Page 314 -

11.4 Analyzing text content 303

in practice. In the online community study example, the term “community members”
may raise some questions. How do you define community members? Do community
members include all the people who have visited the online community website? If
the answer to this question is “yes,” it means you are interested in examining not only
those visitors who have posted messages, but also the people who have visited the
website but have never posted a message (lurkers). If your data set only consists of the
public messages, then the data comes from a subset of your population (those who have
contributed by posting messages) and, therefore, it is not representative of the overall
population. In this case, it may be more appropriate to restrict the target population of
your study to those people who have posted messages. Examination of the data might
reveal a wide range in the number of messages posted by participants. Some people
might visit and post messages on a daily basis. Some people might only post one or two
messages through the entire year. Do you count those extremely infrequent visitors as
community members? If so, there may be concerns over whether the small number of
postings from those visitors’ limits the accuracy of the depiction of their opinions or
behavior. Other factors that should be considered when defining the population include,
but are not limited to, age, gender, profession, education, and domain experience.
Thirdly, you need to know the specific context of the data. Data analysis out of
context is meaningless and highly biased. Any words, terms, and claims need to be
interpreted in the specific context from which they are extracted. Consideration of
the context is an iterative process, occurring at multiple levels throughout analy-
sis. Before data analysis, you need to have a clear understanding of the higher-level
context of your data set. For example, if you are studying the end-user’s attitude
toward security procedures in the organizational environment, you need to be aware
that the type of business or profession may have a notable impact on the topic. An
employee of a government agency who has access to classified information works
in a very different environment from a staff of an entertainment facility. The govern-
ment worker may have to go through security training on a regular basis while the
staff working in the entertainment industry may have no security-related training.
Therefore, the specific context of their work has great impact on the data that they
provide. If you analyze their input without considering the context, it is like compar-
ing apples with pears and the results are tainted. During the data analysis process,
you need to consider lower-level context, such as the phrase, sentence, or paragraph.
We discuss the interpretation of low-level context in Section 11.4.2.

11.4 ANALYZING TEXT CONTENT
11.4.1 CODING SCHEMES

Analyzing text content involves assigning categories and descriptors to blocks of
text, a process called “coding.” A common misunderstanding is that coding is noth-
ing more than paraphrasing the text and counting the number of key words in the text.
Actually, coding is much more than paraphrasing and key word counts. As stated by
Corbin and Strauss (Corbin and Strauss, 2014), coding “involves interacting with