f H
Analysing Focus Groups
Rosa line S. Barbour
As focus group usage has become more widespread, this has sparked sometimes heated debates about the best approach to analysing focus group data. Historically, although focus groups were used in some other contexts, marketing research certainly pioneered the application of this method and has been influential in terms of providing advice on setting up and running such discussion sessions. The marketing tradition, however, has had considerably less to say about analysing focus group data, due, perhaps, to its focus - on gauging the likely success of specific products or advertising campaigns -and, therefore, on producing answers. Focus groups have enjoyed a particularly enthusiastic reception by the health services research community, but here, too, guidance on analysis has been scant.
There is no 'one-size-fits-alP approach to analysing focus group data. This is because approaches to analysis and research aims are inextricably linked. Research utilising focus groups can usefully be envisaged as forming a continuum - with practical or applied projects
at one end and studies which address disciplinary or theoretical concerns at the other end.
The approach to analysis and the degree of sophistication possible are largely determined by the overarching aims of the research and the format and structure of the original focus group discussions - for example, the extent to which the moderator leads the discussion or intervenes; the number and specificity of questions asked; and the content and manner in which any stimulus materials are used. This depends, ultimately, on the epistemological and ontological assumptions underpinning the research. This is the first topic to be addressed in this chapter, which will then outline the initial steps in making sense of focus group data, before presenting further analytic resources. A case is made for employing a composite approach, which blurs the distinction between applied and more theoretical orientations in focus group research, and it is argued that being open to a range of analytic strategies can confer benefits for all types of projects. Finally, the potential benefits and challenges of new developments are considered.
314
TYPES OF DATA AND THEIR ANALYSIS
EPISTEMOLOGICAL AND ONTOLOGICAL UNDERPINNINGS
As Kidd and Parshall (2000) observed, focus groups have been 'relatively agnostic' in that they have not been firmly associated with any one qualitative paradigm. While this has led to an unusually rich and stimulating variety of applications in a wide range of research contexts, this has, inevitably, also led to some confusion - especially in terms of selecting which pieces of advice to follow when embarking on analysing focus group data. This permissive appropriation of focus groups masks important epistemological and ontological differences, which impact on how projects are designed, how data are generated and, most importantly, how they are analysed. The different disciplines that have espoused focus groups as a method have, inevitably, each put their own 'spin' on this, since they have used this approach to interrogate further their own disciplinary and theoretical concerns building on their own distinctive set of techniques and procedures.
It is not especially helpful, then, to take a simplistic view that differentiates between realist and constructivist usages. In effect, the picture is much more complex and focus group research is carried out across a continuum that ranges from realist to constructivist approaches. Given the additional constraints of funding requirements and the need for many focus group researchers to produce findings that are of relevance also to practice situations, many projects are located somewhere in the middle of this continuum and approach analysis of data drawing on 'critical realism' (Bhaskar, 1989) or 'subtle realism' (Hammersley, 1992). This stance acknowledges the need to address practical concerns in presenting findings, but also allows for development of more theoretical explanations. Maxwell (2011) has more recently argued that it is possible - indeed, perhaps, preferable - to marry a relativist epis-temology with a realist ontology (see Maxwell and Chmiel, Chapter 2, this volume).
Focus group researchers, as individuals, however, are likely to lean towards one or
other end of the 'realist-constructivist' continuum, by virtue of their disciplinary training, and this can make for challenging, but potentially invigorating, discussion in multidiscipli-nary teams charged with analysing focus group data. Whether or not these differences are openly acknowledged, such orientations fundamentally impact on the research process, influencing assumptions about what counts as data, and how they should be analysed and presented.
Some commentators (e.g. Wilkinson, 1998) have been critical of the tendency to report focus group findings using quotes from individuals, to the exclusion of longer exchanges between participants. These longer excerpts, it is claimed, showcase the capacity of focus groups to elicit rich interactional data, as participants go about co-producing explanations (Barbour, 2007). Morgan, however, advocates taking a pragmatic approach to this vexed issue, arguing that the choice of which focus group excerpt to use is 'obvious when one quote makes (a particular) point more forcefully' (2010: 719). Quotes from individuals have their advantages in terms of their shortness and efficiency and we should not, perhaps, be too precious about the use to which individual comments are put.
One of the reasons for the emphasis on individuals' comments, however, is the underlying idea that focus groups provide a more efficient means of collecting the views of individuals than do other methods. Some researchers certainly employ focus groups as a 'back door' to obtaining survey-type data relating to attitudes (Barbour, 2007). This involves certain problematic assumptions regarding the measurabil-ity of attitudes and the capacity of focus groups to capture these effectively through recording, as immutable opinions, statements that have been made in a specific context and setting. Such usages overlook the way in which views are debated, defended and sometimes modified, in what is a much more fluid presentation of ideas. However, acknowledging that attitudes are 'performed' rather than being 'preformed' (Puchta and Potter, 2004) need not mean that we should focus exclusively on the
ANALYSING FOCUS GROUPS
315
interaction and performance to the neglect of the content. Morgan recently made the helpful observation: 'saying that the interaction in focus groups produces the data is not the same as saying that the interaction itself is the data' (2010: 721). As Morgan points out, it is entirely fitting that research espousing different goals should involve differing levels of analysis.
At the more applied end of the spectrum are those health services research endeavours, which have used focus groups for a variety of purposes, such as understanding the low uptake of screening programmes or resistance to health promotion or condition-specific treatment plans. Thus, researchers working in this context are, understandably, more interested in examining the content of focus group discussions. Researchers working at this end of the focus group continuum are likely to emphasise outputs, such as the development of appropriate health promotion materials (often for disadvantaged or marginalised groups with specific cultural needs, e.g. Vincent et al., 2006). Action research applications may not involve publication, since such work (according to commentators such as Hilsen, 2006) should be judged on its achievements rather than its methodological sophistication or findings. Occasionally researchers enlist participants as co-analysts, providing them with training, as did Makosky Daley et al. (2010) when carrying out a project with American Indians in Kansas and Missouri.
At the other end of the spectrum is focus group research that is more overtly framed to address theoretical or disciplinary concerns. Here the focus is on form and process, rather than content or outputs. In this iteration, focus groups are prized for their capacity to illuminate empirically a theoretical construct, such as Bourdieu's (1999) notion of 'habitus' (dispositions or lenses through which people view the world - see Bohnsack, Chapter 15, this volume), singled out by several sociologists (e.g. Callaghan, 2005) as being especially amenable to illumination via focus groups, since they allow researchers to access the process through which participants simultaneously manage
their individual identities and make a collective representation to the researcher.
'Conversation analysis' (CA - see Toerien, Chapter 22, this volume) is based on the assertion that 'ordinary talk, mundane talk, the kind of everyday chat we have with one another is fundamental to understanding all kinds of more specialised interaction' (Puchta and Potter, 2004: 9). Focusing on form and process, conversation analysis studies the regularities and conventions that underpin talk and pays particular attention to the sequencing of conversations and the impact that this has on the content of discussions and, crucially, what these exchanges allow participants to achieve. Closely related to conversation analysis, but originating from different disciplinary concerns, is 'discourse analysis' (DA), which also focuses on 'the action orientation of talk' (Willig, 2003: 163; see Willig, Chapter 23, this volume). With regard to such approaches, resources - that is, what counts as data - can be 'words, categories ... or "interpretative repertoires'" (Hepburn and Potter, 2004: 168).
Exponents of CA/DA approaches have sometimes been criticized for their overriding attention to detail. Criticism also includes the lack of attention paid to the broader context in which interactions are played out, resulting in a neglect of issues such as power, and social or political structures, as Rapley (2007) acknowledges. However, this is not a foregone conclusion, since some studies employ CA or DA methods to address such issues. Willig outlines the approach of 'Foucauldian discourse analysis' which, she argues, allows for the study of discourse as a mechanism for enacting, reproducing or challenging "wider social processes of legitimation and power'" (2003: 171; see Willig, Chapter 23, this volume).
Whereas CA has mainly relied on naturally occurring interaction (with a significant body of work relating to doctor-patient consultations), some researchers (e.g. Myers and Macnaghten, 1999; Macnaghten and Myers, 2004) have argued that focus group transcripts (see Kowal and O'Connell, Chapter 5, this volume) can also be analysed as text. In effect, the distinction between naturally occurring
316
TYPES OF DATA AND THEIR ANALYSIS
and researcher-convened groups is not especially helpful. If sufficient preparatory work is carried out by researchers - in terms of focusing the discussion (through careful development of topic guides and selection of stimulus materials) - the moderator can, in the event, take a 'back seat' - more akin to that of a traditional ethnographer - as discussion unfolds (Barbour, 2007).
The outline provided here, however, suggests an overly neat typology, whereas, in practice, there are many similarities and even some 'hybrid' projects. One point of convergence between realist and constructivist approaches is the emphasis on the internal/ alternative logic that informs the views/ perspectives/accounts of respondents or groups. Such ideas will, undoubtedly, be expressed by researchers adhering to various disciplinary contexts through the use of different language (as is suggested by the range of terms used here).
Projects may simultaneously serve realist and constructivist agendas. For example, Angus et al. (2007) convened focus groups to explore the everyday production of health and cardiovascular risk, drawing explicitly on Bourdieu's theoretical construct of 'habitus'. Nevertheless, this work allowed them to address issues of relevance to service providers, including providing insights into the interaction of person, place, social and material circumstances in shaping beliefs and behaviour.
There is also the possibility of different levels of analysis within the one study, drawing on the same data set, as is illustrated by Matoesian and Coldren's (2002) linguistically nuanced CA-informed analysis of focus group data which were also subjected to thematic analysis. Matoesian and Coldren describe the process involved in formulating their analysis for the evaluation report, and translating this into public policy, as 'domesticating' their findings (2002: 471). It is, therefore, possible to present findings in a variety of formats for different audiences (and such possibilities are enhanced by the broader scope afforded by interdisciplinary research teams). Several
studies relating to environmental issues also bridge this gap, addressing contentious political issues and seeking to elicit public responses (Waterton and Wynn, 1999, on views of the nuclear industry; Macnaghten, 2001, on animal experimentation; Collier and Scott, 2010, on industrialized peat extraction) while simultaneously employing some of the tools developed for CA or DA approaches.
INITIAL STEPS IN MAKING SENSE OF FOCUS GROUP DATA
Most of the general advice on analysing qualitative data also pertains to focus groups, although there are some additional challenges and concerns. Some relate to specific focus group usages. Since 'conversation analysis' concentrates on fine-grained analysis of turn-taking, pauses, overlaps of speech, and pauses, it requires that transcriptions be produced according to a specific set of criteria (the Jeffersonian transcription system -see Rapley, 2007: 52-63; also see Kowal and O'Connell, Chapter 5, this volume). This system relies on the use of standardized notation to denote specific, and very detailed, aspects of talk. It allows researchers to take account of such features as the length of silences; the location of micro-pauses; rises and falls in volume (denoted, respectively, by the use of CAPITALS and degree signs); overlaps in participants' talk; faster or slower segments of speech; and even features such as 'sound-stretching' and 'in-breaths' (Rapley, 2007: 60).
Some focus group researchers (e.g. Matoesian and Coldren, 2002) have videotaped discussions in order to ensure that they capture non-verbal communication in addition to talk. They argue: 'an exclusive focus on topic talk ignores the function of the body as it intersects with speech in the conceptualisation of socially embodied action' (Matoesian and Coldren, 2002: 484). However, even when the purpose of producing transcripts is simply to engage in content or thematic analysis, attention to such details can still pay dividends, as
ANALYSING FOCUS GROUPS
317
participants' emphases, tones of voice, facial expressions or gestures can fundamentally alter interpretations of specific statements. The vocabulary of stage directions, borrowed from the theatre, may, on occasion, be more helpful than the standard language employed by methods texts (Barbour, 2008) and field notes are invaluable resources. Interestingly, Matoesian and Coldren, while not overtly adopting a CA approach, have combined many of its features, alongside usage of colourful and extremely detailed descriptions of body language -including terms such as 'lateral head jerk' and 'open palms recoil' (2002: 474).
Producing summaries of discussions, as is routinely done in marketing research, although not a bad starting point, is generally insufficient on its own for analysis of focus groups carried out in a more academic context. A further complication in terms of seeking to summarize complex discussions is that, as Waterton and Wynn (1999) point out, many groups do not reach a consensus. In order to make meaningful comparisons between the content covered in focus groups, however, it is necessary to have information regarding the individuals who have participated. Focus group researchers differ to the extent to which they collect demographic information (such as age or occupation) -with relevant characteristics dependent on the research topic. Usages which rely on snowball sampling (drawing on participants' own networks; see Rapley, Chapter 4, this volume) may not involve recording of such details. Short questionnaires can be extremely helpful in such situations, allowing detailed information to be captured without breaking the flow of discussion or using up valuable discussion time (Barbour, 2007). It is an increasingly common research practice for the principal investigator (i.e. the most senior grant-holder) not to be involved in generating data, although he or she is usually involved in analysis and writing up. In such cases, the additional information possessed by moderators is a valuable resource, leading some commentators to advocate interviewing moderators (Traulsen et al., 2004), or, at least, to involve them
actively as members of the team carrying out data analysis (Barbour, 2007).
One of the hallmarks of qualitative research is its capacity to capture and illuminate context (Barbour, 2008). As pieces of social interaction, focus groups are especially sensitive to and reflective of context. The location of focus group sessions and the associations this has for the group and individuals involved are likely to have an important impact on the discussions. In addition, the composition of the group also influences what is and is not said. It is essential that such information is drawn upon throughout the process of analysis.
Sometimes variation in responses to moderators' questions can alert researchers to important differences between groups. Heikkla (2011) presents findings from a focus group study with Swedish-speaking Finns (themselves a minority) and has analysed these discussions in order to explore the relevance of social class position in relation to their talk about good and bad taste. Heikkla characterizes three clear-cut categories of response to the question 'What do you think good taste is?' These were: (1) astonishment at the difficulty of addressing this question (followed by engagement with the topic); (2) posing of a further question requesting clarification; and, finally, (3) a silence or making a joke. She found that initial responses to the moderator's question broadly prefigured the orientations of the different groups as expressed in the discussion following on from this question, with the first response characterizing the perceptions of upper class groups; the second those of middle-status groups; and the third that of low-status groups. Paying retrospective attention to this sort of patterning can prove to be a valuable aid to analysis.
Heikkla's (2011) work also provides an example of the value of making comparisons between groups. Of course the potential to do this is determined by the attention paid to sampling in formulating the study design. In this case the researchers had convened groups comprising Swedish-speaking Finns with different social class positions (low, middle and high) since they were keen to explore the
318
TYPES OF DATA AND THEIR ANALYSIS
influence of background on ideas about taste. This suggests a broadly sociological orientation, highlighting the importance of disciplinary assumptions in shaping research designs. Heikkla found that discussion in the low-status groups consisted entirely of examples (generally of bad taste) and moral judgements, whereas, in the high-status groups, discussion flowed more freely, probably due to the upper/ upper middle classes' established cultural repertoire, which allowed them to talk more analytically. Of course, researchers' own cultural repertoire and language also frame the way in which they phrase questions and interpret responses. We are sometimes alerted to potentially fruitful lines to pursue in analysis through paying attention to our own reactions to comments that jar with our own understandings and expectations.
The moderator can also play a significant role in shaping data, since participants may react differently to moderators who are or are not perceived to share their own characteristics (and assumed values). This is not an argument for matching moderators and participants; rather a reminder that useful insights may be gleaned by comparing the responses to moderators of differing age, gender, race or ethnicity - among other characteristics - that either pertain to moderators or that are attributed to them by participants (Kitzinger and Barbour, 1999).
The setting where focus groups are carried out can also be a resource for comparison. Green and Hart (1999) used focus groups to study how children's knowledge about accident risks is produced in local contexts. Reflecting on this experience, they highlight the markedly contrasting nature of stories told by children in the classroom (where formal safety messages were emphasised) and in the playground (where risk-taking experiences were recounted - sometimes very dramatically).
The neat - and sometimes overly simplistic -sampling categories (see Rapley, Chapter 4, this volume) we imagine when writing our research proposals are often revealed to be less straightforward once we begin to do our fieldwork.
When planning a study about carer involvement in drug services, a colleague (Orr et al, 2012) decided to hold separate focus groups with carers of drug users, health care professionals and policy-makers, anticipating that there would be important differences in their perceptions. However, on several occasions she discovered that individuals recruited to her health care professionals' groups were also carers of problem drug users, and that some people taking part in carers' groups were also employed within the health or social services sector. Such individuals were frequently an enormous analytic resource, since they were able to comment from more than one perspective and also encouraged other participants in the focus group, who did not have the benefit of these dual identities, to reflect more deeply on the issues being discussed. Further opportunities for comparison can, thus, arise fortuitously and it is important to be alert to such unanticipated bonuses.
FURTHER ANALYTIC RESOURCES
Identification of patterning in data is key to developing explanatory frameworks - that is, paying attention to who says what in which context (Barbour, 2008). Especially important here is critical examination of apparent contradictions or exceptions (as in the approach termed 'analytic induction'). Our explanations can be refined through detailed and systematic analysis of 'confirming' or 'disconfirming' excerpts, taking additional features (e.g. participants' characteristics or focus group settings) into account. For an illustration of the approach of 'analytic induction' in building an explanation from focus group data see Frankland and Bloor (1999) who systematically looked for exceptions in interrogating, building up and continuously modifying their understanding of how peer pressure operated in relation to adolescents' smoking behaviour. As Flick explains, 'analytic induction' is 'a way to take the exceptions as a point of reference rather than the average and normal in the material' and allows researchers to 'further elaborate models' (2007: 32).
ANALYSING FOCUS GROUPS
319
Despite the enormous popularity of 'grounded theory' (see Thornberg and Charmaz, Chapter 11, this volume) as an approach to analysing qualitative data, one of the most under-exploited aspects of Glaser and Strauss's (1967) approach is the notion of returning to the field to generate further data in order to explore emergent and partial hypotheses. It is not necessary to have a grandiose theoretical framework to interrogate, as this can involve little more than a 'hunch'. Hussey et al. (2004) decided to go back into the field to explore whether - as their initial data on doctors' views about issuing sickness certificates suggested - there were important variations in concerns on the part of general practitioners (family physicians) occupying different employment statuses (as locums, registrars and principals). This led these authors to convene another three focus groups which also utilized, as stimulus material, some quotes from earlier focus groups, which allowed emergent hypotheses to be interrogated. A similar approach was employed by Murdoch et al. (2010) who shared data with participants as they sought to develop their analyses.
Although it can be a useful starting point, categorizing individuals in terms of the views they espouse is unlikely to convey the whole story, due to the nuanced and contingent nature of views and perceptions. A detailed examination of the contradictions and shades of meaning conveyed, however, may well go some way towards uncovering the patterns that govern responses - always acknowledging, of course, that such schema are imperfect, provisional, and subject to revision and reformulation as our analyses proceed. A particularly useful resource for analysis is afforded by any tensions and dilemmas reflected in focus group discussions - either as differences of opinion between participants (Farnsworth and Boon, 2010) or as difficulties that are acknowledged and which participants attempt to address collectively. It is not only focus group researchers who 'worry away' at such conceptual puzzles -focus group participants may also charge themselves with this task - and may even 'problematize' our questions and language.
Heikkla (2011), for example, found that some focus group participants found it difficult to define good taste, but spoke at length about bad taste. Paying attention to such 'backhanded' or circuitous ways of discussing specific topics is likely to be fruitful. Silences can also be a valuable resource for further interrogation by the researcher - either by drawing these to the attention of participants and seeking clarification or by subjecting these to detailed analysis. Poland and Pedersen (1998) highlight the potential of what they term 'silences of familiarity', which may escape the attention of the uncritical or unwittingly com-plicit researcher, but which may, nevertheless, be key to understanding the interaction. Moderators may possess - or may acquire along the way - valuable 'insider' knowledge of the unspoken rules governing behaviour, such as conversational turn-taking. For example, when carrying out focus groups with Pacific North West Indian people, Strickland (1999) noted that elders were always allocated the final words in any discussion, but never contributed until that point was reached.
In analysing focus group data researchers should seek to maintain a critical or sceptical focus with regard to what focus group participants say, bearing in mind the potential provided by this setting for self-presentation, offering what Brannen and Pattman (2005) refer to as a 'site of performativity'. It is important to guard against the dangers of taking participants' comments too literally. An example is provided by a recently completed study. This was a health services research project located at the applied end of the research spectrum, which was carried out to inform development of a weight-loss intervention package for women following childbirth. In their discussions women interrogated the 'ideal' of weight loss and were often critical of received health promotion 'wisdom'.
One of the groups, for example, engaged in a lengthy and jointly constructed explanation as to why the weather had a big impact on their ability to address weight management. What is achieved in such exchanges is relatively complex, in that the women, themselves, are aware
320
TYPES OF DATA AND THEIR ANALYSIS
of the justificatory nature of their talk, which is often punctuated by admissions. This was the case with the following excerpt coded under our  'in-vivo' (see Thornberg and
Excerpt 1
Jen
Sally
Jen Sally Jen Sally
Jen
(Post-partum Weight Management Study — Focus Group 4)
(Transcriber's description in italics; Researcher's emphasis in bold.)
(The ellipsis '...' in a block of text denotes a short pause or speech tailing off and '...' between lines indicates that some text has been omitted from the quote in the interests of brevity.)
DIVERSE INSIGHTS: THE CASE FOR A COMPOSITE APPROACH?
Halkier (2010) makes a case for employing a range of tools, derived, variously, from the work of Goffman (1981), conversation analysis (see Toerien, Chapter 22, this volume), discourse psychology (see Willig, Chapter 23, this volume) and positioning theory, and merges this assembly of approaches in which she calls 'a practice-theoretical perspective' or, perhaps more illuminatingly, a 'moderate social constructivist view'. She provides a helpful - but not overly prescriptive - set of
Charmaz, Chapter 11, this volume) code of 'heavy bones' - that is, a code derived from the ideas of those being researched (Kelle, 1997):
suggestions with regard to how to go about analysing focus group data. Essentially, this paves the way for analyses that combine a focus on topic, form and structure of talk.
The reference to Goffman reflects the importance of the performative aspects of interaction, with focus groups viewed as a stage where participants tell, negotiate and reformulate their 'self-narratives'. According to this formulation, focus group participants are engaged not just in presenting their own narratives, but in supporting or challenging others' narratives, forging, testing and occasionally repairing relationships along the way, and, sometimes, in co-constructing accounts.
According to Halkier's approach, it is possible to see how participants draw on strategies - frequently those identified in CA approaches - in order to strengthen the claims that they are making in discussions. Halkier demonstrates how 'positioning analysis' can alert the researcher to the stance that particular focus group participants are affecting, which, of course, aids in interpreting the comments made and the effect that is desired. Halkier also points out that focus group members may also seek to position others in the group,
... Your actual weight I don't think is as important as what clothes are fitting. Because some people can be heavier than others due to heavy bones or, you know. And people often ... even when I went to Weightwatchers the lady used to say to me, I can't believe you actually weigh that, because I must be quite heavy inside. Because you're obviously fitting into a size 12 (US size 10, European size 40) pair of trousers but you can be a lot heavier than another size 12. Hazel was speaking about that the other day as well. Because I was saying, if I never ate for like five years I would never be eight stone (US 112 pounds; European 44.8 kilos). No, well that's it. You're kind of built either ...
I've just... I've never been that, I don't think since ... I can't remember ever being, like, that size. No, neither can I.
I obviously was at one point as a child. But as a grown up person ...
I was like, "What?" I knew I was a bit heavier, but I wouldn't have said ... I didn't feel like I was particularly unfit. So I feel it's how you ... what clothes you fit into rather than your actual weight. Some people are bigger boned than others.
ANALYSING FOCUS GROUPS
321
through overt challenges that question their spectives on post-partum weight loss shows self-presentations. The following excerpt three women negotiating around the first from our focus group study of women's per-     speaker's self-presentation:
Excerpt 2
Veronica I think the way I seem to have lost weight is because, not what I'm eating, but trying to educate myself to think about healthy eating because ... My kids don't get chocolate. Well, they get lots of crisps and biscuits because Daddy eats them.
Helen      See crisps and biscuits are just as bad as chocolate.
Eileen      But kids run around and burn it off, you know ... 'can I go out to play, yeah?' ... 'bye'. Three
hours later they come in. Helen      But if it's not in your cupboard, you're not going to eat it.
Veronica Yeah, yeah, if I don't eat it, they don't tend to get it. I go for the healthy things and I think that's how I've lost weight ... apart from running after them!
(Focus Group 3)
Veronica starts by sharing her experience of trying to put into practice what she has learnt about healthy eating. Helen, however, is quick to point out the flaw in this approach, while Eileen comes in to defend Veronica. Undeterred, Helen reiterates her point about
Excerpt 3
Veronica
Nan
Eileen Nan Nan
Eileen
(Focus Group 3 - Underlining denotes emphasis in the original.)
Another strategy highlighted by Halkier (2010) -that of 'category entitlement' - involves making an appeal based on personal experience and knowledge in order to authenticate a specific comment or perspective. This is what the women in these excerpts are appealing to when they invoke the demands of parenting and domestic
restricting availability of food that is deemed unhealthy. Veronica chooses not to react to Helen's further challenge and returns to her initial topic of how she has achieved some modest success with regard to healthy eating and weight loss. Eileen then joins in to locate the issue within the context of busy family life and this idea is echoed by others:
responsibilities in justifying their disregard of dietary and dieting advice. Veronica can be seen starting to make claims about the impact of knowledge, which she then, in the next breath, goes on to question, appealing to the demands of child-rearing.
As Halkier's examples (provided in her paper) show, language selection is far from accidental and such strategies tend to involve the use of particular linguistic appeals, such
I think if the mums are provided with, say, an idea of what's healthy and what isn't ... and then again who wants to sit and read healthy when they've got kids? On a Friday night after baby's gone to bed sometimes you need that little bit of chocolate cake, or that naughty bag of crisps [voices overlap] You've got to have something.
All the diet groups say that you shouldn't ever deprive yourself or go hungry. Have you tried that Skinny Cow chocolate fudge brownie ice cream by Ben and Jerry's? And it was wicked, I felt naughty eating it, but it tasted so good [voices overlap] ... Oh, I love [voices overlap] ...
322
TYPES OF DATA AND THEIR ANALYSIS
as using the term 'you' to appeal to shared views and experiences. Again, this usage can be seen in the previous excerpts, conveying solidarity and shared assumptions.
Halkier (2010) also recommends that we analyse our focus group transcripts by looking for instances of strategies such as 'factist characterised descriptions and evaluations', whereby
Excerpt 4
Debbie
Kim
Laura Debbie
Laura Debbie
(Focus Group 1 - Researcher's emphasis in bold; Underlining denotes emphasis in the original.)
At first glance this looks like just another challenge to received wisdom. However, what is striking about this example is the shift in register from Debbie's vernacular 'they can stick it where the sun don't shine!' to her 'It's extremely unrealistic ...' and all in the space of a few lines. Interestingly, here she appears to be invoking a 'factist' style to challenge received wisdom -seeking to 'have it both ways', in fact.
These excerpts show the women employing all of the strategies outlined by Halkier (2010),
Excerpt 5
personal opinions are presented as 'shared by most people' or as 'common knowledge' (as with Veronica's initial utterance in Excerpt 3).
Another group in the same study expanded their critique to challenge the accuracy of the Body Mass Index (BMI) charts routinely used by health professionals to determine target weights:
ranging from 'factisf displays of knowledge, with these being set up only to be brought into question by 'positioning' strategies, with 'category entitlement' being invoked through the power of personal experience (of various types).
As outlined earlier, discussions about weight management could be subversive in focus, with women making wry references to 'cheating', focusing on the fruit content in high-calorie foods, or confessing to piling 'portion plates' (designed to aid portion control) as high as possible (thereby subverting their purpose). Sometimes the women appear to be vying with each other, telling funny stories (and the talk is punctuated by shared laughter):
It was because the target weight they had that, that was 'overweight' and the target weight that I arn is classified as 'obese'.
That's like me when I got it done. It said that I was obese and I looked at myself and thought 'eh? that can't be'.
Yeah, I'd be quite happy to be 'overweight' in their categories. [She laughs] The target weight for my height was about nine and a half stone (US 133 pounds, European 53.2 kilos) and I just thought, 'Do you know what? There's no way that I'm going to get down to there so they can stick it where the sun don't shine!'
Mine was something like seven and a half stone (US 105 pounds, European 42 kilos) and I was like no way, I was maybe that when I was at school. Sorry ...
It's extremely unrealistic the actual BMI, it just was not achievable ... yeah, it just seemed so unachievable that it didn't matter, like ideally I'd like to be about ten and a half (US 147 pounds, European 58.8 kilos) maybe eleven stone (US 154 pounds, European 61.6 kilos), and to even be that and still be told you're obese ...
Nan You see, I'm really evil because my cheese sits in the fridge and I've got written on it, 'Nan's diet cheese', so as my husband doesn't like strong cheese, he likes, like, double Gloucester and all these cheeses. I like cheese that you [voices overlap] and I write, like, 'Diet cheese', but inside it's like seriously strong stuff, and he thinks it's diet so he never touches it ... [Several snorts and laughter from other participants] I just put 'Diet' on something even though it's not, he goes, 'Oh, that's my wife's diet stuff, I'd better not touch it', and I get it. [Prolonged laughter] Oh, I come from an Italian family - I'm not stupid ... I let him think I am.
Alison Yeah, they've got to think you're a little bit stupid.
[Nods from some of the others.]
ANALYSING FOCUS GROUPS
323
(Focus Group 3)
The shared laughter here affirms common experiences in relation to struggles with weight and dieting and also acknowledges complicity in 'managing' male partners.
The following excerpt, produced in response to the moderator's question about how best to approach the topic of weight
Excerpt 6
Rose
Nan Eileen Veronica Nan Nan
Eileen Nan
Nan
All Nan
Veronica
(Focus Group 3)
A sense of anger and hurt pervades these discussions with overweight women who consider themselves a beleaguered minority in a world that emphasizes a narrow vision of attractiveness to which they do not conform. The hilarity produced in this discussion echoes Jefferson's (1984) observations about the important role of laughter in talk about 'troubles' and the analysis might well benefit from paying more detailed attention to how laughter is ordered and structured - as Jefferson suggests.
The analytic strategies recommended by Halkier (2010) can certainly produce useful insights into the intent and effect of conversational gambits and exchanges and, ultimately, what is attempted or achieved by participants in the course of a focus group exchange. It has been extremely helpful in looking beneath the
loss, provides a clue with regard to interpreting the comments about BMI made earlier in this group and in other focus group discussions, including the 'they can stick it where the sun don't shine' comment. Here the exchange resembles a comedy improvisation 'riff as the women build on each other's comments to humorous effect:
surface of the 'plucky' talk produced by the women in the post-partum weight management study focus groups. Whether or not it is necessary to label strategies in the ways suggested by Halkier (2010) in order for our analyses to derive benefit is another matter. There is much to recommend in terms of paying attention to such strategies, where this helps to explain analytically troublesome or potentially rich exchanges, although one might stop short of routinely documenting and interrogating all instances that occur throughout focus group discussions. As Halkier, herself, concedes, 'just like pure content analysis of focus group data is relatively uninteresting, and does not take the specific methodological strengths of this kind of data seriously, likewise pure interaction form analysis is a methodological dead-end for most social scientific
Okay. So thinking about just to round off, I think we've just about got through all of our questions here. What advice would you give to us really in terms of developing a weight management intervention? Tread gently. Very gently. Don't be pushy. Or patronizing.
Because we know we're overweight but we just don't need you telling us we are, we've got mirrors in the house as well. [Several affirmative head shakes] And we're not dense.
We know what vegetables and fruit are, we know we should eat them, but at the end of
the day a KitKat (UK manufactured branded chocolate wafer bar) is easier to get through
than an orange. Orange is like 'aarg!' but a KitKat - done; gone. [Laughter]
A lot of it is more time management - the convenience of it ... If you've got a KitKat you
think, 'Well, I should have that lovely fresh healthy orange, but, bugger it, I haven't got time,
and it gets sticky ... wash my hands ...' the KitKat - done ... Yeah, peel it, and wash your
hands, and change your top ...
Yeah. [Accompanied by nods and smiles]
And then you get hacked off about it and think, you know, I still fancy eating a KitKat [Several nods and laughter]
I think you can get orange KitKats now too [Voices overlap amidst lots of laughter]
324
TYPES OF DATA AND THEIR ANALYSIS
uses of focus groups' (2010: 86). Ultimately, the choice for the data analyst will be governed by the aims of the research and the audience for whom the analysis is to be produced.
NEW DEVELOPMENTS AND PERSPECTIVES
New developments - particularly those afforded by the Internet - bring tantalizing new possibilities, but also new challenges, in terms of both generating and analysing focus group data. Online discussion forums (see Marotzki et al., Chapter 31, this volume) are often considered to be in the public domain and thus are likely to be seen as providing ready-transcribed data. Since such forums have an existence independent of the research being carried out, they are also attractive to those who are concerned about the effect of the moderator on the data generated. 'Harvesting' such ready-made data, however, brings its own challenges - in addition to ethical issues - including lack of researcher control over selection of participants, or even access to demographic information which might be useful in analysis. Although asynchronous formats (with a delay between successive 'postings') potentially allow the moderator to ask questions or seek clarification, synchronous (i.e. real-time) discussions do not afford such opportunities. Commentators such as Stewart and Williams (2005) highlight the need for focus group researchers to develop new techniques in response to such challenges, including exploiting the analytic potential of 'emoticons' (symbols as used in texting, e.g. :-) to denote a happy face) (Fox et al., 2007).
Seale et al. also point out that text produced via online forums is 'grammatically and lexically less dense than written language and is often unedited, with numerous contradictions of words and uncorrected typing or punctuation oddities that contribute to the style of this mode of communication in informal context' (2010: 596). This raises several problems with regard to interpretation.
The internet, however, also offers some new approaches to analysing these new forms of data. Computerized data analysis software packages such as NVivo (see Gibbs, Chapter 19, this volume) offer the possibility of counting word frequency and it is possible to customize such searches. A feature of Web 2.0 websites and blogs (see Marotzki et al., Chapter 31, this volume) is what is termed 'tag clouds', which is an approach borrowed from the visual design field and which allows for a visual depiction (utilizing different font sizes and colours) of the relative frequency of selected terms and concepts. Although this makes for arresting graphic displays it is more difficult to make analytical use of these. Social networking sites, such as Facebook, also provide their own network analysis tools, but, similarly, this has produced complex diagrams which, so far, have proved somewhat resistant to incorporation into in-depth analysis. While such tools can aid in establishing broad patterns, a more nuanced analysis of complex texts is still likely to demand yet more sophisticated methods to allow for extrapolation between visual display and explanatory frameworks.
Interestingly, new applications in marketing research have also been focusing on how to capture and use information relating to word frequencies in analysis of focus group data. Reviewing these developments, Schmidt (2010) argues that commercial software can identify 'rule based webs' - of associations between words. Seale et al. (2010) have simultaneously been exploring the potential of keyword analysis for online data. Their approach, of 'keyword analysis', relies on the comparative analysis of two texts, but they acknowledge that this can capture a wealth of information that may, in the event, be largely irrelevant for the research question being addressed. They argue that, ultimately, a qualitative judgement has to be made with regard to choosing the keywords that 'best bring out the characteristics of a particular text' (Seale et al., 2010: 598). There are likely to be significant challenges in terms of utilizing such methods to explore ironic or strategic use of words,
ANALYSING FOCUS GROUPS
325
imagery and metaphors. While it is important to keep a weather eye on new developments, the original focus of the research and the disciplinary and theoretical persuasion of the researcher or research team remain the key to deciding what does and does not work. There is still no substitute for thoughtful research design, and imaginative, but attentive and thorough analysis.
FURTHER READING
Barbour, Rosaline (2007) Doing Focus Groups. London: Sage.
Frankland, Jane and Bloor, Michael (1999) 'Some issues arising in the systematic analysis of focus group materials', in Rosaline S. Barbour and Jenny Kitzinger (eds), Developing Focus Group Research: Politics, Theory and Practice. London: Sage. pp. 144-55.
Macnaghten, Phil and Myers, Greg (2004) 'Focus groups', in Clive Seale et al. (eds), Qualitative Research Practice. London: Sage. pp. 65-79.
REFERENCES
Angus, Jan, Rukholm, Ellen, St. Onge, Renee, Michel, Isabelle, Nolan, Robert P., Lapum, Jennifer and Evans, Sarah (2007) 'Habitus, stress and the body: The everyday production of health and cardiovascular risk', Qualitative Health Research, 17: 1088-1102.
Barbour, Rosaline (2007) Doing Focus Groups. London: Sage.
Barbour, Rosaline (2008, (2nd edition in press)) Introducing Qualitative Research: A Student Guide to the Craft of Qualitative Research. London: Sage.
Bhaskar, Roy (1989) Reclaiming Reality. London: Verso. Bourdieu, Pierre (1999) Culture of a Theory of Practice.
Cambridge: Cambridge University Press. Brannen, Julia and Pattman, Rob (2005) 'Work-family
matters in the workplace: The use of focus groups in
a study of a UK social services department',
Qualitative Research, 5 (4): 523-42. Callaghan, Gill (2005) 'Accessing habitus: Relating
structure and agency through focus group research',
Sociological Research Online, 10 (3): (accessed 9
May 2013).
Collier, Marcus J. and Scott, Mark (2010) 'Focus group discourses in a mined landscape', Land Use Policy, 27: 304-12.
Farnsworth, John and Boon, Bronwyn (2010) 'Analysing group dynamics within the focus group', Qualitative Research, 10 (5): 605-24.
Flick, Uwe (2007) Managing the Quality of Qualitative Research. London: Sage.
Fox, Fiona E., Morris, Marianne and Rumsey, Nichola (2007) 'Doing synchronous online focus groups with young people: Methodological reflections', Qualitative Health Research, 17: 539-47.
Frankland, Jane and Bloor, Michael (1999) 'Some issues arising in the systematic analysis of focus group materials', in Rosaline S. Barbour and Jenny Kitzinger (eds), Developing Focus Group Research: Politics, Theory and Practice. London: Sage. pp.144-55.
Glaser, Barney and Strauss, Anselm (1967) The Discovery of Grounded Theory. Chicago: Aldine.
Goffman, Erving (1981) Forms of Talk. Oxford: Basil Blackwell.
Green, Judith and Hart, Laura (1999) 'The impact of context on data', in Rosaline S. Barbour and Jenny Kitzinger (eds), Developing Focus Group Research: Politics, Theory and Practice. London: Sage, pp.21-35.
Halkier, Bente (2010) 'Focus groups as social enactments: Integrating interaction and content in the analysis of focus group data', Qualitative Research, 10(1): 71-89.
Hammersley, Martyn (1992) What's Wrong with Ethnography? Methodological Explorations. London: Routledge.
Heikkla, Riie (2011) 'Maters of taste? Conceptions of good and bad taste in focus groups with Swedish-speaking Finns', European Journal of Cultural Studies, 14(1): 41-61.
Hepburn, Alexa and Potter, Jonathan (2004) 'Discourse analytic practice', in Clive Seale etal. (eds), Qualitative Research Practice. London: Sage. pp. 180-96.
Hilsen, Anne Inge (2006) 'And they shall be known by their deeds: Ethics and politics in action research', Action Research, 4(1): 23-36.
Hussey, Susan, Hoddinott, Pat, Wilson, Phillip, Dowell, Jon and Barbour, Rosaline (2004) 'The sickness certification system in the UK: Qualitative study of views of general practitioners in Scotland', British Medical Journal, 328: 88-92.
Jefferson, Gail (1984) 'On the organization of laughter in talk about troubles', in J. Maxwell Atkinson and John Heritage (eds), Structures of Social Action: Studies in Conversation Analysis. Cambridge: Cambridge University Press, pp. 346-69.
Kelle, Udo (1997) 'Theory building in qualitative research and computer programs for the management of textual data', Sociological Research Online, 2(1): (accessed 9 May 2013).
326
TYPES OF DATA AND THEIR ANALYSIS
Kidd, Pamela S. and Parshall, Mark B. (2000) 'Getting the focus and the group: Enhancing analytical rigour in focus group research', Qualitative Health Research, 9 (3): 293-308.
Kitzinger, Jenny and Barbour, Rosaline S. (1999) 'Introduction: The challenge and promise of focus groups', in Rosaline S. Barbour and Jenny Kitzinger (eds), Developing Focus Group Research: Politics, Theory and Practice. London: Sage. pp. 1-20.
Macnaghten, Phil (2001) Animal Futures: Public Attitudes and Sensibilities Towards Animals and Biotechnology in Contemporary Britain. London: Agriculture and Environment Biotechnology Commission.
Macnaghten, Phil and Myers, Greg (2004) 'Focus groups', in Clive Seale et al. (eds), Qualitative Research Practice. London: Sage. pp. 65-79.
Makosky Daley, C, James, Aimee S., Ulrey, Ezekiel, Joseph, Stephanie, Talawyma, Angelina, Choi, Won S., Greiner, K. Allen and Coe, M. Kathryn (2010) 'Using focus groups in community-based participatory research: Challenges and resolutions', Qualitative Health Research, 20 (5): 697-706.
Matoesian, Gregory M. and Coldren, James R. (2002) language and bodily conduct in focus group evaluations of legal policy', Discourse and Society, 13 (4): 469-93.
Maxwell, Joseph A. (2011) A Realist Approach for Qualitative Research. Thousand Oaks, CA: Sage.
Morgan, David. L. (2010) 'Reconsidering the role of interaction in analysing and reporting focus groups', Qualitative Health Research, 20 (5): 718-22.
Murdoch, Jamie, Poland, Fiona and Salter, Charlotte (2010) 'Analysing interactional contexts in a data-sharing focus group', Qualitative Health Research, 20 (5): 582-94.
Myers, Greg and Macnaghten, Phil (1999) 'Can focus groups be analysed as talk?', in Rosaline S. Barbour and Jenny Kitzinger (eds), Developing Focus Group Research: Politics, Theory and Practice. London: Sage. pp. 173-85.
Orr, Linda C, Barbour, Rosaline S. and Elliott, Lawrie (2012) 'Carer involvement with drug services: A qualitative study', Health Expectations, DOI: 10.1111/hex. 12033.
Poland, Blake and Pederson, Ann (1998) 'Reading between the lines: Interpreting silences in qualitative research', Qualitative Inquiry, 4 (2): 293-312.
Puchta, Claudia and Potter, Jonathan (2004) Focus Group Practice. London: Sage.
Rapley, Tim (2007) Doing Conversation, Discourse and Document Analysis. London: Sage.
Schmidt, Marcus (2010) 'Quantification of transcripts from depth interviews, open-ended responses and focus groups', International Journal of Market Research, 52 (4): 483-509.
Seale, Clive, Charteris-Black, Jonathan, MacFarlane, Aidan and McPherson, Ann (2010) 'Interviews and internet forums: A comparison of two sources of qualitative data', Qualitative Health Research, 20 (5): 595-606.
Stewart, Kate and Williams, Matthew (2005) 'Researching online populations: The use of online focus groups for social research', Qualitative Research, 5 (4): 395-416.
Strickland, C. June (1999) 'Conducting focus groups cross-culturally: Experiences with Pacific Northwest Indian people', Public Health Nursing, 16 (3): 190-7.
Trauisen, Janine Morgell, Almarsdöttir, Anna Birna and Björnsdöttir, Ingunn (2004) 'Interviewing the moderator: An ancillary method to focus groups', Qualitative Health Research, 14 (5): 714-25.
Vincent, Deborah, Clark, Lauren, Zimmer, Lorena Marquez and Sanchez, Jessica (2006) 'Using focus groups to develop a culturally competent diabetes self-management programme for Mexican Americans', The Diabetes Educator, 32: 89-97.
Waterton, Claire and Wynne, Brian (1999) 'Can focus groups access community views?', in Rosaline S. Barbour and Jenny Kitzinger (eds), Developing Focus Group Research: Politics, Theory and Practice. London: Sage. pp. 127-43.
Wilkinson, Sue (1998) 'Focus group methodology: A review', International Journal of Social Research Methodology, 1 (3): 181-203.
Willig, Carla (2003) 'Discourse analysis', in Jonathan A. Smith (ed.), Qualitative Psychology: A Practical Guide to Research Methods. London: Sage. pp. 159-83.