This article discusses how to use open, axial, and selective coding in the analysis of qualitative data when researchers conduct studies using deductive qualitative analysis (DQA). Unlike grounded theory, DQA begins with preliminary codes that both guide the research and that researchers expect to test and to change in the course of doing the research. This article reports on email exchanges with two students that Jane Gilgun had. Jane is a professor at the University of Minnesota, Twin Cities, USA. the students are Anke Reints, a PhD student at the Vrije Universiteit Brussel, Belgium, and Ben Duncan, a student at Tennessee State University, USA.
1. Coding in Deductive Qualitative Analysis
Jane F. Gilgun, Ph.D., LICSW
Summary
This
article
discusses
how
to
use
open,
axial,
and
selective
coding
in
the
analysis
of
qualitative
data
when
researchers
conduct
studies
using
deductive
qualitative
analysis
(DQA).
Unlike
grounded
theory,
DQA
begins
with
preliminary
codes
that
both
guide
the
research
and
that
researchers
expect
to
test
and
to
change
in
the
course
of
doing
the
research.
This
article
reports
on
email
exchanges
between
Anke
Reints, a PhD
student
at
the
Vrije Universiteit
Brussel, Belgium, and Jane Gilgun, a professor at the University of Minnesota, Twin Cities,
USA.
About
the
Author
Jane
F.
Gilgun,
Ph.D.,
LICSW,
is
a
professor,
School
of
Social
Work,
University
of
Minnesota,
Twin
Cities,
USA.
Professor
Gilgun
has
related
books,
children’s
stories,
&
articles
on
scribd.com,
Kindle,
iBooks,
and
other
on-‐line
booksellers.
2. Coding in Deductive Qualitative Analysis
How to code when doing deductive qualitative analysis (DQA) is a question that many
researchers have. In DQA, researchers begin with theory that guides their research. Sometimes
the theory is a theoretical model, sometimes a set of inter-related hypotheses, and sometimes the
theory is used to guide the research. The initial theory is a source of codes that researchers use to
analyze the data they collect. They also expect the test the viability and usefulness of the codes
and to change at least some of them and add new ones in the course of doing the research.
Negative case analysis is the procedure that guides researchers to look for data that do not
fit with the initial theory. When this happens, researchers change the initial theory to fit their
findings. Deductive qualitative analysis is different from grounded theory in its use of
preliminary theory and coding, but eventually the procedures of both approaches converge in
data analysis and in the writing up of results.
A basic premise in deductive qualitative analysis is that many researchers have theories
they think will help them to focus their research questions. These theories can be based on
combinations of preliminary studies, reviews of research and theory, professional experience and
personal experience. Few dissertation committees and funders of research will approve research
that does not have a well thought out plan of action. Deductive qualitative analysis is responsive
to these issues, while maintaining procedures that allow for the identification of new dimensions
of social phenomena and the concepts and theories that compose them.
Anke Reints, a Ph.D., student in sports psychology at the Vrije Universiteit Brussel,
Belgium, emailed me in January, 2011, to ask about coding in DQA. She had read two articles I
have written on deductive qualitative analysis. One is called “A Primer on Deductive Qualitative
Analysis” and the other is “Qualitative Research and Family Psychology.”
The following is the email exchange that we had. The article Anke refers to is
“Qualitative Research and Family Psychology.”
Dear Jane Gilgun, In your article you mention the possibility of using Strauss and
Corbin’s coding scheme (including open coding, axial coding, and selective coding),
while carrying out DQA. I don’t really understand this, because isn’t it the case that in
DQA you already have set your codes a priori?
In my case, the components of the model I am testing are my codes. So all the
information I gather from the face-to-face interviews are placed under those categories (=
components of the model). When I read about grounded theory, I feel here you create
your own categories. Do you get my confusion?
This is how I answered Anke. I edited this answer for the sake of clarity and
completeness. I did not edit any of the other emails.
Hi, Anke. Good questions. Yes, in DQA you do have prior codes, but you work at trying
to improve the ones you started with and developing new ones if what you see in your
3. research material warrants new codes. Negative case analysis helps you to look for
exceptions to your emerging analysis so that what you come up with is more inclusive
than what you began with. It is easy to find material that supports the prior codes, but it is
just as important in many cases to find material that does not fit into your codes. When
you find material that does not fit your initials codes, you give names to that new
material. In other words, you code the new material. You may also change some of your
initial codes if you have material that supports that.
With grounded theory type of coding, you usually first do open coding, which means you
simply go through the material and mark up the text with any ideas that come to mind. In
deductive qualitative analysis, you also do open coding, but you can do it before or after
you code using your prior codes. When you do negative case analysis, you look for any
material that does not fit emerging new understandings. You then give names to that new
material. In other words, you code that new material.
Negative Cases Analysis
You do negative case analysis in at least two points in your research: while you are still
collecting data and when you are analyzing data.
Negative case analysis during data collection. While collecting data, it is
important to look for and inquire about any exceptions to the general statements that
research informants make and any exceptions to the general description of informant
actions that you are developing. This is within-case negative case analysis. Also, in
choosing your units of analysis—that is, cases to include in your research—it is important
to collect data from persons/situations that differ slightly from the cases you have already
collected data from and on the basis of which you have already developed preliminary
ideas.
Negative case analysis during analysis. When analyzing data, researchers
typically develop a story line/narrative/descriptions of processes that they can show fits
the material they have collected. In developing these narratives, researchers at some
point also look for any material that can show the various patterns that can occur within
these descriptions. Typically, researchers look for material that can add to, undermine,
and even refute what they so far have described.
Through negative case analysis, researchers will produce a description of
processes/concepts that account for patterns, or multiple dimensions of the phenomena of
interest.
DQA and Axial and Selective Coding
Axial coding happens naturally in my experience. This involves seeing
connections between the various codes. You show what connects to what and what
might not connect to other things. You also want to show how they are connected.
Selective coding happens once you have identified the codes or core concepts that
you want to concentrate on. This involves coding again, this time using the codes that
4. stand for the concepts you think are really strong. Some of these codes/concepts can be
prior codes, but could also have some new codes/concepts that you didn't begin with or
you could modify some of the codes you did begin with.
This is Anke’s response to the above email.
Awesome, and yes I do get your point. Just one more (little) thing. Is it true that by using
maximum variability (in my sample I have selected a variety of types), I am following the
idea of negative case analysis. Namely that because there are many different types of
participants, the chance is there that several cases will not fit my prior codes.
The following is my answer.
Yes. The sampling procedures you are thinking of following would give you a variety
that would challenge your coding scheme. Typically negative case analysis involves
doing a series of cases that are similar and then choosing a negative case, but if you are
clear about your sampling, variations are fine.
Discussion
Deductive qualitative analysis is different from how many people think researchers are
supposed to do qualitative research. On the other hand, a surprising number of qualitative
researchers do theory-guided research. They typically are not clear about the procedures they
follow. This article and others I have written on deductive qualitative analysis seeks to articulate
and clarify procedures for doing theory-guided and theory-testing qualitative research.
References
Gilgun, Jane F. (2005). Qualitative research and family psychology. Journal of Family
Psychology,19(1), 40-50.
Gilgun, Jane F. (2009). Deductive qualitative analysis and family theory building.
http://www.scribd.com/doc/26474669/Deductive-Qualitative-Analysis-Family-Theory-Building
Gilgun, Jane F. (2010). A primer on deductive qualitative analysis and family theory building.
http://www.scribd.com/doc/26474669/Deductive-Qualitative-Analysis-Family-Theory-Building
Gilgun, Jane F. (2010). A primer on deductive qualitative analysis: A slideshow.
http://www.scribd.com/doc/40014289/A-Primer-on-Deductive-Qualitative-Analysis-ASlideshow
Gilgun, Jane F. (2010). Case-based research, analytic induction, and theory development: The
future and the past. http://www.amazon.com/Case-Based-Research-Induction-Developmentebook/dp/B004D4YMRQ/ref=sr_1_29?ie=UTF8&m=AG56TWVU5XWC2&s=digitaltext&qid=1295639501&sr=1-29
Gilgun, Jane F. (2010). The intellectual roots of grounded theory.
http://www.amazon.com/Intellectual-Grounded-Qualitative-Research-