Content uploaded by Oliver Robinson
Author content
All content in this area was uploaded by Oliver Robinson on Mar 23, 2021
Content may be subject to copyright.
1
Conducting thematic analysis on brief texts: The structured tabular approach
Dr Oliver C. Robinson
University of Greenwich
Dreadnought Building, London, SE10 9LS
Email: o.c.robinson@gre.ac.uk
Tel: +44 208 331 9630
2
Abstract
In this article I present a structured approach to thematic analysis that is designed for working
with brief texts. It is grounded in both the ecumenical thematic analysis of Boyatzis and the
reflexive thematic analysis of Braun and Clarke. The process of structured tabular TA (ST-
TA) is best conducted in spreadsheet software such as Microsoft Excel. As with other forms
of thematic analysis, it permits inductive, deductive or hybrid approaches to theme
development and analysis. Its logistical processes are well suited to working with the large
samples that can be achieved when gathering brief text data. It can be used to conduct purely
qualitative analyses, and can also elicit frequency data that can, in principle, be analysed
quantitatively too. The process of checking agreement between analysts is an integral feature
of the method. I discuss the practical implications of the approach and its applicability to
various qualitative and mixed-methods research designs.
Keywords: thematic analysis; brief texts, short stories, flexibility, qualitative psychology,
mixed methods
3
Conducting thematic analysis on brief texts: The structured tabular approach
The development of qualitative research methodologies in psychology and the social sciences
has from the outset been bound up with an emphasis on gathering in-depth data. This
emphasis has presented an important counteractive to the reductionist tendencies of
quantitative psychology. Qualitative research initially emerged in psychology in conjunction
with analysing individual cases or critical incidents in depth. Examples of early work include
Erik Erikson’s biographical case studies of Ghandhi and Luther (1958; 1969); Festinger’s
quasi-ethnographic case study of a UFO cult (Festinger, Rieken, & Schacter, 1956); and
Flanagan’s work developing the Critical Incident Technique and his applications of it on
learning to fly (Flanagan, 1954). From the 1980s, as qualitative methodology became
explicitly recognized within psychology and the social sciences, early sourcebooks on
qualitative methods all focused on in-depth data collection from each case (Glaser & Strauss,
1967; Lincoln and Guba, 1985; Miles & Huberman, 1984; Reason & Rowan, 1981).
Interviews and focus groups subsequently became the most widely used data collection
methods in qualitative psychology (Howitt, 2016).
This focus on long texts (i.e. thousands of words per person or per conversational
interaction) has remained integral to qualitative methods in the intervening decades.
Analytical approaches such as Grounded Theory, the Comparative Method, Conversation
Analysis and Interpretative Phenomenological Analysis were all developed with the aim of
analysing these in-depth texts. Until recently, little has been provided by way of
methodological injunctions for how to work analytically with brief texts, and what the
theoretical and practical arguments are for doing so. To meet this need within a flexible
epistemological framework, in this article I set out a variant of thematic analysis entitled
structured tabular thematic analysis (ST-TA), which offers an adaptable technique for
working with brief qualitative data in a relatively structured way.
My own epistemological is informed by the middle-ground approach of critical
realism, which allows for multiple interpretations of a phenomenon but clearly distinguishes
between better and worse interpretations by the relationship of those ideas to a reality beyond
words and texts (Robinson & Smith, 2010). It is also is founded on the importance of
dialectical reasoning. This means that when I investigate any topic that has debate and
disagreement within it, I actively explore whether a hidden consensus, synthesis or unity can
be found behind the plurality of viewpoints (e.g. Robinson, 2020a). This dialectical reasoning
process involves critically examining and deconstructing apparently opposing presentations
4
of complex matters to seek hidden but often unarticulated common assumptions. Finding a
synthesis or unity in this way does not over-ride plurality and difference; these often happily
co-exist, like multiple notes in a single piano chord. In the context of methodology, I contend
that taking a dialectical approach to word-based and number-based methods shows that both
emerge from a complex range of overlapping epistemologies, all of which are founded on the
central importance of resorting to evidence when asserting facts or generalities. This means
that the incompatibility thesis, which argues that (a) positivism undergirds quantitative
research, (b) interpretivism is the foundation of qualitative research, and (c) these are
incompatible (e.g. Wiggins, 2011), is wrong. I elaborate on this point below, but before that I
consider why brief texts matter to psychology and the social sciences.
The forms and functions of brief texts in qualitative psychology
There are various theoretical and practical arguments for acknowledging the
important role that brief texts (i.e. typically one paragraph or less) currently serve in the
social sciences and why they are likely to become even more important to research in the
future. The first argument is the sheer growth in their prevalence since the rise of social
media. Qualitative studies have already been conducted on social media texts in the form of
YouTube comments (Carpentier, 2014; Mejova & Srinivasan, 2012; Schultes, Dorner &
Lehner, 2013); Facebook posts (Vraga et al. 2015); Twitter feeds (Giles, 2017; Lyles et al.,
2013); and forum-based online discussions (Giles, 2016; Giles, 2014). The accounts of life
events and experiences that are conveyed in social media are referred to by some theorists as
small stories (Georgakopoulou, 2014). They have some advantages over depth data that is
elicited in autobiographical interviews. For example, compared with the generally
retrospective nature of interviews, social media postings typically represent events and
experiences that have happened that very day or may be ongoing, hence they are less heavily
filtered by memory. Furthermore, the socially interactive nature of social media postings,
being composed as initial texts with subsequent comments and replies, can convey how
experiences can be framed and interpreted within an intersubjective frame (Georgakopoulou,
2017).
As well as social media, another important phenomenon that has boosted the
availability of short forms of qualitative data is the online survey platform, such as Qualtrics,
Typeform or QuestionPro. Through these, participants can write brief stories, reflections or
respond to open-ended questions. Such data is important for qualitative psychology for at
least the following reasons. Firstly, it allows access to hard-to-reach sample groups or
5
geographically dispersed populations that standard depth methods struggle to reach (Terry &
Braun, 2017). Secondly, such data collection allows for total anonymity, which can be an
ethical strength when asking individuals to disclose information about highly personal or
sensitive topics (Slepian & Moulton-Tetlock, 2018). Thirdly, using online platforms allows
for gathering a larger, and hence potentially more representative, sample than in-depth
methods. This can be an advantage if the aim of a qualitative study is to make inductive
claims about a broader population group from which the sample is drawn. Such an aim is, for
example, often the case in qualitative evaluation studies that make claims about intervention
efficacy (Thomas, 2006).
For an extensive exposition of the functions and potentials of qualitative surveys, the
reader is directed to Terry and Braun (2017). These authors present a theoretical and practical
guide to this form of data collection, exemplifying their approach with a qualitative survey
study on views about body hair removal, which was conducted via this method with a sample
of over 600 participants from New Zealand. Another technique for eliciting data that can be
captured via online survey platforms is the story completion method. In this method, the first
sentence of a story is provided about a specific topic. This must then be completed by
participants, typically of a few hundred words in length (Clarke et al., 2018). This method has
recently been used with data collected online to explore parents’ perceptions of the future for
a child with a chronic pain syndrome termed Complex Regional Pain Syndrome (Coningsby
& Jordan, 2019).
Along with the pragmatic opportunities and benefits of working with brief data, there
is a pluralist epistemological argument for working with brief data alongside depth data.
According to this argument, the more varied forms of qualitative data that can be
meaningfully analysed, the more effectively we can grasp the complexities of human
behavior, inner life and interpersonal interaction that can be conveyed through words and text
(Frost et al., 2010). Put another way, much qualitative data is available in small texts, so to
include them fully within the auspices of qualitative methods is to ensure that psychology and
the social sciences reach out to all possible forms of textual data and the potential insights
they contain.
Structured tabular thematic analysis: A conceptual analytic comparison with existing
methods
ST-TA locates itself in a currently unoccupied niche between (a) existing approaches
to thematic analysis, most specifically those of Boyatzis (1998) and Braun & Clarke (2006),
6
and (b) existing approaches to analysing brief texts, such as the narrative analysis of small
stories (Bamberg and Georgakopoulou, 2017). In order to justify that such a niche exists and
is worth occupying, I here present a conceptual comparison with these other methods,
focusing on key differences and similarities with these established approaches. Central to this
discussion is my contention that doing qualitative research does not entail negating the
language of number as a key tool for science. Numbers are symbols and signs that assume
meaning via complex cultural and cognitive networks of sense-making, just as words do
(Osbeck, 2014). The language of number that we use today has evolved over millennia from
a combination of Arab, Hindu and Roman systems, and, as such, has a cultural linguistic
heritage just as written language does (Seife, 2000). Numbers in the context of scientific data
never interpret or explain themselves. Turning numerical data into scientific understanding
entails complex abductive conceptual leaps and inferences that are typically located in the
discussion section of a journal article. I argue that to support qualitative research with the
judicious use of numbers, particularly in calculations of researcher agreement and theme
frequencies, gives additional clarity, precision and meaning to an analysis. On the flipside,
using words to illuminate quantitative data is equally essential.
The development of ST-TA has been influenced by two established approaches to
thematic analysis: the ecumenical approach set out by Boyatzis (1998) and also the reflexive
approach devised by Braun and Clarke (Braun & Clarke, 2006; Braun, Clarke, Hayfield &
Terry, 2018). Both approaches embrace a pragmatic ethos in which the research problem is
paramount. They both also concur that the objective of thematic analysis is seeking recurrent
patterns across multiple cases that point to some kind of meaningful invariance that can help
understand a class of phenomena or events. In both methods, inductive and deductive
research can be legitimately conducted. The approaches both allow for themes to be extracted
at a descriptive/manifest level or latent/inferential level, via a defined yet flexible series of
analytical phases. ST-TA stands on these foundations, which together can be summarised as a
problem-focused commitment to the flexible seeking of patterns and meaning in data that
serve clearly defined research problems, according to clear and explicit parameters of
transparent and rigorous research.
As well as these similarities, ST-TA also has points of difference with both
approaches. Braun and Clarke (2019) have recently argued that formal processes for
establishing agreement across analysts dilute or pollute qualitative research by drawing in the
positivist agenda of quantitative research that ultimately denies the contextualised
subjectivity of the researcher conducting thematic analysis. In contrast ST-TA adopts the
7
importance of using processes for establishing agreement, including using a simple
quantitative benchmark for determining adequate agreement.
This process of agreement checking has often been referred to previously as checking
inter-rater reliability (Boyatzis, 1998), but this term is problematic for several reasons: firstly,
thematic analysts do not rate data, and secondly reliability is a term that comes laden with
meanings from psychometrics and classical test theory. To call the process of seeking
agreement in a qualitative analysis as a reliability process conflates it with the very different
process of ensuring that psychometric questionnaires and tests give the same response over
time and across item sets. The term that I use instead for ST-TA is inter-analyst agreement.
Boyatzis’ method does incorporate a similar process. He is of the view, as am I, that
undertaking a process of reaching a high level of coding agreement between two or more
researchers means that the eventual description and labelling of themes is more likely to be
based on a consensual and transparent understanding of the subject matter (Hill et al., 1997).
Braun and Clarke (2019) have incorrectly labelled Boyatzis’ injunction to calculate an
agreement metric as quasi-positivist, but this is based on the assumption that quantification is
itself positivist. I discuss below why that assumption is faulty. Boyatzis does not actually
construe the process of reaching agreement as a means of determining objective fact, but
rather as one of bringing about a working consensus that is essential when (a) conducting
research as a team, and/or (b) when research is to be replicated or extended in new directions
by different researchers in the future. Using a constructionist or interpretivist framework,
which is common in thematic analysis, does not mean giving up on reaching agreement with
others, but instead involves interpreting agreement across analysts as the reaching of inter-
subjective consensus within an agreed interpretive or discursive framework, rather than
discovery of an objective ‘fact’. Analysis does not end when agreement is reached, for new
questions may arise in the process of reaching a consensus that lead to new avenues of
enquiry (Ballesteros & Mata-Benito, 2018).
Another area where ST-TA entails an overlap between qualitative and quantitative
processes is in the calculation of theme frequencies. Theme frequencies refer to the number
or proportion of participants who have text allocated to a particular theme. Brief texts allow
for larger samples than depth approaches, and larger samples provide for more meaningful
statements of a theme’s potential prevalence within a target population than smaller samples
do. Neither Boyatzis nor Braun and Clarke provide protocols for calculating such
frequencies, hence the process within ST-TA is a clear point of difference with existing TA
methods. Frequencies in qualitative reports convey some information on the salience and
8
importance of a theme to the study’s message, but do not by themselves convey the centrality
or salience of themes. The relationship between aims, research questions and themes is also
key in discerning theme salience during an analysis (Braun & Clarke, 2016). With that caveat
in mind, theme frequencies do provide important information. If they are misrepresented or
not included, it can lead to major issues of interpretation for the reader. For example, in
qualitative research that evaluates the experience of an intervention across multiple
participants, the proportion of participants who refer to the intervention as leading to positive
rather than negative experiences is essential information for the reader. Another example is in
the growing domain of qualitative research into the experience of psychedelic drugs. It is
important for the transparency of such studies to convey what proportion of participants
reported enlightening or distressing subjective experiences (Davis et al., 2020). Frequencies
provide that information to the reader, and it is up to the reader or future researcher how to
use that information in conjunction with other information provided.
ST-TA is well suited to mixed-methods research. For example, it can be used to
analyse open-ended questions within a survey that can, in turn, be linked to quantitative data
gained within the same survey. A common criticism of mixed-methods research is that
qualitative and quantitative methods are philosophically incompatible, given that the former
is interpretivist and the latter is positivist, and that these paradigms have discrepant
epistemological assumptions (Wiggins, 2011). I present an argument against this contention,
based principally on the point that quantitative research is based on a plural combination of
epistemologies, of which positivism is at best a minority player, in Appendix A.
As well as overlapping with existing forms of thematic analysis, ST-TA finds itself in
methodological proximity to other methods devised to work with certain kinds of brief texts.
As another established method for working with brief data, Bamberg and
Georgakopoulou (2008) have devised a form of narrative analysis for working with short
stories. Short stories are brief written accounts of events or happenings in a person’s life.
These have become the standard currency of many social media platforms that are based on
brief autobiographical reflections and comments from others (Georgakopoulou, 2017). The
form of analysis that Bamberg and Georgakopoulou have devised to analyse short stories
focuses specifically on identity construction in short stories and takes the form of five steps:
1. Who are the characters and how are they relationally positioned?
2. The interactive accomplishment of ‘narrating’
3. How is the speaker positioned within the interactive flow of turns that constitute
the situation as ‘research’
9
4. How are relationships between all characters managed?
5. How is the self portrayed in this brief story telling?
Bamberg and Georgakopoulou’s approach to short story narrative analysis is an
exemplar of taking an existing broad approach to analysis and then making it bespoke to the
challenges of working with brief texts. Its approach is anchored specifically in the tradition of
narrative analysis developed by Labov (1997), and also a model of identity positioning that
involves the analysis of the self as presented in relation to other characters (Bamberg, 1997).
It differs from ST-TA insofar as the former requires brief autobiographical reflections as its
data, while the latter can be used with any kind of brief text, including, for example, answers
to open-ended questions in surveys (that may not have any self-reference, characters or
story).
Although not specifically devised for brief data, content analysis has been used
extensively for analysing brief qualitative data. To give one recent example, Davis et al.
(2020) conducted a qualitative content analysis of 2,561 brief written descriptions of
memorable experiences of taking the psychedelic DMT. The methodological processes of
such content analysis studies show notable similarities with ST-TA, however the outcome of
this kind of content analysis is a list of codes and frequencies with little by way of theme
description, example quotes and discussion of patterns found. In contrast, ST-TA places a
strong emphasis on conveying the meaning and context of qualitative themes, with verbatim
examples taken from the data to support and illustrate any general concepts conveyed. It’s
embracing of some quantification is done in addition to this fundamental qualitative process,
rather than instead of it.
The process of conducting ST-TA on brief texts
Structured tabular thematic analysis (ST-TA) is conducted in spreadsheet software
such as Excel and is designed to meet the challenges and opportunities of working with brief
texts. It requires no specialist qualitative analysis programmes so is accessible to all
researchers, no matter their budget or technical knowledge.
At a procedural level, structured tabular TA follows a hybridized process approach
that incorporates elements of Braun and Clarke’s TA process (2006) and Boyatzis’s TA
phases (1998). Below I describe each phase in turn and whether it applies to inductive
research, deductive research or hybrid inductive-deductive designs. Table 1 summarizes the
phases for inductive, deductive and hybrid options. To illustrate some points, I use data,
10
tables and a figure based on a study on how perceptions of parenting relate to authenticity in
young adults (Ayoola & Robinson, 2017).
INSERT TABLE 1 HERE
Phase A: A-priori theme development (Deductive and Hybrid only)
Thematic analysis that is purely deductive in approach does not require the generation of new
codes and themes. It commences with a set of themes prior to data collection and analysis,
taken directly from a previous study in the topic area and then seeks to apply those to a new
sample. Research objectives suitable for a deductive approach include: (a) replicating an
existing thematic analysis study or (b) developing, extending or testing an existing thematic
framework or theory. In order to develop a set of themes for a deductive study, one can take
either a broadly theory-based approach, in which themes are inferred from a theory, or a
prior-research-based approach, in which themes are taken from the findings of an existing
thematic analysis study (Boyatzis, 1998).
Deductive and inductive approaches can be combined in hybrid designs (Robinson &
Smith, 2010). A hybrid approach is appropriate where there is (a) substantial qualitative
literature on the topic of study to draw on, meaning a purely inductive approach would
potentially omit existing insights and knowledge, but also (b) a clear sense that existing
knowledge is partial, and hence there is a need for continued development of thematic
frameworks and theory.
In such a hybrid approach, the analyst will firstly deploy an initial set of themes or
concepts from existing work to orientate the analysis process. These provide a starting point
as orientating constructs. The process of generating codes and themes is then worked through
with this opening set of constructs or themes in mind, and these are modified or added to
depending on whether the data fits the scheme or not. For example, in a study on admiration
in young adults, myself and colleagues used this hybrid approach to organize our analysis of
brief written descriptions of an admired individual provided by young adults from three
cultures (Robinson et al., 2015). We employed a thematic framework from an existing
qualitative study (Schlenker et al. 2008), and then refined this set of themes as we analysed
the data. So, the final set of themes only partially drew on the initial themes.
In summary, if you are intending to conduct a deductive or hybrid analysis, you will
need to select a set of constructs or themes from existing literature and provide a robust
rationale for why you have done so.
11
Phase B: Deep immersion in the data (Deductive, Inductive and Hybrid)
For this phase, you will need to transcribe or import your data into Excel in such a way that
each brief text occupies one cell in a column, as illustrated in Table 2. You will also need to
include a column with an anonymized participant identifier or number, and columns with
demographic details. Two key injunctions that Braun and Clarke emphasize in their
methodology, which are also essential to this first phase of the tabular approach, are (a)
repeated reading of the data, and (b) taking initial notes for codes. To facilitate repeated
reading of the data in Excel, make sure to select the ‘Wrap Text’ option (right click > Format
Cells > Alignment > Wrap Text). This ensures that all text is shown in each cell. To facilitate
taking notes, next to the column of data, create a column labelled initial notes. See Table 2
for an illustration of the layout of the spreadsheet. Carefully and slowly read each data
segment, adding notes for possible codes, or other initial analytical ideas, as you go. If you
have started with an a-priori theme set, you might make notes on any cases that you think do
not fit the scheme. For this task, you can either do this on screen or print the spreadsheet out,
depending on your preference. Follow this process of immersive reading of the entire dataset
at least twice, until you feel a strong familiarity with all the data and start to get an early
sense of any patterns therein.
INSERT TABLE 2 HERE
Phase C – Generating initial codes and themes (Inductive and Hybrid Only)
After the initial process of familiarising yourself with the data, you can move onto the
development of codes, if you are using an inductive or hybrid design. In a hybrid design, a
priori themes will tentatively inform theme generation.
For the process of generating initial codes, add an additional column to your
spreadsheet and add the title in the top row of ‘Initial Codes’, as shown in Table 2. Based on
your immersive reading and initial notes, add in names of codes into this new column. Enter
terms or words that you think, based on your repeated reading, subsume or describe content
in multiple data segments or texts. By so doing, you are taking the first step towards finding
common patterns, words or ideas, which is always your ultimate goal in a thematic analysis.
Once the process of code development is complete, you will have at least one code
entered in every row. For the next step, copy and paste the full column of code words into
another worksheet in the same Excel file (NB. click the button at the bottom to do this).
12
On this new sheet, use the copy and paste function to move codes around and group them into
clusters. You can use different columns for different clusters to aid visualising the process.
Each cluster of codes is a prospective theme. Then, you need to name your clustered codes
using phrases or terms that are clearly anchored in the data and are as idiosyncratic to your
study as possible (that can sometimes mean using a longer, rather than shorter, theme name).
A common error is to name themes with terms that are so short or generic that they have no
clear relationship to your specific study or research question. For initial codes and subthemes,
using the words and phrases that participants use is key to ensuring that the process of
thematic abstraction is grounded in the language of participants. For higher order themes, it is
avoidance of the ambiguity that comes with excessive concision that is key to idiosyncrasy.
An example is shown in Table 3; the theme name “Perceived negative effect of parenting on
authenticity”, at seven words long, is longer than most themes that one sees in most thematic
analysis studies. However, by using a phrase like this for the theme name, the meaning is far
less ambiguous than if one were to attempt to reduce it to one or two words.
You can continue to move codes between clusters, combine clusters, and re-name
themes until you have a framework that you are satisfied will allow all, or nearly all, of your
brief texts to be linked to at least one theme. A popular way of creating an additional layer of
order in your themes is to have two levels of theme: main theme and sub-theme. Main themes
are more abstract and therefore include more semantic content than sub-themes, hence
provide an additional quality of analytical parsimony, should that be desired. Whether or not
two levels of theme are appropriate to your study depends on the research questions you pose,
and whether a more abstract level of thematising helps to convey clear and coherent answers
to your questions.
The structured tabular approach to thematic analysis is open to searching for semantic
or latent themes (Braun & Clarke, 2006). Semantic themes are manifest in the surface
meanings of the data; they are descriptive and minimize inference from the textual content.
Latent themes require further interpretation, as they are not manifest in the data, but are
implicit beyond or below the surface content.
Phase D – Tabulating themes against data segments (deductive, inductive and hybrid)
Phase D involves attaching data segments to themes in a tabulated form, an example of which
is shown in Table 3. This provides a foundation for the agreement-checking and frequency
calculation processes outlined in Phases E and F. The practical process of Phase D is as
follows:
13
1. If you are working inductively, open a new worksheet in your Excel file and copy a
duplicated version of your Phase B spreadsheet, including participant number,
demographic data and qualitative text data in the left-hand columns. Delete the notes
and themes columns (make sure to keep the original worksheet with those notes and
themes on file).
2. Insert a row at the top of the new spreadsheet. Write your theme names across the top
row, starting with the column to the right of your qualitative text column. If you have
just one level of theme, then one row at the top will suffice. If your themes are
differentiated into main themes and sub-themes, insert two rows at the top and put the
main themes across row 1, and the sub-themes across row 2. For main themes, merge
the cells across the columns that main themes refer to, as shown in the example in
Table 3. Keep theme columns narrow, so that you can fit many on the screen at once –
this helps the process of analytically allocating texts to themes.
3. Select the top row or top two rows (depending on whether you have one or two levels
of theme), then go to View > Freeze Panes > Freeze Panes (based on current
selection). This will mean that your theme names remain visible as you scroll
downwards.
4. Once you are sure that you have your final set of themes, go down through each brief
text and wherever a sub-theme is represented in the data, add a 1 in the relevant
column. Do this until all have been allocated to sub-themes. You can attach each text
to multiple themes if appropriate. Table 3 shows an example in which texts from 5
participants have been allocated to three themes, extracted from the authenticity and
parenting study by Ayoola and Robinson (2017).
INSERT TABLE 3 HERE
This process of tabulation allows the relationship between data and themes to be visually
related in new ways, so may lead to continued theme development. If themes are further
developed at this point, make sure to keep a dated log of all changes. This helps your
analytical process to be fully transparent to others. One option for keeping a log of thematic
developments is by creating an additional worksheet in your Excel file and using it a log. In
this way, it will also be found in the same place as your analysis.
14
Phase E: Checking inter-analyst agreement
Phase E involves the process of checking the level of thematising agreement between
yourself and another analyst. One way of reaching agreement is through an informal
discussion-based approach where the two researchers discuss the themes they have attached
to the brief texts and resolve differences and debates in order to end up with a more
consistent, coherent and clear set of themes. A more structured set of processes for checking
agreement across analysts is as follows:
1. Two individuals are provided with a blank version of the data tabulation spreadsheet with
no. 1s entered. Both individuals should ideally be familiar with the theme names and
codes developed or employed for the study.
2. The two analysts should allocate texts to themes independently of each other.
a. If the dataset is large, an option is to select a subset of participants for this
agreement check (20-30 is an appropriate number).
3. Having both done that, one of the analysts combines the two spreadsheets into one for
checking, by inserting the theme columns from one into the other.
4. For each row, the analysts must then calculate the number of agreements (where both
analysts have a ‘1’ in the same cell), and the number of disagreements (one analyst has a
‘1’ in the cell, but the other does not).
5. The total number of disagreements and agreements should be calculated across all cases.
A percentage level of agreement is calculated as follows:
Total no. of agreements
________________________________
Total no. of agreements + disagreements
The aim of this process is to end up with a level of agreement that supports the
proposition that the analytical scheme and process is transparent, rigorous, coherent and
trustworthy (Nowell, Norris, White & Moules, 2017; Yardley, 2000). If a thematic scheme is
clear and coherent, and themes are described with rigour and transparency, analysts should
have little problem agreeing on which texts are allocated to which theme. Conversely, a weak
analysis, in the words of Braun and Clarke, is where “the themes do not appear to work,
where there is too much overlap between themes, or where the themes are not internally
coherent and consistent” (2006, p.94). If themes are vague, poorly defined, or poorly labelled,
two analysts will find it difficult to tabulate themes against brief texts, and this will be shown
up in this process.
x 100
15
An appropriate rule-of-thumb to aim for, originally put forward by Miles and
Huberman (1994) based on extensive trialling of inter-analyst checking, is 80% agreement. If
this level is not achieved, the two analysts can convene and discuss their disagreements and
consider ways of adapting theme names or theme descriptions to come to a higher level of
agreement. This second stage of reaching consensus need not be done blind, but rather should
be done as a discursive process of continued theme development between the two researchers
until a consensus position is achieved. This may of course lead to theme re-development, in
which case the process cycles back to Phase C.
Phase F – Exploring Theme Frequencies
The use of structured tabular thematic analysis provides for a higher degree of
precision with which statements of a theme’s prevalence across the sample can be made.
Having such prevalence data increases the trustworthiness and transparency of the findings,
in line with other injunctions for trustworthiness in thematic analysis (Nowell et al., 2017). It
is, however, important to emphasize again that frequency of a theme does not equate on its
own to how relevant or salient a theme is within a study (Braun & Clarke, 2016).
To calculate the frequency of participants allocated to each theme, add a frequency
calculation cell at the bottom of each column, as illustrated in Table 3. To calculate this
automatically using an Excel formula, write =SUM( ) in the cell, with the brackets containing
the top and bottom cell code, separated by a colon. So, for example, if a theme is shown in
column D and there are 40 participants, the first of which is in row 2 (because themes occupy
row 1), the formula would be = SUM(D2:D41). The resulting frequency data is primarily to
provide accurate statements about the prevalence of themes when writing up the report in
Phase 6. You can also choose to explore frequencies by comparing them across key
demographic groups, for example, comparing males and females, if that is considered
appropriate to the research question.
Frequency data present the opportunity of further quantitative analysis beyond total-
sample frequencies. For example, if a researcher had data from males and females and was
interested in gender differences in terms of theme prevalence, they could transfer the
spreadsheet into a statistics package, insert 0 for all the instances where a theme has not been
coded, enter gender in a column as a nominal variable, and run a frequency-based test such as
Chi Square to test the difference. This process fits within a form of mixed-methods research
design referred to as the data transformation model (Creswell & Plan Clark, 2010).
16
Phase G: Developing thematic maps and diagrams
Braun and Clarke (2006) emphasize the benefits of thematic maps and diagrams to thematic
analysis. These can aid analysis by presenting a visual representation of relations among
themes that stimulate an integration of themes into a model or a conceptual framework
(Robinson, 2011). Maps and diagrams are also integral to the structured tabular approach to
TA, both as a way of helping to develop and relate themes, and as a way of presenting
analytical patterns concisely and coherently. See Figure 1 for an example of a diagram
developed from the Ayoola and Robinson (2017) study on authenticity and parenting in
childhood.
INSERT FIGURE 1 HERE
Creating diagrams and maps involves examining relationships between themes and then
using the arrows or lines in the diagram to represent those relationships. Through this
process, a list of themes moves towards becoming a model, framework or integrated scheme.
It is recommended that once a list of themes has been provisionally developed, they can be
written on post-it notes or small pieces of paper and combined in patterns with potential
relationships also written onto post-it notes and placed between themes. This process may
lead to further insights in theme development, as when a ‘whole’ is achieved by way of
creating a theoretical frame, it can inform the nature and labelling of the themes to some
degree. Thus, there may be a recursive process between Phase G back to Phase C.
To support this process of relating themes to achieve integration, a ‘bolt on’ method
called Relational Analysis (Robinson, 2011) can be used. Relational analysis presents ten
kinds of ways that themes can relate: descriptive, comparative, semiotic, evocative,
contingency, causal, reciprocal, dialectical, conceptual part-whole and contextual part-
whole. These relational forms can be explored as candidates for making sense of how themes
relate. Researchers can undertake this process of exploring relationships in dialogue or
individually. The outcome of exploring inter-theme relationships feeds directly into the
process of creating a map or a diagram, as lines or arrows in maps visually indicate such
relationships.
Phase H - Producing the report
In any thematic analysis study, writing the report is an active part of the analytical
process, and this holds true of the structured tabular approach. The nature and structure of the
17
report depends on whether a tabular thematic analysis is used as (1) a stand-alone analysis,
(2) alongside in-depth qualitative methods, or (3) with quantitative methods. If brief texts are
the sole form of data, the report will contain a singular results section that presents the themes
using the typical structure of a qualitative results section. If forms of in-depth qualitative data
have been collected concurrently as part of the study, it is recommended that the two are
presented in two subsequent results sections, with an integrative discussion to systematically
compare the brevity-and-breadth findings of the tabular approach with the length-and-depth
findings of the other method.
Another option for a report including a structured tabular thematic analysis is a
mixed-methods paper that combines qualitative and quantitative findings. As mentioned
earlier, a popular option in mixed-methods research is to concurrently gather numerical and
brief textual data about a specific phenomenon by way of an online data collection tool, then
integrating these forms of data to inform findings. For example, the Ayoola and Robinson
(2017) study from which the data extracts in Table 2 and Table 3 are taken, included (a) brief
texts on how parenting during childhood is perceived to influence adult authenticity, as well
as (b) psychometric data on trait authenticity and retrospective ratings of parental care and/or
neglect during childhood. The qualitative and quantitative analyses were discussed in the
report and interpreted in combination.
Sampling concerns
A pertinent issue that relates to ST-TA is the matter of sampling. Qualitative methods
that have traditionally been associated with depth data have been associated with purposive
sampling (e.g. Lincoln & Guba, 1985). Purposive sampling involves the intentional selection
of specific kinds of participant from the target sample to ensure variability of the sample
along key parameters that may differ in their responses (e.g. ensuring a balance of males and
females or young and old). It is thus designed to elicit a sample that represents a broader
population when the N is low (Robinson, 2014b).
Brief text research gains its richness through the diversity of responses, rather than the
depth of responses. Therefore, when ST-TA is used, the sample N will often be larger than
in-depth qualitative studies. Thus, it can and should employ a different sampling approach to
the purposive strategies of small-N interview studies. Random sampling is premised on the
logic that the larger the number of participants in a sample, the more likely they are to be
representative of a target population. Thus, samples of hundreds or thousands may well show
representative parameters in ways that samples of 10 or 20 will not. However, a range of
18
factors mitigate against random sampling even with large samples. The voluntary nature of
psychological research studies means that people who are interested take part. These
psychologically curious individuals may well not be representative of the population. Another
issues is that if recruitment processes are locally situated, for example via a university or via
recruitment posters, they may end up with a convenience sample, limited by geography;
social connections to the researcher; socio-economic background, or a whole range of other
factors, which may mean the sample is not truly random. Online recruitment agencies may
have a greater geographical access, but their participants are those who have signed up for
getting micro-payments through research participation. Such individuals are unlikely to be a
random sample.
One solution to this is to combine random sampling with purposive sampling
(Robinson, 2014b). For example, if it is considered important to have an equal distribution of
males and females in a sample, and also to have an equal distribution of younger adults and
older adults, a researcher can purposively select to have 30-40 young adult males, 30-40
young adult females, 30-40 older adult males and 30-40 older adult females in a sample, but
then randomly sample within each of these cells to reach that target. In sum, a problem-
focused and flexible approach to sampling, which can incorporate purposive and random
sampling or combinations of the two, is appropriate to accompany ST-TA.
Conclusion
I have presented the structured tabular approach to thematic analysis as a way of
flexibly and rigorously analysing brief texts. Such an approach is of growing importance
given both the increasing availability of such data via social media along with the rising
popularity of open-ended survey response or short story elicitation methods (Clarke et al.
2019; Terry & Braun, 2017). The approach synthesizes injunctions from two approaches to
thematic analysis and adds in a range of processes for working with brief texts, including the
practical advantages of using a spreadsheet when dealing with a larger sample and a tabulated
form of analysis that provides opportunities for frequency and agreement calculation. It
requires no specialist analysis software, thus is widely accessible and user-friendly for
researchers at any level. The protocols and processes I have described above are flexible
guidelines, and I encourage the reader to adapt them to their needs and to innovate further as
and when appropriate. Brief texts remain an important frontier for qualitative psychology and
I hope this method will act as encouragement for researchers to explore the full potential of
this type of data.
19
References
Ballesteros, B. & Mata-Benito, P. (2018). The inner process of collective interpretation in
qualitative research. The Qualitative Report, 23 (1), 168-183.
Bamberg, M. & Georgakopoulou, A. (2008). Small stories as a new perspective in narrative
and identity analysis. Text and Talk, 28, 377-396. https://doi.org/10.1515/TEXT
Boyatzis, R.E. (1998). Transforming Qualitative Information: Thematic Analysis and Code
Development. Sage Publishing.
Braun, V., & Clarke, V. (2006). Using thematic analysis in psychology. Qualitative Research
in Psychology, 3, 77-101. http://dx.doi.org/10.1191/1478088706qp063oa
Braun, V. & Clarke, V. (2016). (Mis)conceptualising themes, thematic analysis, and other
problems with Fugard and Potts’ (2015) sample-size tool for thematic analysis.
International Journal of Social Research Methodology, 19, 739-743.
http://dx.doi.org/10.1080/13645579.2016.1195588
Braun, V. & Clarke, V. (2019). Reflecting on reflexive thematic analysis. Qualitative
Research in Sport, Exercise and Health, 11:4, 589-597.
https://doi.org/10.1080/2159676X.2019.1628806
Braun, V., Clarke, V., Hayfield, N., & Terry, G. (2018). Thematic Analysis. In P.
Liamputtong (Ed.), Handbook of Research Methods in Health Social Sciences (pp. 1-
18). Singapore: Springer Singapore. https://doi.org/10.1007/978-981-10-2779-6_103
Carpentier, N. (2014). Fuck the clowns from Grease!!’ Fantasies of participation and agency
in the YouTube comments on a Cypriot Problem documentary. Information,
Communication & Society, 17, 8. https://doi.org/10.1080/1369118X.2013.875582
Clarke, V. Braun, V., Frith, H. & Moller, N. (2019). Editorial introduction to the special
issue: Using story completion methods in qualitative research. Qualitative Research in
Psychology, 16. https://doi.org/10.1080/14780887.2018.1536378
Clarke, V., Hayfield, N., Moller, N., Tischner, I. & the story completion research group.
(2017). Once upon a time…Qualitative story completion methods. In V.Braun,
V.Clarke and D.Gray (Eds.) Collecting Qualitative Data: A practical guide to textual,
media and virtual techniques, pp.45-70. Cambridge University Press.
Creswell, J.W. & Plano Clark, V.L. (2010). Designing and Conducting Mixed Methods
Research. Sage Publications.
Davis, A. K., Clifton, J. M., Weaver, E. G., Hurwitz, E. S., Johnson, M. W., & Griffiths, R.
R. (2020). Survey of entity encounter experiences occasioned by inhaled N,N-
20
dimethyltryptamine: Phenomenology, interpretation, and enduring effects. Journal of
Psychopharmacology. https://doi.org/10.1177/0269881120916143
Deighton-Smith, N., & Bell, B.T. (2018). Objectifying fitness: A content and thematic
analysis of #fitspiration images on social media. Psychology of Popular Media
Culture, 7(4), 467-483. DOI: https://doi.org/10.1037/ppm0000143
Erikson, E. (1969). Ghandi's Truth. W. W. Norton.
Erikson, E. (1958). Young man Luther: A study in psychoanalysis and history. New
York: W. W. Norton.
Festinger, L., Riecken, H.W., & Schacter, S. (1956). When prophecy fails: A social
and psychological study of a modern group that predicted the end of the world.
University of Minnesota Press.
Flanagan, J. C. (1954). The critical incident technique. Psychological Bulletin, 51, 327–358.
https://doi.org/10.1037/h0061470
Frost, N., Nolas, S.M., Brooks-Gordon, B., Esin, C., Holt, A., Mehdizadeh, L., Shinebourne,
P. (2010). Pluralism in qualitative research: the impact of different researchers and
qualitative approaches on the analysis of qualitative data. Qualitative Research, 10,
441-460. https://doi.org/10.1177/1468794110366802
Georgakopoulou, A. (2014). Small stories transposition and social media: A micro-
perspective on the ‘Greek crisis’. Discourse and Society, 25, 519-539.
https://doi.org/10.1177/0957926514536963
Georgakopoulou, A. (2017). Sharing the moment as small stories: The interplay between
practices & affordances in the social media-curation of lives. Narrative Inquiry, 27,
311-333. https://doi.org/10.1075/ni.27.2.06geo
Giles, D.C. (2017). How do fan and celebrity identities become established on Twitter? A
study of ‘social media natives’ and their followers. Celebrity Studies, 8, 445-460.
https://doi.org/10.1080/19392397.2017.1305911
Giles, D.C. (2016). Observing real-world groups in the virtual field: The analysis of online
discussion. British Journal of Social Psychology, 55(3), 484-498.
https://doi.org/10.1111/bjso.12139
Giles, D.C., Stommel, W., Paulus, T., Lester, J., & Reed, D. (2015). The microanalysis of
online data: the methodological development of ‘digital CA’. Discourse, Context and
Media, 7, 45-51.
Giles, D.C. (2014). “DSM-V is taking away our identity”: the reaction of the online
21
community to the proposed changes in the diagnosis of Asperger’s Disorder. Health
18(2), 179-195. https://doi.org/10.1177/1363459313488006.
Glaser, B.G., & Strauss, A.L. (1967). The Discovery of Grounded Theory. Strategies for
qualitative research. London: Aldine Transaction.
Hill, C. E., Thompson, B. J., & Williams, E. N. (1997). A guide to conducting consensual
qualitative research. The Counseling Psychologist, 25, 517-572.
Howitt, D. (2016). Introduction to Qualitative Research Methods in Psychology, 3rd Edition.
Pearson.
Labov, W. (1997). Some further steps in narrative analysis. Journal of Narrative and Life
History, 7, 395-415.
Lincoln Y.S., Guba E.G. (1985). Naturalistic Inquiry. Sage Publications.
Lyles, C.R., López, A., Pasick, R., Sarkar, U. (2013). “5 Mins of Uncomfyness Is Better
than Dealing with Cancer 4 a Lifetime”: an Exploratory Qualitative Analysis of
Cervical and Breast Cancer Screening Dialogue on Twitter. Journal of Cancer
Education, 28, 127–133. https://doi.org/10.1007/s13187-012-0432-2.
James, W. (1907). Pragmatism: A new name for some old ways of thinking.
Hackett Publishing.
Miles, M.B. and Huberman, A.M. (1984). Qualitative Data Analysis. Sage.
Mejova, Y. & Srinivasan, P. (2012). Political speech in social media streams: YouTube
comments and Twitter posts. WebSci '12 Proceedings of the 4th Annual ACM Web
Science Conference, 205-208.
Nowell, L. S., Norris, J. M., White, D. E., & Moules, N. J. (2017). Thematic analysis:
Striving to meet the trustworthiness criteria. International Journal of Qualitative
Methods, 16(1), 1-13. https://doi.org/10.1177/1609406917733847
Ayoola, E.O. & Robinson, O.C. (2017). Perceptions of Parenting and Authenticity
in Emerging Adults: A Mixed-Methods Study. Paper presented as part of symposium
“Mixed methods research on emerging adults: Quarter-life crisis, resilience
and authenticity” at 8th Conference on Emerging Adulthood, Washington DC.
Osbeck, L.M. (2014). Scientific reasoning as sense-making: Implications for qualitative
inquiry. Qualitative Psychology, 1, 34-36. https://doi.org/10.1037/qup0000004
Otto, S., Kröhne, U., & Richter, D. (2018). The dominance of introspective measures and
what this implies: The example of environmental attitude. PloS one, 13(2), e0192907.
https://doi.org/10.1371/journal.pone.0192907
Pavlov, I.P. (1927). Conditioned reflexes: an investigation of the physiological activity
22
of the cerebral cortex. Oxford University Press.
Popper, K. (2002). Conjectures and Refutations: The Growth of Scientific Knowledge (2nd
edition). Routledge.
Reason, P. & Rowan, J. (Eds.) (1981). Human Inquiry: A Sourcebook of New Paradigm
Research. Wiley.
Robinson, O.C. (2011). Relational analysis: an add-on technique for aiding data
integration in qualitative research. Qualitative Research in Psychology, 8, 1977-209.
https://doi.org/10.1080/14780887.2011.572745
Robinson, O.C. (2014a). Re-examining the epistemological foundations of quantitative
psychology: Pluralism, not positivism. Qualitative Methods in Psychology Bulletin,
17, 5-8.
Robinson, O.C. (2014b). Sampling in interview-based qualitative research: A theoretical and
practical guide. Qualitative Research in Psychology, 11, 25-41.
https://doi.org/10.1080/14780887.2013.801543
Robinson, O.C. (2020a). A dialectical approach to understanding the relationship between
science and spirituality: The MODI model. Journal for the Study of Spirituality, 10,
15-28. https://doi.org/10.1080/20440243.2020.1726045
Robinson, O.C (2020b). Development through Adulthood. Macmillan Higher Education.
Robinson, O.C., Dunn, A., Nartova-Bochaver. S., Bochaver, K., Asadi, S., Khosravi, Z.,
Jafari, M., Zhang, X., Yang, Y. (2015). Figures of admiration in emerging adulthood:
A four country study. Emerging Adulthood, 4, 82-91.
https://doi.org/10.1177/2167696815601945
Robinson, O.C. & Smith, J.A. (2010). Investigating the form and dynamics of crisis episodes
in early adulthood: The application of a composite qualitative method. Qualitative
Research in Psychology, 7, 170-191. https://doi.org/10.1080/14780880802699084
Schlenker, B. R.,Weigold,M. F.,&Schlenker, K. A. (2008).What makes a hero? The impact of
integrity on admiration and interpersonal judgment. Journal of Personality, 76, 323–
355. https://doi.org/10.1111/j.1467-6494.2007.00488.x
Seife, C. (2000). Zero: The Biography of a Dangerous Idea. London: Souvenir Press.
Slepian, M. L., & Moulton-Tetlock, E. (2019). Confiding Secrets and Well-Being. Social
Psychological and Personality Science, 10(4), 472–484.
https://doi.org/10.1177/1948550618765069
Terry, G. & Braun, V. (2017). Short but often sweet: The surprising potential of qualitative
survey methods. In V.Braun, V.Clarke and D.Gray (Eds.) Collecting Qualitative
23
Data: A practical guide to textual, media and virtual techniques, pp.45-70.
Cambridge University Press.
Thomas, D. (2006). A general inductive approach for analyzing qualitative evaluation
data. American Journal of Evaluation, 27, 237-246.
https://doi.org/10.1177/1098214005283748
Venditti, S., Piredda, F. & Mattana, W. (2017) Micronarratives as the form of contemporary
communication, The Design Journal, 20:sup1, S273-S282,
https://doi.org/10.1080/14606925.2017.1352804
Vraga, E.K., Thorson, K., Kligler-Vilenchik, N., Gee, E. (2015). How individual sensitivities
to disagreement shape youth political expression on Facebook. Computers in Human
Behavior, 45, 281-289. https://doi.org/10.1016/j.chb.2014.12.025
Zick, A., Granieri, M. & Makoul, G. (2007). First-year medical students’ assessment of their
own communication skills: A video-based, open-ended approach. Patient Education
and Counseling. 68, Issue 2, 161-166. https://doi.org/10.1016/j.pec.2007.05.018
Yardley, L. (2000). Dilemmas in qualitative health research. Psychology and Health, 15,
215-228. https://doi.org/10.1080/08870440008400302
Wiggins, B.J. (2011). Confronting the dilemma of mixed methods. Journal of Theoretical
and Philosophical Psychology, 31, 44–60. https://doi.org/10.1037/a0022612
24
Table 1. Analytical phases for deductive, inductive and hybrid research studies
Deductive
Hybrid
Inductive
Phase A: A-priori theme
development
Phase A: A-priori theme
development
SKIP PHASE A
Phase B: Deep immersion in
the data
Phase B: Deep immersion in
the data
Phase B: Deep immersion in
the data
SKIP PHASE C
Phase C – Developing revised
codes and themes in context
of, and influenced by, a-priori
themes.
Phase C – Generating codes
and themes (as uninfluenced
by existing theory as is
possible)
Phase D – Tabulating themes
against data chunks
Phase D – Tabulating themes
against data chunks
Phase D – Tabulating themes
against data chunks
Phase E: Checking agreement
Phase E: Checking agreement
Phase E: Checking agreement
Phase F – Exploring Theme
Frequencies
Phase F – Exploring Theme
Frequencies
Phase F – Exploring Theme
Frequencies
Phase G: Thematic maps
Phase G: Thematic maps
Phase G: Thematic maps
Phase H - Producing the report
Phase H - Producing the report
Phase H - Producing the report
25
Table 2: Spreadsheet format for Phase 1 with illustrative textual data from parenting –
authenticity study
No.
Gender
Qualitative data segment
Initial notes
Initial codes
1
Male
Yeah. I think so. My parents were honest with
me and about themselves and I think it fostered
that in me, too... So, I try to stay true to myself
as much as I can.
9
Female
My mum made it very easy to be whoever I
wanted to be and I saw how she accepted all my
friends growing up in spite of anything that
could make them different/stand out. She took
an interest in me and who I was and so I had a
strong sense of self from an early age.
27
Female
I believe that it helped a lot. My mum always
encouraged me to be myself and it was fun to
sometimes shock my dad with who I am. So, I
have learnt to know myself and to be myself.
33
Male
The love my parents have for me show that I
don't need to pretend to be someone else as they
love me just the way I am.
NB. Cases selected for this table represent main theme of perceived positive effects of
parenting on authenticity
26
Table 3: Spreadsheet format for Phase 4 – illustrative five cases, one main theme with three
subthemes shown
Main theme: Perceived negative effect of
parenting on authenticity
No.
Gender
Qualitative data segment
Subtheme 1 –
Cultural /
generational
disconnect
Subtheme 2
Parents as
negative
role models
Subtheme 3
Criticism or
disapproval
of
characteristics
10
Male
I feel I am true to myself, but there are some
parts of who I am I feel I have dismissed or
choose to hide from my parents as I feel that
they would disapprove or not fit the image
that they have of me.
1
15
Female
I think they gave me a foundation. However,
I've come to being my own adult sometimes
in disagreement with my parents. I think it's
because they were born and raised in Africa
and I in London.
1
35
Female
Being criticised for my personality by my
family has caused me to feel insecure as an
adult. If I was ever feeling upset about
something that my parents didn't believe to
be a big deal, they would brush it off,
leaving me to feel like I was too sensitive.
1
41
Male
My parents are very particular people and so
the parts of myself that do not match their
picture of me have to be hidden. I try to be
as authentic as possible but it is not always
possible, but only in some aspects of life.
1
48
Female
Seeing how much my father neglected his
own emotions and needs completely, I feel
obligated not to make the same mistakes and
live a life being as authentic as possible but
find it difficult as I feel the impression my
father gave has stuck with me and is
difficult to counterbalance.
1
SUBTHEME FREQUENCY
1
1
3
NB. Cases selected for this table represent main theme of perceived negative effects of
parenting on authenticity
27
Figure 1. A map of themes developed in a study of how parents are perceived to influence
adult authenticity in young adults
Criticism or disapproval
of characteristics
Perceived lack of
importance to parents
Parents as negative role
models
Cultural/generational
disconnect
Self-concealment,
insecurity,
avoidance of self-
disclosure
Support for choices
and decisions made
Parents positive role
models; honest & open
Acceptance of
personality & friends
Unconditional love
and care
Confidence to
self-assert or
self-disclose,
sense of security
Authenticity as a
young adult
undermine
facilitate
Perceived negative effect of
parenting on authenticity
Perceived positive effect of
parenting on authenticity
28
Appendix A. The epistemologies underlying quantitative research: A complex picture
the assertion that quantitative research is positivist is discrepant with historical facts.
The history of psychology clearly shows that quantitative research is based on a plurality of
epistemologies, with positivism being a minority player at best. The first of these paradigms
is Popper’s hypothetico-deductive approach to science (Popper, 2002). Popper was explicitly
critical of positivism; whilst positivism conceives of science as eliciting solid facts and
objective truths, Popper’s approach sees science as eliciting tentative and provisional
hypotheses that are never actually true but can be only said to be not yet proved false. The
second influential paradigm in quantitative methods is the pragmatism of William James
(1907). James supported the use of qualitative and quantitative data. He based this on the
reasoning that all research should primarily be directed towards some productive end, and
thus have an instrumental benefit. We should use whatever kind of empirical information can
help solve that problem, and not determine a priori if that evidence should be verbal or
numerical. A third paradigmatic foundation is the introspectionism of Wundt and his
followers, which formatively influenced the development of psychometrics (Otto, 2018). This
paradigm provides a justification for self-observation and hence for self-report
questionnaires, and this in turn supports the edifice of quantitative psychometric methods.
Self-report questionnaires are not only reliant on the validity of self-observation and
introspection, they also require substantial interpretation on the part of the participant. The
individual completing a questionnaire must read a series of written statements or questions
and then judge which number on the scale accords best to their character or experience in
relation to the statements. This process is clearly a deeply subjective and hermeneutic one,
albeit one that is not frequently recognised as such (Robinson, 2014).
While positivism has had little influence on psychology, where it has had some
influence is in sociology, and crucially, positivists in sociology dismiss attempts at self-
observation or self-report (Comte, 1842). This in turn means the rejection of the countless
quantitative studies based on self-report, which are the foundation of much of neuroscience as
well as psychology.
In sum, there is no neat allegiance between quantitative methods and positivism. Such
an assertion appears to be an over-simplistic and distorting reinterpretation of history. The
binary distinction of ‘qualitative-quantitative’ hides a raft of commonalities and complexities.
Rather than two islands with their own separate methodological ethos, qualitative and
quantitative research are more like two intersecting paths through the same forest of
29
evidence-based sense-making. Hence, I contend that a method such as ST-TA, which
intentionally combines qualitative research with limited quantification, is epistemologically
justifiable and coherent.
A preview of this full-text is provided by American Psychological Association.
Content available from Qualitative Psychology
This content is subject to copyright. Terms and conditions apply.
