Abstract
Keywords
Introduction
Interviewing couples can provide valuable insights across disciplines and topics concerned with relationships. There are several different interview modes available (see Eiskovits & Koren, 2010). For example, separate partner interviews may be held concurrently, consecutively, or on different days, with the same interviewer, or with different interviewers. Joint interviews may be undertaken with both partners interviewed by one or two interviewers, held once or repeated. As others have pointed out, the dominance of the interview as a qualitative method can lead to taken-for-granted assumptions about its choice as an appropriate method, without further consideration of its mode (Brinkmann, 2013; Gray, 2013). A lack of reported reasoning for adopting a particular mode may reflect an assumption that the research aims act as a primary driver for choosing data collection methods (Braybrook et al., 2017). Thus, if the aims are reported, the data collection method is explained. However, this does not provide the rationale behind the choice of mode nor a detailed description. The aim of this paper is to contribute to methodological discussion regarding interview mode, by reflecting on a study which utilized a single combined joint and individual interview approach with couples to explore attributes of satisfying enduring relationships (Blake & Janssens, 2021; Ewing et al., 2020). Echoing an interpretative hermeneutic-phenomenological lens, this reflective approach was guided by the notion that researchers cannot bracket their own experiences to undertake value-free research (King & Horrocks, 2010; Laverty, 2003). We were interested in how our assumptions and decisions may have influenced the production of knowledge.
The paper is divided into two parts. In part one, we consider the context of the study and how the interview mode was conceptualized. We discuss the unit of analysis and object of inquiry, as well as our evaluation of two pilot interviews, before then describing the chosen multi-level interview mode. In part two, we reflect on how the multi-level interview approach worked in practice. Specifically, in relation to participant recruitment, disclosure, and the avoidance of harm. We discuss how a multi-level interview design may balance the strengths and weaknesses of individual and joint interviews, but also leaves us with familiar ongoing challenges regardless of interview mode. We reflect on what it means to undertake a “successful” interview and translate our experience into a list of guiding reflective questions for researchers considering semi-structured interviews.
Part One: Study Design
In this part, we describe the context of the study, how we reached our decision regarding the mode of interview and the multi-level interview design we adopted. Our qualitative study formed part of a wider project that aimed to identify relationship attributes critical for sustaining relationships, to inform the development of a relationship toolkit for young people (Ewing et al., 2020). The specific aim of the cross-sectional study, discussed herein, was to explore attributes of 10 long-term couple relationships (15+ years) across relationship forms (cohabitants, married, civil partnerships, same and opposite-sex). The sample size was predefined before recruitment to reflect the resource intensity of long interviews and detailed analysis which retain the unique circumstances of each case (Gabb & Fink, 2015). Including the two pilot interviews (which we discuss within this paper), the sample comprised 12 couples; six married, four cohabitant, two civil partnerships. Four couples were same-sex and eight opposite-sex. Interviewees’ age ranged from 37 to 73 (mean average 58 years) and length of relationship from 15 to 51 years (mean average 28 years). Of the 24 individuals, 14 had been educated to degree-level or higher and 23/24 described themselves as White British. Ethics were reviewed and approved by the Social Sciences and International Studies College Ethics Committee, University of Exeter [#201617-018]. Interviewees provided both written and verbal individual consent to participate. To preserve anonymity, interviewee characteristics have been left out or changed, without impacting the points under discussion.
The initial plan was to conduct face-to-face semi-structured interviews with each partner separately to align the method with a longitudinal study of married opposite-sex couples within the wider project. Our rationale for choosing in-person semi-structured interviews reflected the adoption of this approach in the longitudinal sample and the exploratory nature of the research design. The use of a similar topic guide, within each interview, would provide a degree of uniformity to enable comparison within the sample and with the longitudinal sample. The flexible nature of semi-structured interviews would allow narrators to purposefully select events for the listener to understand how they make meaning out of their experiences (Riessman, 2008) and provide opportunities for elaboration and clarification (Kelly, 2010; Rubin & Rubin, 2005). It was recognized that the cross-sectional study would not be measuring change over time, as per the longitudinal study. However, the research team discussed the extent to which single individual partner interviews would result in comparable data with repeated interviews of a longitudinal study. The interviewer of the longitudinal study had met with the couple partners in this sample on a prior three occasions over the past decade. In one-off interviews, would individual or joint interviews provide a deeper understanding of attributes of the couple relationships?
Existing Literature and Pilot Interviews
To decide the interview mode, we scoped existing methodological literature and undertook two pilot interviews using different interview modes. Relationship studies typically interview couple partners on a one-to-one basis (Gabb & Fink, 2015; Maclean & Eekelaar, 2005). As Hertz (1995) explains, interviews with wives to represent a household, moved to individual interviews with both couple partners to recognize the two realities in a marriage. However, Eiskovits and Koren (2010) suggest that interviewing couple partners together has gained increased attention in recent years. This has led to an emerging evidence base which largely compares the advantages and disadvantages of individual or joint interviews (Braybrook et al., 2017; Taylor & de Vocht, 2011; Valentine, 1999) or advocates for one preferred approach (for individual interviews, see Eiskovits & Koren, 2010; Mellor et al., 2013; for joint interviews, see Mavhandu-Mudzusi, 2018; Zarhin, 2018). Overall, this literature suggests that neither individual nor joint interviews with couples are superior, with each approach having its strengths and weaknesses. Our pilot interviews with couples in long-term relationships supported this position.
The first pilot involved two separate consecutive interviews, one with each partner. The second pilot interview was carried out jointly with the couple partners together. At the end of each interview, the interviewees were asked how they felt about taking part, whether they would have preferred to have been interviewed together, or separately, and if the mode of interview would have influenced their decision to take part. While on the spot it may have been difficult to be contrary, all four reported that they were glad to have taken part and that they would have taken part had the mode of interview differed. The couple who were interviewed jointly thought that it was nice to do it together as a shared experience, while the couple interviewed separately did not express a preference.
The audio recordings of the pilot interviews were listened to by the interviewer of the longitudinal sample (JE) to independently consider how the interviews flowed and whether joint or separate interviews would best collect comparable data. As per Bjornholt and Farstad (2014), the joint interview was shorter (1 hour compared to 2 hours for separate interviews) and it was possible to get a sense of the couple relationship through observing their interactions and how they support one another. For example, when discussing intimacy, rather than address the interviewer, the husband turned to his wife to tell her how attractive he found her and how this had not changed over time. For practical reasons, the individual interviews were held consecutively. A consequence of this, was that it was less natural to prompt for more detail in the second interview due to having already heard a similar answer in the preceding interview. The interviewer (SB) also found it emotionally draining to authentically express interest when the same stories were being shared in the separate interviews.
However, the individual interviews provided interviewees with an opportunity for self-reflection. For example, one of the partners questioned whether her relationship commitment would change in the future. She spoke movingly about her fear for her relationship as her partner was showing early symptoms of dementia. As her partner did not mention his health in his interview, SB and JE speculated whether she would have brought this up in a joint interview. This was not the only difference in accounts from the pilot individual interviews. One interviewee described how she had felt daunted by her partner’s prior relationships when they met as she had no prior relationship experience. However, in his interview, her partner stated that he had not had any prior relationships. Such differences may indicate forgetfulness or varied interpretations of concepts, such as what counts as a prior relationship in this instance. As per Mellor et al. (2013), when couple interviews are carried out individually, the interviewer must manage confidentiality and avoid disclosing information gleaned from one interview to the other. Whereas with joint interviews, “the researcher does not become the medium through which confidential and possibly sensitive information about conflicts may be pieced together in ways that may be unintended by the individual participants, and which may have a negative impact on their relationship” (Bjornholt & Farstad, 2014, p. 6). With a cross-sectional study design, the interviewer would only be meeting the interviewees once. As interpretations of any differences in accounts from individual interviews would be left to the researcher’s frame of understanding, the pilot interviews confirmed the importance of situating experiences in the context of the couple relationship.
Object of Inquiry and Unit of Analysis
Brinkmann (2013) has suggested that choosing between individual and joint interviews reflects a methodological tension between phenomenological approaches which emphasize meanings are essentially out there to be articulated by an interviewee, and constructionist approaches which emphasize meanings are constructed dialogically. In studies deploying couple interviews, few papers explicitly describe their object of inquiry and whether their unit of analysis is individual or couple experiences. Offering couples a choice as to whether to participate in joint or separate interviews has been seen as an empowering approach and a way to improve participation (Braybrook et al., 2017). In line with a principle of nonmaleficence, it was important to prevent potential harm and secure the well-being of interviewees (Farrimond, 2012). In research with couples, this included considering the effect of study participation on the relationship and a priority to avoid relationship disruption. We were concerned that if we offered a choice to be interviewed jointly or separately, and one partner actively opted for individual interviews, this may be perceived as negative evidence of the partner having secrets to share (Taylor & de Vocht, 2011). In studies where a choice of mode is offered to participants, few have described how they dealt with the added challenge of working with different units of analysis. Sayer and Klute (2005) suggest that due to the dominance of methodological individualism, a research question can be asked at a dyadic-level but data is collected at an individual level and aggregated by the researcher to report couple perspectives. Eiskovits and Koren (2010) describe how they undertook dyadic analysis from separate interviews. Often, however, despite the pivotal placement of the researcher’s frame of understanding, there is little clarity about how data is interpreted and presented from joint or separate interviews.
For our study, were we interested in exploring relationship attributes as a shared experience between a system of two interacting people or, to leave room for asymmetry, two individual experiences? If the object of inquiry is considered a collective perspective and there is interest in how a couple interact, then a joint interview may be more appropriate. Indeed, Mavhandu-Mudzusi (2018) was clear that as she was looking to understand how couples (not individual partners) described their experience in a serodiscordant relationship, joint interviews were the best option. If the object of inquiry is viewed as an individual experience and interest is in how partner views differ from each other, then individual interviews may be more appropriate. Reflecting our epistemological lens, our view was that researching relational data required a method which took account of the complex, messy object of inquiry (Law & Singleton, 2005). We felt that the meaning people ascribe to relationships would be polyvocal and contradictory, permitting multiple readings and interpretations (Brinkmann, 2013). Taking account of the interactional nature of semi-structured interviews and the role the researcher plays in the production of knowledge, we felt that a refined method was required. Bjornholt and Farstad (2014) suggested a solution could be to interview parties together and separately. This approach was adopted by Heaphy and Einarsdottir (2012) whose study shared a similar methodology and object of inquiry to ours. We therefore decided to move beyond our initial assumed choice between individual and joint interviews to adopt a multi-level approach, with the couples interviewed jointly and separately consecutively in one visit by a sole interviewer. By looking at both individual and dyadic perspectives as our units of analysis, we hoped to gain a rich picture of a complex object of inquiry.
Multi-Level Interview Design
The interview design, including the focus of each part of the interview and approximate timings, is set out in Figure 1. We started the interview with the partners together and then undertook separate interviews with each partner, as per the mode adopted by Heaphy and Einarsdottir (2012). Valentine (1999) suggested dividing topic guides into areas that logically make sense to be undertaken individually (such as past relationships and parental relationships) helps couples accept the need for separate confidential interviews. With this in mind and taking cues from the pilot interviews as to which questions generated considerable repetition and where answers were reported as singular “I” and plural “we” experiences, the topic guide was divided into areas that were more likely to be reported as shared (joint interview) or individual experiences (separate interviews).
Multi-level interview design.
The joint interview at the start provided formal time for introductions, verbal checks on consent and discussion of expectations. For example, reassurance from the interviewer that there were no right or wrong answers. Expanding the Heaphy and Einarsdottir (2012) approach, the couples decided who would be interviewed first in the individual interviews and the partners were brought back together to finish with a joint session. We were mindful that interviews end “but couples leave them together. It is the researcher’s obligation to do their best to protect interviewees from harm caused by the research, not only during interviews but also in their aftermath” (Zarhin, 2018, p. 851). So, the final joint session of the interview aimed to leave couples in a good emotional state (Mavhandu-Mudzusi, 2018). Existing literature suggests that separate interviews for couple partners can feel like a test as to whether what they say will match (Mellor et al., 2013; Taylor & de Vocht, 2011). The couple interviewed separately in our pilot interviews and those in the longitudinal sample also expressed such a sentiment, albeit in a light-hearted way. From our experience from prior qualitative studies, interviewees often mention additional points once the interviewer turns the audio-recorder off. So, we anticipated that a joint end session would enable partners to raise something discussed in their individual interview which they wanted their partner to know they had said and add their thoughts.
The interviews were all conducted by the same experienced interviewer during a single visit which lasted from two to three hours for the interviewer and one and a half to two hours for each of the interviewees. We provided three distinct opportunities to empower interviewees within the interview design. Firstly, by arranging the interview to be held at a date, time and location chosen by the interviewees. Ten of the couples were interviewed within their home and two of the couples were interviewed in a private room within one partner’s place of work. Secondly, by offering interviewees the choice as to who went first with the individual interviews. As this choice had been explained in the study information sheet provided before the interview, interviewees seemed to have decided their preferred sequence before the interviewer arrived. For example, one partner went first so that she could leave to fulfill a prearranged commitment to collect her daughter from a local activity center during her partner’s interview. Thirdly, we asked the interviewees to sit where they felt comfortable. Seating arrangements depended on the chairs available in the room chosen by the interviewees. In most cases, the couple sat on a sofa and the interviewer sat on a chair slightly to an angle facing them.
Part Two: A Multi-Level Design in Practice
In this second part, we reflect on how our adopted multi-level interview design worked in practice, specifically focusing on participant recruitment, disclosure, and the avoidance of harm.
Recruitment
We recognized that recruiting a purposive sample of couples who met our inclusion criteria and had both partners willing to share intimate aspects of their life was a potential challenge (Mellor et al., 2013). We were therefore mindful of how the interview mode may impact participation. Wilkinson (1998) suggests that interviews tend to be carried out on a one-to-one basis due to the logistical ease of arranging a time with one rather than two people. However, there is some evidence that joint interviews encourage male partners to participate (Braybrook et al., 2017; Seale et al., 2008). So, how did a multi-level interview mode impact recruitment?
Couples were recruited through study advertisements sent out via staff email newsletters of the biggest local employers (council, hospital, university) and social groups for ethnic and sexual minorities. A snowball technique was also adopted, whereby the research team approached people they knew to see if they knew other people who may be eligible. None of the recruited couples were known to the interviewer beforehand. The advert briefly described the aims of the research and inclusion criteria. A £10 shopping voucher was offered to each interviewee as a small token of appreciation for their time. When interested parties contacted the research team, eligibility was checked and an information sheet, which explained what participation involved and how their data would be used, was then provided.
The response was mixed. In line with Mellor et al.’s (2013) suggestion that women can act as relationship gatekeepers, two female enquirers declined to participate when it was confirmed that we were looking to interview both partners. They stated that while they were interested, their male partner would not be. However, from the eight opposite-sex couples interviewed, in three instances, the male partner initiated participation by asking their partner about taking part and contacting the research team.
As part of checking for informed consent, during introductions at the start of the interview couples were asked if they had taken part in an interview before, what they were expecting from the interview and their reasons for taking part. Guillemin and Gillam (2004) describe ethically sound research as studies wherein the participants take up the goals of the research. In line with this description, all the interviewees expressed a belief in the importance of developing a relationship toolkit for young people. As one interviewee put it “I wish the chance had been there when I was at school.” One couple thought being interviewed would be a new experience to try and another explained that they were taking part to increase visibility of same-sex relationships. The participating couples also shared a sense of pride in their relationship. The most common reasons given for declining to be interviewed by individuals to whom the research team spoke directly with, were that their relationship was “unconventional” or “not what you would be looking for.” This, and the findings from the pilot interviews, suggest that shared beliefs in the research goals and confidence in a perceived match with the criteria being looked for, rather than interview mode, may influence decisions to participate. As an in-depth study with a small sample, further research is needed to explore the impact of interview mode on participant recruitment.
Disclosure
Much of the existing methodological literature focuses on whether interviewing couples jointly or separately best facilitates interviewees to disclose, with the answer subject to debate (Heaphy & Einarsdottir, 2012; Taylor & de Vocht, 2011; Valentine, 1999; Zarhin, 2018). The establishment of rapport to enable disclosure; a warm, trusting but professional relationship between the interviewer and interviewee developed through showing interest, mirroring, and reassuring, is an often-discussed aspect of interviewing (Duncombe & Jessop, 2002; Hutchinson & Wilson, 1994). From a phenomenological perspective, rapport is important as it helps the interviewer and interviewee to sense how each are experiencing the interview (Wagner, 1970). For example, by picking up on nonverbal cues. In this section, we explore influences on rapport and disclosure within the multi-level interview design.
Frames of reference
Our experience suggests that dynamic differences between joint and individual interviews make it easier to develop rapport in individual interviews. With a two-way discussion (instead of three in the joint interviews, where two knew each other intimately), the interviewer was more able to use empathy within shared frames of reference to enhance communication (Schramm, 1974). For example, in one individual interview, an interviewee made a few negative remarks about her intellect. As Kelly points out, an unequal relationship between an interviewer and interviewee based on gender, race, class or education “may affect the willingness of respondents to disclose information” that could be perceived as discrediting (2010, p. 313). To address any perceived educational power difference and to support the interviewee’s confidence in the process, the interviewer took an opportunity when it arose, to explicitly empathize about the challenge of helping children with their homework when you have “no idea what it is about whatsoever.” This shared frame of reference induced laughter and shortly afterward, the interviewee disclosed past issues relating to alcoholism and controlling behavior within the relationship. As per Mavhandu-Mudzusi (2018), the process of building rapport appeared to support the interviewee to feel relaxed to discuss more sensitive aspects. The interviewee may well have made the same negative references to her intellect in front of her partner. However, in a joint interview it is arguably less likely this opportunity to relate empathically would have arisen, as the interviewer would need to take care not to displace the partner’s role in providing emotional support.
Alldred & Gillies suggest that “interaction in the research interview tends to elicit presentations of self which largely conform to dominant cultural forms” (2002, p. 146). The above discussion highlights the merit of shared frames of reference between interviewer and interviewee. However, the use of socially approved systems of typifications, which are often used when strangers meet to facilitate discussion (Wagner, 1970), can paradoxically block the production of descriptions beyond those which are superficial and conventional. For example, reading Extract 1 from an individual interview, one could think from the emphasis given to agreeing with the question and her reference to their cats as “the girls,” that the interviewee would describe her experience as having the “family feel” the interviewer suggests: It’s funny because we have the cats you see and we don’t have children, but the cats have become quite a focus of our lives and our time together and what we talk about and that sort of thing and sometimes I kind of stop and think, “gosh this is what people do with their kids” (laughs). But I say that because it’s kind of made a difference to what our home time is like because it’s not just her and me anymore, it’s us and the girls. So, we probably spend our home time a bit differently now as well. Yes, so there’s a family feel? Yeah very much so. Extract 1. Normative framing.
A phenomenological approach aims to stick as close as possible to the interviewees’ descriptions and draws attention to the importance of examining our bodily relationship to the world (Brinkmann, 2013; King & Horrocks, 2010). What this illustrative quote does not convey, is the interviewee’s physical reaction. While verbally agreeing to the notion of “a family feel,” she took a long pause, her facial expression contorted, and she shifted uncomfortably in her seat. Acutely aware that the interview was over-running in terms of time and based on a quick decision that to follow-up was not pertinent to the research questions, the interviewer did not ask the interviewee to elaborate and move beyond the heteronormative framing of “family” as involving the presence of dependents. This experience can be contrasted with a joint interview, as per Extract 2: So that was the first time, and did you know at that point there was a romantic spark between the two of you?
Yes probably.
Probably wouldn’t call it romantic (laughs)
(laughs) Yes, how do we choose the right word?
Physical attraction
Physical attraction would be better phrase
I am not very good on the romance thing
Extract 2. Cueing phenomenon.
The interaction between the partners in Extract 2 avoided the imposition of the interviewer’s framework of understanding and hegemonic norms regarding how long-term relationships begin (King & Horrocks, 2010). It shows how the distribution of power between a couple and an interviewer in a joint interview can encourage disclosure through the cueing phenomenon (Bjornholt & Farstad, 2014). This concept alludes to the situation within a group interview wherein participants who know each other can prompt each other to provide further details to their answers (Allan, 1980). In this way, rich data can be collected through corroboration, elaboration, or disagreement between partners (Hertz, 1995). As most of the time we act without conscious reflection (Cohen & Omery, 1994), and the object of inquiry may not be something individuals can express without time to reflect, or compare with another, a multi-level approach can help interviewees be more explicit.
However, it needs to be acknowledged that couples in this study presented as confident in their mutually caring relationships. The cueing phenomenon relies on couples having a balance of power which enables them to question the other in front of an interviewer. In line with the notion of couple desirability, partners may not speak openly and freely in front of their partner on all matters, adjusting their answers to partner expectations (Zarhin, 2018). Meeting with the partners individually after the joint interview provided an opportunity to disentangle dyadic and individual perspectives. For example, there was a lot of verbal and physical gestures (nodding) suggesting a shared perspective in relation to the rejection of “romance” in the joint interview as per Extract 2. Yet, while Interviewee 1 went on to strongly explain why she rejects the notion of romantic relationships in her individual interview, Interviewee 2 did not refer to this in his individual interview. Both examples highlight the importance of observational notetaking to record tacit knowledge such as sighs, murmurs, hunched shoulders, darting eyes. All of which may indicate contrasting or shifting perspectives.
Personal backgrounds
Our experience corresponds with other researchers who have suggested there can be a kaleidoscope of rich, complex, commonality and difference between interviewer and interviewee which come to the fore at different moments (Gabb & Singh, 2015; Song & Parker, 1995). Rubin and Rubin (2005) suggest that a certain amount of disclosure on the part of the interviewer is essential to help build empathy and so the interviewee feels less exposed. The joint session at the start of the multi-level interview facilitated this well. Often, interviewees asked questions, not about the interviewer’s research experience, but whether she was in a couple relationship or a parent. This reflected an understandable need for interviewees to feel that their personal experiences would be recognized (Harding, 2006). While this may have aided the building of rapport, it is likely that interviewees also used this limited information and their own assumptions about the interviewer as a gauge as to what was both expected and safe to disclose (Song & Parker, 1995). For example, one interviewee remarked sarcastically on the “need to use the term ‘partner’ now” [instead of husband or wife] in response to a question the interviewer posed which asked about “his partner.” The choice of relationship terms used in the introductions and pre-interview information may therefore have provided clues as to the interviewer’s positionality, although the interviewee’s use of sarcasm suggested a comfort with expressing his own positionality.
The clearest area of discrepancies between individual accounts in this study related to the couple’s sex life. For example, where one partner reported sexual fulfilment, the other reported it as a problem area of their relationship. Gender, as well as age symmetry between interviewer and interviewee may have influenced discussion in relation to sex and intimacy (Song & Parker, 1995). For example, despite the interviewer’s relative maturity in age (late 30s), one male interviewee responded reluctantly to a question as to how affection was shown with “well love, you’re very young.” Female interviewees were typically more forthcoming. While this may have reflected a symmetry between the interviewer and interviewee, cultural norms and individual backgrounds are also likely to influence disclosure. Most individuals reported that they tended not to speak to their partner about their sexual needs (let alone a stranger). A male interviewee who was a similar age to the interviewer responded openly to discussions relating to sex and intimacy. Earlier in his joint interview, he had described how he has “always been really kind of comfortable around girls and women. Like when I was a kid, I used to hang around with girls at primary school…I’m in touch with my sensitive side.”
Avoidance of Harm
Hertz (1995) proposed that individual interviews may be more suitable for sensitive topics which partners do not wish to discuss in front of each other. As found in our pilot interviews, separate interviews for couple partners provided an opportunity for self-reflection which may need to be expressed individually to avoid relationship disruption. For example, in a joint interview, both partners reported that there was nothing they would not discuss with each other. Yet, as per Extract 3, a different perspective was provided in one of the partner’s individual interview: There are a lot of times when I don’t take things into the relationship [quiet and flat voice]. Can you think of an example that you haven’t brought to the relationship or that you then may have gone somewhere else for that support? [after describing the issue]…And I think I probably haven’t raised it…because I can already feel my throat closing up. [Voice falters] I think it’s such a raw need that it feels too difficult to bring up without suddenly, sort of, you know, slopping a great big pile of steaming raw need down in the middle of the table…I am going to need to say…[about issue], which is going to be difficult (laughs) hmmm wow. You’re a good interviewer, you really drag some really painful things out (laughs).
No. no, no and this is very useful because it’s very rare that people have an opportunity to be this open and frank and self-reflective about really big issues…The more I talk to you about it, the more I am thinking that there are a lot of things that we don’t talk about in our relationship…I had always had a certain amount of smug pride that we spent so much time talking, how could our relationship possibly be anything other than perfect, but maybe we are not talking about the right things, or not talking about the right things in the right way or something along those lines. I think talking is very important, but I am now beginning to question the subject of the talks, yes. Extract 3. Going beyond the couple account.
As per Birch and Miller (2000), semi-structured interviews can have parallels with therapeutic encounters. Where rapport and confidentiality are established as in Extract 3, interviews can have potential transformative effects. The interviewee who disclosed alcoholism explicitly remarked at the end of her individual interview that it had felt “like therapy.” A phenomenological interview which seeks to take the hidden out of hiding (Cohen & Omery, 1994), emphasizes interviewee experiences and casts the researcher as listener, is closely aligned with Roger’s (1951) humanistic counseling approach. While the extracts discussed in this paper indicate a degree of acceptance and trust in the interviewer (Freeman et al., 2020), the potential to disturb an interviewee’s way of seeing things through close personal rapport can shake the security of the couple relationship. This means interviewers must judge the needs of the research against the risk of harm reactively. For example, following the discussion in Extract 3, the interviewer chose to pause the audio-recording to give the interviewee a break. During the break, the interviewer reassured the interviewee that the questions were not trying to catch him out, nor imply that there was a right or wrong way to feel about his relationship, or that he needed to take action. As per Mavhandu-Mudzusi (2018), this limited action addressed immediate emotions without changing the interview process into a counselling session.
There is a tension between seeking disclosure and respect for privacy and couple autonomy. The interviews suggested a gulf between a public image of emotional and sexual fulfilment through relationships and the lived up and down reality of coupledom. A challenge of different accounts from individual interviews is, therefore, how collected data will be presented. Hertz termed the keeping of differences in accounts as the “researcher’s dilemma” (1995, p. 441). As there is a possibility of loved ones recognizing interviewees, even if pseudonyms are adopted, publishing differences in partner’s accounts, or accounts which may discredit interviewees may cause relationship disruption (Taylor & de Vocht, 2011). Our epistemological grounding meant that we were not looking for, or expecting, to find a stable accurate representation of reality, but we did want to stay authentic to the essence of an interview (Brinkman, 2013). It can be difficult to judge sensitivity, as values will vary across audiences. For example, including the interviewee’s disclosure of alcoholism could infringe her rights to privacy, yet not including it may add to the perception of this as a taboo and be misleading to the picture of their relationship. Anonymity, as used in this paper, is possible where the research question does not relate to certain interviewee characteristics. However, in reports of the study’s findings, where quotes identified sex and relationship type, we did not present any data we felt could cause harm.
As hypothesized at design stage, several interviewees took the opportunity in the final joint session to share something they had mentioned in their individual interview to see what their partner thought about it. The end joint session reinforced the ethical importance of a debrief for both interviewees and interviewer. For example, an interviewee expressed gratitude for this extra time, as he wanted to clarify a view he felt he had not expressed that well during his individual interview. Seeing the couple come together as they had begun the interview also helped the interviewer leave feeling, even if at times an interview had been emotionally challenging, overwhelmingly privileged to have heard their stories.
Conclusion—Defining a Successful Interview
Undertaking this reflection raised questions as to our assumptions of what we considered a successful interview. Gabb and Fink (2015) describe “intimately revealing” moments in their data as key results. Our choice of examples for this reflection suggests we initially felt similarly as they each describe a moment in the interview process where a level of depth or understanding was reached which was not revealed at first. As many of these revelations occurred in the individual interviews, such a perspective could be seen to advocate for this mode. The confidential nature of the one-to-one setting and greater capacity for building rapport may well increase the likelihood of hearing something deeply personal. However, dominant individualistic perspectives which emphasize an autonomous self may lead to assumptions as to the value of data provided in private (Wilkinson, 1998). For example, Morris (1990) found working-class couples were more likely to air disagreements in joint interviews and censor what they said in separate interviews out of loyalty to their partner. The notion of intimate revelations constituting insightful data fails to recognize a) the role of the interviewer and the specific context of the interview in the construction of data and b) the layers of data which together provide a richer account of a complex object of inquiry.
Narrators select, organize, and interpret events, to be meaningful for an audience (Riessman, 2008). As discussed within the paper, personal backgrounds and information shared about yourself as the interviewer, whether explicit or not, can both encourage and prohibit disclosure. It can be challenging to move beyond socially accepted typifications and ideas of how one “ought” to feel. As per Birch and Miller (2000), interviewees described the interview process as offering a space to consider and emotionally process meanings of their experiences which they had not consciously done before. The therapeutic overtones may mean, as per Extract 3, that interviewees can get swept along with the process. The sense that disclosure is the key “goal” of an interview can make it difficult to judge the ethical boundaries of family accounts and the extent to which to probe.
Our multi-level interview mode attempted to balance an objective for disclosure with interviewee well-being by making participation part of the couple’s relationship narrative (something they did together) and by providing ways to empower the interviewees in the interview process. As per the illustrative extract in Extract 4, many interviewees commented in the end joint interview on how they enjoyed the interview experience:
I feel it is a very positive experience.
Yeah and it made you verbalize and address questions that I have not sort of especially thought about. As you’ve heard, some of them were difficult to answer but yeah, it’s been good.
Extract 4. Positive endings
As per Heaphy and Einarsdottir (2012), we found individual interviews did not simply confirm the joint interview that preceded them. Although assumptions were made as to what was a shared or individual experience, which need to be further explored, the careful separation of the topic guides for the individual and joint interviews appeared to work well and the interview mode did not appear to affect participation. By looking at individual and dyadic levels, this multi-level approach incorporated a form of triangulation producing two individual scripts and one couple script to gain a richer understanding of the phenomenon being explored (Tracy, 2000). The inclusion of joint interviews provided an opportunity to observe how partners interact and support one another, move beyond normative framings via the cueing phenomenon and gain an understanding of what couples reveal to each other and what they present externally. In this instance, we found this triangulation trumped consistency of mode of interview to collect comparable rich data for the wider project.
Limitations and Future Research Priorities
A challenge of a multi-level interview mode is the length of interview. It requires focused attention for interviewees and particularly the interviewer over a prolonged period, making it challenging to pick up on all behavioral prompts. Each individual script, couple script, interactions between interviewees, as well as interviewer and interviewees also require time-intensive transcription and multiple readings which may limit its uses. Otherwise, the challenges which arose were not specific to a multi-level design but reflected familiar concerns with semi-structured interviews regardless of mode: preservation of anonymity in presentation of sensitive data, characteristic match between interviewer and interviewee, recording of tacit knowledge, moving beyond normative expression and balancing privacy with prompts for further detail. Given the finding of common challenges across interview modes, in Figure 2, we offer a list of guiding reflective questions for researchers which are likely to have wider relevance for studies involving semi-structured interviews.
Guiding reflective questions for semi-structured interviews.
Our in-depth cross-sectional study utilized a small sample of couples who all described their relationship as satisfying, with both partners confident and motivated to take part. Further concerns with a multi-level design which may arise with other samples cannot be ruled out. This study was also focused on attributes of the couple relationship, so studies with a different focus may have a different experience with a multi-level design. More research is needed to understand the strengths and limitations of a multi-level interview mode. Word-count for qualitative findings papers offer little room for detailed description of method, with space often prioritized to present the voices of the interviewees. However, future studies should report on how the interview mode affects participation and interviewees’ experiences of different interview modes. Developing a greater understanding of what it means to be interviewed may encourage wider participation in studies and go beyond the researcher’s interpretation of revelatory moments to a greater theoretical understanding as to what makes for a “successful” interview.
We hope that the description and reflection advanced in this paper will encourage other researchers to move beyond binary interview designs and try a multi-level interview mode. When interested in relationships such as parent/child, siblings, employer/employee, where both shared and individual perspectives exist, a multi-level interview mode moves beyond cultural norms of personhood to recognize the complexity of data.
Footnotes
Authors' Note
Acknowledgments
Declaration of Conflicting Interests
Funding
ORCID iD
