Chapter Three Focus groups in context

Contents
Locating focus groups – philosophical and methodological traditions 30
Which qualitative tradition? 32
Disciplinary adoptions and adaptions 36

Objectives
After reading this chapter, you should:

see how disciplinary and professional concerns have influenced the use of
focus groups;
have an understanding of where focus groups sit within epistemological
and ontological debates; and
have an appreciation of where focus groups are located with reference to
the major qualitative traditions.

Although it has been asserted that focus groups can only be properly
appreciated when viewed as a qualitative method, it is important to
acknowledge that qualitative research is, itself, characterized by disagreement
and debate between proponents of a variety of approaches. Each of these
overlapping, but separate, traditions has its own distinctive set of assumptions
– as to what are appropriate research questions; what constitutes ‘data’ or
knowledge; how to generate and set about analyzing data; and the use to
which findings should be put (Barbour, 1998).

In seeking to locate focus group research it is necessary to, firstly, take a step
back from these debates to examine the wider philosophical and
methodological context in which qualitative methods, in general, and focus
groups, in particular, have been embraced and developed. The various
qualitative traditions, their origins and propensity for generating and

49
theorizing about focus group data, are then explored. This is also helpful in
terms of understanding why particular formulations of focus group research
have found favour within the various disciplines – and, indeed, sub-
disciplines – that have embraced the method. The final part of this chapter
explores some of these usages and looks at how different disciplines may use
focus groups, in accordance with their position with reference to major
philosophical and methodological debates. However, there is no neat
typology to be outlined here, as disciplinary alignments themselves shift,
leading some to engage in what is a fleeting flirtation with focus groups,
while others invest considerable time and effort to develop a focus group
variant that can be harnessed to address their specific theoretical concerns.

Locating focus groups – philosophical and

methodological traditions
In order to meaningfully contextualize focus groups with reference to major
philosophical debates it is necessary to explore the concepts of
‘epistemology’ and ‘ontology’. ‘Epistemology’ refers to ‘what we regard as
knowledge or evidence of things in the social world’ (Mason, 1996, p. 13).
‘Ontology’ relates to our views about the social world and how to go about
studying it.

Many texts on qualitative research appeal to the notion of ‘paradigm wars’,

whereby ‘positivism’ and ‘interpretivism’ are juxtaposed with each other,
with qualitative research being located within this latter tradition. These
opposing approaches are also often referred to as ‘realism’ and
‘constructivism’. Whereas ‘positivism/realism’ has attempted to mirror
approaches used in the natural sciences, to allow for the establishment of a
set of universal laws, tested via a set of scientific rules of method,
‘interpretivism/constructivism’, instead, posits that there is no ultimate
objective reality, since all knowledge is socially constructed (meaning that
there are no universal laws to be established and/or tested). It is tempting, in
accordance with this typology, to make a distinction between the applied and
the theoretical orientation to focus group research, and to assign to the
‘applied camp’ all research that seeks to elicit ‘views’ or ‘attitudes’ and
which seeks to address practice-defined issues and to assign to the

50
‘theoretical camp’ all focus group studies that attempt to analyze what
happens during interaction and to relate such insights to disciplinary models
or theories. However, in practice, there are many applications that defy such a
simplistic classification. Not all research that uses focus groups in order to
elicit views of specific groups can be criticized as taking a mechanistic ‘back
door survey’ approach; nor do appeals to theoretical models or characteristics
of interaction necessarily protect against making erroneous assumptions
about the universality of beliefs and/or their consequences.

Rather than viewing ‘realism’ and ‘constructivism’ as two separate and

conflicting approaches, several commentators have, instead, explored the
middle ground between these two (often exaggerated) extremes. Maxwell
(2011) helpfully explains the mid-ground occupied by critical realism:

Critical realism combines realist ontology (the belief that there is a

real world that exists independently of our beliefs and
constructions) with a constructivist epistemology (the belief that
our knowledge of this world is inevitably our own construction,
created from a specific vantage point). (2011, p. 180)

This description is a much more helpful way of viewing many studies that
have employed focus groups, whether alongside other methods, or as a stand-
alone approach. There are a number of distinct, but overlapping, qualitative
traditions which occupy this mid-ground of ‘critical realism’, although some
applications may veer towards one or other end of this broad continuum
between extreme realism or constructivism. The way in which focus groups
are applied and qualitative traditions are invoked depends, ultimately, on the
research question being addressed, and the disciplinary home and
predisposition of the researcher or research team. Focus groups do not exist
independently of the context in which they are utilized and cannot, therefore,
be neatly assigned to any one of many – potentially contradictory –
qualitative approaches.

The multi-faceted antecedents of focus group usage mean that focus groups
have developed without being closely associated with any specific tradition,
so that they can be viewed as being ‘relatively agnostic in terms of the

51
methodologies attending them’ (Kidd and Parshall, 2000, p. 296). Although
this has sometimes led to something akin to a methodological free-for-all,
there are particular features of focus group discussions that lend themselves
to a qualitative approach and it is argued that it is only in the context of such
use that focus groups realize their full potential. Rather than opting, from the
outset, for a specific qualitative tradition, the researcher considering using
focus groups would be well-advised to carefully review the various options
on offer, paying close attention to their potential for generating data that is
‘fit for purpose’ (in terms of its content, focus and possibilities for analysis)
for the research project in hand.

Which qualitative tradition?

Located midway between observational fieldwork and one-to-one interviews,
focus groups have been described as involving ‘structured eavesdropping’
(Powney, 1988). This characterization of focus groups fits well with ‘critical
realism’ (Bhaskar, 1989; Hammersley, 1992). As outlined by Maxwell
(2011), this approach acknowledges both the role of the real world (and the
capacity of the researcher to draw on this in formulating research questions;
selecting study sites; sampling for focus groups; making use of and
disseminating findings) and the fluid and provisional nature of participants’
perspectives (as expressed, contested, justified or modified throughout
interaction and constituting the data to be analyzed).

There has been some spirited debate with regard to exactly where focus
groups fit on the continuum between structure and spontaneity. This depends,
in part, on how active the researcher is in directing discussion, and there has
been considerable disagreement about the relative merits of ‘researcher-
convened’ and ‘naturally occurring’ groups, with some commentators
viewing focus group research as a ‘poor relation’ of the established traditions
of ethnography and anthropology. According to some commentators, ‘If the
researcher “creates” a population, then the research is no longer ethnography’
(Brink and Edgecombe, 2003, p. 1028). This stance conveniently overlooks
certain elements of the practice both of ethnographers and anthropologists –
notably the careful selection of study sites that chime with the research
question or theoretical framework to be interrogated, and the attention paid to

52
identifying ‘key informants’ or ‘stakeholders’ and eliciting their insights,
often through the use of supplementary interviews.

Involving lively interaction between participants, focus groups provide an

opportunity to generate data that are amenable to analysis within the
symbolic interactionist approach, which emphasizes the active construction
of meaning. This approach, according to Blumer (1969), supposes:

that human society is made up of individuals who have selves (that

is, make indications to themselves); that individual action is a
construction and not a release, being built up by the individual
through noting and interpreting features of the situation in which he
acts; that group or collective action consists of the aligning of
individual actions, brought about by the individuals’ interpreting or
taking into account each other’s actions. (1969, p. 184; parenthesis
in original)

This was the approach developed by what has come to be hailed as the
‘Chicago School’ of sociologists. Working in the United States in the period
following the Second World War, they were committed to the idea of human
actions as arising through the active construction of meaning through
interaction in groups with significant others. It was through interaction that
concepts were interrogated, concerns aired, meanings conferred and
rationales for views and behaviour developed. Symbolic interactionism has
gone somewhat out of fashion in more recent years, having been supplanted
by an emphasis on ‘phenomenology’. Both, however, concentrate on the
process of interaction and active construction of meaning.

Viewing language as a form of social action (Burr, 1995) and paying even
closer attention to the sequence and structure of talk, conversation analysts
also view interaction as a site for study that affords the researcher access to
the construction of meaning and social action being performed. Puchta and
Potter (2004) explain:

Conversation analysts in particular have argued that ordinary talk,

53
 mundane talk, the kind of everyday chat we have with one another
is fundamental to understanding all kinds of more specialized
interaction … Talk is … something we use to perform an enormous
variety of the practical tasks of living. (2004, p. 9)

Symbolic interactionism and conversation analysis have both been subjected

to criticism for privileging ideas of agency (the capacity of individuals to
effect change and action) over structure (the wider context and constraints
that affect or limit the possibilities of action). That is, they have been
criticized for concentrating on the ‘micro’ to the exclusion of the ‘macro’ and
ignoring the important relationship between the two. Thus, such approaches
have sometimes been seen as affording detailed insights into the trivial
without the capacity to offer an explanation of how these processes impact on
society at a higher level than that of the small group. A good deal of the effort
of qualitative researchers is expended, therefore, on explaining ‘how the
realities of everyday life are accomplished [with] … the issue of why things
happen in the way they do [being] rarely addressed’ (Seale, 1999, p. 39).

I would argue that focus groups, if used judiciously, can effectively address
this important gap in understanding. Notwithstanding the many projects that
restrict analysis of focus group data to the purely descriptive, a more rigorous
and theoretically informed approach can aspire to providing an explanation.
However, this higher level of understanding does not just come about
magically through some inherent property of focus group discussions: for
focus groups to make the fullest contribution possible requires the thoughtful
and purposeful engagement of the researcher.

A broadly social constructionist approach (Berger and Luckmann, 1966)

holds most promise for bridging the explanatory gap identified by Seale,
since it allows the researcher to combine the micro attention to interaction
advocated by symbolic interactionism and more macro elements (taking into
account the social, economic, political and policy context) in which data are
being generated and which can usefully inform analysis, in order to make
comparisons, to contextualize, and to interrogate emergent explanations. This
is in line with the approach advocated by Gergen (1973), who highlighted
that phenomena are specific to a particular time, place and culture, arguing
for what he called a ‘historical social psychology’.

54
Research design – and, particularly, well-considered sampling strategies –
can render focus groups a particularly effective tool for interrogating the
relationship between agency and structure. According to Berger and
Luckmann (1966), the objective social world is mediated by significant
others who ‘modify [this world] in the course of mediating it. They select
aspects of it in accordance with their own location in the … social structure
and also by virtue of their individual, biographically rooted idiosyncrasies’
(Berger and Luckmann, 1966, p. 151). In addition to arguing that social
phenomena are collectively created by participants, Berger and Luckmann
also referred to social phenomena (the very focus of the qualitative research
endeavour) being sustained through social practice. Interaction in focus
groups constitutes one such form of social practice and focus groups,
therefore, afford a privileged vantage point from which to observe the
processes through which ideas, meanings and discourses are formulated,
contested, debated and modified

Returning to the earlier debate regarding structure and spontaneity, rather

than relying on naturally occurring debates, the focus group researcher can
actively focus discussion (through asking questions and presenting stimulus
materials) and can, further, interrogate these contributions – and, indeed,
emergent explanations (through active and responsive moderating). (More
advice on productive moderating is provided in later chapters.) This can
include paying attention to relevant discourses or analytical frameworks, both
in designing the research project and in the course of generating and
analyzing the data.

Other writers (such as Burr, 1995), exploring the potential of social

constructionism, have proceeded to emphasize the role of ideology in
linking the individual and group interaction processes to wider social
concerns and processes, thereby locating subjectivity in its social context.
Callaghan (2005) argues that focus groups can afford participants the
opportunity of simultaneously managing their individual identities and
making a collective representation to the researcher, thereby providing
valuable insights into the construction of meanings and their impact on
action. She further explains that ‘carefully selected focus groups, can access
knowledge which embodies the “habitus” of the wider community’. The term
‘habitus’ was coined by Bourdieu and refers to ‘dispositions’ or lenses

55
through which individuals view the world, which are ‘socially constituted’
and ‘acquired’ (Bourdieu, 1990). Bourdieu further elaborates on the
‘generative’, ‘creative’ and ‘inventive’ capacities of ‘habitus’ (Bourdieu,
1999), emphasizing the flexibility of the concept. According to Callaghan
(2005), the processes involved in this creative endeavour can be further
illuminated – again, through paying attention to structure – in the form of
strategic sampling, to allow the researcher to explore patterning in relation to
social and cultural categories, such as age, gender, ethnicity and social class.

The practice of discourse analysis – or, in some applications, critical

discourse analysis or Foucauldian discourse analysis (Willig, 2003;
Wodak, 2004), which pays particular attention to potentially conflicting
discourses of power – is congruent with the ‘historical social psychology’
approach outlined by Gergen (1973). It builds on the idea that discourse (as
language in use) intersects with and shapes social action and relationships.
While some discourse analysts concentrate on analyzing naturally-occurring
talk and/or documents, others have employed focus groups and have analyzed
the transcripts produced as text, again paying close attention both to the
language and concepts appealed to and the way in which the construction and
reinforcement of discourses are accomplished through interaction. As with
conversation analysis, close attention is paid to linguistic strategies and
features of talk. Such orientations stem from the origins of the various
disciplines, or sub-disciplines, that have adopted or adapted focus groups, and
the next section takes a brief look at the history of the different disciplines
that have employed focus groups. It looks, specifically, at methodological
and philosophical features of research traditions that render focus groups
more or less compatible with the aims, objectives, preoccupations and
political stance of both established and emergent research groupings.

Disciplinary adoptions and adaptions

The various disciplines that have shown themselves amenable to using focus
groups are located at different points on the ‘realist–
interpretivist/constructivist’ continuum. However, disciplines, themselves,
are in a constant state of flux, with new theoretical directions evolving, which
may lead to focus groups being viewed as more or less relevant. Even within

56
disciplines, the emphasis can vary considerably between research groupings –
particularly where interdisciplinary collaborations are involved. This,
certainly, also has an impact on the way in which focus groups are pressed
into service, the sort of data that are elicited and how they are analyzed.

Until relatively recently, focus groups were generally not used in mainstream
research, even at the more applied end of the spectrum, such as in the context
of evaluation, where researchers relied mostly on surveys and one-to-one
interviews. This was also the case for those working in disciplines such as
anthropology or sociology; although these latter groups frequently employed
observational fieldwork either alongside surveys or interviews, or as a stand-
alone approach. Over the past 20–25 years focus groups have enjoyed an
unprecedented rise in popularity, having been adopted by researchers with a
wide range of disciplinary backgrounds seeking to study an extremely diverse
set of topics. This expansion of focus group research has spanned both the
more theoretical end of the continuum (having been espoused by many
different disciplines) and the more practical end of the continuum.

At the more practical end of the spectrum, as noted earlier, focus groups have
been used extensively in the field of design, which has, itself, witnessed an
‘ethnographic turn’, resulting in focus groups being used in order to review
‘real world’ responses to design prototypes.

By far the most prolific use of focus groups – at least, in terms of the
resulting volume of published papers – has been made by health services
researchers, and this large body of work, as we have already seen, affords a
useful lens through which to explore a broad range of usages and to critically
examine assumptions and practices. There have, however, been several other
areas – namely research into environmental issues, social work research,
psychology, sociology and ethnography – where focus groups have been
adopted and, indeed, adapted, and researchers in these fields have made a
significant contribution in terms of expanding our understanding of the
potential of focus groups and extending the analytic possibilities at our
disposal.

Focus groups have found favour in a growing number of research fields, as

researchers discover the potential they afford for addressing their discipline-
specific theoretical concerns. The adoption of focus groups has, thus, been

57
associated with the growing acceptability of qualitative methods in fields
such as population geography (Skop, 2006), human geography (Crang, 2002),
psychology (Wilkinson, 2003), social work (Linhorst, 2002) and education
(Wilson, 1997). This new-found enthusiasm for focus groups has also chimed
with shifts in disciplinary ‘gazes’ to accommodate new ideas – such as what
has been referred to as ‘the discursive turn’ in organizational research (Grant
et al., 1998); the ‘ethnographic turn’ in design, or ‘the linguistic turn’ in
social work (Hitzler, 2011).

The use of focus groups in the context of researching contentious

contemporary topics, including environmental issues, looks back to the
origins of focus groups in seeking to access public opinion, but, here,
researchers have mostly engaged in questioning the orthodoxies involved in
opinion polling (Myers and Macnaghten, 1999). Such work frequently relies
on conversation or discourse analysis, recognizing the importance of the
context in which data are generated. An interesting development has involved
the adoption of focus groups by other constituencies with an interest in
sustainability and climate change, including natural scientists. One pan-
European project on public perceptions of carbon capture and storage (CCS)
(Upham and Roberts, 2011) held focus groups with members of the public in
six European countries. A research project in Canada used focus groups as
part of a wider programme ‘examining the benefits of introducing a range of
voices into ethical analysis and public debate about the appropriateness of a
novel technology (genomics) and a set of scientific practices’ (Tansey and
Burgess, 2008, p. 481). This elicited information on lay taxonomies.

A further application in a field outside of those traditionally associated with

focus group practice, involved a study of coastal and island communities and
interactions with the marine environment in the shape of seagrass meadows
(Cullen-Unsworth et al., 2013). Rather than simply documenting responses,
this approach allowed the researchers to provide a nuanced account of the
tensions involved. They explain:

Policy makers have taken note of the relationship between

ecological systems and social processes … It is well known that
coastal and island communities provide a clear example of the
interactions between societies and nature, where people both

58
 depend on the marine environment to provide their basic life needs
and where the marine environment is proving to be highly
vulnerable in the face of social pressures. (2013, online)

The disciplines of psychology, social work, education, business studies,

organizational research, and sociology have all been enthusiastic proponents
of focus groups, albeit using them in slightly different ways and to different
ends, depending on whether the ultimate focus is on practice-related concerns
or theoretically driven questions. Sociologists, for example, have been quick
to exploit the capacity of focus groups to illuminate processes of identity
formation, the workings of social capital and ‘habitus’ (e.g. Callaghan, 2005;
Munday, 2006) and such work is discussed in more detail later in this
chapter. Some of these applications span disciplinary boundaries – e.g.
Lindgren and Nelson’s (2014) study of identity construction of inter-country
adoptees, which is sociological in focus, but which simultaneously addresses
social work practitioners.

The use of focus groups within ethnographic studies has generally involved
employing the method alongside other approaches, such as one-to-one
interviews, observational fieldwork and/or document analysis – e.g. Powell’s
(2014) study of neighbourhood relations in an Appalachian community
bordering a university campus; or Espinoza and Piper’s (2014) study of
collective memory sites in Chile (which is discussed further in Chapter 4, in
relation to mixed methods research designs). The community development
tradition has also tended to use focus groups alongside other methods.
Although this approach might, at first sight, appear to resonate with the
anthropological research tradition, there are tensions between the two, as
Baker and Hinton (1999) acknowledge – not least because of the differing
emphasis put on effecting change.

Each discipline is likely to approach focus groups in a distinctive way, using

them for a variety of purposes that relate to the central theoretical concerns
that characterize each discipline, taking account of intra-disciplinary debates
and concerns and building on existing areas of expertise, such as group work
within social work (Cohen and Garrett, 1999). Other disciplines that have
explored the possibilities afforded by focus groups include occupational
therapy (Hollis et al., 2002), and paediatric health research (Heary and

59
Hennessy, 2002).

This has resulted in a situation where novice focus group researchers can be
confronted by a confusing array of advice that can, at times, be somewhat
prescriptive –and, perhaps, also, inappropriate for the research task at hand.
Some of this advice – particularly that directed at researchers working at the
more applied end of the continuum – treats focus groups as merely a set of
techniques, largely removed from the broader context of qualitative research.
Ignoring this context can give rise to problems as researchers struggle to
reconcile the depth and richness of the data they have generated with their,
often inappropriate, expectations.

Marketing texts can provide useful hints on encouraging reluctant

participants to talk and on selecting exercises to stimulate discussion.
However, advice about sampling should be treated with some caution (see
Chapter 5, which is devoted to the topic of sampling), as it is important to
bear in mind the very different purpose that underpins the marketing research
enterprise. Marketing research is big business and is frequently carried out on
a national scale, with the potential for convening many groups in different
locations over a very short period of time. Sampling depends on identifying
target markets for advertising and aims to recruit a sample that is broadly
representative of this target population. In this tradition, focus groups are
prized because of their capacity to provide up-to-the minute responses and
thus to anticipate market trends rather than their capacity to provide detailed
information of the sort generally required by health services researchers and
social scientists.

In contrast to marketing research or the more conventional approaches to

using focus groups to gauge public opinion, research concerned with
contentious issues, such as animal experimentation, environmental issues and
sustainability, has frequently used conversation analysis techniques and
draws extensively on theoretical frameworks in making sense of the data. The
degree of detail involved in the analysis is, of course, likely to depend on
who has commissioned the research and for what reasons. As Macnaghten
and Myers (2004) point out, the background to the project and the time-scale
determine many of the choices involved in using focus groups. (These and
related issues are discussed in more detail in Chapters 4 and 9.)

60
Professional and disciplinary focus and concerns have shaped the ways in
which focus groups have been developed and employed within different
professional and academic circles. Details of focus group applications vary,
depending on the nature of engagement with clients and those being
researched, the services provided, professional models used and theoretical
frameworks employed. Differences also relate to the sort of research
questions posed, the content of topic guides, the questioning style of the
moderator, the approach to data analysis, the way in which findings are
presented and the use to which findings are put. Usage also differs according
to the extent to which interaction itself, or group work, is central to the
practice of a profession or theoretical frameworks – as is the nature of
involvement with the wider society, including funding sources and
government bodies.

Much time and energy has been devoted by researchers to seeking advice
from texts produced by these various traditions, but, I would argue, that they
have frequently become caught up in some of the internal debates within
these specific disciplines and have not always sifted through these as
critically as they might have; selecting what fits their own study and purpose
and rejecting what does not.

Key points
Focus groups are a qualitative research method.
They can be used both in applied and in more theoretically focused
projects.
Most usages are located somewhere in the middle of the realist–
constructivist continuum.
Focus groups are amenable to being used within any of the existing
qualitative traditions, although this may result in variation in terms of
how they are used and, especially, how data are analyzed.
For the new user, or one without a specific disciplinary or theoretical
‘home’, a broadly ‘social constructionist’ approach is likely to be most
helpful.
Focus groups can be harnessed in order to address issues specific to your
particular discipline.

61
Each discipline is likely to approach focus groups in a distinctive way.
The extent to which a focus group project can reflect disciplinary and
theoretical concerns depends on the aims, funding source and timetable.

Further reading

These readings will enable you to meaningfully locate your own focus
group project within the various traditions and usages:

Barbour, R.S. (2014) ‘The scope and contribution of qualitative research’,

Chapter 1 in Introducing Qualitative Research: A Student’s Guide.
London: Sage, pp. 11–27.

Barbour, R.S. (2014) ‘Qualitative traditions: epistemology and ontology’,

Chapter 2 in Introducing Qualitative Research: A Student’s Guide.
London: Sage, pp. 28–45.

Maxwell, J.A. (2011) A Realist Approach for Qualitative Research.

Thousand Oaks, CA: Sage.

These readings look at use of focus groups in specific disciplinary

contexts:

Belanger, F. (2012) ‘Theorizing in information systems research using

focus groups’, Australasian Journal of Information Systems, 17 (2): 109–
35.

Burgess, S. (2010) ‘The use of focus groups in information systems

research’, International Journal of Interdisciplinary Social Sciences, 5 (2):
57–68.

Küster, I. and Vila, N. (2011) ‘Successful SME web design through

consumer focus groups’, International Journal of Quality and Reliability
Management, 28 (2): 132–54.

Jayasekara, R.S. (2012) ‘Focus groups in nursing research: methodological

perspectives’, Nursing Outlook, 60 (6): 411–16.

Lindhorst, D.M. (2002) ‘A review of the use and potential of focus groups
in social work research’, Qualitative Social Work, 1 (2): 208–28.

62
Macnaghten, P. and Myers, G. (2004) ‘Focus groups’, in C. Seale, G.
Gobo, J.F. Gubrium and D. Silverman (eds), Qualitative Research
Practice. London: Sage, pp. 65–79.

O’hEocha, C., Conboy, K. and Wang, X. (2010) ‘Using focus groups in

studies of ISD team behaviour’, Electronic Journal of Business Research
Methods, 8 (2): 119–31.

Skop, E. (2006) ‘The methodological potential of focus groups in

population geography’, Population, Space and Place, 12: 113–24.

Stalmeijer, R.E., Mcnaughton, N. and van Mook, W.N.K.A. (2014) ‘Using

focus groups in medical education research’, Medical Teacher, 36 (11):
923–39.

Webb, C. and Doman, M. (2008) ‘Conducting focus groups: experiences

from nursing research’, Junctures: The Journal for Thematic Dialogue, 10.
www.junctures.org/junctures/index.php/junctures/article/view/49/394.

Wilkinson, S. (2003) ‘Focus groups’, in J.A. Smith (ed.), Qualitative

Psychology: A Practical Guide to Research Methods. Thousand Oaks, CA:
Sage, pp. 184–204.

Wilson, V. (1997) ‘Focus groups: a useful qualitative method for

educational research?’, British Educational Research Journal, 23 (2): 209–
24.

63
Chapter Four Research design

Contents
Mixed methods designs 45
Deciding whether to use one-to-one interviews or focus groups 48
One-off or repeat focus groups? 50
Teleconferencing and video-conferencing 51
Online focus groups 52
Mixing qualitative methods 55
Triangulation 56
Research design practicalities 58

Objectives
After reading this chapter, you should:

understand the rationale behind deciding between interviews and focus

groups;
be able to make an informed decision as to whether to use one-off or
repeat focus groups;
be equipped to weigh up the potential of tele- or video-conferencing and
online focus groups;
see the advantages and problems of using focus groups in mixed methods
approaches and in triangulation; and
have an appreciation of the importance of attending to practicalities at the
outset.

This chapter provides guidance with regard to making choices about research
design (see Flick, 2018c, for more details) and outlines the different options
available, evaluating both their potential and disadvantages. As we have seen
in the previous chapter, researchers are likely to derive most benefit from
focus groups when they appreciate the full range of capabilities of this

64
versatile qualitative approach. However, focus groups can also make a
valuable contribution to essentially quantitative studies, as is evidenced by
their use in order to develop and test instruments. There are, however, other
possibilities, including the use of focus groups at a later stage in mixed
methods studies, in order to explore apparently anomalous findings or simply
to further investigate intriguing findings or associations.

The chapter also provides advice on how to weigh up possible alternative

research designs, whether using mixed methods sequentially, or in parallel,
and how to decide which approach is most appropriate for your research
project: whether this is a mixture of quantitative and qualitative methods or a
combination of purely qualitative methods (with, for example, interviews,
observational fieldwork or document analysis being used alongside focus
groups). It also evaluates the potential for employing focus groups as a stand-
alone method and the debates surrounding the appropriateness of focus
groups as opposed to one-to-one interviews.

Both the strengths and weaknesses of mixed methods designs are critically
examined, as are claims regarding ‘triangulation’. It is argued that a
combination of methods produces parallel data, which should be used to
illuminate differences in focus or emphasis, rather than being prized for their
capacity to corroborate findings produced using various methods of
generating data. Again, the capacity of focus groups to facilitate comparison
and afford insights that would not be afforded by other methods is seen as
their main contribution.

Even once the researcher has opted to use focus groups (either alone or
alongside other methods) there are further questions to be considered. There
are decisions to be made as to whether face-to-face, telephone, teleconference
or online groups are to be held, or whether a combination of these approaches
might be most appropriate. As well as convening ‘one-off’ focus groups,
repeat sessions offer further possibilities, which may be especially useful
where an action research approach is being followed. Recent years have seen
the development of a range of new approaches to generating data, involving
the increased use of visual methods, ‘photo-voice’ and social media
platforms, and the researcher should carefully weigh up the advantages and
disadvantages of such approaches, whether used alone or in combination with

65
each other. However, there have also been interesting developments,
including novel approaches to using longer established methods (such as
ranking exercises – see the example provided by Demant and Ravn, 2010,
which is presented later in the book in Chapter 6, which considers how to go
about generating focus group data).

The choice of research design depends, ultimately, on the research question,

the depth of data sought and the detail of analysis that is envisaged. However,
paying attention to the practicalities involved is also important and the
chapter concludes by providing some tips with regard to selecting venues for
focus groups, matching moderator and participants, and timetabling. Finally,
the often-overlooked potential of dissemination sessions is explored, and it is
argued that these can afford a valuable route to generating further data, since
they allow the researcher to focus down on specific issues.

Mixed methods designs

Focus groups are frequently employed during the exploratory phase of large-
scale quantitative studies. Some researchers, working within a predominantly
quantitative paradigm, have used focus groups in order to validate
measurement instruments for use in a clinical setting (e.g. Kirchberger et al.,
2009). Such work has generally been carried out as preliminary work for a
bigger study. Here focus groups can make a significant contribution in
accessing the views of non-scientists and in terms of broadening researchers’
frames of reference – for example, Kirchberger et al. (2009) report that the
focus groups that they held in order to validate the instrument from the
patient’s perspective identified 47 additional categories for incorporation into
the International Classification of Functioning, Disability and Health (ICF)
core set for diabetes mellitus.

Focus groups have also been used to advantage in developing Quality of Life
(QoL) tools and Patient Related Outcome Measures (PROMs); both of which
areas have frequently involved collaboration between clinical researchers and
social scientists. Den Oudsten et al. (2011) convened focus groups with
people with Parkinson’s disease, their family and health care professionals,
alongside an extensive international QoL collaboration, carried out under the
auspices of the World Health Organization. Capitalizing on the potential

66
afforded by focus groups to access and elaborate cultural norms and
expectations, Zeldenryk et al. (2013) carried out focus groups in order to
develop a culturally relevant QoL tool for use in Bangladesh (in relation to a
previously somewhat neglected tropical disease). More recently, health
services research has turned its attention to PROMs and, here, also, focus
groups have been used within a mixed methods study with the aim of
developing a new measure for use in studying stroke self-management
(Boger et al., 2015).

Expert focus groups have sometimes been employed in order to address

professional concerns, such as the recognition and response to dementia in
several European countries, as studied by Iliffe et al. (2005) who convened
multinational expert focus groups. Within health services research there has
also been a tradition of the use of Delphi methods, which usually involve the
use of an expert panel, who comment (sometimes in focus group
discussions) on results obtained using other methods – most frequently a
survey (Kitzinger and Barbour, 1999). However, as with many usages
building on the community development model, this often forms part of the
grey literature and there are very few published accounts that discuss, in
detail, the use to which focus groups are put within this approach. A notable
exception here is the report provided by Neale et al. (2014) who also
capitalize on the potential of the Internet for carrying out such work and, in
particular, the advantages it offers for recruiting the required number and
spread of ‘experts’ (Bloor et al., 2013). In this case, the objective of the study
was to determine the extent of agreement between experts about how to
measure recovery from addiction. This project involved three iterative rounds
of data collection and sought to capture the perspectives of addiction
psychiatrists, senior staff from residential drug and alcohol services, and
senior staff from inpatient detoxification units. Brüggen and Willems (2009),
explain the principles of ‘e-Delphi’, where the discussion leader is involved
in aggregating and analyzing the initial responses, before sending this
summary to online participants in order to stimulate further discussion. This
follows an approach very similar to that employed in the original formulation
of Delphi studies.

Bloor et al. (2013) contend that Delphi groups are particularly well-suited to
policy research. However, such groups are likely, they argue, to provide a

67
useful additional exercise and they caution that ‘panellists’ agreement does
not [automatically] equate to either public participation or member
validation’ (2013, p. 67). Again, purposive sampling is important in order to
ensure a spread of ‘experts’ and it is important to include individuals with
policy expertise.

Another established approach to embedding focus groups in mixed methods

studies involves their use – particularly in health services research – in order
to develop a questionnaire (e.g. O’Brien, 1993). In a study focusing on
professional responses to women with psychiatric problems, where issues
relating to child protection were also a factor, we opted to use multi-
professional focus groups in order to assist with developing a questionnaire.
Although we concentrated on refining the wording of two particular
questions that were presenting us with problems, the ensuing focus group
discussions suggested that professionals from different professional
backgrounds were employing varying criteria in order to determine the
‘riskiness’ of a situation (Barbour et al., 2002). This, in turn, led us to
develop and incorporate in our questionnaire, a set of hypothetical scenarios,
designed specifically to explore this issue (Stanley et al., 2003).

O’Neill and Nicholson-Cole (2009) combined focus groups with interviews

and Q-methodology (which represents an attempt to quantify qualitative
data). Here the focus groups allowed participants to discuss and elaborate on
prior findings relating to engagement with climate change via visual and
iconic representations. The comparative potential afforded by this research
design highlighted an interesting anomaly: namely, that ‘the very images that
made participants have the greatest sense of climate change being important
were also disempowering at a personal level’ (2009, p. 373).

Despite the considerable potential for focus groups to be employed in order

to illuminate surprising or anomalous findings from quantitative studies
(Barbour, 1999b), this mixed methods usage is still much less common. Box
4.1 provides one of the few examples of focus groups being used to this
potentially very useful end.

Box 4.1 An example of imaginative use of mixed methods

68
 Wilmot and Ratcliffe (2002) report on their experience of using focus
groups to illuminate survey findings. Their study related to principles of
distributive justice used by members of the public with regard to the
allocation of donor liver grafts for transplantation. In common with other
studies in this area, quantitative data had been collected by survey, which
had used hypothetical choice contexts in order to investigate informants’
preferences with respect to the allocation of donor organs. However,
Wilmot and Ratcliffe acknowledged the limitations of such data, which do
not ‘allow the investigator to identify the way informants explain and
justify their particular choices’ (2002, p. 201). Through focus group
discussion they sought to provide an in-depth understanding of the
arguments and explanations used in ‘determining and justifying allocation
decisions and the ethical and moral arguments expressed’ (2002, p. 201).

Drawing on a list of patient criteria (expected prognosis following the

operation; age of the patient; patient’s responsibility for their illness;
length of time on the waiting list; and whether the patient is being
transplanted for the first time or is being re-transplanted) shown by
quantitative research to be significant factors in determining public
attitudes to donor allocation, these researchers designed five hypothetical
scenarios, which were used to generate discussion in focus groups.
Following this the focus group members were provided with further
information concerning the social background of the hypothetical
individuals presented in order to explore the impact of additional
circumstantial information on their responses. The findings highlighted
that the relationship between participants’ reasoning and the three main
principles of equity, efficiency/utility and desert was more complex than
anticipated. Although they were more receptive to some criteria than to
others, they identified difficulties in applying each of the criteria studied.
The study provided insights into how members of the public engaged
thoughtfully and flexibly with the criteria.

Deciding whether to use one-to-one interviews

or focus groups
There are no hard-and-fast rules that determine whether focus groups or one-
to-one interviews are most appropriate and, once again, the answer lies in
carefully weighing up the pros and cons in relation to each new project. Some

69
respondents, if given the choice, will say that they feel more comfortable
talking to a researcher one-to-one and would be reluctant to attend a group
session. For others, however, there may be safety in numbers and coming to a
focus group discussion may allay concerns of some individuals that they may
not have ‘anything of interest’ to contribute to the research. Focus groups
may also be an attractive option for those who are otherwise isolated, or who
crave the opportunity to talk to other people in the same situation as
themselves – especially when there are no relevant support groups available.
Some researchers, such as Brooks (2014) (who carried out research into
young women and safety), recognizing that participants may have marked
preferences, have sensibly opted to give them the choice as to whether they
would like to take part in focus groups or one-to-one interviews. Sometimes,
however, there may be compelling ethical considerations, which mitigate
against bringing individuals together in focus groups. For example, it could
be distressing for someone newly diagnosed with a debilitating condition to
be brought face-to-face with others at a more advanced stage in the disease,
and who are already showing a marked decline in physical or cognitive
functioning. When bringing together existing peer groups or work teams, it is
also important to remember that these relationships are likely to continue
beyond the research project and this may suggest that, in some circumstances,
one-to-one interviews might be a safer option. In other situations, the
researcher would be well-advised to ponder the implications of including in
focus groups individuals who are known for holding particularly strong
views.

When carrying out research with professionals on the topic of advance

directives, or living wills, we (Thompson et al., 2003a, 2003b) decided to
interview some individuals who were prominent advocates rather than to
invite them to take part in focus groups. While including such people in the
group discussions would undoubtedly have stimulated debate, it was highly
likely that the contribution of zealous individuals would overshadow that of
others and that some participants might feel intimidated with regard to
expressing their own views, which were probably not so clear-cut or well-
rehearsed. A combination of shift work and family commitments also meant
that some other individuals could not attend focus groups and so interviews
were carried out instead. This illustrates the need to take into account both
ethical issues and practicalities and to carefully consider what you are hoping

70
to achieve through using focus groups or one-to-one interviews – to visualize
the likely style and content of the exchange. This will help you to decide
which is the most appropriate method. Here it is important not to be
intimidated by other researchers’ choices. Simply because others have
favoured one-to-one interviews does not mean that focus groups are
inappropriate; indeed, using a different method may allow you to make an
original contribution to the knowledge base of your discipline, through
highlighting previously unexplored aspects of the issue at hand by, for
example, unpicking the reasoning behind certain types of behaviour or
beliefs.

In weighing up whether to use one-to-one interviews or focus groups, it is

important to remember that, although the two methods are close cousins and
share many epistemological assumptions, focus groups elicit data that are
also different in content from that generated by one-to-one interviews. This
calls for slightly different skills in terms of analyzing focus group data – in
particular it is important that researchers move away from the tendency to
focus on individual voices to the exclusion of paying close attention to the
interaction. (This is discussed more fully in Chapter 8, which addresses the
topic of analysis.)

In summary, there are no universal guiding principles, save the exhortation to

weigh up the pros and cons of focus groups and one-to-one interviews for
each new project and context (see also, Flick, 2018c, 2018d; Brinkmann and
Kvale, 2018). Crabtree et al. (1993) summarize:

the choice of research style for a particular project depends on the

overarching aim of the research, the specific analysis objective and
its associated research question, the preferred paradigm, the degree
of desired research control, the level of investigator intervention,
the available resources, the time frame, and aesthetics. (1993, pp.
139–40)

One-off or repeat focus groups?

71
Perhaps because much of the discussion about the choice of qualitative
methods has revolved around the relative merits of interviews and focus
groups, there has been relatively little consideration given to the potential of
repeat focus groups. Given the potential of qualitative methods – and focus
groups, in particular – to illuminate process, it may make sense to reconvene
focus groups in order to chart developments. This was the approach taken by
Whitley and Campbell (2014), who combined observational fieldwork with
4-monthly focus group sessions over a 5-year period, in order to study the
process of recovery from mental illness in three recovery communities.

A further possible reason for the apparent dearth of longitudinal research

designs involving focus groups is advanced by Walton (2009) who suggests
that there may be few published accounts of such work, because this is
considered part of the general process of consciousness-raising or
empowerment. One such action-research-focused study (Bronstein and
Mason, 2013), however, did report on the experience of using serial focus
groups: this involved the study of a civic service project looking at the
experience of senior volunteers and not-for-profit agency staff. The
researchers concluded:

The concurrent design allowed the researchers to hear the concerns

from both sides, the volunteers and the non-profit staff, within a
time frame that gave them the opportunity to take action. (2013, p.
82)

Two other examples of the use of repeat focus groups are provided by de
Oliveira (2011) who held a series of 12 weekly meetings with two groups of
adolescent girls in Southern Brazil, in order to explore the potentially difficult
topic of sexual risk-taking. The repeated meetings capitalized on the girls’
shared sociocultural background and allowed trust to be built up over time.
Similarly, Järvinen and Demant (2011) held a number of focus groups with
the same young Danish participants, arranged to take place when they were
14–15 years old, 15–16 years old and when they were 18–19 years old. This
study also addressed a potentially sensitive issue – that of drug use. This
afforded the researchers a useful vantage point from which to explore the
ways in which their focus group participants related to each other and carried

72
out what the authors term ‘symbolic boundary work’ in making sense of and
commenting on each others’ perceptions and behaviour. Again, the
researchers gained additional insights through using existing (mutually
trusting) peer groups over time. As this discussion highlights, these latter two
studies were concerned to generate textured interactional data, which they
then analyzed in-depth, looking in detail at the meanings being applied and
created and at the linguistic strategies employed by participants. Situations
where participants are well-known to each other are more likely to provide
data that are amenable to such micro-sociological analysis (as is highlighted
by Demant and Järvinen, 2011). They explain:

Risk conceptions and risk behaviours (here drug use) is not, in our
understanding, the product of individual decisions and cognitions
alone, but first and foremost a ‘socially interactive enterprise’ …
Young people experiment with drugs because their social networks
do the perceptual coding on risk … and because they, as part of a
specific peer network, come to hold attitudes and conceptions that
encourage drug use. (2011, p.180)

The research design employed here afforded privileged access to these

processes and allowed the researchers to study these fine-grained
interchanges in a way that is seldom possible.

Teleconferencing and video-conferencing

Although sometimes viewed as a forerunner of the more sophisticated use of
the Internet for conducting ‘virtual’ focus groups, telephone focus groups are
still worth considering in certain situations – not least because they may raise
fewer logistical challenges for the researcher. Allen (2014) found telephone
focus groups useful in her research with early childhood mental health
consultants throughout rural and urban Oregon and rural Alaska. Gratton and
O’Donnell (2011) have also highlighted the usefulness of video-conferencing
for accessing remote First Nations communities in Canada.

I have used telephone conferencing to access senior members of professional

73
groups who were geographically dispersed. One interesting aspect of this
latter usage was that individuals were less likely to dominate the discussion
than in face-to-face groups, perhaps because, in the absence of visual contact,
they cannot rely on signifiers of status and body language to stake their claim
for preferential treatment regarding ‘air-time’. Capitalizing on the potential of
focus groups to address sensitive issues, telephone conferencing has also
been used successfully to discuss, for example, the experiences of families
involved in organ donation (Regan, 2003). However, Regan advises that
additional time be allocated to the preparation of a virtual environment that is
conducive to discussing sensitive topics.

Frazier et al. (2010) reflect on their experience of carrying out telephone and
face-to-face focus groups with women with ovarian, endometrial or invasive
cervical cancer. In common with other researchers who have employed
telephone focus groups, they reflect that the telephone option facilitated
recruitment of rural participants and those living in distant cities, who would
otherwise not have been able to attend face-to-face groups. Interestingly, they
note that certain topics, such as sexuality and unconquered fear, were raised
only in telephone groups and did not feature in face-to-face discussions.
However, this is not to suggest that these topics can only be explored via
telephone: once the moderator had been sensitized to these issues, these
could, perhaps, have been raised in further face-to-face sessions. In this
situation it might also have been possible to utilize, as stimulus material,
some of the (unattributed) comments made during telephone focus groups, in
order to give both moderator and participants ‘permission’ to raise these
topics (see further discussion in Chapter 6 on generating focus group data).

Online focus groups

Over the past ten years or so, many researchers have been alerted to the
potential afforded by the Internet, to conduct ‘virtual’ focus groups, whether
this involves the researcher in active moderating of discussion or in
‘harvesting’ naturally-occurring materials posted on web-based discussion
sites. Online focus groups can offer economies in terms of recruitment, travel
costs and transcribing (as online discussions come ready transcribed). They
can also be used simply to furnish a sampling pool for more conventional

74
qualitative research (Stewart and Williams, 2005). Online discussions can
also dispense with some of the problems associated in matching moderators
and groups, since the gender (Campbell et al., 2001) and age of the moderator
need not be made explicit and hence do not have an impact on the data
generated (although, of course, participants will make their own assumptions,
perhaps based on the language used and style of responses). Bloor et al.
(2001) highlight the immediacy and collapsed spatial distance offered by
online methods.

As with telephone focus groups, the Internet is particularly useful when

researching remote populations, or those engaged in criminal activities
(Underhill and Olmsted, 2003), researching potentially ‘difficult’ topics’, or
engaging with any groups that might be reluctant to take place in face-to-face
focus group discussions. For example, Fox et al. (2007) used online focus
groups to carry out research with young people with potentially stigmatizing
skin conditions. Nicholas et al. (2010) convened online groups with children
with cerebral palsy, spina bifida and cystic fibrosis, and Tates et al. (2009)
carried out online focus groups with paediatric oncology patients. Children
and adolescents born with only one arm were also recruited for online
discussions by de Jong et al. (2012). Sometimes potential participants present
a recruitment challenge due to a combination of attributes. Ybarra et al.
(2014) used online focus groups with gay and bisexual male adolescents,
noting that this provided ‘the opportunity to speak to other sexually
inexperienced gay, bisexual and queer young men in a safe and anonymous
environment’ and that this ‘helped reduce their feelings of isolation and
increased feelings of support for their decisions regarding abstinence’ (2014,
p. 561), thus acting as a valuable adjunct to an HIV prevention programme.

An important decision relates to whether to employ, as did Fox et al. (2007),

synchronous online discussion (which unfolds in ‘real time’) or asynchronous
formats (which involve a delay between participants posting comments and
responses from others – whether other participants or the moderator). A
particular challenge of the online format is that it reduces the ability of the
moderator to guide the discussion and request elaborations and this may
require enhanced moderator skills (Stewart and Williams, 2005). Bloor et al.
(2001), however, point out that – as opposed to ‘real-time’ exchanges, which
can be fast and furious – asynchronous discussions are much easier to

75
moderate (although they do not give quite the same immediacy).
Asynchronous discussions involve what Burns (2010) has described as a
‘much slower dance of communication’ (para. 13.1). This does potentially
allow participants to reflect on their statements and to ensure that meaning
has been adequately captured (Nicholas et al., 2010) although some
researchers – depending, again, on their disciplinary leanings – may favour
spontaneity over accuracy. The asynchronous format also allows the
researcher to think carefully about which further questions to pose and how
to word these, and Bampton and Cowton (2002) have argued that it may even
be possible to conduct asynchronous online interviews in a language in which
the moderator is not fluent, which suggests that this may also be the case with
asynchronous online focus groups.

Where researchers rely on established chat rooms to ‘harvest’ pre-existing

discussion material for research purposes, they lose control over the amount
of background information they can collect in order to contextualize
responses (Bloor et al., 2001, p. 78). Alinejad (2011) used blogging sites in
order to access second generation Iranians living in the USA and Canada. She
reasoned that such individuals were particularly likely to suffer dislocation
and were likely to use this medium as a way of sharing their experiences with
each other. Such sites, however, are not designed with the researcher in mind
and the researcher may be hampered by the lack of background information
about participants. Using existing websites to furnish data – either without
any intervention, or minimal participation, from the researcher – begs the
question as to whether these pieces of research qualify as ‘focus group
studies’ or whether they are best regarded as a particularly vivid form of
document analysis, which also recognizes that documents are not inert (Prior,
2008) but, rather, are living and evolving entities.

‘Harvesting’ online data also raises important ethical considerations (Robson

and Robson, 1999). Stewart and Williams (2005) highlight the complications
concerning data storage and anonymizing, since the original data are
automatically available to all discussants. This means, in turn, that
individuals can, at least in theory, be identified by others through deductive
disclosure. Stewart and Williams (2005) also discuss the implications, for
preserving anonymity, of using web-boards that require discussants to
complete a registration process. As Zhao et al. (2008) remind us, ‘it is

76
incorrect to think that the online world and the offline world are two separate
worlds’ (2008, p. 1831).

Online software offers additional features that can be used to advantage when
employing synchronous methods – such as the ability to divide the screen to
allow the discussion leader to speak privately with a participant (Brüggen and
Willems, 2009). However, this requires particular skills on the part of the
moderator and some researchers have opted to employ the services of
specialist companies, both to afford access to software and, also, on occasion,
to run the online focus group sessions, as did Hillier et al. (2012), who used
an international survey firm that had developed a proprietary software
program which they judged suitable for use with the young people they
wanted to access.

As with the other choices and decisions debated here, there is no definitive
answer as to whether online methods are superior to traditional face-to-face
approaches to generating data. Bloor et al. (2001) provide the following
measured summary of the potential of virtual focus groups:

Virtual focus groups are not the future of focus group research …
However, virtual focus groups do offer a useful stablemate in the
focus group tradition, and a worthwhile new tool for the social
researcher. (2001, p. 75)

Some researchers (e.g. Nicholas et al., 2010; Krol et al., 2014) have
combined offline and online focus groups and this may, in some cases, offer a
useful compromise, drawing on the rationale for mixed methods designs that
acknowledges the shortcomings of any method employed on its own and
which seeks to compensate through using parallel approaches (Kelle, 2006).
Online discussions, importantly, may broaden the reach of the research:
Nicholas et al. (2010) report that children (with cerebral palsy, spina bifida or
cystic fibrosis) who were ‘shy or hesitant to be in the spotlight’ expressed a
strong preference for the online format, whereas others, who were more
outgoing, appreciated the opportunity to discuss their experiences face-to-
face with their peers.

77
However, there may be important differences in terms of the form and
content of the data generated. Nicholas et al. (2010) found that participants in
their asynchronous online discussions offered substantially less information
and gained less understanding about the experiences of peers (which
potentially has implications for the use of asynchronous online methods
where the aim is to raise awareness or mobilize action). The very different
type of data generated can also raise challenges for analysis, since a
combination of online and offline focus groups essentially constitutes a
mixed methods approach – albeit within the qualitative paradigm.

Mixing qualitative methods

Some approaches draw on qualitative methods that are highly compatible,
such as the participant observation and repeat focus groups employed by
Whitley and Campbell (2014) – which, incidentally, are not that far removed
from therapeutic interventions practised in mental health settings (where their
research took place). Espinoza and Piper (2014) developed a hybrid method,
which they called ‘dialogical accompaniment’, combining an
anthropological/ethnographic approach with interactive group discussion.
This involved the researchers in selecting both sites and participants (which is
discussed in more detail in Chapter 5 on sampling) and the carrying out of
relatively unstructured observation of naturally-occurring, but focused – and,
indeed, videoed – dialogue, which was then subjected to detailed discourse
analysis. They argue that this allowed them to transcend normal research
practice, as they were able to focus on ‘the double interaction between the
participants and the space, and the interaction among themselves during the
development of the group’ (2014, p. 19). (This study is also discussed further
in Chapter 6 on generating focus group data.)

Most examples of the use of focus groups within a qualitative mixed methods
design are associated with broadly ethnographic approaches, where focus
groups simply afford another method that extends possibilities in an arena
where mixed methods are well-established and relatively unproblematic. An
example of such usage is provided by the ethnographic study of
neighbourhood relations in an Appalachian town adjacent to a university
campus (Powell, 2014), where focus groups were employed alongside semi-

78
structured interviews and ‘photo-voice’. (This latter method is further
discussed in Chapter 6 on generating focus group data.) Action research
initiatives are another area of practice where mixed methods have an
established pedigree, often involving case study approaches. Nilvarangkul et
al. (2013) also worked alongside government stakeholders – in the context of
an action research project aiming to strengthen self-care among women
weavers in North-East Thailand – using focus groups, interviews and
participant observation augmented by ongoing stakeholder meetings over a 5-
year period.

While it might be assumed that there would be less friction between broadly
similar qualitative methods in terms of epistemological and ontological
underpinnings, this can, nevertheless, mask important differences, which may
not come to light until the stage of data analysis. Different methods generate
data that are quite varied in terms of both form and content and this can
provide a further challenge for analysis – particularly for researchers who are
not used to inter-disciplinary or mixed methods approaches. Again, mixed
methods may be conceived of largely with reference to their use within a
quantitative paradigm, and there have been some spirited debates surrounding
the issues of ‘triangulation’.

Triangulation
A reason frequently advanced – at least in grant proposals – for employing a
mixed methods design is the goal of ‘triangulation’ (see also Flick, 2018b).
However, this is fraught with difficulties, even when working exclusively
within either the quantitative or qualitative tradition (Barbour, 1998, 2001).
The idea behind ‘triangulation’ is that data produced through applying
different methods can be compared in order to confirm or disconfirm each
other’s results. The problem, however, arises with regard to how to explain
discrepancies or contradiction. The notion of ‘triangulation’ – borrowed from
navigation and surveying – relies on the idea of a fixed point of reference,
involving a hierarchy of evidence, and assumes agreement between
researchers as to which method is accorded most status in terms of producing
the most ‘authentic’ or trustworthy findings.

Interestingly, within the qualitative paradigm, this status of ‘gold standard’

79
tends to have been accorded to one-to-one interviews (Silverman, 1993),
against which the data produced by focus groups are generally compared.
Rather than becoming caught up in irresolvable debate as to which dataset is
most ‘authentic’, it is helpful to view focus groups and one-to-one interviews
– or, indeed, any other forms of qualitative or quantitative data collection – as
producing parallel datasets. Such an approach allows the researcher to
capitalize on the comparative potential of various datasets, rather than being
caught up in attempts to establish a hierarchy of evidence.

While the concern for quantitative researchers appealing to ‘triangulation’ is

to corroborate or confirm results produced using different methods,
qualitative research thrives analytically on differences and discrepancies. It is
through focusing on these that we can benefit most from comparing data
from parallel datasets. Rather than agonizing about contradictory findings as
a problem, we should be engaged in using these as a resource. As Morgan
(1993) argues, ‘if research finds differences between the results from
individual and group interviews, then the methodological goal should be to
understand the sources of these differences’ (1993, p. 232; my emphasis).

Focus groups provide insights into public discourses (Kitzinger, 1994) and
the views expressed in focus groups may, of course, be different from the
‘private’ views that would be expressed in one-to-one interviews. Michell
(1999) compared ‘public’ and ‘private’ accounts of young people’s
experiences of their social worlds produced through interviews and focus
groups and interrogated differences, using the two datasets to afford
alternative lenses through which to look at the issue in hand. She utilized
comparison of parallel data to explore experiences of the hierarchical
structure of peer groups in school and in the neighbourhood. She highlights
the ‘added value’ of using these two complementary methods to provide
insights into both the process and experiences of bullying and victimization.

This is the approach favoured by Richardson (1994), who argues for the use
of the term ‘crystallization’ in preference to ‘triangulation’. She prefers this
imagery, she explains, because it emphasizes the value of looking
simultaneously at the same issue or concept from a variety of different
angles. Qualitative methods are especially adept at capturing the multiple
voices of different actors engaged in some aspect of social behaviour (e.g.

80
patients, carers and professionals). If we are intrigued rather than worried
when accounts from these various ‘players’ illuminate the very different
situations in which they find themselves and the different concerns they bring
to bear in discussing topics, then why should we react any differently when
complementary methods produce additional insights?

Bristowe et al. (2012) assigned interprofessional groups to four teams who

watched the unfolding of simulated video-recorded obstetric emergencies.
Rather than being concerned about disagreement between the teams, the
researchers (together with the actors involved in the exercise) assessed
responses and identified for further analysis areas of ‘convergence’,
‘complementarity’, ‘dissonance’ and ‘silence’. While this study employed the
same method (in this case focus groups) throughout, it, nevertheless, suggests
a possible template that might be employed for exploring and systematically
comparing the responses produced using different methods.

As well as thinking about how to use complementary methods to advantage

in ensuring that important voices are not muted in our research endeavours,
giving careful thought to selecting our methods also gives us an opportunity
to anticipate analysis. If we see complementary methods as producing
parallel datasets with potential for instructive comparison, then there is some
merit in working backwards from this point to consider which
complementary methods might provide most opportunity for such
comparison (see Flick, 2018a). Working backwards is also a strategy to be
recommended in terms of anticipating and allowing for the many practical
challenges that are likely to be involved in mounting a focus group study.

Research design practicalities

It would be a pity to scrupulously develop your research design only to be let
down by failing to take account of the practicalities involved. Even with
regard to making what are apparently straightforward decisions, the
researcher will be influenced by disciplinary conventions and her/his personal
skills, as well as the aims of the research project in question and the level of
funding and time available to carry out the research. Focus groups do not
necessarily offer an easy option and, as McParland and Flowers (2012)
observe, there are many challenges involved in actually getting your sample

81
to attend the arranged focus groups. Researchers are well-advised to think
carefully at the outset about their choice of research settings, and moderators,
the level of analysis to be carried out and its implications for transcribing or
note-taking (including whether or not an assistant moderator is required),
sampling (discussed in Chapter 5), approaches to generating data (covered in
Chapter 6) and, indeed, ethical issues (presented in Chapter 7).

Settings
Researchers need to be flexible with regard to where they hold focus groups
in order to maximize participation. There is probably no such thing as a
setting that is universally acceptable to all of the people one might like to
involve in the research. It is important, however, to understand that utilizing
too narrow a range of locations may seriously compromise the spread of
participants involved and the range of views that are expressed.

Sometimes the choice of settings is limited, due either to the availability or

cost of suitable premises, and the researcher may have to compromise.
Depending on the participants involved, consideration may need to be given
to the possibility of providing a crèche (to allow parents to attend) or
accessibility (to allow disabled participants to take part). Issues of payment
and insurance coverage may also need to be addressed – especially in relation
to providing child care. Providing food may sound like a good idea, but can
be a culturally fraught issue, and it is essential, for example, to consider
requirements surrounding particular religious festivals and observance (such
as Ramadan).

It is important, also, to give some thought as to the likely impact of specific

location on participants and the focus of the data likely to be generated.
While clinical researchers may scarcely notice the presence of trappings such
as somewhat gruesome posters, the impact on patients should not be
underestimated. However, there is much that the researcher can do to
compensate for a less than ideal setting, such as ensuring that specific
questions and stimulus material are included in the topic guide in order to
steer discussion away from the associations suggested by the chosen setting
on to the topics most relevant to the research. (This is discussed more fully in
Chapter 6, which looks at generating focus group data.)

82
Selecting a venue for your face-to-face focus group discussions also requires
that you bear in mind the potential for unwelcome noise or the possibility of
discussions being overheard. If sessions are to be video-recorded, this calls
for attention to be paid to the physical layout of the room and it is essential
that the researcher familiarize her/himself with any equipment to be used.
However, no equipment can compensate for lazy or inattentive moderating or
poorly-thought-out or unpiloted topic guides and stimulus materials (see
Chapter 6).

Again, rather than being viewed as a limitation in focus group research,

building a variety of settings into the research design can strengthen its
comparative potential, with the differences to which this strategy gives rise
becoming a resource in analysis, rather than a problem.

Matching moderator and group

The researcher’s persona does impact on the form and content of data elicited
using focus groups, as indeed it does with all other qualitative methods. It is
to this aspect of the research endeavour that commentators are appealing
when they refer to the concept of ‘reflexivity’, which involves
acknowledging the ways in which the researcher actively contributes to the
data she or he is generating. Different moderators can generate data that are
different in content and form. It is important to take into consideration the
likely impact of a particular moderator and the match between this
individual’s characteristics – or perceived characteristics (Kitzinger and
Barbour, 1999) – and the group to which she or he is to be deployed. This is
likely to present less of a problem online, although participants will still make
attributions based on the moderator’s style of response and vocabulary used.
However, online methods do not automatically solve such problems and
researchers should ensure that they do not over-stretch themselves with
regard to using online methods, if they are not personally comfortable and
familiar with the form of communication involved.

Differences between moderators can, however, be used to advantage to

provide another window on the research encounter and the resulting data.
Some research teams have consciously drawn on such comparisons in order
to enrich their analysis. Burman et al. (2001), for example, whose research

83
team spanned a wide age range, chose to work in pairs to capitalize on the
characteristics of the various members of the research team. Since their
research on violence involved studying adolescent girls, the presence of a
young female team member who wore ‘trendy’ clothes was invaluable in
establishing rapport and credibility in the eyes of the girls, whereas the
involvement of an older researcher probably served to remind participants
that this was a serious piece of funded research.

However, just as it can be counterproductive to select a group that is too

homogeneous, so, too, can it be unhelpful to choose a moderator who is an
‘insider’. Hurd and McIntyre (1996) point out that there can be ‘seduction in
sameness’, whereby the researcher shares too many of the group’s taken-for-
granted assumptions and is therefore unable to expose these to critical
scrutiny. However, the use of a moderator who is, in some respects, an
‘outsider’ can help to elicit explanations and can serve to contextualize the
data being generated. The group situation, too, can offset the effects of this
mismatch between moderator and group members’ characteristics.
Reflecting, in the context of a Europe-wide study of young people’s
expectations of the future, on her own experience of moderating, as a young
white woman, a focus group comprising young British-Asian women,
Smithson (2000) concludes:

one white woman and one Asian woman would be unlikely to

produce as detailed a picture of young British Asian women’s lives
and debates. Here the group is collectively ‘powerful’ in that they
have access to shared knowledge of which the moderator is
ignorant. Rather than being constructed by the researcher as the
other, these Asian women use the focus group to position
themselves between two cultures at ‘intersecting axes of
identification’. (2000, pp. 111–12)

Although it is not always possible – or even desirable – to match moderator

and group, careful consideration should be given to the impact of the
moderator on the data generated and this should also be used as a resource in
the analysis. Some research teams make strategic use of moderators’ personal
characteristics to generate data for comparative purposes.

84
What is crucial, is that the impact of the researcher on the data is taken into
account in the analysis; that is, that this is used reflexively to analytic
advantage. It is also important to take into consideration the skills of the
moderator and to provide additional training (including, potentially, training
specific to online methods), where appropriate.

The presence of an adequately briefed assistant moderator can also be useful

in ensuring that relevant notes are taken during and immediately after a focus
group session. If the analysis is likely to focus on the process of interaction
(rather than involving purely content or thematic analysis), then an assistant
moderator can help to identify individual speakers, thus providing a valuable
resource for comparative analysis. It may be important to be able to
differentiate between those espousing different viewpoints, in order to trace
the processes through which interaction may serve to challenge participants’
perspectives, or even to cause them to modify their opinions.

Transcription
Even at the outset of a focus group project the researcher should give some
thought as to the level of analysis to be carried out and the implications this
has for decisions about transcription. Many researchers assume that they must
have verbatim transcripts, but this is not always necessary – particularly
where the intention is to carry out content analysis (see discussion in Chapter
8), without paying attention to the minutiae of interaction. A verbatim
transcript, in itself, does not automatically confer rigour any more than
relying on notes or listening repeatedly to tapes implies that the procedure is
necessarily lacking in terms of being systematic and thorough. This is a
property of the research process and is not closely related to the presence or
absence of verbatim transcripts. However, verbatim transcripts do open up
the possibility of returning to your data at a later date, perhaps to re-analyze it
in the light of new insights you have gained from subsequent studies or
through further reading. It is always possible to settle for a compromise –
provided that audio-tapes are retained, it is always possible to re-visit these
during the process of analysis and to fully transcribe only those sections
which are selected for more detailed analysis.

So enshrined are transcripts in the process of qualitative research that we

85
rarely question their value or the ways in which they are produced. However,
transcription requires a range of specialist skills and involves the
transformation of fleeting and lively discussion into a text (Poland and
Pederson, 1998, p. 302). It is, therefore, important to bear in mind what may
be left out of a transcript, as Macnaghten and Myers (2004) also note. Jenny
Kitzinger recommends reading transcripts while simultaneously listening to
the original recording, and noting (with the aid of your field notes) any
significant gestures, emphases and expressions (Kitzinger and Barbour,
1999).

One of the best pieces of advice for the novice focus group researcher is to do
some of the transcribing yourself. This makes you a much more attentive
moderator in the future, as it will bring you face–to-face with the frustration
of noticing where you have neglected to pick up on interesting leads or where
you have failed to seek clarification or to invite participants to finish
sentences that were interrupted. It also has the added bonus of making you
much more appreciative of the skills of audio-typists hired to produce focus
group transcripts – often with little guidance from researchers about their
requirements regarding layout. Some information as to the use to which you
want to put transcripts can also be very helpful for the typist charged with
this responsibility. Carrying out some of your own transcribing also pays
dividends in terms of familiarizing you with the data. For some types of
transcription, however, the services of an expert transcriber are required. This
relates specifically to approaches that involve conversation analysis.

Conversation analysis
Conversation analysis requires that transcripts be produced according to a set
of conventions, utilizing a range of symbols to indicate specific features of
talk. Such attention to detail is crucial, as, according to conversation analysts,
‘no facet of speech, whether it is a pause, a repair, a change in pitch or
volume, the selection of particular words, the point at which one speaker
overlaps another, or even a sniff, should be assumed to be irrelevant to
interaction’ (Puchta and Potter, 2004, p. 3). As Puchta and Potter (2004)
concede, this framework can be difficult to work with at first – both for the
researcher and for the typist, since it is cluttered with symbols indicating
features of speech delivery and intonation. (See Silverman, 1993, or the

86
appendix provided by Puchta and Potter, 2004, for a glossary of the symbols
required for ‘Jeffersonian transcription’, as this approach is called, and also
Rapley, 2018.) For those who are interested in pursuing conversation
analysis, Puchta and Potter (2004) also recommend consulting Hutchby and
Wooffitt (1998) and ten Have (1999).

Even if a rigorous conversation analysis approach is not followed, the general

focus group analyst can learn much from this attention to detail, focusing on
how specific features of talk act as a resource for participants and allow them
to accomplish certain tasks/make particular representations throughout the
process of the interaction. The researcher can also include, for example,
observations on tone, interruptions and body language in order to aid
analysis. (This is in line with the composite approach that is advocated in
Chapter 9, and which allows the researcher to take advantage of a range of
insights derived from different ways of analyzing focus group data.) Again,
provided that the recorded discussion is retained, it may be possible to have
sections of a discussion transcribed using the Jeffersonian method, should
this be considered likely to be helpful.

Key points
Focus groups can be usefully employed either as a stand-alone method
or as part of a mixed methods approach. In mixed methods studies,
focus groups have potential for developing more structured ‘tools’, such
as questionnaires, but they can also be used to advantage in illuminating
quantitative results.
The decision as to whether to employ focus groups or one-to-one
interviews has to be weighed up in the context of each study. While
interviews excel at eliciting ‘private’ accounts, focus groups give the
researcher access to the interpretations and arguments that participants
are willing to present in group situations, whether these are peer groups
or researcher-convened groups of strangers.
Repeat focus groups can afford additional insights into the development
of views over time.
Tele- or video-conferencing or online focus groups can be useful in
accessing otherwise hard-to-engage groups/individuals, but researchers

87
may have to devote time and effort to developing new skills.
Using focus groups within a qualitative mixed methods design also
raises important challenges.
While triangulation is a problematic concept, focus groups can provide
parallel data and, hence, facilitate interrogation of contrasting datasets
through comparison – particularly with regard to exploring and seeking
to explain discrepancies.
Pay attention, at the outset, to the practicalities involved (choosing
settings, matching moderator and groups, making decisions about
transcription and associated requirements).
There is no such thing as a ‘neutral’ setting for a focus group. Instead, it
is important to anticipate the effect of different possible locations on the
content of the data generated and to plan accordingly. Using more than
one setting can provide comparative data.

Further reading

Issues of planning focus groups and of combining them with other

methods are outlined in more detail in the following books and articles:

Barbour, R.S. (1999) ‘The case for combining qualitative and quantitative
approaches in health services research’, Journal of Health Services
Research and Policy, 4 (1): 39–43.

Crabtree, B.F., Yanoshik, M.K., Miller, M.L. and O’Connor, P.J. (1993)
‘Selecting individual or group interviews’, in D.L. Morgan (ed.),
Successful Focus Groups: Advancing the State of the Art. Newbury Park,
CA: Sage, pp. 137–49.

Flick, U. (2018) Designing Qualitative Research (Book 1 of The SAGE

Qualitative Research Kit, 2nd ed.). London: Sage.

88
Chapter Five Sampling

Contents
Identifying sampling possibilities 67
Principles of qualitative sampling 69
Group composition 69
Number and size of groups 70
Pre-existing groups 73
Making the most of the potential for comparison 75
Returning to the field and second-stage sampling 79

Objectives
After reading this chapter, you should:

be able to evaluate the potential of existing sampling pools/possibilities;

understand the principles of qualitative sampling;
appreciate the issues involved in sampling and group composition in focus
groups;
see the advantages and limits of using existing groups; and
be alert to other approaches to sampling that can extend analytic
possibilities.

No text on focus groups would be complete without giving due attention to

sampling. This chapter focuses on sampling strategies, and shows that
decisions about sampling bring together several issues: the aims of the project
and where it is positioned on the realist–constructivist continuum;
practicalities (in terms both of the options open to the researcher via existing
networks/opportunities – including available databases – and the possibilities
afforded by the project timetable); and, last but not least, ethical
considerations (regarding how to approach potential participants and the
likely consequences of taking part/publicizing or sharing findings).

89
As with all qualitative research, focus group projects – and their design in
particular – benefit greatly from researchers’ knowledge of the field of study.
This includes awareness of possible sampling frames, stakeholder groups or
constituencies from which to select focus group participants. Sometimes
researchers have utilized sampling pools afforded by their previous
quantitative research in order to select participants for focus groups, allowing
for particular findings to be more fully explored through focusing on specific
sub-groups of respondents.

In focus group research, sampling decisions are two-tiered: not only does the
researcher need to think about the individuals to be recruited into the study;
s/he also needs to make decisions about the composition (and number) of
groups to be convened. This chapter also gives advice on group composition
and using pre-existing groups. It considers ethical issues and the need to take
these into account in developing sampling strategies and convening groups.
Although much qualitative work has traditionally relied on convenience
samples, there is much to be gained by taking a more strategic approach. Just
because groups already exist, this does not necessarily mean that these same
groups should be the ones that shape the study design, and the rationale for
utilizing pre-existing groups should be carefully thought through.

Sampling is of crucial importance as this is what, ultimately, provides the key

to the comparisons that it will be possible to make when the data are
analyzed. The principles of qualitative sampling are outlined and this chapter
provides guidance on how to plan your sampling strategies in order to
maximize the potential to identify and seek to explain differences and
patterning in the data. Even where comparison is not the explicit aim of the
study in question, introducing this possibility can lead to unexpected insights
that can considerably enrich your analysis.

Identifying sampling possibilities

It is certainly worth considering all possible avenues for recruiting your focus
group sample – including routes more commonly followed in quantitative
studies. Hillier et al. (2012) decided to engage an international survey firm,
since this afforded them access to a national sampling frame, thus
significantly broadening the pool from which they could recruit young people

90
to their study. Even where there is a readily available sampling pool,
however, this may prove to be a laborious way of recruiting participants. Krol
et al. (2014) report that they contacted no fewer than 359 children (aged 8–11
years) and 395 adolescents (aged 12–15 years), via the client lists of two
rehabilitation centres, in order to convene seven face-to-face focus groups
and two online forums.

Wilmot and Ratcliffe (2002), whose study is discussed in Chapter 3, took the
opportunity afforded by their previous survey (and its sampling frame) to
carry out further sampling for focus groups, in order to allow them to explore
the reasoning processes that gave rise to their initial results relating to public
attitudes to organ donation.

Further possibilities are afforded by the Internet, both in terms of recruiting

participants and also, as a platform on which to conduct discussions. As we
have seen in Chapter 4, this may be particularly effective in accessing those
who are deemed ‘hard-to-reach’, or individuals and groups that are
geographically dispersed. Wood and Griffiths (2007), reflecting on their
experience of carrying out online research with gamblers, make the point that
the Internet may afford access to individuals who are ‘socially unskilled’ and
who might not otherwise be readily contacted.

Internet forums, although attractive in terms of geographical coverage and

ready access, nevertheless, may operate to exclude certain individuals, such
as those without proficiency in the language used – as Alinejad (2011)
acknowledged, with regard to her study of second-generation Iranian
bloggers, living in North America. It is important to take into account how
different mediums shape audiences and, therefore, your potential sampling
pool. Researchers can be mistaken in thinking that group communications
online will provide them with data on the topics in which they are interested.
Wojcieszak and Mutz (2009), reporting on a national survey of online
participation of over 40,000 respondents, found that most exchanges centred
on shared hobbies, interests and leisure activities, with only around half
touching on some discussion of political topics. They also observed that
political, civic, religious and ethnic groups were extremely limited in terms of
generating cross-cutting political discourse online. Whereas professional
online groups provided cross-cutting views on a range of topics, ethnic online

91
groups encouraged reinforcing discussions.

Regardless of the options being explored, preliminary fieldwork can pay

dividends in terms of sensitizing the researcher to the criteria that are relevant
and that should inform sampling decisions. Even where it is not practicable to
carry out an extensive ‘scoping’ exercise, researchers can benefit from the
knowledge of community groups, which can play an important role in
educating the researcher as to the diversity, nuances and sensitivities involved
– as is pointed out by Umaña-Taylor and Bámaca (2004). These researchers
found that consulate staff, as well as those employed in community
organizations, were a valuable source of information with respect to the
differences between sub-groups of Latina women of Colombian, Guatemalan,
Mexican and Puerto Rican nationality, with contrasting profiles and
experiences in terms of reasons for migration, length of residency, housing
locality, incidence of poverty, educational levels and income.

Insider knowledge, including that previously acquired as a practitioner, can

be useful in guiding selection of stakeholder groups, and belonging to such
networks can also assist in gaining access. Although they employed
purposive sampling (see discussion later in this chapter) to ensure a spread in
terms of a number of patient characteristics, Boger et al. (2015) recruited
their focus group participants from existing stroke support groups, but
acknowledge that this may have limitations. They conclude: ‘The strategy of
sampling via stroke support groups may mean that the views of participants
who demonstrate effective self-management skills are over-represented.
Alternatively, it may be that people who do not attend stroke groups are
successful self-managers, and therefore do not seek additional support’
(2015, p. 185).

However, it is also important to weigh up possibilities through the eyes of a

researcher, with decisions always being made on the basis of the potential for
comparison afforded. While attempting to be helpful, gatekeepers may
inadvertently ‘select out’ individuals who appear not to fit the required
characteristics, but whose inclusion might add significantly to discussions by
introducing a new dimension. Although it may be tempting to use
gatekeepers in order to carry out selection for focus groups, this is not
generally advisable, since they are unlikely to fully understand the purposes

92
and assumptions underpinning the research. Qualitative sampling employs
criteria and reasoning that differ markedly from those that characterize
quantitative research practice.

Principles of qualitative sampling

Sampling is crucial, as it holds the key to the comparisons that you will be
able to make using your data (see also Flick, 2018c, Chapter 3). Both Kuzel
(1992) and Mays and Pope (1995) stress that the purpose of qualitative
sampling is to reflect the diversity within the group or population under study
rather than aspiring to recruit a representative sample. Thus, such sampling
will capitalize on any ‘outliers’ identified and seek to incorporate these
individuals or sub-groups, rather than dismissing them – as would be done if
sampling quantitatively. An example might be seeking to include parents of
home-educated children or travellers in a study that looked at parenting,
using routes other than schools for identifying a sample, or making an effort
to encourage men with primary childcare responsibilities to participate in the
study. The issue here is not the number of such individuals in the population
at large, but rather the insights that can be provided through including these
exceptions, and their potential to throw into sharp focus some of the taken-
for-granted assumptions or processes that are otherwise unremarked upon.
Sampling decisions already anticipate analysis and the implications of
sampling choices and their potential to facilitate theoretical analysis are
further discussed in Chapter 9.

Qualitative sampling is generally referred to as involving either ‘theoretical

sampling’ (Mays and Pope, 1995) or ‘purposive sampling’ (Kuzel, 1992).
Whatever term is used, it refers to essentially the same process: theorizing –
albeit at an early stage – about the dimensions that are likely to be relevant in
terms of giving rise to differing perceptions or experiences. This was the
basic premise of research carried out by Strandberg et al. (2013) who sought
to establish how the practice context might give rise to particular prescribing
patterns amongst general practitioners (GPs) in southern Sweden. In this
case, sampling involved comparing the views of GPs in urban and rural areas.
Similarly, Boger et al. (2015) sought to recruit a range of individuals to their
study on stroke self-management, selecting them according to both

93
demographic and clinical criteria; i.e. on the basis of time since their stroke,
gender, age, ethnicity, and level of impairment.

‘Purposive’ sampling relates to the anticipated use of the selected criteria in

making comparisons once the data have been generated. In other words,
purposive sampling allows for the data to be interrogated purposefully, that
is, in order to carry out systematic comparison (Barbour, 2001, 2014a).

Group composition
Since the group will be the main unit of analysis in focus group research, it
makes sense to convene these to facilitate comparison, by ensuring that group
members share at least one important characteristic. Not only does this make
good sense in terms of research design, it can also encourage people to attend
and may facilitate discussion on difficult topics, such as those where
participants share some stigma (Bloor et al., 2001).

Morgan (1988) provides the useful reminder that focus groups should be
homogeneous in terms of background and not attitudes. Although some
commentators on focus groups, such as Murphy et al. (1992), view
differences of opinion as potentially disruptive, these are what lend ‘bite’ to
focus group discussions. Provided that we are not cavalier about mixing
together people who are known to have violently differing perspectives on
emotive issues, a little bit of argument can go a long way towards teasing out
what lies beneath ‘opinions’ and can allow both focus group facilitators and
participants to clarify their own and others’ perspectives. Perhaps, in some
contexts, this can even facilitate greater mutual understanding. In terms of
generating discussion, a focus group consisting of people in agreement about
everything would make for very dull conversation and data lacking in
richness. Fortunately, however, this is unlikely; even where the researcher
misguidedly attempts to bring together like-minded people, they are unlikely
to be as one-dimensional as, undoubtedly, are our approximate and somewhat
crude sampling categories.

Number and size of groups

94
As to the question of how many focus groups to hold, this is determined by
the comparisons that the researcher wishes to make. However, discussion of
the reasons behind these decisions is often neglected in reports of focus group
studies. Carlsen and Glenton (2011) reviewed 220 studies published in 2008
and found that only 17% of these provided an explanation for the number of
groups carried out.

There is no magic number and more is not necessarily better, although

holding two focus groups with groups that have similar characteristics may
place the researcher on firmer ground in relation to making claims about the
patterning of the data, since it would suggest that the differences observed are
not just a feature of a one-off group, but are likely to be related to the
different characteristics of participants reflected in selection. Since each
individual participant possesses a constellation of characteristics (age, gender,
socioeconomic and educational background), it is likely that it will be
possible to engage in some intra-group comparisons, since for example, a
women’s group may well comprise individuals with a range of ages. It is
always prudent, however, to leave some leeway for adding other groups, as
further potential comparisons occur.

Another frequently asked question relates to the number of participants that

should be recruited to each focus group. Many of the earlier focus group texts
echoed the advice that tends to be given in marketing research that the ideal
size of a group is 10–12 people. In comparison to marketing research, where
many discussions are summarized – either verbally or in note form – the
focus of social scientists is most often a verbatim transcript, which is then
subjected to detailed and systematic analysis. Both in terms of moderating
groups (picking up on and exploring new leads as these emerge) and in terms
of analyzing transcripts, I would argue that a maximum of eight participants
is generally quite challenging enough. The requirements on the researcher to
identify individual voices and seek clarification and further exploration of
any differences in views that emerge make larger groups, if not impossible,
then exceedingly demanding to moderate and analyze. In terms of a
minimum number, it is perfectly possible to hold a focus group discussion
with three or four participants (Kitzinger and Barbour, 1999; Bloor et al.,
2001), and for some topics this may be preferable – for example, with
terminally ill older people (Seymour et al., 2002).

95
Barrett and Kirk (2000) stress the importance of over-recruiting for focus
groups with elderly participants, having found, as did Owen (2001) – here in
relation to women with serious and enduring mental illness – that such
groups are especially prone to ‘no shows’ on the day. Umaña-Taylor and
Bámaca (2004) also highlight this as a challenge with regard to recruiting
Latina mothers and recommend over-recruiting by at least 50% for groups
who may find it difficult to attend focus groups due to the nature of their
other family commitments.

Although it can be useful to produce a sampling grid to inform your sampling

strategies and to allow you to measure progress, this represents an ideal and it
may not always be possible to recruit a sample that reflects all of the
characteristics or dimensions you have identified as desirable. In a study
(Barbour et al., 2012) looking at the views of women who had recently given
birth with regard to taking folic acid supplements, we initially drew up a
sampling grid, which identified six potential focus groups. Our plans were
based on responses to a screening questionnaire, completed by women
attending baby clinics, and we sought to hold separate groups for women who
had taken folic acid and those who had not, considering that it might be
problematic to bring together, in the same group, women with differing
orientations. We also intended to convene separate groups for first-time
mothers and those with two or more children, as well as some groups that
mixed women of different parity (i.e. number of children) – all in the interests
of affording potentially useful comparisons. We also realized early on in the
process of discussion that questionnaire responses did not necessarily accord
with women’s actual experiences – either in relation to pregnancies or taking
folic acid. Some who had been provisionally allocated to the group of first-
time mothers had had previous stillbirths or terminations, and participants’
behaviour in relation to folic acid reflected a wide range in terms of those
who had taken folic acid intermittently, or during periods that were not
recommended in guidelines, or who had exceeded recommended doses. To
further complicate matters, despite women having indicated their willingness
to take part in the study, we encountered great difficulty in encouraging them
to attend arranged sessions. In the event we opted to run telephone focus
groups and took a more relaxed approach to mixing women of different
parity and reported folic acid behaviours. (For further discussion of our
original sampling template see Barbour, 2010.)

96
Many qualitative projects build their samples incrementally, which allows
researchers to take advantage of new insights developed along the way.
However, this begs the question as to when to stop recruitment. In their
review of published focus group papers, Carlsen and Glenton (2011)
observed that many authors attempted to explain their sampling choices –
especially their decision as to the number of groups to hold – by reference to
‘saturation’. This refers to the concept developed by Glaser and Strauss
(1967), as a further aspect of the iterative process involved in grounded
theory, whereby researchers eventually reach the conclusion that their coding
categories have stopped evolving, and are becoming repetitive or ‘saturated’.
However, in the context of providing a rationale for sampling decisions –
particularly the decision to stop recruitment – the idea is more often cited or
appealed to than it is evidenced, since it is usually employed as a short-hand
– seemingly self-evident – term, without a full explanation being given.
Again, practicalities are likely to intervene and available funding and
timetable constraints are likely to be the most important factors in reaching
this decision.

Green et al. (2005) provide an exception to the shortcomings identified by

Carlsen and Glenton (2011), in that they explain the reasoning behind their
sampling choices. Their experience shows that imaginative use of focus
groups can even afford further comparison in an international context. They
studied public understanding of food risks in people belonging to four
different stages in the life cycle in a variety of settings in Finland, Germany,
Italy and the UK. Although this project involved a relatively small number of
focus groups, convening groups to take account of the different ages and life
stages of participants enhanced the comparative potential of the data. A
common misconception is that purposive sampling necessarily inflates the
number of participants required. However, if you realize that each individual
may potentially meet several of the desired criteria in terms of diversity
(everyone having a gender, age, social class, etc.), then it becomes clear that
multiple comparisons can be made on the basis of fewer participants than a
first consideration of the approach to sampling might suggest.

Although it is useful to sit in a research office and draw up a sampling frame,

it is not always possible to fill all the boxes identified and it is also important
to leave the outline sufficiently open in order to capitalize on any further

97
insights that occur to the researchers as the study progresses, or further
opportunities present themselves. In practice, theoretical models, knowledge
of the existing literature, knowledge of a specific locality, contacts and
gatekeepers, and serendipity all play a role. In some cases, it may be possible
to utilize pre-existing groups rather than bringing individuals together
specifically for the purposes of the research project.

Pre-existing groups
Marketing research texts consistently advise recruiting groups of strangers, in
preference to naturally occurring groups. However, it is important to bear in
mind the context in which such advice is offered. Marketing research, as we
saw in Chapter 1, is primarily concerned with tapping into public preferences
and is charged with making broad recommendations as to whether or not to
develop or market a specific product or to pursue a particular advertising
campaign. In this specific context, pre-existing groups would be likely to
skew the findings in favour of sub-groups within the population, rather than
affording the blanket coverage that is generally the aim of marketing
research.

However, when engaging in health services or social science research, our

goals are somewhat different from those that inform marketing research. We
are, quite simply, asking very different – and usually more complex –
questions, often with the ultimate aim not only of answering straightforward
questions but also of contributing to long-established and constantly
accumulating bodies of disciplinary knowledge. (Moreover, we carry out our
research within a context that is characterized by collaboration – personal and
institutional rivalries notwithstanding – as distinct from the more fiercely
competitive world of business and marketing.) The goal of most health
services and social sciences research involving focus groups is likely to be
that of developing a greater understanding of process rather than that of
predicting outcome in terms of the likely public response to a new product or
marketing campaign.

Rather than viewing pre-existing groups as a potential problem, however,

some commentators, such as Bloor et al. (2001), argue that there are some
advantages in utilizing what they refer to as ‘pre-acquainted’ groups. Seeking

98
to elicit the views of young Danes with regard to the potentially sensitive
topic of drug risk, Demant and Ravn (2010) relied on the recruitment of
volunteers from particular schools, augmented by snowball sampling, which
resulted in focus groups of peers who were already known to each other.
Brown (2015), who carried out research on teenage parenting, argues: ‘using
groups in naturally occurring settings, where participants feel comfortable
and secure, enables complex social data to emerge, in terms of how the
participants speak between themselves, and how that illustrates relational
dynamics’ (2015, p. 94). A further example of the advantages to be gained by
utilizing pre-acquainted groups is provided by Berge et al. (2012), who
included entire families in focus groups convened for their Minnesota inner-
city neighbourhood research. This fitted particularly well with their aim of
exploring family-level perspectives of risk and protective factors. It is quite
surprising that there are not more examples of such usages – especially given
the amount of therapeutic work that focuses on family groupings.

Using pre-existing groups, however, raises important ethical issues,

particularly with regard to ensuring confidentiality. Researchers need to be
mindful that these groups have a life that continues after they have elicited
their data and should seek to minimize possible negative ramifications. It is
essential that the researcher takes time to emphasize the importance of
confidentiality prior to the discussion and that space and time are allowed for
any concerns about disclosures to be raised at the end. Particularly in work
with members of minority ethnic communities, participants in focus groups
may have complex interlocking relationships that can be affected by shared
confidences (Ruppenthal et al., 2005).

As with all other pieces of advice regarding focus group research, the
decision as to whether to incorporate or avoid pre-existing groupings depends
on the scope of the research project in question. For instance, it is likely that
the presence of someone’s line manager will inhibit the frank exchange of
ideas. However, this may be germane if the purpose of the research project
(as in the case of Berge et al., 2012 above) is to provide an understanding of
the ‘real-life’ context in which people work or come together for other
purposes.

One of the questions I am most frequently asked by workshop participants is

99
whether to convene mixed professional or single professional focus groups.
As usual, there is no ready answer, except to point out that groups consisting
only of general practitioners will generate data different in content from that
of groups held with nurses or hospital doctors. reflecting their varying
concerns and the complementary, but distinct, nature of their work roles.
Mixed professional groups will generate data that differs again from that
produced in single professional groups. My advice, in this situation, would be
to convene both single and mixed professional groups and to compare the
data elicited in these two different contexts, and only then to make a decision
as to which type of data is most germane to the research question in hand.

Where researchers decide to try to convene focus groups with pre-existing

groups it can be pragmatic to utilize pre-arranged meeting slots, such as team
meetings or conferences. However, there are some pitfalls associated with
such ‘piggybacking’ (Kitzinger and Barbour, 1999) and it is essential to warn
all potential attendees that the focus – and, perhaps, even length – of the
sessions will be different from that of usual meetings and, if necessary, to
reiterate this during the discussion.

Making the most of the potential for comparison

In drawing up plans for a focus group study, it is important to keep in mind
the need to capitalize on the potential for comparison. This is illustrated (see
Box 5.1) by our experience on a project in which I was involved as a
consultant.

Box 5.1 Reasons for under-reporting of racist incidents

The research team had been commissioned to carry out research in order to
elaborate the reasons for an apparently low rate of reporting of racist
incidents in the Scottish region of Strathclyde. A mixed methods approach,
involving focus groups, interviews and a survey, was employed. Initial
research team discussions about the focus group component identified the
importance of eliciting – and comparing and contrasting – the perspectives
and experiences of members of various ethnic minority groups; men and
women; people of differing ages; individuals belonging to different social
classes; those who had been born in Scotland and more recent immigrants.

100
Thus we developed a notional sampling grid with a wide range of potential
groups. We worked closely with a range of local organizations in order to
establish, firstly, whether there were sufficient numbers in any given group
to allow us to convene the range of groups required. Given the relatively
small number of people in Glasgow belonging to the African-Caribbean
community, we were able to convene only one group, which involved
people from a range of ages and social class positions. In addition to
paying attention to social class differences we also sought to explore the
views of people in specific occupations, including representatives of
minority ethnic organizations, small business owners, and students.
Another potentially interesting dimension related to the locality in which
people lived, and we were able to hold a Chinese focus group discussion
with a number of individuals living in a small town rather than the inner
city.

The minority ethnic groups run were as follows:

Asylum seekers (English-speaking from a range of ethnic

backgrounds).
Representatives of Chinese organizations.
Asian men (range of ages and social classes).
Asian women (range of ages and social classes).
Representatives of Asian organizations.
Asian young men (16 to early 20s).
Asian young women (16 to early 20s).
African-Caribbean (mixed gender, range of ages and social classes).
International students (various ethnic backgrounds).
Chinese (mixed gender and locality, and belonging to lower
socioeconomic groups).
Chinese small business owners.
East European (mixed ages and social classes).
Asian and African-Caribbean researchers.

We reasoned that it might also be illuminating to carry out a focus group

discussion with a mixed ethnic group consisting of small business owners,
in order to tease out which issues were specific to particular groups or
localities and which were common. Not surprisingly, this proved
impossible to set up, due to the long hours worked by people running such
businesses and the necessity of holding such a group in a location that
would have involved travel for some of the participants. We also convened
eight focus groups with members of the indigenous white community.

101
 These included groups of people living in affluent, mixed and deprived
areas, male students, professional women, a church group and a group of
individuals actively involved in local politics. Focus group discussions
were also held with groups of serving police officers throughout the
region.

Seeking also to place her study in the widest possible context, Griffiths
(2014) – who has studied the phenomenon of the influx of Polish immigrants
to a specific locality –also carried out focus groups with members of the
indigenous community. Powell (2014) also took the opportunity to view,
through a number of different lenses, the neighbourhood relations in an
Appalachian town adjacent to a university campus. Alongside a photo-voice
project, she opted to convene a focus group with year-round residents; one
with students; and one mixed group.

Focus groups are often employed within mixed methods ethnographic studies
and this can allow for sampling decisions to be informed by the wider
contextual information that is acquired through the immersion of the
researcher in the setting. Rather than sticking rigidly to your pre-formulated
sampling templates, you would be well-advised to take advantage of
serendipitous opportunities – particularly those that provide scope for further
comparison, or which are likely to extend discussion in directions that are
congruent with your research purpose.

In relation to the focus groups held for a project that was looking at decision-
making about medication, we observed that some participants – notably those
who were recovering from a heart attack – were facing particular problems in
relation to coping with extensive drug regimes. This led us to convene a
further two groups, comprising individuals in this position, and which
identified additional difficulties experienced as a result of one drug being
substituted for another, giving rise both to financial issues (where people
were paying for prescriptions) and concerns about drug interactions (which
sometimes went unreported, due to concerns about the additional expense
that might be incurred). In the course of this project the occasional participant
might bring along a partner or friend. Although this was not something we
had anticipated, we welcomed this, reasoning that it might provide additional
insights since it suggested that discussion was perhaps more likely to
resemble everyday talk.

102
Further analytic potential
In a study seeking to explore the potential for involvement of carers in drug
services, Orr et al. (2013) convened focus groups in order to compare the
perspectives and differing lenses brought to bear on this topic by carers of
drug users; front-line drug services providers; and drugs policy makers.
Although, in terms of planning the study, these were construed as forming
separate groups (defined by the route through which individuals were
recruited), in practice, there was some overlap, since carers could also be
service providers or policy makers and vice versa. Rather than this giving us
cause for frustration, such conjunctions proved to be a valuable resource,
since they helped to initially articulate some of the commonalities (in terms
of the views espoused by these different groups) which we then went on to
explore.

Giving thought to the comparative potential also increases the likelihood of

including groups who might otherwise be overlooked. This ‘pushes
researchers beyond the voices that are most familiar, most obvious, most
articulate, or easiest to recruit’ (Macnaghten and Myers, 2004, p. 71).

C. Wright Mills, writing in 1959 about what he called ‘the sociological

imagination’, exhorted researchers to employ a ‘sociological playfulness of
mind’, which involves, amongst other approaches, turning research questions
on their head. Thus, in attempting to understand why people do not do
something, it may also be useful to problematize the behaviour that we view
as desirable or, at least, not requiring an explanation; for example, why do
people take professionals’ advice? This sheds a potentially very useful light
on those studies (discussed in Chapter 2) which have focused on ‘difficult’ or
‘non-compliant’ patients or those who fail to take up opportunities. A study
on childhood immunization, carried out by Evans et al. (2001) used this
strategy to advantage, by recruiting to their focus groups not only parents
who had refused immunization, but also including – and comparing the views
of – those who had chosen to immunize their children. Interestingly, they
established that the views of these two sets of parents were not as far apart as
might have been anticipated: both groups initially had shared very similar
concerns and uncertainties, but the researchers were able to illuminate the
processes through which parents had arrived at their different decisions.

103
Embedding such ‘Why don’t …?’ questions in a wider discussion also serves
the useful function of not singling out for potential criticism those who have
failed to take up services or follow advice. It therefore avoids the resultant
‘sampling by deficit’ (MacDougall and Fudge, 2001), which threatens to
alienate potential participants and which can raise important ethical questions
(as is further discussed in Chapter 7).

The Box 5.2 study provides an example of a thoughtful and productive

approach to sampling.

Box 5.2 Selecting participants and study sites

Espinoza and Piper (2014) sought to study the ways in which people in
Chile collectively made sense of their country’s fraught past. They chose
to study three cohorts, selected because of their differing experiences:
those over 60 years (who had been young adults at the time of the 1973
coup d’état); 30–60 year olds (who had lived through it) and 18–30 year
olds (born after the end of Pinochet’s dictatorship). They explain their
reasons for this choice: ‘meanings about the past are dynamic, unstable,
and depend both on the social positions from which they are enunciated
and on the contexts in which they circulate and are discussed’ (2014, p.
719). ‘The complexity of studying collective memory processes and
generational discourse at memory sites in countries with traumatic pasts
like Chile demands the construction of innovative methods’ (2014, p. 712).
‘We understand sites of memory as those urban spaces that while making
reference to the past are also used by people as scenery to talk about the
past, constructing and deconstructing interpretations about their own
history and the country’s past’ (2014, p. 716). The researchers chose sites
identified as those ‘having most symbolic relevance in terms of
remembering the violation of human rights during [Pinochet’s]
dictatorship’ (2014, p. 717).

Returning to the field and second-stage

sampling
The original formulation of grounded theory (Glaser and Strauss, 1967)
advocated that researchers return to the field in order to test out emergent

104
hypotheses. However, the current funding climate and tight project deadlines
mean that this is, in many instances, an unattainable ideal. In contrast to other
qualitative methods, focus groups afford unrivalled potential to engage in this
sort of exercise, through ‘second-stage sampling’ or the convening of further
‘wildcard groups’ (Kitzinger and Barbour, 1999) to enhance analytical
sophistication. In terms of those aspects of our research over which we retain
control, it is helpful to remain alert to further differences within groups, not
only with respect to the niceties of social interaction and the need to
minimize distressing encounters, but also in order to develop the analysis.
Although an individual may have been recruited into a focus group study by
virtue of some characteristic (e.g. age or gender), there may be other aspects
of his or her situation that become apparent only during discussion, but which
are illuminating and may provide ideas for further sampling.

An example of the dividends paid by this approach is the study of general

practitioners’ views and experiences of sickness certification (Hussey et al.,
2004; see Box 5.3). Provided that such ‘second-stage’ sampling does not
involve the researchers contacting an entirely new group of people and
merely involves configuring the groups differently to reflect a specific shared
characteristic in preference to others, it is generally possible to allow for such
an eventuality – even in ethics applications – by reserving the option of
convening further groups, depending upon the provisional findings produced
during preliminary data analysis. It may also be possible to convene further
‘wildcard groups’ (Kitzinger and Barbour, 1999), which may involve
recruiting through additional routes – provided that this is allowed for in the
original proposal. Most commonly, this is likely to involve listing a wide
range of potential recruitment sites in the initial project outline and ethics
application.

Box 5.3 An example of ‘second-stage’ sampling

The four general practitioner members of the research team drew on their
own knowledge of what might be factors likely to affect the experiences of
general practitioners (GPs) and we (Hussey et al., 2004) decided that we
would seek to convene groups of GPs practising in urban, rural and remote
localities, and in affluent and deprived areas. Clearly the challenges of
dealing with the potentially fraught issue of providing sickness

105
 certification were likely to be somewhat different for GPs living and
working in a tight-knit island community than for GPs working in a
relatively anonymous inner-city area, where the GP almost certainly would
not have her/his home. Incorporating a range of localities was also likely to
afford potential for comparison in terms of the different types of employer
active in an area and the implications for issuing ‘sicknotes’ (i.e. whether
most of these were destined to arrive in the same office where there was an
identified principal employer, such as a large factory). We agreed, at the
outset, that we wanted to include both men and women and GPs of varying
length of experience and degrees of seniority, as well as those working in
large group practices, smaller practices and, if possible, some single-
handed GPs.

Having sampled according to these criteria and having run the first set of
seven focus groups, the GP moderators compared notes and we began the
process of preliminary analysis of our data. As well as looking at the
patterning (i.e. similarities and differences) between the seven groups, we
also took note of which members of each group were raising particular
issues. This exercise suggested that there might be specific issues for GP
locums (who worked for short periods of time in a number of different GP
practices), GP registrars (who were still undergoing training) and GP
principals (who had management responsibilities, a long-term commitment
– often financial – to one GP practice, and whose remit included providing
continuity of care for patients). We therefore decided to convene three
additional focus groups – one with each of these three groups – to further
explore this hunch, or hypothesis.

Like any other method, focus groups can culminate in high-quality research
only when due attention is paid to developing an appropriate and rigorous
research design. Sampling is the keystone of good qualitative research
design. The main points of this chapter can be summarized as follows.

Key points
Try to be creative with regard to identifying potential recruitment
sources, but remain alert to the emphasis and gaps in coverage that may
result from involving gatekeepers in the recruitment of your sample.
Both top-down and bottom-up recruitment strategies can result in certain
voices being unrepresented or muted.

106
It is important to acquire background information about the group being
studied, either through preliminary fieldwork or by accessing the
knowledge held by local organizations.
Although pre-existing groups can afford access to discussions that more
closely approximate ‘real-life’ situations, these raise challenges in terms
of maintaining the research focus, and the implications for group
members should be fully considered.
Ethical issues are inextricably bound up with research design choices
around sampling. The effect on pre-existing groups of taking part in a
focus group discussion should be taken into account and questions and
exercises designed with this in mind. Concerns about the consequences
for individuals of talking with others with particular characteristics
sometimes have to take precedence over research design requirements.
Sampling is of crucial importance as it holds the key to the comparative
potential of your dataset.
The aim of ‘purposive’ or ‘theoretical’ sampling is to reflect diversity,
not to achieve representativeness.
There is no magic formula regarding the number of focus groups to hold
or the number of participants in each group. Rather, this depends on the
comparisons you wish to make, the research topic, the type of data you
wish to generate and how you plan to analyze this.
While it is helpful to draw up a sampling grid that reflects the
characteristics of your ideal sample, you should remain alert to
additional opportunities for comparison afforded by unanticipated
differences between participants.
You should try to be ‘theoretically sensitive’ throughout the research
process, in order to spot gaps in coverage or potential for exploring
further distinctions/differences.
Second-stage sampling can be extremely valuable in following up
‘hunches’ developed through paying attention to individual voices
within focus group discussions.

Further reading

The following publications give you some further advice about how to
sample in focus group research:

107
Flick, U. (2018) Designing Qualitative Research (Book 1 of The SAGE
Qualitative Research Kit, 2nd ed.). London: Sage.

Kitzinger, J. and Barbour, R.S. (1999) ‘Introduction: the challenge and

promise of focus groups’, in R.S. Barbour and J. Kitzinger (eds),
Developing Focus Group Research: Politics, Theory and Practice.
London: Sage, pp. 1–20.

Kuzel, A.J. (1992) ‘Sampling in qualitative inquiry’, in B.F. Crabtree and

W.I. Miller (eds), Doing Qualitative Research. Newbury Park, CA: Sage,
pp. 31–44.

108