{

 

144 NOTE ON METHOD

nique, in general adequacy, and in quality. Only studies with some
pretension to the status of “research” projects were included.
Actually, this criterion was interpreted—and necessarily so—
quite liberally; very few empirical studies in the field would
satisfy all the requirements of scientific method. A special effort
was made at every point to allow for methodological deficiencies
in interpreting the data. The methodological limitations of the
studies are of several kinds:

1. The samples on which the studies were based varied widely.
Many of them are biased in one way or another, and many are
deficient either in validity or reliability, or both. Many were ap-
parently selected through considerations of convenience and ac-
cessibility, rather than adequacy. The samples are deficient be-
cause they include too many well-educated people,® or because
they are weighted with “reading” occupations,” or because they
are composed entirely of people living within a certain distance of
a certain library,® or because they are heavily weighted with
attendants of adult education classes,? or because of other reasons.
In some cases the samples are so limited and so specialized®
that some of their data are not reported.

In this connection it is important to point out that national data
on some aspects of library service are reliably provided by the
recent study of the Survey Research Center. The sample of 1,151
adults in this study was selected through probability sampling

‘methods as described in SRC, 1948, p. 89—95. The sampling error
ranges from about 1.5 percent to about 5.5 percent, depending

upon the variable and the group involved. This survey provides
the best national data on certain questions of public library use.

2. The questions asked in some interviewing studies were in=
sufficiently concrete. Recent experiments have shown that gen-
eral or abstract questions about such prestige-laden activities as
book reading produce inflated answers, and it is now considered
better to ask concrete questions about behavior during recent
days or weeks. The design of questions in several studies was du-
°Link and Hopf, 1946. "Kelley, 1935.
*Ellsworth, St. Louis, 1937. *Lorimer, 1931.
“For example, Gray and Monroe, 1930.