Identifying Themes and Coding
Interview Data: Reflective Practice
in Higher Education
© 2015 SAGE Publications, Ltd. All Rights Reserved.
This PDF has been generated from SAGE Research Methods Datasets.
Identifying Themes and Coding
Interview Data: Reflective Practice
in Higher Education
Student Guide
Introduction
This example introduces coding as an important process for conducting a
comprehensive thematic analysis of interview data. Coding helps to achieve
all three of the aims of thematic analysis: examining commonality, examining
differences and examining relationships.
The interview transcripts used in this exemplar were provided by Jamie Harding, a
Senior Lecturer at Northumbria University. The research carried out was inductive
and the objectives were to explore lecturers’ motivation in choosing their career,
their experiences of teaching students and their views on reflective practice and
change in higher education.
Coding
Coding helps to achieve all three of the aims of thematic analysis: examining
commonalities within a dataset, examining differences and examining
relationships. This example shows how to use codes to identify similarities and
differences between cases in your data.
Codes are usually notes made in the margin of interview transcripts and can take
a number of forms, including: complex system of abbreviations; systems that use
both abbreviations and numbers; and full words and short phrases. This example
SAGE
2015 SAGE Publications, Ltd. All Rights Reserved.
SAGE Research Methods Datasets Part
1
Page 2 of 14 Identifying Themes and Coding Interview Data: Reflective Practice in
Higher Education
uses primarily words and phrases as codes.
It is helpful to make a distinction between two types of code: apriori codes and
empirical codes (Gibson & Brown, 2010: 132–133). Apriori codes are created prior
to research to reflect categories that are already of interest before the research
has begun. They tend to derive at least partly from the researcher’s previous
reading and are more appropriately used as part of a deductive approach.
Empirical codes are derived after the data evidence has been collected, while
reading through the collected data points of importance and commonality are
identified. Empirical codes are more likely to be used in inductive pieces of
research, where the data is examined and analysed before consideration of the
existing theory and literature. It is important to emphasise that these two forms
of coding are not entirely separate: even when using empirical codes, it is likely
that the researcher’s prior knowledge of the subject will influence decision making
to some extent. Similarly, when using apriori codes, it is almost certain that some
issues and themes will emerge that were not anticipated from the researcher’s
prior reading in the subject area. As Jamie Harding’s research was primarily
inductive, we will look at empirical coding in this example.
Data Exemplar
The interview data was collected in the Faculty of Social Sciences at a case
study university by an interviewer, under the supervision of Jamie Harding, a
Senior Lecturer at Northumbria University. This was primarily an inductive piece of
research, which meant that there was no theory to test and no research questions
to answer. However, there were a number of research objectives. This example
focuses on one of these objectives: to identify feelings about reflective practice
and methods by which it was put into practice.
The interviewee we focus on in this dataset is Thomas. Thomas is a lecturer with
previous work experience in industry. He had been employed at the university
SAGE
2015 SAGE Publications, Ltd. All Rights Reserved.
SAGE Research Methods Datasets Part
1
Page 3 of 14 Identifying Themes and Coding Interview Data: Reflective Practice in
Higher Education
for a substantial period of time. This dataset will also discuss coding across the
excerpts from Thomas and another lecturer named Lewis and the full transcripts
with three other lecturers. Full transcripts are available in the Download Dataset
section.
Analysis: Coding
Jamie Harding describes how he approached the coding of Thomas’s interview
transcripts. The process of using empirical codes can be broken into four steps.
These steps are:
1. Identifying initial categories based on reading the transcripts.
2. Writing codes alongside the transcripts.
3. Reviewing the list of codes, revising the list of categories and deciding
which codes should appear in which category.
4. Looking for themes and findings in each category
Step 1: Identifying Initial Categories Based on Reading the
Transcripts
Codingshould begin with a thorough reading of the full transcripts to be analysed.
This enables you to identify categories that codes can be placed into and so
saves time in the analysis that follows. Identifying categories is a major part of
separating and sorting your data, however, it is difficult to suggest specific tactics
or techniques for creating categories. The researcher can only use their judgment
to identify broad subject areas under which the data could be grouped.
The initial list of categories will almost inevitably be modified in the course of the
analysis. However, the coding process is likely to take less time and to seem less
daunting if the researcher is able to draw up a preliminary category list at the start.
SAGE
2015 SAGE Publications, Ltd. All Rights Reserved.
SAGE Research Methods Datasets Part
1
Page 4 of 14 Identifying Themes and Coding Interview Data: Reflective Practice in
Higher Education
My list of categories around reflective practice looked like this:
1. Mechanisms for undertaking reflective practice
2. Motivation for reflective practice
3. Aims of reflective practice
4. Limitations to reflective practice
It is important to note that these categories were based not just on Thomas’s
interview transcripts, but also on the data collected from other participants in the
study at the case study university. The downloadable dataset contains the full
transcripts of all the interviews should the interested reader wish to read the data
in its entirety.
Step 2: Writing Codes Alongside the Transcripts
After deciding on the initial list of categories, and the form that their codes should
take (e.g. abbreviations, words and phrases), the researcher should begin to write
the codes alongside the interview transcripts. The application of codes involves
three elements:
1. Summarising
2. Selecting
3. Interpreting
I will now use a section of an interview with a lecturer, Thomas, to show how codes
can be applied. This section is available in the dataset download and is called
Coding a Transcript: Thomas. As a reminder, the four categories for codes that
had been identified are:
1. Mechanisms for undertaking reflective practice
2. Motivation for reflective practice
3. Aims of reflective practice
SAGE
2015 SAGE Publications, Ltd. All Rights Reserved.
SAGE Research Methods Datasets Part
1
Page 5 of 14 Identifying Themes and Coding Interview Data: Reflective Practice in
Higher Education
4. Limitations to reflective practice
I followed the three steps of summarising, selecting and interpreting to analyse the
data:
Reducing/summarising: This data reduction step helps the researcher to see
beyond the detail of the individual case and to identify themes. The interview
with Thomas demonstrated that points can often be made more succinctly in
written form than when they are made verbally; reducing/summarising information
through codes can often be quite a simple task.
Selecting: It is better to err on the side of caution and to limit the amount of
selection. It’s preferable to introduce codes that may need to be discarded later,
rather than risk failing to code an idea that could become an important feature of
the analysis. A key element of the inductive approach is that the development of
theory is driven by the research findings, rather than existing theory directing the
nature of data collection and analysis. A helpful guiding principle to decide what to
code is to search for commonality. The creation of categories in advance assists
with the process of selection; on Thomas’s interview transcript I knew to code any
comment about mechanisms, motivations, aims or limitations because these had
already been identified as areas discussed by a number of respondents.
Interpreting: Interpreting phenomena in their context is a key feature of qualitative
research. To correctly interpret the words of respondents, the qualitative
researcher needs to consider the context of what has been said and apply a code
that reflects the most likely meaning of the speaker. Thomas’s comment that:
‘And I think I do that given time but if you’re only teaching the same things, or
doing the same things, with little pressure you can reflect and learn and develop’
was interpreted to mean that time and teaching different subjects were limitations
on his ability to reflect on his practice. I understood that Thomas felt that time
limited his opportunities for reflection from a number of comments that he made
SAGE
2015 SAGE Publications, Ltd. All Rights Reserved.
SAGE Research Methods Datasets Part
1
Page 6 of 14 Identifying Themes and Coding Interview Data: Reflective Practice in
Higher Education
elsewhere, e.g. ‘There’s not time now to sit and think about it’. However, there
were no additional comments to support my interpretation in the area of teaching
different subjects, so I relied on the context. This is the type of occasion where
a methodological memo is useful, in order to facilitate later reflection on the
judgments that were made during the analysis. I wrote this memo:
Some interpretation was needed for Thomas’ comment: ‘And I think
I do that given time but if you’re only teaching the same things,
or doing the same things, with little pressure you can reflect and
learn and develop.’ It was clear from the context of what was
said that ‘do that’ meant ‘reflect on practice’. It also seemed clear
from comments that were made elsewhere in the interview that
Thomas thought that lack of time was a factor that limited his use of
reflective practice. A further interpretation that was made, although
one with less supporting evidence, was that Thomas believed that
being asked to teach different subjects limited his ability to reflect.
The reason for this interpretation was that Thomas seemed to
be discussing the conditions under which reflection could most
easily take place – having time, teaching the same subjects, not
being under pressure – but then implying that, where any of these
conditions did not apply, reflection was more difficult. However, as
the issue of the subjects taught was not referred to elsewhere in the
interview, this was a particularly subjective interpretation.
Step 3: Reviewing the List of Codes, Revising the List of Categories
and Deciding Which Codes Should Appear in Which Category
A number of practical measures can be taken with an initial list of codes and
categories in order to make better sense of the data. These include:
SAGE
2015 SAGE Publications, Ltd. All Rights Reserved.
SAGE Research Methods Datasets Part
1
Page 7 of 14 Identifying Themes and Coding Interview Data: Reflective Practice in
Higher Education
• Identifying codes which should be placed in pre-set categories.
• Creating sub-categories within the initial categories.
• Identifying new categories which can bring together a number of codes.
• Identifying codes that apply to sufficient numbers of respondents to be part
of the findings even though they stand outside any category.
• Identifying codes that stand outside any category and should be discarded
because they do not apply to sufficient numbers of respondents.
Each of these steps is demonstrated below for the data relating to reflective
practice, where the initial list of codes was long and unwieldy. The list is shown
below with the name(s) of the respondent(s) who each code applied to. Despite
its length, it is included in full, in the hope that you will not be disheartened if your
initial list looks equally unmanageable:
Complete List of Codes Used in Relation to Reflective Practice
RP important: Fern, Susan, Rachel, Lewis, Thomas
Mechanism – student feedback: Fern, Susan
Mechanism – personal reflection: Fern, Susan, Rachel, Lewis, Thomas
RP should be constant: Fern, Susan, Rachel, Lewis,
Motivation – pride: Fern
RP for both teaching and research: Fern
Lecturers who do not reflect become outdated and stale: Susan
Motivation – for students and lecturer to enjoy teaching: Susan
Internal and external motivation for RP: Susan
Mechanism – personal teaching reviews: Susan
Mechanism – comparing with practice elsewhere in the faculty: Susan
Danger of dreading teaching: Susan
Aim of RP – to excite and engage students: Susan
Accepts responsibility for students’ reaction: Lewis,
SAGE
2015 SAGE Publications, Ltd. All Rights Reserved.
SAGE Research Methods Datasets Part
1
Page 8 of 14 Identifying Themes and Coding Interview Data: Reflective Practice in
Higher Education
Motivation – to teach well: Rachel
Motivation – to stay updated: Rachel
Failure to reflect leads to outdated practice: Rachel
Change of job helpful: Rachel,
Motivation internal: Rachel, Lewis, Thomas
Need for internal pressure: Rachel
Surprised by lack of external pressure for RP: Rachel, Thomas
Should be collective reflection: Lewis
Limitation to RP – time: Lewis, Thomas
Motivation – to be good at job: Lewis
Motivation – wants to communicate effectively: Lewis
RP should be informal: Lewis
Limitation to RP – teaching new subjects (implied): Thomas
RP has led to improved practice: Thomas
Mechanism – working with colleagues: Thomas
Personal reflection and working with colleagues more effective than
teaching course: Thomas
Used to be greater opportunities for RP: Thomas
Mechanism – watching the teaching of colleagues: Thomas
No opportunity to reflect on bad lecture until next year: Thomas
Delay means reflection will be less effective: Thomas
Motivation – to do the best possible job: Thomas
Motivation internal: Thomas
No oversight of quality of teaching: Thomas
Mechanism – peer review: Thomas
Peer review limited by time: Thomas
Identifying Codes Placed in Pre-Set Categories
The categories identified when first reading through the transcripts were the
SAGE
2015 SAGE Publications, Ltd. All Rights Reserved.
SAGE Research Methods Datasets Part
1
Page 9 of 14 Identifying Themes and Coding Interview Data: Reflective Practice in
Higher Education
obvious starting point in seeking to identify themes. Including the name of the
category in the code makes it easy to bring the relevant codes together. In the
case of mechanisms for reflective practice, the codes that were easy to place in
this category were:
Mechanism – student feedback: Fern, Susan
Mechanism – personal reflection: Fern, Susan, Rachel, Lewis, Thomas
Mechanism – personal teaching reviews: Susan
Mechanism – comparing with practice elsewhere in the faculty: Susan
Mechanism – working with colleagues: Thomas
Mechanism – watching the teaching of colleagues: Thomas
Mechanism – peer review: Thomas
Creating Sub-Categories
A further stage of analysis may be helpful after the codes that should be placed in
a category have been identified. One method of sub-dividing commonalities is the
creation of sub-categories. It may be possible to identify common characteristics
of some codes beyond membership of the main category, meaning that a sub-
category can be created. The grouping together of codes into sub-categories
can contribute substantially to the identification of themes. For example, in the
case of the list of mechanisms for reflective practice above, all except ‘personal
reflection’ and ‘personal teaching reviews’ could be placed into a sub-category of
‘Mechanisms involving working with others’.
Creating New Categories
This is often more difficult than creating sub-categories and may require some
more conceptual thinking. While it may be obvious from the researcher’s list of
codes that some should go together, in other cases they may need to think a little
further about a common factor that could justify the creation of a new category. It
SAGE
2015 SAGE Publications, Ltd. All Rights Reserved.
SAGE Research Methods Datasets Part
1
Page 10 of 14 Identifying Themes and Coding Interview Data: Reflective Practice in
Higher Education
may be reassuring to know that this is a skill, like many others in qualitative data
analysis, which develops with practice.
Identifying Codes That Apply to Sufficient Numbers of Respondents
to be Part of the Findings Although They Stand Outside Any Category
Despite their best efforts to fit as many codes as possible into categories,
qualitative researchers tend to find that they have some codes that simply do not
have much in common with any others. They then have to make an important
decision. Should the codes be retained, because they can contribute to the
findings on their own, or should they be discarded? There is no easy answer to
this question but the simplest method of deciding is by looking at the number of
respondents that the code applies to: I might choose a ‘threshold’ of one quarter
of the respondents.
Identifying Codes That Stand Outside Any Category and Do Not
Apply to Sufficient Numbers of Respondents to Be Considered to
Constitute a Theme
This action is closely related to the previous one. Using the threshold of one
quarter of respondents in this case meant that any code which stood outside a
category and which applied to only one respondent should be eliminated from the
analysis. The codes that were eliminated because they did not fit into any category
and only applied to one respondent were:
RP for both teaching and research: Fern
Used to be greater opportunities for RP: Thomas
No opportunity to reflect on bad lecture until next year: Thomas
Step 4: Looking for Themes and Findings in Each Category
SAGE
2015 SAGE Publications, Ltd. All Rights Reserved.
SAGE Research Methods Datasets Part
1
Page 11 of 14 Identifying Themes and Coding Interview Data: Reflective Practice in
Higher Education
Three pieces of advice are offered when identifying findings:
1. Remember the purpose of thematic analysis
Identifying findings, like every part of thematic analysis, should be guided by
the aims identified by Gibson and Brown (2009: 128–129), i.e. examining
commonality, examining differences and examining relationships. However, not
every dataset, or issue within a dataset, allows for the examination of
relationships, so it may be that only the first two of these aims can be achieved.
Examining relationships is associated more with conceptual findings and building
theory. For the new researcher, identifying similarities and differences within the
data is a very worthwhile first goal of analysis.
2. Be content with simple findings
If the process of creating and modifying categories and codes has been effective,
then identifying findings becomes quite straightforward. Indeed, it is a common
experience for the qualitative researcher to feel disappointed that their findings
are simple and do not seem to be saying anything particularly profound. The skills
that you develop through the analysis will be invaluable when examining more
complex data, which may have a more complicated story to tell.
3. Find ways of expressing trends that avoid the use of numbers
It is rare for qualitative findings to be expressed in terms of specific numbers.
Instead, other words are found to provide indications of trends within the data.
The qualitative researcher must find their own language with which to identify
trends – findings are often expressed in terms such as ‘some’, ‘the majority’ and
‘a number’.
The above three pieces of advice were taken into account when identifying
some of the findings in relation to reflective practice. There were some obvious
SAGE
2015 SAGE Publications, Ltd. All Rights Reserved.
SAGE Research Methods Datasets Part
1
Page 12 of 14 Identifying Themes and Coding Interview Data: Reflective Practice in
Higher Education
commonalities between respondents, which could be simply stated. However,
the findings in relation to some of the other codes – including those in the sub-
category of ‘Mechanisms involving working with others’ – needed a little more
detail when they were noted:
• Several respondents discussed methods of involving colleagues in seeking
to identify best practice. The methods discussed were both formal (e.g. peer
review) and informal (e.g. watching the teaching of colleagues).
• A small number of respondents discussed incorporating student feedback
into their reflective practice.
Reflective Questions
1. Using the template in the downloadable data, code the section of
Jamie’s interview with the lecturer, Lewis, relating to reflective practice.
2. Identify which codes from the list provided should be placed in the
category of motivation for reflective practice. Then sort them into sub-
categories and identify findings in relation to the theme of motivation.
3. This data example has focused on Jamie’s aim to identify feelings
about reflective practice. Another of Jamie’s objectives was to get
lecturers to discuss different types of students and the experience of
teaching them. With this second objective in mind and using the full
interview transcripts available in the download, follow Jamie’s steps to
categorise and code the data.
Further Reading
Charmaz, K. (2006). Constructing Grounded Theory. London: Sage.
Dey, I. (1993). Qualitative Data Analysis. Abingdon: Routledge. The notes made
are considerably longer than the ones that are used in this chapter but they are
SAGE
2015 SAGE Publications, Ltd. All Rights Reserved.
SAGE Research Methods Datasets Part
1
Page 13 of 14 Identifying Themes and Coding Interview Data: Reflective Practice in
Higher Education
helpful as an illustration.
Gibson, W. J, & Brown, A. (2009). Working with Qualitative Data. London: Sage.
SAGE
2015 SAGE Publications, Ltd. All Rights Reserved.
SAGE Research Methods Datasets Part
1
Page 14 of 14 Identifying Themes and Coding Interview Data: Reflective Practice in
Higher Education