Page 322 -

11.4 Analyzing text content 311

when trying to identify the key concepts and categories. Most of the code items at
level three, four, and five were identified through emergent coding, and so were the
three categories listed under “Societal difficulties.” In contrast, the three categories
under “Human capabilities” were derived from existing theory from psychology and
behavioral science.

11.4.2 CODING THE TEXT
When the data set is not large, which is typically true for interviews, focus groups,
or observations, it is recommended to read the text from beginning to end before
starting to do any coding. During the first round of reading, you may find interesting
issues and feel the urge to write among the text or in the margins. Those activities
should wait until you start the coding. The purpose of this first round of reading is
to immerse you into the life and experience of the participants and get a general,
unbiased idea of the data set before focusing on any specific aspects. After this first
read-through, you should be ready to dive in and start coding.
Inexperienced coders may find it difficult to identify anything interesting (or
anything that is worth being coded) in the data, especially when the coding cat-
egory is not established and they are doing open coding to identify coding catego-
ries or themes. Other coders may experience the other end of the scale: they may
feel that the data is so rich that they need to code almost every word or phrase.
Eventually they may be overwhelmed by the large number of coding items that
they are trying to document. They may be distracted by the less important or even
trivial coding items and fail to identify the most interesting or informative patterns
in the data. In order to avoid both situations, we recommend the following steps
for coding:
1. Look for specific items.
2. Ask questions constantly about the data.
3. Making comparisons constantly at various levels.
We'll discuss these steps in the following sections.

11.4.2.1 Look for key items
While coding the data, specific types of “statements” are more likely to carry valu-
able information. A partial list of such statements is given in Table 11.2. In fact, these
categories might prove useful as codes on their own!
Objectives deliver important information. A user's computer usage behavior and
interaction style is largely affected by the objectives that they want to achieve. If a
user uses a specific application just for entertainment, it may be unrealistic to expect
the user to devote a substantial amount of time to learning how to use the application.
It would be totally different if the application is a critical tool at work.
Words, phrases, and sentences that describe actions are also important. They tell
you what the users do with the specific application or technique. They also tell you
what functions are frequently used and what are less frequently used. Once you detect
an action code in the data, you can follow up on that and examine whether the user