from perfect or complete. Nevertheless their amount
may readily be measured by correlating the measurements
obtained from one person with those obtained from others.

As we have seen in the earlier pages of this book, the
device of correlating persons raises many obscure statistical
issues ; and the difficulties are not so easily disposed of as
in the case of correlations between traits. But there can
be little doubt about the practical utility of the procedure.
In clinical work, a psychologist has primarily to analyse the
composition of the concrete individual, not the composition
of abstract abilities or traits; he has to take into account
qualitative characteristics for which there are no standard-
ized tests, quite as much as quantitative characteristics, like
intelligence or educational attainments, which can be readily
measured ; almost inevitably, therefore, he comes to think
in terms of the perceptible resemblances between his
individual cases rather than in terms of supposed affinities
between hypothetical functions or propensities. In an
academic research on educational abilities, the investigator
can ask a teacher to rank or grade the pupils in a single class
in order of merit for the six or seven main subjects of the
school curriculum ; and he can then proceed to correlate
school subjects. But in clinical work on backwardness,
delinquency, vocational guidance, and the like, each fresh
case is reported on by a fresh teacher; and, as we have
already noted, it is much safer to compare judgements by
the same teacher on the different qualities of the same child
than to compare judgements by different teachers on the
same qualities of different children. Consequently, when
the practical psychologist comes to review his material, he
is naturally tempted to begin by arranging his data, not
according to the tests, the school subjects, or the traits
assessed, but according to the several individuals.

Yet, although he has thus altered the customary mode of