Qualitative Research Journal

Coding qualitative data: a synthesis guiding the novice

Mai Skjott Linneberg, Steffen Korsgaard,
Article information:
To cite this document:
Mai Skjott Linneberg, Steffen Korsgaard, (2019) "Coding qualitative data: a synthesis guiding the
novice", Qualitative Research Journal, https://doi.org/10.1108/QRJ-12-2018-0012
Permanent link to this document:
https://doi.org/10.1108/QRJ-12-2018-0012
Downloaded on: 11 May 2019, At: 07:56 (PT)
References: this document contains references to 30 other documents.
Downloaded by University Library At 07:56 11 May 2019 (PT)

To copy this document: permissions@emeraldinsight.com

The fulltext of this document has been downloaded 4 times since 2019*
Access to this document was granted through an Emerald subscription provided by emerald-
srm:485088 []
For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald
for Authors service information about how to choose which publication to write for and submission
guidelines are available for all. Please visit www.emeraldinsight.com/authors for more information.
About Emerald www.emeraldinsight.com
Emerald is a global publisher linking research and practice to the benefit of society. The company
manages a portfolio of more than 290 journals and over 2,350 books and book series volumes, as
well as providing an extensive range of online products and additional customer resources and
services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the
Committee on Publication Ethics (COPE) and also works with Portico and the LOCKSS initiative for
digital archive preservation.

*Related content and download information correct at time of download.

 The current issue and full text archive of this journal is available on Emerald Insight at:
www.emeraldinsight.com/1443-9883.htm

Coding
Coding qualitative data: qualitative
a synthesis guiding the novice data
Mai Skjott Linneberg
Department of Management, Aarhus University, Aarhus, Denmark, and
Steffen Korsgaard
Department of Entrepreneurship and Relationship Management, Received 6 June 2018
University of Southern Denmark, Kolding, Denmark Revised 15 March 2019
Accepted 16 March 2019

Abstract
Purpose – Qualitative research has gained in importance in the social sciences. General knowledge about
qualitative data analysis, how to code qualitative data and decisions concerning related research design in the
analytical process are all important for novice researchers. The purpose of this paper is to offer researchers who
Downloaded by University Library At 07:56 11 May 2019 (PT)

are new to qualitative research a thorough yet practical introduction to the vocabulary and craft of coding.
Design/methodology/approach – Having pooled, their experience in coding qualitative material and
teaching students how to code, in this paper, the authors synthesize the extensive literature on coding in the
form of a hands-on review.
Findings – The aim of this paper is to provide a thorough yet practical presentation of the vocabulary and
craft of coding. The authors, thus, discuss the central choices that have to be made before, during and after
coding, providing support for novices in practicing careful and enlightening coding work, and joining in the
debate on practices and quality in qualitative research.
Originality/value – While much material on coding exists, it tends to be either too comprehensive or too
superficial to be practically useful for the novice researcher. This paper, thus, focusses on the central
decisions that need to be made when engaging in qualitative data coding in order to help researchers new to
qualitative research engage in thorough coding in order to enhance the quality of their analyses and findings,
as well as improve quantitative researchers’ understanding of qualitative coding.
Keywords Transparency, Coding, Qualitative data, Validity, Qualitative data analysis
Paper type General review

Introduction
The analysis of qualitative data is challenging and often frustrating, especially for novice
qualitative researchers. Findings and results do not emerge from your transcripts and
documents by themselves, but require deliberate work to identify the most important
elements and write them up into a coherent and convincing “story” that answers the
research questions and provides insights that are loyal to the data (Miles et al., 2013).
An important tool in the process of turning the raw qualitative data into a communicative
and trustworthy “story” is coding. The core operation of coding involves examining a coherent
portion of your empirical material – a word, a paragraph, a page – and labeling it with a word or
short phrase that summarizes its content. Central to qualitative analysis, coding reduces large
amounts of empirical material and makes the data readily accessible for analysis, while at the
same time increasing the quality of the analysis and findings. In particular, coding in itself is an
early form of analysis “in such a way that the ‘final’ conclusions can be drawn and verified”
(Miles and Huberman, 1994, p. 11).
The process of teasing findings out of qualitative data requires craft and artfulness on
the part of the researcher. With time it therefore becomes less frustrating and more exciting,
even though it remains a cumbersome process (cf. Bochner, 2018). This paper presents a set
of coding techniques that can help pave the way for the researcher’s interpretive judgements
and improve their quality. By using this paper, novice researchers will be able to reflect Qualitative Research Journal
more carefully on the repertoire of choices related to qualitative coding in order to improve © Emerald Publishing Limited
1443-9883
their analyses and draw conclusions while remaining loyal to the data. Thus, this paper DOI 10.1108/QRJ-12-2018-0012
 QRJ discusses how and why coding can help the qualitative data analysis and explains the steps
associated with engaging in coding.
We were motivated to write this paper by the state of the current literature on coding,
which we found to be either too comprehensive and detailed (see, e.g., the otherwise excellent
Saldaña, 2015; Miles et al., 2013) or too superficial in its treatment of the basic craft of coding
(most methods books that we have encountered when teaching qualitative methods) to be
really helpful to novices seeking to familiarize themselves with coding for the first time. In this
paper, we provide a comprehensive yet focused synthesis of the practice of coding in
organization and management studies. The synthesis is an educational and reflexive resource
that also provides researchers with practical hands-on knowledge. Having read this paper,
researchers new to qualitative research will become familiar with the different perspectives
that are present in the literature on coding, enabling informed choices to be made.
Qualitative research comprises heterogeneous approaches encompassing a multitude
of philosophies, paradigms and methodologies and there are significant differences in how
qualitative researchers attend to research designs, analytical techniques and quality
Downloaded by University Library At 07:56 11 May 2019 (PT)

issues (cf. Gehman et al., 2018; Creswell and Miller, 2000; Bochner, 2018; Guba and Lincoln,
1994; Creswell and Poth, 2017). Despite these differences, however, we suggest that
researchers share a concern for immersion and careful attention to the empirical material,
coupled with an insistence on having their findings grounded in or supported by the data
and are able to benefit from transparency in how the findings were developed from the
data and perhaps the unfolding of theory both within a research team and in the
communication of results. Thus, whether you are “coding for” mechanisms, discourses,
social constructions, metaphors or lived experiences, coding offers a useful craft that
enables deep immersion in the data, as well as transparency in the development and
presentation of findings that will be useful in particular to many novice researchers who
are testing the waters of qualitative inquiry for the first time. In some ways, coding is
similar to how the craftsmanship of the stroke, the mixing of colors and the preparation of
a canvas is conducive to the artfulness of both abstract and figurative painting. As such,
structuring your data and achieving an overview of it, carefully considering the relevance,
meaning and importance of segments of data and making the data easily accessible for
subsequent data analysis are all likely to constitute good practice and to enable good
analytical work in most of the approaches used in qualitative research. In the following,
we therefore treat coding as a craft that is useful in most qualitative approaches, but note
when, for example, underlying philosophies (as in grounded theory) have implications for
the coding process.
We start the paper by discussing what coding is, its rationales and what to consider prior to
coding. In the following section, we move on to discuss the actual activity of coding – whether
it is deductive or inductive in nature, and how to implement coding cycles. We then highlight
the usefulness of analytical memos, examples of how and why to display data and thoughts on
interpretation before finally presenting some common pitfalls to keep in mind when engaging
in qualitative coding. The paper concludes by discussing the limitations of and problems
encountered in coding.

What is coding?
Coding in its most basic form is the simple operation of identifying segments of meaning in
your data and labeling them with a code, which can be defined as “a word or short phrase
that symbolically assigns a summative, salient, essence-capturing, and/or evocative
attribute for a portion of language-based or visual data” (Saldaña, 2015, p. 3).
The basic coding operation can be done in various ways. In smaller projects with a limited
amount of data, simple color coding with markers may suffice, with one color for each code.
The copy-and-paste function in software such as Word or Excel will also allow you to copy
 portions of text or images from your data documents into new documents (making sure the Coding
data origin is visible, e.g. “interview no. 3”). With larger amounts of data or research qualitative
groups, dedicated software programs can be useful. Such software programs use a basic data
copy-and-paste operation and have become very easy to use over time. We encourage our
students to use this software in their qualitative projects and have observed how the use of
software allows them to do better data analysis, going more deeply into the data, providing
better presentations of the data and findings, and achieving higher levels of quality regarding
the findings. However, using software is not without its dangers. The ease with which new
codes can be generated often results in the creation of too many codes. Also, as using software
can easily become mechanical in nature, the coding can end up being too rapid and perhaps
lacking in reflective analytical depth and focus. Hence, beneath the simple operations lie a
great number of challenges. So, while coding itself is easy to do, it is difficult to do well. This is
perhaps why some researchers prefer to stick to paper and markers while coding.

Some advantages of coding

Downloaded by University Library At 07:56 11 May 2019 (PT)

While at first glance coding might seem like an unnecessary and laborious step between
data collection and coming up with findings, there are several good reasons for spending a
considerable amount of your time on coding. It represents the gritty craftmanship that
enables artful and creative interpretation and analysis of the data. Basically, the coding
process creates an inventory of your data, which enables you to do the following
(cf. Saldaña, 2015; Miles et al., 2013).

Acquire deep, comprehensive and thorough insights into your data

As you code your data, you have to look at each individual sentence and paragraph in your
data and make a judgement about its meaning. Simply reading through your data as if it
were a book will not prevent you from overlooking potentially new and surprising data.
Thus, coding is an approach which makes you revisit all aspects of the data you have
collected, including those you may not have noticed during the actual data collection. This
will likely trigger analytical ideas that are not simply derived from your (unconsciously)
selective impressions and recollections of the data collection activities (interviews,
observations, etc.).

Make the data easily accessible and retrievable

Qualitative empirical work can often continue over long periods, during which the memory of
the data may fade, thus making access to the data an important condition for analysis. Coding
sorts the data into labeled segments, much like folders on your computer. This enables quicker
access to data and allows the researcher to retrieve it for another look. Moreover, it makes
collaborative work easier when more than one researcher is using the same data.

Sorting and structuring your data

Having your research question in mind, the most essential codes can often be used as a
skeleton on which to build your analysis. While coding, you make judgements about each
individual element in your data in order to decide whether it is relevant or not. This reduces
the amount of data you have to take into the final analysis and makes the analytical
tasks easier. If you are doing a comparative study or have multiple interviews, coding
provides you with a structure that allows comparison of specific dimensions of interest.

Ensuring transparency
As indicated in the introduction, there is an intense debate over the quality criteria that are
appropriate in qualitative research (e.g. Sinkovics et al., 2008; Creswell and Miller, 2000;
 QRJ Kvale, 1995; Bochner, 2018). We believe that, despite these terminological and philosophical
debates, most qualitative researchers are keen on ensuring that their findings are fundamentally
credible and trustworthy. In empirical work, this can be enhanced by observing transparency in
respect to how your conclusions are linked to your data (cf. Guba and Lincoln, 1994; Elo et al.,
2014) because this allows the reader to understand the role played by the researcher in the
shaping and analysis of the data. Also, it justifies the choices made by the researcher and can be
seen as an attempt to even out the knowledge and the access asymmetry between the researcher
and the reader (in particular when the reader is a practitioner rather than another researcher).
In turn, this can also compel a focus on the applicability and meaningfulness of one’s work to
practitioners, which can be understood as a quality aspiration in its own right in many
approaches to qualitative research (Kvale, 1995; Bochner, 2018).
Coding allows you to pick out relevant parts to show the reader. Even though interpretation
will always be a part of the researcher’s task, too much telling and too little showing lowers the
quality of the work. Showing the data to the reader forces you to develop a chain of evidence
depicting your arguments and showing how you have reached your conclusions (Pratt, 2009).
Downloaded by University Library At 07:56 11 May 2019 (PT)

Ensuring validity
Coding is an important step in moving from the raw data to the findings, as well as being a
means to maintain coherence between the objective and the results. Coding is a way to
ensure that the questions asked are the questions that have been answered. Even though the
questions asked initially can change throughout the research process, especially in
inductive approaches (Charmaz, 2014), reconciling the chosen question(s) and answer(s) is
ultimately necessary. Moreover, this provides an excellent opportunity to come to terms
with confirmatory bias by searching for contradictory evidence in the empirical material.

Giving a voice to one’s participants

Although coding confronts the researcher’s work with the empirical material, the coding
process can also be seen as interactive (Charmaz, 2014). Codes are created as a means to
understand the phenomenon and/or participants and their perspectives. As such the
researcher interacts with the phenomenon and the participants time and time again by means
of the empirical material. As a result, through the dynamic development of codes, we come to
understand participants’ views and actions from their own perspectives (Charmaz, 2014). This
view is particularly prominent in the literature on grounded theory, where researchers are not
restricted by preconceived codes but understand codes as possible emerging inductively from
scrutinizing meanings in the data (see below for an overview of inductive and deductive forms
of coding, respectively). However, in deductively inclined coding as well it is essential that the
researcher engages with the data continuously during the analysis.

Preparing for the coding process

While it is always good to have an overview of your research process and to prepare the
different steps in advance, certain things need to be in place before coding is initiated.
First, a research question (or a set of research questions) and a research design are
established. The objectives and the research question are important, as they help define
what kind of data are needed in order to complete the project successfully by answering this
question, as well as serving as an arbiter in respect of any questions that appear during
the research process. The research design outlines the nature of the research and examines
the overall elements of the research project to determine how they fit together. It also defines
the unit of analysis, the context and the data that need to be collected.
Second, in most qualitative research projects, even those with a more grounded
approach, some form of reading or review of the relevant research literature is necessary.
 According to Eisenhardt, “I believe in knowing the literature, and then looking for a problem Coding
or questions where there’s truly no known answer. It’s almost impossible to find those qualitative
problems without knowing the literature” (Eisenhardt in Gehman et al., 2018, p. 4). Also, data
knowing the existing research literature enables you to delimit the area of research.
Moreover, the existing literature can help you develop the tools for collecting the data, for
example, an interview guide (cf. Rowley, 2012; Kvale, 1997; Miles et al., 2013) or the coding
framework, assuming the coding will being using a deductive approach.
Third, you need to collect at least some of the data and document it in a form that
allows systematic analysis. Collecting data requires the development of a research design
and a sampling strategy. In organization and management research, for example, the
typical data for which qualitative coding is relevant are textual, such as transcribed
interviews, written field notes documenting the activities observed and/or other forms of
text, such as newspaper articles, reports, excerpts from social media, etc. Visual data in the
form of images and video are increasingly being used, and the software tools available for
qualitative data analysis are getting better at handling such data. Regardless of the type
Downloaded by University Library At 07:56 11 May 2019 (PT)

of data, it is important to document them in a format that makes them accessible to

labeling so they are visible and retain their form over time. Consequently, researchers
taking field notes while observing must make sure to rewrite the hastily written jottings
into more elaborate and legible notes as soon as possible in order to retain thick data in a
form that is fully retrievable even after the impressions of the moment have faded from
their memory (Spradley, 2016).

Coding qualitative data

Inductive coding
There is a strong tradition in qualitative research of developing codes “directly” from the
data. Researchers develop codes from the data by using phrases or terms used by
the participants themselves, rather than using the, often theoretical, vocabulary of the
researcher. In this way, the codes stay close to the data, mirroring what is actually in them,
rather than the ideas and prior understandings of the researcher, who is working vigorously
to remain open-minded. This approach is most often referred to as the inductive approach or
sometimes as grounded theory. The concept of grounded theory originates from the
strategies developed by Glaser and Strauss (1967) and Strauss and Corbin (1990) and later
Charmaz (2014), but it has since developed into a “big tent,” that is, a set of approaches to the
task of building a theory from data:
It almost invariably involves collecting data, breaking it up […] and then abstracting at a higher
level […] this process is at the heart of what most theory-building qualitative researchers are doing.
(Gehman et al., 2018, p. 5)
The inductive approach is relevant when doing an exploratory study or when no theoretical
concepts are immediately available to help you grasp the phenomenon being studied.
Working systematically with coding allows the inductive researcher to observe transparency
and thus offer credible interpretations of the empirical material (Gioia et al., 2013).
When doing inductive coding, you will often find yourself creating several codes, often
very precise and narrow ones, which is good for capturing the complexity and diversity of
the data. From the grounded theory approach, we have the notion of line-by-line coding,
which depicts this exactly (Charmaz, 2014). Finding a balance between having a workable
number of codes and capturing the complexity and diversity of your data is difficult.
Ultimately you may want to end up with a list of 50–70 initial codes. These codes can be
submitted to a second cycle of coding in which higher level categories are created from the
initial code list (Gioia et al., 2013). This process moves the researcher from having a higher
number of codes to having a smaller number of themes or categories. As you move forward
 QRJ in the coding process toward higher level categories, you may want to draw on some of the
existing theory and concepts that are related to your phenomenon. This helps you anchor
the study in the literature and provides further support for the findings.

Deductive coding
While inductive codes have the advantage of being completely loyal to the data, there is a
risk of the whole process becoming too complicated and lacking in focus, especially for the
novice qualitative researcher. Sometimes researchers will therefore adopt a more narrow
and deductive approach to coding. Here, a pre-defined list of codes is created in a so-called
coding frame before you start coding your data (cf. Miles et al., 2013). This approach helps
focus the coding on those issues that are known to be important in the existing literature,
and it is often related to theory testing or theory refinement. It is also a helpful approach, if
the aim of the study is to generalize analytically across cases (Rowley, 2002; Eisenhardt,
1989). Indeed, if the study is theory-driven, the theoretical framework may be converted into
a coding framework. Even if you engage predominantly in deductive coding, the process can
Downloaded by University Library At 07:56 11 May 2019 (PT)

still remain flexible.

Generally, the codes in deductive coding are theoretical concepts or themes drawn from
the existing literature. In a deductive coding approach, the number of codes will typically be
relatively limited, with maybe just five to ten codes derived from the theoretical framework.
During coding, the coding frame can be adjusted if interesting differences emerge within a
given code or if some new and interesting things come up that are not captured by the
existing codes.
In practice, a combination of inductive and deductive coding is the most commonly used
approach, what is sometimes referred to as a blended approach (Graebner et al., 2012) or
abduction (Alvesson and Kärreman, 2007). In order to get started, it is nevertheless a good
idea to concentrate on one of the two approaches, as they have different strengths and
weaknesses. Starting inductively ensures closeness or “giving voice” to the data, with the
possibility of unfolding theory later. Deductive approaches ensure structure and theoretical
relevance from the start, while still enabling a closer inductive exploration of the deductive
codes in later coding cycles.
The notion of abduction captures this combination of inductive and deductive elements
and suggests a cycling back and forth between data and theory (Pierce, 1978). By doing this,
the researcher remains open to surprises in the data while at the same time staying attuned to
existing theories. Inductive coding stays more loyal to the data but may also be less focused.
Turning a deductive coding process inductive will thus move you closer to the data, yet also
lose theoretical focus, and vice versa. The abductive approach is attuned to looking for
surprises in the data in the way that one has an interest in rethinking the existing theories
(Pierce, 1978). The notion of abduction thus encourages a flexible theoretical framework as
well as a flexible approach to the empirical framework, which manifests itself in a combination
of inductive and deductive coding in accordance with the needs of the emerging analysis.

Cycles of coding
It is helpful to see coding as occurring in two or more cycles. In an inductive approach, the
first coding cycle uses informant-centric terms, whereas the second coding cycle becomes
more researcher-centric in the sense that concepts, themes and dimensions from existing
theories may be introduced to lift the analysis to a higher level of abstraction (Gioia et al.,
2013). This can help you think of the code types that are used in the initial phases as more
descriptive. Here the codes are used to create an overview of the data and enable subsequent
exploration of patterns of similarities and differences in the later cycle. Thus, the code types
used in the second cycle are more analytical in nature and focus more on creating patterns in
the data. Hence, the act of coding is not to be seen as linear: rather, it enters feedback stages,
 which is why we can use the labels “first-” and “second-cycle coding.” Below is an overview Coding
of selected coding methods (adapted from Saldaña, 2015, pp. 59-64). qualitative
Coding methods: data
(1) First-cycle coding:
• descriptive coding (for field notes, documents), inductive and/or deductive; and
• attribute coding (for data base structure and overview, and particularly valuable
in comparative research projects).
(2) Second-cycle coding:
• eclectic coding (for refining first-cycle choices);
• pattern coding (exploring patterns across first-cycle codes);
• categorization (combining first-cycle codes into – perhaps theoretically informed
categories); and
Downloaded by University Library At 07:56 11 May 2019 (PT)

• exploring patterns in across codes (temporal or processual structure, similarities

and differences across descriptive and attribute codes).
Different code types are commonly used in these two coding cycles. The text-book literature
is swamped with code types, which seems to confuse novices more than helping them. For
this reason, here we will stick to just a few. In the first coding cycle, two types of codes are
often of particular importance: descriptive codes and attribute codes.
Descriptive codes are assigned to segments of data based on what the segment
“is about.” Segments of data are thus summarized using a label that indicates the
meaning of the segment of data in relation to the overall research topic. Descriptive codes
are assigned to smaller units of data, usually in the form of sentences or a coherent set of
statements. If we consider an example where entrepreneurs have been interviewed about
their financing, whenever the researcher comes across statements by interviewees about
their use of networks, he or she will code the statements “networks.” If descriptive coding
is done properly, it leads to a categorized inventory of the data providing an overview of
what is in them (Saldaña, 2015).
Descriptive coding may be more or less structured depending on whether the approach is
deductive or inductive, and you may wish to uncover certain types of content in the data.
For some purposes, it may be useful to code for events, which may later be analyzed in a
process-oriented study, or alternatively to code for emotions, such as emotional responses to
certain events. Notably, in deductive coding a higher level of abstraction can be reached in
the second cycle, but it may also be useful to explore differences within codes if the
deductive codes contain interesting sub-themes and variation. This second option will take
you closer to the details of the data rather than to a higher level of abstraction. This depends
on where you feel the analysis needs to go and is something you can experiment with.
Attribute codes are basic information assigned to larger segments of data, typically to
the units in which the data were originally collected such as interviews, sites of observation,
or data sets for a given example. In certain types of project, typically those where the
findings are derived in part from comparing cases or eliciting differences in experiences and
perspectives (e.g. in multiple case studies, cf. Yin, 1989; Eisenhardt, 1989), attribute codes
can help fix an early structure for the data set establishing a good overview and easy access
to data. Later this enables sources of differences and similarities in the data to be explored
more effectively. At the individual level it might thus be useful to assign attribute codes for
age, gender, experience or other attributes that are of relevance in the given study. At an
organizational level, there are similar generic or specific attributes such as industry,
size, etc., that can be considered potential sources of insights into the phenomenon.
 QRJ This corresponds to a replication logic (the constant comparative logic) invoked on the data
level, which involves the process of continuously comparing new data to extant data. The
ongoing effort will confirm, enhance or discount the theoretical propositions under scrutiny.
From the first cycle of coding to the second cycle, codes develop from the initial phases,
being more straightforwardly developed in later stages, when a second cycle of coding
entails “classifying, prioritizing, integrating, synthesizing, abstracting and conceptualizing,
and theory building” (Saldaña, 2015, p. 58). As the research process develops, so does the
type of coding, which also allows the researcher to move from basic descriptive codes
toward answering the research question posed. The trick is to look for an overarching
structure or process that can be understood at a theoretical level (Gehman et al., 2018).
Therefore, when codes are clustered together according to similarity and regularity,
patterns are born, and you can begin to analyze the connections between them. This also
implies identifying concepts inspired by the relevant theoretical framework. In this process,
both theories and extant empirical work come to play yet another role in data management,
as you can compare your own categories and their relations with relevant theories and/or
Downloaded by University Library At 07:56 11 May 2019 (PT)

concepts from previous studies in order to validate or generate further ideas for the analysis.
This is sometimes referred to as theoretical triangulation (Miles et al., 2013)
A coding structure consists of the first, second and perhaps subsequent cycle codes and
shows the progression from data to theory, thereby making visible the grounds on which
you are drawing your conclusions (Gioia et al., 2013). Depicting the data structure is not
promoted equally by all qualitative researchers; for instance, you can argue that comparing
multiple cases through replication (at the case level) makes it difficult to see anything from
the data structure that does not point to the particular elements in the single case
(Eisenhardt in Gehman et al., 2018). In this instance, you can choose to present each case
narrative as the depiction of the empirical material.

After coding
Combining coding with analytic memos
Analytical memos can be described as the researcher’s ongoing reflections during coding
concerning the codes, the phenomenon, the informants and their interrelations. Analytical
memos help you think about these elements, thus eventually bridging the distinctions
between coding, analysis and results. Memos are creatively developed little documents
based on intuition, hunches and serendipitous occurrences (Saldaña, 2015) related to the
above-mentioned elements. We use memos as a tool as early as possible in data collection.
Repeatedly, we find that data collection and coding run in parallel, and the analytical memo
is a great tool to help materialize ongoing reflections, much like a log that can both inform
subsequent data collection and lead to richer explanations in the analysis later on.

From coding to analysis

Coding and interpretation are not two distinct phases but interrelated processes that co-evolve,
yet have different characteristics. While in a sense the first cycle of coding in particular
represents a more mechanical application of labels to segments of data, the associated task of
interpreting the codes and deriving an overall interpretation from them is not mechanical. It
requires insight to get patterns to emerge. While the coding process and analytical memorizing
enable the emergence of patterns in the data, it can only enable, not determine it. Ultimately a
process of immersion in the data is required, and this will take time and be as difficult to control
as the creative process of artists. Nonetheless some notes on what the output of the final analysis
is expected to look like may help guide the novice to the process of qualitative data analysis.
In most conceptualizations of qualitative research, theory-building (as opposed to theory
testing) is considered the overall purpose (Gioia et al., 2013). This means that qualitative
 research should lead to, for example, the identification and/or elaboration of new concepts, Coding
ideas about how concepts relate to each other, or accounts of processes. The output may qualitative
thus come in the form of typologies outlining the different aspects of a phenomenon, a data
concept with rich descriptions of the phenomenon, individuals or organizations that are
categorized under this concept, or theories describing conceptual relations.
Importantly, findings in qualitative research should always emphasize elaborate detailed
descriptions, specific examples and the inclusion and discussion of outliers rather than
count instances of a phenomenon. Good qualitative studies obtain a kind of undeniability to
their results, provided the findings come with detailed accounts of the phenomena.

Displaying your data and codes

Visual display eases the reading and increases the understanding of scientific texts. A data
display is anything in which data can be illustrated more efficiently than in longer pieces of
text. In some methodological approaches, such as grounded theory, the visual representation
of emergent findings is an essential step (Verdinelli and Scagnoli, 2013). Inductive coding in
Downloaded by University Library At 07:56 11 May 2019 (PT)

particular can be visualized nicely using a form of horizontal tree structure. This displays the
initial codes and shows how they are developed subsequently into categories and perhaps
concepts. In qualitative analysis, visual displays have a role to play at various stages of the
analysis (Saldaña, 2015, p. 58). There are multiple types of visual display (see Miles et al., 2013
for a comprehensive overview); here we concentrate on just a few.
Boxed displays are the simplest displays and are “literally, text framed within a box […]
used to highlight a specific narrative […] (it) emphasises the authors’ interest or points of
relevance” (Verdinelli and Scagnoli, 2013, p. 12). This type of display underscores the
extensive narratives that are central to the analysis in any project.
Matrices basically include two or more categories, dimensions or concepts that are cross-
classified. The matrix is probably one of the most commonly used types of display. As with
boxed displays, they can be used from the early stages of an analysis, for instance, to
present a literature review, to illustrate demographics or, later in the process, to illustrate the
results of an analysis.
Network displays are described as collections of nodes that are connected or linked (e.g. Miles
et al., 2013), allowing a focus on many variables simultaneously. They help illustrate the
relationship between themes and categories, and are also suitable for illustrating second-cycle
coding or even the results of the study.

Discussion: a critical view of coding

Despite its widespread use, coding has also been exposed to criticism. First, some argue that
coding splits the data into disjointed elements, and the holistic element so central to
qualitative analysis is thus being lost. This criticism is warranted in so far as coding does
split your data, and there is a risk that in your analysis you focus solely on the relations
between the codes at the expense of holistic and comprehensive understandings of the
examples and phenomena you are studying, or else you neglect the context and situational
factors. It is important to bear constantly in mind that the overall understanding of the
phenomena in order to make sure that the findings emerging from the coding process are
valid and relevant. There are several techniques that can be employed throughout the
process of analysis: rereading the full transcript and initial analytical memos, looking for
negative evidence in the data, keeping an updated logbook that includes situational factors
of relevance, and cooperating with one’s peers.
Second, some criticisms are directed at the subjective nature of the coding process on the
basis that judgements about which codes to develop and where a given segment of data fits
into the coding scheme are subjective to the coder. On the one hand, this is simply a
necessary consequence of doing qualitative research, the drawbacks of which are offset by
 QRJ the richness and detail we can obtain through any qualitative research. On the other hand, it
is important to note and consider this subjectivity, especially through reflexivity regarding
oneself and one’s perspectives. It has been suggested that using language-processing
technology (LPT) for coding can address this subjectivism and increase the trustworthiness
of the result (Crowston et al., 2012). However, while using LPT allows the researcher to
process very large amounts of data, there are obvious consequences to using such a system.
First, having a system requires a trained analyst to develop the rules applied to the system;
second, the system requires substantial inputs of data. It is therefore suggested it only has a
raison d’être in empirical material where manual coding is not feasible due to the volume of
data (Crowston et al., 2012).
Along the same lines, a recurrent theme in the literature is the increase in trustworthiness
that comes from having several people involved in the coding process, or perhaps
establishing inter-rater reliability by, for example, conducting inter-coder reliability tests or
working collaboratively to develop inter-coder agreement (a softer form of inter-coder
reliability) (Campbell et al., 2013). However, researchers’ varied perspectives and previous
Downloaded by University Library At 07:56 11 May 2019 (PT)

experiences are critical elements in the data analysis. This means that, instead of using one
researcher to confirm the reliability of the other, one can draw on “the positional reflexivity
of two researchers, each with a distinct perspective, as a potential strength to cogenerate
themes and theory” (Anderson et al., 2016, p. 44). This suggests that researchers working
independently with thematic analysis and coding cooperation in a team can subsequently
produce joint interpretations allowing for greater dimensionality.
Third, researchers should be careful not to mistake qualitative coding for a mechanical
quick fix neatly and swiftly de-contextualizing the data. In particular, the current literature
on qualitative research reminds us that the urgency and the believability of the findings
should be prioritized over their detachability from their empirical context (Bochner, 2018).
Importantly, coding and a count of the instances of a given phenomenon should not replace
elaborate detailed descriptions, elaboration of contexts, specific examples, or the inclusion
and discussion of contradictory results. This is neither the mission nor the value of
qualitative research. Good qualitative studies obtain a kind of undeniability of results
provided the findings come with detailed accounts of the phenomena that allow readers to
see the stories unfold and to relate to the experiences of the people involved. Thus,
ultimately the achievement of coding that is well done comes with a strong notion of the
qualities of qualitative inquiries, but it can be challenged by the misconception that coding
streamlines and standardizes the data.

Conclusion
This paper has explored the key aspects of the coding of qualitative data: the reasons for doing
coding, the roles coding plays, some of the different types of coding the use and combination of
inductive and deductive approaches to coding, the phases involved in engaging in cycles of
coding, how to display data and the relationship between coding and interpretation.
With experience, focussing on coding provides structure and depth to the analytical
process. In particular, coding should not be “just” a mechanical process chasing specifics. It
is important to remain attuned and sensitive to the data and its context, as well as to one’s
role as a researcher in collecting the empirical material. Engaging in coding allows the
researcher to provide transparency for others and oneself in relation to how existing
concepts are reflected in the empirical material, but also where there might be novel insights
and possibilities for theoretical development. Alvesson and Kärreman (2007) suggest doing
several rounds and types of coding with different framings and to take a particular interest
in the contradictions in the data, rather than merely looking for similarities. To engage in
coding could be seen as an adventurous trip revisiting venues and allowing ongoing
reflection and learning – not merely confirmation, but rather an art skilfully conducted
 (Probst and Bucholtz, 2015). One should not be tempted into considering coding as merely a Coding
pragmatic tool for organizing and reducing empirical material, as this material also allows qualitative
the coding to enhance the quality of one’s research by, for instance, checking for data
inconsistencies or contradictory findings in the empirical material, as well as providing
transparency for oneself through the process, and the reader in the final output.

References
Alvesson, M. and Kärreman, D. (2007), “Constructing mystery: empirical matters in theory
development”, Academy of Management Review, Vol. 32 No. 4, pp. 1265-1281.
Anderson, R.C., Guerreiro, M. and Smith, J. (2016), “Are all biases bad? Collaborative grounded theory
in developmental evaluation of education policy”, Journal of Multidisciplinary Evaluation, Vol. 12
No. 27, pp. 44-57.
Bochner, A.P. (2018), “Unfurling rigor: on continuity and change in qualitative inquiry”, Qualitative
Inquiry, Vol. 24 No. 6, pp. 359-368.
Downloaded by University Library At 07:56 11 May 2019 (PT)

Campbell, J.L., Quincy, C., Osserman, J. and Pedersen, O.K. (2013), “Coding in-depth semistructured
interviews: problems of unitization and intercoder reliability and agreement”, Sociological
Methods & Research, Vol. 42 No. 3, pp. 294-320, doi: 10.1177/0049124113500475.
Charmaz, K. (2014), Constructing Grounded Theory, Sage, London.
Creswell, J.W. and Miller, D.L. (2000), “Determining validity in qualitative inquiry”, Theory into
Practice, Vol. 39 No. 3, pp. 124-130.
Creswell, J.W. and Poth, C.N. (2017), Qualitative Inquiry and Research Design: Choosing among Five
Approaches, Sage publications, London.
Crowston, K., Allen, E.E. and Heckman, R. (2012), “Using natural language processing technology for
qualitative data analysis”, International Journal of Social Research Methodology, Vol. 15 No. 6,
pp. 523-543.
Eisenhardt, K.M. (1989), “Building theories from case study research”, The Academy of Management
Review, Vol. 14 No. 4, pp. 532-551.
Elo, S., Kääriäinen, M., Kanste, O., Pölkki, T., Utriainen, K. and Kyngäs, H. (2014), “Qualitative content
analysis: a focus on trustworthiness”, SAGE Open, Vol. 4 No. 1, pp. 1-10, doi: 10.1177/
2158244014522633.
Gehman, J., Glaser, V.L., Eisenhardt, K.M., Gioia, D., Langley, A. and Corley, K.G. (2018), “Finding
theory-method fit: a comparison of three qualitative approaches to theory building”, Journal of
Management Inquiry, Vol. 27 No. 3, pp. 284-300, doi: 10.1177/1056492617706029.
Gioia, D.A., Corley, K.G. and Hamilton, A.L. (2013), “Seeking qualitative rigor in inductive research:
notes on the Gioia methodology”, Organizational Research Methods, Vol. 16 No. 1, pp. 15-31,
doi: 10.1177/1094428112452151.
Glaser, B.G. and Strauss, A.L. (1967), The Discovery of Grounded Theory, Aldine Publishing Company,
Chicago, IL.
Graebner, M.E., Martin, J.A. and Roundy, P.T. (2012), “Qualitative data: cooking without a recipe”,
Strategic Organization, Vol. 10 No. 3, pp. 276-284, doi: 10.1177/1476127012452821.
Guba, E.G. and Lincoln, Y.S. (1994), “Competing paradigms in qualitative research”, in Denzin, N.K. and
Lincoln, Y.S. (Eds), Handbook of Qualitative Research, Sage, New York, NY, pp. 163-194.
Kvale, S. (1995), “The social construction of validity”, Qualitative Inquiry, Vol. 1 No. 1, pp. 19-40,
doi: 10.1177/107780049500100103.
Kvale, S. (1997), InterView, Hans Reitzels Forlag, Copenhagen.
Miles, M.B. and Huberman, M.A. (1994), Qualitative Data Analysis, 2nd ed., Sage, Thousand Oaks, CA.
Miles, M.B., Huberman, A.M. and Saldana, J. (2013), Qualitative Data Analysis: A Methods Sourcebook
(Incorporated), SAGE Publications, Thousand Oaks, CA.
 QRJ Pierce, C.S. (1978), “Pragmatism and abduction”, in Hartshorne, C. and Weiss, P. (Eds), Collected Papers,
Harvard University Press, Cambridge, MA, pp. 180-212.
Pratt, M.G. (2009), “From the editors: for the lack of a boilerplate: tips on writing up (and reviewing)
qualitative research”, Academy of Management Journal, Vol. 52 No. 5, pp. 856-862,
doi: 10.5465/amj.2009.44632557.
Probst, B. and Bucholtz, J. (2015), “Polyphonic coding in qualitative analysis: conversation as musical
motet”, Qualitative Research Journal, Vol. 15 No. 3, pp. 339-350.
Rowley, J. (2002), “Using case studies in research”, Management Research News, Vol. 25 No. 1, pp. 16-27.
Rowley, J. (2012), “Conducting research interviews”, Management Research Review, Vol. 35 Nos 3/4,
pp. 260-271.
Saldaña, J. (2015), The Coding Manual for Qualitative Researchers, Sage, Thousand Oaks, CA.
Sinkovics, R.R., Penz, E. and Ghauri, P.N. (2008), “Enhancing the trustworthiness of qualitative
research in international business”, Management International Review, Vol. 48 No. 6,
pp. 689-714, doi: 10.1007/s11575-008-0103-z.
Spradley, J.P. (2016), Participant Observation, Waveland Press, Long Grove, IL.
Downloaded by University Library At 07:56 11 May 2019 (PT)

Strauss, A. and Corbin, J. (1990), The Basics of Qualitative Research, Sage, London.
Verdinelli, S. and Scagnoli, N.I. (2013), “Data display in qualitative research”, International Journal of
Qualitative Methods, Vol. 12 No. 1, pp. 359-381, doi: 10.1177/160940691301200117.
Yin, R.K. (1989), Case Study Research, Sage, Newbury Park, CA.

Corresponding author
Steffen Korsgaard can be contacted at: stko@sam.sdu.dk

For instructions on how to order reprints of this article, please visit our website:
www.emeraldgrouppublishing.com/licensing/reprints.htm
Or contact us for further details: permissions@emeraldinsight.com