Open Coding

Shahedul Huq Khandkar

Introduction
We need to give names to our ideas and concepts to define, analyze and share with others. O e it s
defined, we can begin to examine them comparatively and ask questions to systematically specify the
states and to imply possible relations with others. It s also important that we name our concepts
appropriately; because people act toward things based on the meaning those things have for them;
and these meanings are derived from social interaction and modified through interpretation. [1]

To build concepts from a textual data source, we need to open up the text and expose the meaning, idea
and thoughts in it. One of the processes of analyzing textual content is Open Coding. Open Coding
includes labeling concepts, defining and developing categories based on their properties and
dimensions. It is used to analyze qualitative data and part of many Qualitative Data Analysis
methodologies like Grounded Theory.

Qualitative Data Analysis

Qualitative Data Analysis (QDA) consists of three parts: Noticing, Collecting and thinking about
interesting things [2]. QDA is generally a non-liner process and often can be recursive. As you continue
on collecting information, you may notice new things and need to think about them. As a result, you
sometimes have to go back to old data and analyze them again.

Figure 1: Workflow of Qualitative Data Analysis

In general, noticing means taking notes based on observation, recording events or interviews, gathering
documents etc. In the analysis phase, when you are going through the data you often mark important
sections and add descriptive name or code to it. This is the first step of coding and called Open Coding.
This article begins with a description of Open Coding and provides walk through on different open
coding techniques.
Building Concepts
The first step in qualitative data analysis is to go through the data (i.e. text) to break down in to pieces to
examine closely, compare for relations, similarities and dissimilarities. Different parts of the data are
arked ith appropriate la els or odes to ide tif the for further analysis.

A concept is a labeled section of data that a researcher identifies as significant to some facts that data
represent. Concepts are abstract representations of events, objects, actions or interactions and they
allow researchers to group similar information to better understand the data.

Concepts can be of various types; communication, storm or private company includes example of
concepts. Concepts may incite certain natural imagery as they have their own properties. For example,
we can think of data set representing telephone conversations between two participants and we can
label them as Telephone Co u i atio . So a labeled thing is something that can be location and
placed in a class of similar objects. Anything under a classification has one or more familiar properties or
characteristics; like sending information is a property of communication. It is important to understand
that concepts can be classified differently, it depends on the different properties of data the researcher
is focusing on and how he/she is translating them.

Abstracting the Concepts

As we continue to analyze the data by breaking down into distinct ideas, events or objects, we label any
important information in the process. The name of the labels can be decided by the analyzer or can be
take fro the o te t too. The later st le is ofte alled i i o odes Glaser & “trauss, 19 .

Naming Type Description

In vivo codes Wording that participants use in the interview
Constructed codes Coded data from in vivo codes,
Created by the researcher,
Academic terms
Table 1: Different naming strategy for codes

While analyzing the data, we sometimes get events or objects with common characteristics yet other
properties may separate them. We can use the common properties to group them under same concept.
Different researchers may think of different names from the same data set but in general, it should be
based on the context.

Following is a partial transcript of an interview with women in her early 20s and is about drug use by
teens. The interviewer did not have preset questions to ask. He continued his questions based on the
i ter ie ee s respo se.

Interviewer: Tell me about teens and drug use.

Respondent: I thi k tee s use drugs as a release fro their pare ts [ rebellious act ]. Well, I do ’t k o .
I a o l talk for self. For e, it as a e perie e [ experience ] [in vivo code]. You hear a lot about
drugs [ drug talk ]. You hear the are ad for ou [ negative connotation to the drug talk ]. …
Source: Basics of Qualitative Research, Second Edition by Anselm Strauss & Juliet Corbin [6]

As you can see in the above interview transcript, we have grouped the similar information using abstract
labels (i.e. drug talk). Some of the names for the labels are selected directly from the data (i.e. hard-core
use). This process of going through line by line data to assign codes is called line-by-line coding.

Notes for concepts or codes

Sometimes a name with few words is not enough to describe an entire concept. In such scenario, we can
rite otes agai st a o ept that e all Me o . A e o a o tain a paragraph or even more if
needed. If we take a closer look in to the line-by-line coding example, the answer of first question of the
interview has more meaning than we have expressed in the code. With memo, we can record that
information like this:

Interviewer: Tell me about teens and drug use.

Respondent: I think teens use drugs as a release from their parents

Memo: The first thi g that strikes e i this se te e is the ork use . This is a stra ge ter e ause,
when taken out of the context of drug taking, the work means that an object or a person is being
employed for some purpose. It implies a willful and directed act. In making a comparison, when I think
about a computer, I think about employing it to accomplish a task. I think of it as being at my disposal.
Source: Basics of Qualitative Research, Second Edition by Anselm Strauss & Juliet Corbin [6]

Glaser (1978) offered guidelines for preparing effective memos to generate substantive theory including
the following [3]:

 Keep memos separate from data

 Stop coding when an idea for memo occurs, so as not to lose the thoughts
 A memo can be brought to you by literally forcing it, by beginning to write about the code
 When a lot of memos on different code appear similar, compare the codes for any differences
that may have been missed. If the codes still seem the same, collapse to codes into one code.
 When you have two ideas, add two separate memos to avoid confusion.
Source: Nursing research: principles and methods by Denise F. Polit & Cheryl Tatano Beck

Defining Categories
As we continue to create codes for new concepts, it s not unexpected to come to a point when we will
have more than few pages of codes. At that stage, we should analyze the codes to find the similarities
and group them into categories based on their common properties. We may also consider dimensions of
the codes that represent the location of the property along a continuum or range. The name of the
category can be different from the codes to express its scope better and if necessary, we can also create
sub-categories from the codes then link to categories.

Figure 2: A sample set of codes generated from a qualitative analysis

Fro the a o e set of odes, e a group the o epts: E ail , Telepho e Co ersatio , Te t
message & Voi e Mail i to a ategor a d a e it Co u i atio .

When to stop line-by-line coding?

As we can see, the line-by-line coding is a very time consuming and tedious work but at the same time it
also helps to build detail structured conceptual data model. When we are not really finding any new
concepts but only repeating the existing labels, we can stop doing this very detailed analysis. But to
discover more information from the data, we have to continue our analysis. At this stage, we can use
a al ti tools to reak the data a d olle t ore i for atio . This pro ess is alled i roa al sis .

Types of Open Coding

There are a number of ways to do Open Coding. In our previous examples, we have analyzed the data
line by line, every sentence and even word by word. This process of coding is called line-by-line coding
which is important to build concepts and categories. But based on the research requirement, we can
also look into a bit broader scale and code against a sentence, paragraph, chapter etc. There can be
situation where we may just need to define concepts for an entire document.

Research Team Size

Whe doi g ope odi g, it s better to do in a group at the beginning. Sarker from Washington State
University and Lau & Sahay from University of Alberta found that problems can occur if the team
members do the initial coding separately [8]. Researchers from Institut für Informatik, Freie Universität
Berlin also found some key benefits of pair coding that includes:

 Group conversation helps to take important decisions (i.e. single out phenomena for coding,
decide which existing concepts to use for coding or when to create new concept) [Berlin]
 Concept definitions become more exact and differentiations get more precise
 The data perspective is maintained more consistently
 Generally, more number of phenomena are discovered and processed.
Use of Open Coding
Open Coding is generally the initial stage of Qualitative Data Analysis. After completing the Open Coding,
depending on the methodology we use, we can do Axial Coding and Selective Coding. At later stage of
the research, these coding help us to build theories in an inductive process (i.e. Grounded Theory). Open
Coding can be used with inductive, deductive or verification modes of inquiry too.

Figure 3: Workflow of Grounded Theory methodology

Exercise
No that e k o ho to do Ope Codi g, let s tr to use it. Following is part of an interview transcript
with a woman in her 20s and is about drug use by teens. We would like to use open coding to analyze
the data.

Interviewer: Tell me about teens and drug use.

Respondent: I think teens use drugs as a release fro their pare ts. Well, I do ’t k o . I a o l talk for
myself. For me, it was an experience. You hear a lot about drugs. You hear they are bad for you. There is
a lot of the arou d. You just get i to the e ause the ’re a essi le a d e ause it’s ki d of a e
thi g. It’s ool! You k o , it’s so ethi g that is ad for ou, ta oo, a o . E er o e is agai st it. If ou
are a teenager, the first thing you are going to do is try them.

Interviewer: Do teens experiment a lot with drugs?

Respondent: Most just tr a fe . It depe ds o here ou are a d ho a essi le the are. Most do ’t
really get into in hard-core. A lot of teens are into pot, hash, a little organic staff. It depends on what
phase of life ou are at. It’s ki d of progressi e. You start off with the basic drugs like pot. Then you go on
to try more intense drugs like hallucinogens.
Interviewer: Are drugs easily accessible?
Respondent: You can get them anywhere. You just talk to people. You go to parties, and they are passed
around. You can get them at school. You ask people, and they direct you as to who might be able to
supply you.

Interviewer: is there any stigma attached to using drugs?

Respondent: Not among your peers. If ou’re i a group of tee agers a d e er o e is doi g it, if you
do ’t use, ou are fro ed upo . You a t to e a le to sa ou’ e e perie ed it like the other people
around you. Obviously, outsiders like older people will look down upon you. But within your own group of
friends, it definitely is not a stigma.

Interviewer: You say you did drugs for the experience. Do kids talk about experience?
Respondent: it’s ore of shari g the e perie e rather tha talki g a out the e perie e. You talk a out
doi g drugs ore tha hat it’s like he ou take drugs. It depe ds upon what level you are into it, I
guess. Most kids are doing it because it is a trend in high school. They are not doing it because of the
experience in some higher sense. They are doing it because they are following the crowd.
Source: Basics of Qualitative Research, Second Edition by Anselm Strauss & Juliet Corbin [6]

So, how are we going to do that? One option is to printout the content, high light the important
concepts and write the codes. But a manual open coding approach is not a good process especially when
we have to deal with large amount of data. As you are limited to search by reading only, it can also
cause unwanted errors. This process is simply impractical for a large scale data analysis with open
coding.

Figure 4: Open coding with pen and paper

We can use software applications like Saturate [4] or Atlas.ti [5] to do Open Coding. Atlas.ti is a
commercial qualitative data analysis tool and Saturate is a free web based tool developed by Dr. Sillito
from University of Calgary. We will use this tool in our exercise. A video tutorial for the tool can be
found here [4].

No that e ha e soft are a d k o ho to use it, let s start uildi g o epts. Take a look i to the
response: I think teens use drugs as a release from their parents . This looks like an act of rebellion. So
e a ode it as re ellious a t . The ter use looks like ea i g so ethi g ore. If e takeout the
context of drug use for a second and think about it – it may mean that they are being used for some
other reason which e are ot sure at this state. “o, e should take a ote Me o for future
reference. We have to continue to analyze the data line-by-line and add codes as necessary as long as
we find significantly new concepts. Tools like Saturate can help us in both improving the efficiency and
better manage the data.

Figure 5: Adding codes using Saturate

As we continue on coding the data, we may find similar concepts and can classify them under common
o epts i.e. drug talk , egati e o e tio . Following code snippet shows a part of the interview
transcript with codes.

Respondent: I think teens use drugs as a release fro their pare ts [ rebellious act ]. Well, I do ’t k o . I a o l
talk for self. For e, it as a e perie e [ experience ] [i i o ode]. You hear a lot a out drugs [ drug talk ].
You hear the are ad for ou [ negative connotation to the drug talk ]. There is a lot of the arou d
[ available supply ]. You just get i to the e ause the ’re a essi le [ easy access ] a d e ause it’s ki d of a
e thi g [ novel experience ]. It’s ool! You k o , it’s so ethi g that is ad for ou, ta oo, a o [ negative
connection ]. E er o e is agai st it [ adult negative stance ]. If ou are a tee ager, the first thi g ou are goi g to
do is try the [ challenge the adult negative stance ].

Interviewer: Do teens experiment a lot with drugs?

Respondent: Most just try a few [ limited experimenting ]. It depends on where you are and how accessible they
are [ degree of accessibility ]. Most do ’t really get into in hard-core [good i i o o ept] [ hard-core use vs
limited experimenting ]. A lot of teens are into pot, hash, a little organic staff [ soft core drug types ]. It depends
on what phase of life you are at [ personal developmental stage ]. It’s ki d of progressi e [ progressive using ].
You start off with the basic drugs like pot [ basic drugs ]. Then you go on to try more intense drugs like
hallu i oge s [ intense drugs ] [in vivo code].

In the process of line-by-line coding, we will soon be able to group the concepts into categories like
drug use for o epts like hard- ore use a d soft ore . Once we start getting too many old concepts,
we can stop labeling and move on to next step (i.e. selective coding, axial coding) based on our research
methodology.

Benefits of Open Coding

In the process of Open Coding, the concepts emerge from the raw data and later grouped into
conceptual categories. The goal is to build a descriptive, multi-dimensional preliminary framework for
later analysis. As its build directly from the raw data, it process itself ensures the validity of the work.

Problems
Although Ope Codi g is a i porta t tool for Qualitati e Data A al sis ut it s also a er ti e
o su i g a d tedious ork. “o eti es it s hard decide when to stop line-by-line coding and if the
researcher misses any important concept, he/she may have to restart the boring task again.

Annotated Bibliography:
John V. Seidel. Qualitative Data Analysis [2]
This document was originally part of the manual for the Ethnograph v4. It explains the process of Open
Coding and also Qualitative Data Analysis in a broad sense.

Michael Nunes, Saul Greenberg, Carman Neustaedter. Using physical memorabilia as

opportunities to move into collocated digital photo-sharing [9]
A study on how physical memorabilia can be used as opportunities to move into home-based collocated
digital photo-sharing. The researcher used semi-structured contextual interviews, each approximately
one hour long. They used open coding as part of the data collection and analysis.

Sarker, S. Lau, F. Sahay, S. Building Inductive Theory of Collaboration in Virtual Teams: An

Adapted Grounded Theory Approach [8]
This paper outlined how the grounded theory was adapted to develop a theory of collaboration in
virtual teams. The researchers studied virtual teams composed of students from two different
universities and engaged in a 14 week long systems development projects. They analyzed the data using
adapted versions of open coding, axial coding and selective coding.

References
[1] Symbolic Interactionism. Bulmer H. (1969) [Link to Google Books]

[2] Qualitative Data Analysis. John V. Seidel [Link]

[3] Page 582, Nursing research: principles and methods by Denise F. Polit, Cheryl Tatano Beck
[Link to Google Books]
[4] Saturate, a web-based Open Coding tool developed by Dr. Sillito. University of Calgary
http://www.saturateapp.com

[5] Atlas.Ti, A commercial desktop application for Qualitative Data Analysis. http://www.atlasti.com/

[6] Chapter 8, Basics of Qualitative Research, Second Edition by Anselm Strauss & Juliet Corbin [Link]

[7] A Coding Scheme Development Methodology Using Grounded Theory for Qualitative Analysis of
Pair Programming. Institut für Informatik, Freie Universität Berlin. [Link]

[8] Building Inductive Theory of Collaboration in Virtual Teams: An Adapted Grounded Theory
Approach. Sarker, S. Lau, F. Sahay, S. [Link]

[9] Using physical memorabilia as opportunities to move into collocated digital photo-sharing.
Michael Nunes, Saul Greenberg, Carman Neustaedter. [Link]