Page 125 -

5.4 Probabilistic sampling 111

inclusion criteria). Each selected user received a personalized survey, asking
them to respond to recommendations made by the social networking software.
During the period of the research study, 415 out of the 500 users logged in, 258
responded to the survey, and when the data was cleaned (due to incomplete or
missing responses), 230 users had submitted valid surveys (Chen et al., 2009).

It is important to note that the sample frame need not be individuals; it can also be
organizations. A long-term survey has documented the level of Internet access in pub-
lic libraries across the United States. See the “Use of Sampling Frames in Studying
Internet Access” sidebar for an example of a random sampling of organizations.

5.4.1 STRATIFICATION
Sometimes a sample can be stratified. A stratified sample is when you divide your
entire population in separate subpopulations (known as strata) and a separate sample
is selected from within each subpopulation (Frankel, 1983). So, when collected, data
analysis can be made for each subpopulation and can be combined for the entire popu-
lation. Stratification can help ensure that you have an appropriate number of responses
from each subset of your user population (Babbie, 1990). Stratification can also help
when the subpopulations have unequal representation, so that the relative distribution
of the subpopulations in the sample can be increased or decreased (Frankel, 1983). A
simple example is a random sample of university students. A random sample of univer-
sity students is likely to have freshmen, sophomores, juniors, and seniors represented;
however, due to enrollment trends often some class years have much larger classes
of students as compared to other years, and therefore, there very likely would be an
unequal number from each class who were randomly selected. However, a stratified
random sample would have an equal number of respondents invited to participate,
from each of those class years. A stratified random sample was used in a study of how
people use technology to keep in touch after they move (see the Stratification sidebar
for more information), so that long-distance moves would be represented more than
local moves (which in fact are the majority of moves).

USE OF SAMPLING FRAMES IN STUDYING INTERNET ACCESS
The American Library Association sponsors a survey on the implementation
and use of the Internet in public libraries in the United States. The earliest
survey was in 1994 and the survey has been repeated on an annual or
biennial basis since then. The most recent survey was in 2012, after
which this survey was folded into the Digital Inclusion Survey (see http://
digitalinclusion.umd.edu/). The survey started out as a paper-based survey
but, over the years, it moved to a web-based survey.

(Continued)