Page 320 -

11.4 Analyzing text content 309

11.4.1.2 A priori coding and theoretical frameworks
Theoretical frameworks are commonly used in multiple stages of qualitative research
(Corbin and Strauss, 2014). In the research design stage, theoretical frameworks can
help you frame the research questions, decide on the specific research approach to
adopt (i.e., survey, interview, focus group, etc.), and identify the concepts and ques-
tions to be included in each approach. When analyzing text information, theoretical
frameworks can help you identify the major categories and items that need to be coded
and explain the findings of your research. Therefore, at the beginning of a research
project, it is important to study the research literature and find out whether there is
any theoretical framework related to the research topic that you are investigating.
For example, suppose you interview a number of senior citizens to examine the
major difficulties that they experience when using computers. One question you
would like to answer is the underlying cause of those difficulties. You know a large
proportion of the difficulties can be attributed to the gradual decline of human capa-
bilities. According to well-established literature, human capabilities can be grouped
into three major categories: cognitive, physical, and perceptual abilities. You can use
those three types of capability as the high-level categories of your coding scheme and
try to group the participants' responses in each of those three categories.
In the HCI field, theoretical frameworks are also called taxonomies. Numerous
taxonomies have been developed to help guide research and understand the data col-
lected through various user studies. One example of the earlier taxonomies proposed
is about the types of task that users conduct (Norman, 1991). By grouping tasks into
categories, such as “structured and unstructured” or “regular and intermittent,” and
summarizing the different nature and requirements of each type of task, researchers
and designers can study the interaction in a consistent way and make easier connec-
tions between different aspects of the result. For a comprehensive discussion on task
analysis, see (Courage et al., 2007). Another widely cited taxonomy in the HCI field
groups human errors into mistakes and slips (Norman, 2013). Slips can be further
categorized into capture errors, description errors, data-driven errors, associative-
activation errors, loss-of-activation errors, and mode errors. Each type of slip has
different causes, and different design techniques can be used to help prevent, detect,
and recover from those errors. Well-studied and validated taxonomies can provide
great insights for identifying the potential categories to be included for coding.
In both emergent coding and a priori coding, different coding techniques may be
adopted depending on the nature of the data and the study context. Examples include
magnitude coding, in which codes are associated with qualitative or quantitative as-
sessments of the strength or frequency of the concept; process coding, which uses
gerunds (“-ing” verbs) to identify actions; and a variety of affective coding methods
focused on emotions and values. Saldaña (Saldaña, 2012) provides an in-depth cata-
log of different coding approaches and methodologies.

11.4.1.3 Building a code structure
After the key coding items are identified, they can be organized and presented in a
code list (also called a “nomenclature” or a “codebook”). A nomenclature is a list of