Interactional problems in research interviews

Theoretical perspectives

Among a variety of analytic methods to examine how interview talk is produced by speakers are approaches that draw on ethnomethodology (Garfinkel, 1967; Ten Have, 2004). Although their application to research interviews is not without controversy (Nikander, 2012), these include conversation analysis (CA; Sacks, 1995; Schegloff, 2007) and membership categorization analysis (MCA; Housley and Fitzgerald, 2002; Jayyusi, 1984; Stokoe, 2012). CA examines the basic building blocks of talk-in-interaction used by members in everyday interaction, including turn-taking, sequential organization and preference structure (Hutchby and Wooffitt, 2008; Ten Have, 2007), while MCA is an approach to the analysis of talk and texts that examine how reasoning is demonstrated in acts of description. A growing body of work has examined interview data using these approaches (Author, reference omitted) (Baker, 1997; Bartesaghi and Bowen, 2009; Hester and Francis, 1994; Housley and Fitzgerald, 2009; Housley and Smith, 2011; Roulston, 2006; Schubert et al., 2009).

Baker (2002) has described five ways to examine interview data via an ethnomethodological lens. These are (1) to examine the interaction as a conversational sequence using the tools of CA, (2) treat data as accounts in which members of groups speak as members of specific populations to which they have been assigned by the researcher, (3) examine the membership categorization work used by speakers to do descriptions, (4) investigate the identities produced in talk, and (5) examine the ‘versions of worlds’ talked about by speakers (Baker, 2002: 778). In Baker’s view (2002: 792–793), these approaches assist in examining the ‘culture in action’ in a particular social setting—or

the practical and moral reasoning, not only about the topic of the interview, but about what a competent or sensible interview account of events, people, courses, of action, and so on could sound like (italics in original).

This approach to analyzing interview data exposes the moral assumptions of both interviewer and interviewees as they co-construct questions and answers as research data.

Research data

Sequences analyzed in this article are drawn from a qualitative evaluation of a federally funded training program at a family medicine residency program in the United States conducted from 2007 to 2010. The aim of the evaluation, which was approved by Institutional Review Boards at both the residency and university associated with the project, was to examine residents’ perspectives of a training program on Mind Body Medicine and Spirituality (MBM/S) interventions for physicians preparing for careers in family care. Although the study included field notes of observations, surveys, and documentary data, this article focuses on sequences from a corpus of individual interviews (n = 77) with 57 people. Interviewees included 47 residents and 10 faculty members. All but three interviews were conducted face-to-face by the author, ¹ and all were audio-recorded. Individual semistructured interviews averaged 40 minutes in length and were transcribed by several research assistants, a professional transcriber, and the author.

The author began by locating instances in which interviewees either failed to respond to questions, or did not provide the sorts of detailed descriptions provided by other participants. Superficially, it seemed that this was sometimes a problem of achieving mutual understanding, or, on occasion, moments in which it seemed that participants were demonstrating resistance to the intent of the interviewer’s task of generating interviewees’ descriptions of their opinions and beliefs. After retranscribing these sequences using Jeffersonian transcription conventions (Psathas and Anderson, 1990), and analyzing them further, the author identified strategies used by interviewees that served as obstacles to the researcher’s accomplishment of generating ‘rich descriptions’. These included (1) asking questions of the interviewer (e.g. ‘what do you mean by that?’) and (2) providing minimal responses to questions. In some instances, these strategies worked together to function as refusals to expand on topics initiated by the interviewer. In other instances, participants provided in-depth descriptions at later points in the interview that elaborated on questions posed earlier. The purpose of the following analyses is to explore how findings from examinations of these sorts of sequences might complement understandings of topical findings, and inform interview practice and research design. In the next section, Baker’s (2002) guidelines for ethnomethodological analysis of interviews are used to examine sequences that exemplify these strategies. Excerpts frequently illustrate both strategies, and thus, they are not treated as separate phenomena.

A semistructured interview guide was used for all interviews, and covered a number of topics, including residents’ beliefs and perspectives concerning the incorporation of spirituality in patient case, whether they had applied any of the modalities included in the training program in their clinical practice, and physicians’ perspectives concerning how these modalities might be applied in clinical practice with ‘underserved’ populations.

Pursuing the interviewer

Excerpt 1 below exemplifies a problematic interaction in which the interviewee pursues the interviewer for further information. Prior to this excerpt, ² R1 ³ described her discomfort with how spirituality might be applied in clinical practice in response to the interviewer’s question, ‘what are your beliefs about the role of using mind body approaches in working with underserved and minorities?’ In her expressed beliefs, R1 argued that she could not include treatment options that she saw as informed by non-Christian traditions in patient care if they were to ‘potentially lead someone in a different path’ (e.g. referring patients to a free yoga class funded by the project). When accounting for her ‘beliefs’ in relation to spirituality, R1 likewise distanced herself from the description by positioning the viewpoint as hypothetically belonging to ‘somebody like ↑me’ and repeatedly using the subordinating conjunction ‘if’.

Soon after, at the prompting of the interviewer (‘you mentioned that you’ve used ↑breathing with some of your patients? could you give me describe what a specific example of that?’), R1 provided a story illustrating how she had applied her knowledge of diaphragmatic breathing—a modality included in the training program—in patient care (not shown here). R1’s story depicted her encouragement of a patient who she knew to be a Christian through advising her to meditate on some Biblical scriptures to alleviate anxiety in stressful situations. Immediately following this story of R1’s application of a modality incorporated in the training program, the interviewer posed a follow-up question seeking further examples of application in relation to a specific group served by the residency and identified by the project faculty as ‘underserved’ (Excerpt 1).

As noted earlier, R1 had already provided a detailed account of how she viewed the use of MBM/S modalities with ‘underserved’ populations. This explanation functioned as a rationale for why ‘somebody like ↑me’ might not use modalities viewed as contrary to his or her faith. In this earlier sequence, R1 explicitly referenced the need to be ‘understood’ by the interviewer. The interviewer formulated R1’s talk and follow-up questions similarly (e.g. ‘so if I understand you correctly’, ‘I just wanted to make sure that I’ve understood you correctly’; sequence not shown here). These repeated references by interlocutors to ‘understanding’ one another indicate the intensive attention paid by both parties to accomplishing intersubjective understanding.

In response to the interviewer’s follow-up question subsequent to her story exemplifying applications, at lines 110, 112, and 114 (see arrowed lines in Excerpt 1), R1 questioned the interviewer about the fundamental assumptions of the training program. In this solicitation of information from the interviewer in her role as an external evaluator of the program implementation, this participant’s utterances function as a fundamental critique of the very assumptions upon which the project was funded.

Excerpt 1: Third-year resident, June 2008 [34:45-37:27] ⁴

To review, the question posed by the interviewer at lines 98–99 in Excerpt 1 occurred a few minutes after a discussion, in which R1 had provided her reasoning for why she believed MBM/S approaches were not appropriate to use with underserved patients. Here, the interviewer’s follow-up question circled back to the topic of ‘underserved’ patients already discussed in earlier talk. Given that R1 had already responded to a question about ‘underserved’ patients, here, the interviewer may be seen to pursue R1 for an elaborated account. This also implies that the earlier account has been taken as insufficient by the interviewer. After hesitating, R1 sought clarification in how to respond to this question by asking whether she could produce an account that does not refer to a clinical setting. This question is formulated as a closed question (‘it doesn’t have to be in a clinical setting’, line 101) with a preference for an agreement, which is supplied with no delay by the interviewer (‘↑no I don’t think so ↑no’, line 102). Lines 101–102 function as an insertion sequence between an adjacency pair (Q-A). In response to the first of IR’s questions (lines 98–99), R1 asserted that she has had ‘opportunities to encourage people’, before coming to a halt, which is marked by a 3-second pause (‘yeah but I’ve but out even outside of a clinical setting I’ve had (.) opportunities to um (.) encourage people’). The utterances at lines 103 and 105 provide no specific detail concerning how she has used MBM/S with ‘underserved’ populations outside clinical settings and begin with repetition of ‘but’, a preface to a disagreeing response that is implied, but not articulated. What is interesting about this account is that it is produced as an insertion sequence with multiple delay markers (pauses, ‘um’). The interviewer’s question is not formulated to prefer either an affirmative or a negative response but could have been answered in either way (yes ± account), or (no ± account). R1 has responded to the question by providing an affirmative response with a truncated account marked by the delayed characteristic of ‘dispreferred’ responses (Liddicoat, 2007: 111).

This account aligns with the version of affairs that R1 had presented earlier in the interview, in which she had argued that ‘somebody like ↑me’ would not apply some of the MBM/S modalities advocated in the training program with underserved populations because they could possibly lead patients away from Christianity. In Excerpt 1, the interviewer initially orients to R1’s utterances (lines 105, 107) unproblematically to signal that she has understood R1 to have concluded her answer by providing a preclosing comment as she examines the interview guide for the next question (‘let me see I think we’re getting towards the end here’, lines 108–109). At this point, R1 initiates a new topic of talk by posing a question to the interviewer at line 110 (see arrowed lines, Excerpt 1). The interviewer’s initial response at lines 111 and 113 is found insufficient, and at lines 114–115, R1 repeats and develops her question, stressing the word ‘why’: ‘why is mind body medicine focusing on underserved people’.

Throughout the interview as a whole, the interviewer’s questions have cast the interviewee in the omni-relevant category of ‘physician working in a clinical setting’. Sacks (1995) introduced the idea of ‘omni-relevance’ (vol. 1, pp. 594–596)—in which speakers use devices ‘relevant to a setting via the fact that there are some activities that are known to get done in that setting’ (p. 313). In Excerpt 1, the interviewee selects to speak from an alternative category—that of ‘community member who encourages others’, having earlier used the category of ‘faithful Christian’ (lines 25–26 ‘= now somebody like ↑me (.) hhhh. if I believe that (.) there’s an absolute truth and there’s a true heaven and a ↑hell=’). These interactions show that although an omni-relevant category ‘may stand as background to be invoked at any time’ (Fitzgerald and Housley, 2002: 583), in research interviews, categories can shift and be overlaid by other categories as interactions proceed.

In the ensuing questions posed by the interviewee to the interviewer, there is explicit rejection of the link between ‘use of MBM/S’ and ‘underserved populations’ that is assumed in the interviewer’s opening question of Excerpt 1. This position reiterates ideas concerning the role of MBM/S in treating underserved and minority patients that R1 had articulated earlier. R1 pursues the interviewer for an account of why there is a link between underserved populations and MBM/S, and rejects the interviewer’s initial response ‘that’s actually one of the um objectives for which they got um’ as self-evident (‘I know that’). The interviewer, meanwhile, distances herself from the program implementers by referring to them as ‘they’, and suggests that R1 seek further information from one of the faculty members involved in program implementation (lines 125–126). In doing so, she rejects the category of ‘program implementer’ that R1 has assigned to her, taking on a distanced category position of ‘external evaluator of the program’, and by implication, not responsible for programmatic content and implementation. The category of ‘program implementer’ to which R1 has assigned the interviewer had also been indicated earlier when R1 provided the story of her application of breathing in patient treatment (‘it’s probably so off of like wha(.)t yo(.)u all taught us’, lines 63–64, not shown here).

In the version of affairs produced in this sequence, R1 questioned the objectives of the training grant funded by a federal agency whose primary objective is to improve ‘access to health care services for people who are uninsured, isolated or medically vulnerable’ (‘About HRSA’, n.d.). This interaction provides a good deal of insight concerning this resident’s perspectives and opinions of the MBM/S training program. First, how R1 poses questions to the interviewer indicates dissatisfaction with the follow-up question posed (lines 98–99) and the repeated reference to the topic of ‘underserved and minority patients’ by the interviewer—a topic already discussed in detail. Embedded throughout R1’s account is a meta-commentary indicating that she wanted to make her points clearly, and to be ‘understood’ by the interviewer. By posing a follow-up question on the topic of ‘underserved and minority populations’, the interviewer is exposed as ‘not having understood’. R1’s three repeats of the question ‘why is mind body medicine focusing on the underserved’ function to question the assumption that there is any relevance of ‘MBM/S’ for treating ‘underserved patients’. R1’s utterances imply critique of the assumption that these modalities could or even should be applicable to underserved populations—a view for which she has crafted a detailed rationale that indicates sensitivity to the possibility that she may be misunderstood. In this complex interaction, both interviewer and interviewee background omni-relevant categories of ‘physician in a clinical setting’ and ‘program implementer’, selecting to take up alternative category positions, including ‘community member who encourages others’ and ‘external program evaluator’.

Pursuing the interviewee

In Excerpt 2, R2 responds to the interviewer’s initial open-ended question ‘what do you think about when you hear the term spirituality in relation to patient care’ with a delay marker (↑u:::m) followed by a 2-second pause. In contrast to detailed descriptions produced by other participants in this study, rather than providing an in-depth account in response to this question, R2 answers by noting that ‘I don’t have a good answer for that I I it’s it’s patient dependent’. In the sequence that follows, the interviewer pursues the topic through a category search—that is, by using alternative categories via which R2 might produce an elaborated answer.

Excerpt 2: First-year resident: October 2009 [3:20-3:52]

In posing the opening question, the interviewer has assigned R2 to the omni-relevant category of a ‘physician’ for whom ‘spirituality in patient care’ is a relevant topic. In this sequence, the interviewee is assigned to the category of physician who might ‘think’ about the topic of spirituality and patient care (lines 1–2), who might discuss this as a member of a residency cohort (line 9), or notice this in clinical practice with patients (lines 11–12). Initially, R2 carefully avoids providing a description about his ‘thinking’ on the topic by stating that he does not have a ‘good answer’. This formulation avoids any imputation of an ‘unthinking’ response, by noting that he does not have a ‘good’ answer since the topic is ‘patient dependent’. In line 6, R2 locates the issue of spirituality as something of relevance to ‘patients’ (‘I I it’s it’s patient dependent’) but does not elaborate on what form that might take. Implied in these utterances is the idea that ‘a good answer’ could only be provided with reference to patient-specific detail. Interactional trouble is signaled through delay markers (lines 4, 6, and 10) and multiple pauses in responses.

Excerpt 2 shows the interviewer engaged in ‘finding a topic-relevant category’ from which the interviewee might answer. ⁶ However, neither ‘physician in clinical practice’ nor ‘family care physician-in-training’ is taken up as a relevant category by the interviewee, who steadfastly declines to elaborate further. Despite pursuit by the interviewer, through providing polite answers to the effect that he has ‘nothing of value to report’, R2 does not provide descriptions that speak to the researcher’s interest in the topic of spirituality as relevant to the categories of ‘physician’ or ‘family care resident’. This account, then, may be taken as indicative of one possible version of medical training for family care physicians, in that spirituality is not found to be relevant.

Omitting specifics

Excerpt 3 is drawn from an interview in which a resident had repeatedly mentioned the issue of ‘resistance’ among other residents toward the implementation of the MBM/S training program, yet had omitted specific details. Throughout two interviews, R3 expressed himself as knowledgeable about the reasons for others’ resistance to the training program. In Excerpt 3, R3 returned to the theme of resistance that he had focused upon some 13 minutes earlier (that of ‘resistance’ to ‘faculty’ involved in the training program).

Excerpt 3: First-year resident: May 2009 [30:33-32:23]

This sequence begins with a closing question on the interview guide, posed as ‘↑how would you assess the impact of the mind body training program’. The account from R3 is marked by lengthy pauses and minimal detail. The interviewer formulates the talk several times (see lines 10, 14, 22, and 24). R3 provides agreement to the first of these formulations (‘and that’s amongst thee uh residents?’ line 10) but interrupts the interviewer at the second of these (line 14) with further information: ‘it’s not for lack of trying on the part of the mind body educators’. At lines 19–21, R3 elaborates on the information that has been provided both earlier in the interview ⁷ and earlier in this sequence (line 6). This utterance is marked by lengthy pauses, indicating the speaker’s trouble in formulating a sensitive topic that must be presented with great care:

but it’s the resistance (.) I don’t think just to the mind body but to: (5.0) thee (2.0) overall (2.0) program or the (4.0) resistance to the training or the the persons doing the training.

The interviewer orients to this utterance as meaning that residents are resistant first to the ‘idea’ of mind body medicine as part of medicine. R3 begins an utterance before the interviewer completes the second component of her formulation (indicated by ‘both’, line 22). R3 disagrees with the interviewer’s formulation, and indicates that the ‘idea’ of MBM/S in medicine is only a ‘small’ part of the resistance. He then continues the account by asserting, ‘but I think a bigger part of the resistance is to (.) u::m (1.0) those doing the training’. The interviewer reformulates this as ‘faculty’, and R3 agrees (line 32).

That this is both a serious and sensitive complaint is oriented to by the interviewer, who accepts R3’s utterance with delays ending in silence: ‘OK yeah (.) u:::::m (2.0)’. She then poses a follow-up question, which does not seek further detail concerning R3’s assertion (e.g. which faculty members are referred to, or what it is about the faculty members that elicits resistance among residents) but seeks a solution to the problem: ‘do you see any like (.) ways that ways to (.) u::m change that?’ After a sigh followed by a pause of 3 seconds, R3 orients to the recording device: ‘not on tape’. At this point, both speakers dissolve into loud laughter, to which the interviewer responds with ‘oh that’s tough’ and suggests a solution to the problem (‘I guess you can get a whole new bunch of faculty’) that is oriented to as a joke by R3, who provides no further comment. The interviewer’s comment here is ‘leading’ in that it invites some form of response (agreement/disagreement) from R3; however, this invitation is not taken up, and a new topic ensues. In Excerpt 3, R3 has framed a strong complaint about faculty of the program, while demonstrating sensitivity to the implications of ‘going on record’. By refusing to elaborate further ‘on tape’, R3 demonstrates how speakers are accountable for what they say ‘on the record’, especially utterances that are ‘delicate [or] inappropriate’ (Speer and Hutchby, 2003: 318–319). Both interviewer and interviewee orient to the talk as highly sensitive, through lengthy pauses and delay markers, and finally, shared laughter.

In this sequence, R3 asserts himself as knowledgeable about the problems with program implementation and the reasons why residents have expressed resistance to the program. However, few details are provided, and the interviewer elects not to probe further. The interaction is oriented to by both speakers as highly sensitive and is marked throughout by extended delays, and finally laughter, which conversation analysts have found to be associated with troubles-telling (e.g. Jefferson, 1984). In the version of the world portrayed in this interaction, the residents are resistant to the faculty who implement the training program, and that is the reasoning provided by R3 for why he has categorized the impact of the training program as ‘mediocre’. What is interesting about this sequence of talk is what is not said. No information is provided by R3 concerning why the faculty are deemed to be problematic, which faculty members are being referred to, or what might be done to alleviate the situation. Nor does R3 provide details concerning what residents do to resist the program. More interesting still is the interviewer’s response to both the information provided and the lack of detail—apart from formulating her understanding of R3’s talk and seeking advice on what might be done to deal with the problems, she does not probe further.

In this version of the world, any possible ‘solution’ to the problem is not available ‘for the record’. This interpretation is supported by R3’s failure to assess one possible solution (that of getting new faculty), provided by the interviewer. Although the interview context was confidential, R3 demonstrates recognition that any information provided will be used for official reporting purposes, and therefore may have potential implications beyond the interview setting, although these, too, are unstated. Because in this sequence the interviewer asks for few specific details from R3, she cooperates with R3 in the omission of detail. Although the interviewer assesses information provided by this participant about ‘resistance’ as ‘helpful’ (sequence not shown here), ironically, specific information is lacking. It is possible that if asked for an account off the record, R3 may have elaborated. Regrettably, this did not occur, suggesting that gaining deep understanding of settings for the purpose of answering research and/or evaluation questions requires considerable fieldwork in which researchers learn about participants’ perspectives in both ‘on-record’ and ‘off-record’ settings.

Discussion

What is it that participants tell researchers about research topics via problematic interactions? How do speakers manage problematic interactions in research interviews? In this article, examples of problematic interactions have been provided from interviews with three participants. In these sequences, interviewees challenged the interviewer about research purposes, provided minimal answers to questions, or refused to elaborate on a topic for the record.

First, when interactional difficulties occur in research interviews, data may be analyzed with a view to learning more about (1) the category memberships employed and assumed of one another by participants and interviewers, (2) sensitive topics that might be explored further—perhaps using research methods other than formal interviews, and (3) possible ways in which researchers’ and participants’ assumptions about project aims and purposes do not align (see also Roulston, 2011). For example, in this study, the topic of underserved populations emerged as a sensitive topic in multiple interviews, and as a result of ongoing analyses of the sort demonstrated here, this topic was explored in later rounds of interviews to gain more insight into residents’ perspectives.

Second, sequences analyzed here demonstrate how keenly interviewers and interviewees monitor one another’s talk—continually orienting to what came before in efforts to understand one another, and demonstrating turn-by-turn orientation to the categories employed and the sequential work of asking and answering questions for the purposes of doing research interviews. Fitzgerald and Housley (2002) have explored talk in radio phone-in programs, arguing that categorial ‘layers’ are used as a resource that are ‘inextricably entwined’ with sequential organization to organize the flow of talk (p. 580). They argue that ‘the local environment of interaction may usefully be approached as one involving layers of context, with these layers embedded within the flow of interaction’ (p. 585).

Similarly, in the sequences from research interviews analyzed here, the flow of talk is organized sequentially through question–answer sequences that are laminated with layers of complex and shifting category work on the part of both speakers. In managing interactional difficulties as they arise, interviewers—whose default position as ‘researcher generating data’ assumes primary responsibility for navigating delicate topics—may venture forth on topic-generated category searches in order to encourage interviewees to elaborate on answers deemed to be insufficient. For example, although in any research interview, researchers may assume an omni-relevant category from which interviewees might respond to questions (e.g. ‘physician who may use MBM/S modalities in clinical practice’), interviewees may draw on other category memberships from which to speak (e.g. Excerpt 1, ‘devout Christian and community member’), or reject invitations entirely (e.g. Excerpt 2). Furthermore, in order to further the flow of talk when interaction difficulties are encountered, interviewers themselves may reject an assumed category membership (e.g. Excerpt 1, ‘program implementer’) in favor of another categorial incumbency (e.g. ‘external evaluator’).

Methodologically, the generation of research data via interviews is a collaborative endeavor in which interviewees may refuse to participate topically, even while orienting to the sequential organization of the interview by providing answers to questions. This can be accomplished via rejection of omni-relevant categories invoked in an interview, by asking questions of the interviewer, or by declining to speak ‘on record’. Finally, participants also attend to the moral implications of their descriptive practices and how these may be heard by others–—both in the immediate interview context, and by audiences beyond the interview ‘for the record’.

Conclusion

In instances in which research participants decline to elaborate in response to interviewers’ questions, or question interviewers, topical analysis may be problematic in that the substance of talk—the answers provided—will not be helpful as data if, as methodological literature advises, only ‘rich descriptions’ are sought. Often, these kinds of interviews are categorized as ‘failed’ interviews, and anecdotal evidence suggests that sometimes researchers discard these kinds of interviews on the basis that they are topically ‘thin’ (e.g. Yanos and Hopper, 2008). Vitus (2008) has theorized these kinds of difficult interactions that produce ‘uninformative’ data using an ‘agonistic’ approach. This article uses an ethnomethodological approach proposed by Baker (2002) to analyze interview data to explore problematic interactions. I argue that this form of analysis is one way to gain insight with respect to the identities employed by speakers in doing research interviews, and the moral assumptions concerning research topics implicit in both questions and answers provided by interviewers and research participants.

Examination of problematic interactions illuminates issues that participants make relevant about research topics that were, perhaps, initially extraneous to the researcher’s agenda. These kinds of analyses also reveal gaps in the interviewer’s questioning—those points at which different questions might have been asked, and that could be followed up in further rounds of data generation. I argue that such analysis is productive for thinking methodologically about both research projects and the interview as a method for generating data. Secondary analyses of problematic interactions demonstrate that rather than discarding interview data as ‘failed’ or ‘uninformative’, researchers can productively forward their knowledge of the ways in which versions of the world are talked into being in research interviews. These kinds of examinations contribute to what Rapley (2012: 553) calls the ‘social studies of interview studies’ by recovering the ‘extraordinary’ work of ‘mundane, situated, practical action and reasoning’ entailed in the task of apparently simple, yet enormously complex work of asking and answering questions. Finally, this type of analysis attends to the details of how interviewees and interviewers manage interactions while navigating the varied roles and relationships represented by the spectrum of ‘strangers’ to ‘lovers’ (Cicourel, 1964: 76).