Researching with ‘Local’ Associates: Power, Trust and Data in an Interpretive Project on Communities’ Conflict Knowledge in Myanmar

ABSTRACT This article discusses benefits and challenges of qualitative-interpretive research conducted in teams of outside (Northern) researchers and national (Southern) associates, in which the latter have considerable autonomy over research design and data generation. Reflecting on our collaboration with Burmese associates on arts-based workshops with violence-affected communities in Myanmar, we discuss how structures and dynamics of power and trust-building shaped the research process and data interpretation. Our reflective analysis suggests that interpretivist research ‘by proxy’ is possible and can be highly enriching but depends upon sufficient time (and funding) for meaningful, long-term engagement with ‘local’ research collaborators, which our project lacked.


Introduction
The role of 'local' brokers or assistants in fieldwork-based conflict research conducted by outside (often Northern) researchers has recently received welcome attention (e.g., Eriksson Baaz and Utas 2019a). Traditionally a concern in the discipline of Social Anthropology (e.g., Hoffman and Tarawalley 2014), a gradual increase in fieldwork-based conflict research in other disciplines over the last decades has led to more scholars working with research brokers or assistants based in the countries they study. Heightened risk aversion among Northern universities (Russo and Strazzari 2020) has furthermore meant that, beyond logistically facilitating conflict research, Southern research assistants increasingly conduct research as a 'remote method' for their Northern colleagues. Both tendencies have made critical assessments of the ethics and practicalities of such researcher-assistant relationships ever more pressing (cf. Cronin-Furman and Lake 2018, Eriksson Baaz and Utas 2019b). Reflections on these usually asymmetric relationships have revolved around questions of researcher vs. broker access and security in conflict contexts (Eriksson Baaz 2019, Bøås 2020, Sangaré andBleck 2020); inequalities and dynamics of power, positionality, privilege and profits (Cronin-Furman and Lake 2018, Bouka 2019, Parashar 2019, Clausen 2020; and brokers' usually substantive but unacknowledged impact on research data and results (Käihkö 2019, Utas 2019, Eriksson Baaz and Utas 2019b. There are hints in the literature that different epistemologies and methodologies have an impact on the quality of the researcher-broker relationship and the meaning given to its impact on research outcomes. Maria Eriksson Baaz and Mats Utas (2019b, p. 159), for example, note that reflections on brokers are less common in quantitative conflict research than qualitative projects, which suggests that the role of brokerage may not be seen as particularly problematic by quantitative researchers. They also hint at the role which '(different) epistemological positions as researchers, but also the varied aims of . . . research projects and . . . truth claims' have in judging research relationships and data in the field (Eriksson Baaz and Utas 2019b, p. 170). So far, however, the relation between specific types of research and researcher-assistant relationships has not been foregrounded in the literature. This article seeks to start fill this void by reflecting on the three areas of concern identified by the literaturesecurity/ access, power/inequalities, and broker impact on outcomesin the context of a specific approach, namely qualitative-interpretive research.
At the heart of this article is the question of benefits, challenges and limits of qualitative-interpretive conflict research 'by proxy', that is, research in whichfor methodological or security/access reasonslocal research associates hold a central, relatively autonomous role at the research design and data generation stages. We focus on how dynamics of power and trustbuilding can shape the relationship between outside (Northern) researchers and their national (Southern) research associates, and how these dynamics impact upon the qualitative data generated and the outside researchers' ability to interpret it. While power dynamics in researcher-broker relationshipsfrom the power hierarchies based in structural inequalities to the capillary flows of power in researcher-broker interactionshave been highlighted by other authors, we specifically ask after the impact power has on research findings in qualitative-interpretive projects and show how this impact is mediated by processes of trust and mistrust in the research partnership and consequently the data. We argue that interpretivist research 'by proxy' is possible and can be highly enriching in terms of research data and findings, but that for such a positive outcome trust is central: successful interpretivist research 'by proxy' is dependent upon sufficient time and funding to allow for a qualitatively deep, long-term engagement with the 'local' research collaborators and the field to mediate the negative effects of dynamics of power and related sources for mistrust in North-South research collaborations.
Empirically, we draw on our experiences of collaborating with three Burmese research associatestwo artists and a businesswoman who are also civil society activists 1on two arts-based workshops to explore experiential conflict knowledge among participants from violence-affected communities in Myanmar's Kachin and Rakhine states. The research collaboration was part of the AHRC-funded project 'Raising Silent Voices: Harnessing local conflict knowledge for communities' protection from violence in Myanmar', an 18-month research taking place in 2016-18. 2 The project had two central objectives. The first was to develop creative ways for conflict researchers and international nongovernmental organisations (INGOs) to access local knowledge in Myanmar's access-restricted conflict zones. The second objective was to explore whether and how this knowledge could be useful in the work of the project's partner INGO, Nonviolent Peaceforce Myanmar (NPM). 3 Regarding the first objective, within which this article is located, the project centrally relied on two drawing workshops designed to explore people's life experiences, sense-making, protection needs and peace hopes in the midst of violent conflict through their own images and words. The collaboration with Burmese research associates and their relative autonomy was essential to the qualitative-interpretive design, implementation and interpretation of these workshops and the data generated, but also answered to the semiauthoritarian context of research in Myanmar, in which such remote research methodologies help circumvent access restrictions for foreigners.
When using the pronouns 'we/us/our', this article refers to its authors, who are from/based in the Global North (UK and US) and who led the project. Through recurrent reference to a chapter by Katarina Kušić (2020), who was not involved in the project but interviewed us UK/US-based researchers and our Burmese associates for a meta-study on fieldwork-based interpretive research 'by proxy', we also take the Burmese associates' assessments and perceptions of our collaboration into account. While this is not the ideal of coauthored research outputs, in this specific case other projects and life circumstances led our Burmese associates to rather concentrate on their careers in the business and arts sectors. Equally, not naming our colleagues in Myanmar in this article has been a joint decision for reasons of anonymity, not an attempt to erase their voices or work from this research. When referring to them, we use the term 'research associates' or 'collaborators' to mark the difference between the positive, central role we envisaged for them and the sometimes ambiguous associations with the terms 'broker' and 'fixer' found elsewhere, perhaps epitomised most iconically in Joe Sacco's (2009) graphic novel The Fixer (cf. Käihkö 2019). This is an acknowledgement of our Burmese colleagues rather than a descriptive specification of their input. Especially the businesswoman's role also involved a range of typical brokerage tasks such as arranging and interpreting meetings. The Myanmar team, in turn, used local brokers in Kachin and Rakhine to recruit research participants and assist with workshop logistics. Yet, our associates' roles also explicitly included the design and implementation of the art workshops and contributions to data interpretation and to the dissemination of research findings, which are central to the reflections in this article.
In the following, we first briefly describe the project background against which our research collaboration unfolded. Specifically, we introduce its qualitative-interpretivist methodology, the challenges faced by outside conflict researchers in the closed context of Myanmar, and the type of collaboration these two contexts led us to pursue in our project. We then employ economic geographer James T. Murphy's (2006) multi-scalar heuristic model for the study of trust-building processes to trace dynamics of power and trust/mistrust in our research collaboration. Focusing on processes located at and across macro (structures, context), micro (subjective dispositions, expectations) and meso (intersubjective) levels of engagement, this model is useful to unpack the links between hierarchical structures and multi-directional flows of power, and dynamics of trust/mistrust. This helps, we suggest, to prevent simplistic representations of collaboration relationships as either exploitative/mistrusting or mutually beneficial/trusting found in parts of the existing literature, and rather directs attention to the complexity and evershifting, evolving nature of social (research) relationships. In the third section, we explore how these structures and processes of power and trust-building affected the type and quality of data that our collaboration generated in the context of our project during the stages of design, implementation and interpretation. We conclude that the data generated through this process far exceeded what was possible otherwise, and yet it was also different from the original research conceptualisation and intention. Issues of power and trust played out in complex ways, influenced the resulting data and limited the Northern researchers' ability to interpret it. In the conclusions, we suggest that despite challenges particular to qualitative-interpretive conflict research, on balance interpretive research 'by proxy' is possible. What is more, not only does it promise to enable data generation in access-restricted field settings such as Myanmar, but also to considerably enrich the hermeneutical data and its interpretation. Yet, these positive effects can only develop fully if the research conditions and resources allow for long-term engagements.

Interpretively Researching Conflict Knowledge among Communities in Myanmar
In 2017, our project's three Burmese research associates held arts-based workshops with participants from Myanmar's conflict-affected Kachin and Rakhine states. The workshops employed drawing as a hermeneutical method to generate qualitative data about how people living in these conflict zones experience and make sense of their lives amidst armed violence. The workshops were entirely led by the Burmese collaborators with a high degree of autonomy in terms of design and implementationwith (virtual) support, but without physical presence or intellectual control, by us Northern researchers leading the project. There were two reasons for this set-up, one methodological, the other practical.
In deciding to explore the violent conflict in two states of Myanmar through the experiences and knowledge held by members of violence-affected communities, we chose a hermeneutical approach to qualitative-interpretivist research which aimed to avoid imposing preconceived categories and concepts onto local meaning-making (Schwartz-Shea and Yanow 2012, Bevir and Rhodes 2016). Interpretivism is not only interested in an individual's political attitude or behaviour, but in the worlds of meaning through which individuals make sense of the political. For instance, for interpretivists it is only through the study of meaning in action that we can appreciate the wide range of experiences a seemingly simple category such as 'insurgent' or 'soldier' entails (cf. Cheesman 2017aCheesman , 2017b. Such focus on meaning-making involves an awareness that common academic concepts may not be meaningful in specific contexts or that they can be filled with a different meaning to the one assumed. For many Kachin people, for example, 'peace' is a term negatively associated with the ceasefire agreement between Myanmar's military Tatmadaw and the insurgent Kachin Independence Army (KIA) from 1994-2011. This period saw a reduction of armed violence but also an increased exploitation of Kachin resources by Burmese and foreign investors and a de-facto Burmanisation of the state through internal migration of members of the Bamar majority from Myanmar's poor central regions to Kachin state for work (Sadan 2016). It is due to such inappropriateness and/or polysemy of concepts that interpretivists speak of concepts emerging from context and call for an abductive logic of inquiry, in which the researcher goes back and forth between their empirical data and the concepts and theories used to make sense of it (Schwartz-Shea and Yanow 2012).
In order to deliver on its methodological premises, interpretivist fieldwork necessarily involves the researcher's immersion and reflexivity (Kurowska and Bliesemann de Guevara 2020). In the context of our project, however, the possibilities for immersive fieldwork were limited. While all authors had fieldwork experience in conflict contexts, only one had worked on Myanmar before, but only in the form of a desk review for project partner NPM; none of them speak a local language; and the short project duration and limited financial resources provided little possibility for prolonged fieldwork immersion and cultural learning, let alone attempts at a deeper ethnographic engagement to reap the 'benefits of long-term fieldwork' (Millar 2018).
Additionally, access restrictions by Burmese authorities render the conduct of community-level fieldwork in Myanmar's conflict zones difficult for outside researchers without some form of local brokerage. Despite the country's democratic opening since 2011, Myanmar's military remains powerful through constitutionally guaranteed prerogatives (Croissant and Lorenz 2017). One practical consequence is that foreigners' movement is monitored, controlled and restricted to various degrees. While foreigners are free to travel in large parts of the country, physical access to some (especially conflict) areas is tightly restricted for any foreigners, including humanitarian and development workers, journalists and researchers. 4 This rules out embedded research with humanitarian or aid organisations as access strategy for these areas (Lewis et al. 2019, cf. Peter 2020. Searching for alternative forms of access to conflict-affected communities, some researchers enter areas controlled by insurgent armies via neighbouring countries and with the consent of the respective armed group(s) or other local authorities (cf. Cronin-Furman and Lake 2018, pp. 607-608, Lertchavalitsakul 2014). This option was unfeasible for our project not only because it would not have received approval by the project-hosting universities' research ethics panels, but also because the project explored methods of knowing conflict-affected communities for its project partner NPM, which as an INGO is also access-restricted. Working 'remotely' through collaborators of Burmese nationality who are not subject to the same access restrictions and/ or can more easily 'blend in' was therefore the only viable alternative (cf. Myrttinen and Mastonshoeva 2019).
We chose to work through arts-based research shown to help explore participants' meaning-making (Mannay 2016), and with Burmese associates, to counter these various challenges to our interpretive research. For convenience, we decided to adopt a drawing method one of us had helped develop and pilot in a different research context (Gameiro et al. 2018). Yet we also agreed to give our associates wide-reaching autonomy in the adaptation of this method to the context of Myanmar to account for its (cultural, social, political, etc.) specificities. Different from local research assistants in qualitative-positivist research designs hired for bespoke tasks such as conducting household surveys or structured interviews designed by the researcher (e.g., Haer and Becher 2012, p. 10), we thus opted for a co-designing process which would give a central say to our associates in how the workshops would look and be implemented. Consequently, 'training of local researchers', as promised in our grant proposal, did not consist in a systematic exposure to Western research concepts and methods in preparation for a predefined task. Rather we used a reflexive methodology of ongoing face-to -face (when in Myanmar) and face-to-screen conversations among the core team and with the Burmese associates, individually and as a group, of what we were doing, how and why.
Despite these conscious choices, questions of power and trust/mistrust interfered with our intentions of establishing, as far as structurally possible, equitable relationships between us Northern researchers and our Burmese associates. As our research unfolded, we kept coming back to the question of what the quality/meaning of this interaction was and how it affected the interpretive data we were generating. Overall, we developed basic trust in each other and sufficient agreement to proceed with the planned research. Room for improvement notwithstanding, '[t]he whole research team evaluated the project as a success' (Kušić 2020). Still, despite our attempts to break out of the trap of 'neocolonial domination in research' (Bishop 2005), power structures continued to frame the collaboration and generated specific individual expectations and assessments, which in turn created ever-shifting dynamics of trust and mistrust in the research interaction among and between the two 'teams'.

Dynamics of Power and Trust-building in North-South Research Collaborations
The existence of power dynamics between researchers and researched and their influence on research relationships, data and outcomes is not a new observation. Especially feminists have cautioned us to acknowledge that knowledge is 'not just "out there", but the result of a particular engagement in a particular context as a continuous way of "becoming"' (Davids and Willemse 2014, p. 2; cf. Ackerly and True 2008). Power in research relationships is most visible in the common absence of the participant-subjects from the processes of design and implementation of research about them (Janack 1997). Feminist approaches have explicitly worked to counter these effects of epistemic privilege (Bar On 1993, Patai 1991 and to break the assumed hierarchy between 'expert' and 'experiential' knowledge , including in the study of peace, conflict and war (Wibben 2016, McLeod and O'Reilly 2019)an aim they share with many interpretivist approaches (Schwartz-Shea and Yanow 2012, Fujii 2018). While less often addressed in this literature than the relationship between researcher-subjects and participant-subjects, questions of epistemic privilege obviously also affect the relationship between foreign/Northern researchers and national/Southern associates (Lottholz 2018).
In the research brokerage literature, questions of power are often interlinked with questions of marginalisation and exploitation of research collaborators, grounded in the relational asymmetry between resource-rich Northern researchers and resource-poor Southern collaborators. In many post-conflict societies, the resulting commodification of research brokerage has led to an outright fieldwork industry, with methodological consequences for access to and quality of information (Schiltz andBüscher 2018, Käihkö 2019). In conflict settings, the strong resource-asymmetry between Northern researchers and Southern collaborators may incentivise decisions and behaviours which put collaborators and participants at risk of physical harm or repression (Cronin-Furman and Lake 2018, Bøås 2020). Ideas and narratives about who is at risk in the first place are furthermore strongly gendered and racialised, leading to a further precarisation of Southern collaborators (Eriksson Baaz 2019). Additionally, the systematic erasure of Southern researchers from written outputs constitutes a form of structural violence based in an unequal granting of intellectual property rights to Northern researchers and their Southern partners, which goes beyond questions of ethics in that it cements structural inequalities (Bouka 2019).
The studies cited above highlight the need for a heightened awareness of the consequences of structural power and inequality for researcher-broker relationships and for practical guidance on how to tackle both, power asymmetries' consequences and the structures and practices that re-/produce them (e.g., Cronin-Furman and Lake 2018, p. 612). Yet, the literature also suggests that the everyday power dynamics in researcher-broker relationships in the field are more complex and fluid due to the shifting positionalities of Northern researchers and their Southern partners in specific research contexts and situations (Eriksson Baaz and Utas 2019b, p. 171). Financial asymmetries may lead to mutual benefits rather than one-sided exploitation (Käihkö 2019). Lack of local or situational knowledge about a conflict region may put the foreign researcher at higher physical risk than the local broker (Sangaré and Bleck 2020). Services like interview interpretation or document translation may give local assistants and partners a considerable say over how data is understood and analysed (Schiltz and Büscher 2018). These examples point to the need for more complex analyses able to capture both, the structural asymmetries and their effects, but also the dynamics and multidirectional power flows that shape the everyday functioning of research collaborations within these structural frames. Based on our experience in qualitative-interpretive conflict research, we suggest that what is at stake in these power asymmetries and dynamics is the question of trust, which mediates between the researcher-broker relationship and its effects on the generation and interpretation of research data.
Trust is usually discussed in relation to research participants (rather than collaborators), for example concerning its role in researcher access (Norman et al. 2009, Lewis et al. 2019) and interviewing (Fujii 2010). For Romain Malejacq andDipali Mukhopadhyay (2016: 1017; cf. Parashar 2019) trustbuilding with a network of informants in a conflict zone is the central element to what they call 'field research as tribal politics': 'Trust is the currency by which fear can be overcome, entry facilitated, and relationships built.' Nerve Macaspac (2018) reminds us, however, that trust-building in the field can be out of the researcher's control, as it relies on perceptions and positionality. As a researcher of Filipino nationality based at a US university, Macaspac faced 'double suspicion' when doing research in his country of origin: for his Filipino interlocutors he was an 'outsider' and hence suspicious; for the international community of scholars he was 'too close to [his] research subjects' and hence not 'objective' enough (Macaspac 2018, pp. 10-13, cf. Clausen 2020. In this case, global power structures thus interfered in fieldwork interaction mediated by perceptions and limited his ability to shape trust-building both 'in the field' and 'back at the desk'. When it comes to the research brokerage literature, trust and mistrust are less often discussed and mainly concern questions such as the reliability of interpretation/translation services or informants. Schiltz and Büscher (2018, p. 136), for example, relate that some brokers they interviewed in northern Uganda reported to sometimes conceal information from researchers, or to actively convince individuals to participate in a survey or interview, 'in order to avoid disappointment and not to miss out on opportunities for themselves'. Due to a diffuse or acute awareness among Northern researchers of the effects of the structural inequalities inherent to research relationships between them and their Southern assistants, the trustworthiness of the latter is often in doubt. In qualitative-interpretive research, 'where the researcher is a supplicant who learns with others, rather than an expert who tests on others' (Kurowska and Bliesemann de Guevara 2020), dynamics of trust/ mistrust have a specifically profound impact that goes beyond the question of whether a piece of information is trustworthy. When interpretivists work directly with their participant-subjects, unhinging moments of blunder or meta-data such as lies, rumours and silences (Fujii 2010) are opportunities to observe meaning in action and discover properties of the underlying social order (Kurowska and Bliesemann de Guevara 2020). Yet it remains largely unexplored how such moments involving shifting dynamics of trust/mistrust inform the process of data generation and analysis in collaborations where the interaction between outside researchers and participants/data is mediated by research associates.

Tracing Power and Trust-building in Our Research Collaboration in Myanmar
In order to trace the structures and dynamics of power and trust-building in our North-South research collaboration, we use James T. Murphy's 'process-, power-, and context-sensitive conceptualisation of trust' originally formulated for the study of trust in global economics, which draws 'theoretical and methodological links between the micro-social interactions that constitute economic relations and the meso-scale and macro-scale phenomena characterising the diverse geographies of the world economy' (Murphy 2006, p. 429). Murphy sees trust not as a state, but as a dynamic process which takes place within, and interacts with, structures and relations of power at different scales (Murphy 2006, pp. 429, 440-442). In the following, we employ Murphy's model as a heuristic to unpack the relational dynamics of international cooperation and exchange in North-South research collaborations that draws attention to the processual, power-imbued, context-specific and scalar ways in which such relationships unfold, before addressing the question of how these relationships can in turn influence data collection, interpretation and findings of qualitative-interpretive conflict research.

Macro-scale: Research Collaboration in an Institutionalised North-South Context
At the macro-scale, trust-building processes and power flows are influenced by normative and regulative factors, positionality, and the current structural circumstances of the relationship in question (Murphy 2006, pp. 441-442). The most important normative and regulative factors influencing our research collaboration consisted in the terms set out by the funder, institutional research ethics and risk assessment requirements of the projecthosting UK universities, and the values and epistemological orientations with which we Northern researchers entered the research relationship. Power was exerted through grant requirements and institutional regulations but subsequently also the choices we made of how to meet them. For example, the funding terms defined distinct roles and responsibilities for the different parties involved in our collaboration, which in turn created power hierarchies. Only those whose names were known from the outset, i.e. the Northern researchers, were considered 'investigators' and eligible for salary costs. Our Burmese collaborators, who due to the explorative project nature we did not know at the time of applying, received daily fees for their services of 'workshop facilitation', 'interpretation' or 'transcription' out of the project's travel, subsistence and consumables budgets at our discretion. 5 This reduced them formally to service providers and cemented the formal hierarchy built into the project design.
In our case, the funder wielded a significant power that shaped what was possible. While there are now a range of funding schemes that emphasise equitable research relationships, e.g. by building on mixed principal investigator teams from Northern and Southern countries who hold distinct budgets, the specific call we had successfully applied for institutionalised unequal/hierarchical formal relationships between Northern researchers and Southern collaborators. This is also true for the ethics and risk assessment procedures the project underwent. These were geared towards the safety of research participants (principles of informed consent, confidentiality) and the Northern researchers (travel risk assessments based on UK Foreign and Commonwealth Office advice), but did not specifically ask for ethics or risks involved for any 'brokers', 'assistants' or unnamed 'collaborators' on the project. We addressed questions of research associate and participant safety and ethics in an online training between the principal investigator and the Burmese team. We acknowledge that research ethics are assumed to extend to all research interactions, not just participants in surveys or interviews; however, strictly speaking there was an institutionalised hierarchy of risk-bearing and care, which both exerted greater control over but also led to safer positioning of the Northern researchers (cf. Wong 2010, Heathershaw andMullojonov 2020).
We were aware of and worked actively to ameliorate some of these institutional hierarchies. For example, the project took place against the background of a history of British colonial power in Burma and a global economic system in which the UK and US are donors of official development assistance (ODA), while Myanmar is among the least developed countries receiving ODA. This global hierarchy crystallised in pay levels of the different project members and the distribution of official accountability for project outcomes and finances, which lay firmly in Northern hands although the Burmese collaborators felt highly responsible for the successful implementation of the workshops including budgets (Kušić 2020). Despite the Northern researchers' active efforts to ameliorate power and work against exploitative research relationships, the questions of who would benefit (most) and how, did come up in initial conversations. In a meeting with the two Burmese artists to establish the collaboration on the drawing workshops, one of them raised the question of attribution of research inputs, recalling a previous experience with an Asian researcher with whom they had generously shared information and contacts in Myanmar's art world, without ever being acknowledged. Their critical position and the ensuing frank discussion of our power and intentions contributed to trust-building among us (cf. Kušić 2020).
Another critical factor affecting trust on the macro-scale is the political context and, more specifically, the general trust 'climate' (Murphy 2006, p. 442). In Myanmar, a gradual process of democratic opening since 2011 has made trusting research collaborations between foreign and Burmese researchers easier than they had been in Myanmar's authoritarian past (Brooten andMetro 2014, Matelski 2014). Yet, Myanmar's political system has retained authoritarian elements and trust-building with participants in conflict-affected states remains difficult despite an improved freedom of speech, described by a Kachin civilian protection volunteer like this: Fear was not the same in the past as it is in the present. In the past, we feared but did not have the chance to talk about anything. [. . .] Now we can speak freely but we still have to be afraid of being harmed. 6 Our Burmese associates seemed much less concerned about state repression, as visible not least in their unconcerned and frequent use of Facebook to publicly voice political opinions. The differences between them and the Kachin interviewee can be explained by age and life circumstances. The older Kachin man encountered violent conflict and state repression in his past and current everyday life in Kachin state. Our much younger research associates, albeit from similarly conflict-torn family backgrounds, are based in cosmopolitan Yangon, and social interactions with artists, peace activists and the international community in Myanmar as well as frequent travels abroad had created very different perceptions of the general trust climate in the country when we started our project in 2016. It was only in later 2017, after the tense situation in Rakhine state had escalated to mass violence, that the trust climate became a topic of discussion with one of our associates whose family is from Rakhine. She was visibly more wary to discuss politics in public; yet, other than relocating our political discussions to private spaces, changes in Myanmar's general trust climate did not seem to majorly affect our research relationship.
What is more, the ability of our Burmese collaborators to travel to places restricted to foreigners, together with their knowledge of languages indigenous to Kachin and Rakhine as well as Burmese and their rootedness in local cultures, positioned them to be more capable to navigate the fieldwork's political context and thus more powerful than we Northern researchers were in the actual data generation, which in the planned form and locations would have been impossible without them.

Micro-scale: Expectations, Dispositions and Uncertainties in Research Collaborations
At the micro-scale, trust-building processes are influenced by collaborating subjects' social dispositions, their expectations deriving from factors such as their interests or the potential rewards of trust-building, their analysis of the risks and uncertainties involved in extending trust to others, and their perception of power of their own and others' control over the situation (Murphy 2006, p. 441). Among us Northern researchers, there was a general willingness to build trustful relationships. Years of engagement with actors in conflict zones, a feminist and/or interpretivist sensitivity towards these actors' experiential-practical knowledge, and the principles of unarmed civilian protection work (the professional background to two of us) which revolve around relationship-building and the primacy of local actors (Furnari 2015), were factors that contributed to a general trusting disposition. This was strengthened by the rewards to be expected from a trusting, power-sharing collaboration in view of the project's practical constraints and interpretive research design. Without Burmese associates the project objectives would have been in question. The project's limited length and funding provided a further strong incentive for a leap of faith in order to implement as much of the original interpretivist project idea as possible, which outweighed potential risks. These included whether the Burmese associates were equipped to implement the workshops in line with the project's interpretivist methodology and what kind of data the collaboration would ultimately generate. The Northern researchers made a conscious choice to give up control over the research associates, which nevertheless did not entirely undo the (structural) power involved in the collaboration (Kušić 2020).
On the side of our Burmese collaborators, payments and potential benefits deriving from working on an international research project (e.g., involvement in follow-on projects) certainly constituted a strong incentive to join the project. Additionally, two of them identified as peace activists and the other was involved in development work, so all three saw this project as furthering their work. Our Burmese associates are cosmopolitan (e.g., studying or working abroad for some time) and had collaborated with foreigners on projects or work before, suggesting a generally positive disposition to trust in the Northern researchers. At the same time, previous negative experiences (mentioned above) weighed on our collaboration and led the two artists to ensure a certain control over the process within the possibilities of the structural power differentials from the outset. In addition to formal recognition of their work inputs, they also asked to be part of the workshop design to ensure that the generated data would not only address our project objectives but simultaneously also their objectives as peace activists.
On both sides, there were also leaps of faith among the 'teams', whose members did not all know each other (well) before the project and had to come together as teams in the first place (Kušić 2020). At the same time, the general willingness to trust each other and enter into this research collaboration was intensified by the fact that getting to know each other was already embedded in trusting social relationships: the Northern researchers knew each other through conferences or previous work and one Burmese collaborator through a personal contact. She, in turn, brought us in contact with the Burmese artists, who also turned out to be friends of a staff member of our partner organisation NPM. Beyond these direct and indirect personal networks, it is also fair to assume that personal affinitiesand perhaps even the gender of the predominantly female team 7played a role in the trust that, setbacks notwithstanding, slowly emerged among the project members.

Meso-scale: Personal Interactions, Misunderstandings and Mutual Appreciation
A third factor involved in trust-building are intersubjective processes during encounters of the different parties, which include both personal performances and the locations and 'props' involved in these encounters (Murphy 2006, pp. 438, 441). In our project, collaboration could not have worked without this process of personally interacting and getting to know each other during two initial field trips to set up the project and several further ones to jointly work on the workshop material interpretation and preparation of dissemination materials. With the Burmese businesswoman, trust emerged during our joint meetings with artists and peace activists which she organised and interpreted . She increasingly took on a more proactive role in explaining the research objectives and guiding the conversations. We spent long hours in taxis crawling through the heavy Yangon traffic, sharing about our lives and discussing the work. Due to our reliance on her, she exerted considerable and welcome power to shape the research. With the two artists these interactions likewise involved spending long hours in the living/working room of their small flat, sharing tea, food and stories, before the workshop idea even materialised. Trust was mutually reinforced through the exchange of experiences and printed outputs of previous drawing workshops held by the artists and one of the Northern researchers, respectively, which conveyed determination to follow through and seriousness about the chosen method. These personal face-to-face interactions built trust and made it easier later on to support the research of the Burmese associates remotely through Skype, WhatsApp and Facebook Messenger. In the process of repeated personal interaction, all came to appreciate each other for their capacities and knowledge and there were learning processes on all sides.
That said, the interactions were not free of frictions and misunderstandings, and the potential for mistrust always lurked in the background. For instance, right before the first workshop there was a miscommunication about the facilitation fees. This led some of the Burmese associates to believe the Northern researchers were exploiting them as cheap labour, while some Northern researchers were reminded of other work/research contexts in which they had been confronted with local actors' attempts to raise prices for internationals. This situation created tensions not only between but also among 'teams' (Kušić 2020). It is also an illustrative example of the twodirectional flow of power involved in collaboration processes: the Northern researchers had the power of the purse, the Burmese associates the power of doing (or refusing to do) the work of data generation. At one point during the first workshop, one of the artists questioned the Northern researchers' ethics, fearing we would willingly accept the risk of retraumatisation of conflictaffected individuals for our own gains. Both incidents were worked out, building a deeper sense of co-creation, but perhaps also eroding trust a bit. After the workshops, long delays in transcribing and translating the workshop data raised doubts among the Northern researchers about the commitment of some of the Burmese associates and showed us our powerlessness in a situation in which we had willingly conceded the control over important parts of the research process to the Myanmar team.
All these tensions were resolved in some way, never leading to a breakdown of working relationships. They do, however, hint at the difficulties of establishing equitable research relationships in a short time frame and under inequitable circumstances. These challenges impacted the type and quality of data that was generated.

Power, Trust and Data: Designing, Implementing and Analysing the Research
In the following we reflect on how the dynamics of power and trust/mistrust sketched above influenced our qualitative-interpretive research with regard to the two drawing workshops we held in Kachin and Rakhine states, organised around the phases of workshop planning (research design), workshop implementation (data generation), and workshop outcomes (data interpretation).

Workshop Design
The metaphor-centred drawing method DrawingOut that inspired us to explore drawing as a method for this project was originally developed to explore sensitive topics in health research. 8 These included invisible diseases among groups of participants for whom such topics were culturally taboo, difficult to objectify in language due to the nature of the problem, or for those whose language skills make interviews or focus group participation challenging. Specifically designed to avoid working with preconceived categories and concepts, the original method uses metaphor in drawingsthe representation of something abstract such as emotions or thoughts through something concreteto explore participants' experiences (Gameiro et al. 2018). Since the workshop design dovetailed with the qualitative-interpretive project approach, we planned to use this method to explore conflict and protection experiences and knowledge among violence-affected communities in Myanmar.
Aware of the power structures and challenging trust-building processes inherent in asymmetric North-South collaborations, we readily agreed to a cooperative approach requested by the Burmese artists, who would facilitate the activities, to adapt the workshops to fit not only the project but also their interests, and enable all team members to use the workshop results. Perhaps unavoidably, however, the project members differed in their conception of how the workshops should be designed. While the Northern researchers favoured an open approach that would give research participants space to set their own agendas, the Burmese artists' approach as experienced workshop facilitators was of a more pedagogical nature and had the express aim to foster the acceptance of multicultural diversity in Myanmar's society. Likewise, a first workshop outline prepared by one of them revolved around specific predetermined metaphors (e.g., prompting participants to draw trees) rather than keeping the choice of metaphor open to participants and thereby allow for diverse forms of expression (e.g., if your past was a plant or a living being, what would it be?). Suggestions by the Northern researchers to leave the workshops as open as possible, carefully phrased as requests and not demands to express our respect for and trust in them, were perceived by the Myanmar team not as welcome research autonomy enabled by the abductive approach of qualitative-interpretive research, but as a stressful lack of guidance on what the project objectives were (Kušić 2020).
The final design agreed on for the first workshop was a three-day event which required participants to stay at the venue over night and involved a per diem for their time provided to the project. The agenda was a compromise between the open approach and metaphor-centred method favoured by the Northern researchers and the more guided and factual approach our Burmese associates felt comfortable with. The design further evolved during the implementation phase as the Burmese team included questions that the original plan had not contained. The metaphor-centred element was ultimately reduced to one element among many, asking participants to develop a metaphor that represented themselves. 9 Several questions remained openended, but the facilitators also included questions which diametrically opposed the hermeneutical project methodology. For instance, participants in the Kachin workshop were asked to reflect on 'What makes you proud to be Kachin?', rather than a hermeneutical rendering of this question like 'What does it mean to be Kachin?' or 'How would you describe your identity?'. 10 These changes to the workshop design impacted the type of data. Data generated through questions that did not allow for diverse meanings to be expressed and explored, while interesting in themselves, are less useful in interpretive-hermeneutical research. The Northern researchers experienced this as resulting from loss of power to influence the research due to their lack of presence, while at the same time appreciating the initiative taken by the Burmese team to include a range of additional aspects into the workshop.

Workshop Implementation
In order to recruit participants for the two workshops, the associate mainly responsible for logistics relied on 'local brokers'. 11 In Kachin, a local protection practitioner who we had met during a previous joint field trip to Kachin was hired to recruit workshop participants. Some of the participants did not (or refused to) speak Burmese, making it necessary to also organise language translation, as only one of the Burmese associates speaks Jingphaw. Most participants were living in camps for internally displaced persons (IDPs), and most had already participated in workshops held by international organisations and NGOs.
For the Rakhine workshop, our Burmese collaborator who is from Rakhine consulted a local NGO and relied on her husband to travel to remote villages to recruit participants from different ethnic groups living in Northern Rakhine state. Without these connections, the workshop would not have occurred, given that we foreigners were not allowed to these places. Notably, while the participant group involved members of the Arakanese majority in Rakhine and of a number of ethnic minorities in the state who are often overlooked in conflict analyses, there were no Muslim workshop participants. This was an important omission given that one of the mainalthough not the onlycleavage in Rakhine state is between Arakanese and Muslim populations. 12 The lack of a Muslim research associate likely contributed to the lack of trust in the workshop among potential Muslim participants, given the already highly tense conflict situation; only weeks after our workshop, the 2017 mass violence against the Rohingya Muslims of Rakhine exploded. Most of the Rakhine workshop participants had never participated in a workshop and most were from poor rural backgrounds. While a number of them had experiences with violent conflict, the reasons for recruiting some others were less clearly project-related. Again, we relied on our Burmese associates and felt powerless to direct the recruitment process but thankful the project moved ahead. In fact, some of the recruitment process we did not understand until after the workshop.
In the original DrawingOut design, drawing is chosen as arts-based research method because it does not require costly resources and because it is assumed that drawing is a universal human skill that can be used with any group of participants. The artistic quality of a drawing does not matter since it is only a vehicle to express feelings or thoughts and not an end in itself. Drawings are accompanied by participants' verbal explanations and discussions among groups of participants in order to access the intended meaning of the drawer (Gameiro et al. 2018). During the first of the two workshops, it became clear that one of the artists who facilitated the drawing exercises put much more emphasis on teaching drawing skills based on his knowledge about a lack of arts teaching in Myanmar's schools (Kušić 2020). This led to the paradoxical situation that this emphasis on drawing not as method but as skill clashed not only with the Northern researchers' methods-related assumptions, but curiously also with the expectations of some Kachin workshop participants. They requested to draw less and talk more, eager to tell their stories of life amidst violent conflict, the main objective of the project for which drawing was only meant to be a prompt. While the Kachin participants with the help of one Burmese associate organised an evening story-telling session to share their experiences which led to rich empirical data, this episode nonetheless raises the question of how much other potential data could have been generated but was prevented by a rigid workshop implementation that focused on priorities which were not the most project-relevant ones.
In many respects, the Kachin workshop was experienced by the Burmese collaborators as a learning experience, in which they came together as a team, built trust among themselves, experimented with the workshop design, and not least grappled with the question of whether what they were doing was useful for the project and what we, the project leaders, had in mind (Kušić 2020). Online team meetings and exchanges via social media helped to deal with 'difficult' participants or reflect on specific workshop elements, but the main decisions during the workshop implementation/data generation phase lay in the hands of the Myanmar team. Their feedback 13 during and after the second workshop was considerably more positive regarding their internal working relationships, their interaction with participants and the smooth running of the (slightly revised) workshop design. This seemed to be mirrored by the fact that the Rakhine participants asked to be included in future workshops, should such opportunity arise. In addition, one of the artists said she was inspired to learn more about healing and later started exploring research as a new field of engaging with her society. The problems the Kachin workshop faced and the relative harmony that characterised the Rakhine workshop, however, said little about the type and quality of research data.

Workshop Outcomes
Data generated by the two workshops includes all drawings the participants created (some of which have some writing on them) as well as recordings of the sharing sessions in which participants explained their drawings to each other and shared additional stories about their lives. 14 Since the workshops took place in Burmese and additional minority languages, our Burmese associates were paid to organise the transcription and translation of the recordings, a process which took months on end for reasons that remained opaque. This constituted a main source of frustration for the Northern researchers who did not want to undermine the Myanmar team by outsourcing the recordings to other translators but could not proceed to the data interpretation due to a lack of translated materials until after the official project end.
The Kachin data turned out to be very rich, with participants eager to share their life stories and reflect on the different challenges and experiences of life amidst violent conflict. A good range of questions and prompts given by the Myanmar facilitators were open enough to allow to infer meaning from participants' answers unstructured by preconceived concepts. 15 The workshop showed that qualitative-interpretive research is possible even if conducted 'by proxy' and can render results that further our understanding of specific conflicts. This also indicated that the drawing method could be successfully used by protection INGOs like NPM to nuance their conflict assessment and make their programmes relevant to people's lives, and indeed, based on our research NPM and its sister organisation Nonviolent Peaceforce Philippines (NPP) have incorporated drawing into some of their later work.
The Rakhine drawings, which for months were the only data from the second workshop available to us, seemed equally rich and together with the Myanmar team's positive feedback about the workshop implementation raised our expectations regarding the recorded data. When the translations finally arrived, however, these expectations were not met: there seemed to be very little data for a three-day workshop in the first place, and it is not clear to date whether there is still untranslated material, whether some parts of the workshop were accidentally not recorded, or whether some recordings were purposefully not shared. From the reconstruction of the workshop topics and drawing prompts in discussion with our Burmese associates, it is clear that the Rakhine workshop partly lost sight of the project's core interest in local experiences and knowledge of violent conflict and protection, although some images and accounts do address this topic. 16 Most remarkably perhaps, the main cleavage between Buddhists and Muslims in Rakhine state, which is a socio-economic one and embedded in wider grievances and inequalities between different ethnic groups in Myanmar, 17 is hardly thematised in the data generated in Rakhine. It is here that our specific interpretive research 'by proxy' shows a clear limitation. It is not entirely clear whether the workshop prompts encouraged participants to share potential stories around this conflict. If there were such prompts which, however, did not result in respective data, this would be an interesting finding in itself and worth further exploration. Were the team members and participants so wrapped up in the discrimination against their own ethnic identities that they simply ignored this conflict? Or was the situation already so tense that people did not dare raise the subject? Is the climate of fear of the Myanmar authorities much more marked in Rakhine than it is among Kachin people? If so, what are the reason for thisis there factually more repression in Rakhine than in Kachin, can cultural differences account for this, or is the communal nature of violence in Rakhine 18 a deterrent to people speaking out? Or could the difference be that while in Kachin we worked with members of the oppressed minorities, the Rakhine participants included members of the majority group commonly blamed for the tensions?
Alternatively, if the Myanmar team did not prompt any drawing exercises that could have led to reflections on community tensions between Muslims and Buddhists in Rakhine, then why was that? Did the fact that one of the team members self-identifies as a member of one of the conflicting groupsand shares this group's frustration and historical grievances of being ignored by the state of Myanmar and the international communityplay a role in the choice to omit the topic? Does this confirm the suspicion against local researchers described by Macaspac (2018) as being in one way or another 'internal' to the conflict, that as citizens of their country no outside position to the violence is possible (short of ignoring it completely)?
Given the limitations of our research, we were not present to observe, ask questions, explore further, understand, nor did we have the time and resources to follow up later. The questions just posed reflect a gap in trust and perceived power stemming from our absence. While the puzzling omission of one main cleavage of violent conflict in Rakhine could be a highly productive finding from a hermeneutical point of view (Fujii 2010), our inability to follow up on it under the given research circumstances also illustrates the limits of working through local associates in doing interpretive research. Conversations with our Burmese collaborators after the eventual receipt of the transcripts illuminated some aspects but remained ambiguous around others. In the time that had lapsed between the workshops and the receipt of the translated transcripts, violence in Rakhine had escalated, putting one associate in a very stressful personal situation and causing a certain rift among the Myanmar team for a while, and it did not seem ethical or humanly decent to push for more answers where they were unclear or refused. After all, our research journey had involved a lot of personal sharing and we had come to care for one another well beyond work.

Conclusions
Myanmar is only one example of a rising number of contexts in which, for various reasons, peace and conflict researchers are becoming increasingly dependent upon national research associates. This raises the question of what our mixed experiences of an asymmetric North-South collaboration in a qualitative-interpretive project mean more generally for the prospects of doing qualitative-interpretive fieldwork by proxy. Can such collaborations substitute for foreigners' lack of field access, or do interpretive researchers have to accept that their approach hits a wall where direct field immersion is limited?
Our arts-based workshops yielded quite different results in terms of the type and quality of data they generated. In the case of the Kachin workshop, there is lots of rich material and most of it lends itself to hermeneutical interpretation. It is data that gives nuance to understandings of violent conflict in Kachin, but it is also to some extent predictable data holding few surprises. The data generated in the Rakhine workshop, by contrast, is highly surprising. Not only is the material rather thin in comparison; it also curiously lacks reference to the communal clashes between Arakanese Buddhists and Muslim Rohingya and to the violent interventions by state security forces targeting both groups. From an interpretivist's point of view this is an exciting finding, as it may reveal some important insights into inter-communal relations in Rakhine. That we foreign researchers got these insights at all was thanks to our collaboration with Burmese associates. Not only were they able to go to areas forbidden to us; they were also brilliant in building trust and working with research participants, as mirrored in the positive feedback they received from the latter. Furthermore, their language and cultural interpretations made the verbal and visual research data accessible to us in ways that would not have been possible without these cultural translations and insights.
The positive sides of our research collaboration were made possible by a constant albeit imperfect work to mitigate some dynamics of mistrust caused by the impact of power structures and flows on our collaboration. As our analysis suggests, moments of mistrust arose mostly when communication failed or uncertainty negatively influenced perceptions. Yet at no point did this lead to a breakdown in the generally committed working relationships on either side. What is more, many of these moments of mistrust could have been attenuated with more timetime to train the Burmese collaborators not in specific methods or concepts, but in the interpretivist sensitivity underpinning our hermeneutical research interests, and also time to jointly analyse the data and follow up with participants where insights surprised. It was these underlying conditions of the research, rather than any inherent impossibility to do qualitative-interpretive research 'by proxy', that limited interpretation and left a mixed feeling of excitement about the insights obtained and doubt about some of their meaning.
In this sense, interpretive research by proxy is neither just 'good enough' research in the sense of research that is less than ideal but better than none at all. Nor is it 'good enough' in the sense of assuming that by default Northerners conduct 'better' interpretivist fieldwork, if only they have field access. While the data and its interpretation will have been somewhat different had the Northern researchers conducted the workshopsdue to the changed dynamics their presence would have wittingly or unwittingly causedfrom a qualitative-interpretivist epistemology these research outcomes would not have been 'truer' or 'more factual', as such values do not make sense in the social-constructionist worldview of interpretivist approaches (Kurowska and Bliesemann de Guevara 2020). Rather, what our reflective analysis suggests is that those impacts of power structures and flows, which crystallised in dynamics of trust and mistrust and generated some uncertainty and disappointment regarding our research outcomes, could have been better addressed.
There are two core lessons emerging from this article. Firstly, communications and contractual interactions, especially those regarding payments and expectations, should be as clear as possible from the outset to avoid uncertainties and misunderstandings as two main sources of mistrust (cf. Bøås 2020). Secondly, and more demanding, colleagues planning interpretivist research 'by proxy' should plan in considerable field trip time not only to find the 'right' associates (a process which took us a month), but more importantly to spend a long (and ideally paid) time with them discussing and practicing the research methodology, specific methods and strategies for reflexivity. Time is also needed to follow up with associates, ideally face-to-face, and for associates to keep working with participants after first research findings have emerged from analysis, to be able to dig deeper into those surprising moments that the material might contain.
Were such conducive conditions for a more intensive co-working are in place (dependent not least on enough funding 19 ), interpretivist research by proxy will yield rich research findings broadening our hermeneutical knowledge.