Staying Open: The use of theoretical codes in grounded theory

By Barney G. Glaser, PhD., Hon. PhD. with the assistance
of Judith A. Holton


Theoretical codes (TCs) are abstract models that
emerge during the sorting and memoing stages of
grounded theory (GT) analysis. They conceptualize the
integration of substantive codes as hypotheses of a theory.
In this article, I explore the importance of their emergence
in the development of a grounded theory and I discuss the
challenge of the researcher in staying open to their
emergence and earned relevance rather than their preconceived
forcing on the theory under development. I
emphasize the importance of GT researchers developing
theoretical sensitivity to a wide range of theoretical
perspectives and their associated codes. It is a skill that all
GT researchers can and should develop.


The full power of grounded theory comes with staying
open to the emergent and to earned relevance when doing
grounded theory (GT). This is especially so with regard to
writing up a GT with emergent theoretical codes (TCs).
Researchers seem to have the most trouble at this stage of
the generating Process – sorting memos and writing up the
theory with emergent TCs. Substantive coding comes
comparatively easily and is exciting, giving the researcher
the exhilarating feeling of discovery. Theoretical coding
does not come easily as an emergent and has a beguiling
mystique. As one PhD student emailed me: “theoretical
codes and interchangeability of indicators were the two
aspects of GT that I found the most difficult to
comprehend.” (Holton email January 26, 2004). Another
GT researcher writes, “The author of this current paper
suggests that theoretical coding perhaps places the most
demand upon the grounded theorist’s creativity” (Cutcliffe,

Theoretical codes are frequently left out of otherwise
quite good GT papers, monographs, and dissertations. The
novice GT researcher finds them hard to understand. This
article begins the process of trouble shooting this problem
by dealing with many facets of theoretical coding and will
consider several sources of difficulty in using TCs. The goal
is to help the GT researcher stay open to the nonforced,
non-preconceived discovery of emergent TCs.

The reader may consider this article hard to
understand unless he/she has read and studied my several
former books. There will be some repetition of the ideas I
have already written, but they will be in the service of
offering new insights regarding TCs. Readers who are
challenged in staying on a substantively abstract level of
conceptualization may find this article even harder.
Keeping researchers on an abstract or conceptual level is
hard – especially for those in nursing, medicine, business
and social work – since they are trained at the accurate
description level. They tend to slip easily into a theoretical
descriptive level as the trained style and practical
considerations of their professional field take other.
Staying open to TCs will help maintain the substantively
conceptual level required by GT and will increase its power.

This article is grounded in my origination of GT, in
supervising many, many GT researches and dissertations,
in reading many dissertations and GT monographs and in
intense study of noted QDA methodology books. It is
grounded in the hard study of the above caches. It is NOT
a “think up” article. It is grounded in what is going on in
GT research. The focus of this article, as is my many
books, is to help researchers get GT research done –
achieve GT products that receive the rewards of PhD
degree and career moves. It is not an epistemological
rhetorical wrestle that gets wordy and goes nowhere.
People are doing GTs all over the world and GT
methodology helps them achieve their product.
Epistemological discussions are of no potential help to the
actual doing of research. Rather, they can easily have the
negative effects of sowing doubt in the emergence of
categories and causing premature judgements of

As I have defined previously, “Theoretical codes
conceptualize how the substantive codes of a research may
relate to each other as hypotheses to be integrated into a
theory. They, like substantive codes, are emergent: they
weave the fractured story back together again. Without
substantive codes they are empty abstractions.” (Glaser,
1978) TCs are abstract models, allowing the researcher to
talk substantively of categories and properties while
thinking conceptually. The important point is that the
reader should develop a clear notion of their conscious use
and relevance in generating theory. Then she/he can use,
with theoretical sensitivity, an emergent theoretical code or
codes to put a theory together. This consciousness can
help in staying open. Reading my previous books will help
achieve this abstract level. TC abstraction and use come
with GT experience over many researches. It is part of
the experiential growth of GT skill development. This
abstraction avoids the flat, descriptive and often superficial
presentations of QDA products.

Staying Open

Staying open to the emergent, earned relevance of
theoretical codes is the point of this article. Repetitions
that come from sections in Theoretical Sensitivity and
Doing Grounded Theory (Glaser, 1978, 1998) are in the
service of this goal. Staying open to earned relevance
means that theoretical codes are not to be forced by
disciplines, supervisors or pet codes. Trusting to
emergence and one’s own theoretical sensitivity is

For the researcher, staying open to earned relevance
of TCs means being open to the fullest possible array of
TCs. The researcher must learn and master sensitivity to
as many TCs as possible. The more TCs the researcher
learns, the more this requirement becomes moot. There
are hundreds. The lists in Theoretical Sensitivity and Doing
Grounded Theory
(Glaser, 1978, 1998) offer the most
frequently used and familiar ones, but they are a small list
compared to the possible number of TCs to which one can
be open by perusing the literature of many scientific fields.

GT is NOT a methodology guided by one theoretical
perspective and its TCs. GT is a general method, based on
a concept-indicator model that can use any TC derived
from any theoretical perspective. This theme is hard to
sustain in actual research. It is not easy to stay open
because of previous training, the tremendous grab of some
TCs – e.g. basic social process – and the tendency to cling
to a particular theoretical perspective and its attendant
idols or great men—e.g. symbolic interaction. The
researcher sees what he has been trained to see. Breaking
out to being open takes time and is hard both personally
and in a framed research context. I realize that what I am
saying is easier said than done. But it can be done. Many
do. The basic idea is to become open and sensitive to the
emergent, earned relevance of TCs. The procedure is to
stop preconceived forcing based on discipline, supervisors,
pet codes, a “grande” perspective and unwarranted

Hard To Stay Open

Staying open is not easy. It is hard. Most people
attempt a GT research framed, or inculcated in a
theoretical framework, either consciously or unawares.
Perhaps it is hard to truly become open, but it is quite
possible as GT procedures from start to finish are designed
to open up the researcher and keep her/him open to the
emergent and to earned relevance. When the researcher
gets the point, GT procedures provide ways to perpetually
suspend the frameworks of any forcing theoretical
perspective in favour of what substantive and TCs emerge.
Staying open then becomes relatively easy. Not knowing
before the emergent becomes fun and discovery exciting.

Most GT researchers I have read to date get the
staying open point easily for substantive coding, but not for
TCs. They miss the point for TCs for failure to study them,
thus not becoming sensitive to what TC might emerge.
Rather, they use the TC of their theoretical perspective of
trained origin. In restricting TCs to their field of origin,
they miss possible emergent TCs by not being sensitive to
a fuller array of them.

One normal block to staying open is to describe GT by
a popular TC “as if” GT research always yields that TC. “I
have often described grounded theory as an explanation of
some underlying basic social process, and so I guess, in my
mind, the development of a GT is really a qualitative causal
modelling process” (Olsen email March 7, 2003, Institute
for Qualitative Methodology). To be sure, basic social
processes (BSPs) frequently emerge and are pervasive, but
not always, as I clearly said in Theoretical Sensitivity
(Glaser, 1978). In fact, in our now famous book,
Awareness of Dying, the core category was a typology of
dying expectations (Glaser & Strauss, 1965).

In The Grounded Perspective II: Description’s
Remodeling of Grounded Theory Methodology
, I detailed at
length the remodeling of GT by the QDA methodologists
(Glaser, 2003). GT has been used to “jargon up” QDA
methodology and, in the bargain; TCs are caught up in the
method mix jargon. QDA methodology stultifies GT.
Staying open to a full array of sensitively emergent TCs is
restricted to the author’s forced theoretical perspective,
frequently symbolic interaction or systems theory. TCs
become “assumed” by the framed researcher.

Staying open to whatever TC is relevant is the goal in
my effort to extricate the forcing of TCs by the qualitative
methodologists and their “grande” theoretical perspectives.
There is nothing wrong with using structural or symbolic
interactional TCs if they earn relevance, but my effort is to
stop the ascendant default remodelling caused by the
routine forcing of TCs. I especially wish to stop, or at least
curb, the use of a TC to remodel GT to another QDA
method. For example, using Strauss’s conditional matrix
“as if” always relevant and irrespective is pure forcing.
One reads of Strauss’s conditional matrix everywhere in the
QDA literature. Remember, GT is a general methodology
than can use any data and therefore any TC.

Milliken and Schreiber argue for the generality of GT
when they write about the epistemology of GT (Milliken &
Schreiber, 2001). They say, “Epistemology has been
defined more loosely in sociology to encompass the
methods of scientific inquiry used to study knowledge.
Thus, epistemology can be seen both as a philosophy of
human knowing and how one learns about it. Inherent in
different epistemologies are different assumptions and
beliefs about the nature of know, of what can be known,
and who can be the knower “. In applying these thoughts
to GT, they say: “In contrast to quantitative methods, in
which the researcher is the expert, in grounded theory the
researcher defers to the experience of the participant, who
has experience with the phenomenon of study. The
researcher’s job is to investigate the socially constructed
meanings that form the participants’ realities and the
behaviors that flow from these meanings. That is, we want
to know how they understand and act within their worlds.
What can be known of the covert and overt behavior of
participants is negotiated between the researcher and
participant, toward a shared understanding. Clearly, in our
view, the epistemology of grounded theory is steeped in
symbolic interaction.” (Milliken & Schreiber, 2001), p.180)

This view is patently wrong. It is pure QDA rhetoric in
the quest of worrisome accuracy (Glaser, 2002). It
neglects conceptualization. It uses a “grande” theoretical
perspective and its TCs to define GT, thus denying that GT
is a general method that can use any type of data and the
TCs of any theoretical perspective. GT searches for the
latent patterns in any type of data to articulate a grounded
theory. Latent patterns are everywhere and all is data for
GT including the use of any TC from whatever perspective.
To be sure, interactionally constructed data exists BUT it
only a piece or one type of the data used in GT studies. To
be sure, GT as a general method picks up constructed data
in many studies these days, but these researchers must
transcend the data type to see the general use of GT
methodology and enrich their research by using “all as
data” (Glaser, 1998). GT does not need a “grande”
epistemology, as such, to justify its use. It is based on a
latent structure analysis approach using a conceptindicator
model yielding emergent theoretical frameworks to which
the researcher must stay open.

Two experienced grounded theorists express the
staying open requirement well. Phyllis Stern says
“theoretical coding…simply means applying a variety of
analytic schemes to the data to enhance their abstraction”
(Stern, 1980). Holly Skodol Wilson says,”Theoretical codes
are the ways in which substantive codes and data they
express are interrelated. There are innumerable families
of theoretical codes. All are ways of relating variables
theoretically. I attempted to discover multiple and varied
relationships between and among concepts. Such an
approach is designed to yield molecular rather than linear
theoretical models”. (Skodol Wilson, 1977). Thus, the true
nature of TCs has been around for many years and cannot
be allowed to be remodeled by a single theoretical
perspective as others, especially the QDA methodologists,
would try.

Theoretical codes come from all fields and their
theoretical perspectives, whether social psychology,
sociology, philosophy, organizational theory, economics,
political science, history, biochemistry, etc. Staying open
to TCs from these fields is very enriching of GT. For
example, the random walk TC from biochemistry is very
useful in GT. Conjunctural causation from political science
is an eye opener for GT.

Staying open to what can emerge can be turned in on
itself, however, “as if” to be open somehow cannot be
based on the researcher’s ability to suspend knowledge.
This inability is seen as routine and unavoidable and to be
expected of expert knowledge. Katherine May argues that
expert knowledge in qualitative research consists of an
exquisitely tuned capacity to know where to look and the
ability to ferret out similarities and differences based on
experience. Although entering the field with as open a
mind as possible has advantages, she contends that her
experience in the health care arena was an undeniable
asset. She says “expert analysts are virtually always
informed by extant knowledge and use this knowledge as if
it were another informant” (May, 1994). Thus, her view is
that staying open is not possible for the learned and that,
alternatively, experienced preconceptions are useful. Thus
she implies that experienced researchers get formed in
their field and cannot transcend their experienced view.
They see it everywhere, rather than staying open. I say
not so! Experienced people are more able to suspend their
knowledge of a literature and research field based on their
skilled, competent research ability to stay in control of
perceptions and thereby stay open. They can spot
preconceptions both substantively and for TCs quite easily,
since they are more aware (Morse, 1994). While it is easy
for the novice researcher to be open due to lack of
knowledge (Glaser, 2003), it can be just as easy for the
experienced researcher – if not more so – based on
awareness of more subtle forcing.

Learning TCs

By now the reader may be throwing up his/her hands
and feeling that she/he cannot stay open; that it is too
hard to leave the stability, comfort and safety of the
cherished, learned and trusted TCs of their field. Not so!
They are not to be given up. They are to be extended by
learning more TCs, by being sensitive to these and then
letting earned relevance dictates their use. Staying open to
emergent TCs requires learning as many as possible so the
researcher is sensitive to what may earn relevance.

First of all, the researcher should study TCs beyond the
boundaries of his current discipline and keep studying
them. It never ends. There are so many. Learn as many
as possible. The possibilities are endless. As Hans
Thelesius wrote me, “Theoretical codes are tricky and I
have more to learn there for sure”. (Thulesius email,
December 14, 2002). He is open to the endless task and
its possible difficulties.

Start with the TCs I have listed in Theoretical
and in Doing Grounded Theory (Glaser, 1978,
1998). They are exciting to learn because of their abstract
view of data. Take time to assimilate them when they
seem difficult to grasp quickly. The wider the array of TCs
that one learns, the less the tendency to force a pet or
discipline TC on a substantive theory and the easier it is to
stay open and sensitive to the emergent.

The excitement of learning TCs is well put by Walter
Fernandez when he says, quite rightly, “Theoretical coding
conceptualizes how the substantive codes are interrelated
by generating hypotheses that are then integrated into a
theory. The grounded integration of concepts is a flexible
activity that provides a broad picture and new perspectives.
The theoretical flexibility, however, must remain grounded
on data. The concept of flexibility implies theoretical
sensitivity to a number of possible coding paradigms, or
coding families, consciously avoiding over-focusing on one
possible explanation. Glaser (1978, 1998) provides a
comprehensive (but not definitive) list of code families
allowing for this flexibility” (Fernandez, 2003). Fernandez
then provides his reader with a two-page chart of 26 TC
families. Each family includes several TCs. The list is taken
from my books. Being sensitive to all of these possible TCs
immensely increases the researcher’s ability to stay open.
Staying open to the emergent is what Fernandez means by
“flexibility”, while he insists on earned relevance.

The more TCs a researcher learns, the less the
tendency to derail a GT into a routine QDA by diluting the
GT with a pet or discipline TC – e.g., its all constructed
interaction or the conditional matrix – which is so, so
wrong (Glaser, 2003). There is no argument for the routine
discipline use of a TC for, by consequence, it closes staying
open. Stern and Schreiber say, the researcher using GT
needs to exercise care to avoid a departure from the intent
of the authors who developed it, Glaser and Strauss. In
short, there are a number of variations in doing GT, all of
which are acceptable. On the other hand, there are a lot of
wrong ways of doing it”. (Schreiber & Stern, 2001)

Imposing TCs is a wrong way of doing GT. Earned
relevance of one or a mix of TCs is the acceptable way.
There is no “for or against” argument for the discipline TCs
as they are just some of many that may emerge. This is
the GT procedure: Let TCs emerge in mature memos and in
sorting. Do not worry about results and remember – no GT
is better than the skill development of the researcher and,
in the bargain, no TC is better than what the researcher is
sensitive to – unless it is forced. TCs, like substantive
codes, are a result of the researcher’s learning curve.

The TC learning curve requires the study of many fields
and their theoretical perspectives. In Doing Grounded
, I said, “the fact that many do not use or
understand TCs simply means that they should start
learning them. One reads theories in any field and tries to
figure out the theoretical models being used. It is a fun
exercise. It is a challenge to penetrate the patterns of
latent logic in other’s writings. It makes the researcher
sensitive to many codes and how they are used. He or she
should take the time it takes to understand as many
theoretical codes as possible by reading research literature
also. This is a very important part of developing theoretical
sensitivity” (Glaser, 1998). Skimming and dipping in papers
for TCs from other fields is fun and easy. They pop up. Let
me give some examples.

In perusing a biochemical paper, I came upon the
“random walk” model. This means all variables are in
unorganized flux until one crucial variable is introduced and
then, all of a sudden, all the variables fall into stable
organization. This is highly applicable to social life and
action. People mixing around and visiting in all directions
before a meeting, suddenly come to order when a host,
teacher, or lecturer appears. It happens in fancy
seminars, courts, staff meetings, and in kindergarten
classes. In some cases, a gavel is pounded and “come to
order” is announced. The formal and sentimental order of
the occasion is produced almost immediately.

Another powerful TC that comes from economics is
“amplifying casual looping.” This is part of the interaction
of effects family. As consequences become continually
causes and causes continually consequences, one sees
either worsening progressions or escalating severity. This
applies to spousal power abuse or authority power abuse
as the abuse gets worse. It applies to increasing
organizational failure. It applies to falling in love. I am
sure the reader can now see more possible applications.
Causal looping amplified in either direction – positive or
negative. This TC integrates substantive codes nicely,
when it emerges. It applies to the bullying self-
socialization phenomenon that we saw in the Columbine
massacre (Gisburne, 2003). For additional economic
models, see Frederic S. Lee, “Theory Creation and the
Methodological Foundation of Post Keynesian Economics”
(Lee, 2002). Lee focuses on repeatable causation and
mechanisms thereof.

Yet another powerful TC – “conjunctural causation” comes
from political science. Ragin (1987) explains it
clearly: “The other characteristic form of the problem of
order-in-complexity concerns the difficulty involved in
assessing causal complexity, especially multiple
conjunctural causation. When an outcome results from
several different combinations of conditions, it is not easy
to identify the decisive causal combinations across a range
of cases, especially when the patterns are confounded”.
The problem is not to specify a single causal – consequence
model using Strauss’s conditional matrix. The problem is
to determine the character of more complex causal models
that exist in the substantive data. And many causes may
not be relevant; only high impact causes have earned

My three examples show how complex causal models
that emerge can provide integration of substantive codes
that go far beyond simple causation that is forced “as
appropriate”. The reader will find it fun to skim theories
from other fields to pick up their TCs and thereby open
themselves up to many TCs, assimilating and becoming
sensitive to their particular meaning. The more this is
done, the more the researcher will have the realization that
the number of TCs is endless and yet to be named and that
staying open and sensitive to whatever TC emerges is the
only way to do GT. In the alternative, it is a pure shut
down to remodel GT by saying it has only one theoretical
perspective. This learning approach to TCs solves the
problem that Marjorie MacDonald neatly articulates – the
almost total absence of theoretical codes in current nursing
GT research due to a lack of integrating the macro and
micro levels of social action (Schreiber & Stern, 2001).

TCs are Slippery

As I have said above, theoretical coding is the least
understood aspect of generating GT. When GT is used
merely as a legitimating jargon to QDA, then of course,
understanding TCs is a moot issue. But when the
researcher is genuinely trying to do GT, the first confusion
is the general idea of theoretical coding of the data for
substantive categories and TC models with TCs. This is an
unfortunate terminological confusion. Both types of codes
emerge in memos. They occur in mixes, and TC mixes are
often the integrative picture that fits and works. For
example, a causal model can easily be mixed with a zone of
tolerance and two outside cutting points. Learning TCs
emphasizes the earned relevance of these mixes as they
model substantive codes. The possibilities are not as
infinite as it might seem; they are grounded empirically.

Unlike substantive codes, the underlying
“groundedness” of a TC is less clear, since they are
abstract models of integration based on best fit. Their fit is
not as underlying tight with the data as a substantive code.
Their organization of a theory is not wrong so much as
variable, for an abstract level can have alternatives;
whereas the grounding comes out in the work, fit and
relevance of substantive codes.

This “slipperiness” often results in confusion,
depression and anxiety over non-emergence or the best
way of integrating. Commitment to one model is seen as
“dangerous”. Of course, best fit is required in TC
emergence, but given the ready modification of a GT in the
hands of others, the TC model can easily get adjusted,
changed or corrected. The slipperiness of abstract TCs is a
power. Using a theoretical code is not dangerous; it
formulates the confusion around putting the GT into
writing. This is why forcing a TC is often a tendency and a
premature way out of the confusion of waiting and working
for the TC of earned relevance. It is best to let the TC
emerge. Forcing leads to familiarity within a discipline but
also to irrelevancies. For example, every GT is not a BSP
(basic social process) and, rich as this TC is, forcing stages
on a theory can dilute its fit, work and relevance.

The goal of a GT researcher is to develop a repertoire
of as many theoretical codes as possible. There could be
hundreds. The more theoretical codes the researcher
learns, the more she/he has the variability of seeing them
emerge and fitting them to the theory. They empower an
ability to generate theory and keep its conceptual level.

Theoretical Coding: Substantive Codes vs.
Theoretical Codes

To revisit what I have been saying: “If and when the
researcher gets beyond substantive coding and a full memo
bank, he begins to sort and then he will use emergent
theoretical codes, explicit or implicit, to integrate his
theory.” However, “there is confusion between substantive
codes and TCs among some researchers” (Glaser, 1998).
Needless to say, substantive codes are the categories and
properties of the theory that emerges from and
conceptually images the substantive area being
researched. They are used to build the conceptual theory,
but are not theoretical codes. This is a bit confusing to
some, especially those with little or no theoretical training.

In contrast, theoretical codes implicitly conceptualize
how the substantive codes will relate to each other as a
modeled, interrelated, multivariate hypothesis in
accounting for resolving the main concern. They are
emergent and weave the fractured substantive story turned
into substantive concepts – back into an organized
theory. They provide the models for theory generation and
emerge during later coding, memoing and especially in
sorting. Theoretical codes must also pattern out to be
verified and provide grounded integration.

“Without substantive codes, theoretical codes are
empty abstractions; but substantive codes can be related
without theoretical codes. The result, however, is usually
confused, theoretically unclear, and/or typically connected
by descriptive topics but going nowhere theoretically. It is
the interaction between substantive and theoretical coding
which characterizes GT as an analytic inductive research
methodology rather than conceptual journalism” (Glaser,
1998), p.164). This statement is simple enough to say but
leads to confusion in many ways. Everyone understands
substantive coding, but TCs, and how to code for them, are
not well understood. TCs are confused with substantive
codes on a conceptual level, by similar words, in mixing,
and in research action, calling it theoretical coding for both
types of codes, and just missing the TC involved.

Everyone loves and understands the constant
comparative method for generating substantive categories
and their properties. Their discovery produces a high with
tremendous grab for the researcher. As one researcher
wrote me, “your phrase ’fluctuating networks’ has really
grabbed my attention. Thanks for these little flashes of
brilliance” (Holton email June 9, 2003). But this joy and
grab is not so for TCs, except for perhaps discovery of a
BSP. TCs are often ignored; left implicit or just plain
missed and not understood. Researchers generate
categories naming latent patterns all the time. The
patterns are about social action and recognized in life by
the naming with a category. The same researchers often
do not systematically generate TCs except to mumble at
times cause, consequence or process. The reason is
simple. Substantive categories grab by denoting
recognizable patterns whereas TCs seldom have this grab
since they denote abstract models that are usually implicit
in the theory, not consciously used and seldom explicitly
mentioned. Another source of mentioning a TC nonpurposely
occurs when it is virtually the same as the
substantive category, such a balancing or process.

Thus, it is clear that substantive and theoretical codes
are on a different conceptual level of abstraction and TCs
are a more abstract level since they model the integration
of substantive concepts. Thus, substantive codes and
theoretical codes not only differ in abstract level but in
kind. Substantive codes refer to latent patterns and TCs
refer to models. However, many confuse the two types of
codes in different ways by mixes that take figuring out.

First, TCs are confused with core variable in many
writings. A core variable may be TC’d but it is not the core.
For example, becoming or cultivating may be a core
substantive code and they are basic social processes; but
the basic social process is not the core. It is just a TC that
models the substantive code. Jan Morse clearly makes this
confusion when she says, “The theory (GT) is …usually
organized around a central theme (basic social processes or
core variable/categories). Can the theory have two or more
competing major basic processes or major core
variables/categories? Perhaps, but this is rarely seen. The
basic social process or core variables/categories appear to
serve the purpose of focusing the researcher….” (Schreiber
& Stern, 2001). Clearly, she confuses the model with the

Morse also, in the above citation, confuses the level of
GT by mixing the substantive with the theoretical code.
She says, “The theory is usually categorized as mid-range”
to paraphrase Merton’s notion of middle range. This is
patently incorrect. A GT can be generated at any level
varying from a very specific grounding to the general
implications of a substantive theory to high level formal
theory. For example (and there are many), a very
grounded theory of cautionary control generated in the
study of dentists dealing with HIV patients has much
general application to cautionary control in all dentistry and
medicine. Indeed, it can be turned into a high level formal
theory dealing with cautionary control policy and action in
all of society as it seeks to protect its citizens. In short, it
is up to the researcher to choose the level of his GT. But to
be sure, increasing the level of a GT does not just come by
forcing a TC on it like “conceptualization” a – popular QDA
strategy these days.

Ian Dey offers another “authoritative” but confusing
description of theoretical and substantive codes (Dey,
1999). I say “authoritative” as Dey talks with nothing but
self-styled authority. The reader can, if he wishes, figure
out the confusion. I offer it merely as another example:
“First, the distinction between substantive and theoretical
coding is not very clear. Glaser presents theoretical coding
as “implicit” in substantive coding; suggesting that in doing
the latter, one is inevitably engaged in the former. He
presents theoretical coding itself as a separate activity –
that of relating the substantive categories. One question
this raises is whether categories at some level can be
identified which do not already involve some theoretical
elements, for example, such as causation, process, degree
and soon. Do categories “stand by themselves” or are they
not always part of a broader concretization that already
implies relationship among the categories?” (Dey, 1999,
p.108) He then asks two questions about theoretical
coding. “Is theoretical coding an aspect of substantive
coding or a separate activity?” and “How do we select
among theoretical codes that all fit the data?”

These comments by Dey are too descriptive, in which
in pure data everything is involved at once. GT abstracts
out of data substantive categories and theoretical codes
separately. On the abstract level, the two types of codes
are quite different. Also, since he is descriptive and not
following GT procedures, he does know about sorting and
how by sorting a TC emerges that integrates. Dey asks
the question, “Do processes divide naturally into stages, or
is this rather a construct used by the analyst to order
events?” It is not either/or. It is empirically both or only
one source of a process may emerge. If a few TCs emerge,
they can be mixed or the researcher can choose the one he
thinks best articulates the theory. It is his autonomy to
choose which of the emergent and further, it is just
conceptual theory that can be modified, not QDA accurate
description with its concern for worrisome accuracy. At
least the theory is grounded as best possible, NOT
conjectured out of a fertile, reifying mind.

In sum, Dey is not aware of the abstract nature of GT,
being firmly entrenched in the QDA methodology.
Therefore, his ability to discuss GT issues is nil, since it is
on the descriptive level. He has no sense of GT
abstraction. He is using GT jargon on the data level of
description, leading to multiple views and worrisome
accuracy and this “allows” him to doubt GT as a method.
This article and my many books on the GT perspective
easily allow us to discount his binary analysis (good vs.
bad) as not relevant to GT as an abstracting methodology.
His work is a classic case of remodeling GT to a QDA
method. On the abstract level, the distinction between
substantive coding and theoretical coding (modeling) is
easy. On the descriptive level, the distinctions are easily

Are TCs Necessary?

The answer is “no”, but a GT is best when they are
used. TCs help. TCs are always implicitly there even when
not consciously used. But a GT will appear more plausible,
more relevant and more enhanced when integrated and
modelled by an emergent TC. The hypotheses will be
clearer and stand in relief. TCs avoid the superficiality of
QDA methods. Using a TC at the later stages of memoing
makes generating substantive categories and their
properties easier and the resulting theory more complex
and multivariate. TCs are always latent in the substantive
coding, but being sensitive to enough TCs to see one
emerge helps theoretical sampling, theoretical saturation,
delimiting the theory and reaching theoretical
completeness because the TC becomes an emergent
guiding framework.

Of course, the researcher can analyze without an
emerging TC framework, but it is harder. Applying the
emerging TC framework is of great help in the ensuing
analysis. Actually, it is hard not to apply a TC framework
but be cautious. The TC must emerge and not be forced.
Categories and their properties emerge easier when one
can see their relation to other categories within a
framework. Then, memoing on the relations between
categories becomes easier also as the memos capture the
theory with a TC model.

In conclusion, while not necessary, the need for a TC is
great in generating a GT. It is easy, by prior training, to
force one on the theory as a framing tendency. I can only
counsel to let it emerge. For example, every study is NOT
a BSP. John Cutcliffe says this clearly, if somewhat over
strongly: “Few would argue that substantive coding is an
integral part of data analysis within grounded theory, but if
the intellectual rigor halts at substantive coding then it is
debatable that the researcher used a grounded theory
methodology. The author of the current paper would argue
not. Glaser (1978) argues that it is the theoretical coding,
the conceptualization of how the substantive codes may
relate to each other as hypotheses, which enable the
substantive codes to be integrated into a theory. It is the
theoretical coding that can provide the full rich
understanding of the social processes and human
interactions that are being studied. The author of this
current paper suggests that theoretical coding perhaps
places the most demand upon the grounded theorist’s
sensitivity. Further, it is perhaps theoretical coding and the
postulating of previously undiscovered or unarticulated
links that enable the development of the theory.”
(Cutcliffe, 2000) As I said, his statement is a bit zealous,
but its promise is correct. Staying open to emergent TCs is
important, if not totally necessary.


Barney G. Glaser, Ph.D., Hon. Ph.D.
The Grounded Theory Institute,
P.O. Box 400
Mill Valley, CA 94942 USA

Judith A. Holton,
10 Edinburgh Drive,
Charlottetown, PE C1A 3E8 CANADA

Reference List

Cutcliffe, J. R. (2000). Methodological issues in grounded
theory. Journal of Advanced Nursing, 31(6), 1476-1484.

Dey, I. (1999). Grounding grounded theory: guidelines for
qualitative inquiry. San Diego, CA: Academic Press.

Fernandez, W. D. (2003). Metateams in Major Information
Technology Projects; a grounded theory on conflict,
trust, communication and cost. Unpublished Doctor
of Philosophy, Queensland University of Technology.

Gisburne, J. M. (2003). The applicability of the Misomodel
to youth violence and the interpretations of
dangerousness. Unpublished PhD, University of
Nebraska, Lincoln, NB.

Glaser, B. G. (1964). Organizational Scientists: Their
professional careers. Mill Valley, CA: Sociology

Glaser, B. G. (1978). Theoretical sensitivity: advances in
the methodology of grounded theory. Mill Valley,
CA: Sociology Press.

Glaser, B. G. (1998). Doing grounded theory: Issues and
discussions. Mill Valley, CA: Sociology Press.

Glaser, B. G. (2002). Conceptualization: On theory and
theorizing using grounded theory. International
Journal of Qualitative Methods, 1(2), Article 3.

Glaser, B. G. (2003). The grounded theory perspective II:
Description’s remodeling of grounded theory
methodology. Mill Valley, CA: Sociology Press.

Glaser, B. G., & Strauss, A. L. (1965). Awareness of Dying.
Chicago: Aldine Publishing Company.

Glaser, B. G., & Strauss, A. L. (1967). The Discovery of
Grounded Theory: Strategies for Qualitative
Research. Hawthorne, NY: Aldine de Gruyter.

Glaser, B. G., & Strauss, A. L. (1971). Status Passage.
Chicago: Aldine Atherton Inc.

Glaser, B. G., & Strauss, A. L. (1974). Time for Dying. Mill
Valley, CA: Sociology Press.

Glaser, B. G., & Strauss, A. L. (1975). Chronic Illness and
the Quality of Life. Mill Valley, CA: Sociology Press.

Lee, F. S. (2002). Theory creation and the methodological
foundation of post Keynesian economics (Working
Paper). Kansas City: Department of Economics,
University of Missouri.

May, K. (1994). The case for magic in method. In J. Morse
(Ed.), Critical issues in qualitative research
methods. Thousand Oaks, CA: Sage.

Milliken, P. J., & Schreiber, R. S. (2001). Can you “do”
grounded theory without symbolic interactionism? In
R. S. Schreiber & P. N. Stern (Eds.), Using grounded
theory in nursing (pp. 177-190). New York: Springer.

Morse, J. (Ed.). (1994). Critical issues in qualitative
research methods. Thousand Oaks, CA: Sage.

Ragin, C. C. (1987). The comparative method: Moving
beyond qualitative and quantitative strategies:
University of California Press.

Schreiber, R. S., & Stern, P. N. (Eds.). (2001). Using
grounded theory in nursing. New York: Springer
Publishing Company.

Skodol Wilson, H. (1977). Nursing Research, 26, 103.

Stern, P. N. (1980). Grounded Theory Methodology; its
uses and processes. Image, 23.