142 NOTE ON METHOD

the data. In some cases, for example, Table 34, there was a cer-
tain amount of agreement despite the diversity of the studies.
In other cases, considerable reliance was placed upon the better
or more general studies, especially for national data upon the
results of the Public Library Inquiry study conducted by the
Survey Research Center of the University of Michigan. Thus,
for example, the proportions of the library clientele in different
educational groups, as reported in Table 10, vary considerably
from study to study, depending upon local circumstances. How-
ever, evaluation of the data in the light of such local conditions
—for example, from the low income areas of Ellsworth? to the
high income area of Cole*—does aid in general interpretation;
and the data of the Survey Research Center provides a core to
which other data can be referred.

The other kind of general interpretation deals with a relation-
ship between two (or more) factors involved in library use. Is
there any correlation between education and library use, or dis-
tance from the library and library use, or sex and the use of
reference and circulation services in the library? Here the locality
of the study would have less effect upon the generalization.
Whereas findings on the incidence of library use vary widely
with the local situation (for example, Montclair against South
Chicago), findings on relationships ordinarily would not. Let
us take, for example, the generalization that education is corre-
lated with library use. Although the percentages of each educa-
tional group vary from study to study depending upon the lo-
cality and upon the definition given to “library use” and upon
other conditions, nevertheless, in every case the use of the library
is greater when the educational level is higher (Table 9). Whether
the local educational level is high or low or the library is much
used or little used does not matter, since it is the relative use by
different educational levels which is crucial. In such a case, when
the generalization is based, not upon a simple frequency distri-
bution, but rather upon a relationship between factors, disparate
studies can contribute to the same conclusion.

*Ellsworth, Chicago, 1937.
*Cole, 1948.