The use of composite narratives to present interview findings

Abstract

This research note describes the use of composite narratives to present interview data. A composite narrative uses data from several individual interviews to tell a single story. In the research discussed here, investigating how politicians consider climate change, four composites were created from fourteen interviews with Members of the UK Parliament. A method for creating composite narratives is described. Three, linked, benefits of the technique are discussed. First, they allow researchers to present complex, situated accounts from individuals, rather than breaking data down into categories. Second, they confer anonymity, vital when reporting on private deliberations, particularly if interviewees are public figures. Third, they can contribute to ‘future-forming’ research, by presenting findings in ways that are useful and accessible to those outside academia. The main limitation of composite narratives is the burden of responsibility upon the researcher, to convey accurate, yet anonymized, portrayals of the accounts of a group of individuals.

Introduction

Although a great deal has been written about the politics of climate change (Giddens, 2009; Urry, 2011) there has been little research aimed at understanding how politicians themselves understand or respond to the issue (Rickards et al., 2014). In contrast, there are studies exploring the environmental views and behaviours of individuals (Laidley, 2013; Norgaard, 2006) and the outlook of business leaders (Wright et al., 2012). The research described here aimed, therefore, to take a similar, explicitly qualitative and sociological approach to examine how politicians understand their role.

A major part of the study was a set of fourteen interviews with current and former Members of the UK Parliament in 2016, drawing on Riessman’s (2008) narrative method, and conducted under conditions of anonymity. In the first half of the interview, participants were encouraged to reflect on their working life, including their ambitions, and the influences and pressures upon them. They were then asked to consider how they approached the issue of climate change. As the interviewer, I was impressed and somewhat surprised by my subjects’ reflectiveness, with many showing a disarming frankness about their professional and personal struggles. I was left with a set of complex and personal interview accounts, and the tricky question of how to do justice to this richness, whilst maintaining the anonymity which, as many of my subjects told me, was crucial to them speaking out.

The results of the study are published in an article in the Sociological Review (Willis, 2017a). This research note discusses one aspect of data analysis and presentation: the use of what I term ‘composite narratives’, in which a number of interviews are combined and presented as a story from a single individual. Though this approach is rarely used, it seemed to me to be a good response to the tricky question described above: how to present an authentic yet anonymous story.

Below, I firstly describe how I arrived at the decision to develop composite narratives, and the process I engaged in to draw up the composites. I then discuss the advantages of this approach, and the potential drawbacks. The note ends by suggesting that being transparent about the way in which composite narratives are developed might help to overcome these issues.

Using fictions to convey ‘emotional truth’

Glossy hair, shiny shoes, white teeth . . . This woman is beautiful, mid-twenties, successful. She delivers her material as though saying: ‘Pretty gripping stuff, eh? What do you make of that, oh therapist?’ (Blundy, 2017)

Psychotherapist Anna Blundy writes a monthly column for Prospect Magazine, offering intimate accounts of her conversations with patients, to inform and to entertain her readers. But none of these characters actually exist. At the end of every column, there is a small disclaimer: ‘the situation described above is composite, and confidentiality has not been breached’ (Blundy, 2017). The therapist Susie Orbach takes a similar approach. In her book The Impossibility of Sex (2000), she draws on her vast clinical experience to invent seven fictional characters, describing the complexities of the patient-therapist relationship.

Both Blundy and Orbach state that they use composite accounts to prevent breaches of patient confidentiality. Orbach also alludes to a further advantage of composites: she can use a single story to tell a more generally representative account of the experience of her patients. Thus she is not just talking about one individual; she is instead using her judgement to create stories which capture the essence of her patients’ lives, experience and perspectives. Orbach calls this ‘emotional truth’ (2000: 196), defined as ‘an authentic representation of feeling states rather than a strict adherence to narrative truth’ (2000: 197).

In taking this approach, Orbach has much in common with qualitative sociological researchers. With qualitative methods, researchers use their experience and judgement to make sense of large quantities of data, from interviews or ethnographic observations, for example. Researchers select, interpret, order and arrange data, in order to analyse and present research findings (Law, 2004). Creating stories or narratives based on a range of individual accounts is one way to do this.

Like Orbach and Blundy, I was drawn to storytelling by the data I had in front of me. I had conducted fourteen interviews with politicians, and they had talked to me about their life and work, their aspirations and concerns, and the nitty-gritty of how they got through their working day. The picture thrown out by the data was richer, more tangled and more personal than I had expected. It did not seem to me to be appropriate to analyze the data through ‘standard’ categorization, such as party affiliation, social background, gender, age or previous experience. Of course, all these factors are significant, as evidenced by previous research (see, for example, Carter (2013) on the differences between political parties; and the work of McKay (2011) and Puwar (2004) on gender in politics). Yet my aim was not to distinguish between politicians, through comparisons and categorizations. Instead, it was to investigate how politicians, as (a group of) people, navigate their life and work.

An obvious solution would have been to focus on particular individuals, as case studies. I could have picked, say, four individual politicians, and given a full account of each. However, I was prevented from doing this by the need to preserve anonymity. Each subject had agreed to the study under condition of anonymity. As public figures, politicians can be identified easily, even if personal details are disguised. Thus I had the idea of combining individual accounts, to form a composite. As with Orbach’s work, this would also have the advantage of writing narratives which, together, give a picture of the group as a whole.

With this in mind, I undertook a search of literature on qualitative research, assuming that presenting interview data in this way would be commonplace. To my surprise, it appeared to be a technique rarely used. In addition to the psychotherapy examples described above, a small number of other studies emerged. Wertz et al. (2011) describe the use of ‘composite first person narratives’. For example, a story of an obese adolescent girl, told from her point of view, illustrates the ‘body shaming’ that such individuals may feel. As Wertz et al explain,

The individual narratives of each study participant are unified with the reflexive understandings of the researcher . . . [which] affords the reader the ability to explore the ‘felt-sense’ of the informants’ experiences. (2011: 8)

Similarly, a study by Piper and Sikes (2010) of sexual relationships between teachers and pupils uses fictionalized stories, based on ethnographic and interview data. Though dialogue and contextual details are fabricated, the researchers did not make up ‘anything that directly related to people’s experiences and perceptions of allegations of abuse as told to us’ (Sikes and Piper, 2010: 43). Their prime reason for using this approach was protection of anonymity, particularly given the risk of media intrusion. Annette Markham (2012) also describes how she creates composite accounts to present data gathered from online sources, which could otherwise, if quoted directly, be traced easily through search engines. Markham deliberately and provocatively uses the term ‘fabrication’ to ‘interrogate and destabilize the mistaken and often unspoken assumption that invention necessarily represents a lack of integrity’. (2012: 336).

Inspired by these examples, I began to experiment with creating composites from my interview data, that would convey ‘emotional truth’, to use Orbach’s phrase (2000: 196). In doing so, I began to formalize a more rigorous method, in order to give a clear account of how the composites were derived. This is described below.

Developing a methodology for composite narratives

Whilst the use of composites discussed above (Orbach, 2000; Piper and Sikes, 2010; Wertz et al., 2011) are a compelling way of presenting research, the link between the original data and the final story is not clear. The process by which the original speech becomes a written narrative is opaque. Readers are asked to simply put their faith in the researcher’s judgement. To a certain extent, all qualitative research faces this issue; researchers must use their judgement to analyse and present unwieldy amounts of ‘raw’ data in the form of interview data or ethnographic fieldnotes. By documenting the way in which the data has been analysed and presented (describing, for example, the process of coding; decisions about what extracts to include; decisions about any categorization, and so on) the researcher can build trust in the research process and the subsequent findings.

I started to consider ways in which I could do this. I wanted to demonstrate that the narratives were derived directly from the original data. I decided that all the quotations I used would be verbatim quotations from the original interviews, and likewise, all the details from the narratives would be taken from one or more of the original interviews. When I submitted the article for peer review, however, the reviewers were broadly supportive of the composite narrative approach, but all asked for more clarity about how they were derived. One wrote

The idea of composite narratives is interesting, partly as a way of dealing with the difficulties of anonymising politicians’ statements. However, you would need to tackle some serious questions about precisely how you compiled the composites and what we can deduce from them. ¹

As a result, in a subsequent draft, I developed a much more detailed and transparent account of the process of deriving composite narratives. I devised four practices to govern the process of composing the narratives, as follows:

1. Each composite is based on transcripts from interviews. 3–5 transcripts are condensed into one composite narrative. For example, ‘Jonathan’ is a composite of three interviewees from different parties, all relatively new to their roles as MPs and sharing similar concerns about speaking out on the issue. In contrast, another narrative, ‘Stephanie’, is a composite of four experienced MPs, who had all served as ministers.

2. All quotations come directly from these interview transcripts.

3. Other details, such as where the interview took place; how the conversation evolved; and any paraphrasing of discussions, are taken directly from one of the source interviews.

4. In the narrative itself, I avoid imposing any judgement on the interviewees’ experiences and opinions, and do not assume motivations or feelings. Any comments of this nature in the narrative are taken directly from the interviewees. For example, in the composite narrative ‘David’, concerns are raised that if politicians are vocal about ‘controversial’ issues like climate change, this might affect their opportunities for promotion. In the narrative, this is reported as follows: ‘David feels that his commitment has come at a price’, showing that this is his reading of events, not the researcher’s interpretation.

Taken together, these practices ensure that there is a clear link between the original interview transcripts and the final narratives. The only modification is to present data from several interviewees as if it were from a single individual. I use the term ‘narrative’, not ‘fiction’, to emphasize the fact that they are derived solely from the data.

The resulting composite narratives

Following this approach, I created four composite narratives from fourteen interview transcripts. Several different groupings of composites were considered, and of course, the stories could have been combined in any number of different ways. I chose the final groupings as they seemed to me to best convey the range of positions and views that the data revealed. A full description of the method, and the final composite narratives, each of a length of around 500–600 words, is presented elsewhere (Willis, 2017a) and summarized below. The composites mix interviewees of both genders, but in order to tell the story in a readable way, a gender is assigned to each, roughly representing the gender balance of the House of Commons as a whole (71 percent male in the 2015–2017 administration).

‘David’ is formed from a composite of three interviewees, all relatively new to parliament. David speaks out about the need to act on climate change, even though this means that his colleagues see him as a ‘freak’.

‘Jonathan’ is a composite of three MPs, one Labour, two Conservative. In contrast to David, Jonathan worries about being too vocal on issues that he sees as controversial. His primary concern is to support his constituents.

‘Paul’ is formed from four interviewees. He has more experience than David or Jonathan. He wants to tackle climate change, and focuses on the tactics by which action can be achieved.

‘Stephanie’ is a composite of four interviewees, all of whom have served in government. Stephanie’s overriding concern is the slow pace of change, and the need to focus on practical, achievable steps.

Following the presentation of the four narratives, the article (Willis, 2017a) examines three themes common to each: questions of identity, or how politicians understand their own role and life; the question of how politicians interpret and execute their role as representatives; and the issue of how each MP manages the pressure and complexity of everyday life.

Why use composite narratives?

This research highlights three potential benefits of using composite narratives, discussed below.

Complex, situated accounts

Given that the aim of this project was to examine the outlooks and motivations of politicians, as individuals within a wider system of democratic governance, there was a need for a method that allows the complexities of an individual’s position to be presented and explored.

The urge to resist typologies and categorizations, and instead to find ways to convey the richness and complexity of an individual’s perspective, is common to much qualitative research. Examples include the body of work on illness narratives which moves beyond medical categorizations to focus on the lifeworld of the patient (for a dicussion see Atkinson, 2010); the studies described by Wertz et al. (2011) discussed above; and feminist research such as Puwar’s (2004) study of female and ethnic minority MPs.

In his review of ‘narrative ethics’, Tony Adams asserts that, in any such research which focuses on eliciting narrative, it would be wrong to categorize or de-personalize accounts in presenting the data: ‘we must not approach stories with a prescription or typology for analysis; an evaluation of narrative must remain contingent on the stories, authors, and audiences as they interact’ (Adams, 2008: 179). Instead, as John Law writes in describing ‘messy’ methods, the challenge is ‘to keep the metaphors of reality-making open, rather than allowing a small subset of them to neutralize themselves and die in a closed, singular, and passive version of out-thereness’ (2004: 139). In keeping with this orientation, narratives aim to embrace the account provided by the individual, rather than searching within it for an objective ‘truth’ (Adams, 2008; Riessman, 2008; Thomas, 2010). As Riessman writes, ‘narratives are useful in research precisely because storytellers interpret the past rather than reproduce it as it was’ (2005: 6). Narratives allow research to be presented in a way which acknowledges the complexities of individual motivations and outlooks, whilst drawing out more generalized learning and understanding.

Anonymity

The more context and detail offered in a narrative account, however, the greater the chance of compromising anonymity. An important advantage of composite narratives is that they allow the presentation of contextualized stories, without resorting to fiction, and whilst offering a significant degree of anonymity.

Members of Parliament are public figures, with a great deal of information about them in the public domain, particularly given the ease of online data gathering. Ensuring anonymity is, therefore, much more complex than changing names and withholding basic personal data (Saunders et al., 2015; Tolich, 2004). MPs can be identified by their political party, constituency represented, stated views on a particular issue, voting record, or, most likely, some combination of these. To ensure anonymity, details must therefore be removed or altered. However, doing so removes vital context, and works against the aims of providing a situated account, described above. Composite narratives protect anonymity through mixing accounts, thereby preventing the possibility of identifying a research subject through a combination of details.

Much discussion of anonymity has focused on the need to protect vulnerable research subjects, such as Piper and Sikes’ (2010) teachers and pupils involved in accusations of abuse, and Saunders et al.’s (2015) family members of people in minimally conscious states. Politicians may not be seen as ‘vulnerable’ in this sense. Yet their status as public figures confers a different type of risk or vulnerability. In her discussion of elite interviewing, Morris (2009) reports that whereas elite interviewing is typically held to be problematic because the interviewees are powerful figures, she instead argues that researchers ‘have power over the respondent through the process of research . . . control of what is published and control of meaning’. (Morris, 2009: 213) In this sense, elite interviewees are particularly vulnerable to breaches of anonymity.

In my case, politicians spoke openly to me about the dilemmas they faced in deciding whether to speak out on issues that they saw as controversial or difficult. For example, one admitted to feeling lost and unsupported as a new MP. Many were critical of the stances taken by their colleagues. These interviewees would have been embarrassed at best, and possibly severely compromized, if their identity was revealed.

Future-forming research

Last, a crucial additional advantage of composite narratives is that, by providing contexualized and personalized accounts, they can help to build understanding of particular people and groups, in ways that are accessible to non-academic audiences.

For example, my work has proven useful for people who want to work with politicians to build political responses to climate change, including scientists and advocates from pressure groups. Because composite narratives are based on anonymized conversations, they provide insights that might not be forthcoming if MPs were speaking in the public domain. I have presented this research to audiences of scientists and practitioners, who commented that it gave them an insight into how MPs think through their working life and strategy. The work is now being used by a coalition of advocacy organizations, to evaluate their parliamentary campaigns. Similarly, many illness narratives aim to provide healthcare practitioners with an understanding of patients’ lived experience, thereby improving care practices (see Mishler, 2005 for a discussion).

Such research can be seen as part of a wider project of what Gergen (2015) terms ‘future-forming’ research. Gergen distinguishes between, on the one hand, social sciences which take from the natural science tradition an aim of mirroring reality; and on the other hand, a ‘reflexive pragmatism’ (2015: 287) which sees research as situated within the social world, with an explicit ethical stance and aim for intervention, rather than just description or analysis. The research described in this article, a collaborative project between a university and an environmental advocacy organization, was designed to do precisely this: using academic methods to improve outcomes for climate policy and politics. Whilst not all research using composite narratives is necessarily ‘future-forming’, this project, as well as examples from the field of healthcare, demonstrates that composite narratives lend themselves well to this sort of applied research.

Evaluating the use of composite narratives

My approach to composite narratives, and particularly the development of a transparent route from the data to the narrative, as described above, helps to build confidence that the composites are a meaningful representation of the original data.

Yet even with this more formalized approach to composite narratives, there is a danger that in merging accounts from different individuals and presenting them as one, the narratives becomes simplistic or caricatured. For example, I could have combined interviews in a different way, to create a narrative about a timid, newly-elected MP who feels completely overwhelmed by their role. Many interviewees did express feelings of doubt or worry to me, and I could have combined these to create one composite of a nervous, overwhelmed politician. However, this would not have been true to the data, as everyone I interviewed had a certain degree of self-assurance and confidence, unsurprisingly given the job they were doing.

As this example illustrates, when creating composite narratives, there is a considerable burden on the researcher to develop composites that ‘fit’ the underlying data. In common with much qualitative research, including narrative techniques and ethnography, the reader relies on the researcher’s interpretation and presentation of data. The researcher needs a level of understanding and familiarity with the context of the study, in order to judge what makes a meaningful composite. In my case, I had worked with politicians for many years, and understood the world of parliament in general and the politics of climate change in particular; this meant that I had a depth of understanding which enabled me to write up data in this way.

There may also be a danger of privileging narrative, relying too much on accounts provided by individuals, and not seeing the wider context or structure within which the narrative is set. For example, one of my interviewees told me that he had not been promoted within his party, because of his outspokenness. As a peer reviewer of my article commented rather archly, he could have been overlooked for promotion simply because he is not very talented. This difficulty can be overcome not by trying to get to the ‘truth’ buried in each account, which is an impossible task, as discussed above; but by using a range of different methods to triangulate and situate narrative accounts. In this instance, there is other evidence (see for example Rickards et al., 2014; Willis, 2017b) that politicians and other leaders do indeed feel the need to modify their views in order to progress in their careers.

In summary, this research illustrates that composite narratives provide an effective means of presenting anonymized interview data, while maintaining the richness and complexity of personal stories. Researchers can demonstrate the validity of the narratives through following a transparent process in drawing the interviews together to form composites.