172 THE FACTORS OF THE MIND

problems in education and vocational guidance—e.g. for studying
the preferences of children and adults, for assessing the agreement
among school or university examiners, for estimating the reliability
of psychological observers, for analysing the nature of alleged tem-
peramental and clinical types, and generally for all those inquiries
in which the performances of the persons examined had to be
compared with a subjective rather than an objective standard, i.e.
was itself a set of personal reactions or judgments. Here, however,
we shall be concerned, not so much with the concrete results, as
with the exceedingly instructive controversies to which the proposal
has lately given rise, and with the light those controversies seem to
shed upon the whole nature of factor-analysis.

Criticisms.—The legitimacy of the procedure has recently
become the subject of some debate. Several com-
petent authorities have strongly criticized the extension of
correlation to problems of this type ; others have argued
that it might be developed into a new instrument of research
and made the basis of an " entirely new branch, of psycho-
metry." Objections, I think, were first explicitly raised by
Dr. E. C. Rhodes, investigator to the English Committee
of the International Institute Examinations Inquiry, when,
in 1935, I suggested applying the method to data collected
for the Committee, and so studying the reliability of the
examiners' marking along lines we had previously adopted
in evaluating tests [41] and in reviewing our own College
results.1 About the same time, Professor Thomson, who
had himself used correlations between persons for investi-
gating teachers' marks for essays [55], expressed grave
doubts about any wider extension of the method ([87],
pp. 75-6). He holds that " probably correlations between
persons will be in the general case impossible to calculate "
([132], p. 201); nevertheless, he admits that in certain
special cases, e.g. where the examinee is required to rank
the c tests' (pictures, essays, etc.) in order of preference,
such correlations may be legitimately calculated.

1 The essential idea was to test the presence of a general order of merit,
influencing all the examiners in varying degrees, by looking for a hierarchical
order among the inter correlations. This principle had been followed, not
only for the Binet tests ([41], 1921), but also for junior county scholarship
examinations and university examinations (for the Teacher's Diploma, etc.)
with which I had been connected.