VARIOUS USES OF FACTORS 59

practical problem is again rather different. It is not so much to pre-
dict the future as to decide on the immediate treatment—a matter
to which I shall return in a moment; and I cannot help thinking
that our critics have confused the logical requirements of two

different issues.1

Report of the Psychologist to the London County Council' (1922) a more
detailed factor-analysis of the results of the London Revision of the Binet-
Simon tests was attempted by means of partial correlation and the ' group-
factor method.' (The results are briefly indicated on pp. 184 and 195 of
Mental and Scholastic Tests.} Besides the general factor, we found the same
group-factors as were previously reported in the analysis of scholastic tests
—viz., verbal, numerical, and manual, and, in addition, a memory-factor,
a visual or spatial factor, a relational or reasoning factor, and a series of
factors depending upon special knowledge acquired at home or at school,
e.g. familiarity with money, etc. Several of my educational colleagues who
have criticized the earlier report (including Dr. Hill) evidently assumed that
the object of the factor-analysis was to justify the use of the scale as a test
of intelligence ; that, however, was dealt with on the basis of direct correla-
tion in another section (pp. 199 et seq. and Table XXXI, as cited above).
The purpose of the analysis into group-factors was rather to discover " how
far the scale could be used for the incidental diagnosis of more specialized
mental abilities." The results were by no means wholly favourable to the
Binet scale, particularly in its original form. A similar analysis is being
carried out jointly by Miss Simmins, Miss Davidson, and myself for the new
Terman-Merrill revision ; and this may provide us with an opportunity for
replying more fully to our critics.

1 This confusion seems discernible, not only in the articles by Hill and
Cattell cited above, but also in two critical papers, which I have recently
received, by Dr. R. F. Reed and by Profs. H. A. Reyburn and J. G. Taylor.
Like Hill, Reed regardscf the appeal to factors as inserting the weakest possible
link into a chain that is none too strong." " All the educational predictions
of the factorist," he argues, " are purely hypothetical . . .; he first assumes
that the factor extracted from his tests is identical with what teachers call in-
telligence, and then he assumes that limitations in hereditary intelligence as
thus tested will necessarily limit-school progress, regardless of what the teacher
and the school medical officer can do. Thus, Dr. Burt, having tested
Arthur and found his mental ratio to be less than 85, assures the teacher that
Arthur's case is hopeless and that Arthur is backward for life." But Reed,
like Hill, entirely overlooks the fact that both the ' assumptions* that he
attributes to the factorist have also been verified by direct correlation.
Besides extracting a * general factor' by analysis, I also correlated it, first
with independent assessments of intelligence by competent teachers, and
secondly (in showing that it was largely 4 hereditary') with the intelligence
of the children's parents; and finally, since I hold that factors as such need
not come into the picture, I correlated the intelligence tests directly with the
children's subsequent achievements. As for Arthur and the other children
named in the chapter criticized by Reed, they were kept under observation