On Brokers, Biases and Leaving the Veranda: Working with Research Brokers in Political Science Based Field Research

ABSTRACT

In this paper I critically discuss the pros and cons of working with research brokers from a political science perspective. I do this by sharing my experiences of collaborating with two types of local “fixers” – cultural and communal brokers. I argue that even if similar approaches risk introducing a number of broker-induced biases, countermeasures can be taken to mitigate their effect: e.g. continuous triangulation, interviewing “sleepers” and asking in-depth questions to exclude interviewees who misrepresent themselves. However, researchers must be aware that working with multiple brokers can also generate methodological and ethical challenges that need to be taken into consideration.

Introduction

There has recently been an upsurge in interest amongst conflict scholars to critically discuss the usage of research brokers – the local ‘fixers’ who help scholars generate knowledge about the researched ‘other’ (Eriksson Baaz and Utas 2019: 157, Jennings 2020, Myrttinen and Mastonshoeva 2019: 240, Schlitz and Büscher 2018: 125). On the one hand, such broker figures are indispensable in order to gain an understanding of the ‘local rules of the game’¹ and build trust between apprehensive informants and researchers. Commonly living in, or originating from, the research environment in question, brokers possess the local knowledge and networks that researchers lack (Eriksson Baaz and Utas 2019: 161). These types of services have become increasingly valuable as donor agencies are restricting travels to war fraught countries and many host governments confine researcher’s access to the field (Eriksson Baaz and Utas 2019: 158). On the other hand, the very qualities that make brokers so valued – access to different knowledge worlds, the entrepreneurial skills needed to circumvent unexpected obstacles, and political connections – also entail that they are well-placed to bias the research process by controlling the narrative scholars get to hear (Parashar 2019: 253, Schlitz and Büscher 2018: 124). As such, brokers are not only facilitating the research process, but also ‘creating new realities’ (Schlitz and Büscher 2018: 124).

The call for greater transparency when it comes to research brokers constitutes a healthy development. However, this debate has primarily been driven by anthropologists who are arguably also the category of researchers who are the least dependent on brokers in the first place (Eriksson Baaz and Utas 2019: 159; Schiltz & Büscher 2018: 130). Cast in the ethnographic tradition, with its emphasis on ‘deep hanging out’, ‘immersion’ and lengthy field work, anthropologist are not only well-placed to accumulate the rich local knowledge needed to mitigate broker-generated biases, but also to work independently from brokers once initial introductions have been made (Molony and Hammett 2007: 298). So far, the call for a more critical stance has found limited resonance amongst political science scholars (Eriksson Baaz and Utas 2019: 159).² This is problematic, since broker-induced biases are likely to be more pronounced when conducting research employing political science based research designs. This is because political scientists more commonly: (a) rely on theoretical deduction; (b) place more emphasis on comparing multiple cases; and (c) leave behind a larger monetary footprint. The purpose of this paper is to draw on my field experiences from Liberia to highlight some of the pitfalls associated with working with brokers from a political science perspective and discuss the strategies that can be used to minimise broker-induced biases. I thereby adhere to Eriksson Baaz and Utas (2019: 159) call for greater ‘[…] consultation and debates between various disciplines engaged in conflict research’ when it comes to sharing experiences of working with brokers.

More specifically, my reflections are based on work that I have done with ex-command structures³ – defined as a group of individuals made up of an ex-commander and those ex-subordinates that the former superior meets on a regular basis (Themnér and Karlén 2020: 2) – in Liberia.⁴ The objective of this project was to explain why some ex-commanders were more successful in retaining control over their ex-fighters, than others. In this work, I collaborated with two kinds of local fixers: cultural brokers (urban-based agents who possess the ability to bridge, link or mediate between researchers and the researched belonging to different cultural backgrounds) and communal brokers (gatekeepers into target groups who share a strong in-group identity and distrust outsiders). Being well versed in international peacebuilding discourses, the cultural brokers – in the form of an ex-combatant NGO and a university research assistant – provided valuable cultural translations and made important introductions to central figures in local communities (especially ex-commanders). However, since they lacked close relations with a key segment of the target group, the ex-combatants, I was also obliged employ the ex-commanders to convince the ex-fighters to take part in interviews. Due to the ex-commanders’ role in sustaining hierarchical structures within the target group, it was vital to employ their services as communal brokers. Not only did this allow me to mobilise ex-combatant interviewees, but also to ensure that they shared their own insights as key informants.

In order to minimise the risk that my broker-based study generated harmful biases, a number of countermeasures were taken. These included actions such as (a) working with multiple cultural brokers in order to prevent selection bias and hearing one sided narratives; (b) continuous triangulation (conducting multiple interviews with the communal brokers to confront the latter with new information received from their informants), (c) interviewing ‘sleepers’ (i.e., interviewees not identified by the communal brokers as key informants), and (d) asking in-depth contextualised questions to identify and remove ‘false’ informants presented by the communal brokers. These measures allowed me to minimise the risk of social desirability bias (whereby the cultural and communal brokers sought to please me) and of the latter exaggerating the size of their networks or controlling the narrative presented to me. An unanticipated challenge was that it proved more difficult than expected to collaborate with brokers operating at different stratums of society. My layered approach of working with cultural and communal brokers sometimes generated tensions, with the latter accusing the cultural brokers of taking advantage of them. The hierarchical relations that often exist between brokers is largely overlooked in the literature and points to the need for scholars to analytically distinguish between different categories of brokers. This is crucial in order to develop strategies that can alleviate the risk of inter-broker conflicts negatively affecting the data generating process and hinder international scholars from cementing local forms of exploitative knowledge production.

This paper proceeds by further developing why broker-induced biases may be particular harmful when conducting political science based fieldwork. I thereafter provide a brief overview of my work with ex-military networks in Liberia and my experiences of collaborating with both cultural and communal brokers. This is followed by a discussion of the anticipated biases that my design risked generating and the strategies developed to mitigate them. Next, I outline some unexpected challenges that arouse during the implementation of the project. This foremost concerns problems generated by my layered approach of working with both cultural and communal brokers. The paper will end with some concluding reflections.

Political Science Based Field Research and Broker Induced Biases

Research brokers can be seen as agents ‘being in-between the researcher and the researched who regulates the access and flow of knowledge between them’ (Eriksson Baaz and Utas 2019: 160). The type of knowledge that brokers regulate can take numerous forms. The most common is to make introductions to informants of the target group that the researcher is studying (Eriksson Baaz and Utas 2019: 161, Parashar 2019: 250). This is possible thanks to the localised networks that many brokers possess. Without such facilitations, most interview-based projects are difficult to implement. Another form of knowledge that brokers disperse is cultural translations (Eriksson Baaz and Utas 2019: 158, Schlitz and Büscher 2018: 125). Operating ‘at the “interfaces” of different world views and knowledge systems’, most brokers do much more than merely introduce people (Schlitz and Büscher 2018: 125). Oftentimes researchers and the researched lack the linguistic, but perhaps more importantly, cultural capacity to comprehend each other. To prevent misunderstandings brokers – often accustomed to the discourse of ‘Western’ researchers – play a crucial role in repackaging and contextualizing what is said during researcher-interviewee interactions. Finally, it is not uncommon that research brokers are key informants in their own right and must be considered as being part of the field (Jenkins 2018: 147, Knott 2019: 144, Middleton and Cons 2014: 280, Molony and Hammett 2007: 292, Parashar 2019: 260). This is often the case when working with distinct social groups who distrust outsiders. Under such circumstances, it may only be central figures within the group who can convince other members to take part in interviews (Jenkins 2018: 147–150). However, since these broker figures may not only possess unique knowledge about the group, but also play a key role in sustaining in-group structures, they must also be treated as informants. Experiences have shown that there is a plethora of local actors who have the capacity to transmit the three above-mentioned forms of knowledge to foreign scholars. These range from NGO representatives, local researchers and university students, to community and youth leaders.

There has recently been an upsurge in interest amongst conflict scholars to critically reflect on how researchers work with and relate to research brokers (see e.g., Schlitz and Büscher 2018, Eriksson Baaz and Utas 2019, Myrttinen and Mastonshoeva 2019, Parashar 2019, Jennings 2020). In this debate, focus has both been on the methodological and ethical pitfalls that research-broker collaborations can generate. While the former primarily concern broker induced biases, the latter touches upon questions of exploitative forms of knowledge production and exposing brokers to risks that researchers are unwilling to take. Interestingly, this self-reflective debate has been spearheaded by anthropologists and can be seen as being part of a larger intra-disciplinary effort to debunk the myth of the heroic and lone anthropologist who embeds her/himself in and learns about local communities without the help of local actors (Barley 1983, Turner 2010, Schlitz and Büscher 2018, Eriksson Baaz 2019). Unfortunately, this type of self-critical assessment is largely missing amongst political scientists studying war- and peacetime dynamics. This is worrying for two reasons. First, most political scientists do employ some kind of brokers when conducting field research in war-afflicted countries. Second, broker-induced biases can be particularly toxic when using political science based research designs. By this, I do not mean that similar hazards cannot undo findings generated by anthropologists, it is just that broker-induced biases are likely to be more pronounced in the realm of political science. It is therefore high time for a more critical discussion concerning the pros and cons of working with brokers from a political science perspective. There are, more specifically, three reasons for why broker-induced biases may be particularly severe for political scientists.

First, even if political scientists sometimes conduct inductive studies, it is more common that they steer the data-gathering process via theoretically deduced interview questions. As such, the research design is informed by existing theory or applications of theoretical findings from adjoining fields of interests (King, Gary, Robert Keohane and Sidney Verba 1994). Although deductive-based studies have clear benefits – for instance, facilitating the accumulation of knowledge by more clearly building on previous research – it does make field research more vulnerable to social desirability bias. Studies have highlighted how brokers sometimes conceal uncomfortable information from researchers in order to keep them happy (Schlitz and Büscher 2018: 137). There may be a particularly high risk that brokers engage in similar censoring if it is obvious that the researcher has a ‘pet’ theory. Even if researchers try not to disclose their key hypotheses, they may inadvertently be revealed if, for instance, a non-proportional number of interview questions revolve around a certain topic. Such types of biases may be less problematic in anthropology-based studies. The latter has a stronger tradition of utilising more open-ended inductive approaches.⁵ Cast in the ethnographic tradition of ‘deep hanging out’ and ‘immersion’, more emphasis is put on active listening and informal conversations, than formal interviews and surveys (Davies 2002, Molony and Hammett 2007: 298, Utas 2019). Induction can also make researchers more attuned to understanding local dynamics that can hold explanatory value. Hence, by using more open-ended designs, anthropologists may not only exert less (indirect) pressure on brokers to confirm stipulated hypotheses, but also create a more enabling environment where researchers and brokers can have a dialogue on which direction the project should take.

Second, political scientists commonly sacrifice additional empirical detail in order to gain analytical leverage by comparing multiple cases – e.g., rebel groups, social movements or local communities (King, Gary, Robert Keohane and Sidney Verba 1994). While such approaches are helpful to generalise findings, it also makes researchers extremely dependent on brokers (Jenkins 2018: 149–150). Due to the need to move between field sites, few political scientists have the luxury to fully immerge themselves into local communities and without brokers they may find themselves stuck on the ‘veranda’. Even if many anthropologists are also dependent on brokers, the former usually have the benefit of spending considerable time in the same locality. Not only can this allow them to eventually build up relations with informants independently of their brokers, but also develop a deeper understanding of the target community (Molony and Hammett 2007: 298, Parashar 2019: 264). As such, anthropologists are often in a better position to scrutinise the information provided by brokers and the relevance of the informants presented by the latter.

A final reason for why political scientists need to be particularly wary is that their studies often leave a larger monetary footprint. Even if there may be consensus amongst anthropologists and political scientists that research brokers should be paid for their labour, there are important differences when it comes to providing compensation to informants. By spending extended periods of time in the social setting of the target group, often via so-called participatory observations, anthropologists can gain empirical insights via informal conversations (Davies 2002, Molony and Hammett 2007: 298). As such, knowledge can be generated without major interruptions in the everyday lives of informants and there is often less of a need to compensate the latter for incurred costs. It is more difficult for political scientists to engage with informants in a similar manner. Due to the need to visit several research sites, few political scientists have the time and resources to embed themselves in multiple local communities. Instead, it is more common for political scientists to work via formal interviews, commonly organised by their brokers (Kvale 1996). In order to minimise the risk of outing informants, such interviews are often done outside the latter’s area of residence. Taking part in interviews may therefore not only entail travelling costs, but also loss of income. For economically marginalised groups, the cost of participating in interviews can therefore be substantial. There is a growing realisation amongst conflict scholars that there are strong ethical reasons to compensate informants, especially those with limited economic means, for incurred costs (see e.g., Jennings 2020).⁶ However, besides the obvious problem of social desirability bias – whereby compensated informants provide the information they assume scholars want to hear (Jennings 2020: 223) – the distribution of economic resources can create incentives for brokers to manipulate whom the researcher gets to interview. Interested in boosting their local Big Man status, brokers may be tempted to mobilise their own or potential new clients as informants, rather than the target group identified by the researcher (Utas 2019: 273–274). This constitutes a particular danger if the compensation offered to informants is disproportionally high.

Working with Cultural and Communal Brokers

My work with ex-commanders and fighters was part of a larger project – involving a research colleague and a PhD student – studying the prevalence of ex-command structures in Liberia. While my two colleagues were tasked with studying these ex-military networks through ethnographic research, I employed a more traditional political science approach. This entailed spending less time in the field, working more closely with brokers and focusing on systematically comparing different ex-command structures.

To answer the research question guiding my part of the project, I set out to compare ex-military networks situated in different milieus, such as mining regions, urban areas (ghettos and middle-income neighbourhoods), as well as semi-urban communities. This would allow me to assess assumptions found in the literature that ex-fighters flock around ex-commanders operating in spaces of marginality (e.g., ghettos penetrated by the drug economy and mining areas). More concretely, this entailed investigating aspects such as: (a) the number of ex-combatants in an ex-commander’s network, (b) how frequently ex-commanders’ and combatants met, and (c) the level of asymmetry in ex-commander-fighter relations. Even though the study focused on ex-command structures, there was no prior assumption that all or even most ex-fighters still interacted with their ex-commanders. It merely entailed that the study spoke to those ex-combatants who, for one reason or another, still had some relations with their ex-commanders.

Although I had previous experiences of working with ex-combatants in other war-ravaged African countries, I had never visited Liberia before. As such, I lacked the connections needed to access my target group. With the help of colleagues, I was able to identify two local actors who had experience of interacting with ex-fighters in Liberia: an ex-combatant NGO and a research assistant from a local university. My first meetings with these brokers were characterised by a good amount of posturing in order to define our relationship, expectations and pay. Initially, the brokers repeated commonly used policy discourses about the threat emanating from ex-combatants and that more international assistance was needed to prevent ex-combatant violence. The reflexive manner in which these ideas were presented indicated that the brokers assumed that this was what I, as a ‘Westerner’, expected to hear. When the NGO employed such simplified descriptions, it also felt like they were pitching a project idea to me in the hope that I would convey it to international funders (somewhat ironically making me their broker). According to Bliesemann de Guevara and Kostić (2017: 14), the production of such simplified policy solutions – which de facto mask the palette of interventions available to policymakers – is becoming ever more common, increasing the need for researchers to maintain a critical approach when working in conflict settings. My reaction to these depictions was to emphasise, and at times exaggerate, my experience of working in other post-war contexts to signal that I was not gullible. I do believe that this kind of posturing was crucial in order to move our discussion away from policy clichés, to more fruitful discussions about both the positive and negative role ex-fighters can play in communities and how local mechanisms (e.g., patronage networks, saving clubs) can help reintegrate ex-combatants. However, my positionality as an ‘expert’ on post-war dynamics also came with a price. My need to signal competence and worldliness not only risked making me blind to the uniqueness of the Liberian case, but also make the brokers unwilling to prove me wrong.

It soon became clear that the NGO and research assistant could disperse two types of knowledge to me – introductions to key informants (especially the ex-commanders) and cultural translations. As a newcomer, I initially lacked the networks and experiences to efficiently navigate the complex web of international, national and local actors with an interest in issues relating to ex-combatants. Having previously worked with both international researchers and ex-fighters, the brokers provided me with an introduction of key actors and the main trends in the country. They were also instrumental in implementing an initial survey⁷ that gathered testimonies from 58 ex-commanders and 37 of their ex-combatants. The survey provided a general picture of ex-commanders’ and fighters’ situation, and an initial understanding of why they continued to interact. Partly based on connections generated from the survey, the brokers helped me to identify five suitable ex-commanders (a miner, a security provider, a drug distributer, a military recruiter and farmer/motorcycle-taxi rider) who headed networks of relevance for the study. It would have been difficult for me to identify these five ex-military networks on my own. Due to the suspicious disposition of ex-military actors – who are often fearful of being outed as ex-combatants or prosecuted for war crimes – most ex-commanders would probably have refused to talk to me if I had approached them. In addition, as relatively new to the field, I may have ended up working with individuals pretending to be ex-commanders, reciting commonly available ‘meta’ narratives of war and peacetime dynamics.

Even if the NGO and assistant played a crucial role in identifying suitable ex-military networks to engage with, they were unable to help me to carry the project into the next phase – convincing its members to take part in interviews. In fact, the brokers lacked the ability to generate sufficient trust in the project. This was due to three reasons. First, the professional interests of the research brokers meant that they had rather weak ties to the target group, in particular the ex-fighters. In order to identify job assignments and funding opportunities the brokers had devoted much of their time building ties to Monrovia-based elites and international donors, rather than engaging themselves with local communities. This was particularly evident when it came to the NGO. After the end of the war, the NGO had successfully raised funding for various reintegration programmes. This had allowed them to actively engage with different ex-combatant groupings. However, at the time of my arrival, the ex-combatant issue had disappeared from the agenda of donor agencies and the NGO was struggling economically. In the midst of rebranding themselves as a more general developmental NGO, the latter had largely ceased with its ex-combatant activities. As I would later discover, some of my ex-combatant interviewees were disappointed with the NGO, expressing bitterness that they had forgotten them. This is part of a general problem for many NGOs working in a developmental context; due to their dependence on international funding, NGOs become more accountable towards foreign actors, than the communities they claim to represent (Puljek-Shank and Verkoren 2017). Consequently, many NGOs only possess thin legitimacy in the eyes of community members, at times hampering their ability to mobilise informants for researchers.

A second impediment for these research brokers was the suspicious disposition of the ex-fighters. This is a common problem when conducting research on social groups with strong in-group identities and who tend to distrust outsiders (Norman 2009). In such situations, it may not be enough that brokers come from the same country or ethnic group as the researched. In my experiences, ex-fighters are just as often suspicious vis-à-vis unknown ‘locals’ as foreign nationals. Rumours of state agents and mercenary recruiters operating in ex-combatant communities often abound amongst ex-fighters, making the latter edgy whenever unfamiliar local actors start asking questions. As such, being ‘local’ and representing ostensible neutral institutions – such as an NGO and a university – may not be enough to ease the fears of target groups.

Finally, the reluctance of the ex-commanders to extend support for the project – for instance by making introductions to their ex-fighters – also curtailed the work of my research brokers. The ex-commanders were generally adamant that they should be awarded a more direct role in the project. This was perhaps not surprising when considering that studies have highlighted how ex-commanders often jealously guard access to their ex-fighters, functioning as gatekeepers for outside actors seeking to access the latter (Themnér 2012).

Even if the introductions to the ex-commanders made by the NGO and assistant were important, it was their ability to provide cultural translations that proved to be most valuable. Having previous experience of working with foreign researchers on matters relating to peacebuilding and ex-combatant reintegration, they were accustomed to my academic jargon sprinkled with theoretical and methodological terms. As such, they could grasp what I wanted to do. What is more, they helped me to reformulate what was sometimes culturally irrelevant interview questions derived from existing theory. Finally, during the ex-combatant interviews the brokers had a facilitating role; not only were they often obliged to repeat what I said in more correct Liberian English, on multiple occasions they had to intervene to clear up confusions between the interviewees and me.

Based on these experiences, how should we conceptualise and understand these research brokers? I argue that they can best be described as a type of cultural brokers – urban-based agents who possess the ability to bridge, link or mediate between researchers and the researched belonging to different cultural backgrounds.⁸ More specifically, they do this by (a) facilitating introductions between foreign researchers and leading figures within the target group, and (b) providing researchers and the researched with a common cultural understanding of each other. However, as these cultural brokers commonly operate at the meso-level, between elites and local communities, they seldom possess the ability to engage with the target group without the active collaboration of more locally embedded brokers. In addition, since they are not part of the target group, cultural brokers do not have to be treated as key informants. Finally, in order to move the researcher-broker dialogue beyond the mere recycling of dominant policy discourses, it may be necessary for researchers to engage in a good amount of posturing.

The inability of the cultural brokers to make inroads into and mobilise members of the ex-military networks entailed that it became necessary to more actively collaborate with the ex-commanders. In fact, I ended up employing them as brokers in their own right, facilitating the implementation of the project. This was an efficient strategy. Since the ex-commanders knew all the members of their networks and was often, but not always, respected by their ex-combatants it was possible to build a minimum of trust with the latter. The collaboration with the ex-commanders was based on the idea of conducting a number of egocentric or personal network studies.⁹ This is an established approach in social network analysis, and is standard practice in sociology, and increasingly so in political science, economics and history (Borgatti et al. 2018). In short, this entailed working closely with the ex-commanders and asking them to describe the network of ex-combatants surrounding them. This was primarily done through the creation of a roster, whereby the ex-commanders listed the ex-fighters whom they had regular contact with. Using this strategy, we established the size of each ex-commander’s network – the drug distributer (12 ex-fighters), farmer/motorcycle-taxi rider (23), security provider (28), miner (29) and military recruiter (33). In a follow up study of the security provider and military recruiter, their networks were composed of 32 respective 30 ex-combatants. Hence, a key form of knowledge that the ex-commanders provided was introductions to the ex-fighters in their networks. As the focal point of the study, the ex-commanders also had to be interviewed in order to gain an understanding of their role in sustaining the network. As such, the latter also dispersed valuable knowledge as key informants. However, one type of knowledge that the ex-commanders were less apt at providing was cultural translations. They had little experience of working with foreign researchers and policy makers. It was therefore difficult for the ex-commanders to repackage and clarify the narratives that came out of the ex-combatant interviews in a format that was accessible for me as a foreigner.

I soon realised that the interpersonal dynamics were very different when working with the ex-commanders, as compared to the cultural brokers. My position as a privileged ‘Westerner’ – assumed to possess both wealth and valuable international connections – was much more pronounced when engaging with the former. It was more common for the ex-commanders to voice expectations that I should assist them in the future – helping them to attain a Schengen Visa, or set up a business – even after our collaboration had officially ended. Being more accustomed to international actors, the cultural brokers were presumably aware that many ‘Westerners’ saw such requests as inappropriate. From a patrimonial perspective, the ex-commanders’ appeals were legitimate; our exchange of labour and economic resources signalled the beginning of a patrimonial relationship. As such, the key question was whether I, as a Big Man, had settled the accrued debt or whether future recompense was called for (Utas and Christensen 2016, Cronin-Furman and Lake 2018). This observation raises both ethical and methodological issues. Concerning the former, to which extent did the ex-commanders accept their broker assignment based on the assumption that further assistance would be given, and to which extent did I, as a privileged foreigner, use this to my advantage? Concerning the latter, did the prospect of future rewards affect the ex-commanders’ willingness to provide truthful accounts? In hindsight, it is difficult to answer these questions. What can be said is that these experiences underline the importance of taking patrimonial dynamics into consideration when collaborating with brokers.

Another way in which I related differently vis-à-vis the ex-commander brokers – as compared to their cultural counterparts – was how I employed my body during our meetings. It was not uncommon that the ex-commanders tested me (or at least what I interpreted as a test) – seeking confrontation, raising their voice, staring me down, or claiming the physical space at my expense. It never turned physical, but at times made me uneasy. However, I soon learned that I could, when needed, use my physique (being a tall and sturdy man) to my advantage in order to signal determination and willingness to defend my space. Kovats-Bernat (2002) has described how similar forms of posturing may be essential when navigating conflict-ridden contexts; both in order to create a sense of security for the researcher, but also to gain access to key informants. An interesting question is, however, whether the ex-commanders (all male) felt a need to test me precisely because of my gender and size. Put differently, to which extent would the dynamics have been different if the ex-commanders engaged with a short, female researcher? Using my body to express confidence was, however, not unproblematic. It was sometimes difficult to read the room and assess when determination and strength was called for, and when similar tactics risked escalating the situation and creating an unconducive environment for discussions. However, similar tactics would have been uncalled for and unproductive in my relations with the cultural brokers (and also many of the ex-fighters).

I argue that the ex-commanders can best be described as a type of communal brokers – gatekeepers into target groups who share a strong in-group identity and distrust outsiders. Due to their role in sustaining hierarchical structures within the target group, such broker figures have the capacity to mobilise support for any given research project. They can therefore be likened with Hannerz’s (1980: 190) urbanite brokers – actors who more or less monopolise interactions between two sets of people and without whose services contact would be ‘negligible or non-existent’. Even if such skillsets make them valuable research allies, it also entails that they possess the power to steer the study in an undesired direction. In addition, by presenting themselves as indispensable for project implementation, they de facto claim to speak on behalf of the target group. Since they cannot be meaningfully separated from the latter, they need to be treated as key informants in their own right. Arguably, ex-commanders are not the only type of communal brokers that researchers may come across. In war-ridden societies there is an abundance of other actors – such as chiefs, vigilante leaders, heads of criminal groupings and youth leaders in urban ghettos – who have the power to control researchers’ access to various target groups (isolated ethnic minorities, as well as vigilante, criminal and youth groups). What these actors have in common is that they profoundly shape in-group dynamics in the groups they claim to represent. As such, researchers may have little option but to use their brokering services, while at the same time investing ample of time to document and critically assess their personal narratives. What is more, since some communal brokers – not only ex-commanders, but arguably also heads of vigilante groups and criminal leaders – often operate in milieus characterised by strong masculine norms and everyday violence, they are often accustomed to challenging and testing unknown ‘outsiders’. Consciously, or unconsciously, this may oblige researchers to shapeshift and change their persona.

It is vital to analytically separate communal brokers from their cultural counterparts. Cultural brokers may claim to have access to the target group. However, since they do not share the same key social traits as members of the latter, they essentially become ‘outsiders’. Despite the agency generated from being an ‘insider’, most communal brokers lack the cultural capital and connections to directly access international actors, such as researchers. This makes them dependent on cultural brokers, who are better placed to identify research broker opportunities. What is more, due to their lack of experience when it comes to interacting with international actors, communal brokers generally lack the capacity to bridge the realm of the research world, with its specific language, and the real life world of the target group. It is also vital for researchers to keep in mind that they may have to (un)consciously adjust their positionality depending on whether they are engaging with cultural or communal brokers. Openly discussing researchers’ shapeshifting is also an exercise in humility and can hopefully make scholars more understanding of why it is often difficult to nail down the persona of their informants (Robben 1995).

When it came to the actual ex-fighter interviews, it was decided not to hold interviews in the residential areas of the latter. This choice was taken to minimise the risk that the study outed and stigmatised the ex-fighters. Based on recommendations from the ex-commanders, the NGO and the university research assistant, secluded venues were selected where we could meet our informants, e.g., the NGO office, rented rooms at entertainment centres or local hotels.

The downside to conducting interviews at these locations was that there were incurred costs for the ex-fighters. Not only were many obliged to pay to transport themselves to our meetings, participation also entailed a loss of income. Depending on where the ex-fighters lived, they were usually away from their hustles anything from a couple of hours to the good part of a day. This constituted an important economic loss for individuals already living on the margins and largely sustaining themselves through activities in the informal economy, such as selling dry goods, loading cars at markets, driving motorcycle-taxis or doing sex work. As such, there were strong ethical reasons to provide compensation to the participants. Based on recommendations from the NGO, all ex-fighters were offered five USD,¹⁰ as well as a can of soda and crackers. As research brokers, the ex-commanders also had a number of expenses. In order to sensitise the ex-fighters, and at times bring the latter to the interviews, it was necessary for them to travel to or call the ex-fighters. During such sensitisation meetings, ex-commanders were often obliged to offer food, drinks and cigarettes to their ex-fighters. It was therefore decided that I would cover the ex-commanders’ costs, as well as reimburse them for the time taken away from their hustles. As brokers, they de facto worked part-time on the project during two to three weeks. Here it is important to stress that even if the ex-commanders were better off than most ex-combatants, they were not necessarily doing well; none had formal employment and all were engaged in multiple licit, and at times illicit, economic activities to get by. Depending on the size of their ex-combatant networks and the amount of work they did, the ex-commanders were provided with 70 to 130 USD each.

Addressing Broker Induced Biases

It was anticipated that my dependence on cultural and communal brokers risked introducing a number of biases into the study. This was particularly true since my research approach was based on principles of economic recompense, theoretical deduction and limited stays at each field site. A major concern was that the project was too intrusive, particularly when it came to generating a monetary footprint, and that it risked altering the very networks being studied. This is a common source of bias when studying social relations in insecure and poverty-stricken environments (Malejacq and Mukhopadhyay 2016: 1012–1014). I was, for instance, worried that the ex-commanders would seek to inflate the number of individuals in their network by including ex-fighters who had not been under their command or even civilians. The ex-commanders could have seen the project as an opportunity to boost their status as local Big Men and expand their networks of clients by presenting non-ex-combatants as interviewees. This constituted a particular danger if the compensation offered to informants was disproportionally high, underlining the importance of understanding the patrimonial context that the research is conducted in (Utas 2019: 273–274). Another possibility was that my work would cement the ex-commanders’ control over their networks. Again, this could happen if ex-fighters were provided with too generous compensation, whereby the former felt indebted to their ex-commanders. Similar dynamics could, however, also appear without compensation (Hoffman 2014: 6). The mere fact that the ex-commanders were associated with a foreign researcher could have enhanced the former’s standing amongst their ex-subordinates.

My deductive approach also constituted a potential source of bias. A key assumption of the study was that ex-military networks continued to be an important social setting for certain groups of ex-fighters, be it for socioeconomic, political or security reasons. There was a possibility that I, as their employer, would influence the NGO, research assistant and ex-commanders into confirming this assumption. Out of a desire to please me, the two former could have had incentives to only introduce me to ex-commanders who headed strong networks or embellish cultural translations to exaggerate how devoted the ex-fighters were to their former superiors. There were, for similar reasons, a risk that the ex-commanders primed their ex-fighters into overstating their role in sustaining the network. Such concerns must be taken seriously when there are asymmetric relations between research brokers and informants (Utas 2019: 280).

A final worry was that my comparative approach – with its emphasis on analysing multiple ex-military networks – would negatively affect the data generating process. When it came to the cultural brokers, I was particularly concerned that they would present ex-military networks that were too similar when it came to aspects such as exposure to international actors or integration into elite patronage networks. I would thereby lose the analytical leverage gained by comparing contrasting cases. Such forms of selection bias were particularly likely if I confined myself to only collaborating with and tapping into the connections of one cultural broker. What is more, my comparative approach also made me more vulnerable to narrative manipulation by the ex-commanders. Without ‘emerging’ myself for long periods of time at each research site, I had less detailed empirical knowledge to fact check claims made by the ex-commanders. Just like researchers have a tendency to adjust their persona depending on the environment they are navigating, so do ex-fighters and commanders often engage in shape-shifting and adjust their narratives depending on the context and individuals they interact with (Theidon 2009: 23–24). The reasons for engaging in such forms of role-playing can be numerous: a desire to uphold a moral façade, impress on researchers or maximise the chance of gaining access to assistance. I was particularly concerned that the ex-commanders would seek to leave out illegal, or sensitive activities, which they were engaged in.

In order to minimise potential biases caused by collaborating with cultural brokers, my main strategy was to work with two brokers who were independent of each other – i.e., the NGO and university research assistant. I thereby hoped to gain access to different parts of Liberia’s ex-military networks, and minimise the risk of selection bias and hearing one-sided narratives. Studies have shown that working with several parallel networks of informants constitute an efficient strategy to gain a more holistic understanding of dynamics in war-ridden societies (Wood 2006: 375). An added benefit of working with two cultural brokers was that I could ask each broker what they thought about the cultural translations of the other. This constituted an efficient tool to help me detect whether the cultural brokers were censoring certain information in order to please me. Finally, I devoted much time to impress on the NGO and research assistant that the goal of the project was to analyse ex-military networks of various strengths. As such, I needed their assistance to identify both ex-commanders who headed strong and faltering networks.

However, most of my attention was devoted to developing ways to mitigate ex-commander induced biases. As communal brokers, they were both facilitators and key informants, and it was assumed that they had the greatest capacity and interest to unduly steer the research process. More specifically, six strategies were developed to address this challenge. First, all ex-combatant interviews were conducted bilaterally, without the presences of other ex-fighters or their former superiors. This prevented the latter from censoring what was said and created a more conducive environment to speak in.

Second, emphasis was put on triangulating the information provided by the ex-combatants and particularly the ex-commanders. All of the former were interviewed at least once, and some two or three times, while the ex-commanders were interviewed on several occasions. Hence, when new information appeared, I could go back and confront the ex-commanders and ask them to confirm the claims or provide their version of events.

Third, great care was taken to talk to ex-fighters who were not part of the network, but had fought for the ex-commander in question. Since these individuals were, per definition, no longer in the ex-commanders’ orbit, it was necessary to develop alternative ways to identify them. This was done through the help of already interviewed ex-fighters.¹¹ I asked the latter whether they knew any former military colleagues who had fought under the ex-commander, but who were not on the roster. Interviewing these informants, or ‘sleepers’, was seen as crucial for the successful implementation of the project; in case the ex-commanders were priming the members of their networks, it was vital to gain access to independent information. In all, a total of 24 sleepers were interviewed, distributed between the five networks.

Fourth, while the ex-commanders provided information on network membership, details outlining the strength and dynamics of the network were predominantly based on the testimonies of the ex-fighters. This concerned information about how frequently ex-commanders and fighters met, who contacted whom, and who asked whom for assistance. The rational for focusing on the ex-fighters’ depictions, was to minimise the risk that the ex-commanders could manipulate the network narrative.

Fifth, I actively excluded those informants who I suspected to be lying about their background from the study. This included ex-fighters who I judged had not fought for the ex-commanders, or individuals who appeared to lack a military background. In order to identify these individuals, I began each interview with a set of detailed questions concerning issues such as when the ex-combatants had been part of their ex-commanders’ units and where they fought. This was necessary in order to ensure that the ex-commanders were not inflating the size of their ex-command structures.

Finally, since I did not want to give the ex-commanders and combatants the impression that I was expecting them to have strong relations with each other, the interview guide included many questions that had nothing to do with ex-command structures. These questions were geared towards issues such as their wartime experiences and socioeconomic and political situation. Having this broader battery of questions was also vital to ensure that I kept an open mind about the relative importance of ex-command structures. To further avoid priming the ex-commanders and fighters, all questions concerning ex-military network membership were saved for the second half of the interview.

Thanks to these precautions, it was possible to alleviate a number of the biases. It soon became obvious that it was difficult for the ex-commanders to prime their ex-fighters and control the narrative presented to me. Even if some ex-combatants were hesitant to talk about their ex-commanders, others spoke freely about the latter. Several ex-fighters divulged information about illicit activities that their former superiors were or had been engaged in – e.g., selling marijuana, moving stolen goods and mercenary activities – and ties they had to various elites. When I confronted the ex-commanders with this information, they generally verified the truthfulness of these accounts.¹² Perhaps most interestingly, there was no unanimous hailing of the ex-commanders as generous, influential and respected individuals. Even if some ex-commanders were more appreciated than others, there was always an assembly of ex-fighters who gave unflattering reviews of their former superior. I especially remember one interview, where an ex-fighter vividly laughed at me when I asked whether his ex-commander was a respected figure in his community. For him, this was an absurd question and he could never conceive of going to his ex-commander for any kind of help.

Furthermore, through the detailed background questions, I was able to identify a handful of informants who appeared to be misrepresenting themselves. These were either regular civilians or ex-combatants who had fought in other units than the ex-commanders’. These individuals were subsequently excluded from the study and not included in the final network analysis. To this day, I do not know what caused this mishap – perhaps it was due to miscommunication between the ex-commanders and me concerning whom actually constituted the target group, or maybe the former actively sought to include irrelevant informants. Whatever the reason, there was no swarm of individuals seeking to gain access to the study and I have no reason to believe that there were any major efforts by the ex-commanders to inflate the size of their networks. This seems to indicate that the five USD constituted a suitable level of compensation for the ex-combatants. Another fact that speaks against an inflation affect, was that the size of the ex-command structures varied substantially – from 12 to 33 ex-fighters. If my research project offered an exceedingly lucrative opportunity for the ex-commanders, the number of informants should arguably have been more similar across the networks.

Having said this, it is essential to acknowledge that some biases may have persisted. Perhaps there were additional ‘imposter’ informants who I was unable to identify, and maybe the ex-commanders were involved in activities that I failed to detect. When interpreting the results of my egocentric network studies – like most types of social science research – it is therefore vital to see all numbers as indications, rather than hard empirical truths. One benefit of conducting comparative case studies is, however, that biases of a systematic nature – those affecting all cases somewhat similarly – may not be that problematic. This is particularly true when the objective is, as in my study, to explain the relative differences in outcomes (i.e., strength of different ex-command structures).

Unexpected Challenges

Even if several precautions were taken to prevent broker-induced biases, and ensure a somewhat smooth implementation of the study, I inevitably encountered a number of unexpected challenges.

These foremost revolved around problems associated with working with multiple brokers. To recall, a key objective of the study was to systematically compare a number of ex-command structures. This entailed that I had to collaborate with ex-commanders at multiple sites. What is more, because I had limited time to emerge myself in each ex-military network, it was difficult, at least initially, for me to grasp the contextual specificities of each network without the cultural translations of the NGO and research assistant. This obliged me to simultaneously work with both a cultural and ex-commander broker at each respective site.¹³

After commencing with the ex-combatant interviews, I soon realised that this layered approach sometimes generated tensions within the research team. As I built up a rapport with the ex-commanders, I became less dependent on the cultural brokers to facilitate talks, and at times met the ex-commanders without the latter. Even if this was a welcomed development, since it made me less dependent on the cultural translations of the NGO and research assistant, it also generated uncomfortable questions about economic compensation and possible exploitation. On one occasion, a verbal argument broke out between an ex-commander and an NGO official over who should play the role as ‘lead’ assistant. The dispute was instigated by the ex-commander who urged me to fire the NGO assistant, arguing that he was doing most of the work. Presumably, the ex-commander saw an opportunity to increase his cut of the money I was spending. As a relative newcomer to Liberia, it was difficult to assess whether my economic agreements with the cultural and ex-commander brokers were reasonable. In the end, the conflict was resolved with the ex-commander accepting the status quo. However, this arrangement came with a price, as the latter became less cooperative and seemingly less inclined to share his stories. This example highlights that researchers must not only be aware of that inter-broker conflicts may negatively affect the data-gathering process, but also that their work can, at worst, cement local hierarchies of knowledge exploitation. Thanks to their familiarity with peacebuilding discourses, and near monopoly on international connections, cultural brokers possess a clear advantage vis-à-vis their communal counterparts. As such, the former are often well placed to take the lion’s share of the resources, recognition and future referrals that come from being ‘lead’ broker. The methodological and ethical implications of working with multiple brokers is seldom discussed in the literature. Commonly a wide variety of local fixers – ranging from NGOs, local researchers and university students, to youth leaders and various community leaders – are clumped together under the generic label research brokers. This obscures how brokers differ when it comes to education, knowledge of policy/research discourses, and embeddedness in local and international networks. Ultimately, these differences do not only affect brokers’ ability to disseminate knowledge to international researchers, but also their bargaining range vis-à-vis the latter. It is therefore high time that researchers begin to analytically disaggregate between the multiple types of brokers operating at different levels of analysis.

What is more, the fact that I worked with multiple ex-commander brokers, made it more difficult than expected to ensure anonymity between the networks. Even though I engaged with ex-commanders living in three counties, and when it came to Monrovia three different parts of the city, the networks at times converged. When recollecting the war, one rebel ex-commander described how his unit was engaged in fierce fighting against the unit of a famous pro-government commander. To my surprise, the latter was one of the ex-commanders who I was collaborating with. Meanwhile, on a few occasions, the ex-combatants would refer to their relations to other ex-commanders who I studied, but whom they had not fought for. This happened when I asked the ex-fighters whether they had any Big Men who they could go to if they had a problem. This type of information was never the result of prying on my part, but was rather the product of spontaneous recollections divulged by my informants. When similar information appeared, I chose not to dwell on it in order to minimise the risk that the interviewees understood that I had worked with the ex-commander in question. However, it cannot be excluded that facial expressions of surprise may have revealed me. It is also important to note that ex-fighters do not only interact with members of the same unit, but also with ex-combatants of other units and armed groups. As such, there is a possibility that ex-fighters from the different networks shared information about my work, and thereby outed the identity of some project participants. It is therefore crucial to see ex-military networks as being embedded in larger webs of relations, much like Bayart’s (2009) underground rhizomes, rather than operating in isolation from one another. This underlines the need for researchers to seriously consider how to ensure anonymity, or at least minimise the risk of outing informants, in contexts where ‘foreigners tend to “[draw] attention to those in their orbit”’ (Malejacq and Mukhopadhyay 2016: 1018).

Concluding Discussion

Research brokers have become indispensable allies for conflict scholars in their quest to gain access to the field. Even if there is a burgeoning debate amongst particularly anthropologists concerning the pros and cons of working with local fixers, much less has been written from a political science perspective. This is worrying, considering that political scientists are arguably one of the categories of researchers who are most dependent on the services provided by brokers. In this paper, I have shared my experiences of working with research brokers in Liberia. A central take away from this study is that there are measures that scholars – working with political science based research designs – can take to mitigate broker-induced biases. This includes actions such as (a) working with multiple brokers in order to prevent selection bias and hearing one sided narratives, (b) continuously triangulating information given by brokers, (c) interviewing informants not provided by brokers, and (d) identifying and removing ‘false’ informants presented by the latter. By employing similar strategies, it becomes easier for political scientists to dare to leave the ‘veranda’.

A key challenge for many political scientists is that they de facto collaborate with research brokers operating at different stratums of society – cultural and communal brokers. While such layered approaches may be necessary in order to gain access to social groups with strong in-group identities, it also raises important methodological and ethical issues seldom discussed in the literature. First, to what extent can inter-broker dynamics hamper and unduly affect the data-gathering process? Second, what can international researchers do to ensure that their fieldwork does not reinforce exploitative forms of knowledge production that systematically benefits cultural brokers, at the expense of their communal counterparts?

Acknowledgments

I am grateful to Maria Eriksson Baaz, Lotta Themnér, Mats Utas, and the editors and external reviewers at Civil Wars for their helpful comments. This article would not have been possible without the financial assistance provided by the Swedish International Development Cooperation Agency. I am is also grateful for the support provided by the Centre for Gender and Africa Studies, University of the Free State, where I am a research fellow.

Disclosure Statement

No potential conflict of interest was reported by the author(s).

Correction Statement

This article has been republished with minor changes. These changes do not impact the academic content of the article.

Additional information

Notes on contributors

Anders Themner

Anders Themnér is an Associate Professor at the Department of Peace and Conflict Research, Uppsala University. His research focus is on post-civil war democratization; disarmament, demobilization and reintegration of ex-combatants; and transitional justice. His work is predominantly based on extensive field research in Liberia, the Republic of Congo and Sierra Leone.

Notes

1. Schlitz and Büscher (2018: 125).

2. Noteworthy exceptions are Cronin-Furman and Lake (2018) and Vlassenroot (2006).

3. The terms ex-combatant/military networks and ex-command structures are employed interchangeably.

4. The project received ethical approval by Regionala Etikprövningsnämnden i Uppsala, reference no 2011/259.

5. It is important to note that purely inductive studies do not exist. Researchers always depart from some kind of preunderstanding of the social phenomenon they are interested in, and this will affect where and how information is gathered. Hence, it is a question of degrees, where the risk of social desirability bias may be greater in deductive studies.

6. There are additional reasons for why scholars should consider offering compensation to informants who have economic costs due to research participation. First, without reimbursement, scholars risk reinforcing exploitative forms of knowledge production, where the time of the salaried researcher is held as more valuable than that of the researched local. Similar ideas are often based on the problematic assumption that poor people – often making a living in the informal economy – have an abundance of free time at their disposal. Second, there is a tendency to question the appropriateness of providing reimbursement to poor subjects in the developing world. The presumption is that such forms of recompense risk affecting their willingness to give truthful accounts. Interestingly, similar concerns are seldom raised when scholars study ‘Western’ subjects (Head 2009). By refusing to provide reimbursements to informants, researchers risk reinforcing stereotypes of the ‘untrustworthy’ ‘Other’ from the global south (Eriksson Baaz and Utas 2019: 168).

7. I am grateful to Mats Utas who helped to develop the questionnaire.

8. This definition is based on Jezewski and Sotnik’s (2001) discussion on cultural brokers.

9. When working with egocentric network analysis to study persons, the objective is to identify all individuals (alters) that a given individual (ego) has relationships with. As such, the focus is on all alters that are in the direct orbit of the ego (i.e., one step away). This allows the researcher to study the interactions between the ego and alters. This can be compared with snowball sampling, which is a tool primarily used to identify informants, rather than a method to study social interactions. What is more, the objective of snowball sampling is seldom to identify all the alters of an ego, but rather identify the most crucial ones and thereafter ask the latter for referrals to their most important alters. As such, snowball sampling usually generate links to actors who are located several steps away from the ego.

10. The ex-fighters were informed before the interviews that they would receive five USD. This seemed appropriate in order to ease any qualms they may have had that participation would entail an economic cost. The money was given after the completion of the interview. In hindsight, it would have been better if I had offered the money at the start of the interview. This may have reduced misgivings that the ex-fighters may have had and created a more trustful encounter.

11. The ex-fighters who helped me with this task were usually able to identify one, and more rarely two, sleepers. Technically, the former could also be seen as research brokers.

12. When confronting the ex-commanders, I never told them who had given this information.

13. My objective was to work with two research brokers at each site – one cultural and one communal broker (i.e., ex-commander). However, when working with the ex-military network in a semi-urban community, the NGO had to employ a youth leader to reach out to the intended ex-commander, creating a three-layered broker link.