Step
Data analysis process
Description
4
Third cycle coding:
Direct answers to
research questions
This step was similar to Step 1 and Step 2 combined, but
this time I was looking specifically at how participants’
responses helped answer the research questions of the
study. All transcripts were recorded and new categories
aligned to answer the research questions.
5
Synthesizing themes:
Frequency of codes
Run queries in NVivo coding using nodes from Cycles 2
and 3 to identify areas of divergence and convergence
among all participants
6
Grouping and theme
selection
After all cycles of coding were complete and nodes were
combined and synthesized into fewer categories, I
applied the theoretical framework to the findings.
During Step 2, following grounded theory methodology (Charmaz, 2014), I began the
coding process performing one round of first cycle coding. I used NVivo software for coding all
interview transcripts. Once all transcripts were uploaded into NVivo, I began creating codes, or
nodes as they are called in NVivo. This process of first cycle coding involved identifying
sentences, passages, and paragraphs of data and using gerunds instead of nouns as codes. The
reason for using gerunds, as Charmaz (2014) explained, is gerunds help bring to surface
underlying processes that may not be revealed by using nouns as codes.
I moved to second cycle coding to begin developing categories for Step 3 in my process.
During this step, I combined code/nodes from into categories and recombined them to narrow
down to a more manageable number of categories. At this point in the process, I began to look
for areas where codes and/or categories converged or diverged and how and wrote memos
reflecting on ways I could combine further in a cohesive and coherent manner. Nodes that
converged were combined into categories, and nodes that diverged were kept as their own
category. Some categories became dead ends or unviable, and at this point I began to solidify
what would be my final themes.
71
Step 4 was similar to Step 1 and Step 2 combined, but this time I was looking specifically
at how participants’ responses helped answer the research questions of the study. During this
step, I was not looking to create or refine codes; instead, I was looking at what their direct
responses to the research questions were. All transcripts were recoded and new categories
created to align to the research questions.
During Step 5, I ran queries in NVivo using nodes from the second and third cycles to
identify areas of divergence and convergence among all participants. I was hoping to find
extreme examples that may bring insights by nature of being outliers. I did not find any instances
that could be considered extreme divergences, so that become a dead end, and I returned to
looking for commonalities or areas of convergence. This proved more fruitful; thus, I moved to
the next step in the process.
During this sixth and last step, I began reorganizing codes into more condensed
groupings and as themes began to emerge there was a moment when I thought I had too many
themes. I took a couple of days away from the data and came back with an idea for narrowing
down to the three most salient themes that would speak to the research questions and be
interesting to instructional design and technology professionals as new lines of inquiry and
practice. At this point I revisited the theoretical framework to apply it to the findings. Rather than
weave the theoretical framework throughout the findings narrative, it made more sense and
seemed more practical to write a subsection title “Manifestations of Agency” in which I apply
the framework to the findings.
In this section I have detailed my data analysis process and how I arrived to selecting my
final themes. In the next section I introduce the participants, not as sources of data, but rather as
72
people with complex lives and diverse circumstances with HyFlex learning as a common thread
in their lived experiences.
|
