Abstract
In the light of the changing landscape of social research, this article explores the role of the analytic imagination in the process of qualitative data analysis. It argues that while team research, secondary data analysis and the use of computerized qualitative data analysis packages may be altering the ways in which research and analysis are carried out, this need not change the processes of interpretation that are at the heart of qualitative data analysis. Here, as the article explores, imaginative acts are key to the analytical craftsmanship involved in interpretive analysis. This a process illustrated through the analysis of parent and child narratives gathered during a project about families and food.
Keywords
Introduction
In 1973, in his famous essay ‘Thick description: toward an interpretive theory of culture’, the anthropologist Clifford Geertz argued that interpretations ‘must be cast in terms of the interpretations to which particular persons of a particular denomination subject their experiences, because that is what they profess to be descriptions of’ (1973: 15). By this he suggested that sticking close to the texture of everyday life, rather than indulging in abstract speculation, is key to the interpretive act. And so he talked of ‘inspecting events’ (p. 17), examining the ‘flow of social discourse’ (p. 20) and the inevitable ‘awkward fumbling’ (p. 25) that occurs when anthropologists try to make ‘constructions of other people’s constructions of what they and their compatriots are up to’ (1973: 9). For him, interpretive analysis involved ‘sorting out the structures of signification’ (p. 9) in order to reduce ‘the puzzlement’ (p. 16) involved in understanding the lives of other people while, at the same time, acknowledging that ‘it is not necessary to know everything in order to understand something’ (p. 20).
In Geertz’s essay, then, one senses the craftwork that is necessarily involved in the process of interpretation. Indeed, Geertz himself acknowledged that it is the power of what he called ‘the scientific imagination [that] brings us into touch with the lives of strangers’ (1973: 16). This calls to mind the crafting process that C. Wright Mills also saw as critical to data analysis. In The Sociological Imagination, he writes of the skill, rather than simply the technique, that is needed to carry out sociological research. It also echoes Atkinson’s (1990) claim, in The Ethnographic Imagination, that processes of interpretation and the rhetorics of persuasion are core to the production of sociological accounts. It is the significance and importance of these techniques of crafting that I want to explore in this article through demonstrating the workings of what I want to call, the analytic imagination.
This concept raises two key issues about the process of data analysis. First, it suggests that the sharp edges of analysis, which appear to offer rigour to the process of understanding qualitative data, might also be open to acts of imagination – free spirits shaking the analytical scaffolds of processes of data analysis. Second, it draws attention to the creativity that emerges in the spaces and times between the ‘doing’ of research, the coding of data and its subsequent writing up. In comparison to the amount of words that, over the years, have been dedicated to the ‘how to’ elements of the research process – how to design research, how to do this method or that, and how to organize one’s data ready for analysis – curiously, relatively little recent attention has been given to considering what is actually involved in the processes of interpretation. This, then, is the main focus of this article and, in its third part, I reveal the workings of the analytic imagination through an extended analysis of parents’ and children’s narratives. First, however, I explore the background to this urgent need to understand what is going on in the process of interpretation.
Part 1: background noises
My interest in unpicking the acts of interpretation that constitute data analysis has been provoked by a number of changes that have occurred in recent years in the landscape of qualitative research. First, are the economic pressures on research funding, pressures that are gradually re-shaping the social relationships of research, especially in the UK but also elsewhere (for an overview see Denzin and Lincoln, 2005). It is becoming, for example, increasingly difficult for lone researchers, whether anthropologists or sociologists, to continue to go ‘to the field’ in person to collect data. With more emphasis now being given to the creation of larger research teams, combined with the increasing use of assistants to do the primary data collection, the data that a researcher ends up analysing may not have been gathered by him/herself (Wuchty et al., 2007). Having been generated by others – perhaps by post-doctoral students or by research assistants on short-term contracts – the data have been filtered through these people’s experiences and actions in the field. The grant holder who has overall responsibility for analysing the field data may therefore have had little, if any, personal involvement in the process of data production itself. With that experiential context lacking, I am interested in how – at the moment of analysis – the researcher begins to engage with those data in order to produce an analytic account.
Second, the economic pressures on primary research funding have had other consequences for sociological research. They have drawn attention to the large body of data that already exists in archives and which is ripe for secondary analysis or data re-use. These data are even more second-hand, having been collected some time ago by others, asking particular questions about the social world; indeed, they have often already been analysed and certain conclusions reached. What happens in the interpretive process here?
A third and rather different concern also haunts this article. It stems from the increasing use of computer-based systems for the analysis of qualitative data. While clearly they offer tremendous scope for handling large bodies of data, the fear is that they also run the risk of removing the craft element that, in my view, is an essential component of the analytic imagination that lies at the heart of interpretation. The crafting of data into a coherent account of other lives and worlds is, as Geertz indicates, what data analysis is supposed to be about. Might our increasing reliance on computerized packages be turning us from crafts people into mere technicians? (see below and Mills, 1959; Sennett, 2008).
In the sections of this article that follow, therefore, I first explore each of these current issues in turn to discover what they reveal about the process of data analysis. Then, by drawing on an extended empirical example, I want to demonstrate that despite these changes in the research landscape and in the methods by which we gather and organize qualitative data, the crafting skills involved in interpreting these data are as necessary as they always were. And, these skills are core to the analytic imagination.
Not being there
In her handbook written for novice researchers, Mason (2002) notes that core to the analysis of much qualitative research is the interpretive reading of data – making ‘a version of what you think the data mean or represent, or what you think you can infer from them’ (2002: 149). As Mason notes, this often involves the process of reflexivity:
a reflexive reading will locate you as part of the data that you have generated and will seek to explore your role and perspective in the process of generation and interpretation of data, You will probably see yourself as inevitably and inextricably implicated in the data generation and interpretation processes, and you will therefore seek a reading of data which captures or expresses those relationships. (2002: 149)
Indeed, for anthropologists, such reflexivity is traditionally regarded as such a fundamental part of the ethnographic tradition of participant observation that reflexive explorations of ‘being there’ (Watson, 1999) – understanding the role and experiences of the self in the field – are entrenched as a core part of anthropological praxis. It is claimed that ‘being there’ – participating and observing –offers the fieldworker vital experiences of immersion, the ‘physical emotional and psychological experience of being in the field that gives that time a unique quality’ (Watson, 1999: 2). It also creates a sense of otherness – the possibility of ‘being simultaneously ‘in’ but not ‘of’ the other culture’ (Watson, 1999: 2). This sensibility brings a kind of heightened awareness that, it is argued, is fundamental to the process of interpreting the everyday lives of other people.
And now that the one-off interview, whether formal or informal, has even become a recognized part of the armoury of anthropological fieldwork methods (yet another indicator of the changing landscape of research) the being-there-ness of the interviewer is also being emphasized. Hockey (2002), for example, has shown how the embodied experience of doing interviews – entering into people’s homes and gaining an awareness of their material surroundings combined with the sensory experience of sharing in people’s emotional reflections as they talk – adds an additional powerful layer of data for reflection, beyond the spoken words of the interview itself. Indeed, within sociology, the ways in which power and authority relations – the result of gender, age, class or ethnic differences and similarities – can shape an interview has long been recognized as core to the reflexive interpretation of qualitative data (Finch, 1984; Graham, 1984; Scott, 1984).
In both traditional ethnographic fieldwork and in interviews, then, the importance of the contextual experience – the being-there-ness of the researcher – for the process of data analysis is highlighted. Half-forgotten memories of carrying out the interview, sensory recall of an ethnographic context or the implicit knowledge gained through participating in the social worlds of other people can help fill in the missing pieces of the intellectual puzzle that data analysis brings to the fore (Mason, 2002: 149). What happens, then, when the researcher is not there, when their engagement with ‘the field’ is potentially reduced to shuffling pieces of paper as, in the office, they read through transcripts of interviews done by others on their behalf? Are the invaluable insights gained from immersion in the field or reflexive readings of data simply lost to the potential detriment of the research?
An earlier article argued that they are not (Author A and others, 1996). Indeed, it suggested that, through the continual dialogue that is involved in translating ‘the field’ for those who are not there by those who are there, even greater clarity and deeper insights can be engendered in team-based research (see also Gudeman and Rivera, 1990). Such collaborative exchanges can help with the process of interpretation. Indeed, new forms of reflexive readings can potentially emerge precisely through not being there, for there is some value to be gleaned from the anthropological strangeness that unfamiliar data produces in the reader. Moreover, since it is clear that many excellent research accounts are produced by research teams, and have been so for many years (Wuchty et al., 2007) interesting questions arise about what – precisely – is going on therefore during the process of interpreting data?
Is being there always best?
The privileging of contextual experience has been recently thrown into even sharper relief by debates about the value – indeed, the possibility – of carrying out secondary data analysis. Sometimes referred to as data re-use, this methodological practice has been made possible by the archiving of interview data sets 1 and has given rise to a divergence of opinion about the richness of such archives for social researchers, given the potential problems involved in their use.
On the one hand, in answer to the question – is being there always best – Mauthner et al. (1998), for example, would say, firmly, ‘yes’. They argue that the secondary analysis of interview data is a fool’s errand since – even in relation to the later re-analysis of one’s own interview data – the vitality of the experiential moment always informs the later process of analysis and interpretation. This can never be recovered by proxy. Even when there are extensive fieldnotes and personal reflections accompanying a data set, for Mauthner and colleagues, the implicit context – part of what Ottenberg (1990) calls ‘headnotes’ – remains critical; it is an essential part of the co-construction of interview data that needs always to be reflexively analysed.
Fielding (2004), on the other hand, rejects this view. He argues that this is largely a practical, rather than an epistemological, problem since any process of data analysis (whether of primary or secondary data sets) always requires reflexivity: ‘qualitative researchers have always been in the position of having to weigh the evidence’ he says (2004: 99). During the secondary analysis of data sets, the absence of such implicit knowledge just means that achieving reflexivity is a bit more of a problem. On the plus side, however, Fielding sees some additional virtue in secondary data analysis since, he says, ‘primary data analysis is always subject to the problem that researchers will have entered the field and collected their data with particular interests in mind’ (2004: 100). When using second-hand data, however, this problem disappears. The interview transcripts already exist, to be used as a data set for answering different questions. Or, indeed, they can be used to re-evaluate the interpretations that have already been made by a previous research team (see e.g. Savage, 2005).
And, as Fielding goes on to note, indeed such distancing practices also take place in some forms of primary analysis: for instance, in qualitative evaluation research, precisely ‘to overcome affinities developed in the field, the fieldworker hands over the data to a second team member who will carry out the analysis’ (2004: 100). In this kind of research, then, the head notes and the field interviewer’s tacit knowledge are deliberately avoided, being construed as a hindrance, rather than a help, in the analytic process.
There are, then, some clear parallels between the experiences of the secondary data user and team researchers who analyse interviews that they themselves have not carried out. Like the historian, both are delving into archival memories lodged by unknown others; and, for both, the subjective, contextual knowledge is missing. However, whether or not this absence is regarded as problematic hinges on what is held to constitute ‘data’ and how those data are to be interpreted.
In this respect, Hammersley (2010) makes an interesting distinction between ‘data’ – the stuff that is generated as a resource through fieldwork (e.g. interview transcripts or fieldnotes) and ‘evidence’ – the narrow selection that is later made from that data set during the process of analysis. This is used to make inferences in relation to the particular research questions posed. Rarely is all the data generated by a research project used; arguments and conclusions are, therefore, in this sense, always partial.
But Hammersley makes a further distinction. The first kind of data, he says, is in many ways given. As honest social researchers we don’t tend to make up our data; instead, we write down what our informants tell us or what we have seen in the field (see Geertz, 1973). In doing so, we try to offer a faithful account of the world that is out there, albeit an account that is shaped by our questions and interests and by the conditions of the data generation process itself (Mason, 2002). These data are in this sense always constructed. For Hammersley, however, evidence is also constructed, or in his words re-formed, during the analytic process. Data are used, he says, ‘selectively and deliberately . . . to generate evidence’ in order to construct a particular argument. But although a distinction can therefore be made between, as it were, initial raw data and the particular data that is used as evidence, both continue to speak to each other throughout the process of interpretation since ‘we are reforming something that already exists, not making it up’. Thus, placing a marker in the ground against those who would argue for either an extreme social constructionist or extreme empiricist position, Hammersley (2010) concludes as follows:
Data are, then, in an important sense given as well as constructed; they are not created out of nothing in the research process, nor should we construct whatever inferences we wish on the basis of them.
In agreement with Hammersely, I suggest, therefore, that data re-use is eminently do-able, with the ‘problem’ of context being more – or less – troublesome depending on for what purposes the secondary data analysis is taking place (see also Moore, 2007).
In sum: the lesson I take from these discussions is that the issue about not-being-there is more a matter of degree than of kind. As Hammersley acknowledges, even ‘having been a participant in an interview does not mean that one has an exhaustive knowledge of what occurred’. Being-there does not give unbridled access to some special kind of unequivocal ‘truth’. And, I would add, neither does it mean that the interpretations made of primary data are necessarily more robust than those made of secondary data; they are simply different. And they are different, as I shall demonstrate later, precisely because of the ways in which the analytic imagination intervenes.
Computing an account?
The third prompt to seeking to understand exactly what goes on in the process of data analysis is, as I noted at the outset, the increasing use of qualitative data analysis packages by researchers. As Mason notes, ‘a whole industry has sprung up . . . [that] both facilitates and enhances the indexing and retrieval process’ which ‘may or may not be a good thing’ (2002: 151; see also Winsome and Johnson, 2000). This note of caution I echo. While computer-aided analysis tools are extremely versatile and useful for organizing and managing large and complex data sets, there are, nonetheless, some downsides to the ways in which they can, at the push of a button, generate neatly packaged chunks of data. Mason warns, for example, against the danger of seeing ‘categorically indexed slices of data as more concrete, uniform and static’ than they are (2002: 158). Such data do not, she says, constitute ‘tidy and labelled variables’ that can be used for cross-sectional analysis, albeit that it is tempting to do so (2002: 158). Neither can they necessarily help us understand complex social processes such as reciprocity in family relationships. Indeed, Mason suggests, working with the discrete data blocks thrown up by the nodes or indices of such software packages can hamper, rather than help, interpretive analysis. Understanding such an abstract concept as ‘familial reciprocity’ requires, she insists, getting to know about those ‘complex systems of exchange, or of give and take, in family life’, those patterns of social relationships that are embedded in a family’s unique history and that shape the ways in which individuals play out their relationships with one another across a range of social contexts (Mason, 2002: 158). As Mason says: ‘any one small section of text taken in isolation, or even taken together with others of a similar type, may therefore not express a complex interpretative concept in any meaningful sense’ (2002: 158).
Computer-aided systems may sometimes be useful, however, for theory generation and hypothesis testing through their ability to pull together quickly, for example, co-occurring codes (Kelle, 1995). To discover such linkages manually would be extremely time-consuming. However, notwithstanding the usefulness of this facility for some researchers (see e.g. Richards, 2002), for others it can also lead to the temptation to quantify qualitative data in order to make an analysis seem more rigorous – knowing how many people said X with Y appears to prove a strong connection between X and Y. As Mason (2002) warns, however, cross-sectional indexing categories are not the same as variables. They cannot assist with making valid empirical generalizations. More importantly, though, the temptation to quantify qualitative data, which derives I suggest from the inbuilt functions of these software tools, may also change the nature of the intellectual processes involved in interpreting qualitative data. It risks detracting, therefore, from the purpose of employing a qualitative approach in the first place (see Atkinson, 1992; Gilbert, 2002). 2 Dealing with the blocks of often de-contextualized and dis-embodied data segments that computers can churn out may, if we are not mindful, lead us to forget the huge complexities of our subjects’ lives which, as analysts, we set out to understand. The more interesting question may be not how many people said X with Y, but why they said it at all. As Geertz so eloquently described the interpretative process: ‘what generality it contrives to achieve grows out of the delicacy of its distinctions, not the sweep of its abstractions’ (1973: 25).
Part 2: craftsmanship and the imagination
The changes that are occurring in the landscape of research in the 21st century have, therefore, generated some intriguing questions about how the process of data analysis actually takes place – i.e. what is involved in doing interpretative analysis. For Geertz, it is a messy business. The commitment to interpretation involves committing to the view that (1) any understanding of the social world is ‘essentially contestable’ and (2) that any progress we make ‘is marked less by a perfection of consensus than by a refinement of debate’ (1973: 29). As he wryly observes, ‘what gets better is the precision with which we vex each other’ (1973: 29).
This echoes Mills (1959) insistence that social science research is first and foremost a craft that involves the sociological imagination. This he describes as follows:
the capacity to shift from one perspective to another, and in the process to build up an adequate view of a total society and its components . . . the combination of ideas that no one expected were combinable – say, a mess of ideas from German philosophy and British economics. There is a playfulness of mind . . . as well as a truly fierce drive to make sense of the world. (1959: 211)
For Mills, the craftsman (sic) is not a technician and it is
this imagination, of course, that sets off the social scientist from the mere technician . . . who is too well trained, too precisely trained. Since one can be trained only in what is already known, training sometimes incapacitates one from learning new ways; it makes one rebel against what is bound to be at first loose and even sloppy. (1959: 211–12)
And, says Mills, it is in the form of ‘vague images and notions . . . that original ideas, if any, almost always first appear’ (1959: 212). For this reason, he insists, social scientists must strive to be craftspeople, rather than simply to develop technical skills.
More recently, Sennet has argued for the continued significance of that endangered species – the craftsman – in modern life. For him the craftsman ‘represents the special human condition of becoming engaged . . . practically but not necessarily instrumentally’ by developing a skill and a level of mastery through continued dedication to a craft (2008: 20). But, like Mills, Sennet notes that developing a skill is not the same as mastering a technique. The craftsman’s skill development is ongoing, he says. It involves hours of repetition and practice that lead to refinement which, in turn, open up new creative possibilities. The technician, having acquired the technique, is, by contrast, going nowhere new.
In Sennett’s account of the craftsman are reflected, therefore, Geertz’s ideas about the practice and effort necessary to make an argument convincing and also Mills insistence that creativity is central to the craftsmanship involved in the sociological imagination. Returning to the questions raised earlier, then, the significance of crafting to the process of interpreting data becomes clear. Crafting is hard work; it involves repetition; it takes time. Thus, although data analysis packages may aid the quick sorting, organizing and retrieval of data, what their use risks losing is the refinement of understanding that can be gained through the repetitive acts of crafting noted by Sennett (2008) – through, for example, the reading and re-reading of transcripts in order to locate pieces of data or wrestling with data to try to understand why. These processes lead to the kinds of immersion in data that Geertz sees as critical to the thick descriptions that are necessary for its interpretation.
However, Mason goes even further down the path of crafting. She advises that qualitative data analysis involves ‘reading through or beyond the data’ (2002: 149, emphasis in the original). Thus, when turning ‘raw’ data into ‘cooked’ evidence by making particular selections from it (see Hammersley, 2010 above), we must first ensure that these choices are not arbitrary by becoming familiar with that data, whether it has been generated first or second hand. As Geertz insists, alchemy has no place in analysis; we must always keep in ‘touch with the hard surfaces of life’ by immersing ourselves in it (1973: 30). But Mason’s idea of ‘reading beyond the data’ speaks to something more: to the necessity of the imagination in the process of analysis and thus to the analytic imagination I identified at the outset. To imagine, the shorter Oxford English dictionary tells us, is to: ‘form an idea or notion with regard to something not known with certainty’. This is what Geertz described as the to-ing and fro-ing of interpretive practice that involves ‘guessing at meanings, assessing the better guesses, and drawing explanatory conclusions from the better guesses’ (1973: 20). And this creative process, as Grimshaw (1999) observes, is work that goes on in the office later, rather than during the process of fieldwork itself.
Part 3: the analytic imagination at work
The issues raised by teamwork, data re-use and the disputed potential of software packages for analysing data act as timely reminders of the necessity of the craftwork involved in the analytic imagination. Thus, in this final section, I want to chart the doing of a particular piece of data analysis and show how this crafting tool enabled me to make a particular sense of that data. This is a process of sense-making that not only draws on evidence created from the data (Hammersley, 2010) but one which, in Geertz’s terms, is imagined by me ‘in terms of the interpretations to which persons of a particular denomination subject their experience, because that is what they profess to be descriptions of’ (1973: 15).
I have in front of me three transcripts of interviews with members of the same family that I did not do, which belong to a project about families and food. 3 This was a team-based project carried out by me, one other senior researcher and two research assistants. The project asked children and their parents about family food practices in order to explore children’s roles as participants in the family. For the most part the data were gathered during interviews with some children, and then later with a selection of their parents, though usually the interviews took place with mum. The two research assistants carried out all of the children’s interviews in schools. They then interviewed a sub-set of these children again at home, where the interviews with their parents were also done. Following a team discussion about the shape of the coding frame, the research assistants then coded the data using NVIVO. Later, working as a team within and across these nodes, we subsequently provided various accounts of children’s different roles within family food practices. We analysed mothers’ efforts to negotiate the idea of family for their children through food, described fathers’ rather different roles in relation to food provisioning and, through our analysis, learnt that some of the differences between family food practices could be accounted for by the different ways in which children were positioned within families. 4
However, as we worked with coded data segments during the process of data analysis, I became intrigued by the different kinds of lives that these families lived. An odd turn of phrase, a pointed observation or wry comment sparked further questions for me. Why did that mum say that? What did she mean? In the context of what the other children are saying, that’s odd? The bits of text contained within the NVIVO nodes held whispers from the people who spoke the words, traces of their lives that hinted at other things, beyond their immediate thoughts on family food practices. I began to imagine the people and places behind the words.
And now, with the project long over and the research assistants moved on, I am driven back to the original interview transcripts, to try to understand more about the particularities of particular family lives, rather than – as the project had aimed to do – to explore a range of commonalities and differences in family practices. I find myself, in effect, with a set of secondary data in the form of narrative interviews with parents and their children that I want to re-use. How do I make sense of them? If I were a conversational analyst I might examine the twists and turn-taking that structure the halting flow of the conversation in order to draw conclusions, for example, about the distribution of power between the interlocutors – in this case parents and children (see Silverman, 1993). However, this does not interest me so much as wanting to try to get to know the people whose words I am reading. I am curious about who they are, rather than just what they say. I want to know about the kinds of lives they lead, for, as Rapport has argued, ‘individuals personalise discourse within the context of their own discrete perspectives on life, using it to make and express a sense particular to them at a particular time; discursive exchange is never unmediated’ (1994: 25).
And, as I read the transcripts, in my position now as a secondary data analyst, my understanding is not fettered by memories of the interview context or efforts at the co-construction of meaning within the interview. I don’t have to smile and nod or give implicit assent to ideas that I disagree with, out of politeness, gratitude or the fear of causing offence. Indeed, here in my study, I am free to dissent, to disagree, to question. I am not constrained in making my analysis either by recalling the sweet smell of scented candles that permeated a room or the cat that rubbed incessantly against my leg. In short, not having been there gives me the luxury to think freely, imaginatively and creatively, a process described by Rapport as ‘continuing acts of making worlds afresh (1994: 262).
Thus, as I read the words on the page I make sense of them within the context of my own questions and perspectives. I might be making a slightly different sense of them than the researcher did at the time of the interview. This, in turn, might be a bit different from the sense made by the interviewees themselves. But, clearly, in some part, I share both the worlds of the interviewer and interviewees, otherwise I could not begin to understand the transcripts at all. I know the locality a little; I know about current policy issues in relation to food and healthy eating that pepper parts of the shared discourse between the researcher and the interviewees; I also know some things about family life in the contemporary UK. These form the backdrop of my understanding. However, like historians must do, I also start to imagine a particular cultural reality that the words on the page denote for me – the reader and the interpreter. 5
The three interviews I want to focus on were carried out with a mother, with her 10-year-old son and with him and his two friends from school. Two of the interview transcripts have the labels ‘PS11MPP0SRD’and ‘mumPS11MPP0SRD’. From this coding system, I know that these are interviews with an 11-year-old boy and his mum who live in a rural area. I know that the boy lives with both his parents, that he has no siblings and that he follows a restricted diet. 6 But since PS11MPP0SRD feels a little impersonal I immediately want to rename them – the pseudonyms Paul and Jayne inexplicably come to mind. Or do these names somehow seem fitting? On reflection later, I note that these names are names of some members of my own family, but that the spelling that I have chosen for the mother’s name is that of a friend of mine, rather than that of my relative. How far is this imagining already playing into my sense-making I wonder? Is there another backdrop that I need, reflexively, to set my account against?
As I begin to read the interview with Jayne it becomes clear that she is Paul’s father’s second wife and that Paul, in fact, has a much older half-sister. She is 33, the daughter from his father’s first marriage. But Paul told us he had no siblings – that much is clear from the interview coding system, which is derived from the children’s own descriptions of their families. In his interviews Paul never mentions his half-sister. She is clearly not part of his life. Indeed, unlike the other interviews in the data set as a whole, grandparents, aunts, uncles also make no appearance in what Paul has to say. This is most definitely a nuclear family of three.
Simple maths also means that Paul’s father, at least, must be a lot older than those of Paul’s contemporaries and, as I read on, it is clear that Jayne is also an older mum. Indeed, this theme runs as a red thread through their interviews, providing both Paul and Jayne with explanations for why their lives are as they are and enabling me, looking from a distance, to begin to understand particular aspects of their lives. Paul, for example, says he rarely has friends home for tea because his parents stand watching us, asking embarrassing questions and that in any case they’re old as parents go so they’re a bit boring. They are not, he seems to suggest, the kind of parents who would be fun for his mates to meet. Paul, I surmise, sees himself as coming from a rather different kind of family.
At various points in his interview, and also in the group interview he participated in at school, Paul voices his opinion on a range of matters besides those connected with food. He thinks violent computer games are a bit mindless, that you have to think carefully about spending 10 pounds on something in case you don’t like it when you get it home, that some advertising on TV is subliminal, that he can’t see why people want to go to fast-food outlets and that although he doesn’t want to go climbing mountains with his father in the summer it might just be worth it, for the view.
Later he says that he doesn’t like sweets or chips and it’s so long since he’s eaten a doughnut he can’t remember what it tastes like:
I don’t have many things that sounds a bit odd but, erm, (pause) I don’t, yeah, I don’t really have things like Sunny Delight or, erm, sweets or anything really . . . It’s not because I can’t have it it’s ’cause I don’t really want it . . . Generally, things which don’t taste nice are not very good for you. Well, people say chips are lovely I don’t like them they’re not that interesting.
What about doughnuts? Do you like those?
Ooh, I can’t remember I’ve not had them for ages.
In the context of the data set as a whole, such views are very distinctive, something that Paul himself acknowledges – ‘that sounds a bit odd’, he says. Indeed, they do stand out from those of the majority of the other children and I wonder why. Then I remember what Jayne says: as a family they have interesting discussions. Suddenly, I think I can hear Paul’s dad (and perhaps his mum?) speaking through Paul, about the value of a mountain view or the pointlessness of video games. Paul, I begin to think, must live a quiet life of relative solitude with these older parents where trashy food like doughnuts appear but rarely on their vegetarian menu.
I also know from Jayne that Paul is an obedient child who ‘pushes the boundaries a little bit . . . but wouldn’t go beyond . . . even within the house’, that he doesn’t demand pocket-money except ‘if he remembers to ask on a Saturday’, that he is supposed to ‘pull his duvet over his bed in the morning’ but that ‘he doesn’t always’. This is a small act of rebellion in the scheme of things, I find myself thinking. I learn also from Jayne, that although Paul ‘watches children’s television it is usually BBC’. This Jayne thinks, might make them appear ‘snobby’, but she does not apologize for that. Added to this, I learn that although they do tend to eat together as a family, they usually do so while watching a DVD. And just in case she has given the wrong impression, Jayne quickly adds ‘we’re not sitting there it’s just sort of like literally while we’re eating we can just watch something’. Family meals in this family, unlike other families in our data set do not seem therefore to provide a time of togetherness – or they only do so in the very literal sense of the three of them sitting on the sofa side-by-side. There are few meal-time conversations in this family.
As I read and re-read across the three narratives, I begin to suspect that Jayne is a bit anxious about Paul. She is aware that his family life and his vegetarianism might make him feel an outsider among other boys at school. This sneaking feeling finds confirmation when Jayne says that Paul is also not allowed to have a mobile phone. I know from Paul’s own interview that he would dearly love to own one. And I suspect Jayne knows that too. She says, with a sigh:
I think he’s the only one in his class . . . that hasn’t got a mobile phone. (pause) And he manages that very well.
But once again, I hear the ghost of Paul’s dad shading into the interview as Jayne continues to relate her views, about Paul’s views, about having a mobile phone:
He [Paul] sort of, you know he says, ‘Well actually what would you use it for?’ and he’d like it as a toy.
Jayne goes on to find further explanation – or are these excuses? This is, after all, the year 2007, a time when the use of mobile phones is now commonplace, even for quite young children. With pay-as-you-go payment tariffs, it is no longer quite the luxury item it used to be. And this family is clearly not hard up.
Jayne says that she has a mobile phone but that she never uses it; she just leaves it in the car for emergencies and Paul’s dad, well, for a long time he didn’t have one but they made him have one at work. So, Jayne, seems to say, we are not really being mean and making Paul feel different from his peer group. It’s simply a matter of practicality. But, in her very next sentence, she appears to doubt their parenting approach. With a nervous laugh she says:
Yeah. Erm, (pause) erm, we haven’t got the Internet either (laugh) and, you know he would, he would like that
She goes on to say that, although she doesn’t know for sure, she suspects that Paul has to make excuses for all of this differentness:
probably he has a, he’s part of the group and he, he has a moan he says, ‘Oh, it’s because my Mum and Dad are so old.’
And so in and across these three interviews, a picture of Paul’s home life unfolds for me as, like the data re-user or historian, I form ‘mental concepts, schema, projects for what is not present’ (Rapport, 1994: 262). Jayne’s and Paul’s narratives come to constitute a kind of extended case-study that allow me to reflect on the ways in which they are talking about family life and intergenerational relationships (Savage, 2011). And, from these forms of expression, I imagine what Paul’s life is like at home. This imagination is stimulated by knowledge gained from years of research with other children, from newspaper stories as well as from academic accounts of class, family life, parenting practices and the commercialization of childhood. All this feeds my imagination of what it must be like to be Paul – or Jayne, for that matter. She is always trying to mediate – as I increasingly come to believe as I re-read the narratives – between the strictures set out by Paul’s dad for the way that their family life should unfold and her dawning realization that this particular way of life might be making things tricky for her son. 7
Conclusion
This article has explored the role of the analytic imagination in the craft of interpreting data. It has shown that the experience of not-being-there – whether the result of carrying out team-based research or secondary data analysis – does not have to mean the abandonment of reflexivity nor the loss of the experience of deep immersion in the data. It has also argued for the need for researchers to continue to get to know their data well and shown that it is as profitable to ‘hang out’ in a set of interview transcripts of conversations with people one does not know, as it is to hang out in their living rooms. Though different, both involve what Rapport calls the ‘rewriting of social reality’ that is achieved through the acts of creativity and imagination that enable us to craft stories of people’s lives (1994: 277). The interpretations that result may be different in some respects, but interpretive analyses should make no claim to singularity.
Recalling once more Mason’s (2002) advice to read through the data means drawing on the analytic imagination to see, not just what ‘evidence’ there is, but how one snippet of conversation might relate to another spoken some time later; it means building a picture of the lives narrated, rather than simply documenting their component parts; it involves looking for the absences, as well as the presences to see what is missing and perhaps explain why. In the detailed illustration given above, it means imagining how and why the accounts given by mother and son about the same ‘reality’ can be so different, but also understanding the moments and spaces where their intergenerational perspectives meet. Mason’s (2002) insistence that qualitative data analysis also involves reading beyond the data, means trying to imagine the reasonableness of what people say, why they say what they do, however unreasoned it might at first seem. Such reasonableness has, therefore, to be imagined by, for example, situating particular narratives within the broader social context, by drawing on wider empirical and theoretical understandings, by digging deep into the repertoire of implicit knowledge that researchers themselves possess. These, then, are the tools of the analytic imagination, the processes through which we come to understand our data, rather than simply report our findings.
What this exploration of the analytic process has also underlined is that while computer-based data analysis packages may enable us to produce ‘findings’ from our data more quickly, there are no short-cuts to its interpretation. Interpreting qualitative data is a craft that needs practice; it is not just a technique to be mastered. So, just as the debates of the 1980–90s exposed the significance of writing to the analysis of culture (Clifford and Marcus, 1986) and interrogated the ways in which ‘textual practices themselves constitute the social realities constructed and reconstructed in ethnographic writing’ (Atkinson, 1990: 178), this article has interrogated the role of the imagination in analysis. It has shown that the process of data analysis requires more than the acts of data framing, coding and systematizing. What is needed is the reflexive and creative crafting of that data.
Footnotes
Funding
The data on which parts of this article draws comes from a project Children as Family Participants (Allison James and Penny Curtis) which was part of the research programme Changing Families, Changing Food, funded by The Leverhulme Trust carried out at the University of Sheffield.