Serial Interviews

Abstract

Many qualitative social scientists conduct single-session interviews with large numbers of individuals so as to maximize the sample size and obtain a wide range of study participants. Yet in some circumstances, one-shot interviews cannot produce information of adequate quality, quantity, and validity. This article explains the several conditions that call for an alternative approach, serial interviewing, that entails interviewing participants on multiple occasions. This method is appropriate when studying complex or ill-defined issues, when interviews are subject to time constraints, when exploring change or variation over time, when participants are reluctant to share valid information, and when working with critical informants. A further benefit is the opportunity it provides for verifying and cross-checking information. This article delineates the general features of this technique. Through a series of encounters, the researcher builds familiarity and trust, probes a range of key topics from multiple angles, explores different facets of participants’ experiences, and learns from events that happen to take place during the interviews. This helps overcome biases associated with one-off interviews, including a tendency toward safe, simple answers in which participants flatten complexity, downplay sociopolitical conflict, and put themselves in a flattering light. This article illustrates the utility of this approach through examples drawn from published work and through a running illustration based on the author’s research on elected neighborhood leaders in Taipei. Serial interviewing helped produce relatively accurate and nuanced data concerning the power these leaders wield and their multiple roles as intermediaries between state and society.

What Is Already Known?

Serial interviews offer a means to obtain rich and deep accounts of study participants’ life histories and changing perceptions.

What This Paper Adds?

It explains why even social scientists who have reasons to prefer onetime interviews and who are not epistemologically committed to collecting ethnographically immersive, panoramic accounts of participants’ lives might nonetheless wish to conduct serial interviews. It lays out a comprehensive set of conditions calling for this technique, with special reference to political and social research. And it demonstrates an application of this technique to understanding the power relationships of local state–society intermediaries in Taiwan’s cities.

Introduction

Interviewing is a pervasively common research technique in the social sciences. In one recent survey, political scientists reported using interviews more than any other fieldwork technique (Kapiszewski, MacLean, & Read, 2015, p. 190). ¹ But how much time should we spend with each inter-viewee? In particular, when might it be worthwhile to hold multiple interviews with the same individual?

Many researchers expect to conduct a single interview, perhaps 45 min to 2 hr in duration, with each participant in a study. This maximizes the number of individuals who can be interviewed. Typically, such one-and-done interviews are based on a set of fundamental assumptions: The desired information exists and is relatively straightforward to conceptualize and understand; the research participants know this information; and given the right assurances and motivation, they can and will convey it to the researcher in one fairly expeditious interview session. ² These assumptions underpin Beckmann and Hall’s (2013) thoughtful prescriptions for interviewing Washington elites, for instance. In one core example, a central purpose of their brisk, single-session interviews is to obtain very specific behavioral data, such as the number of times Senate aides contacted executive branch officials (the president, the chief of staff, the commerce department, etc.) about a particular piece of legislation (pp. 196–208, 263–265).

In many research projects, however, these assumptions do not hold. Sometimes, the “data” do not take the form of straightforward facts. The questions may concern topics that the inter-viewees themselves have never consciously considered or require exploring a large set of past experiences that must be retrieved from memory episode by episode. Alternatively or additionally, inter-viewees often have reason to balk at disclosing information and may even give partial or misleading answers, especially at first. Perhaps not everything the researcher wants to learn can be covered in a single session.

This article discusses serial interviewing as a strategy for coping with such problems. In this approach, the researcher interviews participants on multiple occasions. The additional time investment this entails can reap numerous rewards, each involving data that are more valid and more extensive than what one-off interviews may produce. Through a series of encounters, the researcher builds familiarity and trust, so that—ideally—inter-viewees become comfortable sharing information they otherwise would not. Key topics are probed from multiple angles and on more than one occasion. Over time, it thus becomes possible to explore different facets of participants’ past experiences, to cross-check among pieces of information received at different sittings, and to leverage insights that can only be obtained at particular times—for instance, during specific parts of an electoral cycle or when the inter-viewee confronts particular problems. Finally, multiple sessions provide greater opportunities to observe and learn from participants’ incidental interactions with their physical environments and with other people.

This article begins with a discussion of my own research on local politics in Taiwan, showing why serial interviews were useful for answering particular kinds of questions. It then briefly reviews published works about the technique and explains the rationale for this article: to specify all the reasons why even skeptical scholars might choose this approach. The following section lays out six circumstances calling for serial interviews. After that, I expand on how to apply this method by explaining four ways in which serial interviews differ from onetime interviews. Along the way, I point to examples from other political scientists’ research on topics from ethnic violence to coups to demonstrate the power and wide utility of this technique. ³ In places, I present aspects of my Taiwan research to illustrate particular points.

An Example: Serial Interviewing in the Neighborhoods of Taipei

A research project I began in 2006 has examined a category of community leaders who act as intermediaries between state and society (Read, 2012, 2018). Specifically, these leaders are local neighborhood heads whom citizens choose in competitive elections and whose positions the government defines and funds. (They are called lizhang, a term I translate as “neighborhood warden.”) ⁴ As elected representatives, they speak on behalf of the roughly 6,000 people who live in each of their officially defined constituencies, known as li. But they are also, in part, extensions of one of East Asia’s famously strong states. By law they serve as grassroots-level agents of the urban government, technically obliged to accept the “command and supervision” of district officials. They work in state-equipped offices, side-by-side with a civil servant (liganshi) who implements the programs of city agencies. The wardens are involved in many kinds of political, social, and economic processes, from negotiations over urban redevelopment plans to community policing to volunteer groups, festivals, and recreation. ⁵

Their formal duties are set forth in national law and city guidelines. Yet such documents scarcely begin to answer the essential question of what kinds of power these leaders actually wield in practice, across various circumstances and in dealings with innumerable constituents and other actors. Each of Taipei’s 456 neighborhoods contains a complex landscape of political stakeholders. Many wardens emerge from extended family networks with property and historical roots dating back several generations. Other players in neighborhood politics include local businesses, the parent groups connected with public schools, housing developments, and associations of many stripes, from neighborhood watch patrols to historical preservation nongovernmental organizations. Furthermore, Taiwan’s two major political parties vie to cultivate supportive wardens. The Kuomintang (KMT), which wielded authoritarian control over the island from 1945 until democratization in the late 1980s, once had a lock on the neighborhood heads. Today, a growing number of wardens have no partisan affiliation or favor the KMT’s rival, the Democratic Progressive Party (DPP). Those with party allegiances help get out the vote in elections for city councils, the national legislature, and the presidency.

Several analytical questions—the nature of grassroots democracy, the prevalence of vote-buying, the ability of the state to implement developmental and administrative programs, and the adaptability of this once autocratic system to pluralist politics—hinge on the seemingly simple descriptive question of what kinds of power these leaders exercise. Political scientists and sociologists ask similar questions about power holders all over the world. Do local leaders advocate for the interests of ordinary people? Or is their purpose really to capture constituencies for the state or dominant parties, colonizing space and civic energy at the community level that might otherwise be used by grassroots movements? Are they clientelistic brokers who deliver political support to patrons at higher levels? Such questions are found in many lines of research (examples include Auyero, 2001; Collier, 1976; Kasza, 1995; Kitschelt & Wilkinson, 2007; Stokes, Dunning, Nazareno, & Brusco, 2013; Walder, 1986).

Answering these questions is challenging. It is not difficult to meet a neighborhood warden, and basic data about their backgrounds are accessible online. Yet obtaining a precise understanding of the power they actually wield required sifting through the possibly biased, self-serving accounts that they themselves provided; assessing clues embodied in a vast array of commonplace events and interactions; and interpreting relationships between the wardens and countless constituents, officials, and other urban stakeholders. Doing all this necessitated serial interviews.

The research was initially approved by the institutional review board (IRB) at the University of Iowa; after I later moved to the University of California, Santa Cruz, its IRB approved my ongoing follow-up fieldwork. In both cases, the protocol called for obtaining informed consent from participants by sharing with them a consent form in both Chinese and English describing the research; explaining their rights; and discussing risks, benefits, and confidentiality. Both IRBs agreed that participants could provide informed consent orally rather than with written signatures. In this article and other publications, pseudonyms are used for wardens and their neighborhoods. My interviews have taken place over the course of about 6 months of fieldwork, spread across eight separate trips. I have studied 13 different neighborhoods in Taipei but with differing degrees of depth. I have made 20 visits to one neighborhood, between 2 and 15 visits each to nine others, and a single visit to three others. ⁶

To begin an open-ended program of interviews with a new participant, I start by visiting a warden in her office. These spaces—often located in the warden’s home or small business—make excellent settings for interviews. The wardens’ duties are such that they usually spend much of the day in these headquarters, which are arranged and furnished expressly for hosting visitors. Sometimes they chat with friends and supporters or discuss neighborhood matters with the liganshi. As a researcher, it is possible to fit oneself into the warden’s environment, if not unobtrusively, at least without necessarily imposing an extraordinary burden. Questions can be fit around the participant’s other activities and the ambient office conversation.

After the first encounter, I might call a week or two later and ask if I could drop by again. This has led to an arrangement where I would come by on occasion for more tea, conversation, and questions in the office. Most participants have accepted that the interview is not a simple matter of running through a set list of questions but a process of exploring their everyday work and accumulated experiences. A few have declined to take part or indicated that they wanted no more than one or two interviews. While taking every opportunity to prolong my visits, I have left or stayed away when my presence would create awkwardness, such as when a warden has personal business to handle.

Divergent Epistemologies and Competing Imperatives

The large social science literature on interviewing contains a number of discussions of serial interviewing and related techniques. Health researchers, for instance, have done studies in which cancer patients or their families are interviewed on multiple occasions (Grinyer & Thomas, 2012; Murray et al., 2009). In the “ethnographic interview” approach of anthropologist James P. Spradley (1979), questions are posed over a series of at least six 1-hr sessions in order to understand an informant’s way of life (pp. 51, 55–68). For research on fears about crime, Hollway and Jefferson (2000) developed a two-interview “free-association narrative” protocol (pp. 43–44). In this and other narrative research, multiple interviews may be necessary to explore details of the stories that are the very essence of what the investigator seeks (Earthy & Cronin, 2008; Riessman, 2012; Shenhav, 2005; Wengraf, 2001). Related methods such as life story interviews or oral history interviews also involve listening to participants’ accounts at length, likely in more than one sitting (Atkinson, 2012; Charlton, Myers, & Sharpless, 2008; Giles-Vernick, 2006).

These contributions bring out the important point that multiple interviews can build deeper and more trusting relationships with participants and plumb their lives and experiences extensively over time. Valuable as they are, they do not fully express the utility that serial interviews can have to a wide range of researchers. It may not be immediately evident to, say, comparative politics or environmental studies scholars how they might apply techniques used in studying families of cancer patients or crime victims. Further, most if not all existing accounts speak to researchers within particular epistemological perspectives, notably those that privilege ethnographic immersion, extensive narratives, and/or close relationships with participants as inherently desirable features of research. They would not necessarily convince all researchers, including some with a more positivist orientation, or those who simply wish to devote no more time than necessary to each inter-viewee. Such researchers might pursue serial interviews in their own distinctive ways—for instance, by focusing the conversation on topics central to the research question, rather than giving inter-viewees wide latitude to recount their life stories in their own terms, as oral historians might.

Even in projects built around onetime interviews, it is probably common to revisit a few participants for follow-ups. ⁷ Nonetheless, when researchers take single-session interviews as the norm, they accept a constraint that has many implications. Often, they quickly pare away layers of the participant’s belief system or accumulated experience to focus on just one specific topic or event. ⁸ Similarly, they may adopt a “just the facts, please” style, asking narrowly framed questions intended to pin the inter-viewee down on essential points (Beckmann & Hall, 2013, pp. 202–206). Whole subjects may be avoided, or treated in broad-brush fashion, for lack of time. This is entirely appropriate in some projects, but in others, this constraint should be questioned from the start.

Why One Might Interview the Same Participant More Than Once

Even researchers accustomed to in-depth, qualitative inquiry may not embrace the idea of holding multiple sessions with each inter-viewee. Yet depending on the research question, the study participants, and the circumstances, serial interviewing can be suitable and even necessary. Consider the following when deciding whether to make the time investment that serial interviews represent.

Cognitive and Memory Demands Stemming From Complexity

A particularly compelling reason to employ serial interviewing concerns the complexity of the information sought. Complexity holds, for instance:

When there is not just a single topic of interest (i.e., dependent variable) that can be assessed with a few simple questions, but multiple topics of interest or dimensions of a single topic. For example, a mayor’s dealings not merely with her citizen constituents but also with political parties, interest groups, and city officials.

When these topics of interest are not cut-and-dried, straightforward things (e.g., the number of meetings a given group of officials held), but rather contain layers of subtlety and ambiguity. Examples of the latter include participants’ identities, power relations, interpersonal connections, beliefs, and meanings.

When the interviewer seeks to learn about not a single, discrete event but a larger category of past events that might include dozens or hundreds of experiences. Examples might be instances in which a citizen wrote to, visited, or called a local government office, or occasions when a politician attended fund-raisers.

The presence of any of these conditions has profound implications for an interview. It means in particular that it is folly to expect to obtain full “answers” all at once. Instead, a faithful and well-rounded picture of the answer will emerge only over time. It will probably not come in a succinct, coherent statement by the inter-viewee—the cognitive and memory demands for which would likely be forbidding—but rather in an assortment of reflections and vignettes that the researcher must piece together like a mosaic. The result will likely contain some degree of inconsistency or even paradox. Complexity means that an inter-viewee may well express one aspect of his beliefs and experiences today and then give different statements, perhaps seemingly contradictory ones, weeks or months later.

Particularly when dealing with complicated, unfamiliar topics—some involving technical intricacies of law, policy, or technology—we often begin our research ignorant. We do not fully understand the empirical or conceptual lay of the land; we do not yet know the right questions to ask, and initial interviews spend time establishing elementary background. But later, the researcher has refined her questions and can probe exactly what is most important. Serial interviews make it possible to benefit from the researcher’s own gradually improving grasp of complex subjects, by returning to ask participants the questions she “should have asked” in earlier rounds.

Time Constraints

Complex research topics require more time, yet any one interview session is usually time-constrained to some degree. Inter-viewees who have positions involving power and responsibility likely have tight schedules. Often, it is simply infeasible to give serious treatment to a complex set of issues in a single session. This is especially true when allowing time for getting to know each other and for permitting the inter-viewee to present her own perspective in an open-ended fashion.

Variation Over Time

One-off interviews capture information from a participant at a single point in time. That is not always sufficient; it may be worthwhile or necessary to employ serial interviews in order to understand longitudinal change, that is, variation over time. This broad concept can be disaggregated into two types of change, each a distinct rationale for repeated interviews.

In some projects, a core purpose of serial interviews is to understand how research participants themselves are changing over time. For instance, the health researchers cited above interviewed lung cancer patients on multiple occasions as their disease progressed, precisely in order to understand their changing experiences, perceptions, and needs. Thus, “some participants went from initial enthusiasm about having chemotherapy to regret, and others from refusal to deep appreciation of hospice care in later interviews” (Murray et al., 2009). In political or sociological studies, the reason might be to understand how newly elected legislators change as they acquire experience or how social movement activists evolve over successive campaigns.

Over-time change can mean, second, not that the inter-viewees themselves change but that different circumstances bring to light different aspects of their experiences or perceptions. For example, in research that was as much observational as interview-based, Fenno, Jr. (1978) took advantage of opportunities to see congressional representatives interacting with and reflecting on different constituent groups in their home districts, from core supporters to hostile audiences. The phases of the electoral cycle may bring out contrasting facets of politicians and their staff. A city council member in a tight reelection race may well say different things a few months later after the pressure fades. What bureaucrats say in interviews may be shaped by problems they are currently facing or reports they are writing. In an even more immediate way, unique events that happen to take place during the interview itself can provide specific referents on which to focus what would otherwise be abstract questions. For instance, street-level functionaries in Beijing might give bland answers when asked how they mediate community disputes. But if a breathless constituent happens to burst in during the interview and demand help resolving a squabble with his brothers over ownership of their family’s nearby home, this provides a concrete case to explore (Read and Michelson, 2008, p. 744). Thus, the same kinds of questions asked at different times can trigger responses that vary in illuminating ways.

Building Trust to Counter Bias and Concealment

Accounts of interviewing often do not sufficiently problematize the participant’s candor—his or her willingness to provide valid information. Typically, the assumption is that the inter-viewee will likely share what he or she knows on the topic at hand, given that guarantees of anonymity are extended, that the right tone is set in warm-up conversation, that questions are worded appropriately, and so forth. Alternatively, it may be taken for granted that the inter-viewee will simply not give useful answers on certain topics. Neither perspective accounts for the common circumstance that participants may be tight-lipped or disingenuous at first (uncertain, for instance, that the researcher is who she says she is; unsure that what they say will stay confidential; unwilling to give answers that reflect negatively on themselves) but open up more over time. In such circumstances, single-session interviews would naively generate biased results.

In general, inter-viewees can be expected to become more trusting and confiding in successive interview sessions. ⁹ In her study of the Rwandan genocide, Fujii (2009) carried out 231 interviews with 82 different people, nearly 3 interviews with each participant on average (pp. 26–28). In presenting examples from these encounters, she often notes which session it came from because earlier and later interviews with the same individual differed substantially. She comments: “After talking to some women multiple times, for example, the interviews ceased being structured, formal enterprises and took on the quality of friendly conversations among a group of women friends or acquaintances” (p. 34). Establishing such familiarity clearly mattered a great deal for research on this highly sensitive topic. In a later account, Fujii (2018) writes that returning repeatedly to certain “key participants” enabled her to “move past superficial accounts of people’s lives before and during the genocide, and glean stories that were more detailed and substantive” (p. 45).

Repeated interview sessions are not guaranteed to build trust in all cases, and obviously, there are some secrets that participants might not divulge to anyone, let alone to a researcher. Nonetheless, as the above examples show, the experiences of many social scientists demonstrate the value of repeated contacts for increasing candor and decreasing bias.

Critical Informants

Serial interviewing is particularly appropriate for projects that involve what we might call critical informants—that is, participants possessing information of special value, who are particularly or uniquely well positioned to shed light on the topics of interest. The more critical the informant, the more it may be worthwhile to invest time in gaining trust and thoroughly exploring her experiences. Singh’s (2014) research on military coups in Ghana and elsewhere provides one example. Many of the 275 hr of interviews he conducted were with critical informants: retired officers who had personally seen or taken part in coups and coup attempts. Singh generally interviewed each one 3 times for 2 hr per session. The first session sorted out background information and “let us establish a relationship”; this then allowed for deeper questioning on coup processes in subsequent encounters (personal communication with Naunihal Singh, April 13, 2015). To some extent, the complexity of the information at issue—such as precise sequences of events in a coup, and perceptions held by actors in a risky and chaotic crisis—may also have necessitated serial interviews. Singh’s work certainly shows the value of spreading questions across more than one session, saving more challenging lines of inquiry for later in the sequence, when working with critical informants.

Verification, Triangulation, and Cross-Checking

Finally, serial interviews provide rich opportunities to challenge or verify information given in previous interviews and to triangulate and cross-check the participant’s answers in relation to other sources. The interviewer can pose again questions that have previously been asked to see whether the responses change, or reframe them in varying ways. The interviewer may gently confront the inter-viewee with information found elsewhere that conflicts with something he previously said; this can help resolve contradictions in the data, obtain deeper insight into a contested point, and assess credibility.

Why I Used Serial Interviews in Taipei

All six of the above conditions and considerations applied in Taipei. The neighborhood wardens were critical informants in my research in that it was precisely their role that I wished to understand. The wardens have unique knowledge of their own power, a subject that turned out to be quite complex. As with many state–society intermediaries, their power cannot be fully expressed as a list of legal prerogatives. Much of it is informal, constituting possibilities for exercising influence in varied settings. Situational in nature, it is shaped by many factors: a warden’s own social stature and resources, political calculations of more powerful city officials or council members who may seek a warden’s support, and citizens’ expectations, to name a few. Power and influence are not easily observable but are revealed or at least suggested in actual encounters between a warden and other actors. In my interviews, I sought to access a useful selection of such past events, both typical and exceptional. No warden, of course—even a highly motivated, forthcoming inter-viewee—could possibly recall from memory all relevant past events at once. Thus, the complexity of the subject matter and the demands I made on respondents’ memories were important reasons for serial interviews.

Trust building was another reason. Like many kinds of political actors, Taiwan’s neighborhood wardens are not eager to talk about every aspect of what they do. Certain activities that some take part in are illegal (e.g., vote buying in various forms) or touch on legal gray areas (e.g., persuading city officials to overlook building code violations). Others look unsavory or suggest favoritism. They often downplay their partisan activity because it antagonizes constituents who back rival parties and because they dislike any suggestion that they are mere foot soldiers in an election-machine army. They may also soft-pedal intervention with the city government on the part of individual constituents. Wardens thus are likely to steer inquiries toward uncontroversial, benign, and politically simple aspects of their work. They may obscure the very occasions when they exercise the most influence and downplay sociopolitical conflict in general. Although some inter-viewees were quite forthcoming even in the first interview session, serial interviews allowed for developing trust and familiarity over a period of time. For example, I have found it common for local politicians to wonder, at first, whether information they disclose would leak to their electoral rivals. Over time, such participants gained faith that I was not part of some opposition-research ploy and that my assurances of confidentiality were real. At an interpersonal level, repeated interviews generally made it possible to build genuine rapport with inter-viewees.

Finally, exploring over-time variation through repeated interviews deepened the data in multiple ways. I followed some neighborhood wardens over the course of years and saw some of them change from rookies to seasoned local politicians. Exposure to different stimuli through events that cropped up during interviews helped bring out different parts of their roles. As my own experience grew and I developed deeper contacts in the city government and political parties, it became more and more valuable to return to previously explored topics to cross-check wardens’ accounts and to challenge them with contradictory information that others provided. For instance, liganshi often provided independent, dispassionate perspective on topics where the warden himself might give biased or self-serving accounts, such as election campaigns, controversies, and missteps. I also drew extensively on written sources, such as election records and newspaper accounts, and sometimes interviewed competing warden candidates or members of other community groups. Of course, gathering this kind of information is part of preparing for onetime interviews, too. But in serial interviewing, it is not necessary to do all of one’s homework prior to the first encounter with a participant. It provides more opportunities to consult other sources and tack back-and-forth between those sources and the inter-viewees.

How Serial Interviewing Differs From Single-Session Interviewing

Serial interviewing differs substantially from single-session interviewing; it does not mean “doing more of the same.” The following are basic guidelines for how to conduct serial interviews.

The Dimension of Time

Interviews take place over a period of time; their frequency, and the total number of interviews, need not be defined in advance. It would be possible to compress the sessions into a few consecutive days, but more likely the pattern would allow days, weeks, or months to elapse in between. This has several implications. There is no pressure on you or the inter-viewee to get to everything all at once. The inter-viewee gets time off from you between sessions; he or she comes back to the topics fresh. Although the study may or may not become truly longitudinal, time is an opportunity. New events take place in the interim and provide material for discussion. Finally, interviews with multiple participants can proceed in parallel. For instance, one might spend the morning with one inter-viewee, then take the bus across town to spend the afternoon with another.

My interviews with Chen Boyu, a three-term warden of Taipei’s Shengfeng neighborhood, illustrate the utility of conducting a sequence of interviews over time. Coming from a white-collar professional background and a holding a degree from a respected university, the good-natured and articulate Chen was in his late 40s when I met him in 2006. Since then, we have had 15 encounters. Most were in his home living room, which doubled as his warden office; others took place over dinner or tea in restaurants and cafes.

Chen initially won his warden post as an independent candidate. In the first interview, he barely mentioned political parties at all. Instead, he emphasized—as wardens often do—his ability to connect with constituents of all persuasions on the basis of a distinctive platform (in his case, promising “arts and humanities” projects and environmental protection). In later interviews, however, it emerged that Chen was in fact a KMT member, one who had drifted away from the party but was gradually brought back into its fold. Chen ran with a KMT nomination in his subsequent reelection campaigns. As of our second interview, Chen still minimized the importance of the party’s financial support, saying that he had refrained from displaying campaign banners carrying the party logo and stressed that even many DPP voters backed his reelection effort.

Later sessions revealed how deeply Chen had become enmeshed in his party’s electoral machine. It was partly that his relationship to the party evolved over time, as its lieutenants tapped him to support their candidates for higher level offices. It was also that Chen became much more forthcoming on these topics. He explained the gritty details of his efforts to win votes for KMT city council members as well as then-mayor Hau Lung-Bin. These included, for example, hosting a campaign dinner for the mayor and supporters, which he held in a nearby public school, contrary to guidelines for such venues. Chen even received a phone call from the very apex of the party, then-president and KMT chairman Ma Ying-jeou. His partisan alignment seemed to cost him dearly when he lost his seat to an independent challenger in the 2014 local elections, which proved catastrophic for KMT candidates all around Taiwan.

The Interview Format

The interviewer does not follow a scripted sequence of questions but rather explores key themes and lines of questioning from different perspectives over the project. Thus, a loose yet disciplined form of semistructured interviewing is pursued. The format is tailored to obtaining deep coverage of core topics of interest. This could mean, for instance, exploring the participant’s memory of one important sequence of events (a social movement, or a decision-making process). Or, it could mean inquiring about an extensive set of occurrences (a political consultant’s various electoral campaigns, or a lobbyist’s multiple efforts to influence legislation). In the case of experiences that took place in the past, new layers can be excavated in successive interviews. If the events are current and ongoing (a vote mobilizer’s activities during the weeks before an election), the researcher can discuss fresh instances in each conversation with the participant. The interviewer prepares for each session by reviewing notes from past encounters, as well as related data, to determine what to cross-check, what previous points deserve follow-up, and what new information or questions to introduce.

In my Shengfeng interviews, for example, I made a point of inquiring each time about a simmering controversy over the holding of weekend fairs on a busy public square, featuring food stands, carnival games, and vendors of art and clothing. With each festival generating substantial revenue, much was at stake in an issue that pitted local businesses seeking extra customers against residents who didn’t want the noise and restaurants that didn’t want the competition. Warden Chen initially opposed the fairs, then reached a compromise with the sponsoring group, according to which the fairs went forward but without food or games, thus benefiting the neighborhood’s eateries. As this issue unfolded over the course of four interviews, a clear picture emerged of the warden’s degree of influence over such activities. Control over the square lay with the city’s parks and commerce departments, and thus, Chen had no ability to dictate the outcome. But the city offices were open to advice from a warden and from city council members upon whom he could prevail. Repeated inquiries gave me a chance to probe into the delicate question of whether he received any payments or kickbacks from interested parties. He consistently denied this, and the extent of my contact and acquaintance with him gave me more confidence in the truth of this claim than I would have had after just one interview.

The Observational Component

Observation of the inter-viewee’s interaction with his or her surroundings and associates can also enrich the interview process, providing what is sometimes called “latent content.” This is not a necessary component of serial interviewing, which could take place in a neutral, uninformative setting such as a coffee shop or the researcher’s office. But an advantage of conducting a series of interviews is the opportunities it provides to observe people in the settings within which they live or work, and thus to learn from their “natural” behavior—things that research participants would ostensibly be doing regardless of the observer’s presence. The inter-viewee’s demeanor with respect to superiors, subordinates, or others in the room, for instance, or what she says in phone calls that interrupt the interview, can provide important insights.

Moreover, what starts off as a straightforward interview may morph into something far more observational in nature. One political scientist noted that her formal interviews with law-enforcement officers eventually led to opportunities to ride in their cars: “I didn’t have to ask [permission] after a while. I drove around with them on patrol, doing whatever they were doing. This changed the interviews from semi-structured to a free-flowing format” (Interview, August 21, 2012). More commonly, interviews may lead the way to invitations to have lunch with research participants, sit in on meetings, or attend political rallies. What might initially be conceived as a simple question-and-answer encounter can become a rich blend of one-on-one questioning, group discussions, taking part in activities together, and the like.

In my own in situ version of serial interviewing, the setting and the warden’s interaction with visitors and callers have provided rich sources of material that enhance my questions. The neighborhood office—a crossroads where state and society meet—provides endless opportunities to observe how demands and requests from constituents are handled, how state programs (such as welfare, policing, and conscription) play out at the grassroots level, and so forth. I inquired about topics and incidents that came up during an interview session as a way of grounding my questions in specific, real-life events. For instance, I could ask how the warden dealt with Mrs. Li’s exemption request for her illegal add-on bedroom or Mr. Pan’s effort to obtain a curbside parking space rather than asking in general about residents’ efforts to bend the rules.

Over time, I obtained an increasingly precise sense of what a warden could and could not do. For instance, it was a cinch to get a city councilor’s office to donate cases of bottled water for an outing to a nearby park, especially if this would give the councilor an opportunity to join the tour bus and mingle with voters. It was possible to get someone moved up on the waiting list for a procedure at a city hospital but not to have them jumped to the very front of the line. Similarly, infrastructure projects such as renovating sewers and burying power lines could be given higher priority if a warden prevailed on the right people. Or they might be rescheduled for political convenience; in one case, major electrical work that would have closed stores and irritated several hundred residents was delayed until after an election. Yet generally speaking, Taipei wardens could not, for instance, get an ineligible family onto the welfare rolls, as the process requires verification by the civil service.

The Researcher–Participant Relationship

In serial interviewing, the interviewer and study participants likely develop closer relationships than in onetime interviews. As this article has emphasized, establishing familiarity and trust is integral to the method. When the researcher returns time and again, she is no longer a stranger. The interaction tends to become more relaxed and comfortable; a productive “working relationship,” in the term used by Fujii (2018), whose book explores aspects of such connections. ¹⁰ In some instances, it may take on the tone of a kind of collaboration or even friendship. Grinyer and Thomas (2012) write eloquently about the bonds that formed between Grinyer and her research participants over the multiple phases of a project in which she interviewed parents whose young adult children had been diagnosed with cancer. They consider potential problems, such as the possibility that repeated follow-ups concerning emotionally traumatic topics might harm the participants, but emphasize several benefits. These include strong commitment of the inter-viewees to the research, intense motivation of the researcher to maximize the quality and impact of the study, and multiple opportunities to present research findings to the participants themselves.

This closer relationship thus is not merely useful for obtaining better information, it is also part of what makes this method satisfying and even pleasurable. But it introduces extra considerations. First, the interviewer must retain the ability to view participants impartially. To become so sympathetic to an inter-viewee that one adopts his perspective uncritically, or fails to ask uncomfortable questions, would harm the integrity of the research, undermining the very point of the method. Second, as in related techniques like ethnography, it may introduce a deeper sense of obligation on the researcher’s part. After all, the participants are giving generously and repeatedly of their time; what are you doing in return? Some possible kinds of answers could include gifts, favors, dinners, sharing the results of one’s research in some form, or less tangibly, keeping in touch via mail, e-mail, or social media, and thus acknowledging and continuing the human connection that the research process has created.

Even though serial interviewing may not differ much from other types of interviewing in terms of what an IRB would require in a human subjects protocol, the relationship with one’s inter-viewees deserves extra thought. This includes taking care not to impose on participants or to prolong the interview sequence if they no longer wish to continue. For any kind of interview, obtaining participants’ informed consent means explaining to them that they have no obligation to participate and may end the interview at any time. In serial interviews, it is important to reiterate this point and to look for any signs that a participant would prefer to stop.

Despite the comfortable rapport I built, I was periodically reminded not to expect full disclosure of things that were embarrassing to my participants or put them in a negative light. For example, on one evening visit to Chen Boyu’s home, after I was given dinner and a long update on city politics, Chen’s wife asked him: “Did you tell Ben about the accusations?” He waved her off and changed the subject. I had to make inquiries elsewhere to find out what she was referring to; later, I brought it up with Chen and he shared his side of the story. The episode involved allegations by a rival that Chen had violated election law in distributing small gifts to a group of voters. Although he was eventually cleared of wrongdoing, he had endured a humiliating home raid by government investigators. The fact that Chen tried to avoid sharing this painful story with me, even after eleven previous interviews over the course of years, was noteworthy. It showed that even in the context of what can become highly trusting relationships, access is never total and participants are selective in what they reveal. But without serial interviews, I never would have learned about this incident at all.

Conclusion

Just how much time and effort is required to understand the perspectives and experiences of a given inter-viewee? A single encounter suffices for some purposes. But in other circumstances, the researcher might walk away from a onetime interview with a notebook containing many misimpressions, half-truths, partially understood anecdotes, and missed opportunities. Such would have been the case for me had I visited Chen, or other wardens, only once. Therefore, it pays to consider the extra depth, breadth, and validity that can be gained through multiple sessions. In particular, when the information sought involves substantial complexity—when multiple or multidimensional topics are at issue, when these topics involve subtlety and ambiguity, or when information is sought on a long series of past experiences—then just 1 or 2 hr may be far from enough. Projects involving critical informants, over-time change, busy participants, and special potential for bias and concealment are also good candidates for the serial interview approach. Moreover, this approach provides opportunities for triangulating among multiple sources of information.

For all these reasons, serial interviews were essential in my research on neighborhood wardens who play such complicated roles in Taiwan’s grassroots politics. It would be easy either to underestimate their influence or to perceive them as all-controlling “bosses.” Serial interviews brought out a nuanced picture of their actual power in Taipei. Their ability to exercise arbitrary control of their li is constrained by several factors including their limited formal powers, their vulnerability to voters’ displeasure, and the oversight of a relatively neutral civil service. The wardens nonetheless have certain kinds of influence. They accrue stature by winning competitive elections, and from the official nature of their positions, lowly though they may be in the city hierarchy. Much of their sway stems from their close contact with large numbers of city residents. They exercise influence indirectly, through city council members whom they repay with opportunities to win voters’ favor. The wardens have no power to lord it over their constituents, to dictate the terms of housing renewal projects, or to “deliver” the votes of hundreds of supporters (though they may influence some). But they form a key part of the dense fabric of state–society ties in a context where a powerful urban government is expected to be both close at hand and responsive.

Clearly, serial interviews have applicability in many settings involving power relationships or other challenging, complex topics. This article has explained their particular relevance for studies of relationships between voters and politicians, patrons and clients, intermediaries, brokers, and the like. We know that these phenomena are complicated, but often their multidimension nature is disregarded in favor of easily quantifiable proxies. To make progress on these issues, and many others, social scientists need methods that provide qualitative insight into complexity.

Serial interviews, of course, do not necessarily work miracles. There is no guarantee that using this technique will result in participants telling you everything you want to know. Although I have presented reasons to believe that serial interviewing can produce richer, more valid, and less biased information than onetime sessions, it does not of course nullify all forms of bias. Moreover, in any project, successive rounds of interviewing will eventually produce diminishing returns; if it feels that you are not obtaining new, useful material, it is probably time to stop.

Serial interviews and related techniques are familiar in some parts of the social sciences but mainly among researchers with epistemological commitments to deep, perhaps ethnographic study of participants’ lives and narratives. They are used by others in a more ad hoc or instinctual fashion (as was the case in the early phases of my own work). By spelling out the logic of this approach, this article has aimed to help make more researchers aware of the range of options available in designing interview research and the trade-offs involved. This approach entails costs, in the form of time, expense, and the opportunity costs of forgoing a larger pool of inter-viewees. As this article has argued, these costs are more worth incurring in some projects than in others. For each research endeavor, we should ask ourselves whether this kind of investment might be justified—if not indispensable.