CHAPTER XVI
A REPLY TO CRITICISMS OF THE METHOD

Problem.—The inquiry that follows has grown out of an
attempt to analyse part of the large collection of psychological
data accumulated during my work for the London County
Council. It is included here because it provides in concrete
form a practical illustration of some of the more contro-
versial problems dealt with in the preceding pages and of
the way I have proposed to meet them. The question with
which it is primarily concerned is the distribution of
temperamental traits. Are temperamental characteristics,
like intellectual characteristics—general intelligence, for
example—distributed in accordance with the normal
curve ? Or does their distribution indicate a number of
independent groups or temperamental types ?

In order to assess each individual's approximation to the
type to which he apparently belongs, I have used a simplified
method of factor-analysis, involving the correlation of
persons. With the aid of the figures so procured, an ideal or
hypothetical graph, as it were, is empirically constructed
for the temperamental type, similar to the ' psychographs'
I have published for intellectual types ; analogous graphs
are obtained for the temperamental characteristics of each
individual: and, since each graph represents a set of
numerical measurements for the same series of traits, the
approximation of the individual's contour to the theoretical
contour can then be measured by a coefficient of correla-
tion.1 The method will thus serve to illustrate by a

1 The use of suck psychographs as a " means of recording or classifying
individuals as members of a given type " was described in an early L.QC.
Report ([35], p. 64, and Figs. IX, 1-4), and has been found exceedingly useful
in work on educational and vocational guidance. The employment of
" f standard personalities*—typical individuals, that is, who are made to
serve as common and constant points of reference M (The Toung Delinquent,