Reflections and recommendations on transitioning from pre- to post-disaster research

: Fieldwork often takes place in dynamic, uncertain environments. This is especially true of fieldwork in developing countries. Occasionally events can occur which have significant repercussions for ongoing research involving human participants. For example, political and social unrest, terror attacks, economic crises, epidemics, and natural disasters all have the potential to derail fieldwork plans and to radically alter the circumstances in which researchers operate. However, literature on how to anticipate and


INTRODUCTION
There has been little discussion of how researchers could and should respond when crises occur during ongoing fieldwork, despite data collection often being susceptible to disruption by exogenous events (Dominey-Howes, 2015). Potential for disruption is particularly great in research conducted in locations vulnerable to social and political unrest, and prone to natural hazards. Events such as terror attacks, protests, violent conflict, earthquakes, and tropical cyclones all have the potential to derail fieldwork plans and to radically alter the circumstances in which researchers must operate. As Hu (2015, p. 165) notes, such events can create an environment of "uncertainty, vulnerability, and complexity," greatly complicating data collection. Given the increasing volume of fieldwork that is now being conducted in such locations and the increased frequency of certain hazard events (NASEM, 2016), there is an urgent need to share learning and best practice so that the potential repercussions of research shocks can be better anticipated and navigated (Raven et al., 2018).
At present, a modest literature examines the challenges of conducting research in post-crisis contexts, but most of this discusses research that has been initiated post hoc (e.g. Adams- Hutcheson, 2018;Bell, 2009;Brun, 2009;Dominey-Howes, 2015;Hu, 2015;Miyazawa, 2018;Sloan, 2008;van Zijll de Jong et al., 2011). Literature addressing the particular challenges of transitioning from conducting pre-to peri-or post-crisis research when crises occur mid-fieldwork is much more limited. Raven et al. (2018) and Indah (2018) are among the few who discuss such experiences. In the former, the authors detail how studies of health

Accepted Article
This article is protected by copyright. All rights reserved workers in Sierra Leone and Nepal that had been ongoing were adapted in response to the West African Ebola outbreak of 2014 and the Nepal earthquake of 2015, to incorporate research more pertinent to the crises. They contend that the trusting relationships they had already established meant they were ideally placed to conduct such research and to navigate the emerging ethical challenges associated with it. Indah (2018), meanwhile, shares her experience of conducting fieldwork in Banda Aceh, Indonesia, when a tsunami early warning system was triggered by a strong earthquake. She discusses the dilemma she faced between satiating her academic curiosity in respect to the event and ensuring the safety of herself and her research participants. While these articles provide valuable insights and recommendations, the diverse forms that fieldwork and crises can take mean there is a need for more researchers to share their reflections on dealing with crisis disrupted fieldwork.
This article presents our personal experiences of transitioning from pre-to post-disaster research. Drawing on autoethnographical and reflexive methods, we explain how the 7.8M w earthquake that struck Nepal in 2015 profoundly affected our relationship with the community we had been working in, forced us to reassess our research agenda, and evoked challenging ethical questions in respect to our obligations to our research participants. In reflecting on our experiences, we hope to assist other researchers in anticipating the challenges they may face when crises occur, and to provide practical guidance around how to manage such situations.

THE PROJECT, THE EARTHQUAKE AND THE AFTERMATH
The original aim of our research was to explore the long-term combinatory impacts of ongoing social, economic, and environmental changes for smallholder communities in mountainous areas (Roxburgh, 2019). We planned to create a socio-ecological agent-based model of a specific Nepalese village that was experiencing stressors typical of many other mountain communities as a virtual laboratory for examining a range of potential future scenarios. This required us to carry out in-depth primary fieldwork in a case study village in order to develop the necessary level of understanding of the system needed to recreate its fundamental features in code. Specifically, fieldwork was to involve focus groups, participant observation, gathering detailed household survey data, and developing a village wiki. i The first phase of the fieldwork was carried out between 3 February 2015 and 22 April 2015 in the district of Dolakha, north east of Kathmandu. The village was home to fourteen

Accepted Article
This article is protected by copyright. All rights reserved households. Our relationship with the villagers was good from the outset. The majority seemed at ease talking to us, with many villagers enthusiastic for us to spend time with them. This was part driven by curiosity in us as outsiders, but there was also a genuine warmth and hospitality which helped us quickly feel at ease. Umesh, the only fluent Nepalese speaker in the research team, felt our decision to base ourselves in a local guesthouse and to eat with the villagers each day was particularly helpful in cementing our rapport with the community. Our up-front candour that participation in the project would not lead to direct benefits for the villagers helped manage expectations from the outset. On 22 April 2015, after concluding the initial fieldwork phase, we assured villagers we would return the following year to update them on the research and to collect more data to aid model validation. Our pledge to return was met well. One man in his 50s even said to us that it would not be right to exploit their generosity then not return.
On 25 April 2015, three days after we left the village for Kathmandu, a 7.8 M w earthquake struck, devastating a large swathe of the country, including our fieldsite. In the days immediately after the initial tremor, we attempted to make contact with the villagers at the study site, but without success. Only on the 18 May 2015 did Umesh finally receive word: a twenty-six-year-old research participant sent him a message stating that they had suffered no casualties, "but now haven't village." All of the homes had been razed, numerous sections of terracing had collapsed, several animals had died, the village's water supply had been damaged, and the road connecting the site to Kathmandu and other urban centres had been temporarily blocked by landslides.
Given the circumstances in the disaster afflicted region and that only Umesh remained in Nepal by the time we re-established contact, we felt it best to not return to the fieldsite immediately. Instead we opted to wait until the crisis was abating so that we would not interfere with relief work or add to the pressures the villagers were already facinga decision consistent with actions of Raven et al. (2018). We did, however, feel an obligation on humanitarian grounds to provide assistance to those who had hosted us in their community a community that had gone out of its way to assist us. Soon after re-establishing contact, we arranged for approximately $500 of food aid to be distributed to households throughout the village as well as to neighbouring communities. This helped alleviate some of the food stress that had arisen as a result of a food price spike following the earthquakes.

Accepted Article
This article is protected by copyright. All rights reserved We then returned for a week in March 2017, after a degree of stability been restored and reconstruction had commenced. The return visit provided an opportunity to validate aspects of our modelling work as we had originally intended. It was also a chance to check-in with the villagers socially, see how they were getting on, and assess how the events had affected finances, livelihoods, demographics, and village life more broadlyinformation that we then incorporated into the study. We did this through a combination of household visits and focus groups. The outcomes of our study are presented in Roxburgh (2019).

CHANGE IN RELATIONSHIPS AND POSITIONALITY
Fieldwork like ours, grounded in ethnographic approaches, typically involves spending extended periods in communities, observing social, cultural, and economic life (Dennis, 2009;Jeffrey & Troman, 2004). Many of the insights that we glean are only accessible thanks to the cooperation of the participating community. There are then additional factors which influence the nature of the information elicited and how it is interpreted by us as researchers.
The relationship researchers have with the communities they are engaged with and the positionality of the researchersboth actual and perceived (Turner, 2010) are therefore of central importance. Positionality recognises "we all speak from a particular place, out of a particular history, a particular experience, a particular culture, without being contained by that position" (Hall, 1992, p. 258). In other words, our identities and biographies affect how we see and make sense of the world and, by extension, how others view and interact with us (Ceaser, 2015;Moser, 2008;Raven et al., 2018;Sharma, 2018;Turner, 2010). Relationships and positionality can change over time (Miyazawa, 2018;Sharma, 2018). As we discovered during our fieldwork, disasters can have a profound impact on each.

No longer outsiders
By the time we finished the first phase of the fieldwork on 22 April 2015, Umesh had developed a warm relationship with a number of the villagers. Despite coming from a different part of Nepal, belonging to a different ethnic group, and not speaking the local dialect, he was very much accepted in the community and able to socialise with ease. Indeed, he was told by some of the men that they were proud of him as a young Nepalese man for assisting international researchers in their work, suggesting that at least on one level he had come to be seen as an insider. There was also warmth for the non-Nepalese research team members, though language barriers and our very different backgrounds meant the

Accepted Article
This article is protected by copyright. All rights reserved relationship was more based around mutual respect than personal closenesswe were still outsiders, albeit welcome ones.
When we returned to the village in March 2017, we assumed that we would have been byand-large forgotten given the events that followed our departure. However, we learnt that there had been a great deal of concern about our welfare after the earthquakes and many villagers were eager to exchange stories, aware we had been in Kathmandu at the time.
Nepalese or non-Nepalese, there was something of a reunion spirit; a camaraderie borne of shared experience that had not been there during the first visit. While the foreign members of the research team did not wholly transition to insider status, there was now a sense of common understandingthat we appreciated what they had been through and what they had lost, having also lived the initial events and knowing what life had been like beforehand. This contributed to participants opening up to us to an extent that we had not previously experienced, indicating that the earthquake represented an upward turning point in our rapport development with the villagers (Pitts & Miller-Day, 2007).

A new empathy
In the immediate aftermath of the earthquakes, we were extremely concerned about the villagers' welfare, exacerbated by our initial inability to make contact with them and reports of widespread devastation in Dolakha district. When we finally established that there had been no human casualties in the village but the impact had nevertheless been immense, our initial concerns took on more of an empathetic bent with a whole constellation of sometimes contrasting emotions triggeredrelief, sympathy, compassion, sadness, concern and grief (Batson, 2011). Because we had been able to share in the community's sense of place and peer into people's life worlds during the first phase of the fieldwork, and because some of us had also been caught up in the earthquakes, we felt a deep kinship with them. We could also empathise, to a degree, with the feelings of loss and unmooring that the earthquakes had shrouded them in.
This new empathy aided us in renewing our rapport with the villagers during the 2017 follow up visit, but more crucially, it helped us in navigating an uncertain research context in which we recognised participants would likely be hurting and vulnerable. In particular, it aided us in being sensitive without being patronising. We had a good sense of how to deal with the topic of the earthquake because we had already thought deeply about it ourselves and processed Accepted Article many of the same feelings the villagers felt. We had also spoken with numerous other people who had been through the events prior to our return, so we understood that people would have processed the events in different ways, and we prepared ourselves to be responsive to this. For example, we knew from our own experiences that many people are uncomfortable with being viewed as victims, yet others need their pain and loss to be recognisedboth perspectives being very much understandable.
Alongside empathy and emotional intelligence, personality also proved important. While Moser (2008) talks about the value of an extrovert personality, we found that a quiet, calm, and compassionate approach was helpful in our particular case given people's potential vulnerability and complex emotionsit was vital that we be sensitive and steer the conversations with care. That said, there were still many moments of joy and laughter, much valued by all parties. In the post-disaster circumstances, such times are all the more appreciated.

IMPLICATIONS FOR KNOWLEDGE AND UNDERSTANDING
The occurrence of the earthquakes, though horrendous for the affected communities, presented us with a chance research opportunityone that we were well positioned to take advantage of thanks to the work we had already done at our fieldsite.

Chance research opportunities in disasters
The original aim of our research had been to better understand the impact that multiple stressors could have on mountain communities in the period up to 2030 by creating an agentbased model of an actual mountain village and simulating a range of plausible stressor scenarios within it. During the initial fieldwork, we had learnt of a handful of social, economic, and environmental stressors that were particularly pertinent, so were intending to focus on these. While we were aware of the potential for seismic events in the region, we were not planning to incorporate them within the scenarios as we lacked the data needed to depict such events in a credible fashion. The last major earthquake in the region, which could not be recalled by the villagers, had occurred in 1934 when circumstances were very different (Whelpton, 2005). Furthermore, the other stressors felt more pressing at the time. However, the events of 25 April 2015 changed our judgement.

Accepted Article
This article is protected by copyright. All rights reserved We immediately recognised the relevance of the earthquakes to our research agenda. Not only did they constitute stressors themselves, it was clear they could affect how other stressors might be experienced. We therefore decided to incorporate them into the model scenarios. This allowed us to empirically validate our simulations against a trajectory grounded in reality and it enhanced the relevance of the research. The March 2017 fieldwork was therefore not only the team meeting their obligations to the community, but also an opportunity to gather the data necessary to depict the earthquakes in the model scenarios, bringing the events we had experienced into the core of the project. While the change in direction caused substantial disruption to our research timelines, we were able to glean insights into the impact of multiple stressors and, in particular, earthquakes, that would never have been possible had events not panned out the way they did. In particular, we were able to examine, in unprecedented detail, the socio-ecological repercussions of the events at multiple scales, and at a high temporal resolution, over a multi-year period.

Benefit of pre-disaster locational knowledge
Having a detailed understanding of how a community functions ordinarily is extremely helpful when it comes to establishing both the short-and long-run effect of a disaster, as is having access to pre-event baseline data for a comprehensive set of social and economic variables. Without this, it might be possible to get a rough qualitative sense of how things have played out and where things might be going, but quantifying impact, bounding trajectories, picking up on fine details, and assessing importance becomes much harder. In the absence of pre-event data, there is also much greater reliance on recall and backcastingmethods that can be particularly vulnerable to error (Bradburn, Rips, & Shevell, 1987). Preexisting knowledge of a location is of further value when it comes to understanding the right questions to ask, and it reduces the amount of information needed to establish situation understanding. As we had completed our initial data collection phase immediately prior to the earthquakes we had well-rounded and in-depth knowledge of the village as it was. The impacts of the events were therefore readily apparent, and villagers were able to reference our time with them when discussing what had changed.
The pre-event data allowed us to initialise our model to the pre-disaster context and then simulate the knock-on effects of the tremors over time, as well as conduct counterfactual runs in which the earthquakes did not occur. This enabled us to directly compare the two. While Accepted Article serendipitous, this highlights the potential advantages that come from being willing and able to pivot a project in response to real-world events.

Benefit of pre-disaster relationships
Our pre-disaster rapport with the community also helped us on our return in March 2017. The trust and friendships that we had previously developed meant we were enthusiastically welcomed back. The villagers felt at ease talking to us from the off, aiding us in eliciting the required information and our familiarity with the people and culture meant we had a good sense of how to be both respectful and sensitive when engaging with themimmensely valuable given the traumatic events that had occurred (Gaglio, Nelson, & King, 2006;Pitts & Miller-Day, 2007;Sloan, 2008). Furthermore, the care previously taken to manage expectations about benefits from research participation meant there was, as far as we could tell, no presumption that we would be providing aid or reconstruction support. That we lived frugally during our fieldwork stay, and travelled by bus rather than using private vehicles, perhaps also helped temper suppositions about our capacity to assist. Should other outsiders have entered the village in the post-disaster context, the expectations on them would almost certainly have been different. Indeed, researchers seeking to enter without offering assistance could well have been perceived as exploitative.
None of this is to say that researchers without prior relationships with crisis-afflicted communities should not seek to work in such contexts. However, they should be mindful of the elevated expectations that might be placed on them and the difficulties that can arise when seeking to establish trust and comprehension in a fast-moving environment where suspicion of, and frustration with, outsiders can be greater than normal. At our fieldsite, for example, several men expressed annoyance at how aid distribution was being managed.

ETHICS OF TRANSITIONING TO POST-DISASTER RESEARCH
Inevitably, transitioning from pre-to post-disaster research raises a number of ethical issues, which we discuss below.

Provision of aid to the fieldsite
Once we had re-established contact with the village and learned of the heavy toll the earthquakes had inflicted, our thoughts turned to how we could/should respond. Focusing

Accepted Article
This article is protected by copyright. All rights reserved initially on the near-term, we discussed whether or not we should provide aid. Research such as ours is only possible thanks to the generosity of communities like the one in Dolakha.
They welcome us into their homes, are charitable with their time, and willingly share their knowledge, thoughts, and experiencesknowing that they will not accrue direct benefits, monetary or otherwise, from participating. Preserving a culture in which people are willing to participate in research without expecting compensation is important because compensating participants can create perverse incentives and generate expectations that other researchers may not be able to meet (Hammett & Sporton, 2012). We therefore knew that providing aid might be viewed by some as a breach of research convention. However, these considerations felt subordinate to what we perceived as our humanitarian obligations, which is why we ultimately arranged delivery of approximately $500 of food aid. As Brun (2009, p. 197) states, "it is well established that all actors involved in a disaster play a role." As several of us were in Nepal at the time of the earthquakes and we had an ongoing relationship with the village, there was no doubt that we were "involved." We also knew we had the ability to coordinate aid delivery to the site without detracting from efforts elsewhere. The humanitarian imperative therefore applied to us. It would also have felt exploitative to accrue research benefits from the goodwill of the villagers, only to then abandon them in a moment of great need. We use the word "abandon" here because of the temporal proximity of our visit to the earthquakes, the extended period of time we had spent with the villagers, and the links we were planning on maintaining. Our decision to provide food aid was discussed with and approved by members of our Faculty ethics committee. The aid package was funded personally in the absence of agreement from our University to provide any financial support.
Arguments against providing aid include the logistical challenges in arranging relief packages; the limited financial means that researchers (and institutions) may have to fund such aid; fear of obligation creep; and concern about tainting future research data through intervening in community affairs (Dennis, 2009). These issues nevertheless seemed neither insurmountable nor of greater weight than our perceived humanitarian and ethics obligations.
On our return we did not sense that expectations had been raised, nor did it seem that people felt an obligation to engage with us. Gratitude was expressed to us by a number of the villagers, but no requests were made for additional material or financial support. The work that Umesh had previously done to establish expectations about what we could and could not offer likely helped, and we took great care to stress that there was no obligation to participate in our 2017 research as a result. The aid did perhaps help cement our bond with the villagers, Accepted Article but it was our shared experience of the earthquakes that seemingly had the more profound impact on our relationship.

Discussing the disaster
Prior to returning to the village in March 2017, we decided to focus on establishing how village life had changed over the previous two years and how the villagers saw things developing in the years ahead, rather than discussing the earthquakes directly. We were concerned that raising the events in conversation could trigger traumatic memories. In practice, however, villagers frequently brought up the topic themselves, both socially and in interviews, often in the form of personal vignettes. This was perhaps inevitable. The repercussions of the events were so great that few topics could be legitimately divorced from them. It became clear that even prior to our return, it was a common theme in village conversation. The reminders were so pervasive that hiding from it was clearly not an option.
Instead people seemed to have consciously or otherwise opted to take control of the narrative.
Ming Hu (2015) similarly found that people were keen to share their stories during his fieldwork following the 2008 Sichuan Earthquake.
We primarily encountered stoicism and acceptance. This was perhaps a result of the fatalistic attitude that was common in the village and which pre-dated the earthquakes. Life frequently throws up challenges in rural Nepalchallenges that rural communities usually have little control over but must deal with. Therefore, to an extent, people were already psychologically primed. That said, we did occasionally encounter lament and sorrow, but even then, there was a desire to talk and people seemed in need of an empathetic ear. Although discussing traumatic events is often assumed risky, our experiences chime with those of other researchers who found the process to be cathartic for their interviewees. Studies have shown that while subjects in post-disaster research can experience short-term distress during their participation, most individuals ultimately perceive the benefits of their participation as outweighing the costs (Collogan, Tuma, Dolan-Sewell, Borja, & Fleischman, 2004;Newman & Kaloupek, 2004;Parslow, Jorm, O'Toole, Marshall, & Grayson, 2000). In many respects, our experience demonstrated that you cannot presume how people will respond to disasters (Carlin & Park-Fuller, 2012;Hu, 2015).

Accepted Article
This article is protected by copyright. All rights reserved

Risk to researchers and fieldwork assistants
When conducting research on potentially sensitive topics, efforts are made to assess the possible risks posed to research subjects and to minimise these through careful research design, contingency planning, and ongoing vigilance. Rather less consideration is usually paid to the possible psychological harms that researchers and field assistants may experience as a consequence of exposure to distressing events, direct or indirect (Dominey-Howes, 2015). The topic is often absent from fieldwork risk assessment protocols despite explicit requirements to consider environmental risks and potential for violent encounters. Yet, the injury, death, trauma, destruction, and loss that may well be witnessed is unlikely to be without impact (Hutcheson, 2013). Indeed, Dominey-Howes (2015), Miyazawa (2018) and Indah (2018) all reflect on the trauma they experienced during fieldwork in post-disaster contexts.
Two of us were present during the earthquakes of April 2015. We witnessed buildings pancake, the dead and injured arrive at a city hospital, cries of horror from bystanders, and mass erection of temporary shelters where tens of thousands spent the subsequent nights. The rest of the team observed from afar, watching powerlessly as news trickled in, unable to contact those in Kathmandu. Thankfully, none of us experienced any lasting trauma, and our return visit, while evoking a mix of emotions, also had no lasting detrimental effects. The experience nevertheless raises some important considerations. As Dominey-Howes (2015) observes, postgraduate and early career researchers can be particularly vulnerable in situations like these due to financial inaccessibility of counselling services, lack of paid sick leave, insecure contracts, and other pressures that come with academic life. Field assistants are likely to be left in an even worse position. Our experiences highlight the need for institutions to take more proactive responsibility for the care of those affected when the worst does happen, as this can affect future relationships with research assistants, as well as the overall well-being of research staff and students.

CONCLUSIONS
Crises can disrupt fieldwork plans and transform the circumstances in which researchers must operate, yet the literature on navigating such events is limited. This makes it difficult for researchers caught up in crises to determine how they should respond. In this paper we have shared our personal experiences of transitioning from pre-to post-disaster research. In

Accepted Article
This article is protected by copyright. All rights reserved particular, we have outlined how the earthquakes that struck Nepal in 2015 affected our relationship with our research participants, how the events influenced our research agenda and findings, and how they prompted ethical dilemmas. Not all aspects of our experience will be applicable to other crisis-afflicted research projects, and we acknowledge that even events that are similar on face have the potential to play out very differently. However, we hope that our reflectionsalongside the insights of others on this topiccan at least reduce some of the uncertainties and help inform decision making. With this in mind we suggest that researchers who are planning data collection in areas susceptible to crises, or who are transitioning from pre-to post-disaster research, give consideration to the following: 1. Where possible, ensure there is scope to pivot when planning research and fieldwork, and be open to potential changes in direction so that chance research opportunities can be harnessed (Raven et al., 2018).
2. Prior to commencing data collection in post-crisis contexts, seek situational awareness from local colleagues, confidantes, and other sources. Read accounts by others who have worked in similar circumstances to aid preparedness.
3. Previously established relationships with participants could well be affected by crisis events. Recognise that expectations placed on research may have shifted and that other aspects of engagement may also be affected. Clear communication is important in such cases.
4. If conducting fieldwork in a crisis-afflicted area that you have not previously visited, be mindful of potentially elevated expectations and the difficulties that can arise when seeking to establish trust and comprehension in a fast-moving environment. Suspicion of and frustration withnewcomers may be greater than normal. 5. People respond to crises in different ways. Avoid making assumptions about how people will behave, and display empathy.
6. Fieldwork risk assessments should ensure explicit consideration of psychological risks, while institutional assurances are needed that procedures are in place to mitigate these.
Where such risks are deemed high, researchers should be provided with training to recognise symptoms of trauma in both themselves and others.

Accepted Article
7. Open discussion is needed around the obligations on researchers and institutions when research participants and field assistants are caught up in crises during ongoing fieldwork.
8. Institutions should ensure appropriate support is available (e.g. counselling, paid leave) for researchers traumatised during fieldwork at all stages of their research career.

NOTES
1 A wiki is a publication format that facilitates the collaborative documentation of knowledge. We used it during the fieldwork to manage information about a wide range of topics related to the village as we gathered our data. Wikis are particularly suited to such tasks as they can readily handle "non-linear, evolving, complex and networked text" (Wikipedia, 2016). The format had the additional benefit of facilitating participatory involvement in the knowledge documentation process as the literate villagers could personally review and suggest edits to the topic articles.