P-, Q-, AND R-TECHNIQUES 199

psychology, the human genus; in group psychology, the
particular group selected for assessment (cf. above, p. 113).

Curiously enough, in his earliest discussions of correla-
tions between persons, this was the line of work that
Stephenson most vigorously criticized. As we have seen,
his chief criticism was that we interpreted our results in
terms of " abilities " instead of " types "—in terms of
" universal factors " common to all the persons correlated
instead of " type-factors " limited to particular groups.
It was even said that the method had led us to " deny the
existence of psychological types " ([92], p. 295). Thus,
when he now comes forward to show that " correlations
between persons may themselves be a direct measure of
ability" he is not opposing my original view, but reverting
to itx : for his new application of Q-technique has nothing
to do with < typology,' but solely with the measurement of
the general factor of intelligence.

It is not difficult to see how the double misunderstanding
has arisen. The * two-factor theorems' on which Stephen-

1 I am tempted to say that our procedure seems to fulfil Stephenson's
present requirements much more completely than his own. His article in
its opening paragraph claims to be an illustration of " Q-technique, that is,
the factor-study of persons as variables " and to be " concerned with the
general theory of factor-analysis." Actually, however, there is neither factor-
analysis nor any study of factors as such. His ideal order is not obtained by
factorization of correlations, but is directly deduced from " the logical order
of decretion on which the test was constructed" (p. 32). Although, therefore,
he sets out to show the unique conclusions to be attained by analysing inter-
correlations between persons, in the end he does not correlate -persons at all!
What he correlates is one person's arrangement with an objective standard,
much as Binet compares a child's arrangement of given weights with the order
of heaviness assigned by the balance. The £ standard person,' who is set up
to serve as a reference value " much as length is measured in terms of a yard,"
though several times referred to in the introductory argument, is never
actually derived or used, and seems wholly unnecessary to his procedure. In
our own procedure the " standard personality *—the hypothetical ' typical
individual who serves as a common point of reference*—is a kind of weighted
average, specified by the ( factor-measurements for tests/

At the same time I do not wish for one moment to decry the value of
Stephenson's highly ingenious test. If I understand his figures correctly,
it would seem to yield a decidedly closer agreement with independent
criteria of intelligence than my own, and at the same time to be freer from
the complicating influence of verbal ability.