Ambassador,you’re on Mute: Methods for Co-interviewing Elites Online

Abstract

This article draws on the experiences of two previously unacquainted researchers working at distance who collaboratively used online platforms to co-interview respondents across a dozen countries in 2021–22. Rather than our research plans being disrupted by the constraints of COVID-19, we planned from the outset to conduct our research entirely online. Given our geographic separation, personal circumstances, and globally dispersed respondents, online methods were imperative to our research collaboration. At the same time, conducting co-interviews online introduced the challenge of structuring both effective interviewer-to-interviewer and interviewer-to-interviewee interactions in a virtual environment, where conventional methods might not suffice to create productive engagements, and where non-verbal communications are both more difficult to convey and interpret. In this article, we explore and reflect on our strategies and tactics for co-interviewing in contexts where online methods are a deliberate, welcome choice, rather than the fallback option. We focus on the nuances of interviewing elites, offering insights to help researchers prepare for the process, rather than only the content, of such interview encounters.

Introduction

More than four years since the advent of COVID-19, the constraints and consequences of the pandemic for social science research are becoming normalized. The disruption of research practices precipitated by the pandemic is by now well documented (Nind et al., 2021; Teti et al., 2021). Meanwhile, in recent years, many researchers have gained greater familiarity with platforms such as Google Meet, Microsoft Teams, Skype, and Zoom, making internet-based tools a progressively more accepted feature of social science research (Deakin & Wakefield, 2014; Lo Iacono et al., 2016; Seitz, 2016).

When COVID-19 struck, qualitative researchers with projects underway often reluctantly transitioned their methods of data collection from in-person, face-to-face encounters to internet-based platforms as restrictions ebbed and flowed (Abdul, Tuckerman, & Gherhes, 2021; Lobe et al., 2020; Oliffe et al., 2021). Some researchers completely abandoned planned methods in lieu of these online alternatives (Carter et al., 2021; Wahl-Jorgensen, 2021). Despite the obvious inconveniences and disruptions, it is now widely recognized that the COVID-19 pandemic also prompted methodological innovations in research (Keen, 2022; Maulana, 2022; Roberts et al., 2021; Sy et al., 2020). Now, however, researchers explicitly face choices about research design and methodology without necessarily being obliged to shift course out of necessity, as was the case at the onset of the pandemic.

In this article, we discuss our experience of conducting research online, not as the result of a transition of approach, but as a deliberate, positive choice. Shared research interests precipitated the formation of a team of two researchers who, while familiar with some of each other’s work, had never met face-to-face or previously worked together. Throughout the research period, we had no opportunity for an in-person encounter, living continents apart. Even if pandemic-era travel restrictions had not been in place, with one of us raising a young family, and the other constrained by bureaucracy with the effect of restricting travel, it is unlikely that we would have found the opportunity to meet in person or conduct research from the same physical location for an extended period. Moreover, we recognised at the outset of our research that many of our intended respondents – personnel who had formerly worked with, and/or who were knowledgeable about, the Organisation for Security and Cooperation in Europe’s Special Monitoring Mission in Ukraine (OSCE SMM) – came from more than 40 of the OSCE’s 57 participating states, and were not only distributed throughout these countries but were also likely to be found in other parts of the world, given the peripatetic nature of international employment. While occurring during the pandemic, our methodological approach to research was therefore chosen not because of the circumstances of COVID-19, but for these two reasons: the characteristics of our geographically dispersed research population and our own personal and professional circumstances.

While the literature in qualitative research has begun to discuss both the strategies and tactics of online interviews (e.g., Roberts et al., 2021; Salmons, 2015; Sy et al., 2020), our understanding of how two (or more) researchers can effectively employ these and other strategies as online co-interviewers remains limited. This is an important gap for two reasons. First, as our own research demonstrates, online settings enable researchers to co-interview regardless of locality. Therefore, we anticipate that such forms of research collaboration will become more common, making it even more important to understand and articulate online co-interviewing strategies, and consider how these may intersect with researcher personalities, identities, skills, and experiences. Second, it is widely argued that interviewing in online settings inhibits sending and interpreting of subtle gestures and body language (Chen & Neo, 2019, p. 2; De Villiers et al., 2022; Irani, 2019; Oates et al., 2022; Oliffe et al., 2021). However, the literature generally focuses on the interviewer-to-interviewee relationship, and not on interviewer-to-interviewer interactions—a crucial aspect of co-interviews. Although co-interviewing is common, and the literature has long recognised this method (Bechhofer et al., 1984; Groholt & Higley, 1972; Kincaid & Bright, 1957a, 1957b), examining online co-interviewing has not been the focus of the emergent methodological literature on online research. We argue that in online co-interviews, the importance of deliberate planning and preparation for the interview process, as well as the in-interview approaches co-interviewers take, is heightened. This is especially so if researchers lack a shared social history and close knowledge of, or proximity to, their co-researchers.

This article consequently makes two main contributions to the literature. Firstly, it identifies distinct strategies for co-interviewers to effectively elicit information during online interviews, a research context that presents particular challenges. These strategies, designed to leverage the advantages of co-interviewers, include: the ‘insider-outsider approach,’ where we strategically emphasize aspects of our professional and/or personal backgrounds relative to the interviewee, to show variation between interviewers; ‘good cop, bad cop,’ involving one interviewer in a more supportive posture and the other more critically probing or sceptical of the interviewee’s responses; and ‘knowledge ambiguity to elicit information,’ which entails not revealing knowledge too early in the process while demonstrating competence when answers remain too general or basic. Secondly, the article introduces tactics specific to managing the triadic relationship of interviewer-interviewer-interviewee in online settings, where non-verbal cues are more difficult to interpret. These tactics—including in-interview signalling, tactical interruption, and adapted turn-taking, are tailored to virtual environments and take advantage of the fact that two interviewers are collaborating. By articulating these distinct strategies and tactics for online co-interviews, we aim to equip researchers with approaches and tools to prepare for the process (as opposed to only the content) of conducting effective online co-interviews. Our reflections also show how deliberate pairing of co-interviewers independent of location enable a dynamic approach that facilitates data collection in service of clearly defined research questions.

In our analysis, we focus on elite interviews, for both practical and methodological considerations. Practically, our contribution is grounded in our own research, which primarily involved interviewing elites. From a methodological standpoint, concentrating on one type of respondent—elites, in our case—enables us to identify the challenges interviewers are facing more precisely, and to demonstrate how our strategies and tactics can address these challenges. Although our focus is on elite interviews within our specific field of research in peace and conflict studies, many of the insights and implications we discuss are relevant for other types of respondents and research areas, as we elaborate in the concluding section.

The article proceeds as follows. In the first section, we review the guidance on conducting in-person elite interviews, the emerging literature on conducting interviews online, and how the two bodies of work interact. We then show how extant literature fails to engage with the specificities of online co-interviewing. In the second section, we profile our research study, offering the rationale and context for conducting elite co-interviews online. In the third section, we describe three strategies we used throughout our co-interviews to manage and steer the conversation, building on and modifying existing methodological guidance and practice. In the fourth section, we focus on techniques within interviews that allowed us to navigate the complex interactions between both interviewers and interviewee. In the conclusion, we discuss factors that condition the suitability of our online co-interviewing strategies and tactics and how researchers may adapt them to address their own scientific inquiries.

What We Know About Elite Interviews

While there are many definitions of elites, for the purposes of this article we adopt the definition used by Huggins (2014, p. 3), in which elites are ‘actors who are in a privileged position in relation to a particular activity or area of policy.’ Within this somewhat broad definition, we particularly focus on what might be termed organizational elites: those persons who officially represent an organisation or institution or adopt the position that they do (or did) represent such an institution (Odendahl & Shaw, 2002, p. 20; Thomas, 1993, p. 88), and thus had institutional expertise (von Soest, 2023, p. 278). Elites have long been considered as more challenging interview subjects (Odendahl & Shaw, 2002; Ross, 2001). While it is important to not overly generalize assumptions about elites (Mason-Bish, 2019, p. 266), nor consider them as a monolithic category (Glas, 2021, p. 438; Smith, 2006, p. 645), organizational elite respondents may generally have interests in withholding information or being secretive (Adler & Adler, 2003, p. 6; Hunter, 1993, p. 39). This is particularly the case where information respondents hold is deemed sensitive – at least in the mind of the respondent (e.g. Liu, 2018; Thuesen, 2011, p. 618) – or when even granting an interview exposes or is perceived to expose a respondent to professional risks, as it may be understood to violate confidentiality or non-disclosure agreements of the organization with whom they are affiliated. In addition, elites are often selective about the information they reveal, and may frame issues or events in a way that serves their strategic purposes (De Graaf et al., 2015; Freedman, 2015; Roselle et al., 2014).

Elites may purposefully erect barriers to researchers (Figenschou, 2010; Mikecz, 2012, p. 483). Even if such initial gatekeeping hurdles are overcome and access to interviewees obtained, prospective participants may be suspicious of researchers, their methodologies, and their interests, if not the premise and purpose of the research itself (Bergman & Wettergren, 2015, p. 699). Elite respondents may not cooperate or see the interview as a challenge they need to overcome (Harvey, 2011, pp. 433–34), or have strategies to avoid answering by instead offering a general consensual view (Bergman & Wettergren, 2015, p. 693). Such respondents may be dismissive of researchers, see them as unknowledgeable or unworthy of their time (Conti & O’Neil, 2007, p. 73), or perceive researchers or their work as unimportant. Even when elite respondents are not suspicious nor dismissive, researchers may still have to prove their credibility and professional standing (Thomas, 1993; Welch et al., 2002), be subject to gatekeeping questions even once an interviewee has consented to speak with the researcher (Solarino & Aguinis, 2020, pp. 652–53) and have limited time to conduct the interview. Elite respondents may wish to lead the interview, have an agenda to promote in the interview (Freedman, 2015; Thuesen, 2011, p. 615), or draw the conversation on to topics tangential to the researcher’s interests (Stephens, 2007, p. 208).

In short, researchers may face multiple challenges when seeking to elicit relevant information from elites. Importantly, they must ensure that the responses serve to answer the research question, rather than the agenda of those interviewed. Many researchers argue that the development of rapport is key to eliciting information from elite respondents (Brounéus, 2011; Harvey, 2011; Ostrander, 1995; Stephens, 2007; Weller, 2017). We agree that establishing rapport is often desirable and may facilitate research. However, we argue that this is more complex in a co-interview setting; rapport between interviewees and interviewers is both dynamic and contingent on the strategies and tactics used within the constraints of a time-bound interview (Abell et al., 2006). To ensure optimal outcomes, these strategies and tactics ought to be adapted to the specific setting of an interview and be cognizant of the dynamics introduced in a co-interviewing situation.

What we know and do not know about online interviews

Interviews conducted online have become more common, particularly in recent years. Videoconferencing tools gained great prominence during the COVID-19 pandemic, and in many places, reliable Internet access is no longer an issue, especially for elites. Two decades before COVID-19, however, an early literature on online interviews was already emerging, in line with the technological developments of the time. In this first wave of literature, researchers often focused on establishing the validity and legitimacy of such approaches and extolling the potential advantages of accessibility, lower cost, and time saving, while conceding that online approaches might not fully substitute for in-person interviews (Anderson & Kanuka, 2003; James & Busher, 2006, 2009; Madge & O’Connor, 2002; Mann & Stewart, 2000; O’Connor et al., 2008). To a large extent, however, online interviews were conceived of as a facsimile of the in-person, face-to-face interview. Even a decade after some of these initial works, some methodologists still spoke in terms of the ‘translation’ of face-to-face methods to the online environment (Hooley et al., 2012, p. 66), rather than conceiving of the online interview as a form of qualitative research with distinct attributes. Meanwhile, some skeptics questioned the reliability of online methods (Beaulieu, 2004) or argued that the authenticity of online respondents’ views might be problematic (Markham, 2008).

As online interviews became more common, the second wave of literature on online methods emerged. This work engaged more deliberately with the methods of both asynchronous and synchronous online interviews. Much of this literature discusses mitigating the shortcomings of online interviews, such as the technological limitations of both participants and various web and software platforms (Deakin & Wakefield, 2014; Hanna, 2012; Janghorban et al., 2014; Lo Iacono et al., 2016). These studies highlight a range of technical challenges, including the potential for glitches to interrupt the flow of the interview and the quality of data collection (Deakin & Wakefield, 2014; Lo Iacono et al., 2016), as well as the effects of differing levels of technological competence and accessibility on the representativeness of the respondent pool (Deakin & Wakefield, 2014). Common approaches to mitigation of technical challenges include allowing respondents more time to answer, being prepared to repeat or simplify questions, and being as clear and concise as possible in research encounters. Beyond technical limitations, some authors offered a more critical perspective on the implications of conducting online interviews. Salmons (2015) asked researchers to reflect critically on their motivations for conducting interviews online, rather than via conventional means. Weller (2017, p. 622) argued that some of her research participants seemed to treat online interviews as more ‘informal encounters’, making the boundaries between researcher and participant more blurred. Scholars differed in their assessments of the advantages and disadvantages of online research for discussing sensitive subjects (Deakin & Wakefield, 2014; Seitz, 2016, p. 232) or for engaging hard to reach populations (Jenner & Myers, 2019; Wilkerson et al., 2014).

What we term the third wave of literature on online methods developed as a result of COVID-19. Beginning in 2020, work emerged on how to continue qualitative data research despite social distancing, travel, and other public health restrictions affecting both researchers and research participants (Keen et al., 2022; Lawrence, 2020; Lobe et al., 2020; Lupton, 2020; Saarijärvi & Bratt, 2021; Self, 2021; Sy et al., 2020). With the normalization of practices of working from home and at distance both because of pandemic-era restrictions as well as shifting societal norms, there is a clear temporal demarcation between earlier methods literature on online interviewing and this more recent wave of scholarship. What characterizes the third wave of literature is a growing recognition that online interviews need not be considered second-rate compared to in-person interviews, and may in some circumstances even be superior (Archibald et al., 2019; Hyde & Rouse, 2022; Zadkowska et al., 2022, p. 2). For example, Oliffe et al. (2021, p. 3), recounted online research participants as being more comfortable speaking from their homes, thus leading to richer conversations. Similarly, Wahl-Jorgensen (2021, p. 375) identified ‘novel intimacies’ in speaking to respondents via Zoom (see also Maulana, 2022, p. 397). Moreover, several scholars argue that discussing sensitive subjects was often easier in an online setting (Opara, Spangsdorf, & Ryan, 2023, p. 567; Sipes, Roberts, & Mullan, 2019). While some scholars voice scepticism or concerns with online interviews (e.g., Howlett, 2021; Teti et al., 2021, p. 3), in the current wave of literature, there appears to be a growing understanding that online interviews are methodologically different than in-person interviews, and have their own logic and conventions (Gray et al., 2020; Khan & MacEachen, 2022; Roberts et al., 2021; Thunberg & Arnell, 2022).

One conspicuous absence from the literature on online interviewing is a focus on co-interviewing, and in particular the interaction between and strategies of interviewers in co-interview settings. Although co-interviewing – when two researchers actively take part in the same interview – is a common technique, the literature on online co-interviewing is limited (Velardo & Elliott, 2021, p. 1). Nonetheless, the advantages of this type of interview, sometimes called the tandem interview (Groholt & Higley, 1972; Kincaid & Bright, 1957a, 1957b) or the ‘two-to-one interview’ (Monforte & Úbeda-Colomer, 2021) have long been recognized (see also Bechhofer et al., 1984).

Advantages of co-interviewing include the increased efficiency of data collection processes (Kincaid & Bright, 1957a), the application of interviewers’ varying perspectives to be brought to a single interview (Kincaid & Bright, 1957a, p. 306; Monforte & Úbeda-Colomer, 2021, p. 6; Rosenblatt, 2012, p. 103), better follow-up during the interview on unanticipated lines of inquiry (Kincaid & Bright, 1957b, p. 308), more gender sensitive and culturally situated research (Redman-MacLaren et al., 2014) and improved rapport between interviewers and respondents (Velardo & Elliott, 2021, p. 4), as well as making interviews more conversational (Bechhofer et al., 1984, p. 98). Inclusion of specific roles or persons, such as more senior researchers, may increase the willingness of respondents to participate in an interview (Redman-MacLaren et al., 2014). However, unequal power distribution between interviewers may lead to dominance by one interviewer (Velardo & Elliott, 2021, p. 5). Co-interviews also bring some logistical challenges, such as higher demands for coordination and preparation, which should be considered in the design and planning of an interview (Monforte & Úbeda-Colomer, 2021; Velardo & Elliott, 2021).

With the research on co-interviewing focusing squarely on in-person interviews, we argue that outlining strategies and tactics for co-interviewing in online settings is valuable for two key reasons. First, online settings enable researchers to collaborate regardless of their geographical location, making co-interviewing more feasible and, we anticipate, potentially more common. Thinking through how to conduct co-interviews online is thus an important step in developing the online interview methodology, and could help researchers to think more purposively about both how to work with their collaborators, and to prepare for interviews, beyond substantive content. Second, interpreting non-verbal communications can be more challenging in online settings. While the impact on interviewer-respondent dynamics is well-documented (Salmons, 2015; Seitz, 2016), this challenge also extends to interactions between interviewers in co-interview settings. Conducting co-interviews online therefore requires specific strategic and tactical approaches that take advantage of the opportunities while mitigating the challenges posed by this setting. Based on our own experiences conducting online co-interviews with elites and informed by online interview research and the limited in-person co-interviewing literature, we now outline our research study, before turning to explain our co-interview strategies and tactics.

Our Research Study

Our collaboration originates from a mutual interest in ceasefires and related themes, such as the use of technology in ceasefire monitoring. One of us wrote a doctoral dissertation on ceasefires (Sticher, 2021). The other had served as a senior official overseeing a ceasefire monitoring mission, and later pursued the theme of monitoring in a research context (Verjee, 2019, 2022). This shared subject interest led to the authors focusing on the largest dedicated ceasefire monitoring mission operating at the time our research began: the Organisation for Security and Cooperation in Europe Special Monitoring Mission (OSCE SMM) in Ukraine. Given its size and importance, and the fact that the mission was the first to deploy remote sensing technology on a large scale, the OSCE SMM was an obvious case of ceasefire monitoring to study. At its peak, the SMM comprised as many as 1400 personnel from nearly 50 countries (OSCE, 2021, p. 7).

Our main research question was when and how the use of remote sensing technology by the SMM improved ceasefire compliance. We used process tracing as our main research method, requiring us to collect a variety of evidence about steps taken by the mission, the conflict parties, and third-party actors to confirm or disconfirm our hypothesized mechanisms (Sticher & Verjee, 2023).

One of our key sources for such evidence was interviews (von Soest, 2023). Perhaps the most obvious approach would have been to conduct these interviews during field research in Ukraine. However, we were interested in broadening our pool of respondents beyond actively serving SMM personnel, for two main reasons. First, to answer our research question and get nuanced across-time variation, we wanted to cover the entire period in which remote sensing technology was used by the mission. This required us to talk to interviewees serving at various times during the mission, including those who had left the OSCE. Second, at the time our research began, the SMM was still operational, which made it more sensitive for serving personnel to provide information to outside researchers. We also anticipated that staff no longer actively serving would tend to be more open in sharing information and, whilst their recall of details might be incomplete, would be often able to provide a bird’s eye perspective and reflective context, which would be helpful for our type of research. In approaching both serving and formerly serving staff, we also were attentive to ethical concerns, such as ensuring that our interviewees participated voluntarily and with informed consent, and that we did not request interviews with local, Ukrainian staff, given their relative professional precarity and proximity to the conflict context. As co-researchers, we jointly agreed an understanding of research ethics to appropriately manage and safeguard respondent data and confidentiality across multiple sites, both physical and virtual, as was the case in our study. Our study was conducted in accordance with the research ethics procedures in force at the time at our respective institutions.

Given that the SMM was staffed by individual OSCE member states seconding personnel, our potential pool of research subjects was distributed worldwide. Given the breadth of secondment, our respondents could have resided in more than 40 countries in Europe, Asia, and North America (if they were in their home countries), and potentially elsewhere in the world should they have taken on new assignments elsewhere. This made it impractical to conduct interviews in person.

In the search for an alternative to in-person interviews, we recognized that the COVID-19 pandemic had accelerated adoption of online videoconferencing tools. The pandemic had normalized requests for online meetings and, with most international organizations moving meetings and communication online, even tech-averse senior officials had been exposed to platforms such as Zoom. Online platforms enabled us to conduct research that would otherwise not have been feasible, given the geographically dispersed nature of our interviewees. It allowed us to pair up as researchers driven by shared interests and complementary profiles, rather than choosing an interview partner based on location. At the same time, some of the challenges other researchers have faced were accentuated. We often had only one chance to talk to a specific individual, had limited time to gather relevant information from our respondents, and faced situations where it was difficult to use and interpret subtle nonverbal cues, as interviewers and interviewees stared straight into a camera and could not straightforwardly signal to an individual.

To ensure variation in terms of positions, seniority, gender, ¹ nationality, and length of service in the mission, we used cold contacts as a method of recruiting interviewees, thus often starting the conversation without a shared experience or introduction. As a result, we could not be sure of the next respondent’s attitudes and responsiveness. Some interviewees were forthcoming and only needed an invitation to set out their thoughts. However, this did not guarantee that we would gain the type of information that we sought, as interviewees sometimes meandered in their thoughts or provided basic or banal information about the mission. Other respondents appeared withdrawn, patronising, or reluctant to answer certain questions. In such circumstances, conventional attempts to build rapport by drawing on shared interests or mutual acquaintances were insufficient. In turn, interviewees had little knowledge of our specific approach, personalities, or professional experiences and background. Rather than seeing these as inhibiting factors, we sought to use our circumstances advantageously, taking a dynamic and sometimes playful approach to the interview.

In what follows, we describe the strategies and tactics we explored to steer and manage our interviews online in service of our overall research question.

Strategies

Knowing that in many cases, we only had one opportunity at speaking with any individual respondent, we developed several interview strategies premised on our co-interviewing approach. Although as academics of a similar age both based in universities, we had some socioeconomic and professional attributes in common, we also had different characteristics and backgrounds. Our commonalities, however, did not preclude the use of quite different interview strategies by different interviewers in a co-interviewing approach, sometimes even in the same interview. Regardless of whether we perceived commonalities among ourselves, however, we were likely perceived differently by our interviewees, because of our different mannerisms, as well as our identity characteristics of nationality, gender, ethnicity, and even accents, among other features (Hampshire et al., 2014).

We here set out our strategies as a framework for other researchers, noting that individual researchers may feel more comfortable with some approaches than with others, and that the appropriateness and efficacy of any strategy may vary, depending on the circumstances of a study and the nature of the respondent pool. At the same time, we seek to develop the craft of interviewing, rather than suggest that every interviewer need develop strategies anew. An awareness of different approaches can help the researcher maintain flexibility during an interview and adapt as necessary, rather than requiring a premeditated view on what will work in any specific interview.

Strategy 1: Insider-Outsider

The insider-outsider distinction relates to the perceived identity of the researchers from the perspective of the interviewees. In our case, the main insider-outsider dimension was between being peace practitioners and academic researchers in peace and conflict studies—roles that share an interest in understanding conflicts and their resolution, but pursue quite different activities, goals, and priorities. We anticipated that respondents would mostly treat us as outsiders, as we assumed the role of academic researchers in our interview requests and were not visibly practice orientated. As others have recounted, being an outsider in a qualitative research interview is not necessarily problematic (Lawrence, 2020, p. 156; Maksutova & Janoff, 2022; Martinez-Merino et al., 2023), but in our experience, some respondents who perceived us as outsiders would assume that we were essentially ignorant of basic dynamics and constraints of the OSCE’s SMM operation, or would be unable to understand more subtle details of their experience. Accordingly, some responses were quite general, defensive, or overly simplified.

At the same time, we agree with those scholars that reject conceptions of insider or outsider as clearly binary (Aguinis & Solarino, 2019; Dhillon & Thomas, 2019; Hampshire et al., 2014; Herod, 1999; Markiewicz, 2023; Mikecz, 2012); we agree that researchers can and do position themselves along a continuum of insider-outsider perspectives. In our study, establishing an insider positionality was largely based on our prior experience as practitioners, having worked for international organizations, and, in one of our cases, for a different OSCE programme in Ukraine. Insider interviewing, while requiring negotiation and reflexivity, has long been established as an effective interview strategy (Court & Abbas, 2013; Salmons, 2015). Since we were two interviewers, however, it was not necessary that both of us consistently assumed or attempted to adopt insider positionalities to elicit information or steer the interview. Instead, being co-interviewers allowed us to assume insider or outsider roles more purposively and individually, as the situation befit, and even switch roles on occasion in the same interview. At the outset, we tended to emphasize our practitioner and research experience to different degrees. If we felt that our interviewee perceived our positionality as academics as inhibiting information sharing, we brought up relevant practitioner experiences, moving dynamically alongside the insider-outsider spectrum.

While we identified the practitioner-academic distinction and the primary insider-outsider dimension for our study, it is important to note that other dimensions are likely more relevant in other research contexts. Certain researcher characteristics, like gender, are often viewed as fixed. Other attributes, such as cultural or socio-economic background, are certainly more fluid, and perhaps less obvious online. However, what an online co-interview provides is additional possibility for researchers to emphasize, or-de-emphasize, the relevance of these characteristics to varying degrees. For example, they may highlight their personal experiences related to being a woman, a man, or having a non-binary gender identity. The point we would make is that co-interviewers can anticipate such possibilities in advance, and plan their interview approach accordingly, rather than only be reactive, while recognizing that there will still be a need to adapt to the specific context of the interview and responsiveness and attitude of the interviewee (Hampshire et al., 2014; Hathaway et al., 2020, p. 110; Raheim et al., 2016).

Strategy 2: Good Cop, Bad Cop

In this strategy, we explicitly recognized that the power differential between researchers and respondents could lead to a frustrating interview. In line with other researchers’ experiences, we anticipated that interviewees might not fully engage with our questions, adhere to well-rehearsed lines, refuse to accept the premise of our questions, demonstrate limited reflexivity in their responses, or demonstrate a combination of some, or all, of these behaviours. In our study, a common example of this arose when a respondent was asked, if, with the benefit of hindsight, anything in the SMM could have been done better, and the response was that everything was done as well as it could have been. While this in the abstract could well be a legitimate response, it was our impression that at least some elite respondents were reluctant to be self-critical, or critical at all, even constructively, of the organization for which they had worked. This belief was often substantiated when the respondent was prompted further, and modified or made conditional their earlier answer.

Consequently, and particularly where there was a hint of an interesting insight that had yet to be fully developed in a respondent’s reply, we sometimes employed an approach we call good cop, bad cop (see Abbe & Brandon, 2014; Hathaway & Atkinson, 2003). The good cop interviewer would adopt a more sympathetic persona; the bad cop interviewer would be more openly critical of the interviewee’s responses (Berreman, 1962; Hathaway & Atkinson, 2003). Hathaway and Atkinson (2003, p. 164) argued that social science researchers can learn from the techniques of other professionals, such as journalists and the police, and suggested that the strategy is suited to a sequential approach, where the researcher is first more friendly, but becomes more skeptical or critical over the course of the interview. Kvale (2006, p. 487) posits that for the interviewer to provoke conflict, identify contradictions or weaknesses in assertions by respondents, and emphasize divergence, can be productive. We argue that this approach is particularly effective when employed in a co-interview setting, and that the clear sequence of critical escalation identified by Hathaway and Atkinson (2003) becomes less necessary when two interviewers are present. An explicitly challenging approach by a solo researcher might have risks: ultimately, even a respondent who is prepared to engage may withdraw if they perceive critiques to be too strident. Such risks are moderated when the other interviewer is, in comparison, seen to be much less abrasive, or indeed sympathetic to a respondent (Kincaid & Bright, 1957a, p. 307). Using the good cop, bad cop strategy online also mediates the more negative aspects of nonverbal communication that might be perceived if such an interaction was occurring in person, particularly with respect to the bad cop. As with our first strategy, this approach is dynamic, and roles can be switched during an interview. Playing bad cop does not imply any aggressive or abusive behavior, but rather asking more challenging and probing questions, repeating questions more stridently if initial answers were evasive or openly disagreeing with an interviewee’s responses if they appear to be dishonest or disingenuous. Elites tend to be accustomed to these types of challenges (Hathaway & Atkinson, 2003, p. 179; Ntienjom Mbohou & Tomkinson, 2022), facilitating this strategy. Conversely, playing the good cop role does not imply accommodating all views of the respondent or accepting statements at face value. It does, however, require a positive demeanour, which manifests in two ways. First, directly toward the respondent: the good cop interviewer creates a welcoming atmosphere by asking open-ended questions and providing affirmative signals, including through nodding or empathetic language. Second, in interacting with the co-interviewer: the good cop may periodically soften tough questions or elaborate with examples and qualifications, signalling to the respondent that a broad spectrum of responses is welcome. We point to Bechhofer et al.’s (1984, p. 97) experience, noting that a debate between interviewers can also be a productive way of engaging a respondent to react. In our view, the online setting provides the means for the co-construction of such an interview, while still providing safeguards and assurances to the respondent.

Strategy 3: Knowledge Ambiguity to Elicit Information

Since neither of us had worked in the SMM and were not personally known to most of our respondents, our interviewees did not know what we knew about the mission and its evolution over time. We did not want to disclose too much of our own knowledge at the outset in case this risked pre-empting lines of response. We wanted to ensure that respondents had the chance to present their own views of what they deemed important. We noticed that some respondents enjoyed sharing information they believed was new to us, as they wanted to demonstrate their competence or help us understand the conflict. Some, however, would dwell on basic facts and focus on recounting well known descriptive details such as the SMM’s mandate and role, the composition of the OSCE, or superficial information about Ukraine. This limited the time available for us to delve more deeply into their experiences of the SMM, or probe beyond what was largely public information. Yet other respondents appeared reluctant to share more profound insights if they felt the researchers were not competent in the basics.

Again, we adopted a flexible approach to mitigate this challenge. We showed interest without demonstrating too much knowledge early in the interview, and started exhibiting knowledge if the responses were too basic or general. However, with fewer non-verbal cues online, we could not always be sure if our signalling of information was effective in such cases. To elicit information, one way we demonstrated prior knowledge was to explicitly mention or repeat insights about the conflict or the mission, sometimes explaining that other interviewees had raised these points. For example, in asking what happened to information monitors gathered, we might first ask a general question about reporting procedures. If the discussion remained overly general, we might eventually ask how the regional hubs or national headquarters processed reporting information, mentioning the precise names of the relevant organizational units and senior officials. Providing this detail gradually, rather than at the outset, tended to prompt more detailed reflections. Acting as collaborative co-interviewers allowed us to reinforce each other’s points. We would then ask how respondents related to these insights. This gave us the opportunity to steer the conversation and ensure greater depth to responses, while maintaining an open attitude to respondents’ views on a particular issue.

Socratic irony can also be applied to an interview context. As McDowell has pointed out (1998, p. 2138), shifting an interviewer’s perceived competence can be effective: an interviewer might understate or downplay their knowledge with older patriarchal figures, for example, while showing that they are better informed with other respondent populations. What can enable this strategy online is the ability to have a dynamic exchange of information between co-interviewers without alerting the respondent. This we could do through the online messaging system of the interview platform, or via a parallel communication means (usually WhatsApp messages). To preserve the integrity of (and focus on) the interviewer-to-interviewee dynamic, this approach requires coordination between interviewers. Co-interviewers should have a good understanding of each other’s knowledge, and judge when to demonstrate different degrees of competence during the interview, using parallel messaging only sparingly to deal with remaining challenges.

Adapting Strategies and Preconditions

The three strategies are dynamic and not mutually exclusive; they can be employed in combination or consecutively, with the emphasis and application adapted to the interview dynamics. In our case, in some interviews, the flow was natural, while others posed more challenges, necessitating a focused application of a particular strategy or adapting along a spectrum throughout the interview. While there are no definitive triggers for selecting a specific strategy, certain challenges indicate the need for a shift in gears. For instance, seemingly rehearsed responses to questions may call for the ‘good cop, bad cop’ approach. Conversely, overly general, or basic answers may warrant a progression along the knowledge ambiguity spectrum. Effectively implementing the three strategies requires practice, preparation, flexibility during the interview, and debriefing for future learning. Awareness of the strategies can guide researchers through all these steps, enhancing their reflections and deliberate choices. This is particularly important when co-interviewers work online, where decisions often require more deliberate consideration than in in-person settings.

Besides preparation and awareness, there are conditions to effectively employ each of the strategies. Clearly, interviewers need the relevant background to assume a (partial) insider role, as they cannot simply manufacture experience, which might limit the application of this strategy to all qualitative researchers. However, in co-interviews, researchers who lack a specific personal or professional experience can deliberately pair up with researchers who complement their own background, taking advantage of the fact that online settings no longer require a shared locality. Regarding the ‘good cop, bad cop’ strategy, not all researchers may feel comfortable adopting a bad cop persona. Some co-interviewers may be better prepared to assume or take on that role. Alternatively, understanding that this does not involve abusive behaviour, but rather a more challenging and probing attitude that draws a contrast with the co-interviewer, and that the two-cop personae can be along a spectrum rather than fixed and extreme positions, may help researchers make use of this strategy. Regarding knowledge ambiguity, researchers can employ this strategy more effectively if they are well prepared for the content of the interview, and if they have complementary expertise. Researchers may also find that they can employ this strategy more effectively over time, as they accumulate knowledge and understanding of different perspectives once they have conducted multiple interviews on a specific subject.

Tactics during an Online Interview

While strategies refer to overarching approaches to an interview, tactics are facilitative tools conducted in service of these strategies. Online co-interviews are particularly challenging as they require researchers to manage the complexities of the triadic interviewer-interviewer-interviewee relationship in a setting where non-verbal cues are difficult to send and interpret. In this section, we explore how in-interview signalling can facilitate interviewer-to-interviewer interactions, how the tactical interruption can support interviewer-to-interviewee relations, and how deliberate turn taking takes advantage of the triadic setting to manage the conversation online.

Tactic 1: In-Interview Signalling

In co-interviews, it is crucial that interactions between interviewers are effective in establishing the foundation for a successful interview. The importance of nonverbal cues in interviews is well established (Hathaway & Atkinson, 2003, p. 180), and extends to relations between interviewers. In a physical setting, through subtle gestures and expressions, interviewers can signal to each other when they want to interrupt, if they want the other to continue their thought, or even more complex messages such as which part of a question may require a follow-up. In the context of online interviews, however, these nonverbal cues are much more difficult to send and interpret, as all participants stare straight into the camera and interviewers cannot address a signal specifically to the co-interviewer.

Occasionally, we used written communication on a different software system, such as an instant messaging platform, to check if the other interviewer had any pressing issue, or to indicate that we should move forward. However, to keep the focus on the interview, we sought to routinize in-interview signalling as a means to manage interviewer interactions. For example, we would toggle the mute function on the videoconferencing software to signal when we wanted to interject, even if the respondent was still speaking. We knew that the other interviewer would use this as a signal to ask the next question and came to look out for this signal between ourselves. We would often ask the other interviewer prompting questions in front of the interviewee, acknowledging that the online setting required more explicit and verbal coordination between researchers, while also showing to the respondent our active collaboration.

Tactic 2: The Tactical Interruption

Like communication between interviewers, nonverbal cues also serve as an important communication tool between interviewers and interviewees. Interviewers can signal to the interviewee that they should continue their thought, or that the response is sufficient. Interviewees can actively look for such signals or send their own signals to check if they are on track. Even in physical settings, this can be difficult, and some interviewers and interviewees are more effective in using nonverbal cues than others. In online settings however, these challenges are accentuated, and there is a risk that an interview might lose focus if the interviewee seeks to set the agenda or dominate and the conversation shifts away from the main focus. While we did sometimes use subtle ways of signalling—such as using the unmute button discussed above—we would also at times interrupt more assertively, by simply interrupting while the respondent was talking. We used this when interviewees decided to ignore more subtle signs, and especially in situations where they discussed something tangential to our research interests in overabundant detail. Here, we drew on the social expectation that subtle interruption in video-based settings is more challenging, as mitigation to interject in a way that may have been seen as impolite or even hostile in an in-person, face to face setting.

Tactic 3: Turn Taking

Finally, and implied in the descriptions above, we used turn taking between co-interviewers as a tactic to steer interviews. While there was no hierarchy between researchers and, while we both collected evidence for all steps of the process tracing, our interests and questions varied, leading to a natural role division. Before each interview, we would decide on a rough order of questions, and who would take the lead for that interview. The selection of questions and order in which these were posed was determined by the function and seniority of the interviewee, as respondents had different responsibilities and different exposures to actors whose activities were of interest to us. Because we used informal Bayesian updating (Beach & Pedersen, 2016, pp. 154–226; Fairfield & Charman, 2017, p. 364) in our evidence collection, we prioritized questions about activities and causality to which we had less information relative to others.

Turn taking did not imply strict alternation between interviewers for every question or equality in the number of questions asked by each person in each interview. One person would often ask a few questions and follow up where necessary, with the other taking a backseat if the interview was seeming to be insightful on a specific issue. However, if either of us felt that the conversation was not moving forward, or that the interviewee stuck to well-rehearsed lines or meandered in their thoughts, we would indicate that we would like to move on to ask the next question (see in-interview signalling). Alternatively, the active questioner would explicitly pass the floor to the other interviewer. This often allowed us to bring the conversation back in focus, and shift from the interview’s impasse. It also provided for more active participation by everyone, compared to a situation where only one person is asking questions, or where the entire sequence of question-asking is predetermined and rigid.

Conclusion

This article showcases how online co-interviews open the possibility for research encounters that might otherwise be unfeasible. Well beyond our own field of research, online co-interviewing offers a method for researchers attempting to access geographically dispersed respondents, reduce travel to lower carbon footprints, or with financial and other travel-related constraints. An online setting widens the opportunities for researchers to pair strategically, enabling them to ensure complementarity and diverse positionality, regardless of physical location. We propose strategies and tactics that make use of such paired constellations, showing how they serve to address common challenges in elite research interviews more generally, and in online settings more specifically.

We used these approaches dynamically, adapting to the interview situation as we encountered different challenges. Some might argue that our strategies risk being deceptive. We would argue that, while we were deliberate about what information we revealed about ourselves and how we positioned ourselves vis-à-vis the respondent, we did not deceive respondents, but, rather, recognized that co-construction of the interview encounter required planning and ongoing reflexivity by the researchers (Karnieli-Miller et al., 2009). Another possible criticism against our strategies is that they have the potential to manipulate respondents, increasing the likelihood of cognitive biases (such as confirmation bias) to affect the research outcome (Morris, 2009). Indeed, researchers may deliberately or unintentionally act to elicit responses that align with their preconceptions (Beach & Pedersen, 2016, pp. 154–226; Fairfield & Charman, 2017, p. 364). Researchers must be mindful of this possibility. However, co-interviewing can mitigate such risks: researchers with different positionalities and approaches working in the same research encounter are better able to bring their different perspectives to an interview, intervene if an exchange becomes trapped in confirmation bias, and help reflect more critically on the interview with colleagues once it has concluded.

In our discussion of online co-interviewing strategies and tactics, we primarily focus on elite interviews. However, while some challenges may be more prominent in elite interviews, navigating and managing interviews effectively is a widespread challenge in online co-interviews of all types of respondent. We anticipate that such difficulties would vary depending on factors such as the topic or the personality of the interviewee. Some challenges, including the perception that questions are sensitive, withholding of socially undesirable responses, provision of unspecific answers, or deviation from the topic, can occur with a wide variety of respondents. Many of the approaches we outline can be directly applied to these situations. Our strategies require adaptation to different research contexts, but suggest an outline for how research interviews can be constructed, beyond the content based approach of the specific questions that are asked.

Nevertheless, there are certain factors that condition the overall suitability of our approach. Our strategies and tactics are suitable for semi-structured and unstructured interviews, but not for structured interviews where questions are fixed and there is no space to adapt to the attitude, personality, and characteristics of the interviewee. The suitability of the approach also depends on the type of respondent. While our tactics are generally applicable beyond elite interviews, some strategies—such as the ‘good cop, bad cop’ strategy—may be less suitable for non-elite interviews, due to considerably different power dynamics between researchers and respondents. Our approaches are particularly suitable when a conversation is likely to be difficult to manage, where respondents are reluctant or resistant, or where interviewees’ agendas do not readily correspond to those of the interviewer or their research questions (Nickerson, 1998). While our strategies and tactics are suitable for positivist research endeavours, where researchers seek to collect evidence confirming or contradicting theoretical expectations, they are also relevant to interpretivist approaches to research. Clearly, researchers must carefully assess how the use of each strategy may affect the way in which interviewees position themselves and respond. There are, potentially, trade-offs in terms of practicalities, speed of information gathering, and the comprehensiveness of information gathering.

Evidently, technological advancement and widespread use of videoconferencing tools during the COVID-19 pandemic has opened new possibilities for conducting interviews online. Our argument is that online interviews can both be a method of first choice, and that online co-interviews offer advantages as a distinct method of data gathering. While challenges remain, explicit strategies for conducting such interviews can make collaborative research co-production both more meaningful and more effective.