METAPHYSICAL STATUS OF FACTORS 247

inquiry on artistic taste (alluded to above), which was
conducted with the aid of the Listener and a series of wire-
less talks, was partly intended to test the possibility of
social inquiries of this kind; and I think it may be said that
the outcome at least showed the suggestion was feasible.
In their formal character, it will be observed, all such inves-
tigations are essentially researches on the existence and
nature of social c types 5 and * type-factors.'l

1 In an earl7 memorandum that I was asked to submit, when a psycho-
logical committee was formed in connection with the B.B.C., I suggested
that tastes and preferences of radio listeners might be studied by sampling
methods along the lines previously used in psychological and educational
surveys. For the opportunity of collecting data on artistic taste by a direct
appeal to the listening public, I am indebted to my friends, Mr. Charles Siep-
mann (formerly Director of Programme Planning at the B.B.C.) and Miss
Margaret Bulley (who assisted in preparing the test material) : a preliminary
analysis of the results will be found in her book Have Ton Good ^laste P

A somewhat similar survey had been carried out for the L.C.C. on child-
ren's tastes and preferences for films at the cinema; and, at the request of
certain firms, analogous methods were used later on in an endeavour to ascer-
tain the varying attitudes of particular types of customers towards different
types of goods, wrappings, and advertisements. On problems of this kind
a large amount of work has been carried out in America, but rarely with a
factorial technique. G. Gallup has more recently familiarized us with the
notion of sampling popular views in the surveys carried out by the American
Institute of Public Opinion (for a description of the methods employed,
see Katz and Cantril, Sociometry, I, pp. 155-79): ^ would be highly in-
structive to plan one or two inquiries of this sort with a more adequate
sampling and factorial technique.

In all such surveys a number of interesting theoretical issues are involved
to which the factorist might well turn his attention. The most important
are the nature of the best method of sampling and of the best method of
weighting. In this country such problems have chiefly arisen in connexion
with investigations for education authorities. Thus, in an early survey of the
abilities and attainments of the London school population, it was obviously
impracticable to carry out a complete or exhaustive survey (like the census),
or even a survey by simple or random sampling (conforming to the require-
ments of the statistical textbook). Accordingly, a twofold procedure was
attempted which should combine the merits of an intensive £ complete'
survey with those of an extensive * representative * survey. Thec representa-
tive * principle meant the sampling of schools of certain c types' from certain
social strata or districts, and an effort by factorial methods to ascertain the
best weighting for each. Now that education authorities are beginning to
ask for psychological assistance in conducting such reviews, it is essential that
the psychologist should be ready to outline an adequate technique. My own
view is that, in general, the existence of such social, economic, or localc types9