Using Thematic Analysis in Psychology

Download as pdf or txt
Download as pdf or txt
You are on page 1of 26

Qualitative Research in Psychology

ISSN: 1478-0887 (Print) 1478-0895 (Online) Journal homepage: www.tandfonline.com/journals/uqrp20

Using thematic analysis in psychology

Virginia Braun & Victoria Clarke

To cite this article: Virginia Braun & Victoria Clarke (2006) Using thematic analysis in
psychology, Qualitative Research in Psychology, 3:2, 77-101, DOI: 10.1191/1478088706qp063oa
To link to this article: https://doi.org/10.1191/1478088706qp063oa

Published online: 21 Jul 2008.

Submit your article to this journal

Article views: 673754

View related articles

Citing articles: 39706 View citing articles

Full Terms & Conditions of access and use can be found at


https://www.tandfonline.com/action/journalInformation?journalCode=uqrp20
www.QualResearchPsych.com Qualitative Research in Psychology 2006; 3: 77 /101

Using thematic analysis in


psychology
Virginia Braun1 and Victoria Clarke2
1
University of Auckland and 2University of the West of England

Thematic analysis is a poorly demarcated, rarely acknowledged, yet widely


used qualitative analytic method within psychology. In this paper, we
argue that it offers an accessible and theoretically flexible approach to
analysing qualitative data. We outline what thematic analysis is, locating it
in relation to other qualitative analytic methods that search for themes or
patterns, and in relation to different epistemological and ontological
positions. We then provide clear guidelines to those wanting to start
thematic analysis, or conduct it in a more deliberate and rigorous way, and
consider potential pitfalls in conducting thematic analysis. Finally, we
outline the disadvantages and advantages of thematic analysis. We
conclude by advocating thematic analysis as a useful and flexible method
for qualitative research in and beyond psychology. Qualitative Research in
Psychology 2006; 3: 77 /101

Key words: epistemology; flexibility; patterns; qualitative psychology;


thematic analysis

Thematic analysis is a poorly demarcated plication and evaluation of thematic ana-


and rarely acknowledged, yet widely used lysis, and one which does so in a way
qualitative analytic method (Boyatzis, accessible to students and those not parti-
1998; Roulston, 2001) within and beyond cularly familiar with qualitative research.1
psychology. In this paper, we aim to fill That is, we aim to write a paper that will
what we, as researchers and teachers in be useful as both a teaching and research
qualitative psychology, have experienced tool in qualitative psychology. Therefore,
as a current gap / the absence of a paper in this paper we discuss theory and
which adequately outlines the theory, ap- method for thematic analysis, and clarify

Correspondence: Virginia Braun, Department of Psychology, University of Auckland, Private Bag 92019, Auckland,
New Zealand.
E-mail: [email protected]

# 2006 Edward Arnold (Publishers) Ltd 10.1191/1478088706qp063oa


78 V Braun and V Clarke

the similarities and differences between manifestations of the method, from within
different approaches that share features in the broad theoretical framework. Second,
common with a thematic approach. there are methods that are essentially in-
Qualitative approaches are incredibly dependent of theory and epistemology, and
diverse, complex and nuanced (Holloway can be applied across a range of theoretical
and Todres, 2003), and thematic analysis and epistemological approaches. Although
should be seen as a foundational method often (implicitly) framed as a realist/experi-
for qualitative analysis. It is the first ential method (Aronson, 1994; Roulston,
qualitative method of analysis that re- 2001), thematic analysis is actually firmly
searchers should learn, as it provides core in the second camp, and is compatible with
skills that will be useful for conducting both essentialist and constructionist para-
many other forms of qualitative analysis. digms within psychology (we discuss this
Indeed, Holloway and Todres (2003: 347) later). Through its theoretical freedom, the-
identify ‘thematizing meanings’ as one of a matic analysis provides a flexible and use-
few shared generic skills across qualitative ful research tool, which can potentially
analysis.2 For this reason, Boyatzis (1998) provide a rich and detailed, yet complex,
characterizes it, not as a specific method, account of data.
but as a tool to use across different meth- Given the advantages of the flexibility of
ods. Similarly, Ryan and Bernard (2000) thematic analysis, it is important that we are
locate thematic coding as a process per- clear that we are not trying to limit this
formed within ‘major’ analytic traditions flexibility. However, an absence of clear and
(such as grounded theory), rather than a concise guidelines around thematic analysis
specific approach in its own right. We means that the ‘anything goes’ critique of
argue thematic analysis should be consid- qualitative research (Antaki et al ., 2002) may
ered a method in its own right. well apply in some instances. With this
One of the benefits of thematic analysis is paper, we hope to strike a balance between
its flexibility. Qualitative analytic methods demarcating thematic analysis clearly / ie,
can be roughly divided into two camps. explaining what it is, and how to do it / and
Within the first, there are those tied to, or ensuring flexibility in relation to how it is
stemming from, a particular theoretical or used, so that it does not become limited and
epistemological position. For some of these constrained, and lose one of its key advan-
/ such as conversation analysis (CA; eg, tages. Indeed, a clear demarcation of this
Hutchby and Wooffitt, 1998) and interpre- method will be useful to ensure that those
tative phenomenological analysis (IPA; eg, who use thematic analysis can make active
Smith and Osborn, 2003) / there is (as yet) choices about the particular form of analysis
relatively limited variability in how the they are engaged in. Therefore, this paper
method is applied, within that framework. seeks to celebrate the flexibility of the
In essence, one recipe guides analysis. For method and provide a vocabulary and
others of these / such as grounded theory ‘recipe’ for people to undertake thematic
(Glaser, 1992; Strauss and Corbin, 1998), analysis in a way that is theoretically and
discourse analysis (DA; eg, Burman and methodologically sound.3 As we will show,
Parker, 1993; Potter and Wetherell, 1987; what is important is that as well as apply-
Willig, 2003) or narrative analysis (Murray, ing a method to data, researchers make
2003; Riessman, 1993) / there are different their (epistemological and other) assump-
Using thematic analysis in psychology 79

tions explicit (Holloway and Todres, 2003). So in Virginia’s example, if she was inter-
Qualitative psychologists need to be clear ested in how ‘sexual pleasure’ was talked
about what they are doing and why, and to about, her data set would consist of all
include the often-omitted ‘how’ they did instances across the entire data corpus that
their analysis in their reports (Attride- had some relevance to sexual pleasure.
Stirling, 2001). These two approaches might sometimes
In this paper we outline: what thematic be combined to produce the data set. Data
analysis is; a 6-phase guide to performing item is used to refer to each individual
thematic analysis; potential pitfalls to piece of data collected, which together
avoid when doing thematic analysis; what make up the data set or corpus. A data
makes good thematic analysis; and advan- item in this instance would be an indivi-
tages and disadvantages of thematic analy- dual surgeon interview, a television docu-
sis. Throughout, we provide exam- mentary, or one particular website. Finally,
ples from the research literature, and our data extract refers to an individual coded
own research. By providing examples, we chunk of data, which has been identified
show the types of research questions and within, and extracted from, a data item.
topics that thematic analysis can be used to There will be many of these, taken from
study. throughout the entire data set, and only a
Before we begin, we need to define a few selection of these extracts will feature in
of the terms used throughout the paper. the final analysis.
Data corpus refers to all data collected for
a particular research project, while data set
refers to all the data from the corpus that
What is thematic analysis?
are being used for a particular analysis.
There are two main ways of choosing the Thematic analysis is a method for identify-
data set (which approach you take depends ing, analysing and reporting patterns
on whether you are coming to the data (themes) within data. It minimally orga-
with a specific question or not / see ‘A nizes and describes your data set in (rich)
number of decisions’ below). First, the data detail. However, frequently if goes further
set may consist of many, or all, individual than this, and interprets various aspects of
data items within your data corpus. So, for the research topic (Boyatzis, 1998). The
example, in a project on female genital range of different possible thematic ana-
cosmetic surgery, Virginia’s data corpus lyses will further be highlighted in relation
consists of interviews with surgeons, to a number of decisions regarding it as a
media items on the topic, and surgeon method (see below).
websites. For any particular analysis, her Thematic analysis is widely used, but
data set might just be the surgeon inter- there is no clear agreement about what
views, just the websites (Braun, 2005b), or thematic analysis is and how you go about
it might combine surgeon data with some doing it (see Attride-Stirling, 2001; Boyat-
media data (eg, Braun, 2005a). Second, the zis, 1998; Tuckett, 2005, for other exam-
data set might be identified by a particular ples). It can be seen as a very poorly
analytic interest in some topic in the data, ‘branded’ method, in that it does not appear
and the data set then becomes all instances to exist as a ‘named’ analysis in the same
in the corpus where that topic is referred. way that other methods do (eg, narrative
80 V Braun and V Clarke

analysis, grounded theory). In this sense, it can be misinterpreted to mean that themes ‘re-
is often not explicitly claimed as the side’ in the data, and if we just look hard enough
they will ‘emerge’ like Venus on the half shell.
method of analysis, when, in actuality, we If themes ‘reside’ anywhere, they reside in
argue that a lot of analysis is essentially our heads from our thinking about our data and
thematic / but is either claimed as some- creating links as we understand them. (Ely et al .,
thing else (such as DA, or even content 1997: 205 /6)
analysis (eg, Meehan et al ., 2000)) or not
identified as any particular method at all / At this point, it is important to acknowledge
for example, data were ‘subjected to quali- our own theoretical positions and values in
tative analysis for commonly recurring relation to qualitative research. We do not
themes’ (Braun and Wilkinson, 2003: 30). subscribe to a naı̈ve realist view of qualita-
If we do not know how people went about tive research, where the researcher can
analysing their data, or what assumptions simply ‘give voice’ (see Fine, 2002) to their
participants. As Fine (2002): 218) argues,
informed their analysis, it is difficult to
even a ‘giving voice’ approach ‘involves
evaluate their research, and to compare
carving out unacknowledged pieces of
and/or synthesize it with other studies on
narrative evidence that we select, edit,
that topic, and it can impede other research-
and deploy to border our arguments’. How-
ers carrying out related projects in the
ever, nor do we think there is one ideal
future (Attride-Stirling, 2001). For these
theoretical framework for conducting quali-
reasons alone, clarity on process and prac-
tative research, or indeed one ideal method.
tice of method is vital. We hope that this
What is important is that the theoretical
paper will lead to more clarity around
framework and methods match what the
thematic analysis. researcher wants to know, and that they
Relatedly, insufficient detail is often gi- acknowledge these decisions, and recognize
ven to reporting the process and detail of them as decisions.
analysis (Attride-Stirling, 2001). It is not Thematic analysis differs from other ana-
uncommon to read of themes ‘emerging’ lytic methods that seek to describe patterns
from the data (although this issue is not across qualitative data / such as ‘thematic’
limited to thematic analysis). For example, DA, thematic decomposition analysis, IPA
Singer and Hunter’s (1999: 67) thematic and grounded theory.5 Both IPA and
discourse analysis of women’s experiences grounded theory seek patterns in the data,
of early menopause identified that ‘several but are theoretically bounded. IPA is at-
themes emerged’ during the analysis. Rubin tached to a phenomenological epistemology
and Rubin (1995: 226) claim that analysis is (Smith et al ., 1999; Smith and Osborn,
exciting because ‘you discover themes and 2003), which gives experience primacy
concepts embedded throughout your inter- (Holloway and Todres, 2003), and is about
views’. An account of themes ‘emerging’ or understanding people’s everyday experi-
being ‘discovered’ is a passive account of ence of reality, in great detail, in order to
the process of analysis, and it denies the gain an understanding of the phenomenon
active role the researcher always plays in in question (McLeod, 2001). To complicate
identifying patterns/themes, selecting matters, grounded theory comes in different
which are of interest, and reporting them versions (Charmaz, 2002). Regardless, the
to the readers (Taylor and Ussher, 2001).4 goal of a grounded theory analysis is to
The language of ‘themes emerging’: generate a plausible / and useful / theory
Using thematic analysis in psychology 81

of the phenomena that is grounded in the knowledge of approaches, such as grounded


data (McLeod, 2001). However, in our ex- theory and DA, it can offer a more accessible
perience, grounded theory seems increas- form of analysis, particularly for those early
ingly to be used in a way that is essentially in a qualitative research career.
grounded theory ‘lite’ / as a set of proce- In contrast to IPA or grounded theory (and
dures for coding data very much akin to other methods like narrative analysis DA or
thematic analysis. Such analyses do not CA), thematic analysis is not wedded to any
appear to fully subscribe to the theoretical pre-existing theoretical framework, and
commitments of a ‘full-fat’ grounded theory, therefore it can be used within different
which requires analysis to be directed to- theoretical frameworks (although not all),
wards theory development (Holloway and and can be used to do different things
Todres, 2003). We argue, therefore, that a within them. Thematic analysis can be an
‘named and claimed’ thematic analysis essentialist or realist method, which reports
means researchers need not subscribe to experiences, meanings and the reality of
the implicit theoretical commitments of participants, or it can be a constructionist
grounded theory if they do not wish to method, which examines the ways in which
produce a fully worked-up grounded-theory events, realities, meanings, experiences and
analysis. so on are the effects of a range of discourses
The term ‘thematic DA’ is used to refer to operating within society. It can also be a
a wide range of pattern-type analysis of ‘contextualist’ method, sitting between the
data, ranging from thematic analysis within two poles of essentialism and construction-
a social constructionist epistemology (ie, ism, and characterized by theories, such as
critical realism (eg, Willig, 1999), which
where patterns are identified as socially
acknowledge the ways individuals make
produced, but no discursive analyse is
meaning of their experience, and, in turn,
conducted), to forms of analysis very
the ways the broader social context im-
much akin to the interpretative repertoire
pinges on those meanings, while retaining
form of DA (Clarke, 2005). Thematic decom-
focus on the material and other limits of
position analysis (eg, Stenner, 1993; Ussher
‘reality’. Therefore, thematic analysis can be
and Mooney-Somers, 2000) is a specifically
a method that works both to reflect reality
named form of ‘thematic’ DA, which iden-
and to unpick or unravel the surface of
tifies patterns (themes, stories) within data, ‘reality’. However, it is important that the
and theorizes language as constitutive of theoretical position of a thematic analysis is
meaning and meaning as social. made clear, as this is all too often left
These different methods share a search unspoken (and is then typically a realist
for certain themes or patterns across an account). Any theoretical framework carries
(entire) data set, rather than within a data with it a number of assumptions about the
item, such as an individual interview or nature of the data, what they represent in
interviews from one person, as in the case of terms of the ‘the world’, ‘reality’, and so
biographical or case-study forms of analy- forth. A good thematic analysis will make
sis, such as narrative analysis (eg, Murray, this transparent.
2003; Riessman, 1993). In this sense, they
more or less overlap with thematic analysis. A number of decisions
As thematic analysis does not require the Thematic analysis involves a number of
detailed theoretical and technological choices which are often not made explicit
82 V Braun and V Clarke

(or are certainly typically not discussed in might appear in relatively little of the data
the method section of papers), but which set. So, researcher judgement is necessary to
need explicitly to be considered and dis- determine what a theme is. Our initial
cussed. In practice, these questions should guidance around this is that you need to
be considered before analysis (and some- retain some flexibility, and rigid rules really
times even collection) of the data begins, do not work. (The question of prevalence is
and there needs to be an ongoing reflexive revisited in relation to themes and sub-
dialogue on the part of the researcher or themes, as the refinement of analysis (see
researchers with regards to these issues, later) will often result in overall themes,
throughout the analytic process. The and sub-themes within those.)
method section of Taylor and Ussher’s Furthermore, the ‘keyness’ of a theme is
(2001) thematic DA of S&M provides a not necessarily dependent on quantifiable
good example of research which presents measures / but rather on whether it cap-
this process explicitly; the method section tures something important in relation to
of Braun and Wilkinson (2003) does not. the overall research question. For exam-
ple, in Victoria’s research on representa-
What counts as a theme? tions of lesbians and gay parents on 26
A theme captures something important talk shows (Clarke and Kitzinger, 2004),
about the data in relation to the research she identified six ‘key’ themes. These six
question, and represents some level of themes were not necessarily the most pre-
patterned response or meaning within the valent themes across the data set / they
data set. An important question to address appeared in between two and 22 of the 26
in terms of coding is: what counts as a talk shows / but together they captured an
pattern/theme, or what ‘size’ does a theme important element of the way in which
need to be? This is a question of prevalence, lesbians and gay men ‘normalize’ their
in terms both of space within each data item families in talk show debates. In this in-
and of prevalence across the entire data set. stance, her thematic analysis was driven by
Ideally, there will be a number of instances this particular analytic question. How she
of the theme across the data set, but more ‘measured’ prevalence is relevant, as pre-
instances do not necessarily mean the valence can be determined in a number of
theme itself is more crucial. As this is different ways. Prevalence was counted at
qualitative analysis, there is no hard-and- the level of the data item (ie, did a theme
fast answer to the question of what propor- appear anywhere in each individual talk
tion of your data set needs to display show?). Alternatively, it could have been
evidence of the theme for it to be considered counted in terms of the number of different
a theme. It is not the case that if it was speakers who articulated the theme, across
present in 50% of one’s data items, it would the entire data set, or each individual
be a theme, but if it was present only in occurrence of the theme across the entire
47%, then it would not be a theme. Nor is it data set (which raises complex questions
the case that a theme is only something that about where an ‘instance’ begins and ends
many data items give considerable attention within an extended sequence of talk / see
to, rather than a sentence or two. A theme Riessman, 1993). Because prevalence was
might be given considerable space in some not crucial to the analysis presented, Vic-
data items, and little or none in others, or it toria chose the most straightforward form,
Using thematic analysis in psychology 83

but it is important to note there is no right or scription is maintained. This might be a


wrong method for determining prevalence. particularly useful method when you are
Part of the flexibility of thematic analysis investigating an under-researched area, or
is that it allows you to determine themes you are working with participants whose
(and prevalence) in a number of ways. What views on the topic are not known.
is important is that you are consistent in An alternative use of thematic analysis is
how you do this within any particular to provide a more detailed and nuanced
analysis. account of one particular theme, or group of
There are various ‘conventions’ for repre- themes, within the data. This might relate to
senting prevalence in thematic (and other a specific question or area of interest within
qualitative) analysis that does not provide a the data (a semantic approach / see below),
quantified measure (unlike much content or to a particular ‘latent’ theme (see below)
analysis, Wilkinson, 2000) / for instance: across the whole or majority of the data set.
‘the majority of participants’ (Meehan et al ., An example of this would be Victoria’s talk
2000: 372), ‘many participants’ (Taylor and show paper, discussed previously (Clarke
Ussher, 2001: 298), or ‘a number of and Kitzinger, 2004), which examined nor-
participants’ (Braun et al ., 2003: 249). malization in lesbians’ and gay men’s ac-
Such descriptors work rhetorically to counts of parenting.
suggest a theme really existed in the data,
and to convince us they are reporting Inductive versus theoretical thematic
truthfully about the data. But do they tell analysis
us much? This is perhaps one area where Themes or patterns within data can be
more debate is needed about how and why identified in one of two primary ways in
we might represent the prevalence of thematic analysis: in an inductive or ‘bot-
themes in the data, and, indeed, whether, tom up’ way (eg, Frith and Gleeson, 2004),
if, and why prevalence is particularly im- or in a theoretical or deductive or ‘top
portant. down’ way (eg, Boyatzis, 1998; Hayes,
1997). An inductive approach means the
A rich description of the data set, or a themes identified are strongly linked to the
detailed account of one particular aspect data themselves (Patton, 1990) (as such, this
It is important to determine the type of form of thematic analysis bears some simi-
analysis you want to do, and the claims larity to grounded theory). In this approach,
you want to make, in relation to your data if the data have been collected specifically
set. For instance, you might wish to provide for the research (eg, via interview or focus
a rich thematic description of your entire group), the themes identified may bear little
data set, so that the reader gets a sense of the relation to the specific questions that were
predominant or important themes. In this asked of the participants. They would also
case, the themes you identify, code, and not be driven by the researcher’s theoretical
analyse would need to be an accurate reflec- interest in the area or topic. Inductive
tion of the content of the entire data set. In analysis is therefore a process of coding
such an analysis, some depth and complex- the data without trying to fit it into a pre-
ity is necessarily lost (particularly if you are existing coding frame, or the researcher’s
writing a short dissertation or article with analytic preconceptions. In this sense, this
strict word limits), but a rich overall de- form of thematic analysis is data-driven.
84 V Braun and V Clarke

However, it is important to note, as we Semantic or latent themes


discussed earlier, that researchers cannot Another decision revolves around the ‘le-
free themselves of their theoretical and vel’ at which themes are to be identified: at
epistemological commitments, and data a semantic or explicit level, or at a latent or
are not coded in an epistemological va- interpretative level (Boyatzis, 1998).6 A
cuum. thematic analysis typically focuses exclu-
In contrast, a ‘theoretical’ thematic analy- sively or primarily on one level. With a
sis would tend to be driven by the research- semantic approach, the themes are identi-
er’s theoretical or analytic interest in the fied within the explicit or surface meanings
area, and is thus more explicitly analyst- of the data, and the analyst is not looking for
driven. This form of thematic analysis tends anything beyond what a participant has
to provide less a rich description of the data said or what has been written. Ideally, the
overall, and more a detailed analysis of analytic process involves a progression from
some aspect of the data. Additionally, the description, where the data have simply
choice between inductive and theoretical been organized to show patterns in seman-
maps onto how and why you are coding the tic content, and summarized, to interpreta-
data. You can either code for a quite specific tion , where there is an attempt to theorize
research question (which maps onto the the significance of the patterns and their
more theoretical approach) or the specific broader meanings and implications (Patton,
research question can evolve through the 1990), often in relation to previous literature
coding process (which maps onto the in- (for an excellent example of this, see Frith
ductive approach). and Gleeson, 2004).
For example, if a researcher was inter- In contrast, a thematic analysis at the
ested in talk about heterosex, and had latent level goes beyond the semantic con-
collected interview data, with an inductive tent of the data, and starts to identify or
approach they would read and re-read the examine the underlying ideas, assumptions,
data for any themes related to heterosex, and conceptualizations / and ideologies /
and code diversely, without paying atten- that are theorized as shaping or informing
tion to the themes that previous research on the semantic content of the data. If we
the topic might have identified. For exam- imagine our data three-dimensionally as
ple, the researcher would not look to the an uneven blob of jelly, the semantic
influential research of Hollway (1989), approach would seek to describe the surface
identifying discourses of heterosex, and of the jelly, its form and meaning, while the
code just for male sexual drive, have/hold latent approach would seek to identify the
or permissive discourse themes. In contrast, features that gave it that particular form and
with a theoretical approach, the researcher meaning. Thus, for latent thematic analysis,
may well be interested in the way permis- the development of the themes themselves
siveness plays out across the data, and involves interpretative work, and the ana-
focus on that particular feature in coding lysis that is produced is not just descrip-
the data. This would then result in a tion, but is already theorized.
number of themes around permissiveness, Analysis within this latter tradition tends
which may include, speak to, or expand on to come from a constructionist paradigm
something approximating Hollway’s origi- (eg, Burr, 1995), and in this form, thematic
nal theme. analysis overlaps with some forms of ‘DA’
Using thematic analysis in psychology 85

(which are sometimes specifically referred instead seeks to theorize the sociocultural
to as ‘thematic DA’ (eg, Singer and Hunter, contexts, and structural conditions, that
1999; Taylor and Ussher, 2001)), where enable the individual accounts that are
broader assumptions, structures and/or provided. Thematic analysis that focuses
meanings are theorized as underpinning on ‘latent’ themes tends to be more con-
what is actually articulated in the data. structionist, and it also tends to start to
Increasingly, a number of discourse analysts overlap with thematic DA at this point.
are also revisiting psycho-analytic modes of However, not all ‘latent’ thematic analysis
interpretation (eg, Hollway and Jefferson, is constructionist.
2000), and latent thematic analysis would
also be compatible with that framework. The many questions of qualitative research
It is worth briefly noting that qualitative
Epistemology: essentialist/realist versus research involves a series of questions, and
constructionist thematic analysis there is a need to be clear about the relation-
As we have argued, thematic analysis can be ship between these different questions.
conducted within both realist/essentialist First, there is the overall research question
and constructionist paradigms, although or questions that drive the project. A re-
the outcome and focus will be different for search question might be very broad (and
each. The question of epistemology is exploratory), such as ‘how is lesbian and
usually determined when a research project gay parenting constructed?’ or ‘what are the
is being conceptualized, although episte- meanings of the vagina?’. Narrower research
mology may also raise its head again during questions might be ‘how and why is lesbian
analysis, when the research focus may shift and gay parenting normalized?’ (Clarke and
to an interest in different aspects of the data. Kitzinger, 2004), or ‘what are the discourses
The research epistemology guides what you around vaginal size?’ (see Braun and Kit-
can say about your data, and informs how zinger, 2001). These narrow questions may
you theorize meaning. For instance, with an be part of a broader overarching research
essentialist/realist approach, you can theo- question, and if so, the analyses they inform
rize motivations, experience, and meaning would also provide answers to the overall
in a straightforward way, because a simple, research question. Although all projects are
largely unidirectional relationship is as- guided by research questions, these may
sumed between meaning and experience also be refined as a project progresses.
and language (language reflects and enables Second, if data from interviews or focus
us to articulate meaning and experience) groups have been collected, there are the
(Potter and Wetherell, 1987; Widdicombe questions that participants have responded
and Wooffitt, 1995). to. Finally, there are the questions that
In contrast, from a constructionist per- guide the coding and analysis of the data.
spective, meaning and experience are so- There is no necessary relationship between
cially produced and reproduced, rather these three, and indeed, it is often desirable
than inhering within individuals (Burr, that there is a disjuncture between them.
1995). Therefore, thematic analysis con- Some of the worst examples of ‘thematic’
ducted within a constructionist framework analysis we have read have simply used
cannot and does not seek to focus on the questions put to participants as the
motivation or individual psychologies, but ‘themes’ identified in the ‘analysis’ /
86 V Braun and V Clarke

although in such instances, no analysis has (Ryan and Bernard, 2000: 780). Analysis
really been done at all! involves a constant moving back and for-
To sum up, thematic analysis involves the ward between the entire data set, the coded
searching across a data set / be that a extracts of data that you are analysing, and
number of interviews or focus groups, or a the analysis of the data that you are produ-
range of texts / to find repeated patterns of cing. Writing is an integral part of analysis,
meaning. The exact form and product of not something that takes place at the end, as
thematic analysis varies, as indicated above, it does with statistical analyses. Therefore,
and so it is important that the questions writing should begin in phase one, with the
outlined above are considered before and jotting down of ideas and potential coding
during thematic analyses. Those appro- schemes, and continue right through the
aches which consider specific aspects, la- entire coding/analysis process.
tent themes and are constructionist tend to There are different positions regarding
often cluster together, while those that when you should engage with the literature
consider meanings across the whole data relevant to your analysis / with some
set, semantic themes, and are realist, often arguing that early reading can narrow your
cluster together. However, there are no hard- analytic field of vision, leading you to focus
and-fast rules in relation to this, and differ- on some aspects of the data at the expense
ent combinations are possible. What is of other potentially crucial aspects. Others
important is that the finished product con- argue that engagement with the literature
tains an account / not necessarily that can enhance your analysis by sensitizing
detailed / of what was done, and why. So you to more subtle features of the data
what does one actually do? We now provide
(Tuckett, 2005). Therefore, there is no one
what is, we hope, a straightforward step-
right way to proceed with reading for the-
by-step guide to conducting thematic ana-
matic analysis, although a more inductive
lysis.
approach would be enhanced by not enga-
ging with literature in the early stages of
analysis, whereas a theoretical approach
Doing thematic analysis: a step-by-step requires engagement with the literature
guide prior to analysis.
We provide an outline guide through the
Some of the phases of thematic analysis are six phases of analysis, and offer examples to
similar to the phases of other qualitative demonstrate the process.7 The different
research, so these stages are not necessarily phases are summarized in Table 1. It is
all unique to thematic analysis. The process important to recognize that qualitative ana-
starts when the analyst begins to notice, lysis guidelines are exactly that / they are
and look for, patterns of meaning and not rules, and, following the basic precepts,
issues of potential interest in the data / will need to be applied flexibly to fit the
this may be during data collection. The research questions and data (Patton, 1990).
endpoint is the reporting of the content Moreover, analysis is not a linear process of
and meaning of patterns (themes) in the simply moving from one phase to the next.
data, where ‘themes are abstract (and often Instead, it is more recursive process, where
fuzzy) constructs the investigators identify movement is back and forth as needed,
[sic] before, during, and after analysis’ throughout the phases. It is also a process
Using thematic analysis in psychology 87

Table 1 Phases of thematic analysis

Phase Description of the process

1. Familiarizing yourself Transcribing data (if necessary), reading and re-reading the data, noting down
with your data: initial ideas.
2. Generating initial codes: Coding interesting features of the data in a systematic fashion across the entire
data set, collating data relevant to each code.
3. Searching for themes: Collating codes into potential themes, gathering all data relevant to each
potential theme.
4. Reviewing themes: Checking if the themes work in relation to the coded extracts (Level 1) and the
entire data set (Level 2), generating a thematic ‘map’ of the analysis.
5. Defining and naming Ongoing analysis to refine the specifics of each theme, and the overall story the
themes: analysis tells, generating clear definitions and names for each theme.
6. Producing the report: The final opportunity for analysis. Selection of vivid, compelling extract
examples, final analysis of selected extracts, relating back of the analysis to the
research question and literature, producing a scholarly report of the analysis.

that develops over time (Ely et al ., 1997), research will become apparent / the read-
and should not be rushed. ing and re-reading of data is time-consum-
ing. It is, therefore, tempting to skip over
Phase 1: familiarizing yourself with your this phase, or be selective. We would
data strongly advise against this, as this phase
When you engage in analysis, you may have provides the bedrock for the rest of the
collected the data yourself, or they may have analysis.
been given to you. If you collected them During this phase, it is a good idea to start
through interactive means, you will come to taking notes or marking ideas for coding
the analysis with some prior knowledge of that you will then go back to in subsequent
the data, and possibly some initial analytic phases. Once you have done this, you are
interests or thoughts. Regardless, it is vital ready to begin, the more formal coding
that you immerse yourself in the data to the process. In essence, coding continues to be
extent that you are familiar with the depth developed and defined throughout the en-
and breadth of the content. Immersion tire analysis.
usually involves ‘repeated reading’ of the
data, and reading the data in an active way / Transcription of verbal data
searching for meanings, patterns and so on. If you are working with verbal data, such as
It is ideal to read through the entire data set interviews, television programmes or poli-
at least once before you begin your coding, tical speeches, the data will need to be
as ideas and identification of possible pat- transcribed into written form in order to
terns will be shaped as you read through. conduct a thematic analysis. The process of
Whether or not you are aiming for an transcription, while it may seen time-con-
overall or detailed analysis, are searching suming, frustrating, and at times boring, can
for latent or semantic themes, or are data- or be an excellent way to start familiarizing
theoretically-driven will inform how the yourself with the data (Riessman, 1993).
reading proceeds. Regardless, it is impor- Further, some researchers even argue
tant to be familiar with all aspects of your it should be seen as ‘a key phase of
data. At this phase, one of the reasons why data analysis within interpretative qualita-
qualitative research tends to use far smaller tive methodology’ (Bird, 2005: 227), and
samples than, for example, questionnaire recognized as an interpretative act, where
88 V Braun and V Clarke

meanings are created, rather than simply a ing and interpretative skills needed to ana-
mechanical act of putting spoken sounds on lyse the data (Lapadat and Lindsay, 1999). If
paper (Lapadat and Lindsay, 1999). your data have already been, or will be,
Various conventions exist for transforming transcribed for you, it is important that you
spoken texts into written texts (see Edwards spend more time familiarising yourself with
and Lampert, 1993; Lapadat and Lindsay, the data, and also check the transcripts back
1999). Some systems of transcription have against the original audio recordings for
been developed for specific forms of analysis ‘accuracy’ (as should always be done).
/ such as the ‘Jefferson’ system for CA (see
Atkinson and Heritage, 1984; Hutchby and Phase 2: generating initial codes
Wooffitt, 1998). However, thematic analysis, Phase 2 begins when you have read and
even constructionist thematic analysis, does familiarized yourself with the data, and have
not require the same level of detail in the generated an initial list of ideas about what
transcript as conversation, discourse or even is in the data and what is interesting about
narrative analysis. As there is no one way to them. This phase then involves the produc-
conduct thematic analysis, there is no one set tion of initial codes from the data. Codes
of guidelines to follow when producing a identify a feature of the data (semantic
transcript. However, at a minimum it re- content or latent) that appears interesting
quires a rigorous and thorough ‘ortho- to the analyst, and refer to ‘the most basic
graphic’ transcript / a ‘verbatim’ account of segment, or element, of the raw data or
all verbal (and sometimes nonverbal / eg, information that can be assessed in a mean-
coughs) utterances.8 What is important is ingful way regarding the phenomenon’
that the transcript retains the information (Boyatzis, 1998: 63). See Figure 1 for an
you need, from the verbal account, and in a example of codes applied to a short segment
way which is ‘true’ to its original nature (eg, of data. The process of coding is part of
punctuation added can alter the meaning of analysis (Miles and Huberman, 1994), as you
data / for example ‘I hate it, you know. I do’ are organising your data into meaningful
versus ‘I hate it. You know I do’, Poland, groups (Tuckett, 2005). However, your
2002: 632), and that the transcription con- coded data differ from the units of analysis
vention is practically suited to the purpose of (your themes), which are (often) broader.
analysis (Edwards, 1993). Your themes, which you start to develop in
As we have noted, the time spent in the next phase, are where the interpretative
transcription is not wasted, as it informs analysis of the data occurs, and in relation to
the early stages of analysis, and you will which arguments about the phenomenon
develop a far more thorough understanding being examined are made (Boyatzis, 1998).
of your data through having transcribed it. Coding will, to some extent, depend on
Furthermore, the close attention needed to whether the themes are more ‘data-driven’
transcribe data may facilitate the close read- or ‘theory-driven’ / in the former, the

Data extract Coded for


it's too much like hard work I mean how much paper have you got to sign 1. Talked about with partner
to change a flippin’ name no I I mean no I no we we have thought about it
2. Too much hassle to change name
((inaudible)) half heartedly and thought no no I jus- I can’t be bothered,
it’s too much like hard work. (Kate F07a)

Figure 1 Data extract, with codes applied (from Clarke et al ., 2006)


Using thematic analysis in psychology 89

themes will depend on the data, but in the lost (Bryman, 2001); and (c) remember that
latter, you might approach the data with you can code individual extracts of data in
specific questions in mind that you wish to as many different ‘themes’ as they fit into /
code around. It will also depend on whether so an extract may be uncoded, coded once,
you are aiming to code the content of the or coded many times, as relevant. Note that
entire data set, or whether you are coding to no data set is without contradiction, and a
identify particular (and possibly limited) satisfactory thematic ‘map’ that you will
features of the data set. Coding can be eventually produce / an overall conceptua-
performed either manually or through a lization of the data patterns, and relation-
software programme (see, eg, Kelle, 2004; ships between them9 / does not have to
Seale, 2000, for discussion of software smooth out or ignore the tensions and
programmes). inconsistencies within and across data
Work systematically through the entire items. It is important to retain accounts
data set, giving full and equal attention to that depart from the dominant story in the
each data item, and identify interesting analysis, so do not ignore these in your
aspects in the data items that may form coding.
the basis of repeated patterns (themes)
Phase 3: searching for themes
across the data set. There are a number of
Phase 3 begins when all data have been
ways of actually coding extracts. If coding
initially coded and collated, and you have a
manually, you can code your data by writ-
long list of the different codes that you have
ing notes on the texts you are analysing,
identified across the data set. This phase,
by using highlighters or coloured pens to
which re-focuses the analysis at the broader
indicate potential patterns, or by using level of themes, rather than codes, involves
‘post-it’ notes to identify segments of data. sorting the different codes into potential
You may initially identify the codes, and themes, and collating all the relevant coded
then match them with data extracts that data extracts within the identified themes.
demonstrate that code, but it is important in Essentially, you are starting to analyse your
this phase to ensure that all actual data codes and consider how different codes
extracts are coded, and then collated to- may combine to form an overarching theme.
gether within each code. This may involve It may be helpful at this phase to use visual
copying extracts of data from individual representations to help you sort the differ-
transcripts or photocopying extracts of ent codes into themes. You might use tables,
printed data, and collating each code to- or mind-maps, or write the name each code
gether in separate computer files or using (and a brief description) on a separate piece
file cards. If using computer software, you of paper and play around with organizing
code by tagging and naming selections of them into theme-piles. A thematic map of
text within each data item. this early stage can be seen in Figure 2 (the
Key advice for this phase is: (a) code for as examples in Figures 2 /4 come from the
many potential themes/patterns as possible analysis presented in Braun and Wilkinson,
(time permitting) / you never know what 2003 of women’s talk about the vagina).
might be interesting later; (b) code extracts This is when you start thinking about the
of data inclusively / ie, keep a little of the relationship between codes, between
surrounding data if relevant, a common themes, and between different levels of
criticism of coding is that the context is themes (eg, main overarching themes and
90 V Braun and V Clarke

Figure 2 Initial thematic map, showing five main themes (final analysis presented in Braun and
Wilkinson, 2003)

sub-themes within them). Some initial You end this phase with a collection of
codes may go on to form main themes, candidate themes, and sub-themes, and all
whereas others may form sub-themes, and extracts of data that have been coded in
others still may be discarded. At this stage, relation to them. At this point, you will start
you may also have a set of codes that do not to have a sense of the significance of
seem to belong anywhere, and it is perfectly individual themes. However, do not aban-
acceptable to create a ‘theme’ called ‘mis- don anything at this stage, as without
cellaneous’ to house the codes / possibly looking at all the extracts in detail (the
temporarily / that do not seem to fit into next phase) it is uncertain whether the
your main themes. themes hold as they are, or whether some

Figure 3 Developed thematic map, showing three main themes (final analysis presented in Braun and
Wilkinson, 2003)
Using thematic analysis in psychology 91

Figure 4 Final thematic map, showing final two main themes (see Braun and Wilkinson, 2003).

need to be combined, refined and separated, rework your theme, creating a new theme,
or discarded. finding a home for those extracts that do not
currently work in an already-existing
Phase 4: reviewing themes theme, or discarding them from the analy-
Phase 4 begins when you have devised a set sis. Once you are satisfied that your candi-
of candidate themes, and it involves the date themes adequately capture the
refinement of those themes. During this contours of the coded data / once you
phase, it will become evident that some have a candidate ‘thematic map’ / you
candidate themes are not really themes (eg, are ready to move on to level two of this
if there are not enough data to support them, phase. The outcome of this refinement
or the data are too diverse), while others process can be seen in the thematic map
might collapse into each other (eg, two presented in Figure 3.
apparently separate themes might form Level two involves a similar process, but
one theme). Other themes might need in relation to the entire data set. At this
to be broken down into separate themes.
level, you consider the validity of indivi-
Patton’s (1990) for dual criteria judging
dual themes in relation to the data set, but
categories / internal homogeneity and ex-
also whether your candidate thematic map
ternal heterogeneity / are worth consider-
‘accurately’ reflects the meanings evident in
ing here. Data within themes should cohere
the data set as a whole. To some extent,
together meaningfully, while there should
what counts as ‘accurate representation’
be clear and identifiable distinctions be-
tween themes. depends on your theoretical and analytic
This phase involves two levels of review- approach. However, in this phase you re-
ing and refining your themes. Level one read your entire data set for two purposes.
involves reviewing at the level of the coded The first is, as discussed, to ascertain
data extracts. This means you need to read whether the themes ‘work’ in relation to
all the collated extracts for each theme, and the data set. The second is to code any
consider whether they appear to form a additional data within themes that has been
coherent pattern. If your candidate themes missed in earlier coding stages. The need for
do appear to form a coherent pattern, you re-coding from the data set is to be expected
then move on to the second level of this as coding is an ongoing organic process.
phase. If your candidate themes do not fit, If the thematic map works, then you
you will need to consider whether the moves on to the next phase. However, if
theme itself is problematic, or whether the map does not fit the data set, you need
some of the data extracts within it simply to return to further reviewing and refining
do not fit there / in which case, you would of your coding until you have devised a
92 V Braun and V Clarke

satisfactory thematic map. In so doing, it is you do not just paraphrase the content of
possible that you will identify potential the data extracts presented, but identify
new themes, and you will need to start what is of interest about them and why.
coding for these as well, if they are of For each individual theme, you need to
interest and relevent. However, a word conduct and write a detailed analysis. As
of warning: as coding data and generating well as identifying the ‘story’ that each
themes could go on ad infinitum , it is theme tells, it is important to consider
important not to get over-enthusiastic with how it fits into the broader overall ‘story’
endless re-coding. It is impossible to pro- that you are telling about your data, in
vide clear guidelines on when to stop, but relation to the research question or ques-
when your refinements are not adding any- tions, to ensure there is not too much
thing substantial, stop! If the process of overlap between themes. So it is necessary
recoding is only fine-tuning and making to consider the themes themselves, and
more nuanced a coding frame that already each theme in relation to the others. As
works / ie, it fits the data well / recognize part of the refinement, you will need to
this and stop. Consider it as similar to identify whether or not a theme contains
editing written work / you could endlessly any sub-themes. Sub-themes are essentially
edit your sentences and paragraphs, but themes-within-a-theme. They can be useful
after a few editing turns, any further work for giving structure to a particularly large
is usually unnecessary refinement / similar and complex theme, and also for demon-
to rearranging the hundreds and thousands strating the hierarchy of meaning within the
on an already nicely decorated cake. data. For instance, in one of Virginia’s
At the end of this phase, you should have
analyses of women’s talk about the vagina,
a fairly good idea of what your different
she identified two overarching themes in
themes are, how they fit together, and the
women’s talk: the vagina as liability, and the
overall story they tell about the data.
vagina as asset (Braun and Wilkinson,
Phase 5: defining and naming themes 2003). Within each theme, three sub-themes
Phase 5 begins when you have a satisfactory were identified: for liability the sub-themes
thematic map of your data / see Figure 4 for were ‘nastiness and dirtiness’, ‘anxieties’
the final refinements of Virginia’s thematic and ‘vulnerability’; for asset the sub-themes
map. At this point, you then define and were ‘satisfaction’, ‘power’ and ‘pleasure’.
further refine the themes you will present However, these eventual final themes and
for your analysis, and analyse the data sub-themes resulted from a process of re-
within them. By ‘define and refine’, we finement of initial themes and sub-themes,
mean identifying the ‘essence’ of what as shown in Figures 2/4.
each theme is about (as well as the themes It is important that by the end of this phase
overall), and determining what aspect of the you can clearly define what your themes are
data each theme captures. It is important and what they are not. One test for this is to
not to try and get a theme to do too much, or see whether you can describe the scope and
to be too diverse and complex. You do this content of each theme in a couple of sen-
by going back to collated data extracts for tences. If not, further refinement of that
each theme, and organizing them into a theme may be needed. Although you will
coherent and internally consistent account, already have given your themes working
with accompanying narrative. It is vital that titles, this is also the point to start thinking
Using thematic analysis in psychology 93

about the names you will give them in the study to study. As a first step, we recom-
final analysis. Names need to be concise, mend looking at published examples of
punchy, and immediately give the reader a thematic analysis, particularly of the speci-
sense of what the theme is about. fic version you are planning to use (this is
made somewhat more difficult in that the-
Phase 6: producing the report matic analysis is often not a named method,
Phase 6 begins when you have a set of fully but you can find examples, eg, Ellis and
worked-out themes, and involves the final Kitzinger, 2002; Kitzinger and Willmott,
analysis and write-up of the report. The 2002; Toerien and Wilkinson, 2004). In
task of the write-up of a thematic anal- order to provide a sense of the sorts of
ysis, whether it is for publication or for a questions you should be asking of your data,
research assignment or dissertation, is to and the sorts of analytic claims you should
tell the complicated story of your data in a be seeking to make, we will discuss a
way which convinces the reader of the merit particularly good example of an inductive
and validity of your analysis. It is important thematic analysis, which emphasizes un-
that the analysis (the write-up of it, includ-
derstanding men’s experiences in relation to
ing data extracts) provides a concise, coher-
the broader social context (see Frith and
ent, logical, non-repetitive and interesting
Gleeson, 2004).
account of the story the data tell / within
Frith and Gleeson (2004) aim to ex-
and across themes. Your write-up must
plore how men’s feelings about their
provide sufficient evidence of the themes
bodies influence their clothing practices,
within the data / ie, enough data extracts to
and they use data gathered in qualitative
demonstrate the prevalence of the theme.
questionnaires from 75 men to answer
Choose particularly vivid examples, or ex-
this question. They report four themes:
tracts which capture the essence of the
practicality of clothing choices; lack of
point you are demonstrating, without un-
necessary complexity. The extract should be concern about appearance; use of cloth-
easily identifiable as an example of the ing to conceal or reveal the body; use of
issue. However, your write-up needs to do clothing to fit cultural ideals. Each theme
more than just provide data. Extracts need is clearly linked back to the overall
to be embedded within an analytic narrative research question, but each is distinct.
that compellingly illustrates the story you They provide a clear sense of the scope
are telling about your data, and your analy- and diversity of each theme, using a
tic narrative needs to go beyond description combination of analyst narrative and
of the data, and make an argument in illustrative data extracts. Where relevant,
relation to your research question. they broaden their analysis out, moving
from a descriptive to an interpretative
level (often relating their claims to exist-
ing literature). For example, in ‘men
Pinning down what interpretative value practicality’, they make sense of
analysis actually entails men’s accounts in relation to gender
norms and stereotypes, linking the ac-
It is difficult to specify exactly what inter- counts individual men provided to the
pretative analysis actually entails, particu- expectations that men / as members of
larly as the specifics of it will vary from society / face. What they do, as analysts,
94 V Braun and V Clarke

is relate the patterns of meaning in men’s does not require the same detailed theore-
responses to an academic analysis of how tical and technical knowledge that ap-
gender operates. In so doing, they de- proaches such as DA or CA do. It is
monstrate the dual position that analysts relatively easy to conduct a good thematic
need to take: as both cultural members analysis on qualitative data, even when you
and cultural commentators. Their ‘discus- are still learning qualitative techniques.
sion’ section makes broader analytic However, there are a number of things that
statements about the overall story that can result in a poor analysis. In this section
the themes tell us about men’s relation- we identify these potential pitfalls, in the
ship with clothing. This story reveals hope that they can be avoided.
that men ‘deliberately and strategically The first of these is a failure to actually
use clothing to manipulate their appear- analyse the data at all! Thematic analysis is
ance to meet cultural ideals of masculi- not just a collection of extracts strung
nity’ (Frith and Gleeson, 2004: 45), in a together with little or no analytic narrative.
way more traditionally associated with Nor is it a selection of extracts with analytic
women. This analysis makes an impor- comment that simply or primarily para-
tant contribution in that it challenges phrases their content. The extracts in the-
perceived wisdom about clothing/appear- matic analysis are illustrative of the analytic
ance and masculinity. points the researcher makes about the data,
As this example demonstrates, your ana- and should be used to illustrate/support an
lytic claims need to be grounded in, but go analysis that goes beyond their specific
beyond, the ‘surface’ of the data, even for a content, to make sense of the data, and tell
‘semantic’ level analysis. The sort of ques- the reader what it does or might mean / as
tions you need to be asking, towards the end discussed above. A second, associated pit-
phases of your analysis, include: ‘What does fall is the using of the data collection
this theme mean?’ ‘What are the assump- questions (such as from an interview sche-
tions underpinning it?’ ‘What are the im- dule) as the ‘themes’ that are reported. In
plications of this theme?’ ‘What conditions
such a case, no analytic work has been
are likely to have given rise to it?’ ‘Why
carried out to identify themes across the
do people talk about this thing in this
entire data set, or make sense of the pattern-
particular way (as opposed to other
ing of responses.
ways)?’ and ‘What is the overall story the
The third is a weak or unconvincing
different themes reveal about the topic?’.
analysis, where the themes do not appear
These sorts of questions should guide the
to work, where there is too much overlap
analysis once you have a clear sense of your
between themes, or where the themes are
thematic map.
not internally coherent and consistent. All
aspects of the theme should cohere around a
central idea or concept. This pitfall has
Potential pitfalls to avoid when doing occurred if, depending on what the analysis
thematic analysis is trying to do, it fails adequately to capture
the majority of the data, or fails to provide a
Thematic analysis is a relatively straight- rich description/interpretation of one or
forward form of qualitative analysis, which more aspects of the data. A weak or un-
Using thematic analysis in psychology 95

convincing analysis can also stem from a The fifth involves a mismatch between
failure to provide adequate examples from theory and analytic claims, or between the
the data / for example, only one or two research questions and the form of thematic
extracts for a theme. This point is essen- analysis used. A good thematic analysis
tially about the rhetorics of presentation, needs to make sure that the interpretations
and the need for the analysis to be convin- of the data are consistent with the theoretical
cing to someone who has not read the entire framework. So, for instance, if you are work-
data set: ‘The ‘‘analysis’’ of the material. . . is ing within an experiential framework, you
a deliberate and self-consciously artful crea- would typically not make claims about the
tion by the researcher, and must be con- social construction of the research topic, and
structed to persuade the reader of the if you were doing constructionist thematic
analysis, you would not treat people’s talk
plausibility of an argument’ (Foster and
of experience as a transparent window on
Parker, 1995: 204). In so doing, one avoids
their world. Finally, even a good and inter-
(the appearance of) what Bryman (1988) has
esting analysis which fails to spell out
referred to as ‘anecdotalism’ in qualitative
its theoretical assumptions, or clarify how it
research / where one or a few instances of a was undertaken, and for what purpose, is
phenomenon are reified into a pattern or lacking crucial information (Holloway and
theme, when it or they are actually idiosyn- Todres, 2003), and thus fails in one aspect.
cratic. This is not to say that a few instances
cannot be of interest, or revealing; but it is
important not to misrepresent them as an
overarching theme. What makes good thematic analysis?
The fourth pitfall is a mismatch between
the data and the analytic claims that are One of the criticisms of qualitative research
made about it. In such an (unfounded) from those outside the field is the percep-
analysis, the claims cannot be supported tion that ‘anything goes’. For instance, this
by the data, or, in the worst case, the data sentiment is echoed in the first sentence of
extracts presented suggest another analysis Laubschagne’s (2003) abstract: ‘For many
or even contradict the claims. The re- scientists used to doing quantitative studies
searcher needs to make sure that their the whole concept of qualitative research is
interpretations and analytic points are con- unclear, almost foreign, or ‘‘airy fairy’’ / not
‘‘real’’ research.’ However, although ‘quali-
sistent with the data extracts. A weak
tative’ research cannot be subjected to the
analysis does not appear to consider other
same criteria as ‘quantitative’ approaches, it
obvious alternative readings of the data, or
does provide methods of analysis that
fails to consider variation (and even contra-
should be applied rigorously to the data.
diction) in the account that is produced. A
Furthermore, criteria for conducting good
pattern in data is rarely, if ever, going to be qualitative research / both data collection
100% complete and non-contradicted, so an and analysis / do exist (eg, Elliott et al .,
analysis which suggests that it is, without a 1999; Parker, 2004; Seale, 1999; Silverman,
thorough explanation, is open to suspicion. 2000; Yardley, 2000). The British Psycholo-
It is important to pick compelling examples gical Society offers relatively succinct on-
to demonstrate the themes, so give this line guidelines for assessing quality in qua-
considerable thought. litative research (see http://www.bps.org.
96 V Braun and V Clarke

uk/publications/journals/joop/qualitative- whose assumptions are congruent with the


guidelines.cfm). ‘Criteria’ for assessing qua- way one conceptualizes the subject matter’
litative research is a not uncontroversial (Reicher and Taylor, 2005: 549). A concise
topic, with concerns raised about rigid checklist of criteria to consider when deter-
criteria limiting freedom and stifling meth- mining whether you have generated a good
odological development (Elliott et al ., 1999; thematic analysis is provided in Table 2.
Parker, 2004; Reicher, 2000). Reicher (2000)
takes the critique further, by asking whether
the incredibly diverse range of qualitative So what does thematic analysis offer
approaches can and should be subject to
psychologists?
the same criteria.
Bracketing these critiques off, the issues We now end this paper with some brief
raised in many general qualitative research comments on the advantages and disadvan-
assessment criteria can be more or less tages of thematic analysis. As we have
applied to thematic forms of analysis. As shown throughout this paper, thematic ana-
thematic analysis is a flexible method, you lysis is not a complex method. Indeed, as
also need to be clear and explicit about what you can see from Table 3, its advantages are
you are doing, and what you say you are many. However, it is not without some
doing needs to match up with what you disadvantages, which we will now briefly
actually do. In this sense, the theory and consider. Many of the disadvantages de-
method need to be applied rigorously, and pend more on poorly conducted analyses
‘rigour lies in devising a systematic method or inappropriate research questions than on

Table 2 A 15-point checklist of criteria for good thematic analysis

Process No. Criteria

Transcription 1 The data have been transcribed to an appropriate level of detail, and the transcripts
have been checked against the tapes for ‘accuracy’.
Coding 2 Each data item has been given equal attention in the coding process.
3 Themes have not been generated from a few vivid examples (an anecdotal approach),
but instead the coding process has been thorough, inclusive and comprehensive.
4 All relevant extracts for all each theme have been collated.
5 Themes have been checked against each other and back to the original data set.
6 Themes are internally coherent, consistent, and distinctive.
Analysis 7 Data have been analysed / interpreted, made sense of / rather than just paraphrased
or described.
8 Analysis and data match each other / the extracts illustrate the analytic claims.
9 Analysis tells a convincing and well-organized story about the data and topic.
10 A good balance between analytic narrative and illustrative extracts is provided.
Overall 11 Enough time has been allocated to complete all phases of the analysis adequately,
without rushing a phase or giving it a once-over-lightly.
Written report 12 The assumptions about, and specific approach to, thematic analysis are clearly
explicated.
13 There is a good fit between what you claim you do, and what you show you have
done / ie, described method and reported analysis are consistent.
14 The language and concepts used in the report are consistent with the epistemological
position of the analysis.
15 The researcher is positioned as active in the research process; themes do not just
‘emerge’.
Using thematic analysis in psychology 97

Table 3 Advantages of thematic analysis allow the researcher to make claims about
language use, or the fine-grained function-
Flexibility.
Relatively easy and quick method to learn, and do. ality of talk.
Accessible to researchers with little or no experience of Finally, it is worth noting that thematic
qualitative research. analysis currently has no particular kudos
Results are generally accessible to educated general
public. as an analytic method / this, we argue,
Useful method for working within participatory re- stems from the very fact that it is poorly
search paradigm, with participants as collaborators. demarcated and claimed, yet widely used.
Can usefully summarize key features of a large body of
data, and/or offer a ‘thick description’ of the data set. This means that thematic analysis is fre-
Can highlight similarities and differences across the quently, or appears to be, what is simply
data set. carried out by someone without the knowl-
Can generate unanticipated insights.
Allows for social as well as psychological interpreta- edge or skills to perform a supposedly more
tions of data. sophisticated / certainly more kudos-bear-
Can be useful for producing qualitative analyses suited ing / ‘branded’ form of analysis like
to informing policy development.
grounded theory, IPA or DA. We hope this
paper will change this view as, we argue, a
rigorous thematic approach can produce an
the method itself. Further, the flexibility of insightful analysis that answers particular
the method / which allows for a wide range research questions. What is important is
of analytic options / means that the poten- choosing a method that is appropriate to
tial range of things that can be said about your research question, rather than falling
victim to ‘methodolatry’, where you are
your data is broad. While this is an advan-
committed to method rather than topic/
tage, it can also be a disadvantage in that it
content or research questions (Holloway
makes developing specific guidelines for
and Todres, 2003). Indeed, your method of
higher-phase analysis difficult, and can be
analysis should be driven by both your
potentially paralysing to the researcher try-
research question and your broader theore-
ing to decide what aspects of their data to
tical assumptions. As we have demon-
focus on. Another issue to consider is that a
strated, thematic analysis is a flexible
thematic analysis has limited interpretative
approach that can be used across a range
power beyond mere description if it is not
of epistemologies and research questions.
used within an existing theoretical frame-
work that anchors the analytic claims that
Notes
are made.
Other disadvantages appear when the- 1. Boyatzis (1998) provides a much more
matic analysis is considered in relation to detailed account of thematic analysis. However,
some of the other qualitative analytic meth- we do not feel that it is a particularly accessible
ods. For instance, unlike narrative or other account for those unfamiliar with qualitative
biographical approaches, you are unable to approaches. Moreover, his approach differs
from ours in that, although he acknowledges
retain a sense of continuity and contradic-
the subjective dimension of qualitative analysis,
tion through any one individual account, his approach is ultimately, if often implicitly,
and these contradictions and consistencies located within a positivist empiricist paradigm.
across individual accounts may be reveal- 2. Dey’s (1993) account of on ‘qualitative data
ing. In contrast to methods similar to DA analysis’, which aims to identify shared techni-
and CA, a simple thematic analysis does not ques across the diverse range of qualitative
98 V Braun and V Clarke

methods, and demonstrate how to do ‘qualitative 6. The definition by Boyatzis (1998) of latent
analysis’, reinforces this point in that his focus is and manifest is somewhat narrower than our
largely thematic / but not claimed as such. identification of latent and semantic, and he
3. Some authors, such as Potter (1997: 147 / identifies thematic analysis as incorporating
48) argue that one should not simply provide both latent and manifest aspects. However, this
‘recipes’ for qualitative methods, such as DA, results from the fact that he associates the
because ‘a large part of doing discourse analysis process of interpretation with latent analysis /
is a craft skill, more like bike riding or sexing a whereas we would argue that it should also be an
chicken than following the recipe for a mild important element of a semantic approach.
chicken rogan josh. . . This makes it hard to 7. We are assuming that you will be working
describe and learn’. While we do not disagree with a ‘good quality’ data corpus and data set.
We would argue that ‘good data’ are defined by a
that the skills needed for qualitative analyses of
particular set of criteria regarding what, why,
all types need to be learned, others, such as
and how they were collected, and offer rich,
McLeod (2001), argue that by not discussing the
detailed and complex accounts of the topic.
‘how to’ of analysis, we keep certain methods
Good data do not just provide a surface over-
mysterious (and thus elitist). Instead, if we want view of the topic of interest, or simply reiterate a
to make methods democratic and accessible / commonsense account. The challenge for the
and indeed, to make qualitative research of all novice researcher is to interact with research
forms more understandable to those not trained participants in such a way that they generate
in the methods, and arguably thus more popular rich and complex insights. Producing a good
/ we need to provide concrete advice on how to analysis of poor quality data is a far more
actually do it. We are not questioning the demanding task for the analyst, although it can
importance of ‘non-recipe’ forms of training, potentially be performed by a skilled and ex-
but while ‘recipes’ necessarily diminish the perienced analyst.
complexity of certain methods, they are impor- 8. See Poland (2002) for a discussion of the
tant for making methods accessible. problems with the idea of a ‘verbatim’ transcript,
4. Foster and Parker (1995) suggest one and what is left out, and retained, through this
way to acknowledge the creative and active role process.
of the analyst is to use the first person when 9. What we mean by thematic map is similar
writing. to, but less detailed than, the ‘codebook’ Ryan
5. Content analysis is another method that and Bernard (2000) refer to, which involves a
can be used to identify patterns across qualitative detailed account of the hierarchical relationship
data, and is sometimes treated as similar to between codes, as well as a description of each,
thematic approaches (eg, Wilkinson, 2000). How- their criteria, exemplars and counter / examples,
ever, content analysis tends to focus at a more and other such details. Like Boyatzis’s (1998)
account of a thematic code, this model is then
micro level, often provides (frequency) counts
applied to (and revised in relation to) the data.
(Wilkinson, 2000), and allows for quantitative
See Figures 2 /4 for visual representations of a
analyses of initially qualitative data (Ryan and
thematic maps and its refinement. Another ex-
Bernard, 2000). Thematic analysis differs from
ample of a thematic map / this time in table
this in that themes tend not to be quantified form / can be found in Frith and Gleeson (2004).
(although sometimes they may be; and Boyatzis
(1998) suggests thematic analysis can be used to
transform qualitative data into a quantitative References
form, and subject them to statistical analyses;
and the unit of analysis tends to be more than a Antaki, C., Billig, M., Edwards, D. and Potter, J.
word or phrase, which it typically is in content 2002: Discourse analysis means doing ana-
analysis. lysis: a critique of six analytic shortcomings.
Using thematic analysis in psychology 99

DAOL Discourse Analysis Online [electronic Clarke, V. 2005: ‘We’re all very liberal in our
version], 1(1). views’: Students’ talk about lesbian and gay
Aronson, J. 1994: A pragmatic view of thematic parenting. Lesbian and Gay Psychology Re-
analysis. Qualitative Report 2(1). view 6, 2 /15.
Atkinson, J.M. and Heritage, J. 1984: Structures Clarke, V. and Kitzinger, C. 2004: Lesbian and gay
of social action: studies in conversation parents on talk shows: resistance or collu-
analysis . Cambridge University Press. sion in heterosexism. Qualitative Research
in Psychology 1, 195 /217.
Attride-Stirling, J. 2001: Thematic networks: an
analytic tool for qualitative research. Quali- Clarke, V., Burns, M. and Burgoyne, C. 2006:
tative Research 1, 385 /405. ‘Who would take whose name?’ An explora-
tory study of naming practices in same-sex
Bird, C.M. 2005: How I stopped dreading and
relationships. Manuscript under submission.
learned to love transcription. Qualitative
Inquiry 11, 226 /48. Dey, I. 1993: Qualitative data analysis: a user-
friendly guide for social scientists . Routledge.
Boyatzis, R.E. 1998: Transforming qualitative
information: thematic analysis and code Edwards, J.A. 1993: Principles and contrasting
development . Sage. systems of discourse transcription. In Ed-
wards, J.A. and Lampert, M.D., editors,
Braun, V. 2005a: In search of (better) female
Talking data: transcription and coding in
sexual pleasure: female genital ‘cosmetic’
discourse research. Lawrence Erlbaum As-
surgery. Sexualities 8, 407 /24.
sociates, 3 /31.
Braun, V. 2005b: Selling the perfect vulva. Manu-
Edwards, J.A. and Lampert, M.D., editors, 1993:
script under submission.
Talking data: transcription and coding in
Braun, V. and Kitzinger, C. 2001: The perfectible discourse research . Lawrence Erlbaum As-
vagina: size matters. Culture Health & Sexu- sociates.
ality 3, 263 /77.
Elliott, R., Fischer, C.T. and Rennie, D.L. 1999:
Braun, V. and Wilkinson, S. 2003: Liability or Evolving guidelines for publication of qua-
asset? Women talk about the vagina. Psycho- litative research studies in psychology and
logy of Women Section Review 5, 28 /42. related fields. British Journal of Clinical
Braun, V., Gavey, N. and McPhillips, K. 2003: The Psychology 38, 215 /29.
‘fair deal?’ Unpacking accounts of recipro- Ellis, S.J. and Kitzinger, C. 2002: Denying equal-
city in heterosex. Sexualities 6, 237 /61. ity: an analysis of arguments against low-
Bryman, A. 1988: Quantity and quality in social ering the age of consent for sex between
research . Routledge. men. Journal of Community and Applied
Bryman, A. 2001: Social research methods . Social Psychology 12, 167 /80.
Oxford University Press. Ely, M., Vinz, R., Downing, M. and Anzul, M.
Burman, E. and Parker, I., editors, 1993: Dis- 1997: On writing qualitative research: living
course analytic research: repertoires and by words . Routledge/Falmer.
readings of texts in action . Routledge. Fine, M. 2002: Disruptive voices: the possibilities
Burr, V. 1995: An introduction to social con- for feminist research . University of Michi-
structionism . Routledge. gan Press.
Charmaz, K. 2002: Qualitative interviewing and Foster, J.J. and Parker, I. 1995: Carrying out
grounded theory analysis. In Gubrium, J.F. investigations in psychology: methods and
and Holstein J.A., editors, Handbook of statistics . BPS Books.
interview research: context and method . Frith, H. and Gleeson, K. 2004: Clothing and
Sage, 675 /94. embodiment: men managing body image and
100 V Braun and V Clarke

appearance. Psychology of Men and Mascu- Murray, M. 2003: Narrative psychology. In Smith,
linity 5, 40 /48. J.A., editor, Qualitative psychology: a practi-
Glaser, B. 1992: Basics of grounded theory cal guide to research methods . Sage, 111 /31.
analysis . Sociology Press. Parker, I. 2004: Criteria for qualitative research in
Hayes, N. 1997: Theory-led thematic analysis: psychology. Qualitative Research in Psy-
social identification in small companies. In chology 1, 95 /106.
Hayes, N., editor, Doing qualitative analysis Patton, M.Q. 1990: Qualitative evaluation and
in psychology. Psychology Press. research methods , second edition. Sage.
Holloway, I. and Todres, L. 2003: The status of Poland, B.D. 2002: Transcription quality. In
method: flexibility, consistency and coher- Gubrium, J.F. and Holstein, J.A., editors,
ence. Qualitative Research 3, 345 /57. Handbook of interview research: context
and method. Sage, 629 /49.
Hollway, W. 1989: Subjectivity and method in
psychology: gender, meaning and science . Potter, J. 1997: Discourse analysis as a way of
Sage. analysing naturally occurring talk. In Silver-
man, D., editor, Qualitative research: theory,
Hollway, W. and Jefferson, T. 2000: Doing quali- method and practice . Sage, 144 /60.
tative research differently: free association,
Potter, J. and Wetherell, M. 1987: Discourse and
narrative and the interview method . Sage.
social psychology: beyond attitudes and
Hutchby, I. and Wooffitt, R. 1998: Conversation behaviour. Sage.
analysis: principles, practices and applica-
Reicher, S. 2000: Against methodolatry: some
tions . Polity Press.
comments on Elliott, Fischer, and Rennie.
Kelle, U. 2004: Computer-assisted analysis of British Journal of Clinical Psychology 39, 1 /
qualitative data. In Flick, U., von Kardorff, 6.
E. and Steinke, I., editors, A companion to Reicher, S. and Taylor, S. 2005: Similarities and
qualitative research . Sage, 276 /83. differences between traditions. Psychologist
Kitzinger, C. and Willmott, J. 2002: ‘The thief of 18, 547 /49.
womanhood’: women’s experience of poly- Riessman, C.K. 1993: Narrative analysis . Sage.
cystic ovarian syndrome. Social Science &
Roulston, K. 2001: Data analysis and ‘theorizing
Medicine 54, 349 /61.
as ideology’. Qualitative Research 1, 279 /
Lapadat, J.C. and Lindsay, A.C. 1999: Transcrip- 302.
tion in research and practice: from standar-
Rubin, H.J. and Rubin, I.S. 1995: Qualitative
dization of technique to interpretive posi- interviewing: the art of hearing data . Sage.
tionings. Qualitative Inquiry 5, 64 /86.
Ryan, G.W. and Bernard, H.R. 2000: Data man-
Laubschagne, A. 2003: Qualitative research / agement and analysis methods. In Denzin,
airy fairy or fundamental? Qualitative Re- N.K. and Lincoln, Y.S., editors, Handbook of
port [electronic version], 8(1). qualitative research , second edition. Sage,
McLeod, J. 2001: Qualitative research in counsel- 769 /802.
ling and psychotherapy. Sage. Seale, C. 1999: The quality of qualitative re-
Meehan, T., Vermeer, C. and Windsor, C. 2000: search . Sage.
Patients’ perceptions of seclusion: a qualita- Seale, C. 2000: Using computers to analyse
tive investigation. Journal of Advanced Nur- qualitative data. In Silverman, D., Doing
sing 31, 370 /77. qualitative research: a practical handbook .
Miles, M.B. and Huberman, A.M. 1994: Qualita- Sage, 155 /74.
tive data analysis: an expanded sourcebook , Silverman, D., editor 2000: Doing qualitative
second edition. Sage. research: a practical handbook . Sage.
Using thematic analysis in psychology 101

Singer, D. and Hunter, M. 1999: The experience Tuckett, A.G. 2005: Applying thematic analysis
of premature menopause: a thematic dis- theory to practice: a researcher’s experience.
course analysis. Journal of Reproductive Contemporary Nurse 19, 75 /87.
and Infant Psychology 17, 63 /81. Ussher, J.M. and Mooney-Somers, J. 2000: Nego-
Smith, J.A. and Osborn, M. 2003: Interpretative tiating desire and sexual subjectivity: narra-
phenomenological analysis. In Smith, J.A., tives of young lesbian avengers. Sexualities
editor, Qualitative psychology: a practical 3, 183 /200.
guide to methods . Sage.
Widdicombe, S. and Wooffitt, R. 1995: The
Smith, J.A., Jarman, M. and Osborn, M. 1999: language of youth subcultures: social iden-
Doing interpretative phenomenological ana- tity in action . Harvester Wheatsheaf.
lysis. In Murray, M. and Chamberlain,
Wilkinson, S. 2000: Women with breast cancer
M.M.K., editors, Qualitative health psychol-
ogy: theories and methods . Sage. talking causes: comparing content, biogra-
phical and discursive analyses. Feminism &
Stenner, P. 1993: Discoursing jealousy. In
Psychology 10, 431 /60.
Burman, E. and Parker, I., editors, Dis-
course analytic research: repertoires and Willig, C. 1999: Beyond appearances: a critical
readings of texts in action. Routledge, 94 / realist approach to social constructionism.
132. In Nightingale, D.J. and Cromby J., editors,
Strauss, A. and Corbin, J. 1998: Basics of quali- Social constructionist psychology: a critical
tative research: techniques and procedures analysis of theory and practice . Open Uni-
for developing grounded theory. Sage. versity Press, 37 /51.
Taylor, G.W. and Ussher, J.M. 2001: Making sense Willig, C. 2003: Discourse analysis. In Smith,
of S&M: a discourse analytic account. Sex- J.A., editor, Qualitative psychology: a prac-
ualities 4, 293 /314. tical guide to research methods . Sage, 159 /
83.
Toerien, M. and Wilkinson, S. 2004: Exploring
the depilation norm: a qualitative question- Yardley, L. 2000: Dilemmas in qualitative health
naire study of women’s body hair removal. research. Psychology and Health 15, 215 /
Qualitative Research in Psychology 1, 69 / 28.
92.

About the authors


VIRGINIA BRAUN is a senior lecturer in the Department of Psychology at the University
of Auckland, where she teaches, supervises and conducts qualitative research. Her
research interests are primarily focused around women’s health, gendered bodies, and
sex and sexuality, and the intersections between these areas. She is currently working on
projects related to ‘sex in long-term relationships’, ‘female genital cosmetic surgery’ and
‘the social context of STI transmission’.
VICTORIA CLARKE is a senior lecture in social psychology at the University of the West
of England. She has published a number of papers on lesbian and gay parenting, and
co-edited two special issues of Feminism and Psychology on Marriage (with Sara-Jane
Finlay and Sue Wilkinson). She is currently conducting ESRC-funded research on same-
sex relationships (with Carol Burgoyne and Maree Burns) and co-editing (with Elizabeth
Peel) a book on LGBTQ psychology (Out in psychology, Wiley).

You might also like