Researching the researcher-as-instrument: an exercise in interviewer self-reflexivity

Abstract

Because the researcher is the instrument in semistructured or unstructured qualitative interviews, unique researcher characteristics have the potential to influence the collection of empirical materials. This concept, although widely acknowledged, has garnered little systematic investigation. This article discusses the interviewer characteristics of three different interviewers who are part of a qualitative research team. The researcher/interviewers – and authors of this article – reflect on their own and each other’s interviews and explore the ways in which individual interview practices create unique conversational spaces. The results suggest that certain interviewer characteristics may be more effective than others in eliciting detailed narratives from respondents depending on the perceived sensitivity of the topic, but that variation in interviewer characteristics may benefit rather than detract from the goals of team-based qualitative inquiry. The authors call for the inclusion of enhanced self-reflexivity in interviewer training and development activities and argue against standardization of interviewer practices in qualitative research teams.

Introduction

Inner Silence

Writing, Reflecting, Hoping

Slipping into Truth

Interviewing moments

Take me by surprise

Like Sunlight

(Janesick, 1998: 53)

The level of researcher involvement in qualitative interviewing – indeed, the embodiment of the unique researcher as the instrument for qualitative data collection – has been widely acknowledged (e.g. Cassell, 2005; Rubin and Rubin, 2005; Turato, 2005). Because the researcher is the instrument in semistructured or unstructured qualitative interviews, unique researcher attributes have the potential to influence the collection of empirical materials. Although it is common for scholars to advocate for interviewer reflexivity (Ellis and Berger, 2003; Pillow, 2003) and acknowledge the researcher as the primary instrument in qualitative interview studies (Guba and Lincoln 1981; Merriam 2002), with some notable exceptions (e.g. Pitts and Miller-Day, 2007; Watts, 2008) few have actually examined the qualitative interview as a collaborative enterprise, as an exchange between two parties, reflecting on the ways in which the interviewer affects the organization of this talk-in-interaction and the processes by which the talk is produced. Given this, the first aim of this study is to provide a reflexive account of how three different interviewers (authors Jonathan, Annie, and Michelle) individually facilitate unique conversational spaces in their qualitative interviews.

Understanding the qualitative interview as social interaction is important for any sole qualitative researcher, but as Miller-Day et al. (2009) pointed out, this may be particularly germane for qualitative research teams (QRT). Herriott and Firestone (1983) argued that when there is more than one interviewer on a QRT, inconsistencies in interview style and approach may affect the quality of the research conversation and ultimately the study findings. Indeed, several published resources on QRTs suggest that interviewers should receive the same standard training with an eye toward producing consistent strategies and credible findings (Bergman and Coxon, 2005; United States Agency for International Development’s Center for Development Information and Evaluation, 1996). Unfortunately, current literature addressing QRTs has primarily focused on the relationship dynamics among research team members (e.g. Fernald and Duclos, 2005; Rogers-Dillon, 2005; Sanders and Cuneo, 2010; Treloar and Graham, 2003) and on group analytical procedures (e.g. Guest and MacQueen, 2007; MacQueen et al., 1999; Olesen et al., 1994) rather than on the team member roles (e.g. interviewer, analyst) or data collection practices (e.g. strategies for building rapport). As QRTs are becoming more prevalent, especially in funded research (Barry et al., 1999; Ferguson et al., 2009), there is a need for more information about how to maximize the use of multiple interviewers and maintain a focus on the unified research goals while respecting the flexibility of the in-depth qualitative interview as talk-in-interaction (Mallozzi, 2009; Miller-Day et al., 2009). Toward that end, the second aim of this study is to reflect on and discuss implications of the study findings for qualitative research teams.

Researcher-as-instrument

The phrase researcher-as-instrument refers to the researcher as an active respondent in the research process (Hammersley and Atkinson, 1995). Researchers ‘use their sensory organs to grasp the study objects, mirroring them in their consciousness, where they then are converted into phenomenological representations to be interpreted’ (Turato, 2005: 510). It is through the researcher’s facilitative interaction that a conversational space is created – that is, an arena where respondents feel safe to share stories on their experiences and life worlds (Owens, 2006).

Across the years, scholars have considered the nature of researcher-as-instrument as interpreter of empirical materials and as involved in the construction of ideas (Janesick, 2001; Singer et al., 1983). This consideration began to grow after feminist UK scholars such as Oakley (1981) and Graham (1983) criticized quantitative-based research methods that assumed a detached and value-free researcher in the acquisition and interpretation of gathered data, and was further developed by feminist ethnographers such as Stack (1995), who offered seminal research on ‘dramatizing both writer and subject’ in fieldwork on neighborhoods and communities (p. 1). More recently, scholars have extended their interest of researcher-instruments to consider specific interviewing strategies. Conversation analysis tools have often been used to examine the intricacies of interview conversations, studying the ways in which the ‘how’ of a given interview shapes the ‘what’ that is produced (Holstein and Gubrium, 1995; Pillow, 2003).

While qualitative scholars agree that a conversational space must be created, they often disagree as to what that space should look like. Some scholars argue for a Rogerian interviewing space, where empathy, transparency, and unconditional positive regard are felt (Janesick, 2001; Mallozzi, 2009; Matteson and Lincoln, 2009). Pitts and Miller-Day (2007) documented specific trajectories experienced by qualitative interviewers when establishing rapport with research participants, and the authors argue that a feeling of interpersonal connection was necessary for the qualitative interviewer and interviewee to develop a partnership. These claims are grounded in the feminist or postructuralist perspective, which hold that ‘the essential self … is not automatically revealed in a neutral environment but can and might need to be benevolently coaxed out into a safe environment, where it can be actualized’ (Mallozzi, 2009: 1045).

Others advocate against a feminist approach to interviewing. Tanggaard (2007), for example, viewed empathy to be a dangerous interviewer quality because it tends to create a superficial form of friendship between interviewer and respondent. Self-disclosure has been similarly critiqued (Abell et al., 2006). These critics hold that self-disclosure may actually distance the interviewer from the respondent when the self-disclosure portrays the interviewer as more knowledgeable than the respondent. These studies question the popular assumption that displays of empathy or acts of self-disclosure are naturally interpreted by the respondent as a means of establishing a conversational space of rapport and mutual understanding.

So where do these opposing viewpoints lead us as researchers? For the three of us who are authoring this article, the answer to that question is an unsatisfactory, ‘we are not sure.’ Working as part of a QRT, we were trained in a systematic manner, provided with clear procedures for carrying out our qualitative interviews, and educated in the ultimate goals of the research project. The interviewees in this team project were a fairly homogenous group – rural 6–7th grade students – and all three of us interviewed youth in both grades, both male and female, gregarious and stoic. Yet, the interviews we conducted all turned out to be very different. What stood out to us was that our individual attributes as researchers seemed to impact the manner in which we conducted our interviews and affected how we accomplished the primary objective of the interviews, which was to elicit detailed narratives from the adolescents. Hence, we set forth to better understand how we, as research instruments, individually facilitated unique conversational spaces in our interviews and determine if there were some researcher attributes or practices that were more effective than others in eliciting detailed narratives from the adolescent respondents. Additionally, we sought to reflect on the emergent findings and offer a discussion of how unique conversational spaces might impact QRTs.

Gathering and analyzing empirical materials

The team-based qualitative research

Participants

The empirical materials for the current study came from a larger study designed to understand the social context of substance use for rural adolescents in two Mid-Atlantic States. A total of 113 participants between 12 and 19 years old (M = 13.68, SD = 1.37) were recruited from schools identified as rural based on one of two main criteria: (a) the school district being located in a ‘rural’ area as determined by the National Center for Education Statistics (NCES, n.d.; and (b) the school’s location in a county being considered ‘Appalachian’ according to the Appalachian Regional Commission (ARC). Participating schools served a large population of economically disadvantaged students identified by family income being equal to or less than 180 percent of the United States Department of Agricultural federal poverty guidelines and these guidelines start at an annual salary of $20,036 but increase by $6,919 for each additional household member (Ohio Department of Education 2010).

Interview team

Eleven interviewers comprised the qualitative research team for this team-based study. All underwent at least four hours of interviewer training, which reviewed interview protocol and procedures, summarized guidelines for ethical research, and included interview practice and feedback. During training, interviewers were given a clear interview schedule. Because the interviews were semistructured, the interviewers were instructed to use the schedule as a guide. They were instructed not to read the questions word-for-word from the interview schedule, but instead to use their own phrasing for asking each question, use additional probes or prompts if necessary, and use a communication style that felt comfortable and natural to them. Interviewers were also instructed to interact with their participants as learners attempting to understand the participants’ experiences and realities from their perspectives (Baxter and Babbie, 2004). All interviewers on the team participated in mock interview sessions and were provided with initial feedback about their interview skill.

Interviews

The interviews themselves were conducted in private locations within the schools such as guidance counselors’ offices or unused classrooms or conference rooms. In most cases, either the adult school contact or the study liaison brought students to their interview site to ensure that the interviewer did not know the students’ names – only their unique identification number. Researchers assured all students their responses would remain confidential, in accordance with Institutional Review Board standards, and the interviewee was permitted to withdraw his/her data from the study at any time. All interviews were digitally recorded and ranged from 18–91 minutes in length. This length is typical of interviews dealing with sensitive topics such as drug use in a school-based setting (Alberts et al., 1991; Botvin et al., 2000).

The present study: Three Voices in the Crowd

Interview sample

For the purpose of the present study we all agreed that self-reflexivity was necessary to ‘understand ourselves as part of the process of understanding others’ (Ellis and Berger, 2003: 486), increase the transparency of our findings, and increase the legitimacy and validity of our research. Therefore, we elected to limit our analysis to only those interviews that the three of us conducted, excluding transcripts from the other eight interviewers in the team-based study. Transcripts of the interviews were provided by a professional transcriptionist who was blind to the purpose of the study. A total of 18 interviews were transcribed (six per interviewer). Further refining the sample, we elected to analyze only interviews that we deemed to be of sufficient quality. Transcript quality was based on two indicators: (a) the level of transcription detail; and (b) the ability of the respondent to speak and understand English. Transcripts that were poorly done (i.e. that failed to include sufficient detail from the interview audio file) or that indicated that the respondent did not understand English were rated as low quality and were not included in final analyses. We took this step to ensure that all transcripts in the study sample were of sufficient quality and provided adequate detail to decipher our interviewer practices. From the 18 originally submitted transcripts, we found 13 to be of sufficient quality, and retained them for analysis.

Analysis procedures

Following Baptiste’s (2001) advice, the first step in our analysis was to acknowledge our interpretivist orientation and to honestly discuss among ourselves the risks involved with self-reflexively examining our own work. If you think it is difficult to listen to your own voice in an audio-recording, imagine listening to your own voice and simultaneously reading the text illustrating your own interview errors, dysfluencies, and awkward pauses! This first step was perhaps the most difficult, but it resulted in a shared agreement for honest self-reflection and analysis.

The next step involved restricting our analysis to three specifically selected topics from the research interview. The three discussion topics included rural living, identity and future selves, and risky behavior. We identified these topics of discussion because they each represented a different level of emotional risk for the respondents (Corbin and Morse, 2003), based on the assumptions that (a) respondents were all relatively similar in their emotional well-being – specifically, that none were too emotionally fragile to engage in a conversation with us, and (b) discussing topics of illegal or private activities would arouse more powerful emotions in respondents than would topics of legal and mundane activities. Across the entire sample of interviews, conversations on rural living were seen as fairly low-risk topics of discussion. The topic often served as a warm-up for many interview conversations because the topic was easy for respondents to discuss. Conversations on identity and future selves were typically perceived as moderately uncomfortable for respondents. Respondents were asked to talk about their personality characteristics and who they wanted to become in the future. Although some respondents appeared to enjoy the opportunity to talk about themselves, many appeared mildly uncomfortable doing so, perhaps because they were being asked to talk about themselves with someone they did not know. Conversations on risky behavior were often perceived to be more dangerous. Despite being reassured that their stories would remain confidential, respondents were nevertheless being asked to disclose information about potentially illegal activities in which they had taken part. These topics of discussion were not always mutually exclusive (e.g. respondents often talked about risky behavior when they discussed rural living); but, because every interview in the larger study included topics of discussion that were low, moderate, and highly sensitive, we believed that the three chosen topics of discussion represented an appropriate cross-section of the interview.

Dividing interviews into topics of discussion provided a way to organize long transcripts into relatively distinct topical areas. It also allowed us to examine interviewer practices across comparable topics of discussion, and to assess the ways in which particular characteristics facilitated different conversational spaces.

The next step involved identifying and labeling the discussion of each of the three topics within each of the 13 transcripts. As we labeled the related passages in the transcripts, each of us followed the same iterative analytic process, commencing with an analysis of our own individual transcripts and followed by a cross-case analysis of each others’ transcripts. Our individual, within-case analysis proceeded along four main steps: reading through our own transcripts 2–3 times before extracting the separate topics of discussion; then within each topic of discussion across all of our own interviews, we inductively identified, interpreted, and labeled what we each saw as important in the utterances, sequencing, and details of the conversational interaction, assessing the ways in which interviewer practices seemed to facilitate and to inhibit respondent disclosure. For our purposes, we defined an interviewer practice as an action performed repeatedly. These practices were eventually categorized into groups of interviewer characteristics. We conceptually defined an interviewer characteristic as a distinguishing general feature or overall quality of the interviewer. Throughout this process we individually developed and refined our code lists, discussing our emergent codes with one another via weekly meetings and email correspondence. As part of this process, we coded our own transcripts and then shared and discussed our code list with the others. Next, each of us (re) coded a portion of each other’s transcripts and calculated the percentage of raw coding agreement. Disagreements were negotiated until we all reached consensus on a working list of codes. This cross-case analysis did not commence until we had reached a minimum coding agreement of .80. Within the topic of rural living, for example, if two of us each generated five codes to describe one interviewer’s researcher-as-instrument characteristics, consensus was necessary on at least four of those codes before a trustworthy assessment could be made.

During the cross-case analysis we compared and contrasted the coded material within and across the entire sample of transcripts to identify discrepancies and consistencies in our codes. From this process, we reduced the code list to a common set of researcher-as-instrument characteristics and interviewing practices that were present in the utterances, sequencing, and details of the conversational interactions. Throughout this process we explicitly identified evidence (excerpts from the interview transcripts) for any research claim to connect the empirical materials with any findings (Maxwell, 1996). The three of us met periodically to conference, share ideas, and challenge and refine emergent findings. We used Nvivo 8 to manage and analyze the interview data. In the end, we were able to (a) identify and describe individual interviewer practices that served to characterize each of us as individual interviewers, and (b) compare and contrast our individual differences within and across the different topics in the interview conversation. During this comparison we paid special attention to the adolescent’s contribution to the conversation and his or her level of disclosure.

Findings

Interviewer characteristics

Annie

Annie’s general interviewer characteristics were coded as affirming, energetic, and interpretive. The affirming characteristic was defined as ‘showing support for a respondent’s idea or belief’ and is illustrated in the following excerpt:

Annie: What do you do?

Resp: I help the milkers, I help –

Annie: You know how to milk a cow? That’s so cool, that’s great.

Resp: Yeah, but you have to watch out ’cause they kick sometimes. ’Cause they don’t want you messing with their teats – they kick, it’s, uh …

Annie: Have you been kicked?

Resp: I got kicked in the arm, but I’m scared I’m gonna get kicked in the face one of these days.

Annie: Yeah, that would really hurt, huh? Oh, wow, that’s amazing.

Comments like ‘that’s so cool, that’s great,’ and ‘Oh, wow, that’s amazing’ illustrated the affirmation. Annie’s affirming characteristic could be seen in other transcript passages in phrases such as ‘great,’ ‘awesome,’ ‘amazing,’ and ‘excellent.’ Annie’s interviewer characteristics were also coded as energetic, defined as ‘showing wonder, astonishment, or confusion by something respondent said that was unexpected, or remarkable.’

Annie: So you like dirt bikes. Do you have one of your own?

Resp: Yeah, I have a, it’s a one, it’s a two-fifty. It’s like a, it’s a CRX 250, it’s like …

Annie: Oh, wow! Is it a pretty big bike? Wow, what do you like to do on it?

Resp: I just ride around in the fields and usually chase after deer on it.

Annie: Really!

Annie: Um, is your sister older or younger?

Resp: She’s younger, she’s ten.

Annie: So you kinda look out for her?

Resp: Yeah. She likes to feed the calves.

Annie: Oooooh!! Cute little baby calves. That’s neat. Wow! How unique. That’s really, really cool.

Annie: What’s a – dwarf bunny? What is that?

Resp: Yeah, they’re like little bunnies – they’re about that big.

Annie: Like real bunnies?

Resp: Yeah, they’re about that big –

Annie: Oh, dwarf bunnies. Oooh!

The sheer number of exclamation marks in Annie’s transcripts illustrated her energetic interviewer characteristic, but the words she used (wow, really, oooooh) also illustrated the lively quality of her interview approach.

Lastly, Annie was also characterized as being interpretive, conceptually defined as ‘expressing a personal opinion or interpretation regarding something a respondent said.’ For example:

Resp: And I chugged it and like, I passed out.

Annie: Did you have to go to the hospital?

Resp: Oh no. We were in the middle of the woods and we weren’t saying anything ’cause we all would get busted.

Annie: Oh my gosh, oh, you must have felt terrible.

Annie: Do you think that he drinks beer, or does chew or smokes cigarettes?

Resp: He probably does, but –

Annie: Do you think so? Um, and so when he offered this to you, were you, were you uncomfortable? Like, did you feel kind of weird?

In all of the above passages, Annie’s interpretive nature is evident in instances where she offers her own construal of the respondent’s story (e.g. ‘you must have felt terrible’), or when she creates a hypothetical scenario for the respondent to comment on (‘do you think he drinks beer?’). Such utterances illustrate her tendency to offer an opinion, either in response to a respondent’s story or before a conversation formally began.

Jonathan

Jonathan’s interviewer characteristics were characterized by neutrality and naivety. The neutral interviewer characteristic, defined as ‘not engaged on one side of argument or another; neither affirming nor disapproving of respondent’s stories,’ was best illustrated by the lack of extensive commentary Jonathan provided in his interviews. In comparison to Annie’s transcripts, Jonathan’s transcripts were characterized by shorter utterances, fewer opinionated responses, and very few exclamation marks:

Jonathan: Who were you living with in [name of town]?

Resp: My mom. But she, my grandma got custody of me, so.

Jonathan: What, what happened to do that? Like, what, what brought you?

Resp: Well, I got put in [the local in-patient treatment facility] ’cause I said I was gonna kill myself.

Jonathan: Oh, okay.

Jonathan: Okay. What, um, so does your dad mind if you drink then? Like, if he found out that you were going to the bar party and that you had gotten drunk, what would he say?

Resp: He probably wouldn’t do anything because, like, I used to have parties at his house, at my dad’s house. But then he got, then he went to jail, so we stopped [lowers tone, quieter] In case, like, ’cause they were keeping a good eye on him after he got out.

Jonathan: Mm hmm.

Resp: So we stopped having parties there, just so that, like, my dad wouldn’t get in trouble for, like, the underage drinking.

Jonathan: Okay.

It was often difficult to even see evidence of Jonathan’s ‘footprint’ in his transcripts because he maintained a fairly minimal presence in his interviews. As seen from the illustrations above, Jonathan kept many of his responses or comments to single-word phrases, ‘Okay,’ or ‘Mm hmm,’ or ‘Yeah.’ When Jonathan did offer more extensive commentary, it was often to acknowledge his lack of understanding about a subject matter. His transcripts often included passages like ‘I’ve never been here before’ or ‘I don’t know anything about that.’ It was in these instances that Jonathan’s interviewer characteristic of naive, defined as showing a lack of knowledge or information about respondent, was best illustrated:

Jonathan: Is it like illegal? Or is it like the whole town shuts down, they do racing down the streets?

Resp: It’s illegal.

Jonathan: Yes? I don’t know – you got tell me these things. I am learning.

These illustrations of naivety were most likely uttered to give the respondent a sense of mastery over the interview topics of discussion, and to elicit the respondent’s interpretations of the events or topics of discussion.

Michelle

Michelle’s interviewer characteristics illustrated different qualities than either Jonathan or Annie. Michelle’s qualities as an interviewer were coded as being high in affirmation and self-disclosure. Michelle’s transcripts were filled with encouragement and compliments toward her respondents. The following utterances from Michelle illustrate this characteristic:

My goodness, you are smart for a seventh grader …

It sounds like you are very helpful …

Yes, that is a skill that you have there, that not a lot of people do have …

These instances of affirmation, defined as ‘showing support for a respondent’s idea or belief,’ were found in almost every topic of discussion. Michelle’s transcripts were also filled with instances of self-disclosure. Michelle often used stories of her adolescent son when she was explaining a topic that she wanted to discuss with the adolescent respondents:

Resp: On Friday nights, tonight I’ll go to my gran’s and we usually have a get-together and just play cards, it’s just a thing we do. I like it. It’s just time to spend with family.

Michelle: Absolutely. Well, that sounds really nice. And I have a 14-year old in eighth grade. And every Sunday night, we do the game night sort of thing and I look forward to it.

The passages above illustrate three distinct interviewer characteristics: one high in affirmations, energy, interpretations; another characterized by neutrality and naivety; and another high in affirmations and self-disclosure. Although all three interviewers demonstrated other instrument qualities in their interviews, the few qualities associated with each interviewer above were found in nearly every topic of discussion (e.g. in almost every conversational topic for Annie, there was evidence of her affirming, energetic, and interpretive interviewer characteristics). These qualities seemed to characterize the unique style of the interviewers rather than reflect reactions to specific contexts. These qualities also persisted in our other interviews not included in these analyses.

Topics of discussion

In the following section, we compare our general interviewer characteristics across the three topics of discussion: rural living, identity and future selves, and risky behavior. We also examine the ways in which our respective interviewer characteristics appeared to influence the conversational space of our interviews. Specifically, we assess how the various interviewer characteristics seemed to facilitate or inhibit respondent disclosure.

Low threat topic: Rural living

Rural living was generally a low-risk topic. In her discussion of this topic with one adolescent, Michelle tended to utilize her self-disclosing characteristic:

Michelle: Are there groups or, like, not cliques, I don’t wanna say, but groups in school; kids who are more like you, who are more into the computers, versus the kids who are huntin’ and fishin’, versus the jocks? I know at my son’s school there are.

Resp: There’s not really anybody like that here. Like all of my friends who are like that, they’re in a higher grade than me. But there are some people in my grade where I can relate to in a sense, yeah.

Michelle: Okay, so most kids you can relate to are older but most o’ the kids, your peers and your age, are more into the four wheeling and hunting and fishing and kinda stuff like that? That must feel, well, I don’t know, I’m, I’m projecting now unto my own son because sometimes he feels like, that you know, it’s just ridiculous.

Resp: Yeah.

Michelle: It, eh, ya’ know – and you feel kinda stuck.

Resp: Mmm hmm.

Michelle: Yeah?

Resp: Yeah. I just, like I’ll be sitting there in class and then they’ll start talking about hunting or fishing and I just wanna pull out my hair’ cause I, I don’t know how you can like that stuff. Like it’s just sitting there for a couple of hours doing nothing.

Michelle: Right, right.

From the excerpt above, the respondent’s experience with school crowds did not appear to coincide with Michelle’s understanding of her son’s with school crowds. However, Michelle’s self-disclosure seemed to open up the conversational space for the respondent to respond in kind. In the final passage, the respondent offered a different perspective on the nature of crowds in his school.

Conversely, in his conversations with respondents about rural living, Jonathan tended to demonstrate his naive interviewer characteristic:

Jonathan: Is this [name of X town]? Is that where you live now? I don’t even know where I am. Okay, okay. I thought this was [name of Y town] is why, but it’s just the name of the High School.

Resp: Well, this is [name of Y town], but [name of X town] is out near.

Jonathan: Uh, I’m not, I don’t know this area so well …

Resp: And then, like, when you hit, there’s this big huge fire station … and then there’s the [name of X town] Elementary School. And then if you go down there and then you turn and you go up, and then that, like, that whole area in there is [name of X town].

Jonathan: Okay.

Resp: And then you go back and where there’s classified as [name of X town], but it’s actually [name of Z town].

Jonathan: Okay.

In response to Jonathan’s naivety (‘I don’t even know where I am’ and ‘I don’t know this area so well’), the respondent appeared to seize the opportunity to teach Jonathan about the area. The respondent did not simply answer Jonathan’s questions; he provided information about which Jonathan did not ask (e.g. the whereabouts of the fire station, elementary school, and nearby towns).

In contrast, Annie’s conversations about rural living were filled with her energetic interviewer characteristic:

Annie: What do you mean by hang out, like what do you ha-, what do you do when you hang out?

Resp: We go four wheeler riding.

Annie: Oh, four wheeler riding! Cool! Is that dangerous? Is it?

Resp: Yeah, and we go up to our camp we built. Um …

Annie: That you and your friends built?

Resp: Mmm hmm.

Annie: Wow! How did you know how to do all that?

Resp: Um, my brother and a couple of his friends, that we’re really good friends with, helped us. And like, over the summer we camp out like every night. Like, I’m never home in the summertime, ever.

Annie: Wow!

Resp: There are three bedrooms and it’s, has a wood burner and it, yeah.

Annie: That’s like, that sounds like a real house. That’s amazing.

Resp: We built it out of trees. We had our, couple of our friends and our dads help us. We’ve had it for three years and it’s really nice.

After Annie’s lively reply to the respondent’s interest in four wheeler riding (‘Oh, four wheeler riding! Cool!’), the respondent opened up about a different, but related topic: her summer camp house. Moreover, Annie’s energetic comment about the house (‘Wow! How did you know how to do all that?’) seemed to open the conversational space even more, as the respondent explained the ways the house was built, the amenities of the house, and the amount of time she spent in the house during the summer.

Moderate threat topic: Identity and future selves

Conversations about the adolescents’ identity and future selves were considered moderately uncomfortable for adolescents. The interview questions prompted the adolescents to talk about the qualities that described their personal and social identities, along with any hopes and aspirations they had for the future. Although the interview questions were designed to be as unobtrusive as possible, the topic was fairly personal. The interview questions required the adolescent respondent to be introspective with someone with whom they had no personal history:

Jonathan: After you’re all done with school, so you go through and you graduate from a high school. What do you want to do after that?

Resp: Go back to Mexico and visit my family, and um get a job.

Jonathan: Back in Mexico?

Resp: It doesn’t really matter where, but just like get a job.

Jonathan: Yes. What kind of job?

Resp: Probably like a secretary or whatever job they give me, except prostitute.

Jonathan: None a’ that. Is there anything you worry about in that transition of how you’re going to go get a job and what kind of job you’ll get, things like that?

Resp: Not really, because like, you just have to like – I dunno, just like – just like – go on with life and whatever happens, just, take it.

Here again, Jonathan’s neutrality was demonstrated not by what he said, but what he did not say. Despite the fact that the adolescent shared a potentially troubling disclosure, that she would consider any job except prostitution, Jonathan kept his personal reactions to a minimum and provided only a short response (i.e. ‘None a’ that’). After this instance of neutrality, Jonathan moved on to a different topic (i.e. asking the respondent if she had any concerns about getting a job in the future), and the respondent moved on, as well, dutifully answering his questions. She provided no more information on her prostitution comment.

In comparison to Jonathan, Michelle and Annie’s utterances in their conversations on identity and future selves were replete with codes for affirmation:

Resp: I wanna be a pediatrician nurse or something. Like, I love kids to death. Like, I’ve, I learned how to change – I’ve been changing diapers – this is no lie – I’ve been changing diapers since I was like seven years old. ‘Cause my mom, step-mom, had a baby before my dad left again, and like I was always changing her diapers and stuff, and like, I babysit constantly.

Annie: Aww, I bet you’re really good with kids.

Resp: Oh, I’m amazing. Like, there’s this one little boy, like he goes to my church, he’s just like four, and I took him to my house one day and like he asked his mom to buy him a toy at the toy store, I cried, she’s like, she’s like, ‘Aww, I can’t sweetie, I don’t have the money’ and he was crying, he and he’s like ‘All my friends have toys. He was like two and he, like he, like he goes over to this daycare and he’s like ‘All my friends have these toys but I don’t have any.’ Like he had no toys at all and like my mom gave them, handed me a hundred dollars and she’s like ‘Go to, go, go buy toys. We gave him a hundred dollars, like we gave him all this money and they went out and bought like a b-, toys and stuff. It was really nice.

Annie: That is, that’s really neat.

Michelle: So the first question that I have here is which of these things that you wrote down are you most proud of?

Resp: Well, being helpful.

Michelle: How are you helpful?

Resp: Well. In school, there are some people that don’t like speak English that well. And I help them by translating.

Michelle: Oh okay. Like you are doing for your teacher in there. You are helping do that. So how long have you been bilingual your whole life? Do both of your parents speak Spanish?

Resp: Well, yes, they are Mexicans. They barely know English.

Michelle: And they barely know English. And when did you come here?

Resp: When I was nine months old.

Michelle: When you were a baby. And before that you lived where?

Resp: In Mexico.

Michelle: Mexico. So you are 13, so that was when you were a year old. Okay, got it. Okay, so you learned here. So you speak English better than they do it, sounds like. Okay and then you translate. What’s that like translating for them?

Resp: Well, for me it’s like sometimes difficult because I never went to school in Mexico and I know more English than Spanish and when I am translating it’s difficult for me. The big words my parents tell me to try to translate it in English.

Michelle: Okay. So you’re doing both ways. You’re doing from English to and from Spanish to English. Both. Does that feel like a lot of responsibility for somebody your age?

Resp: Yeah, especially when I got field trips stuff like that. I need to tell my parents, that my parents or if my parents needed something that comes in the mail, may be bills or something like that.

Michelle: It sounds like you are very helpful. Who do you want to be when you are out of after high school?

Resp: Since I like to help out people a lot, I mean, maybe be a translator and maybe in a hospital or in a school so –

Michelle: Yes, that is a skill that you have there, that not a lot of people do have. So that’s – I’m glad you realized that, in terms of that.

Annie’s affirming characteristic could be seen in her affirmation of her respondent’s compassion for children (‘I bet you’re really good with kids’); for Michelle, the characteristic could be seen in her affirmations of her respondent’s willingness to help her parents, teacher, and classmates with their English or Spanish (‘… it sounds like you are very helpful’). Both Michelle and Annie’s affirmation seemed to foster a conversational space that was conducive for uninhibited self-disclosure. In response to Annie’s affirmation about owning a daycare someday, the respondent opened up to talk about her talents in working with children, and her compassion for the children in her community who were less fortunate than she. In response to Michelle’s affirmations about the responsibilities of translating for so many people, the respondent expounded on the difficulties of such a responsibility, and the tasks she must perform for various people (e.g. helping her classmates on field trips, assisting her parents with bills).

High threat topic: Alcohol, tobacco, and other drug use

Discussions about alcohol, tobacco, and other drug usage (ATOD) were considered highly sensitive topics of discussion, as adolescents were often encouraged to disclose information about their own or their peers’ drug use. Although the respondents were continually reassured that the information they provided was confidential, disclosing information about illegal activity to a stranger was likely a highly sensitive activity. When discussing ATOD with adolescents, each interviewer utilized a different interviewer characteristic. Jonathan’s dominant characteristic when discussing this topic was neutrality:

Resp: Her parents’, like, bar. Like, they own this big, huge bar. And then, like, in the back where the kids can go.

Jonathan: Oh, okay.

Resp: And her parents don’t really care if you drink.

Jonathan: Oh, okay.

Resp: Just as long as you do it in the bar. You don’t just go outside, or you don’t tell your parents.

Jonathan: Okay.

Resp: She doesn’t really know that we drink, but we usually crash in the van, in the RV.

Jonathan: Uh huh.

Resp: … or out in the yard. And we only do the RV in the summer or in the spring. And then at my other friend’s house who has the bar, we stay at, we do the, we have parties there all the time.

Jonathan: Mm hmm.

Resp: Just cause her parents don’t care.

Jonathan: Yeah.

Even in the midst of some fairly controversial topics of discussion (e.g. underage binge drinking), Jonathan’s neutral characteristic was consistently demonstrated in his calm, even responses (‘okay,’ ‘uh huh’). These neutral responses seemed to provide an unobtrusive backdrop for the respondent to discuss her experiences. Indeed, Jonathan did not even need to ask any questions to the respondent. With minimal prompting, the respondent shared her story.

In comparison to Jonathan, when discussing ATOD, Annie’s approach was coded as interpretive; she often interjected commentary about the respondents’ stories of risky behavior:

Annie: Do you think that he drinks beer, or does chew or smokes cigarettes?

Resp: He probably does …

Annie: Um, and so when he offered this to you, were you, were you uncomfortable? Like, did you feel kind of weird?

Resp: Mm hmm.

Annie: Um, and, and maybe that boy’s brother – like, that guy’s brother – he might smoke or drink from time to time, but, um, that’s about it?

Resp: Mm hmm.

Annie: It doesn’t seem like too many kids around here do that stuff.

Resp: Not as I know.

Annie’s interpretive characteristic stands in stark contrast to Jonathan’s neutral characteristic. Whereas Jonathan’s responses were short and dispassionate, Annie’s responses were somewhat opinionated. These interpretive comments did not seem to generate a conversational space conducive for the respondent’s continued disclosure. Indeed, the transcript above shows that most of the commentary came from Annie, not the respondent.

In discussions on risky behavior, Michelle’s self-disclosing characteristic was evidenced by her stories of her 14-year-old son, and appeared to serve as a point of identification with respondents:

Resp: My parents get mad because I listen to music a lot and I don’t do anything than watch TV. Just hang out with my friends.

Michelle: Then your parents get mad because that’s all you do. You know but the good thing about me is I’m not your parent and I don’t care. So I just want to know what kids are doing. It’s, you know, I have an eighth grader actually he’s 14. And that’s exactly what he does. And in the winter it stinks, though you are right because what else is there to do? You know it’s the question, um any way, okay. So, do you know my question to you is, and again, this is purely confidential, we don’t know names we don’t want names or anything. Has anybody ever offered you any alcohol or cigarettes or marijuana or any of those? And have you said yes or no to that?

Resp: Yes, they offered me and I’d always told them ‘no’ and what it does.

Michelle: Okay, so tell me … pretend that we’re shooting this video. Okay tell me the who when what where why and how. Right? Where were you, not who, not a name. But was it a friend who was older, younger, male, female? That kind of thing. Tell me the story of at least one of these offers.

Resp: Okay. I was hanging out with my friends, just walking around, and there is this bigger kid that we know and he was joined by these smokers, and they would always, he would always tell me never to smoke and we just saw him … And then he offered us and we said no. This is not good for you and he plays soccer and he is not really good at soccer.

Michelle’s self-disclosure about her son experiencing similar challenges as the respondent was initially met by the respondent with a short response. However, Michelle’s subsequent question, framed as a hypothetical task (‘pretend that we’re shooting this video’), seemed to create an opening in the conversational space for the respondent to share a story.

Summary and discussion

In looking closely at the different practices we employed as interviewers, we were able to identify a variety of distinguishing features that seemed to characterize each of us uniquely. If we were characters in a novel or play, Annie’s character name would be energy, Jonathan’s neutrality, and Michelle’s self-disclosure. Across the different conversation topics in the interview, from low to high risk, these interviewer characteristics functioned differently in eliciting detail from adolescent respondents.

When the adolescents and researchers discussed the low-risk topic of rural living, the three interviewer characteristics (i.e. energy, neutrality, or self-disclosure) generated sufficiently detailed responses from the respondents. Variance across interviewers did not seem to have much impact on the quality of the responses obtained from the adolescent participants. This may have been due, in part, to the low-risk nature of the topic. This is a topic many adolescents can talk easily about, have talked about with others, and do not perceive the information they share as particularly threatening.

When the topic was moderately risky, as was the topic of identities and future selves, Jonathan’s neutral approach contrasted with Michelle and Annie’s affirming approach. Although neutrality appeared somewhat effective in facilitating an open conversational space for respondents, the affirming interviewer characteristic seemed to offer a more nurturing environment for conversation. Rich, detailed disclosures from adolescents about their identities occurred more often when the interviewer utilized an affirming approach and set a tone of acceptance for the respondents. Affirmation may be particularly important with adolescents, since adolescence is a notoriously vulnerable time in development.

When discussing a high risk topic such as alcohol and other drug use, Annie’s interpretive approach appeared to be the least effective in providing a satisfying conversational space for respondents. Jonathan’s neutral characteristic and Michelle’s self-disclosing characteristic appeared to elicit detailed information from their respondents, while Annie’s interpretive characteristic only served to inhibit her respondent’s stories. Michelle’s disclosures, while also interpretive, did not appear to limit responses from the adolescents. Couching Michelle’s interpretive language within a personal narrative may have mitigated its presence, although it still presented leading information. Hence, it could be argued that neutrality (displayed in this context by Jonathan) may be most effective when discussing high risk topics, because this neutrality provides the respondents with the most freedom to disclose what they want and how they want.

An important factor to note in this discussion is that of gender. While we did not explicitly study the role of gender in our analyses, our interviewing styles were rooted in traditional gender norms: Jonathan’s minimalist and neutral styles could be characterized as stereotypically masculine, and Annie and Michelle’s effusive and affirming interviewing styles could be characterized as traditionally feminine. These qualities suggest that interviewing styles cannot be disentangled from one’s gender, and that conversational spaces are influenced by more than simply an interviewer’s words. To this end, practices of reflexivity must acknowledge the implications of what an interviewer says and how it is said, as well as the ways in which those utterances are connected to one’s gender.

Although this study provides some intriguing findings, it was limited in a variety of ways. For one, we did not employ detailed conversation analysis procedures on each individual utterance in the interview. And despite the range of conversational segments in the interviews (i.e. introductions, research explanations, establishing rapport, soliciting honesty and openness, a period of questions and answers on six core topics, summarizing the discussion, and closings), for the purposes of this study, we elected to limit our analysis specifically to three topics in the question and answer segment. Nor did we examine other conversational features, such as the role of silence or turn-taking. Conversational features such as those, while certainly worth our attention, were beyond the scope of this exercise.

Lessons learned

Learning about interviewing and doing interviews are different tasks. This lesson was highly relevant for us when conducting this study. Even though we were all trained in interviewing, we still found ourselves displaying the classic mistakes of a novice researcher: asking long, complicated questions, posturing closed yes-or-no questions, and leading respondents (deMarrais, 2004). While humbling, these mistakes forced us to reflect on how to develop our skills and have guided our interviewing work since that time. Indeed, the kind of self-reflexivity involved in conducting an analysis of your own interviews, and then comparing and contrasting them with others, could be beneficial for individual interviewers as they are honing their craft, and QRTs desiring to identify unique characteristics of their resident interviewers.

In considering our findings, we agree that researchers are indeed the ‘instruments’ in qualitative interview research. After all, it is through the researcher’s facilitative interaction that a conversational space is created where respondents share rich information about their lives. Yet, we argue that qualitative researchers are differently calibrated instruments.

In QRTs, in particular, the goal is often to calibrate all instruments to one standard of accuracy. However, the results of this study illustrate that variation in interviewer characteristics may be a benefit rather than a detriment to team-based qualitative inquiry. All interviewers in this study were effective in conducting engaging conversations with participants and eliciting information, but we did these things employing different practices, and sometimes to different ends. Each interviewer demonstrated a relatively consistent interviewer style across all of his or her interviews – Jonathan was consistently neutral, Michelle consistently self-disclosive, and Annie consistently energetic. This finding leads us to suggest that QRTs might benefit from learning what ‘natural style’ characterizes a possible interviewer and then staffing their teams with interviewers who have complementary styles. Interviewers may then be assigned interview tasks commensurate with their strengths. For example, our team needed to learn both about rural identity and about alcohol and drug use, so Michelle and Annie could have been assigned to interview respondents about rural identity (a ‘safe’ topic) and future selves (a moderately risky topic), which both fit our energetic style. This approach could have helped to engage participants in the research and establish rapport with them among the research team. Then, Jonathan could be assigned to the task of summarizing the information learned about the less risky topics and bringing that information into a second interview to pursue the high risk topic of drug use, implementing his neutral style for a non-evaluative conversational space. This suggestion is founded on a premise similar to utilizing information from personality inventories (e.g. Myers Briggs) to establish work teams in organizations (Furlow, 2000).

Since many interviews must occur during a single visit, however, interviewer ‘profiling’ may not be realistic for QRTs. Another suggestion would be to audio-record interview trainees in mock interviews, share those recordings among the team, then devote some time for team members to offer commentary on (a) the ways in which their teammates embodied similar or different instruments in their interviews and (b) how those instruments seemed to create different conversational spaces. This process need not involve detailed conversation analysis tools; nor should it be formal or performance-based. Instead, it should be congenial and constructive, driven by efforts to respect interviewer flexibility while maintaining fidelity to the research approach. These recommendations are in line with calls issued by Mallozzi (2009) and Miller-Day et al. (2009), who argued that consistency efforts be focused on research procedures (e.g. securing consent, managing empirical materials) and not on standardizing interviewer characteristics.

In carrying out these recommendations, more research will be needed to understand the complexities of how and under what conditions interviewer characteristics may impact respondent responses. More research will also be needed on the ways QRT practices may change if reflexivity was incorporated at other stages of the process (e.g. forming research questions and gaining access). Yet this study provides a running start toward that end. Through our exercise, we call for greater interviewer reflexivity and acknowledge that researchers are the primary instruments in qualitative interview studies – but differentially calibrated instruments. We disagree with claims that interviewers in qualitative research teams should receive the same standard training with an eye toward producing consistent interview strategies (Bergman and Coxon, 2005) and argue, instead, that diversity of approaches among members of a research team has the potential to strengthen the team through complementarity.

Biographies

Annie Pezalla is the Academic Skills Director at Walden University. Her research addresses identity development across adolescence and young adulthood.

Jonathan Pettigrew is a research analyst and project coordinator for the Drug Resistance Strategies project at Penn State University. His research examines how interpersonal and family interactions correspond with adolescent health.

Michelle Miller-Day is an Associate Professor of Communication Arts and Sciences at the Pennsylvania State University. She is the Founding Director of the Penn State Qualitative Research Interest Group, an interdisciplinary community of researchers involved in and supporting qualitative inquiry at Penn State University. Her research addresses human communication and health, including areas such as substance use prevention, suicide, and families and mental health. Her community-embedded research has involved numerous creative projects to translate research findings into social change. For the past 20 years she has served as the principal qualitative methodologist for a National Institute on Drug Abuse line of research.