Shifting analytic ontology: using I-poems in qualitative longitudinal research

Introduction

Qualitative methods of analysis are concerned with transforming and interpreting data to capture and understand the complexities of the social world. What constitutes the moment of data analysis, however, is subject to different understandings (Coffey and Atkinson’s, 1996, elaboration of this point, although some years old now, still stands). Some stress analysing as a process that stretches across the whole research project and even beyond, while others variously focus attention on systematic procedures of data handling such as coding, or the reflexive creation of unique or patterned interpretation. Commentators have long noted the way that analysis is or should be integrally linked to, both shaped by and shaping, the version of knowing ‘social reality’ or epistemology that informs the research project and process as a whole (Coffey and Atkinson, 1996; Grbich, 2007; Mason, 2002 [1996]). What is not so clearly discussed, however, is that the re/production and re/presentation of meaning through analysis places the researcher in a particular ontological position – that is, a mode of being in relation to the interviewee and their social reality is constructed through the process of data analysis.

In this article, we focus on the interpretive analytic ontology involved in analysis of interview data. We explore the implications of using the I-poems method of interview analysis – a process that traces how participants represent themselves in interviews through attention to first person statements – to look at change and continuity in young people’s sense of self over time. We draw on detailed case study material from a qualitative longitudinal research project that we are conducting with children and young people, to consider how different methods of data analysis place the analyst – ontologically – in different positions in relation to the interviewee’s account of themselves. We argue that I-poem analysis, at the same time as it draws attention to the research subject’s subjectivity – how they understand and speak about themselves, it also produces a particular analytic ontology – how the researcher is placed in relation to the subject and their social reality. We will be comparing our broadly thematic analysis of extracts from consecutive interviews with one of the participants in our study, Anne, with our analysis of the same extracts based on the I-poem method. We aim to show how methods of qualitative data analysis provide angles on the nature and constitution of social reality (ontology), and in particular conceptions of self–other relations for the researcher.

Indeed, in contrasting the ontologies invoked by different methods of analysis (thematic and I-poem), this article is something in the way of a (belated) response to Coffey and Atkinson’s call for researchers to try out different analytic angles on their data:

There is much to be gained from trying out different analytic angles on one’s data. New insights can be generated, and one can sometimes escape from analytic perspectives that have become stereotyped and stale. We therefore want to encourage a (modestly) playful approach to the diversity of research approaches . . . [This] may help to reveal different facets of the data. (1996: 13, 15)

In this spirit then, it is not our intention in the following discussion to present one method of data analysis as superior to another. Rather, we believe that the construction of varying ontologies of self in relation to other through different processes of data analysis are complementary rather than exclusive. Moving between analytic angles can provide valuable illumination, contributing different stories about participants and their social realities.

The case study that we use to illustrate varying analytic angles and implicit ontologies, as we have said, is part of our qualitative longitudinal research with children and young people. The ‘Your Space’ project aims to document and track the meanings, experiences and flows of prescribed (sibling) and chosen (friendship) relationships for children and young people, and how these relate to their sense of self as their individual and family biographies unfold (http://www.lsbu.ac.uk/families/yourspace/). It has followed just over 50 children and young people born between 1989 and 1996, from different family, class and racial/ethnic backgrounds, living across Britain. As children and as young people, the participants have been interviewed three times: Wave 1 interviews when they were aged between 6 and 13, Wave 2 when they were 10 to 17 years, and Wave 3 when they were 12 to 19 years. Our project is part of the larger ‘Timescapes’ study, designed to shed light on the dynamics of personal relationships over time, and the identities that flow from those relationships across the lifecourse (http://www.timescapes.leeds.ac.uk).

We begin with an explanation of the different methods of analysis involved in our discussion here – thematic and I-poem – before proceeding to present the interview transcript and I-poem material from our ‘Your Space’ qualitative longitudinal case study that serves as the foundation for our subsequent arguments about the varying analytic ontological positions – and substantive insights – constructed through different, thematic and I-poem, analytic strategies. This foundation involves carefully built up detail and evidence to enable the process of analysis, and to demonstrate the shifting ontological position of the analyst.

Interpreting data: thematic analysis and I-poem analysis

Thematic analysis of qualitative data, broadly, is data led. It involves identifying key topics and patterns, regularities and contrasts, in the material in order to create interpretive meaning. Thematic analysis can be guided by issues that stem from the research topic and conceptual approach, and/or manifest or latent leitmotifs that can be discerned in the data, as broad categories or as repeated words, phrases, understandings or experiences in the data. It usually involves processes of segmenting and categorizing particular parts of the text, and then linking related categories together, from which themes can be refined and connected (e.g. descriptions in Boyatzis, 1998; Gray, 2009). Analysts from different epistemological positions adapt this general process in various ways. For example, the focus on lived experience and understanding taken by phenomenologists may lead them to follow a process of general familiarization with the data, identifying and extracting significant statements and meanings that are then organized into clusters of themes (based on Colaizzi’s idea, 1978). Narrative analysts, however, usually focus on reconstructing the holistic meaning of stories that people tell about themselves, such as through using thematic field analysis to access the underlying and recurring order of their stories about the past (see Chamberlayne et al., 2000).

In contrast to a stress on the creation of meaning through thematic analysis, its initiators pose I-poems more as concerned with accessing meaning in relation to self. I-poems are part of a specific method for analysing interviews developed by Carol Gilligan and colleagues in the context of a qualitative longitudinal psychological research project that was concerned with adolescent girls’ resistance to, or silencing of, their sense of self as part of dominant ideas of femininity (Mikel Brown and Gilligan, 1992). Gilligan and colleagues variously refer to the overall method as the ‘Voice-Centred Relational Method’ or the ‘Listening Guide’ (Gilligan et al., 2003; McLean Taylor et al., 1996; Mikel Brown and Gilligan, 1992). The process involves at least four main sequential readings of an interview transcript, with each reading highlighting a particular aspect of understanding the interview and interviewee.

The first reading involves two elements. First, attention is paid to the overall story that the interviewee is relating, almost like a rich synopsis of content or plot. As part of this, recurring images and words, key metaphors, and dominant themes are identified. This reading can be captured in the question ‘who is telling what story?’ (Byrne et al., 2009: 69). This element is not so different from the thematic analysis discussed above. The subsequent, reflexive element of this first reading concerns the researcher documenting their own emotional and intellectual response to this story. This process is the ‘who is listening?’ partner of the ‘who is telling what story?’ question (p. 69). The aim is to uncover how the researcher’s response to the interviewee might affect their understanding and the analysis they produce. The second reading is concerned with tracing how the participant represents or speaks about themselves in the interview. It is this reading that provides the basis for the production of I-poems, which we explain further below. The third reading concentrates on how the interviewee talks about their relationships with other people and what they see as the consequences of these relationships, in particular tracing the different or multiple subjectivities associated with these relationships. And the fourth reading involves paying attention to the specific cultural and political contexts and social and economic structures in which the interviewee is located, and which shape their sense of self.

Our main focus here, as we have said, is on the second reading of the transcript, in terms of the production of I-poems and on their use in longitudinal research. This reading is the one concerned with tracing how participants represent themselves in the interview and concerns the stream of consciousness that is carried by the first person references that run through the interview, rather than being contained by the full structure of sentences. The method pays detailed attention to the use of the personal pronoun – ‘I’ – to identify the different subjectivities from which the participant speaks.

Practically, the creation of an I-poem involves two main steps. The first involves reading through an interview transcript and highlighting each use of the first person ‘I’ and associated verb or seemingly important accompanying text. The run of words associated with the ‘I’ statement that are highlighted is quite an intuitive process, with the analyst judging what is important to understanding the interviewee’s sense of self. The second step involves cutting and pasting – lifting the highlighted phrases out of the transcript in the exact sequence that they occur originally in the interview, and placing them in separate lines, like the lines of a poem. We will detail and illustrate our discussion of this process below.

The I-poem can be constructed into stanzas based on breaks in the topics and ‘voices’. Several senses of self – contrapunctal ‘voices’ – can be identified for one person, which may be conflicting or complementary, resisting or capitulating, confident or distressed, firm or struggling to make themselves heard. Some of these voices may be unique to the particular individual whose account is being analysed, but other of their voices may be senses of self that are echoed across many participants in a research project. In this sense, ebbs and flows of change and continuity across the course of one interview can be tracked using I-poems. This can be represented graphically, with different voices highlighted in different colours on the transcript to trace their movement throughout the interview.

This identification of voices overlaying and interweaving within an interview seems to be the way that most researchers who have followed Gilligan and colleagues have used the method. Researchers have used I-poems to analyse data collected in projects studying, for example, older women with dementia (Proctor, 2001), and young women students’ literary practices (Woodcock, 2005). In a study of sibling incest, the main voices identified for one young woman included an ‘I’m best friends with my brother’ voice; a ‘My brother abused me’ voice; an ‘I take care of others’ voice; and an ‘I do what I’m told’ voice (Kiegelmann, 2000). Another study, of women professionals who had been made redundant, identified three main voices across everyone interviewed using the more abstract terms of: ‘voice of silence’, ‘awakened voice’ and ‘dissonant voice’ (Balan, 2005). Thus, researchers seem to have found the method relatively flexible, adapting it to suit their particular subject matter (Mauthner and Doucet, 1998). The examples that we have found and referred to here all focus on women (and we do this also in our case study of Anne below), but I-poems can and have been used with men’s interview transcripts (e.g. Byrne et al., 2009).

Although the method, and certainly the I-poem aspect, appears mainly to have been used to identify voices within one-off interviews, Gilligan and colleagues (2003) originally developed the method to track changes and continuities in young women’s subjectivity longitudinally, across participant interviews over time. It was this facility that attracted us to use it as one way of understanding the interview data for the children and young people whose lives we have been following over the past seven years or so. As well as the conceptual ability to trace changes and continuities in our participants’ sense of self across interviews carried out at different points in time, the method is attractive practically in providing a systematic series of steps to follow, rather like a recipe. The mysterious and confusing process of data analysis is made clear. Indeed, the accessibility of the guiding framework has led to it being used as part of participatory research. Researchers have used the method to involve young people (male and female) who had left school early in constructing and interpreting their own and others I-poems as part of the research process (Byrne et al., 2009).

There are, however, both practical and epistemological issues in using the I-poem method. On a practical level, as will become clear in our discussion below, even concentrating on just one aspect of the overall voice-centred relational method is quite a time-consuming process, more suited to using in a few cases studies that ‘tune your ear’ for entering a larger sample (Mauthner and Doucet, 1998). On an epistemological level, the method encourages what is said to be the reinsertion of listening into the analysis of the interview transcripts. During an interview, the researcher listens to the person being interviewed and responds to them; the method attempts to recreate that ‘listening’ during the process of analysis – hence one of the names for the voice-centred relational method as a whole, of which the I-poem is a part: the Listening Guide (Gilligan et al., 2003; Mikel Brown and Gilligan, 1992).

As we have said, the I-poem in particular, as part of the overall analytic process, focuses attention on the voices or self of the interviewee. The listening to how the interviewee talks about themselves that is fundamental to the I-poem is supposed to create a space between the interviewee’s own self-perception and the analyst’s perception of them. The idea is that researchers are confronted by how the person who has been interviewed understands their self before the researcher produces an analytic account of who they are: ‘how she speaks of herself before we speak of her’ (Mikel Brown and Gilligan, 1992: 27–8). Thus, the method is supposed to allow a researcher to experience an explicit shift from listener to interpreter.

In their discussions of the voice-centred relational method, Gilligan and colleagues do not appear to be explicit about the definitions of self and voice (Mauthner and Doucet, 1998). It seems to us, however, that the implication is that there is some pure ‘voice’ and essential research subject, however subdued, that we can have access to in an unmediated fashion through their words: the ‘silent and invisible inner world’ to which Gilligan and colleagues refer (2003: 157). Mikel Brown and Gilligan’s (1992) introduction of the method, for example, builds on their argument that young women become dissociated from themselves as they move into adolescence, taking on dominant ‘selfless’ versions of acceptable womanhood and speaking in a ‘no-voice voice’.

Such a tendency towards essentialism may well be why the method focuses on the ‘I’ statements that interviewees make. The notion of ‘I’ invites us to think of something that is part of a person’s make-up that drives or influences what they do, think or say, and that ‘I’ gives access to that. Yet, the use of ‘I’ is a particular form of speech. ‘Me’ can be used instead of ‘I’ in Black English for instance (Sebba, 1997), and people may well talk about themselves in the second person in their interviews. For example, here is Anne – the case study whose interviews we discuss in detail below – talking about working in a daycare centre:

I thought nursery work would be easy with just watching the children, but no! You’ve got to plan activities and it’s just so much work . . . You plan like a cooking activity for the children but you have to make sure all the ingredients are in the kitchen and there’s enough spoons to go around the children, and it’s just silly little things sort of thing but it’s hard work . . . When you come up and go up the stairs you have to remember to close the baby gate, and if you’re changing nappies you have got to put gloves on and then change the child, and then you have to spray down the thing . . .

Presumably Gilligan and colleagues would not consider Anne’s ‘you’ style of talking as referring to her authentic inner self. In our view, however, taking account of people’s different speech patterns to produce you-poems or me-poems, or indeed we-poems, is equally enlightening about senses of self.

Further, the implication is that analysts can understand someone separate from their own theoretical orientation and personal experience – an extremely debatable stance, especially for researchers (such as ourselves) who combine a commitment to understanding people’s lives from their own perspective with regarding interview accounts as co-constructions between interviewee and interviewer. Indeed, there are major debates over the issue of subjectivity and the ontological status of subjects as constituted by or situated in context (structured or agentic). These are accompanied by debates about how researchers may access this subjectivity. In this latter respect, Doucet and Mauthner (2008) have made a strong case for narrated subjectivity that researchers can come to know through the layered readings of the Listening Guide. Such a process has the strength of acknowledging and reflecting upon the integrality of researcher subjectivity in ‘knowing’ their subjects in the ‘who is listening?’ part of the first reading. In our view, however, it then puts researcher subjectivity aside and falls short of following the implications through into analytic ontology. It is an understanding of interviews as co-produced that alerts us to the production of researcher subjective status as well as that of the subject, and enables us to consider our experience that the I-poem method propels the researcher as analyst into a different position in relation to the interviewee through their transcripted data.

We will return to these concerns about authenticity and researcher self–participant other relations after we have explored the use of I-poems and thematic analysis through the case of Anne below. As we elaborate, moving from a thematic analytic approach to an I-poem voice approach seems to involve a shift of the researcher’s analytic ontology, moving from looking at the interviewee to looking at their social world alongside them.

Analysing Anne’s interviews

In this section we present interview transcript and I-poem material from our case study, Anne, which illustrate the process of constructing an I-poem described above, and provide the foundation for our discussion of comparing the analyses and researcher ontology constructed in interpretive thematic and I-poem approaches to creating meaning from data.

Anne is a white British young woman, born in 1991, who has been interviewed three times as part of our research project, between 2002 and 2009. For all of that time, she has lived with her mother in a small flat on a post-war council estate in a major city. Anne’s mother is a support worker in a school. Her father, who lives nearby, has a disability and is no longer able to work. Anne has two half siblings: a sister, Natasha, who is 10 years older than her and a brother who is 8 years older. Natasha has lived some distance away in another region of the country for the whole period that we have known Anne, and she and her partner, Dean, now have a young daughter. Anne’s brother lived with her and their mother when we first knew her, then moved out to live locally, and latterly has been living in another town. By 2009 Anne was working in social care and studying part time for a qualification in the field.

Anne was 11 years old when she was interviewed for Wave 1 of our data collection in 2002, age 15 for Wave 2 in 2007, and 17 for Wave 3 in 2009. Our thematic analysis of Anne’s case across these waves has largely focused on her relationships with her siblings; tracing recurring themes in her account of these relationships: ‘possessions’, ‘talk’, and ‘protection’ (emotional and physical); identifying how she sees their presence as equating with emotional connection and absence with disconnection; and how gender weaves its way through these leitmotifs. Over time the themes have shifted around in various ways – for example, when Anne’s brother was no longer so available physically, five of his friends seemed to take on his mantle of protector, looking out for her in his absence, and indeed Anne referred to them as ‘extra brothers’. One of the interesting aspects of qualitative longitudinal research is not only that it provides in-depth access to the complexity of people’s lives and understandings as they play themselves out over time, and the dynamics and processes involved, but also that analysis is always contingent. Closure in the form of definite pronouncements and ‘findings’ about outcomes in a particular case or set of cases is not a realistic option when the next wave of data can reveal twists and turns in circumstances and subjectivity as well as continuity with the previous wave (McLeod and Thomson, 2009).

For the purposes of this discussion we are going to focus on change and continuity in Anne’s relationship with her older sister across the three Waves, working only with the parts of her interviews where Anne talks about Natasha in an explicit and sustained fashion. From these sections we have selected and present interview material with the ‘I’ statements highlighted, followed by the I-poems that we constructed from the interviews, sufficient to illustrate the analyses and to support the arguments about researcher ontology that we will make later on in this article.

Figure 1A presents material from the transcript of Anne’s Wave 1 interview, where she discusses visiting her older sister, Natasha, with their mother, doing things together and for each other, talking about life, her last birthday at her sister’s and the presents that she received, her sister moving away from home, and keeping in contact. It is followed by the I-poem constructed from the interview extract in Figure 1B.

OPEN IN VIEWER

Figure 1A.

Wave 1 – Transcribed interview extracts: Anne age 11

OPEN IN VIEWER

Figure 1B.

Wave 1 – I-poem: Anne age 11

Figure 2 draws on Anne’s Wave 2 interview, where she discussed the type of person that her sister is and the type of person that she herself is in comparison, what she receives from her sister, and reflection on how her childhood would have been different if she and her sister had been nearer in age and location. For reasons of space, we have dispensed with the interview extracts and go direct to the I-poem we constructed.

OPEN IN VIEWER

Figure 2.

Wave 2 – I-poem: Anne age 15

Figure 3 then presents an I-poem constructed from the parts of Anne’s Wave 3 interview where she discussed visiting her sister, how their relationship had changed, help she did and did not give on her visits, her own and her sister’s baby’s position in the family, and differences in character between her and her sister.

OPEN IN VIEWER

Figure 3.

Wave 3 – I-poem: Anne age 17

Contrasting thematic and I-poem analyses of continuity and change

Having set out the interview and I-poem extracts for our case study of Anne, we now move on to compare two descriptive overviews of change and continuity over time that we produced in 2008, based on this material using, respectively, thematic and I-poems analyses. The left hand column of Figure 4 is a summary of an analysis of recurring and counter-posed themes that we identified in Anne’s accounts across the first two Waves, while the right hand column describes the interplay of different ‘voices’ that we identified in how Anne spoke about herself across the two points in time. These overviews were produced separately and not with the intention originally of comparing the different methods of analysis. It was only after we had produced them that we realized that there appeared to be a distinction in how the interpretive analyst is positioned in each of them.

OPEN IN VIEWER

Figure 4.

Contrasting thematic and voices analyses across time: Anne

Our thematic analysis seems to be gazing at Anne, identifying a range of images and resources and their shifts over time in how her sister was part of her life and sense of self. So we focus on the way that, when she was 11, regular talk (in various forms) was important for Anne’s understanding of emotional closeness to her big sister (such as knowing what her voice sounds like), as was having been the giver and recipient of gifts and acts that symbolized that closeness (such as the ‘special sister’ necklace). By the time she was 15, our thematic analysis draws out the way that talk was still an important demonstration of emotional closeness, but Anne had become nostalgic about this, and built a scenario of the dangers involved if that closeness was not spatial and cohort too. Anne constructed a contrasting image of herself as bad and her big sister as good that had its causal roots in the past and the age gap between them.

In contrast, our voices analysis seems to be standing alongside Anne, identifying some of the voices she spoke in, or selves from which she spoke, and their shifts over time. So, when she was 11, we recognized two key voices in which Anne spoke of herself: ‘little sister’ (‘. . . helps me with my homework’) and ‘fond memory’ (‘I remember it quite well’). The little sister voice continues across time, and we identify it as a voice used by a 15-year-old Anne from which to speak (‘I think I’m a bit naughty compared to her’). The fond memory voice had shifted somewhat, becoming more nostalgic, more wistful and regretful, and was associated with the emergence of a naughty voice as part of how Anne thinks about herself (‘I think it would have been better for my life’).

Looking at the I-poem constructed from Anne’s interview when she was 17 (Figure 3), we can see that a new ‘grown up’ voice has taken hold. Anne is a working adult on holiday who deserves to rest. Further, she is an adult who understands that not only is her niece now the baby of the family but that she herself does not need spoiling. Interwoven and battling with this grown up voice, however, are traces of Anne’s naughty and little sister voices, coming together into a resentful voice (‘why do I have to do that? I’m on holiday. I do enough’). This voice speaks Anne’s resentment that her new grown up status and need for rest are not being recognized. She is still subject to the expectations that she will act as helpful little sister but no longer reaps the benefits (‘I don’t get spoilt no more’). ¹

A good indicative illustrative analytic contrast is provided by the difference between the central focus of the detached thematic analytic discussion of Anne talking about her ‘big sister’: ‘She was regarded as kind, helpful, supportive, someone with whom she could talk’, and the intense and immediate voices analytic identification of Anne’s ‘little sister’ voice derived from the I-poem: ‘I went to my sister’s house/I slept round my sister’s house’. It was this particular point that alerted us to the ontological shift that we seemed to be undergoing as researchers in moving from a thematic to a voices approach. We felt a very tangible and fairly fundamental shift in where we were placed as interpretive analysts. This was not a particularly comfortable experience or an easy shift to make for us (a point we return to below), steeped as we are in interpretive perspectives that understand interviews and interview data as co-constructions (Gubrium and Holstein, 2003; Kvale, 1996), albeit as we argued earlier that it was this very understanding that enabled us to reflect on such a shift in relation to analysis. We felt that, as analysts, we had moved ontologically, from looking at Anne to make our interpretations about her sense of self in relation to her older sister, to standing alongside Anne to interpret how she saw herself as a little sister in various ways into and through her mid-teens, and then standing alongside her as these senses of self then faded away or evolved in her late teens. We moved from looking at Anne as a young person who has a (shifting) relationship with her older sister, to focusing on Anne as a little sister and her emergent ambivalent sense of self in relationship in this respect.

Concluding issues: better or different?

Our intention in this article has been to highlight the way that different analytic methods implicitly place the interpretive analyst in different sorts of relationship to their interview subject and their data. The process of data analysis constructs an analytic mode of being in relation to the interviewee and their social reality. In particular, we have illustrated this point through a detailed consideration of the analytic process involved in producing I-poems from qualitative longitudinal interview data, to explore change and continuity in a case study young person’s sense of self over time. We compared how we understood those changes and continuities through the different analytic angles provided by the gaze of thematic analysis and the voices identified through I-poems.

I-poems are one way of tracing changes and continuities over time in a research subject’s sense of self. Thematic analysis provides another, equally important, angle on change and continuity in key topic motifs in the subject’s account. As we have shown through our detailed examples of analysing Anne’s interviews over time, at the same time as an I-poem analysis draws attention to the research subject’s stream of consciousness – how they understand and speak about themselves – and a thematic analysis focuses on recurring, arising and evolving issues, each also produces a particular, implicit, mode of analytic ontology – how the researcher is placed in relation to their research subject and their social world. We have characterized – and indeed experienced – these as ‘standing alongside’ for I-poems and ‘gazing at’ with thematic analysis. It is the case that intellectually, as we have said, we conceive of interview data as co-constructions between researchers and interviewees. Further, from this intellectual stance, the process of conducting an I-poem analysis cannot enable access to and centre an inner voice or set of voices, nor does the process of thematic analysis render a distanced subject. Both are dealing with a produced account and produced analysis; they merely tell different stories about the research subject in question, or at least emphasize different aspects of the subject in the story. Intellectually, in both cases, the data analyst is positioned in the same way ontologically to the research subject – constructing an interpretation of her/him and her/his social reality. And yet, there are qualitatively different and tangible resonances involved in each data analysis process for the felt ontological position of researcher-self in relation to interviewee-other.

In part, such experience of a shift in our ontological positioning as interpretive analysts may relate to the ‘who is listening?’ element of the Listening Guide (referred to in our earlier outline of the first reading of the method overall). One aspect of this relates to the issue that the tangible ontological shift that we felt ourselves making as we moved from a thematic to a voices approach was not a comfortable experience for two researchers who regard interview accounts as co-constructions. Within this shared epistemological stance, though, the interpellation to ‘standing alongside’ was easier for Ros Edwards than for Susie Weller. Susie had interviewed Anne and thus already had literally ‘gazed at’ and heard her, fixing Anne in her mind. In contrast, Ros had little prior image or knowledge of Anne having neither seen her nor listened to the audio recordings of her voice. Another aspect of ‘who is listening’ is that both of us are ‘big sisters’ (Ros to a sister and Susie to a brother). It may be that conducting I-poem analyses focused on sibling relationships (whether of Anne or other cases in our data) resonated in particular ways with us, creating a feeling of being closer to the interviewee as a ‘little sister’ through the constructed subject-self of I-poems (see also discussion of the implications of researchers’ sibling positioning when investigating relationships between brothers and sisters in Edwards et al., 2006). Whatever the substantive focus or foci of I-poems, it is likely that research analysts will have some personal experience for which the I-poem process has ontological connotations.

Nonetheless, we are not arguing that one analytic mode is superior to another – that it is better to stand alongside your interviewee than to gaze at them, even if only emotionally rather than intellectually. Quite the opposite. Every analytic process has its drawbacks. As we have already noted for I-poems, for example, these are practical and epistemological. On a practical level working with them is time-consuming and best used with a small or sub-sample, which may mean that pertinent ‘voices’ outwith the I-poem sub-sample are lost. On an epistemological level, the method explicitly can lead researchers to believe that they can be reflexive about their own positioning and reactions initially, and then put theoretical orientation and personal experience aside through an I-poem exercise that separates off their interpretations from an interviewee’s ‘authentic’ self-perception and inner world as contained in a transcript. But in addition to this, on a methodological level, I-poems cannot be regarded as the answer to either single snapshot or longitudinal data in qualitative research – clearly not all such research is wholly or partially concerned with participants’ sense of self in this way.

We believe that the thematic and I-poem forms of analysis that we have discussed here each provide a valuable angle of understanding, revealing different facets from which to undertake the making of meaning that is the basis for interpretation. Importantly, as part of the interpretive contribution, we contend that all forms of data analysis involve analytic ontologies, varying by approach, and potentially equally insightful if reflected upon. The crux is for researchers to recognize and reflect on the implications of those varying analytic modes of being in relation to the interviewee and their social world for how they make sense of their data.