{"paper_id":"c2dd9883-666a-4194-857d-7a22295fcb2e","body_text":"Despite a growing literature on the value of relational data in studies of social\nphenomena ( May, 2011 ;\n Smart, 2007 ),\nindividuals still commonly constitute the basic unit of analysis in qualitative\nresearch. Studies of illness experiences commonly give primacy to the account of the\nperson living with the condition, and while this allows a dedicated focus on\npersonal, subjective accounts, in some cases it neglects an important opportunity to\nexplore the relational nature of health and the specific ways in which social\nnetworks and intimate relationships can configure lives and experiences.\nInterviewing couples can offer rich relational data, which may be especially\nrelevant in research about chronic illness because of the potential impact of\nillness on relationships with family members and partners and because of partners’\nroles in managing illness and coping day to day ( Culley et al., 2017 ;  Hudson et al., 2016 ).\nRelationality, and the ‘relational turn’ in sociology, is concerned with\nunderstanding the constitution of a phenomenon through figurations, networks or\nsocial worlds and sees social relations as dynamic and fluid processes ( Dépelteau and Powell, 2013 ;\n Finch, 2007 ;  Roseneil and Ketokivi,\n2016 ;  Smart,\n2007 ). A relational framing therefore requires the use of methods that are\nable to effectively explore the points of connection between social actors ( Hudson et al., 2016 ;  Springer et al., 2012 ).\nThere exists a long tradition in the sociology of the family of interviewing\nmultiple family members which has directly informed the use and development of\nrelational methods more generally ( Finch, 2007 ;  Roseneil and Ketokivi, 2016 ;  Smart, 2007 ;  Valentine, 1999 ). However,\nthe literature remains incomplete and inconclusive with regard to the question of\nwhether interviewing jointly or separately is most desirable, with notably fewer\nauthors addressing the latter approach. Furthermore, few authors have tackled the\nspecific complexities inherent in analysing data from couple interviews. The\ndevelopment of relevant methods and approaches is therefore needed in order to\nfurther the advancement of a relational sociology of health and illness more\ngenerally.\nThis article offers methodological reflections arising from a study of the impact of\nthe gynaecological condition endometriosis on couples (findings from the empirical\ndata are reported elsewhere; see  Culley et al., 2013a ,  2017  and  Hudson et\nal., 2016 ). It describes our rationale for interviewing partners\nseparately, as well the related implications, benefits and challenges. This article\nseeks to contribute to the small body of methodological literature on interviewing\ncouples and on dyadic analysis and to offer insights to inform future relational and\ndyadic research. Despite ethical and interpretive complexities, it is argued that\nthe approach we employed presents a number of advantages. We begin with an outline\nof the study on which this article is based, which is followed by the rationale for\nthe methodological approach we adopted and the outcomes it achieved. The second half\nof this article describes our approach to undertaking dyadic analysis, as well as\nchallenges associated with reporting the data derived from these methods. We use\nexamples throughout to illustrate the challenges and advantages presented by our\napproach.\n\nEndometriosis is a common gynaecological condition, which occurs when endometrium\n(the lining of the womb) grows outside the womb ( De Nardi and Ferrari, 2011 ;  Dunselman et al., 2014 ).\nThe main symptoms of endometriosis are pelvic pain, heavy and painful periods,\nfatigue, dyspareunia (pain during sex) and subfertility ( De Nardi and Ferrari, 2011 ;  Lemaire, 2004 ;  Meuleman et al., 2009 ).\nThere is no definitive cure, but there are several treatments aimed at suppression\nof the disease and symptom relief with varying degrees of success ( Dunselman et al.,\n2014 ).\nEndometriosis has a significant impact on the lives of women, and while research\nsuggests that couple relationships are affected ( Denny and Mann, 2007 ;  Seear, 2009 ), few studies have included\nthis as a specific focus or have included women’s partners ( Culley et al., 2013b ). The ENDOPART study\nwas designed to explore the experience of heterosexual couples living with\nendometriosis and was conceptualised from a relational, dyadic perspective. Using a\nqualitative methodology, it aimed to explore the impact of endometriosis on women\nand their male partners, contribute to the development of theory in chronic illness\nand contribute to improving the wellbeing of people living with endometriosis by\nproviding an evidence base for improving couple support. It comprised three main\nphases: first, a ‘context-setting’ phase, comprising interviews with key informants\nand a systematic literature review ( Culley et al., 2013b ); second, in-depth,\nsemi-structured interviews with women with endometriosis and their male partners;\nand finally, a stakeholder workshop was held to inform the recommendations and\noutcomes from the study.\nA total of 22 couples were recruited to phase 2, and partners were interviewed\nseparately (n = 44). Couples were recruited via the national charity, Endometriosis\nUK (n = 11), NHS clinics (n = 5), other support or information groups or\norganisations (n = 3) and word of mouth (n = 3). Participants were given written\ninformation about the study and provided written consent. Interview schedules for\nthe women and their male partners were developed dyadically, for example, comprising\nsimilarly themed questions as well as a sub-set of questions to allow for direct\ncomparison of perspectives (see further details below). Interviews were recorded,\ntranscribed verbatim and entered into NVivo for analysis. Ethical approval was\ngranted by the host university and by the East Midlands Leicester NHS Local Research\nEthics Committee UK (reference 12/EM/0015).\nIn designing the study, the decision was made to use individual interviews but to\ndesign and analyse them according to a dyadic approach. We argue that dyadic\nresearch can be defined as that which takes a dyad – that is two people in a\npre-existing relationship – as the unit of study and foregrounds this dyadic\nrelationship in the study design and analysis ( Eisikovits and Koren, 2010 ;  Morgan et al., 2013 ;  Ummel and Achille, 2016 ).\nIn dyadic studies, the study aims, design and analytic focus treat relational\naspects (e.g. social ties, networks, interactions, processes, etc.;  Crossley, 2015 ) as the focus\nof inquiry, investigating accounts dialogically and looking at partners’ meanings in\ndialogue with one another. In keeping with previous literature, we argue that a\ndyadic approach does not necessitate joint interviews, but that dyadic analysis can\nbe undertaken on separate interviews with partners in a dyadic unit, with this\napproach comprising one of several possible ‘dyadic data collection modalities’\n( Eisikovits and Koren,\n2010 : 1643).\nAt the stage of study design, we purposefully adopted a two-part model of\n‘methodological best practice’ in order to establish rigour during interviewing.\nWithin this model, we determined that where it was feasible (1) participants would\nbe interviewed separately and (2) interviews would occur simultaneously, and\ntherefore, a different interviewer would be used for each partner. While we aimed to\nadhere to this model for each couple, this was not always possible due to\nrestrictions in participant availability (e.g. relating to childcare), researcher\navailability or funding for travelling long distances (the instances in which we\nwere not able to adhere to this model and the reasons for this are discussed below).\nThis was a pragmatic approach which acknowledged that it may not always be possible\nto follow best practice, but that where logistically possible we could ensure a\nrobust and repeatable data collection process within the resources and timescale of\na publicly funded research project.\nIn the following section, we discuss the rationale, implications, challenges and\nbenefits associated with these two aspects of the model: interviewing partners\nseparately and interviewing partners simultaneously with different interviewers.\nFollowing this, the second half of this article describes our approach to\nundertaking dyadic analysis, as well as challenges associated with reporting.\nWhile separate and joint interviews each have relative advantages, family\nresearchers have often advocated separate interviews when researching\ncouple-sensitive topics (e.g. sex and intimacy) and when discussing dynamics and\npower relationships ( Valentine, 1999 ). Interviewing partners separately enables each\nparticipant to ‘tell the story from his or her own perspective, without having\nto consider the reaction of the other when voicing criticism or bringing up\nsensitive topics’ ( Eisikovits and Koren, 2010 : 1643–1644; see also  Morris, 2001 ;  Ummel and Achille,\n2016 ). Using this approach increases the likelihood each interviewee will\ndisclose information that they would be unwilling to share in a couple interview\n( Valentine, 1999 )\nand potentially reduces the related ‘social desirability effect’ that would come\ninto play if their partner was present ( Taylor and De Vocht, 2011 ).\nIn an insightful piece of work, specifically exploring the gendered nature of\njoint versus individual interviewing,  Seale et al. (2008)  compared data from\njoint and solo interviews in studies about health, pregnancy and parenting\nexperiences. They found that in joint interviews, women were ‘more likely to\nachieve quantitative dominance’ and suggest that this may be because they were\nconsidered, by both their partner and the interviewer, to be the more\nappropriate spokesperson on these topics ( Seale et al., 2008 : 124). The\nexperience of living with or alongside endometriosis is, of course, focused on\nthe health and wellbeing of the affected woman. It follows therefore that many\naspects of couples’ everyday lives are constructed around the impact of\nendometriosis on the female partner. In planning our study, we felt that this\nwould also very likely have affected the interview dynamics, since in most cases\nthe female participants would likely (and justifiably) have dominated the\ninterviews. Seale and colleagues therefore advise that if researchers want to\nfind out more about men’s experiences, they should strive to design their\nresearch, and specifically their interview questions, in a way that focuses on\nmen’s lives and experiences ( Seale et al., 2008 ).\nFollowing the rationale in these studies – that individual interviews afford more\nspace to discuss sensitive issues and that women living with the condition may\nhave achieved ‘quantitative dominance’ – we interviewed all 44 participants in\nour study in separate, individual interviews. This approach afforded us a range\nof benefits in line with our specific objectives of wanting to address the\nabsence of men in previous studies on endometriosis and importantly, to hear\nmen’s accounts separately from those of their female partners. Data from our\ninterviews confirmed this decision, since participants expressed feelings and\nexperiences unlikely to have been forthcoming in a joint interview ( Valentine, 1999 ). For\nexample, both female and male partners reported concealing some aspects of\nliving with endometriosis from their partners. In the following excerpt, one\nwoman talks about how she hides the pain she experiences during sex from her partner: I’d love to be one of these couples who put everything on the\ntable and say ‘this is how it is, this and this and this’ but it’s\njust not how we work. If I told him, if I put everything out and\nsaid ‘when we have sex this is how I feel’, he’d run a mile, he\nwould never want to touch me again.  (Female partner)\nI’d love to be one of these couples who put everything on the\ntable and say ‘this is how it is, this and this and this’ but it’s\njust not how we work. If I told him, if I put everything out and\nsaid ‘when we have sex this is how I feel’, he’d run a mile, he\nwould never want to touch me again.  (Female partner)\nIn the following example, this male partner discusses how he chooses not to\nreveal to his partner his concerns about whether or not they will be able to\nhave children as a result of her endometriosis: 1\n‘It’s a worry to me that we might not be able to have\nchildren’ .\n‘Have you talked to [partner] about your worry?’\n‘No probably not, not as much as, it’s something I am not as open\nwith because, I don’t know, I don’t want her to feel like it’s her fault\nbecause it’s not. I think if I just said that I was really worried and\nthe worries that I do have might make her feel bad and I don’t want that\n… it’s probably something I have kept more concealed just for the sake\nof sparing her feelings’ .\nIn discussing the impact endometriosis had on them, men frequently positioned\nwomen’s needs as paramount and as a result some felt that expressing their own\nemotions, concerns or needs within the relationship would be inappropriate or\nselfish. Separate interviews therefore allowed men to express their own views\nand feelings, whereas joint interviews may have resulted in a paucity of such\ndata. Furthermore, the interviews highlighted that men’s needs in relation to\nliving alongside endometriosis may be marginalised as the focus of support and\ntreatment resides with the female partner ( Culley et al., 2017 ). This\nmarginalisation may have been further exacerbated with the use of joint interviews: I’m so glad you’re asking because … like I said before the chaps\njust get, you know, with doctors, nurses, everything going on, and\nthey’re just so worried about their wives, girlfriends and they’re\njust pushed to one side and it’s so important to get their point of\nview.  (Male partner)\nI’m so glad you’re asking because … like I said before the chaps\njust get, you know, with doctors, nurses, everything going on, and\nthey’re just so worried about their wives, girlfriends and they’re\njust pushed to one side and it’s so important to get their point of\nview.  (Male partner)\nInterviewing partners separately also potentially reduces the related ‘social\ndesirability effect’ that would come into play if their partner was present,\nthat is, the tendency for interviewees to perform or present a particular ‘self’\ndeemed acceptable to their partner ( Taylor and De Vocht, 2011 ;  Werner et al., 2004 ).\nHowever, while this tendency is reduced in separate interviews, we found that\nsocial desirability effects do not disappear completely and that this can\ninclude the way in which an interviewee presents both themselves and their\npartner to the interviewer ( Taylor and De Vocht, 2011 ). As  Eisikovits and Koren (2010)  have\nsuggested, even when separate interviews are conducted, there exists ‘a joint\nrelationship and history, and as a result the partner is often virtually present\nin the interview space’ (p. 1644).\nIn our study, this manifested in participants’ desire to portray their partner\nand relationship in a positive light, and this was done by both men and women\nalbeit in slightly different ways. Men appeared defensive and protective of\ntheir partner especially when talking about the legitimacy of their symptoms and\ntheir partners’ experiences with unsupportive healthcare practitioners,\nemployees, colleagues and friends and family. Women in comparison were more\nwilling to speak negatively about their partner but when they did, were likely\nto include qualifying statements about how hard their partner was trying or how\nsupportive they were and so repositioning them in a positive light. Many\nappeared to avoid or minimise speaking in a way that could be perceived as a\nbetrayal and to avoid being judged as a ‘bad or unmatched couple’ ( Valentine, 1999 : 71).\nNonetheless, alongside these displays, participants also spoke at length about\nnegative aspects of their relationship and in many cases did voice criticism of\ntheir partner; therefore, while displays of social desirability were evident,\nthese did not dominate the interview and may have been more frequent in joint\ninterviews.\nSeparate interviews have been criticised for missing the opportunity to capture\ninteraction and observe negotiation, mediation and dominance between partners\nwithin an interview ( Arksey,\n1996 ;  Morris,\n2001 ;  Seale et\nal., 2008 ;  Valentine, 1999 ). However, we would suggest that any\n observational  data gathered in joint interviews are\nsubstantively different in nature from the  verbal  accounts\nprovided by interviewees and that this may therefore present challenges in\nanalysis; something which is seldom explored. As  Morgan (2010)  argues, treating\ninteraction as something which produces data is different from treating\ninteraction  as  data. This study employed a constructivist and\nrelativist philosophical position, avoiding the pursuit of an underlying ‘truth’\n(see further discussion below). In keeping with this position, we argue that\ninterview accounts should be seen as social constructions, subject to a range of\nshifting positionalities and contextual factors, and that joint accounts are not\ntherefore any more ‘valid’ than separate ones.\nA second feature of our methodological model was that different interviewers\ninterview partners within a couple unit simultaneously where possible, allowing\nfor the logistical restrictions described above. Interviewing partners in this\nway avoids the possibility of partners discussing issues with each other between\ninterviews and thus prevents the possibility that an interview with the second\npartner is influenced by such a discussion ( Eisikovits and Koren, 2010 ;  Ummel and Achille,\n2016 ). As suggested above, while we are not assuming in epistemological\nterms that accounts can ever be ‘untainted’ or unbiased, nevertheless we wanted,\nwhere feasible, to give participants the opportunity to tell their story\nuninhibited by prior expectations or discussions about the interview, or by\ntheir partner’s experience of taking part.\nAs stated above, we were not always able to adhere to this model for practical\nreasons. In practice, we achieved this intention in 17 out of 22 cases. In the\nremaining 5 cases, this was not possible due to participants’ childcare\nrequirements and/or researcher availability or travel resources. However, in 4\nof these cases, the interviews were conducted in immediate succession by the\nsame interviewer. In one final case, we were unable to time the interviews to\nhappen simultaneously or in immediate succession, and instead, they took place\n4 ;days apart. Therefore, we achieved our objective of ensuring that couples\ncould take part without prior discussion of a partner’s interview in 21 out of\n22 cases.\nThis approach had very practical implications. Using two researchers increases\nthe time, cost and organisation needed for data collection. In most cases, we\nwere required to co-ordinate four diaries (two researchers and both members of\nthe couple). This approach also doubles the travel (and if needed,\naccommodation) costs for the research, something which needs to be factored into\nfunding applications for dyadic research. There were also practical issues for\nsome couples who wanted to do the interviews at home and were therefore required\nto find two suitable rooms in the house for interviews to take place. This was\nnot always possible and so in some cases led to the need for an alternative\nlocation for the second interview. For those with children, booking the\ninterviews at an appropriate time where they could both take part raised\nchildcare issues. Finally, in some cases, the logistics made it just too\ndifficult and/or expensive for two researchers to attend; for example, in two\ncases, the distance to the couple was so great that to send two researchers was\nnot possible. One interviewer attended and conducted both set of interviews\nsequentially. This adds further challenges for the researcher when interviews\nare lengthy and sensitive.\nThe decision to use different interviewers in simultaneous interviews was also\ninformed by a desire to reduce the possibility of the interviewer bringing prior\nknowledge, gained in the first interview and relating to the couple unit, into\nthe subsequent interview ( Eisikovits and Koren, 2010 ;  Ummel and Achille, 2016 ). Doing so\nraises ethical concerns and specifically presents potential difficulties for\nconfidentiality if the interviewer inadvertently discloses information\npreviously given by their partner ( Tolich, 2002 ;  Zarhin, 2018 ). There is also the\npossibility that interviewees directly ask the interviewer to tell them what\ntheir partner said or that the interviewer is asked to take sides ( Zarhin, 2018 ). Indeed,\nthis was raised by one participant, despite the interviews taking place\nsimultaneously:\n‘I wonder what sort of questions your friend’s going to ask him. Tell\nme’ .\n‘Quite similar questions really, it’s quite\nconversational’ .\n‘Will she ask him questions like “how does your wife cope with the\nhousework?” and things like that or would she ask him questions just\nabout him? … Tell me what he thinks. He won’t mind’ .\nParticipants may also have felt reassured and more comfortable expressing\nthemselves knowing that their interviewer would not, following their interview,\nbe then also meeting with their partner:\n‘[Partner] will, we’ll be sitting there on the sofa watching the\ntelly – [partner] doesn’t get to hear this does she?’\n‘No absolutely not’ .\n‘There are certain little things that she does that I’m aware of and\nI’ll say ‘oh do you need a tab [tablet]?’, and she’ll say ‘how did you\n[know]?’\nUsing different interviewers therefore avoids a situation of a researcher being\n‘stuck in the middle’ of a couple and the ethical implications this presents\n( Forbat and Henderson,\n2003 ).\n\nWhile there is now a small but growing literature on the relative benefits of\nseparate or joint interviews for the data collection process, much less has been\nwritten about the process of analysis. In dyadic research, the couple is the unit of\nanalysis; it is not enough to sample both partners – it is the focus on\nrelationships and patterns within couple units that makes research dyadic – the ‘we’\nof the experience ( Eisikovits\nand Koren, 2010 ).  Eisikovits and Koren (2010)  appear to offer the most in terms of\noutlining a detailed and systematic dyadic analytic method. They suggest that dyadic\nanalysis alters the individually based interpretation of the data and gives an\nadditional dimension to understanding. Epistemologically they suggest that the\nresearcher creates a third dyadic version, while retaining the individual accounts,\nhighlighting that ‘the dyadic version is more than the sum of two individual ones’\n( Eisikovits and Koren,\n2010 : 1652). In order to achieve this, they propose the following\nstrategy. First, analysis of the individual interviews in line with usual principles\nof qualitative analysis, that is, a thematic ‘horizontal’ analysis, which seeks to\nconstruct a number of common themes across the cases. In the second, dyadic stage,\nthe focus is on each couple unit and comprises a systematic identification of the\ncontrasts and overlaps between partners’ accounts. This is carried out at two\nlevels: the ‘textual and descriptive’ level, which they suggest reveals what they\nterm the ‘open reality’ of the situation, and the ‘sub-textual and interpretive’\nlevel, which seeks to explore what they term the ‘hidden reality’ in the accounts,\nthat is, how the individuals are interpreting phenomena ( Eisikovits and Koren, 2010 : 1653). In the\nENDOPART study, we adapted Eisikovits and Koren’s model, as described below.\nOur approach to dyadic analysis began at the level of question design ( Eisikovits and Koren,\n2010 ). As well as employing a general dyadic method to questioning\n(which involved asking female and male participants similarly themed questions),\nwe also devised a sub-set of interview questions that were designed to allow us\na more direct comparison of perspectives in the analysis. For example, we asked\nparticipants ‘ what is the single biggest issue for you and what do you\nthink is the single biggest issue for your partner? ’ which allowed\nus to explore the similarities and differences in the direct responses to this\nquestion. While we recognise the limitations of such directive questions, this\nstrategy provided some focused data on the relative agreement within the couple\nabout the global impact of endometriosis on their lives as well as participants’\nlevels of awareness regarding their partner’s experience, when compared to the\nmore open data collected about a number of more specific domains such as home\nlife, work, childcare, healthcare and fertility (see  Culley et al., 2013a ,  2017  and  Hudson et al., 2016  for\nfindings).\nSecond, we used the themes that emerged from the stage 1 thematic analysis to\ndirectly inform our stage 2 dyadic analyses. While Eisikovits and Koren suggest\nthat dyadic analysis is conducted throughout the entire data set, our stage 1\ndata analysis demonstrated that certain themes were more dyadic in character\nthan others and that a more detailed, intense focus on these themes would\nprovide a richer relational account of the couple experience of endometriosis\n(see  Hudson et al.,\n2016 , for discussion of these findings). Two themes were selected\naccording to their dyadic nature: that is how significant they were for the ‘we’\nrelationship. They were ‘sex and intimacy’ and ‘planning for and having\nchildren’. In contrast to research where the questions are centred exclusively\non the substantive matter of couple  relationships  (as in  Eisikovits and Koren’s\n(2010)  study), in our study, we were also interested in gendered\nindividual experiences of living with/alongside a specific chronic condition and\ntherefore we also sought to retain an empirical focus on women’s and men’s\nindividual accounts as well as exploring the impact at the couple level. This\ndecision was therefore led by our specific study objectives, but means that only\nspecific elements of our data were analysed dyadically and as such our approach\nsignifies a slight departure from Eisikovits and Koren’s approach.\nEisikovits and Koren’s model focuses on the identification of contrasts (opposing\ndescriptions of situations, phenomenon, feelings and experiences) and overlaps\n(converging descriptions of situations, phenomenon, feelings and experiences) in\npartners’ accounts, and we followed this approach for our two dyadic themes. A\nfurther adaption we made to their model was the identification of\n‘ omissions ’ as a particular kind of contrast in partners’\naccounts, and one which we suggest represents a new analytical category in\ndyadic research. Omissions were represented by topics that were discussed by one\npartner but not the other. This allowed us to further interrogate the different\nemphasis each partner gave to particular issues. For example, with regard to sex\nand intimacy, analysis revealed that men frequently omitted several topics that\ntheir partners discussed, including bleeding during or after sex, seeking advice\nand support externally and the loss of intimacy. With regard to planning for and\nhaving children, men frequently omitted discussion about biographical\ndisruption, the challenges in negotiating fertility-compromising pain treatments\nand trying to conceive, and a lack of support or pressure from families, whereas\nwomen were more likely to omit discussions about the financial implications\n(e.g. of in vitro fertilisation (IVF)) and either their or their partners’\ncoping strategies. These were not identified at the general level, that is, by\ncomparing the data set from all men with the data set from all women, but at the\n couple level  by identifying contrasts, overlaps and\nomissions relating to these factors between partners’ accounts. Exploring the\ndegree of completeness in accounts ( Van Dijk, 2001 ) illuminates the ways in\nwhich participants’ narratives are partial. The identification of what is\nomitted is insightful in understanding the factors that do and do not constitute\nindividuals’ social worlds: ‘what is left unsaid is often more important that\nwhat is said’ ( Huckin,\n2002 : 162). In a dyadic study, this is of paramount importance:\ncomparing partners’ accounts in this way allows for an exploration of the\nfactors which do and do not constitute each individual’s experience of\nendometriosis and how these might differ between partners within a couple unit.\nFurthermore, including the analytic category ‘omission’ was necessary in order\nto avoid an incomplete coding exercise. It became apparent early on in the\nactivity that if the data could only be coded to ‘contrast’ or ‘overlap’, this\nwould have left a considerable amount of data uncoded.\nFinally, as described above,  Eisikovits and Koren (2010)  propose that stage 2 of the analysis\nshould involve the exploration of contrasts and overlaps at two different\nlevels: the ‘textual and descriptive’ and the ‘sub-textual and interpretive’\nlevel (p. 1653). We opted not to systematically apply the two levels of analysis\nin our study for two reasons. This was partly practical, due to the size of our\nsample (n = 22 couples, 44 interviews) and the resource implications of focusing\nin such a detailed way on a data set of this size, but mainly this related to\nthe ontological and epistemological tensions that this approach presents ( Bjørnholt and Farstad,\n2014 ;  Manning and\nKunkel, 2015 ). Seeking to explore the interpretive or ‘hidden\nmeaning’ in the accounts was not in keeping with the constructivist\nphilosophical position we had adopted in the research more generally ( Braybrook et al., 2017 ).\nWe proposed that the cultural repertoires which people draw upon are important\nin understanding illness experiences and therefore decided to keep our analysis\nof overlaps, contrasts and omissions at the textual and descriptive level in\norder to compare the ways in which couples discussed and presented their\nexperiences and perceptions in the interviews, but not to attempt to ‘go beyond’\nthese representations in our analysis. While we recognise the existence of a\nrelationship (albeit a complex and contested one) between people’s\nrepresentations and their experiences ( Bjørnholt, 2011 ), we were not attempting\nto directly assess or evaluate these ‘real life’ experiences via the couples’\naccounts.\nOverall, this approach to dyadic analysis – exploring contrasts, overlaps and\nomissions in partners’ accounts – enabled a detailed insight into the impact of\nthe condition on the couple relationship. 2  It illuminated a range of shared and different interpretations,\nexperiences, understandings and meanings within each couple unit. These kinds of\nanalyses allow a consideration of how life with or alongside the condition can\nresult in markedly different experiences for each partner and contrasting coping\nstrategies which may cause challenges for couple wellbeing, and can produce and\nre-produce particular gendered subjectivities.\nInterviewing members of a dyad separately causes complexities in subsequently\nreporting the data (see, for example,  Saunders et al., 2015 ;  Ummel and Achille,\n2016 ). An accepted convention when reporting quotations from interviews\nis to assign a descriptive label to the quotation, providing the reader with\ncontextual or demographic information about the participant (e.g. individual’s\nage, length of relationship and time since diagnosis). This has not been\npossible because identifying the data in this way, especially in a small study,\nmay allow partners to identify one another and thereby know exactly what their\npartner said in the interview.\nAn alternative might be to use a unique identifier code or a pseudonym; however,\nif couple quotations are presented together (e.g. male 1 and female 1) and a\nparticipant recognises their own quotation, they will therefore also be able to\nemploy ‘jigsaw identification’ to identify the other quotation as coming from\ntheir partner ( Forbat and\nHenderson, 2003 ;  Saunders et al., 2015 ). If a participant could identify one\nquotation from their partner, they would then also know that all other\nquotations assigned to that particular unique identifier or pseudonym were also\nfrom their partner.\nThese issues are all the more significant in a study which included distressing\naccounts and the potential for this information, if identities were uncovered,\nto have profound effects on relationships (see also  Ummel and Achille, 2016 ). For example,\none partner spoke several times about considering ending their relationship.\nAlthough we alerted couples, in the information we provided prior to them giving\nconsent, to the slight possibility that they may identify one another from their\nquotations, we could not have fully anticipated the extent of the impact of\nendometriosis on some couples and the potential implications of participants\nidentifying their partner’s accounts. Data management and anonymity therefore\nbecame of added importance when reporting findings at conferences and in\npublications. Because of the challenges associated with descriptive labels,\nunique identifiers or pseudonyms, and due to the enhanced relational sensitivity\nof the data, we have used ‘female participant’ and ‘male participant’ to present\nfindings in a way that does not compromise internal confidentiality ( Tolich, 2004 ). In\naddition, we have limited the extent to which we present accounts from both\npartners side by side, only doing so when necessary. In these cases, we have\nalso avoided the use of direct quotations and removed or modified specific\ndetails deemed to be highly identifiable and/or relationally sensitive, while\nstill seeking to provide a valid illustration of the arguments presented. While\nthis approach may compromise the integrity of the data, and is a difficult\nbalance to strike, ultimately we have prioritised ethical defensibility over\nrichness in reporting dyadic analysis ( Saunders et al., 2015 ;  Ummel and Achille,\n2016 ).\n\nThe relational turn in the social sciences requires us to consider more effective\nways of understanding the complexity of social life and social relations between\nindividuals. In the context of research about long-term conditions such as\nendometriosis, this can facilitate an improved understanding of the ways in which\nindividuals, couples and families navigate a complex web of symptoms, diagnoses and\nmedical and non-medical management. While individual, solo interviews offer\nprivileged access to individual narratives about lives shaped by chronic illness, a\ndyadic approach can enrich research accounts and offer a means by which to\nunderstand the wider social and relational implications of health and illness as it\nimpacts on social networks and how these connections in turn may act as a source of\nsupport or a source of additional stress.\nDespite these potential advantages, there is a paucity of literature on interviewing\ncouples and the specific methodological challenges of this approach, especially in\nrelation to data analysis. There is a particular gap in the methodological\nliterature regarding the unique contribution and advantages offered by interviewing\npartners in a dyad  separately . Reflections from this study\ntherefore contribute to a small body of methodological scholarship on interviewing\ncouples and offer novel insights on the practical, methodological and ethical\ndilemmas involved in conducting separate interviews about chronic illness with both\npartners in a heterosexual couple. We were able to explore accounts that were\nunlikely to have emerged in joint interviews, and which validate the decisions made\nregarding separate, simultaneous interviews. This approach allowed participants to\narticulate perceptions and experiences considered to be relationally or emotionally\nhighly sensitive and to share problematic aspects of relationships, and permitted\nmen’s accounts to be heard unmediated by women’s participation. The best practice\napproach employed, coupled with a dyadic analysis employing a careful and detailed\nfocus on partners’ accounts in relation to one another, allowed a unique insight\ninto how couples navigate this common chronic condition.\nThis approach is not without its challenges, including the display of social\ndesirability ‘talk’ within the interviews, the need to carefully consider logistical\nand resource implications in planning and managing the project and specific ethical\nconcerns regarding anonymity when reporting. Some of the challenges and limitations\nof separate interviews may be addressed by combining both joint and separate\ninterviews in the same study (see for example  Butt and Chesla, 2007 ). Such a strategy\ncould offer a way to capture relational accounts while minimising the unique\nlimitations of each approach. However, it may present additional challenges and\ncomplexities and its appropriateness will depend on the aims of the study in\nquestion ( Ummel and Achille,\n2016 ).\nDespite the challenges presented, interviewing partners separately and undertaking\ndyadic analysis offered us an effective method for the exploration of couple\nexperiences of endometriosis. Adopting a model of methodological best practice\nafforded us some flexibility when negotiating the realities of carrying out funded\nhealth research and allowed us to make decisions guided by a pre-agreed protocol,\nsupported by existing research evidence, which was tailored to our specific research\nobjectives. This approach also allowed us the space to systematically record,\nreflect upon and report the success of our design and its associated limitations and\noffered us a means to gain detailed insight into the impact of the condition on the\ncouple relationship. We would therefore recommend this approach to others who wish\nto consider the use of such a method.","source_license":"CC0","license_restricted":false}