P-, Q-, AND R-TECHNIQUES 171

cf. [116], pp. 347 I)- Each man's deviations from the
average order in regard to visual, auditory, or motor imagery
will then obviously indicate whether he is a marked visual-
izer, audile, or motile, or whether he simply belongs to the
same neutral type as the average person. If by factor-
analysis or other means we have already obtained character-
istic orders for the three special types, his correlations with
those orders will serve to measure the extent to which he
resembles the typical visualizer, audile, or motile. We
may call the average order for the general population the
' general factor for persons'; and the secondary orders for
particular groups or ' types,' ' group-factors for persons'
or more briefly ' type-factors.'

In such an inquiry the more usual method of correlating
the data by tests or traits instead of by persons—' hori-
zontally ' instead of ' vertically,' as Stern once put it
([21], p. 17)—would yield unreliable results, since there
would be no means of equating one man's notion of vivid-
ness with another's. And generally, wherever the assessor
is the same for all the traits of the same person, but
different for different persons, there the method of
correlating by persons would seem the more trustworthy
procedure.1 This was, in fact, one of my chief reasons for
adopting it in the researches just described.

In an ordinary academic research, the investigator can usually
get a single observer (e.g. the class teacher) to assess every person
in the group he is studying. But for investigations in a vocational
institute or an educational clinic, where almost every case comes
from a different school, and has therefore to be reported on by a
different informant, it appeared necessary to adopt the alternative
approach. The method itself was regarded merely as a supple-
mentary device, with manifest limitations and defects, and not in
any way as constituting a * new technique.' Nevertheless, in our
practical work at the Psychologist's Department of the London
County Council and in the Vocational Department of the National
Institute of Industrial Psychology, and later in more theoretical
investigations at the London Day Training College and University
College, my fellow-workers and I found it both practicable and useful.
In particular it appeared specially suited for numerous incidental

i a., for example, [23], p. 251, [53], p. 66.