Avoiding Risk

The first strategy is avoiding higher-risk areas or subjects by adjusting the research design to “select themselves out” (Sluka Reference Sluka1990, 124). IRBs, insurance requirements, and relevant legislation contribute to the predominance of avoidance strategies. Researchers manage risks, costs, and potential stress by designing credible research programs that can be carried out without seeking subjects in higher-risk areas. This circumscription of the population under study requires a researcher to use different methods and ask different questions about the world that remains within the (sampling) frame. When data quality is the primary issue, research is shaped to minimize this constraint. This avoidance, however, can amplify the accessible voices and neglect questions that increase risk and complexity. For example, urban and metropolitan areas often are considered safer (Baird Reference Baird2018; Chambers Reference Chambers1979). Avoidance approaches can perpetuate an “urban bias” leading to a “general tendency to interpret phenomena a-contextually and in an exclusively top-down manner” (Kalyvas Reference Kalyvas2004, 166).

The political and security environment constrained the voices represented in Swenson’s research on reconstruction efforts in Afghanistan. In an example of shaping research questions to mitigate, although by no means eliminate, risk in Afghanistan, Swenson—facing physical risks, visa limits, heightened scrutiny from funders and university officials, and high financial costs—chose to tackle a question that required in-country travel but that could be addressed through elite interviews in Kabul. For safety reasons, the research approach largely excluded individuals opposed to international state-building or the current regime, particularly those willing to engage in violence. The difficulty of accessing some respondents, even within the capital, amplifies the voices that are accessible, which raises questions about researchers’ participation in an “echo-chamber” (Kuus Reference Kuus2013, 118) whereby certain views are replicated across different research projects. This fieldwork dynamic raises concerns over researchers’ roles in potentially reproducing exclusionary dynamics.

Attempts to avoid urban bias can bring their own challenges. Independent researchers may seek to maintain access while mitigating risk by linking up with non-governmental organizations (NGOs), development agencies, military organizations, or the United Nations. These external organizations often possess the vehicles, contacts, security protocols, and local knowledge to facilitate research in more remote or dangerous areas. As Duffield (2014, S86) noted, university IRBs may push this because “gaining consent often involves demonstrating researcher inclusion within the inhibiting security and logistical systems of a collaborating aid agency.” Yet, collaboration may involve significant tradeoffs because logistical support or funding often implies reciprocal obligations, and subsequent criticism can be seen as betrayal (Mosse Reference Mosse2006). When a researcher crosses the line between the academic and NGO sectors, the politics of research becomes even more fraught (Carpenter Reference Carpenter2012).

Internalizing Risk

The second strategy is internalization—the practice of accepting higher levels of risk and assuming the personal costs of doing so. This internalization may be dismissed as a normal part of “putting up” with poor conditions; however, we instead view it as a strategy whereby researchers make decisions that deprioritize their well-being in favor of pursuing research that they value. The personal costs and stresses of fieldwork have been discussed in the literature on the realities of research (Nordstrom and Robben Reference Nordstrom and Robben1995) as well as the techniques to manage these issues (Sluka Reference Sluka1990). We further identify internalization—that is, personal coping—as a common and integral risk-management approach in fieldwork. This is particularly true for PhD students and early-career researchers who have strong professional incentives to produce novel research but less access to the resources necessary to pursue outsourcing strategies.

The decision to spend time in hostile settings means accepting a degree of risk and assuming the personal cost—and even the “secondary trauma” of doing research (Wood Reference Wood2006, 384). At one end of the spectrum, harm can take the form of anxiety and post-traumatic stress.^{Footnote 3} Even in less extreme cases, researchers can experience feelings of isolation, including when they return, finding it difficult to share their experiences with colleagues and loved ones. These strained feelings can be further complicated by guilt. Green (Reference Green1999, 19–22) described the fear she endured while living in rural Guatemala despite the fact that she could leave if the situation required, which she noted with discomfort is unlike the population she was studying. How researchers respond to the risks they internalize subsequently influences the research product. As Shesterinina (Reference Shesterinina2019, 191) explained, both empathy and fear experienced by research participants “affect the dynamics of interaction, accounts of research participants, and ability of researchers to probe and interpret the accounts.”

The potential for both physical—particularly in terms of health—and emotional harm to researchers in the process of fieldwork has been recorded and increasingly discussed in the context of funders’ and academic institutions’ responsibilities to researchers (Bloor, Fincham, and Sampson Reference Bloor, Fincham and Sampson2010). This also has been an active concern within humanitarian-aid organizations, which increasingly have systems in place for “reintegrating” workers. It is thus notable that there remains little work on the effects of stress on research activities, design, and outputs. Research on conventional workplaces suggests that working under stress requires either increased resource use or a reduction in performance (Hockey Reference Hockey1997). What this dynamic looks like in a high-risk research setting merits further investigation. Beyond requiring researchers to avoid certain areas or activities, we see a need for more acknowledgment of the consequences and implications of risk internalization for both researchers and their work as well as openness to mitigating strategies such as team-based work.

Risk Imposition and the Ethics of Outsourcing

A core, starting observation is that outsourcing risk does not remove the risks of conducting research in dangerous settings. Although the risk is “managed” by the researcher, it may be shifted to the research assistants. Writing about her fieldwork in Burma with local research assistants, Boyden (Reference Boyden, Boyden and Berry2004, 240) reflected candidly on the discomfort of being simultaneously “dependent on others for security assessments” and “responsibl[e] as lead investigator for the safety of both researchers and the researched.” She expressed the conflicted position of being both powerful and vulnerable. However, the fact that it was her research that created that dynamic in the first place is not fully explored.

The core concern in the use of outsourcing is the imposition of risk by the principal researcher on the research assistant. This has been the subject of careful analysis in the ethics literature, particularly by Teuber (Reference Teuber1990) and Hayenhjelm and Wolff (Reference Hayenhjelm and Wolff2012). As Teuber (1990, 236) noted, we care if harm is due to natural causes or the actions of a person and, if so, whether those actions are deliberate: “From a moral point of view, the key question to ask is not ‘What should we do about risk?’ but ‘Which risks do we want to have on our conscience?’” Philosophers examining the ethics of imposing risk on others have focused on the importance of informed consent and compensation. Power asymmetries, however, can weaken these lines of reasoning (Smith Reference Smith1999) because contract researchers may feel compelled to take on assignments or find the risks acceptable due to economic precariousness, for example.

Another ethical issue that arises concerns who benefits from the research. Hayenhjelm and Wolff (2012, 26) argued that risk management has evolved around a consequentialist framework that balances costs against benefits. The challenge of this approach is “that the greatest risks would fall on those least able to influence the decision procedure or protect themselves in other ways. Hence, there is a worry that risk cost-benefit analysis will pile up risks for the vulnerable while the benefits accrue elsewhere”—namely, to the researchers and their university (Hayenhjelm and Wolff Reference Hayenhjelm and Wolff2012, 33). This analysis, which resonates with work on participatory methodologies and who benefits from research, suggests the importance of awareness of the researcher’s own positionality and the extent to which research assistants have a voice in the study’s execution.

These ethical questions are not immaterial, as underscored by the wounding of an enumerator in Afghanistan due to an improvised explosive device (Lyall, Blair, and Imai Reference Lyall, Blair and Imai2013, 8). Although these researchers took precautions and sought permissions from state authorities, local elders, and the Taliban, “[i]n some cases, having permission from the elders” and other authorities “did not protect the research team from threats by younger community members” (Osorio Reference Osorio2014, 3–4). Local researchers may experience some security advantages. At the same time, however, their participation in international research programs exposes them to increased risk, particularly of “transgressing political, social, or economic fault lines of which the researcher might not be aware” (Haer and Becher Reference Haer and Becher2012, 10). Again, we see the limitations of informed consent in risk imposition. Whereas the impact of research practices on subjects is a clear concern of research ethics review boards and fieldwork training, attention to the impact of research practices on hired assistants appears to be limited.

Knowledge Production and Bias

Outsourcing risk affects the way that research is produced. For researchers seeking to do large-scale survey research, particularly in higher-risk areas, outsourcing strategies are essential. Outsourcing reduces risk to the principal researcher while allowing the use of conventional tools; these studies can generate large datasets and engage populations outside of urban centers. These large-N studies tend to be expensive, particularly in higher-risk areas, and therefore are concentrated within institutions able to underwrite such studies or secure funding from entities such as the National Science Foundation, the US Department of Defense, and Homeland Security. Given the immense cost of undertaking this type of research, US government funding also tends to be concentrated in places viewed as high strategic priorities (e.g., Afghanistan and Pakistan). This raises questions about the role of these actors in shaping research agendas in areas of strategic importance as well as concerns about the deprioritization of important research on areas of less political interest.

Even projects that have ample funding and employ local emulators may be restricted by ground conditions, which produces sampling bias. Lyall, Blair, and Imai (2013, 7), for instance, noted that in their survey research, “[f]our [villages] proved inaccessible due to a combination of Taliban hostility, the presence of criminal elements and, in two cases, the inability of the enumerators to locate the selected village.” In another example of such research, Fair, Malhotra, and Shapiro (2014, 745) discussed how in their endorsement experiment on militancy in Pakistan, they were forced to address significant “safety and empirical issues…at the cost of precision about the variable being measured.” They stressed that “[g]iven the prevailing conditions in Pakistan,” the use of an endorsement experiment methodology was a necessary “tradeoff that must be made in order to study specific militant organizations, particularly in rural and economically underdeveloped areas” (Fair, Malhotra, and Shapiro Reference Fair, Malhotra and Shapiro2014, 745).

A final consideration with outsourcing strategies on knowledge production reflects familiar principal–agent problems. Due to lead researchers’ reduced ability to monitor survey results, solutions such as hiring local researchers also may cause or compound data-collection and quality issues. In an example from our own work, Roll hired two research assistants to collect additional survey data on former combatants living in western Timor-Leste for a survey of ex-combatants’ experiences with reintegration programs and postconflict trajectories. One researcher, however, falsified survey results—an effect that became clear through comparison with other sites. Discouraging data falsification and also identifying falsified results requires researchers to invest in time-consuming monitoring and data-checking procedures. Thus, survey outsourcing in difficult settings inherently involves high agency costs or increased risk of bad data. Such research is expensive and time-consuming even when the process goes well; when precautions do not work, it runs the risk of passing on invalid data as correct.