Professional Documents
Culture Documents
2017 - Clarke Et Al - Tell Me A Story. Story Completion Methods
2017 - Clarke Et Al - Tell Me A Story. Story Completion Methods
c 2017 Not known
Copyright and Moral Rights for the articles on this site are retained by the individual authors and/or other copyright
owners. For more information on Open Research Online’s data policy on reuse of materials please consult the policies
page.
oro.open.ac.uk
Chapter 3
Victoria Clarke, Nikki Hayfield, Naomi Moller, Irmgard Tischner and the Story Completion
Research Group
Overview
This chapter introduces the story completion (SC) method of collecting qualitative data, a
novel technique that offers intriguing potential to the qualitative researcher. Since the
method is new to qualitative research, it has fewer published research studies than some of
the other methods covered in this book. For this reason, the chapter aims not only to provide
a description of the method and recommendations for how best to use it, but also to explore
some of the unresolved theoretical and practical questions about SC. These questions have
been identified by the chapter authors, who comprise the Story Completion Research Group.
We are a group of researchers who have come together to share our experience of using
and further developing the method (see Box 3.1). Our view is that SC has the potential to
‘reach the parts that other methods cannot reach’ (Pope & Mays, 1995); it therefore has
advantages over and above being enticingly resource-lite in terms of data collection,
1
SC originally developed as a form of projective test, for use by psychiatrists and clinical
psychopathology of clients (see Rabin, 1981). Projective tests involve asking people to
respond to ambiguous stimuli – such as inkblots, as in the famous Rorschach inkblot test
(Rorschach, Lemkau and Kronenberg, 1921/1998). The assumption is that because the
respondent cannot know unequivocally what the stimulus ‘is’, they have to draw on their own
understandings (personality, needs, life experiences) to make sense of it, and ‘fill in the
blanks’. In doing so – as the theory of projective tests goes – the participant reveals things
about themselves that they may not be conscious of, or would feel uncomfortable revealing if
asked directly about. Projective tests are rooted in psychoanalytic theory (Rabin, 2001),
which assumes that large portions of the self are blocked off to consciousness, and thus
unavailable to both clients and clinicians through conventional means such as self-report.
The psychodynamically-informed promise of projective tests taps into this ‘blocked off’
information, providing what Murray (1943/1971: 1) compares to ‘an x-ray picture of [the]
inner self’.
The key projective method of interest for the current chapter is the Thematic Apperception
Test (TAT), the most famous – but not the first (Morgan, 2002) – projective test based on SC
(Murray, 1943/1971). The TAT involves showing a client a series of evocative but ambiguous
images and asking them to ‘to make up a story’ for each picture presented. Although there
are scoring methods available, the typical approach to the TAT in therapeutic settings is for
the administrator to use their clinical judgement to interpret what the stories reveal about
their clients.
2
Projective tests are used predominantly in clinical settings to provide insight into individual
clients, rather than as an empirical method for research data collection. In other settings,
however, projectives have also been used as a research method – for example, in consumer
and business research (e.g., Donoghue, 2000; Soley and Smith, 2008) and developmental
psychology (e.g., Bretherton, Oppenheim, Emde and the MacArthur Narrative Working
Group, 2003; Bretherton, Ridgewa and Cassidy, 1990; George and West, 2012). Projectives
are typically used in quantitative designs – complex coding systems have been developed
that allow researchers to iron out the variability in individual responses to the projective
stimuli, and turn the rich narrative detail into numbers and categories suitable for quantitative
analysis (e.g., Exner, 2002, for the Rorschach Inkblot Test). It is difficult not to regret the loss
As highlighted, projective tests make the assumption that hidden truths are revealed about
the test takers: ‘indeed it is often because projective methods are supposed to be better at
getting at what people “really” think, that they are recommended’ (Kitzinger and Powell,
1995: 349). For some, this is what underpins SC as a method, as it is assumed that there is
a truth that can be discovered through the research process. Therefore, those who use
essentialist stance on the person and on the data. Such an approach doesn’t sit well with
the rest of the chapter. First suggested in a 1995 study by two feminist psychologists
(Kitzinger and Powell, 1995), this approach situates SC within a qualitative framework
3
Celia Kitzinger and Debra Powell (1995) used SC to examine how 116 undergraduate
research, the (ambiguous) stimulus the participant has to respond to is the opening lines to a
story (the ‘story stem’), which they are instructed to complete. Kitzinger and Powell used a
comparative design to explore differences in responses when the unfaithful person was a
man versus a woman. The ‘unfaithful male partner’ version of the story stem read: ‘John and
Claire have been going out for over a year. Then Claire realises that John is seeing
someone else’ (p. 352). In the ‘unfaithful female partner’ version, the names in the second
Equal numbers of participants responded to each version of story stem. The researchers
also made sure that roughly equal numbers of male and female participants completed each
version, to allow them to explore differences in how the male and female participants made
sense of the scenarios. In contrast to existing frameworks, the authors suggested that it was
not necessary to read the stories as (only) revealing the psychological ‘truth’ of the
discourses upon which subjects draw in making sense of experience’ (Kitzinger and Powell,
1995: 349-350). This approach to SC is a social constructionist one that rejects the idea that
it is possible to access ‘real’ or ‘true’ feelings or thoughts, and assumes instead that realities
Kitzinger and Powell (1995) illustrated the differences between the two approaches by
to infidelity, with a social constructionist one, in which the data were read as replicating
various discourses about the meanings of infidelity for men and women. In this context, male
4
participants’ propensity to write more about sexual than emotional infidelity did not reveal
‘young men’s preoccupation with sex’ (p. 350) but rather said something about their greater
fiction. One of the aims in the current chapter is similarly to hand researchers the choice of
which ‘lens’ to apply to their data, something that makes the SC method eminently adaptable
In common with all of the techniques and approaches discussed in this book, SC methods
have the advantage of being less demanding of time and resources than established face-to-
face interactive methods, such as interviews and focus groups. Hard copy stories, for
instance, can be handed out to a large group of people and the completed stories returned
in 30 minutes or so; online stories can be distributed (and then downloaded) with a few
mouse clicks.
The advantages of SC are not limited to being resource-lite, however. We now outline some
and focus groups – in which small numbers of participants are asked to provide their
asking about the hypothetical behaviour of others (Will, Eadie and MacAskill, 1996; also
5
hypothetically, and in the third person, they do not have to take ownership of, or justify,
their stories in the way they would if they were being asked directly about the topic.
Therefore, they are more likely to 'relax their guard' and engage with the research topic
with less reserve. This gives SC the unusual advantage of breaking down the ‘social
desirability “barrier” of self-report research’ (Moore, Gullone and Kostanski, 1997: 372).
Traditionally, this has posed a problem for essentialist research, which has sought to tap
into participants’ ‘real’ views or perceptions: participants not responding truthfully creates
a validity concern for such research. (It’s important to note that not all SC researchers
ask participants to write in the third person; an example of a first person SC is discussed
below.)
2) SC ideally suits sensitive topics: SC also offers a particularly accessible way for
participants to take part in research, because it does not necessarily require personal
experience of the topic (also see Chapter 4, on vignette research). The use of
hypothetical scenario story telling also means participants are slightly 'removed' from the
topic. This makes SC especially useful for exploring sensitive topics - if questioned
directly about their own experiences, some participants feel uncomfortable, or even
unwilling, to discuss such topics. Sensitive topics that have been explored utilising SC
include orgasmic ‘absence’ (Frith, 2013) and sex offending (Gavin, 2005).
3) SC gives participants control and allows for creativity: Many qualitative researchers value
methods – like focus groups – that are more participant-led and ‘hand back’ some of the
that affords participants more control and creativity than other methods. The ambiguity of
some story stems, for instance (see design section), means that participants have lots of
6
scope to choose the direction and style of their story. They are the sole authority of what
psychologists Jennifer Livingston and Maria Testa (2000), for example, used qualitative
male aggression in a heterosexual dating scenario. The participants completed the story
stem under different experimental conditions (one group was given alcohol to drink
before completing the story, another a placebo, and the third, was not given a drink). The
participants were presented with a first story stem with a male character, Mark. They
were told that ‘you think he’s really good looking’ (p. 741); Mark later phones sounding
drunk and then ‘shows up at your door’ (p. 741). Thus the researchers asked women to
imagine themselves as the female character in their story, and to write in the first person;
they treated the women’s responses as representing their beliefs about this topic.
Third person SC has also been interpreted through an essentialist lens. Psychologist
Susan Moore (1995), for instance, explored girls' beliefs about menarche by asking Year
Six (eleven year old) Australian girls to each complete five different menstruation story
stems.
The second way in which SC data have been interpreted is through the identification of
7
constructionist research on orgasmic ‘absence’, which treated SC data as capturing the
cultural discourses available to participants. She used two versions of a story stem,
featuring a heterosexual couple – Lisa and Tom. In one version, Tom realises Lisa has
not had an orgasm; in the other version, it is Lisa who realises Tom has not had an
orgasm. Frith identified three themes in the data these stems generated. The analysis
explored how the stories drew on and reinforced various gendered discourses, including
women’s responsibility to be sexually attractive to maintain men’s sexual interest and the
notion that men’s sexual desire is unbridled and easy to satisfy. Contextualist research,
which sits somewhere between essentialism and constructionism, and where multiple
truths or situated realities are understood to exist within particular contexts (Braun and
Clarke, 2013), is also possible using SC. However, to date there are no published
5) SC offers robust and easy-to-implement comparative design options: This feature (which
between groups of participants or between versions of the same story and how they are
made sense of. As outlined above, Kitzinger and Powell’s (1995) ground breaking study
example, critical psychologists Virginia Braun and Victoria Clarke (2013) used two
versions of a story to explore people's perceptions of trans parenting. The story stem
described a parent telling their children that they are uncomfortable living within their
assigned gender and want to start the process of changing sex. Roughly half of the
participants completed a male parent (Brian) version and half an otherwise identical
female parent (Mary) version. Having two versions enabled the researchers to compare
8
the responses both according to the gender of the parent character and the gender of
the participant. This was important because mothers and fathers tend to be perceived
very differently in the wider culture, and women tend to be more tolerant of gender
recently begun to fully explore the possibilities that SC offers. For example, critical
psychologists Nikki Hayfield and Matthew Wood (2014) recently piloted a SC using
visual methodologies (Frith, Riley, Archer and Gleeson, 2005) in their research on
perceptions of appearance and sexuality (see ‘steps to using SC’ below). The stem
described a dating scenario; once they had completed their stories, participants were
directed to the website Bitstrips to create a cartoon image of the main character. A
preliminary analysis of the images indicated that participants recognised the existence of
lesbian and gay appearance norms, which was not necessarily as apparent in their
written responses. Hence, visual data may provide an anchor for, or 'bring to life', textual
responses, and can also be analysed in their own right. This allows the potential for
different understandings of, insights into, and interpretations of the findings (Frith et al.,
2005).
including the ease of implementing comparative designs – means that it fits well with
different social groups make sense of a scenario, and whether participants respond
differently to variations in, for example, the story character’s gender or sexuality. We
9
document examples of our and others' gender and sexuality research throughout the
The flexibility of SC is one of its key advantages and accordingly it can be used to research
understandings and social constructions. However, questions that focus on people's lived
experiences are not well suited to SC research, because you’re not gathering stories of their
experiences (see ‘methods of analysis’ below). When developing your research question(s),
as in any qualitative project, you will need to ensure it is both focused on a specific topic, but
also broad and open-ended (typically asking exploratory 'what' or 'how' questions). For
example, Kitzinger and Powell (1995) aimed to 'explore young men's and women's
representations of "unfaithful" heterosexual relationships' (p. 345), and Frith (2013: 312)
examined 'how people account for and explain orgasmic absence during heterosex'. These
question are specific enough to guide the research and design, but open enough so that
there is plenty of scope for fully exploring participants' responses. It is also important to
ensure that the type of question you create 'fits' with your epistemological approach;
'representation' questions are most often used in constructionist and critical research. Table
10
The most important design consideration in SC research is the design of the story stem: the
‘start’ of a story that participants are asked to complete. A careful balance needs to be struck
between providing the participant with a meaningful story stem, and leaving enough
ambiguity for tapping into their assumptions (or ‘perceptions’ or ‘psychological projections,’ in
essentialist research). Braun and Clarke (2013) discussed five considerations in story stem
design:
1) Length of the story stem: How much of the beginning of the story will you write? There
are no hard and fast rules here; it depends on your topic and participant group. If the
necessary for the scenario to be meaningful to them. For example, in Victoria Clarke’s
was safe to assume the participants had knowledge of the topic, so a very short stem
was used (this is the female version): ‘Jane has decided to stop removing her body
hair…’ For a less familiar or more complex topic, such as one focused on the character’s
psychology, your participants may need more detail to understand the scenario that is
the focus of the stem. For instance, critical psychologist Irmgard Tischner’s (2014)
decided that he needs to lose weight. Full of enthusiasm, and in order to prevent him
from changing his mind, he is telling his friends in the pub about his plans.’ Although
weight-loss is a familiar topic to most people, the main focus of the research was on
social perceptions and interactions around weight-loss intentions; this necessitated the
story stem including the protagonist’s interaction with other people, i.e., him telling his
11
2) Authentic and engaging scenarios and characters: Unless the story, its protagonists, and
the context resonate with your participants, it is unlikely they will write a useful story.
Your stem should engage your participants and be easy for them to relate to. Using
names and scenarios that sound authentic and believable will help your participants
imagine or ‘see’ the characters and the scenario, and thus to write a rich and complex
story.
3) Amount of detail: The most difficult design decisions revolve around the issue of detail in
the story stem. Too much detail and direction will potentially limit the variation and
richness of the data; not enough could mean the participants will not know ‘where to
take’ the story, resulting in data that do not address your research question. You need to
design a story stem that stimulates a range of complex and rich stories. To achieve this,
give the participants adequate directions by giving them a context or background to your
story, and some detail about the characters, what the topic of the story should be about
(and what you are actually asking participants to do, which is discussed below). At the
same time you also want to avoid overly constraining their responses, by describing the
background and characters in too much detail. Participants need to know what their story
should be about, but you don’t want to give them the plot or ending. So if you want them
to write about motivations for exercise, for instance, a very open story stem like ‘Toby
decides to become more physically active… What happens next?’ may take the stories
in too many, and possibly undesired, directions, and not focus on Toby’s motivations. On
the other hand, giving participants a particular motivation in the story stem (e.g., ‘Toby
your data, as participants follow your lead and don’t describe the range of understood
12
motivations to take up exercise (a further example is given under the heading ‘what can
go wrong’, below).
assumption that a couple consists of a man and a woman. This can often be achieved by
characteristics of your protagonists (e.g., class, sex, race, sexuality, age). However, if
of the story, this shouldn’t be left ambiguous. For example, Victoria Clarke, Virginia
Braun and Kate Wooles’ (2014) study comparing constructions of same-sex and
5) First or third person: The final design consideration concerns the standpoint you want
your participants to take. Do you want them to step into the shoes of, and empathise
Although to date qualitative SC has involved mostly third person story stems, first person
stems are possible (e.g., Livingston and Testa, 2000). These can be useful if it is
important for the participants to write from the perspective of a specific character. From a
desirable responses (Rabin, 1981). Therefore, if you want to gain a broader range of
stem.
13
6) Completion instructions: Think carefully about the completion instructions provided to
participants (see ‘steps to using story completion’ also). Is it necessary that they write
about a particular aspect of the scenario? Do you want to know about how the story
develops (in the future)? Or the ‘back story’ to the scenario? For instance, if it’s
particularly important that your participants provide a description of the characters, you
need to include this in your completion instructions. For example, the stem on weight-
loss intentions discussed above (Tischner, 2014) was followed by the instructions:
‘Please complete and expand on this story by describing Thomas to us, and telling us
how the story unfolds: what is Thomas saying to his friends about his reasons and
So how many participants or stories should you aim for? In existing SC research, there is a
large variation in sample sizes – from 20 (Walsh and Malson, 2010) to 234 (Whitty, 2005)
your design – more stories generally require more participants to be able to say something
meaningful about each version, especially if you intend to make comparisons; b) the
richness of individual stories – richer stories mean fewer participants (note however than you
may not be able to predict in advance how rich the stories will be); and c) the purposes of
your research. For a small student project, with a one stem design, and no comparison
between different participant groups, around 20 – 40 participants is likely to provide you with
data that are rich and detailed enough for a meaningful analysis. The more comparisons you
make, the bigger your overall sample will need to be. Braun and Clarke (2013) advise
recruiting at least 10 participants per story stem variation, but should you aim to publish your
14
report, you may find that journal editors and reviewers require higher participant numbers
than that.
Of course, as with any research, recruiting enough participants can be a challenge, which is
why many studies are carried out with a student population. Students, however, are a very
specific population, and often not very diverse in terms of demographics. At the same time,
students are used to discussing and describing ideas in writing, tend to be fairly literate, and
thus will not struggle with the task of writing a story (Kitzinger and Powell, 1995); the same
cannot be assumed for all other participant groups. Think carefully about the needs and
expectations of your participants. For example, busy professionals may require very clear
As a general rule, SC research raises fewer ethical concerns than research that involves
direct interaction with participants and asking them about their personal lives; this is
particularly the case for online SC studies that make it even easier for participants to be
anonymous and reduce risk for both participants and researchers. However, participant
comfort with the topic is still an important ethical consideration, particularly for sensitive
topics, and standard accepted ethical practice still needs to be adhered to (e.g. British
Psychological Society, 2009). Follow the relevant ethical guidance of your institution and/or
professional body.
Step One: Decide if you want to use a comparative design: With a comparative design you
can explore and compare the assumptions made, or perceptions held, about certain social
groups or scenarios. If this is your aim, you need to design versions of your story which
15
reflect the specific differences you are concerned with, and allocate roughly equal numbers
of participants to each of these. For example, Tischner (2014) used a comparative design to
explore the gendered constructions of body weight concerns and weight-loss motivations.
This necessitated two story stems, with a male and female protagonist respectively. Clarke
et al.’s (2014) research on infidelity employed a more complex comparative design. Their
aim was to explore how same- and different-sex emotional and sexual infidelity were
conceptualised in the context of heterosexual marriage. This required four story stems. We
do caution against having too many versions of a story in one study, and the use of overly
complex and dynamic) meaning, rather than compartmentalisation. Two to six is the
manageable maximum number for small and medium-sized projects in terms of both
Another level of comparison involves different participant groups, and exploring the
differences between the stories written by people who are, for instance, from different
perceptions of fat therapists (which will be described more in Step Four) included responses
from 18-21 year old undergraduate psychology university students and 16-18 year old sixth
formers. This design made it possible to consider both the salience of counsellor body
weight for the whole group of young people, but also how small differences in age and
educational experience impacted on the expression of fat stigma. Whereas the stories of
16
both groups clearly reiterated anti-fat cultural narratives, the younger cohort were much
Step Two: Determine how many stories each participant will be asked to complete: When
using a comparative design with multiple versions of the story stem, you have the option of
asking participants to complete one, or more than one, story. In psychologist Helen Gavin's
(2005) research on the social construction of sex offenders, each participant was asked to
complete six different versions of a story stem. She did so to explore how individual
participants’ narratives surrounding sex offenders varied when presented with different
situations. Similarly, in a study on adolescent risk-taking the researchers asked all of the
Asking participants to complete more than one stem may reflect a more pragmatic concern
to maximise the number of stories in the data-set. For example, Iduna Shah-Beckley’s (see
heterosex asked participants to complete two versions of a story stem. This halved the
number of participants she needed to recruit. One concern when asking participants to
respond to multiple story stems is that there may be order effects, with participants writing
their longest story for the first story stem. However, In Iduna’s research, the opposite was
true, with participants writing longer stories in response to the second stem.
Step Three: Write your instructions: After you have designed your story stem(s), you need to
write completion instructions for participants. In the participant information sheet, you should
provide participants with some information about the nature of the task, and what they are
17
expected to do, emphasising the necessity of writing a story. Here is an example from
You are invited to complete a story – this means that you read the opening
sentences of a story and then write what happens next. There is no right or
wrong way to complete the story, and you can be as creative as you like in
completing the story! I am interested in the range of different stories that people
tell. Don’t spend too long thinking about what might happen next – just write
the story are deliberately vague; it’s up to you to be creative and ‘fill in the
blanks’!
Then, ideally just before or after you present participants with the story stem, you need to
provide specific instructions on how they should complete the story (unless you do not want
to constrain their responses in any way). Completion instructions can vary from the broad
and open to the more prescriptive and directive. For example, Victoria (Clarke, 2014)
instructed participants to simply 'read and complete the following story'. Another common
instruction is to ask participants to write 'what happens next'. Nikki Hayfield and Matthew
prescriptive approach. Because they wanted participants to focus on the events before,
during and after the female character’s date, they instructed participants to write their story in
three sections. Their story varied by character sexuality (bisexual, lesbian and heterosexual);
18
Jess is a 21 year old lesbian woman. She has recently met someone, and they
Please write about the run-up to the date and how she prepared for it…
Please write about what happened next… (Please feel free to write as much
as you like about the characters and as far into the future as you like)
You may also want to provide participants with clear instructions on the length of story you
wish them to write, or a time-expectation, to help ensure you get the quality of data you
need. For example, we have instructed participants to spend a certain amount of time writing
their story (e.g., 'please spend at least 10 minutes'), or to write stories of a particular length
(e.g., see Victoria’s [Clarke, 2014] example above). Such instructions are particularly
important for participant groups who are not necessarily highly motivated, such as
individuals who take part in order to access particular benefits associated with participation.
assess whether participants interpret the stem and instructions in the way you intended (see
Step Five below). In Victoria’s (Clarke, 2014) study, for instance, the instructions ‘you are
asked to WRITE A STORY THAT IS AT LEAST 10 LINES/200 WORDS LONG’ were added
after piloting, because the pilot stories were often very brief or did not seriously engage with
the task.
Potential Step: Write additional questions: Although one of the key features of SC is that it is
provides an indirect approach, some researchers have combined the use of a story stem
with a small number of direct questions (in a way that combines some aspects of vignette
19
research, see Chapter 4). For example, Naomi Moller’s (2014) research on perceptions of fat
Please read and complete the following story: Kate has been feeling finding it really
difficult to cope with life so she has decided to go for counselling. As she walks into
the counselling room for the first time, her first thought is: ‘Oh, my counsellor is fat!’
What happens next? (Please spend at least 10 minutes writing your story)
After completing the story, participants were asked a direct question about the counsellor
featured in the story stem: ‘What weight did you think the counsellor was?’ The answers to
this question allowed Naomi to understand how the participants’ defined ‘fat’ – a variable
You should also consider whether it is important to ask participants demographic questions
beyond the ‘standard’ questions about age, sex/gender, race/ethnicity, sexuality, disability
and social class (see Braun and Clarke, 2013). Such questions can provide a useful
‘baseline’ for interpreting and contextualising your stories. For example, in her research on
body hair, Victoria Clarke (2014) asked a series of questions about whether participants had
currently or previously removed or trimmed body hair in particular areas and their reasons
for doing so. Given that for women, but increasingly for men too, body hair removal is a
dominant social norm (Terry and Braun, 2013; Braun, Tricklebank and Clarke, 2013), an
overview of the participants’ own body hair practices provides important information for
Step Four: Determine mode of data collection: Another consideration is whether to conduct
your study using ‘paper and pen’ completion, or electronically either online using (free or
20
subscription) survey software such as Qualtrics (www.qualtrics.com) or SurveyMonkey
body of an email. An advantage of hard copy completion is that you can hand the SC directly
events), and, providing you have ethical approval, offer participants a small ‘reward’ (such as
chocolate) for returning their story. However, you then need to manually type up participants'
The key advantage of electronic data collection is that responses require little preparation for
analysis – emailed stories will need to be cut and pasted and collated in a document; online
can complete the study at a time and place that suits them. However, online SC research
that requires participants to have Internet access can limit who can take part; it is the least
privileged members of society that tend to have limited or no Internet access (Hargittai,
2010), and some groups (such as older participants) may be uncomfortable with, or find
difficult to use, certain types of technology (Kurniawan, 2008). The fact that participants can
now complete online studies on smart phones and tablets (there is a Qualtrics 'app' that
users can download for free) may also impact on data quality. Mobile devices often utilise
'soft' keyboards that do not necessarily facilitate accuracy of typing, or indeed typing full
stop. Features such as auto-correct may mean that unless participants look closely at their
responses as they are typing, inaccurate 'corrections' can be made. Therefore, detailed (and
coherent) responses may be restricted by the need to constantly check the screen, as well
However, some research has indicated that as long as participants do not need to enter
21
numerical as well as alphabetic data (thereby requiring switching between soft-keyboards)
completion on mobile devices will not necessarily take participants much longer, nor impact
on errors (Sears and Zha, 2003), and this may also apply to tablets which are generally
larger and more ‘typing-friendly’ than mobile phones. Finally, another important
consideration is achieving a good fit between your mode of data collection and your
participant group. You don’t have to restrict yourself to one mode – it may be most
appropriate to ask some participants to complete the study online and others on hard copy.
Step Five: Pilot your SC: Given the open-ended and exploratory nature of SC research,
piloting your stem and instructions to ensure they elicit relevant and useful data is vital
(Braun and Clarke, 2013). We have often made minor (but transformative) amendments to
stems or instructions following piloting. The resource-lite nature of SC means that piloting is
not generally an onerous task. We recommend piloting your stem on the equivalent of 10-
20% of the intended final sample; the precise number should be determined in relation to the
diversity within your participant group: greater diversity = larger pilot sample. You can pilot in
one of two ways: 1) by treating early data collection as a pilot and using their responses to
judge if the stem and your instructions have been interpreted in the way(s) you intended, or
2) by asking participants to both complete the study and comment on the clarity of the
instructions and the study design. If you make no (or minimal) changes to the stem following
piloting, the pilot data can be incorporated into your sample. Once all these steps are
completed, you are ready to keep calm and collect your data!
22
The generation of poor quality data is a concern across most qualitative data collection
methods; SC can also ‘go wrong’ in this way, and it can result from a number of different
factors. Participants can sometimes ‘refuse’ the task by not completing the story as
requested – for example by not writing their response as a story. This may result from a
simple failure to understand the task. For instance, in Iduna Shah-Beckley and Victoria
about what Ben and Kate might be feeling, and what might happen to their relationship, but
not in the format of a story (see Story 1 in Box 3.2). Therapists are busy professionals, and it
seems likely that they did not spend much time reading the detailed participant information,
and thus did not understand what was being asked of them. This shows the importance of
providing clear but not overly long instructions, and repeating and highlighting key
instructions. Participants may also generate short or shallow stories (see Story 2 in Box 3.2).
This is often the result of low participant motivation – as noted above, we have found that
individuals participating for reasons other than wanting to contribute to the study (e.g.
benefits associated with participation such as students gaining course credit) often write very
short stories unless given explicit (and repeated) instructions to produce stories of a certain
length. But such instructions can iron out variability in story length – eliminating both very
short and longer, richer and more complex, and thus highly desirable, stories. One way to
manage this is to over recruit, so you can eliminate stories under a certain length from the
final data-set.
23
Short or shallow responses can also result from the design of the story stem. Story stems
that constrain participant’s creativity in how they continue and complete the story, or suggest
a very likely single outcome, often produce rather thin and narrow data. For example, a
student project using a story stem about a student feeling anxious about giving an assessed
presentation produced shallow stories, which mostly ended with the student successfully
giving the presentation (Braun and Clarke, 2013). The data did not provide the basis for a
rich and complex analysis. In sum, the lessons we have learned are that: 1) it is important to
write story stems that allow for a range of possible outcomes, hence maximising the
potential for participant creativity; and 2) piloting of the stem is crucial (see above).
Another potential problem is that participants can sometimes write stories that contain
elements of humour and fantasy. Virginia Braun and Victoria Clarke (2013) found this in their
research on perceptions of a parent coming out to their children as transgendered, with one
story containing the memorable line ‘Brian rubs his nipple and then David Beckham
appears’. You don’t need to know much about the study to appreciate the participant’s failure
to take the task seriously! Such stories may potentially reflect participant discomfort with the
topic. In the instance of this study, the prevalence of 'transphobia' in the wider society
(Nadal, Skolnik and Wong,, 2012), and the content of some of these stories (‘Brian's… over
the moon that the tax payer is picking up the bill for a completely unnecessary procedure’)
suggests this as a potential explanation. However, ‘fantasy’ stories are only a potential
problem; for some research questions and approaches, they may actually provide useful
data. For example, in Victoria Clarke’s (2014) social constructionist research on non-
normative body hair practices, fantasy stories about Jane stopping removing her body hair
24
and running away to live as a yeti in the wild were highly pertinent, providing useful
Such humorous or fantasy stories highlight another challenge with the SC method – the data
are potentially more difficult to interpret than self-report data. We’ve noticed that some
student-researchers get confused about what SC data represent, treating the fictional
characters as real people and equating the stories with self-report data. This meant that, for
example, creative responses to hypothetical scenarios about a parent coming out to their
children as transgender were treated by some students (analysing the data for an
assignment) as providing information about the real world impact of a parent undergoing a
gender transition on child development. It’s important to remember that SC produces just
that – stories – which may (depending on your epistemological standpoint) reveal something
about what participants think and feel about a particular topic. Because of the nature of SC
data – in our qualitative context, creative stories about hypothetical scenarios rather than
What Methods of Analysis Are Suitable for Use With Story Completion Data?
To date, two methods have been used to analyse SC data – thematic analysis (TA) (e.g.,
Clarke et al., 2014; Frith, 2013; Livingston and Testa, 2000) and discourse analysis (DA)
(Walsh and Malson, 2002). Following Kitzinger and Powell, TA (Braun and Clarke, 2006,
2012) is often used slightly differently from how it is used to analyse self-report data. Rather
than simply identifying patterns across the stories as a whole, researchers have identified
patterns in specific elements of the story (both of these can be thought of as a variant of
25
horizontal patterning, in the sense that the patterns intersect the stories). For example, SC
research on perceptions of relational infidelity has identified themes in how the relationship
(both that between the primary partners, and that between the unfaithful partner and the
‘other’ man/woman) is presented, how infidelity is accounted for, and how the responses to
and consequences of infidelity are depicted (Kitzinger and Powell, 1995; Whitty, 2005). This
means that SC researchers have identified particular questions they want to ask of the data
(in advance of the analysis, or after data familiarisation) and used the techniques of TA to
As noted above, Kitzinger and Powell (1995) demonstrated that both essential and
constructionist readings of SC data are possible, and TA has been used to analyse SC data
approach for constructionist approaches to SC (Braun and Clarke, 2013). For example,
critical psychologists Eleanor Walsh and Helen Malson (2010) used post-structuralist DA
(e.g., Wetherell, Taylor and Yates, 2001) to interrogate some of the ways in which their
participants made sense of anorexia and bulimia, and constituted the causes of, and
recovery from, eating disorders. They explored how the participants constructed ‘dieting’ as
normal and healthy, for instance, and the ways in which recovery from eating disorder was
framed in terms of a return to ‘normal’ dieting rather than (say) a return to unrestricted eating
examined vertical patterning – patterns in how stories unfold. One approach very useful for
this type of ‘narrative’ analysis is Braun and Clarke’s (2013) story mapping technique that
involves distinguishing patterns in the key elements of a story’s progression. Braun and
26
Clarke provide the example of a study exploring perceptions of a young woman ‘coming out’
to her parents as non-heterosexual. The story map for this study identified patterns in: (1)
the parent’s initial reactions to the coming out; (2) the development of the stories; and (3) the
ending or resolution of the stories. After an initial expression of shock, the parents’
responses to their daughter coming out were categorised as either (broadly) positive or
negative; the negative reaction stories either ended positively, negatively or ambiguously,
and the positive reaction stories always ended positively (see Figure 3.1). Depending on
your research question and approach, this story mapping technique can be a useful
complement to a standard pattern-based analysis (e.g., TA), which helps the analysis to
retain a sense of the storied nature of the data. This technique also lightly captures
(Western) cultural conventions around story-telling (beginning, middle, end) and the
dominance of particular genres (e.g., ‘happily ever after’, ‘triumph over adversity’).
One analytic approach that has yet to be used to analyse SC data, but nonetheless seems
particularly apt, is narrative analysis (Riessman, 1993; 2007). Narrative techniques could be
productively used to identify narratives types and genres, and the structures and styles of
particular narrative types, thus extending and developing Braun and Clarke’s (2013) story
mapping technique.
recommend the use of frequency counts in the analysis of self-report data, because of the
organic and participant-responsive nature of self-report data collection (Braun and Clarke,
2013). However, frequency counts are often used in the analysis of SC data. For example, in
27
their research on perceptions of infidelity, Kitzinger and Powell (1995) asked how many
participants interpreted Claire ‘seeing someone else’ as Claire being unfaithful – a full 10%
rejected the implications of infidelity. When asking such concrete questions of the data (and
when participants have been set an identical task), reporting numbers or percentages rather
than using looser words such as ‘most’ or ‘some’ to capture patterning in the data is entirely
appropriate.
Certain analytic approaches are not suited to the analysis of SC data, including approaches
such as Interpretative Phenomenological Analysis (Smith, Flowers and Larkin, 2009) and
(Reissman, 2008). Because participants are not asked for their views directly, and are often
asked to write stories about things they may have little or no personal experience of, it’s
unclear whether SC data tell us anything meaningful about participants’ lived experience.
Without some big interpretative leaps, SC data would need to be combined with another
data source to be suitable for use in research focused on lived experience. Grounded
Theory has similarly not been used to analyse SC data, and the focus on theory generation
and the examination of the social processes and factors that shape particular phenomenon
conversation analysis (e.g., Schegloff, 2007) and discursive psychology (e.g., Wiggins and
Potter, 2010) – are not well suited to SC data. These approaches typically focus on ‘talk-in-
interaction’; the ‘what’ and ‘how’ of ‘real’ talk – both everyday ‘real’ talk and that produced in
institutional contexts such as courtrooms or consulting rooms – which is rather different from
28
Conclusion
In sum, SC produces data that provide a major, and accessible, alternative to self-report
methods of data collection. SC allows participants control and creativity, and the resulting
data can be fun, rich and complex. SC also offers researchers new and exciting ways to
generate data that provides compelling insights into the topic at hand.
Have a Go…
1) Develop a research question suitable for use with SC and determine your participant
group. Design a story stem that could be used to address this research question with this
participant group. Think carefully about what details should be included (will your
participants know anything about the topic?), and whether any aspects of the stem
should be ambiguous.
2) The following story is from Victoria Clarke’s (2014) research on perceptions of non-
normative body hair practices. Code the data in relation to the research question ‘how do
young people make sense of a woman stopping removing her body hair?’
What are your main analytic observations about this story? Next, consider whether
Braun and Clarke’s (2013) story mapping technique could usefully be applied to this
story? How would you code the opening, development and resolution of the story?
Jane has decided to stop removing her body hair… After years
29
body and the way she looks. Jane does not feel the need to
does keep herself fit and healthy and wears make-up and feels
body hair. Jane only removes her body hair on her legs and
haired bodies.
The companion website for Braun and Clarke’s (2013) book Successful qualitative research
www.uk.sagepub.com/braunandclarke
The paper that introduced SC as a qualitative method: Kitzinger, C. and Powell, D. (1995).
30
Virginia Braun and Victoria Clarke further developed the SC method for qualitative research
(see Chapters 6 and 10): Braun, V. and Clarke, V. (2013). Successful qualitative research: A
Read about one of the examples discussed in more detail: Clarke, V., Braun, V. and Wooles,
K. (2015). Thou shalt not covet another man? Exploring constructions of same-sex and
different-sex infidelity using story completion. Journal of Community & Applied Social
absence: Exploring heterosex using the story completion method. Psychology and Sexuality,
4(3), 310-322.
constructions of eating disorders: A story completion task. Feminism and Psychology, 20(4),
529-537.
Reference list
Clarke, V. and Braun, V. (2008). Gender. In D. Fox, I. Prilleltensky, and S. Austin (Eds.),
Braun, V. and Clarke, V. (2013). Successful qualitative research: A practical guide for
31
Braun, V. and Clarke, V. (2012). Thematic analysis. In H. Cooper, P. M. Camic, D. L. Long,
Braun, V., Tricklebank, G. and Clarke, V. (2013). ‘It shouldn’t stick out from your bikini at the
beach’: Meaning, gender, and the hairy/hairless body. Psychology of Women Quarterly,
37(4), 478-493.
Bretherton, I., Ridgeway, D. and Cassidy, J. (1990). Assessing internal working models of
Theory, research and intervention (pp273-308). Chicago, IL: University of Chicago Press.
Bretherton, I., Oppenheim, D., Emde, R. N. and the MacArthur Narrative Working Group
(2003). The MacArthur Story Stem Battery. In R. N. Emde, D. P. Wolfe, and D. Oppenheim
(Eds.), Revealing the inner worlds of young children: The MacArthur Story Stem Battery and
British Psychological Society (2009). Code of ethics and conduct. Leicester: British
Psychological Society.
32
Charmaz, K. (2006). Constructing grounded theory: A practical guide through qualitative
Clarke, V. (2014). Telling tales of the unexpected: Using story completion to explore
Clarke, V., Braun, V. and Wooles, K. (2015). Thou shalt not covet another man? Exploring
Frith, H. (2013). Accounting for orgasmic absence: Exploring heterosex using the story
Frith, H., Riley, S., Archer, L. and Gleeson, K. (2005). Editorial. Qualitative Research in
Gavin, H. (2005). The social construction of the child sex offender explored by narrative. The
George, C. and West, M. L. (2012). The Adult Attachment Projective System, Attachment
Theory and Assessment in Adults. New York, London: The Guildford Press.
33
Guest, G., MacQueen, K. M. and Namey, E. E. (2012). Applied thematic analysis. Thousand
Thomas.
Hargittai, E. (2010). Digital na(t)ives? Variation in Internet skills and uses among members of
Hayfield, N. and Wood, M. (2014). Exploring sexuality and appearance using story
completion and visual methods. Paper presented at Appearance Matters 6, 1-2 July 2014,
Bristol, UK.
Holaday, M., Smith, D. A. and Sherry, A. (2000). Sentence completion tests: A review of the
literature and results of a survey of members of the Society for Personality Assessment.
Kidder, L. H., & Fine, M. (1987). Qualitative and quantitative methods: When stories
converge. In M. M. Mark & L. Shotland (Eds.), New directions in program evaluation (pp. 57-
constructionist readings of a story completion task. Feminism and Psychology, 5(3), 345-
372.
34
Livingston, J. A. and Testa, M. (2000). Qualitative analysis of women’s perceived
Mays, N. and Pope, C. (1995). Reaching the parts other methods cannot reach: An
introduction to qualitative methods in health and health services research. British Medical
Moller, N. (2014). Assumptions about fat counsellors: Findings from a story-completion task.
Morgan, W. G. (2002). Origin and history of the earliest Thematic Apperception Test
Nadal, K. L., Skolnik, A. and Wong,,Y. (2012.) Interpersonal and systemic microaggressions
35
Rabin, A. I. (Ed.), (1981). Assessment with projective techniques. New York: Springer
Publishing Company.
Riessman, C. K. (2007). Narrative methods for the human sciences. Thousand Oaks, CA:
Sage.
Diagnostic Test based on Perception, 10th Revised Edition. Verlag, Switzerland: Huber.
Rotter, J. B., Lah, M. I. and Rafferty, J. E. (1992). Rotter Incomplete Sentences Blank. San
Sears, A. and Zha, Y. (2003). Data entry for mobile devices using soft keyboards:
Understanding the effects of keyboard size and user tasks. International Journal of Human-
36
Shah-Beckley, I. and Clarke, V. (2015). Exploring male and female therapists and
Terry, G. and Braun, V. (2013). To let hair be, or to not let hair be? Gender and body hair
Wetherell, M., Taylor, S. and Yates, S. (2001). Discourse theory and practice: A reader.
Rogers (Eds), The Sage handbook of qualitative research in psychology (pp. 73-90).
London: Sage.
Wilkinson, S. (1999). Focus groups: A feminist method. Psychology of Women Quarterly, 23,
221-244.
37
Will, V., Eadie, D. and MacAskill, S. (1996). Projective and enabling techniques explored.
38
Box 3.1: Introducing the Example Studies
We have been ‘experimenting’, in the broadest sense, with SC for the last decade or so,
and in this chapter we share what we have learnt about, and our enthusiasm for, the
method, drawing on a wide range of different SC studies from our, and our students’,
research. The example studies reflect our interests in gender, sexuality, appearance and
sexuality and appearance… Again, reflecting our shared interests in gender, most of these
studies use a comparative design to explore gender variation – both with regard to the
responses of male and female (or other gendered) participants, and responses to male
39
Box 3.2: Examples of Story Completion Data (from Shah-Beckley and Clarke, 2015)
The story stem: “Ben and Kate have been together for a few years. For quite some time
they have not been having sex because Ben does not want to. Kate has tried talking to
Ben but he has been reluctant to talk. Tonight Kate is making sexual advances but Ben
says he is tired and turns over… What happens next?” (In a second version of the story
Kate refuses sex.) We have corrected all the spelling errors and typos in the data.
1) Story ‘refusal’: “If that will happen from now on, she will challenge him to talk about it
2) Example of a short and thin story: “Kate is then upset as she feels unattractive. Ben
doesn't want to discuss it further so becomes defensive and dismissive. They have an
3) An excerpt from a longer and richer story: “Kate then decides that enough is enough -
what's wrong with him? Am I unattractive? Is there someone else? Is he worried about
something he hasn't told me? Kate challenges Ben, ‘I can't keep doing this - you need
to tell me what's going on. Is there something you're worried about? Something you
feel you can't tell me? Please try - I just want to understand’. Ben sighs and turns back
over to face Kate. He places his hand on her face and looks at her – ‘It’s not you’ he
says, ‘I just feel like I've lost the urge to have sex…” (The story continues for another
216 words.)
40
Box 3.3: Personal Reflections on Using Story Completion From Iduna Shah-Beckley
refusal in heterosexual relationships; Shah-Beckley and Clarke, 2014) and now for my
doctoral research, which explores how therapists and non-therapists’ make sense of
heterosex. I have just finished collecting my data (200 stories) and begun the process of
research. Before using SC, I only had experience of quantitative research, which often left
me feeling very dissatisfied because the kind of data quantitative methods produced was
simply not useful for addressing the kinds of research questions I was interested in.
Broadly speaking, I am interested in how social norms around sexuality are produced and
perpetuated, and the ways in which men’s and women’s sexuality are differentially shaped
and constrained by social norms. SC is very useful for addressing these kinds of
questions. For me, it’s the best of both the quantitative and qualitative worlds, as it retains
an ‘experimental’ element through the use of comparative designs, and it can generate a
large amount of data, while also allowing for in-depth analysis. For both of my studies, I
have collected data online using the Qualtrics survey software, which has the huge
out hours of transcribing time. The online environment grants participants maximum
anonymity and allows people across the world to be reached. For me the main challenge
of using such a novel method as qualitative SC has been having to explain to other people
why SC produces meaningful data. I have encountered questions and confusion from both
quantitative and qualitative researchers, as well as lay people. So if you choose to use SC
41
methods, you may encounter scepticism from other researchers. But this has really helped
me to develop clear arguments about why I think SC really is a very exciting and useful
42
Table 3.1: Examples of Existing Story Completion Research
framework
constructionist
constructions)
Eating Disorders How are ‘anorexic’ and ‘bulimic’ young women
43
Figure 1: An Example of a Story Map (from Braun and Clarke, 2013)
Initial Shock
Sarah’s parents are usually portrayed as being in a state of
shock and there is silence around the table as they try to
come to terms with this unexpected news
44