(Dis)courtesy Bias: “Methodological Cognates,” Data Validity,and Ethics in Violence-Adjacent Research

Research as Politics: “Methodological Cognates” in Complex Crises

When designing field-based research, scholars generally consider potential sites and populations in terms of their methodological and empirical appropriateness. They may choose specific cases because they illustrate variation in key outcomes of interest (Lieberman, 2005); allow scholars to robustly test hypotheses via experiments or randomized controlled trials; hold utility for theory building (George & Bennett, 2005; Gerring, 2004); represent compelling sites to ethnographically examine systems of meaning-making and practice (Schatz, 2009; Wedeen, 2009; Yanow & Schwartz-Shea, 2006); or generate representative large-N samples for surveys. “Real world” considerations—such as the researcher’s language abilities or a geographic site’s accessibility—may informally influence the researcher’s choices, but are rarely articulated as part of the formal site selection process or justification.

Especially in environments affected by mass violence, forced migration, or natural disaster, politics come into play in ways that affect ethical research practice, data validity, robust analysis, and generalizability. Researchers may encounter bureaucratic hurdles, government intimidation, or social unrest that disrupts even well-planned research visits (Clark, 2006; Thomson, 2009, 2010). Populations may be hidden or otherwise hard to identify and reach (Khoury, 2020). Respondents’ political biases, whether with reference to US or European government sponsorship of research or to local enumerator ethnicity, may influence participants’ willingness to join a survey or remain part of it (Adida et al., 2016; Corstange, 2014, 2015). Researchers may find that they lack training in how to avoid retraumatizing research participants who have experienced distressing events (Lake & Parkinson, 2017). For any number of reasons, interviewees may misrepresent or avoid discussing their roles in political processes (Corstange, 2009; Fujii, 2010). Government agencies may manipulate official statistics or simply lack the resources to compile them (Andreas & Greenhill, 2010; Jerven, 2013).

Scholars may approach these challenges from any number of angles, ranging from emphasizing the importance of trust-based, credible, and/or working local relationships (Al-Faham, 2021, p. 5; Fujii, 2017; Thomson, 2010) to relying on sub-national comparisons (Giraudy et al., 2019; Snyder, 2001), employing sophisticated quantitative analysis (Hoover Green & Ball, 2019; Price et al., 2014), or practicing data triangulation, especially via mixed-methods or multi-epistemological research (Davenport & Ball, 2002; Driscoll, 2021; Khoury, 2020; Thachil, 2014, 2017; Thaler, 2017, 2019). Yet, the difficulties associated with conducting research in these locales are often understood as uniform, intrinsic characteristics that influence researcher-participant relations in predictable, blanket, and consistent ways across populations of interest. This assumption often cuts across approaches. For example, researchers learn that they can neutralize the potential for inaccurate or “fake” responses by adopting particular techniques in survey research or interviews, for example, asking survey or interview questions in multiple ways or by using list experiments.

These claims rest on a second assumption that research in fragile settings and with vulnerable populations occurs independently from other actors’ data- and evidence-gathering activities. This assumption, however, shields scholars from critical dynamics that affect data validity and reliability. Namely, scholars working in settings affected by mass violence, humanitarian intervention, and crisis are almost always situated within a broader political and professional class that seeks information from affected populations in order to disseminate news, provide relief and services, create reports (whether humanitarian or intelligence), or advocate. Due to similarity in techniques deployed to gather data in these spaces—from survey instruments to games, semi-structured interviews, and participant observation ² —potential participants often see researchers not as independent actors, but rather as part of a larger body of outsiders who are often better educated, wealthier, more mobile, and culturally or politically distinct from the milieu in which they work (Lewis et al., 2019). This fact is most consistently manifest in the broad trend of potential participants confusing researchers for aid workers or journalists (Foster & Minwalla, 2018; Lewis et al., 2019), though scholars also report potential respondents assuming that researchers may be intelligence agents (Driscoll & Schuster, 2017; Thaler, 2019).

Short Questions, Complex Answers: Studying Humanitarian Actors in Ongoing Crises

I use ethnographic, observational, and interview-based research to ground my conceptualization and analysis of the effects of regurgitation, redirection, reluctant participation, and resistance on diverse research approaches in crisis- and violence-affected contexts. The goal is to better understand how structural, institutional, and environmental conditions present in such spaces shape data. Specifically, this article draws from intensive fieldwork conducted over the course of a decade in the Middle East, including two non-consecutive years of ethnographic, interview-based, and archival research on Palestinian and Syrian refugees in Lebanon between 2007 and 2018. ⁴ Spending several months living in Palestinian refugee camps and over a year volunteering in them allowed me a unique view of how research, journalism, ⁵ and humanitarian activity interacts with displaced persons’ everyday lives and provided the foundation for more focused research in the more security-restricted field in Iraq.

Two months of participant observation, general observation, and interviews with humanitarian and journalistic actors in the Kurdistan Region of Iraq (KRI) and Ninewa Province in August 2016 (during the early stages of the military campaign to retake Mosul), in July−August 2018 (following Islamic State’s—IS’s—defeat, in a setting of mass displacement and hyper-politicization of the aid environment), and in May 2019 (when many displaced Iraqis remained in camps or had returned to them following re-displacement) complement this prior research experience. I conducted in-depth or informal interviews with 54 humanitarian aid workers (including people working in front-line healthcare, education, and camp logistics as well as people in administrative roles), public-facing Kurdish officials, journalists, and academics. Most of these interviews lasted between one and three hours; at the interviewee’s discretion, they took place in cafes, private offices, clinics, during long car rides, and in hotel lobbies.

I also spent time shadowing humanitarian workers. In 2016, I observed an international non-governmental organization’s (INGO) humanitarian operations during the early stages of the Mosul Operation, a multi-country coalition effort to defeat IS. During my time with this INGO (henceforth INGO 1), and with the consent of regional leadership and on-the-ground staff, I conducted daily observations with operational staff, including accompanying them to projects in Erbil, Ninewa, and Duhok provinces. I also conducted open-ended, informal, and semi-structured interviews with INGO 1’s employees. Also in 2016, I spent several days observing doctors, clinicians, and community health workers employed by healthcare-specific INGO 2. With INGO 2’s local management and staff’s consent, I followed healthcare workers as they completed assessments and checked medical facilities in two internally displaced person (IDP) camps, one in Erbil province and one in Duhok province. In 2018, I visited a third IDP camp in a former IS-occupied area of Ninewa Province. There, I conducted similar interviews and observations with healthcare providers and administrators of a locally-based NGO. ⁶ Contacts made during prior research in the Middle East introduced me to foreign and Kurdish friends working with other INGOs in Iraq, who allowed me to speak informally (in some cases, contacts explicitly used the journalistic term “on background,” meaning not for attribution or quotation) with them about their on-the-ground experiences with the humanitarian sector. I took extensive written notes during interviews and recorded field notes several times daily. Due to the in-country and international sensitivity of the situation, I did not digitally record any interactions and secured confidential written interview records and field notes digitally.

Additionally, in August 2016, I attended United Nations (UN) humanitarian cluster meetings in Duhok and Erbil. In July−August 2018, I observed meetings of both the Kurdistan Regional Government’s (KRG’s) Joint Crisis Coordination Centre with INGOs and a meeting of the central Iraqi government’s crisis coordination body with humanitarian responders working across the country. I secured invitations to attend these meetings in advance and clearly identified myself as an academic researcher during introductions. Joining these events provided me a deeper view into practices of humanitarian data collection as well as meso-level governance of humanitarian work in KRG and Iraq more broadly. It also made it possible for me to join relevant mailing lists that detailed program planning and execution. Furthermore, UN and (I)NGO employees, as well as journalists, granted me access to internal planning documents, reports, and WhatsApp channels with the understanding that I keep their detailed contents and provenance confidential. Finally, in May 2019, I conducted a dozen interviews with English-speaking journalists in Sulaimaniyah and Erbil; the background information they provided further informs my understanding of on-the-ground dynamics and ethical tensions that influence data collection in conflict settings. I was thus able to collect an unusual wealth of data that provided a multi-faceted picture of the nature of humanitarian and journalistic work in northern Iraq, the interactions between recipients of humanitarian aid and actors from outside their communities, and the manner in which these inform and affect data collection.

Four interludes ground the rest of this article. In line with standard practices in interpretive research siting, each interlude embodies a micro-study of dynamics that I have used to build a case (Soss, 2018) in order to illustrate a specific mechanism. Drawing on previous work on conflict and humanitarian crisis, I coded field notes and interview transcripts with the goal of identifying patterns in on-the-ground researcher-participant dynamics that influenced data validity and reliability in my own field sites. By centering on practices that explicitly influenced the direct relationship between the investigator(s) and research participants, rather than the larger context (e.g., how gatekeeping or security concerns might prevent access to participants), I subsequently identified the four mechanisms at the center of this article. Each interlude has been selected to reflect dynamics that consistently emerged across my research. In line with an ethnographic research sensibility (Pader, 2006; Schatz, 2009; Simmons & Smith, 2017), each interlude emphasizes banal, everyday patterns rather than exceptional moments; their exposition is used to unfold nested interactions in a way that clarifies the potential effects on researchers’ data.

Regurgitation: Humanitarian Assessments in Iraqi Kurdistan

The early morning streets of Erbil are empty as an Iraqi medical team crams gear into two large vans before piling in themselves. We are heading to Debaga, a formerly sleepy Kurdish town of 2500 south of Erbil. In the past months, the number of residents has ballooned to approximately 30,000, with hundreds or even thousands more arriving each day. Most of these arrivals are Sunni Arabs who had been living in IS-controlled areas since the summer of 2014. When people arrived at the Kurdish-controlled Debaga camps, they immediately entered a “screening” system, which separated “fighting age” male IDPs from their families and sent them for intensive questioning by security agencies. It could be days or weeks before they would be permitted to join the camp’s population. Women and the elderly consequently managed the tasks of registering with the camp managers and settling families into large United Nations High Commissioner for Refugees (UNHCR) tents. This system meant that from their first days in the camp, men and women had vastly different experiences of being asked personal questions and providing information to authorities. In Debaga, it was clear that women often managed families, and thus personal information and its distribution.

IDPs had different experiences of humanitarian services depending on the time of their displacement and hometown, both of which were used as proxies for the potential threat they posed. We arrived mid-morning in Debaga, when the temperature outside was already over 40C. The medical team set up operations in a large tent that served as the camp section’s clinic. Most of the equipment had traveled with our convoy; INGO 2 shared a huge clinic tent with several other INGOs that rotated through over the course of the week. Each would supply and staff the clinic on their days, leaving very little in the massive tent after hours. Upon our arrival in Debaga, a team including doctors and health educators scanned a section of the camp that hosted new arrivals, planning an assessment.

Assessments constitute a central aspect of systematic information gathering and operational planning for humanitarian operations, especially when (I)NGOs are working in a cluster system. They are effectively a means of gathering basic demographic information along with data of interest for actors operating in a specific sector. Repeated observations of assessments in camps and relief sites revealed that assessments train people to tell stories in a certain way. In Debaga, for instance, the team leader leans downward into the tent ⁷ doorway, greets people inside, identifies the INGO and its purpose, and asks if the team may conduct an interview; it approximates a consent process. We sit on foam sleeping mattresses, an INGO-branded team facing a family across a few meters of floor space. Team members click pens and begin the interview. They ask how long people have been there, where they are from, how many children are in the family, how old they are, whether the children have any health problems, whether the family closes or uses screens (to block the sand flies that carry leishmaniasis), about the condition of water sources and latrines, and about any critical, emergent, or chronic health conditions people in the family face (e.g., a fresh wound, a cold, type 2 diabetes). Shelter assessments, which I also observed, might note the number of people in each household, its age, the dwelling’s condition, any environmental issues (e.g., flooding, exposure to elements), whether it had windows and doors, and the status of any associated cooking areas. They would also attempt to assess unmet needs (e.g., families for whom tents were unavailable). After between five and ten minutes we say our thanks, often politely refusing a lunch invitation, and move down the row to the next tent, repeating the entire process.

This interlude demonstrates the pre-existing patterns of interaction that lead to regurgitation. In Iraq, camp residents almost always participated in multiple, repeated assessments. Such assessments encourage, or even require, people to routinely relay extraordinarily personal stories in a straightforward, systematic, compartmentalized way to strangers. Though the new arrivals in Debaga were not yet accustomed to the system, interactions that I later observed in other Iraqi IDP camps and program sites immediately revealed a pattern; a beneficiary would encounter a humanitarian worker and readily anticipate questions, often with a sheaf of documents ready for verification. ⁸ They might state their name, if they were married, if they were widowed, if someone in the family was a survivor of gender-based violence, ⁹ if a family member had been killed while serving in the Iraqi military, the number of people in their family, the number of children under a certain age (an indicator of special needs), any health problems in the family, the physical state of their house, if they had returned home, and so forth.

IDPs’ entry into camps such as Debaga thus exposed them to at least two cognate practices: interviews (by security forces and humanitarian actors) and surveys (by humanitarian actors). Camps’ built environments often facilitate both types of interactions; they separate populations at limited entry points and assign people to numbered tents in clusters or rows to facilitate information gathering, resource distribution, and monitoring. This same physical organization also makes camps attractive locales for research requiring randomization and interviews; people are usually home, with very little to do, and seemingly few excuses to say no. They are understood to have been randomly assigned to tents as they arrived, even if their time of departure and ability to reach that camp reflected highly non-random processes. Yet they also live in spaces where refusing interventions is frequently viewed with suspicion and where sharing personal information is the only pathway to receiving life-sustaining aid; if they do not understand that academics are distinct from security officials or humanitarian workers, then their consent to participate in academic studies is likely compromised.

Due to the organization of humanitarian deployments, displaced persons participate in multiple assessments of varying standards, further engraining a regurgitative response. In Iraq, the technical divisions of labor—for example, groups focused on working with shelter, protection, sanitation, or women’s issues—within the UN cluster system meant that multiple humanitarian actors operating in each cluster were designated to work only in specific camps on specific issues. As a result, in Duhok, for example, INGOs, UN agencies, the KRG provincial government’s Board of Relief and Humanitarian Affairs (BRHA), and local NGOs independently gathered data for only the subject areas and populations to which they were assigned by the clusters. A healthcare INGO would only conduct a healthcare assessment, and would only do so when and where they were assigned to work. Among frontline healthcare providers, some INGOs worked in certain camps on certain days while others worked complementary shifts. Even among health organizations, the assessment instruments and level of training given to enumerators differed greatly. As a consequence, an IDP living in Karbato 1 Camp in Duhok Province would participate in a different battery of assessments than an IDP living in Debaga (where BRHA did not operate), even if each participated in assessments related to the same subject clusters.

The act of interviewing itself in this setting creates and reinforces power-infused categories of “aid worker” and “beneficiary,” which are relevant for their effects on displaced persons and their future engagement with academic research, especially hybrid national-local teams. This is even true when local enumerators do the work, as in the assessments I observed. Like the displaced families, several of the medical workers in Debaga were IDPs from Mosul themselves, identified themselves as such to beneficiaries, and spoke with Moslawi accents. Yet their medical backgrounds, their families’ decisions to flee IS early, and their employment placed them in a different category, both for INGO 2 and for the IDPs they interviewed. In other cases, Iraqi Kurds or Iraqis from nearby regions such as Kirkuk (both of whom speak with different accents) conducted assessments, worked as fixers, and were part of research teams. Those accents may telegraph a variety of signals, including intimidation (e.g., a Kurdish accent in Arabic for a Moslawi man who had been interrogated by the Peshmerga) or access to aid (e.g., a Moslawi accent from someone working with foreigners), each shaping the form of regurgitation. ¹⁰ Overall, the repetitive question-asking and contextual signifiers produced practiced and circumscribed responses. Such responses call into question academic data based on cognate practices in such contexts, particularly if researchers do not relay clear understandings of how both methods and social relations operate in a specific site.

Redirection: Evolving Modes and Demands of Interaction

There was one site where the stories people told repeatedly diverged from the curated forms that assessments encouraged: clinics. Conducting observations and informal interviews with front-line healthcare workers such as primary care physicians revealed that they often performed a form of “healthcare plus,” spending extra time interacting with and listening to IDPs’ concerns, fears, annoyances, and anxieties. For example, two primary care doctors working in the Ninewa Plains in 2018 noted that they had several regular (as in, several visits a week or month) patients who often had nothing medically wrong with them, but who would sit around talking for twice to three times the length of an appointment. Several physicians in Karbato, as well as Yazidi psycho-social workers, had alluded to similar dynamics in 2016; however, they were reported to me in a more widespread fashion in 2018, when many people had been displaced for over Four years or had fled the Battle of Mosul. For example, an INGO information hotline operator, whose job was simply to tell people what services were available and how to apply, told me in 2018 that beneficiaries often spoke to her for up to an hour about their emotional frustrations and challenges. Doctors who I interviewed attributed these patterns to people having nowhere to go and few people to whom they could speak honestly; the clinical environment and doctor-patient confidentiality, they hypothesized, allowed patients to feel protected and to medicalize difficult aspects of their days in a way that the scripted nature of assessments actively foreclosed. Being able to take a medical encounter, and to make it an emotional one, allowed them a form of care to which they did not otherwise have access. The helpline operator, too, hypothesized that her anonymity and gender (apparent from her voice) made her a rare, accessible outlet for people who otherwise did not have private, safe space to vent or complain.

These behaviors reveal how redirection operates in practice; people enter into specific types of clearly defined interactions—one-on-one health consultation or informational request—and shift it to what they need (a confidential or anonymous space to talk about their feelings). Redirection might consequently be characterized by an interviewee granting initial consent followed by their growing hesitation/frustration/lack of cooperation as the interaction proceeds. In some cases, the interlocutor may come to believe that the researcher’s focus is inappropriate and seek to reshape the conversation. For example, Moe Ali Nayel, an interpreter who worked with US-based academics in Lebanon, has written about arriving in a refugee camp to interview people about their experiences fleeing Syria. He describes the researchers as asking people to relate how and where they traveled during each stage of their journey. However, one interviewee instead started telling them about the health problems her son was experiencing following the journey. She became upset when the researchers ignored her attempt to redirect, attempted to re-start their line of questioning, and requested that she stop smoking (Nayel, 2013).

The ways that redirection affects data quality depend heavily on the research design, epistemological commitments, and the researchers’ approach. Some researchers might insist on responses to their original questions and ignore the redirection, thus demonstrating an inability to productively engage with research participants or to gauge their needs. Depending on the degree and topic of redirection, there are several potential ramifications, from a slightly frustrated interviewee, to refusal to continue the interview, to retraumatization. In other methodological traditions, redirection may open new avenues for questioning by suggesting important aspects of experience that a researcher may not have previously considered and indicating the importance of relevant and accessible care referrals. A flexible researcher, especially one drawing upon interpretive and relational interviewing methods (Fujii, 2010, 2017; Soss, 2013) might improvise, follow a new line of questioning, and generate robust individual- and collective-level data as a result. However, redirection of the sort experienced by the doctors and the helpline operator could also pull social science researchers into a problematic role for which they lack training and that may harm participants. Interactions of this sort are fundamentally products of participants’ vulnerability, rather than their express consent and intentional engagement. Researchers who continue data collection under these circumstances without explicit grounding in trauma-informed research practices are both actively committing harm and analytically hamstrung. Rather than “providing grateful people an outlet,” researchers in this situation are not adequately attuned to participants’ reactions to questions, lack context for their responses, and are unable to adjudicate truth claims (if that is a methodological goal). This leaves academic researchers fundamentally unable to competently analyze their own data. ¹¹

Reluctant Participation: “Getting Stories Out” at a Cost

Almost two years after my trip to Debaga, I found myself in a white Toyota SUV speeding out to an IDP camp in the Ninewa Plains. Accompanied by Rizgan, a public affairs representative for a humanitarian NGO, I was going to interview and observe NGO workers in the camp, which predominantly hosted IDPs who had fled Mosul during the 2016−2018 military operations. We bumped over the bridge at Kalak, the former front line between the Peshmerga and IS. Huge defensive earthworks rose on each side of the road, marking the Peshmerga’s advance up the road towards the town of Bartella and, further along, Mosul.

Before the camp came into sight, I asked Rizgan if he accompanied many foreigners. He cheerfully replied that he did so all the time; especially since the Mosul Operation, there had been a lot of journalistic interest. Curious, I asked how IDPs responded to being interviewed by so many journalists. “Oh, we pay them,” Rizgan clarified. He continued:

We put them in a certain section of the camp and pay them to do the interviews each month. It’s a lot of their time and a lot of stress on them, you know? The journalists ask about a lot of difficult things. The displaced people, they’re vulnerable, they’re tired, so we rotate which families do interviews each month and pay them for their effort. We don’t want the foreigners wandering around, going into tents, and asking private questions.

Rizgan’s tactic in effect recognized how reluctant participation operates; without paying displaced persons, they would not want to engage in interviews with journalists. Working in spaces characterized by heavy securitization, repeated assessments, and high journalistic interest often means that people tire of and potentially fear repeating personal details. Reluctant participation can stem from the fatigue this repetition causes in terms of the sheer volume of questions or in repeated questioning that people experience as misguided or traumatizing (Foster & Minwalla, 2018; Minwalla et al., 2020). The broad journalistic, humanitarian, and law enforcement use of methodological cognates such as surveys, structured and unstructured interviews, and even games (used in some psycho-social activities) prime participants for researchers who arrive later as well as exhausting them.

This argument squares with previous findings. Scholars have noted that “[r]esearch participant fatigue can potentially also undermine the study as the absence of interest can manifest itself in superficial, incomplete or poor answers. Excessive research might contribute to lack of generalizability of the results if the environment is one of super-experimentation with poor response rates, half-hearted research participation, and high attrition” (Cleary et al., 2016, p. 381). Furthermore, “[t]here is some evidence to suggest that research fatigue is increasingly being mobilized as a reason to decline or withdraw from qualitative research” (Clark, 2008, p. 955). Lack of language skills, securitized settings, and short research trips may all compromise researchers’ abilities to take the time to recognize these impacts and appropriately contend with them, leading to poor-quality and unreliable data when fatigue and frustration remain unidentified. This dynamic may also mean that disproportionate numbers of potential participants select out of studies over time and that newly arrived academic researchers are unable to obtain the random samples necessary to many research designs.

Many (I)NGOs have rules prohibiting the kinds of engagements that Rizgan described; both journalists and academics have guidelines that restrict payment for interaction. However, smaller, local NGOs might feel they need to facilitate meetings with their beneficiaries to bring attention to their work. Unlike many, Rizgan’s organization felt obligated to actively compensate IDPs for the time and stress that journalistic and diplomatic interviews caused. In the mind of its managers, there was no other real reason for them to speak with outsiders; it did not get them anything.

In the case of Rizgan’s NGO, humanitarian actors paid participants without clearly informing journalists, diplomats, or researchers that they were doing so, introducing a second effect on data quality and reliability. Hidden payments mask sources of coercion and compromised consent, as well as injecting inequalities within the site (some people benefited, some did not), which fundamentally undermine ethical academic research conduct. But the NGO also selected families with extreme bias; they were neither randomly selected nor representative of the camp’s population. Rather, officials selected obedient families (e.g., where women no longer wore full niqab or men beards, both considered signs of sympathy for IS) who would demonstrate proper appreciation of the NGO, relay that they still had needs, but maintain an anti-IS ideological line. Despite potential appearances, it would be nearly impossible to obtain a random sample of IDP households in a camp under these conditions, despite camp officials assuring researchers otherwise.

Resistance: Local Agency in Action

“ I had several conversations about your project with Palestinian, Syrian, and Lebanese friends who have worked on-the-ground with both researchers and/or humanitarian agencies. I won’t go into details, but everyone strongly suggested looking outside of the Palestinian camps, particularly when it came to South Beirut…[t]he other responses were more complex. To be honest, three were outright hostile to the project’s framing and design, particularly around the conceptualization of “dignity” outlined in the [project] description (which I realize isn’t of your making, I’m just relaying reactions when I read from the website and described your email, I did not share your name).

This final interlude highlights the fourth mechanism: resistance. The verbal refusals and discussions relayed in the email above in part stem from general practices of research design and siting, but are also broadly associated with disconnection and feelings of general exploitation in a specific site. My interlocutors in refugee and IDP camps described additional modes of resistance, including deliberately giving false answers to survey questions. As one friend, a Palestinian NGO worker who had degrees in business and statistics, noted in 2014 as we discussed research on refugees in Lebanon: “we know how surveys work, and we know how to mess with their data. Give us the software, let us do the study.” ¹³ Resistance can affect any project; it takes myriad forms from refusing consent to spreading rumors about researchers to providing deliberately misleading responses to interviews or surveys to overt threats. It might bias samples, cause a researcher to inaccurately represent political dynamics, constrain opportunities for immersion, cause a scholar harm, or completely prevent research.

Resistance often stems from social dynamics associated with researcher ignorance of local history and politics. Projects where locals perceive researchers as “parachuting” into the field (e.g., for one week to train enumerators or to interview elites) may be particularly likely to prompt resistance. For instance, in the summer of 2014, Abu Karim, one of my contacts in a Beirut refugee camp’s administration, stopped by the house where I was living, laughing about a foreign public health researcher with whom he had just met. He explained to me and my hosts that the scholar in concern had very carefully and patronizingly explained to Abu Karim that he thought the water in the camp “might” be contaminated and causing health issues. ¹⁴ He wanted Abu Karim’s permission to test it, to offer him “proof” of contamination so that Abu Karim could do something. The researcher, Abu Karim related between derisive snorts, repeatedly insisted that his work “wasn’t political,” thinking that Abu Karim would then consent to it. At the time, factions in the camp were stalling an internationally funded overhaul of the water system; it was thus patently absurd to claim that water “wasn’t political.” Fed up with a steady train of what he experienced as uninformed researcher requests, Abu Karim relayed that he told the man that anyone too stupid to know that water in the refugee camps was political was too stupid to do public health research. Though he wanted to see something done about the water, Abu Karim refused the researcher permission to conduct tests and declined consent for an interview. This dynamic is not unique; different locales have set heuristics for determining whether to resist a researcher or project. In Iraq, for example, one of the questions I was repeatedly asked was how long I planned to stay. This question was a way of assessing whether I deserved people’s time and effort, because they were sick of “people who come for a week and write a book based on four interviews,” as more than one contact put it.

Interactions of this sort evince that, far from being passive subjects of research, participants in fragile and violence-affected settings, particularly those with repeated exposure to humanitarian actors and media, draw upon a multitude of strategies to invoke agency in their interactions (Lake et al., 2016). As this particular vignette illustrates, resistance may also reveal local power dynamics, particularly among educated participants or in NGOs who have the ability to gatekeep. These techniques are not always visible to newcomers; even experienced survey enumerators may overlook them. Members of a research team might even engage in forms of resistance themselves, especially in locales where foreign researchers are associated with unethical research practices, repeatedly conduct research seemingly divorced from local realities and wishes, and/or have omitted local collaborators and research teams from publications, accolades, and grant applications (Bouka, 2018; Cronin-Furman & Lake, 2018; Mwambari, 2019a; Sukarieh & Tannock, 2019). The Palestinians, Syrians, and Lebanese with whom I consulted regarding my colleague’s project had all worked with multiple foreign researchers in different roles, and had usually engaged with the hope that their contribution to a research project would ultimately help others. In the case above, one potential research assistant with whom I consulted resisted by demanding the project provide immediate benefit to the community; what he was actually attempting to do, by his own acknowledgment, was to make the researcher feel so unwelcome and guilty that she would abandon the project. He had no expectation that the researcher would seek to fulfill his request.

Recommendations and Extensions in a World of Perverse Incentives

The relationships that affect participation include those between researchers and participants as well as the broader social networks that constitute both types of actors’ roles and positionality within larger power structures. There are extremely feasible steps that researchers can take to anticipate and plan for dynamics such as regurgitation, redirection, reluctant participation, and resistance. To start, researchers must to be competently trained in research design, field methods, ethics, and research team management. They should pursue original research siting, language acquisition, continuous consent procedures, and thorough literature reviews that incorporate work done by local researchers. Rejecting pre-set “cases” and categories of analysis and treating them instead with a critical eye (Ghosn & Parkinson, 2019) may open up avenues to higher quality and more unique data. Creative siting practices encourage researchers to examine dynamics of scholarly interest—for example, life-or-death decision-making—where it is less obvious, such as studying wealthy residents of coastal resort areas threatened by tropical cyclones. This is not to say that displaced people who have recently fled war cannot speak to these dynamics; it is to argue that other actors’ use of methodological cognates will shape research interactions with these populations in ways that likely produce suboptimal data, often while involving significant risk and stress to the population. Researchers from North America and Europe should also recognize that the very traits that might make certain field sites attractive—places where many people speak English or French, that are close to airports, that are considered relatively “safe” by institutions such as the United Kingdom's Foreign, Commonwealth, & Development Office or the United States Department of State—are potentially over-researched. Advisors who are unsure of whether a proposed student project is in an over-researched space can do a quick Google Scholar search and consult with other academics who work in the region.

High-quality methodological training is only a start; the role that methodological cognates from across epistemological traditions play in shaping populations’ interactions with outsiders, and thus their data, means that methods alone will not resolve these issues. As scholars have noted, methodological innovations designed to deal with the issue of over-research, including participatory approaches, “all can themselves contribute to the overall experience of over-research by the researched” (Sukarieh & Tannock, 2013, p. 497). Clark (2008) and Cleary et al. (2016) emphasize that the sites that are over-researched are likely that way because change has not occurred following prior research. Continuing projects in over-researched spaces also allows for the ossification of commodifying and exploitative institutions such as fixer and survey firm markets (Ghosn & Parkinson, 2019; Mwambari, 2019b). It is worth considering the role that time plays in creating the structural conditions for such interactions. Resistance is comparatively likely in spaces experiencing forms of protracted crisis and dispossession that have repeatedly been the subject of academic interest and have experienced little benefit from research. Reluctant participation is especially likely in contexts where populations experience intense state surveillance. Redirection might occur with relative frequency in contexts where there is comparatively little researcher interest, contributing to misunderstanding of researcher goals, roles, and boundaries.

Research must be engaging, meaningful, and respectful if people are expected to participate, especially without remuneration and particularly in spaces shaped by large journalistic, humanitarian, and security presences. Scholars should not conflate people wanting to tell their stories more generally with researchers being entitled to them. Part of the issue with the “dignity” project proposed above was that locals simply did not find it conceptually meaningful. To put it more bluntly, some projects simply have no benefit to local communities and very little potential payoff in terms of changing conditions on the ground. While they might be read as having “academic import,” the more relevant question is whether pursuing this research in these sites given the challenges to data reliability and validity will actually produce viable, high-quality data and robust analysis. The evidence referenced above indicates that the answer is very likely “no.”

This conclusion brings into question whether such projects fulfill core principles of research with human subjects, including beneficence and justice (Office of the Secretary of Health and Human Services, 1979) and whether they adhere to disciplinary ethics codes. For instance, the American Political Science Association’s ethics code explicitly states that researchers must consider “the broader social impacts of the research process as well as the impact on the experience of individuals directly engaged by the research” [emphasis added] (American Political Science Association, 2020, p. 13). Academics should consequently focus on finding ways to engage local agency, meanings, and experiences, then link these factors to topics of greater scientific import in their research questions, designs, and practices (Consulo & Hudson, 2018; Thachil, 2017). Notably, this strategy might necessitate broader use of inductive and abductive approaches to inquiry, at least in projects’ early stages. ¹⁵

Another promising strategy is to research “up,” or “across,” rather than “down”—that is, to study powerful actors rather than vulnerable populations—as scholars such as Nader (1972) have long advocated. Explicit calls in this realm have been made for many specific field sites (Al-Hardan, 2017). Such a move might practically involve studying mid-level government officials and the effects of their policies, rather than those for whom they make policy, as Lama Mourad’s work on Lebanese government policy towards Syrian refugees does (Mourad, 2017). However, it must be accompanied by increased recognition of the skill and time it can take to access powerful populations and build working relationships, as well as accompanying ethical considerations (American Political Science Association, 2020, p. 3–5). Such a commitment must be made to avoid reifying elite voices and power structures.

Methodological cognates are used in spaces beyond those described in this article. Regurgitation, redirection, reluctant participation, and resistance could consequently manifest for researchers in any number of contexts where “outsiders” repeatedly ask questions, shape narratives, constitute a separate class of actors, and intervene in politics. Sites of protracted, collective legal processes surrounding labor or land disputes, where workers or residents might engage legal counsel, be deposed, and adopt relevant language (e.g., integrating legal terms into speech with non-legal entities) constitute such settings. In other environments, resistance might manifest as a product of public discourse that, for example, targets journalists, thus classifying people who ask questions “like journalists” (that is, researchers) as threats. Acute humanitarian crisis is not a precondition for any of the four mechanisms to affect scholarly research.

The purpose of this article is not to discourage research, but to emphasize that sound research ethics are constitutive of and a necessary precondition for quality data generation. While the relative ease with which marginalized and underserved communities can be accessed by researchers may be tempting, this article shows that the relationships posited by many projects are often not built on trust and participant openness, but rather on latent coercion and practiced scripts developed before the researcher even arrives. This article’s conclusions are based on research in refugee and displaced person camps, but they also hold lessons for broader research trajectories in sub-fields focused on numerous areas including immigration and migration, urban politics, and the politics of disaster. Above all, this article challenges claims that academic research in fragile spaces and among vulnerable populations can be conducted under the assumption that academic researchers are understood or treated as independent actors and that consent equals enthusiastic, genuine participation. Moreover, it demonstrates that populations living in these contexts are often extensively primed ahead of academic interventions. It strongly indicates that data collected uncritically in these settings should be challenged in terms of its validity and reliability. Yet above all, this article encourages scholars to re-think why and for whom they conduct research among vulnerable populations, especially if they cannot reasonably elicit change.