The Persistence of Power: Reflections on the Power Dynamics in a Merging of Knowledge Research Project

Collaborative research approaches emphasise the need to transform the way the academic community produces science by integrating knowledge from different disciplines, but also by including non-academic knowledge in order to address the challenges of sustainability and social justice. This approach – known in the literature on sustainability science as transdisciplinarity – has been used increasingly in research to resolve sustainability problems, including those related to poverty and socio-economic inequalities. This article seeks to shed light on the power dynamics that exist and emerge in transdisciplinary processes by analysing a case study on food poverty. Following Fritz and Meinherz’s (2020) approach, I use Amy Allen’s (1998) typology of power to track and trace the way that power played out between and within actor groups in a project that applied a transdisciplinary methodology known as the ‘Merging of Knowledge’. Although the Merging of Knowledge model seeks to identify and address power differentials between the participating groups, power relations remain complex, dynamic and – to some extent – inevitable. Collaborative processes would benefit from an analysis of the way that power dynamics emerge, persist and evolve to enhance awareness of different forms of power that coexist in research, and to ensure that imbalances present outside the research process are not reproduced within it.


Introduction
In the face of an unprecedented environmental crisis, rising inequalities and persistent poverty, the need for an ecological and social transition has become urgent. Collaborative research approaches (see e.g. Hirsch-Hadorn et al. 2006) emphasise the need to transform the way in which the academic community produces science by integrating knowledge from different disciplines, but also by including non-academic knowledge in order to address the challenges of sustainability and social justice. This approach is known in the literature on sustainability science as transdisciplinary research (TDR). It has been used increasingly in diverse areas of research, including those related to poverty and socio-economic inequalities (see e.g. Conde 2014; Marshall, Dolley & Priya 2018).
Interrogation of how power operates within transdisciplinary research has received relatively little attention in the literature, despite its central links to knowledge (Fritz & Meinherz 2020). In this article, I draw on Amy Allen's (1998) typology of power to analyse the power relations in a case study conducted using a transdisciplinary methodological approach known as the 'Merging of Knowledge'. I first define what is meant by 'transdisciplinarity'; I then present the historical background to the Merging of Knowledge approach, outlining the tools and methods used to ensure the active involvement of all participants in research processes. Next, I present the case study: a research process conducted between 2018 and 2019 involving (current or former) users of food assistance programs, social workers involved in the distribution of food assistance, and academic researchers from different disciplines. Using material from ex-ante and ex-post interviews that I conducted with the project participants, I analyse the power relations that emerged, persisted and evolved during this research process. Through this exercise, I hope to shed light on the multidimensionality of power and how it plays out in a research process. I argue that it is insufficient to put in place tools and methods that attenuate only the most obvious power dynamics at play; instead, a more nuanced understanding is necessary if we are to construct a research process in which each individual and group's knowledge is integrated. It should be noted that the analysis and views expressed in this article are my own attempts at a reflexive exercise on the power dynamics at play in a research process that I initiated and co-led and may not reflect the perspectives of other participants in the research process. This article is based on several sections of my doctoral thesis.

Theoretical Framework
In this article, I seek to adopt a Foucauldian understanding of power that sees it as dispersed and distributed across social relations rather than as a force exerted unilaterally by an individual or a group onto another (Gaventa & Cornwall 2009, p. 467). As rendered explicit by Barnaud et al. (2016), beyond a naive dichotomy between strong and weak actors, power can intervene in many different ways and forms. In 'Rethinking Power', Amy Allen (1998, p. 22) revisits the feminist literature to offer a typology of power that seeks to overcome the 'conceptual one-sidedness of existing approaches'. In other words, Allen refuses the idea that power exists only in a single form, instead arguing that it can take several forms, all coexisting within relationships. She distinguishes between two different conceptions of power that were typically considered in opposition to one another (i.e. existing either as one or the other, but never concurrently): power as domination and power as empowerment (Allen 1998). The first type of power -'power-over' -Allen argues, refers to a 'master-subject' relationship, whereas the second refers to agents' 'power to transform themselves, others, and the world' (Allen,1998, p. 7). 'Power-over' is thus to be distinguished from the second conception of power ('power-to'), which she defines as 'the ability of an individual actor to attain an end or a series of ends' (p. 34). Empowerment or resistance constitute forms of 'power-to' because they demonstrate 'how members of subordinated groups retain the power to act despite their subordination ' (p. 35). In addition to these two types of power, Allen introduces a third dimension: 'power-with', which she defines as the 'ability of a collectivity to act together for the attainment of a common or shared end or series of ends' (p. 35). Thus, rather than a binary, oppositional arrangement, Allen sees power as taking three possible forms (power-over, power-to and power-with), which can and do coexist: as she writes, 'Each tells only one side of the story ' (p. 22). Similarly, in a transdisciplinary process, it may be too simplistic to view experts (e.g. researchers) as holding power. Indeed, while they may control some or part of the process, they are also likely to be responding to the requirements and demands of funders, for instance, who, in turn, also constrain their choices.
In this article I thus seek to go beyond the 'one-sidedness' of each of the types of power that Allen identifies, acknowledging, as she does, the coexistence and persistence of various forms of power within the same relations. For example, just as a woman may not only be dominated (the subject of 'power-over') by a man, but may also at the same time resist that power (demonstrating 'power-to'), I acknowledge the complexity and multidimensionality in transdisciplinary research, where certain actors or groups may simultaneously hold and lack different forms of power. Indeed, borrowing from and building upon the literature on Participatory Action Research, an analysis of power in transdisciplinarity must go beyond the simplistic idea that power is 'an attribute that some have and others [lack]', with structures, organisations and experts holding power and the oppressed, grassroots or marginalised lacking it (Gaventa & Cornwall 2009, p. 465). By drawing on ex-ante and ex-post interviews with the groups involved in the Merging of Knowledge process, I show that the story is, indeed, far more complex. This echoes the work of Fritz and Meinherz (2020) who conducted a multidimensional analysis of power in transdisciplinary sustainability research, identifying dynamics of 'power-over ', 'power-to' and 'power-with' in the different phases of the research, from the development phase to the dissemination phase.

Merging of Knowledge as a Transdisciplinary Approach
Transdisciplinary research (TDR) is defined by its key authors as a 'reflexive, integrative, method-driven scientific principle aiming at the solution or transition of societal problems and concurrently of related scientific problems by differentiating and integrating knowledge from various scientific and societal bodies of knowledge' (Lang et al. 2012). In this article, we consider the Merging of Knowledge, as applied in this case study, to be a transdisciplinary approach because it seeks to integrate knowledge from academics specialising in different disciplines, practitioners and persons experiencing poverty, thus both the participatory and interdisciplinary conditions are satisfied.
Participatory approaches to poverty research arguably began with the use of Participatory Rural Appraisals (PRAs) and Participatory Poverty Assessments (PPAs) by international organisations in the 1990s. These approaches sought to complement household surveys by capturing poor people's perceptions of their own situation and possible solutions with a view to improving the effectiveness of poverty eradication policies. Around the same time as the World Bank was scaling up its use of PPAs, the international poverty eradication movement, known as ATD Fourth World -short for All Together in Dignity -began experimenting with a different kind of participatory approach. The Merging of Knowledge (MoK), as it came to be known, was based on the intuition of ATD's founder, Joseph Wresinski, as to the capacity of persons experiencing poverty to be teachers of others and to engage in a dialogue with other members of society (Ferrand et al. 2008, p. 15). Like feminist epistemologies and decolonisation theories, which consider each person as possessing knowledge coming from their own situated experience of reality, ATD Fourth World's philosophy began from the conviction that persons experiencing poverty have the potential to bring unique thoughts and actions to the knowledge production process, based on their life experience (Ferrand et al. 2008, p. 15).
From its early days, the international movement has been firmly rooted in this conviction, and it seeks to integrate the knowledge of persons experiencing poverty by involving them at all levels and in all spheres of the organisation. Already in 1980, speaking at a UNESCO committee session composed of academics, Joseph Wresinski spoke of the duty of researchers working on questions related to poverty to 'make room for the knowledge held by the very poor and excluded about their own condition and the world that imposes it on them, to rehabilitate [this knowledge] as unique, essential, autonomous and complementary to all other forms of knowledge, and to help it to develop …' (Ferrand et al. 2008). To this, he added that the knowledge stemming from practitioners working alongside people in poverty, as well as the knowledge of researchers as 'external observers', are both complementary to that of persons experiencing poverty themselves. In his speech, Wresinski conveyed the idea that the knowledge of researchers is 'partial' (Moosa-Mitha 2015) and that they should assume an 'attitude of a learner, one who does not know but through the act of empathetic imagination and by possessing critical self-consciousness comes to garner a sense of what the other knows' (Moosa-Mitha 2015, p. 89).
Three years later, in 1983, Wresinski once again turned towards universities at a conference held at the Sorbonne. In his speech, he evoked the 'moral and political responsibility and the scientific rigor that obliges Universities to turn towards the Fourth World, not to teach, but to learn and to engage in a dialogue' (Ferrand et al. 2008, p. 16), noting that 'the time has come for a reciprocity of knowledges … It is time to ask a population that is at the foot of the social ladder to deliver us its thoughts that only it knows' (Ferrand et al. 2008, p. 16). From 1993 to 1995, a working group composed of ATD Fourth World activists (persons experiencing poverty), permanent volunteers (members of the ATD volunteer corps) and academics became involved in designing a research program that would seek to elaborate the conditions necessary for reciprocal dialogue among three types of knowledge: the knowledge of those with the experience of poverty and social exclusion, the knowledge of those working with persons in poverty and the knowledge of academics.
Since the launching of the research program, the MoK approach has been used throughout the world to support work in health, social work, education, the humanities and social sciences. Most recently, an international study led by ATD Fourth World and Oxford University, titled 'The Hidden Dimensions of Poverty', made use of the principles and practice of the MoK to contribute to the debate on defining and measuring poverty. In Merging of Knowledge projects, people experiencing poverty are included in every phase of the research. The projects are often governed by a steering or coordination committee. These committees are responsible inter alia for recruiting participants, designing the methodology, facilitating meetings, and taking and transcribing notes during peer group and plenary sessions. While the steering committees ensure that the project goals are met throughout the phases of research, the remaining participants (i.e. persons experiencing poverty, practitioners and academics) also participate actively, for instance in the collaborative definition of the research question, the collection of data, analysis of findings, and the drafting of key project outputs such as reports.

Applying the Merging of Knowledge: Tools and Methods
The principles applied in MoK research are outlined in the 'Guidelines for the Merging of Knowledge and Practices when working with people living in situations of poverty and social exclusion'. They outline the prerequisites for merging knowledge as well as the conditions for implementing the approach.
According to the first of the prerequisites, every actor engaging in the MoK must 'be aware that change is necessary'. In other words, he or she must recognise that poverty is not inevitable and that the current social, economic and cultural realities must be changed. Secondly, one must 'see each and every person as possessing knowledge' and be capable of distancing themselves from their own situation to reflect on it. Third, nobody should be left on their own, that is, each participant in a MoK process should be part of a group 'which reinforces and consolidates' their knowledge and provides security, as well as space and time, to reflect and express themselves. These groups are known as 'peer' groups, which operate as 'safe spaces' for participants, particularly for those who are experiencing poverty or whose knowledge is more fragile and/or Osinski Gateways: International Journal of Community Research and Engagement, Vol. 14, No. 2 January 2022 4 marginalised. As Carrel et al. (2018) note, considerable attention is paid in the MoK to ensure that persons experiencing poverty (as well as the other participants) do not contribute to the research based on their individual experience alone, but that they construct knowledge based on their collective experience of living in poverty. Therefore, before knowledge from the three groups of participants is effectively 'cross-pollinated' between groups in plenary sessions, it must first emerge through a collective effort. This is achieved in these peer groups, in which activist, practitioner and academic participants meet on their own. Referring to such safe spaces, philosopher Iris Marion Young (1990, p. 168) notes that: … Members of oppressed groups need separate organizations that exclude others, especially those from more privileged groups. Separate organization is probably necessary in order for these groups to discover and reinforce the positivity of their specific experience … Contemporary emancipatory social movements have found group autonomy an important vehicle for empowerment and the development of a group-specific voice and perspective.
Young's contribution thus highlights two important functions of these peer groups. Safe spaces are important to develop a 'group-specific voice and perspective', but they are also 'an important vehicle for empowerment'. Indeed, for members of disadvantaged or oppressed groups, such safe spaces allow them to find reassurance and comfort in the presence of others who may have lived through similar experiences, and for all groups, they enable the construction of collective thought through the confrontation and triangulation between individual knowledge and experience. This has been identified by Patricia Hill Collins as 'the power of self-definition' (Collins 2000, quoted in Carrel et al. 2018).
The fourth and final prerequisite for the MoK is that each and every person must be seen as being part of the research team. This entails each participant being involved -to varying degrees -in each stage of the research process. Participants are the subjects, not the objects, of the research, and they participate in its elaboration and execution. In many cases, MoK processes are initiated by ATD Fourth World, or as a joint initiative between a research institute and ATD Fourth World. This is not common practice, even in transdisciplinary research. As Rosendahl et al. (2015, p. 23) note, in transdisciplinary projects, the 'initiative is often taken by scientists alone, who become responsible for engaging other actors more deeply connected to the practicalities of the issue'. This situation potentially leads to an unbalanced ownership of the project and fails to empower the non-academic participants, in particular the persons experiencing poverty.
Once the prerequisites are met, the MoK can take place under specific conditions outlined in the guidelines. First, the physical presence of people living in poverty must be guaranteed: they cannot participate through 'simple verbal or written testimonies or video presentations'. Moreover, nobody can intervene on their behalf. Second, the conditions must be met to ensure that the different knowledges can be shared: on one hand, participants should be independent of one another (i.e. no teacher-student or social worker-client relationships) and peer groups must be formed. Third, a space of security and trust must be created, including a form of contract specifying the confidentiality rules, as well as an ethical framework outlining the values inherent in a dialogue between people: 'active listening, respect for what the other says, a willingness to be critical about one's own knowledge and ideas, and a conviction that all knowledge is always "under construction"'. Fourth, the conditions for a true dialogue must be guaranteed by applying a range of tools and methods necessary for 'creating the conditions where everybody's voice carries the same weight', i.e. for seeking to attenuate power differentials between participants and participant groups.
These tools and methods include the presence of discussion facilitators (one per peer group and ideally two for plenary sessions), who must hold extensive knowledge and have considerable experience in working together with persons experiencing poverty. Most often, the facilitators are members of the permanent volunteer corps of ATD Fourth World. This function has been described by some as facilitation, support, mediation or 'gatekeeping' (Carrel et al. 2018) without a clear formalisation of the role. While the image of a 'bridge' enabling the mediation between two different worlds or thought styles has been evoked, it has also been rejected by some as problematic because, according, for example, to Francoise Ferrand, one of the early practitioners of the Merging of Knowledge, the role of the ATD permanent volunteer is not to mediate, but rather to be 'on the side of the persons experiencing poverty' (Carrel et al. 2018), i.e. to fully support them and advocate in their favour. Others have pointed out that the role of the facilitator is to avoid arriving at a situation in which the persons experiencing poverty become silent and unable to express their criticism, when faced with the expertise and eloquence of the other, more powerful participants (Carrel et al. 2018).
The guidelines specify that these conditions should be put in place by a steering or coordination team, which 'should be made up of people who have known, over a good number of years, those living in poverty, their difficulties and resources, and also people from the world of academics or professionals'. The role of discussion facilitators is to 'ensure that all participants can express themselves and be understood, and also respect the time given to each person to speak'. Other foundations of the MoK approach include using one's life experience as a starting point for exchanging knowledge because ' [g]iving an account of one's life experience allows all the participants to start on an equal footing'), taking the time to build trust among participants and to establish meaningful dialogue, and building together in order to improve the relationships between the different participant groups seeking to build the potential to experience 'power with'.

The Case Study: 'Food Assistance: What Alternatives?' Project
The project that serves as a case study for this analysis was initiated by the author (a doctoral researcher) who contacted ATD Fourth World in Belgium to launch a collaborative research project on the way that persons experiencing poverty participate in the sharing economy. Initially, the problem field was left partly undefined, leaving ATD to decide whether it preferred to focus on the sharing of objects, mobility, food or housing. After a period of reflection, the organisation specified that food-sharing, and food assistance more specifically, constituted a key problem for its members. On the one hand, the use of food banks and soup kitchens by persons experiencing poverty may be a source of humiliation. On the other hand, such places remain critical for the survival of increasingly large portions of the population. The term 'food assistance', as used in this article, refers broadly to 'a continuum of practices between welfare and charity initiatives that aim to alleviate food poverty and hunger' (Hebinck et al. 2018).
The research project received funding from several sources. Part of the project was financed by the doctoral student's university through funds provided for a research project about the sharing economy. The funds to cover the materials and meals for participants were provided by a foundation, ATD, the Federation of Social Services (FdSS) and the university (UCL), and each assigned a staff member to work part-time on the project. The funding bodies had little say over the project and no explicit expectations. They did not intervene throughout the process.
The project was conceived as a small-scale Merging of Knowledge research project. A coordinating team was constituted, consisting of the doctoral researcher, a member of the permanent volunteer corps at ATD Fourth World and a staff member of a federation representing food assistance organisations (the FdSS). Participants were recruited by the coordinating team: between five and seven persons experiencing poverty (former or current users of food assistance), between five and seven social workers involved in the distribution of food aid, and five academic researchers, hailing from different disciplines (law, political science, sociology). The three members of the coordinating team were responsible for recruiting participants, defining the content and methodology of the workshops, taking and transcribing notes, and coordinating the research project. A legitimate question concerns the lack of activists (persons with direct experience of poverty) in the coordinating team. In Merging of Knowledge processes, activists are sometimes -but not always -involved in the coordinating teams of such participatory studies. Whether they are included in the coordinating teams and the extent of their participation depends on their experience, their legitimacy vis-à-vis their peers, the time taken for the study and decisions regarding methodological components, and their willingness to take part in decisions concerning methodology, strategy, etc. In this context, no specific activists were identified (or self-identified) who could ensure such participation on the coordinating team, also given the short time-frame of the study.
However, as noted below, the activist peer group was intensely involved throughout the process. Seven participants in the research project (later this number was reduced to five when two members of this group left the project) were persons with experience of poverty. Two key conditions guided the selection of participants in this group. First, the participants were required to be or had been involved in ATD Fourth World's people's universities for some time. ATD Fourth World's people's universities are 'a forum where people from different backgrounds can come together … The basis of this encounter is the recognition that the knowledge and experience of people living in poverty has value. Therefore, this type of forum creates a space where people freely express themselves. This gives birth to new ideas, projects and proposals' (ATD Fourth World n.d.). Second, they had to be former or current users of food assistance programs (including soup kitchens, social pantries and food parcel distributions). This ensured that their participation was based on and grounded in an actual experience of poverty and of food assistance programs; however, they also had to have undergone a form of 'training' through participation in people's universities, which provides an experience of public speaking and analysis of a wide range of subjects relating to poverty.
It is important to note that the group of activists met extensively within their peer group outside the five Merging of Knowledge workshops to prepare future sessions and debrief previous ones. They worked closely with the member of the permanent volunteer corps of ATD Fourth World -who was also part of the coordinating team -to provide feedback on the methodology followed. In this respect, the role of ATD Fourth World was key: by preparing the participants for each workshop and supporting them throughout the process, engagement and trust were established early on in the process.
Initially, seven participants in the research project (then five, as with the above group) were social workers or other employees of food assistance programs. The projects in which they were involved varied considerably: some social workers were employees of large-scale international organisations involved in the distribution of food assistance; others were working for neighbourhood soup kitchens or local food pantries. While four of the five social workers came from 'institutionalized' food assistance programs, one participant worked at a grassroots initiative that distributed unsold organic food products without any conditions and on a pay-what-you-want basis.
Five participants were academics involved in research and/or teaching activities. The academics were four doctoral students and one professor, hailing from different disciplines (law, political science and sociology). Each of the academics focused on themes directly or indirectly related to food assistance in their research: human rights and discrimination, sustainable food systems, poverty and homelessness, and food assistance policies.
The research process was organised as five full-day workshops that were held over a six-month period at the headquarters of ATD Fourth World in Belgium. The first two workshops were devoted to collectively defining the specific research question that would guide the rest of the research project. A number of tools were used to obtain a consensus on this research question among all above-mentioned participants. After two full-day workshops dedicated to this step, the research questions were ultimately phrased as follows:

'How can we understand the violent and degrading situations (in food assistance)? Through what other possibilities can we improve the relationship between food assistance users, services and society in order to contribute to creating a real place in society and a better life for those experiencing the most difficulties?'
The next three workshops were devoted to answering the research questions to the extent possible. Again, a number of tools were used, based on the methodology prepared by the coordinating team. The coordinating team first asked the persons experiencing poverty and the social workers to produce narratives/ stories about difficult situations in food assistance that they had experienced first-hand. Two stories were selected by the coordinating team and these were collectively analysed in the workshops. The analysis was organised in two steps: first, the same story was analysed separately by each of the three peer groups, i.e. persons experiencing poverty analysed the story in their group, as did the social workers and academics. Second, the analyses were shared in a plenary session. A second tool that was used was a simplified version of photovoice, a technique that facilitates expression and dialogue using images. By asking participants to select a photo which best captured the message they wanted to deliver to the other peer groups, the photovoice exercise enabled individuals to express themselves openly about the perceived problems related to their experiences and knowledge of food assistance. It also enabled the full involvement of those participants who could not read and write.
During the fourth workshop, peer groups were mixed. The two 'mixed groups' worked on analysing a second story, which was presented to the other group as theatre scenes, drawing on 'forum theater', a tool developed by Augusto Boal and others in the favelas of Sao Paolo. In this exercise, the audience (i.e. the other group watching the theatre scenes) intervenes by replacing characters in the scene in order to redress a situation perceived as unjust. A facilitator assists the group in analysing the situations, the alternatives explored through the interventions, and the effects of these on the characters and the unfolding of the scene. The aim of the exercise is to enable 'participants to try out courses of action which could be applicable to their everyday lives', thus becoming a 'political tool for change' (Farmer 2021). In the project, forum theatre was used to illustrate two scenes related to the violent and degrading situations that arise in the distribution of food assistance.
The fifth workshop was devoted to summarising the evidence collected throughout the research process, formulating conclusions and discussing the next steps of the project (the preparation of a public presentation and the drafting of a report).

POWER-OVER
Amy Allen (1998, p. 33) defines 'power-over' as 'the ability of an actor or set of actors to constrain the choices available to another actor or set of actors in a nontrivial way'. There are several ways in which groups or individuals exerted power-over others in this research project. First and foremost, the coordinating team played a central role throughout the process, including by inviting and recruiting participants based on their profile and interest in the project, defining the research topic and designing the methodologies for the five transdisciplinary workshops. The coordinating team also held considerable power over the process and its participants by facilitating the workshops, distributing speaking time, choosing to emphasise or accentuate certain aspects rather than others, etc.
Beyond the power exerted by organisers of a participatory process (who may have control over the inclusion/exclusion of participants, the capacity to shape the theme and the agenda, etc.), a common risk in participatory processes relates to the power that dominant individuals exert over others. This can occur when an individual or group of individuals that are well-endowed in 'bases of power' (French & Raven 1959) impose their ideas on others by ignoring or dominating the views of others (Barnaud et al. 2016). Based on previous analyses of power relations in participatory processes, a common dynamic that arises in projects involving academic and non-academic participants relates to the power exerted by the former over the latter. For example, van der Riet and Boettiger (2009, p. 4) note that in the context of rural South Africa, 'the power dynamics between resourced, urban, educated researchers and under-resourced and marginalized research participants are often accentuated'. However, in the case of the project presented in this article, it seems that the researcher peer group did not exert significant power-over the non-academic participants. In other words, the academics did not unilaterally impose their preferred research question onto the group (it was negotiated by the three peer groups), nor were they involved in designing and leading the methodological choices of the research process (this was done by the coordinating team). They also did not take the floor in a disproportionate degree, nor serve to either validate or discredit the experiential or action-based knowledge of the activists and practitioners. I suggest that this may be explained by several interconnected reasons. First, the members of the academics' peer group seemed to be sensitive to power relations themselves. As one academic put it: Especially at the beginning, as researchers, the place of research was questioned. We were holders of knowledge, but we had to be careful not to become holders of power to orient the group too much.
Second, the researchers were not experts in food assistance per se; instead, they were specialised in relevant subjects that were directly or indirectly related to the theme of the research. As a result, despite their status as 'holders of knowledge', they did not seem to be 'holders of power' since they did not, in fact, possess specialised expertise on the subject at hand. Third, most of the academic participants were early-stage researchers who had not taken part in participatory research in the past. Interviews with the members of the academics' peer group suggest that the researchers were eager to engage with the activists and practitioners and to learn from their 'real-world' experience. Their posture thus seemed closer to that of students than that of experts. As one academic later recalled: Before, I had the intuition that in the academic world, we are a little detached from the real world. And here, I saw that it is true.
In a separate interview, a practitioner expressed the position of the academics in the following way: I think most of the academics were a little detached from these problems. They didn't know what food assistance is and how it is organized. They had a very simplistic image of food assistance and they had trouble understanding the main issues behind it. During the forum theater exercise, they were the first to say that their solutions were imperfect, and that they had difficulties finding possible answers. I think they revisited their positions and they have a better understanding of reality now.
Indeed, as one academic put it: I saw the system from the inside. I had never done research specifically on food assistance before.
Based on the interviews and my own observations of the interactions between the three peer groups, the academics' group did not exert significant power-over the two remaining groups. Instead, the researchers' group often served as a buffer -or mediator -between the activists' and the practitioners' groups, particularly in conflictual situations.
On the other hand, the activists played a central role in the process, given that Merging of Knowledge seeks specifically to integrate the knowledge of persons with the experience of poverty. Without them, the process could not have taken place. Had the activists decided to leave the research, the project would have ended. This gave the persons experiencing poverty (the activists) significant power-over the other participants, because the activists' presence was required for the legitimacy of the project. It may explain why some persons participating in or observing the process expressed the feeling that the two other peer groups (the practitioners and the researchers) made significant efforts to please, or comply with, the activists' requests. Moreover, the research process physically took place within the walls of the organisation that sought to represent and empower the persons experiencing poverty -ATD Fourth World. This may have given the activist group additional security throughout the process. This impression is confirmed by the reflections expressed by one of the activists in the ex-post interview: What I realized at times is that the academics and activists were on the same wavelength, and at other times the practitioners and activists were on the same wavelength. It depended on the subject … For me, I did not feel that our group of activists was isolated.
This excerpt shows that the activists played a central role; at times, the practitioners 'sided' with them, at other times, the academics and activists grew closer.
Finally, negotiation theories often discuss the BATNA (Best Alternative to a Negotiated Agreement) of participants. The higher a negotiator's BATNA, the lower his or her incentive for negotiations to succeed, because the possible alternative to a negotiated agreement is advantageous to them. In the Merging of Knowledge process, the persons experiencing poverty had the lowest BATNA of all other participants: they had the least leverage for producing an outcome favourable to them outside of the negotiated research process. In other words, the status quo situation -food assistance as it is currently organised -is least favourable to them, marking the importance for a 'negotiated' or collective research process to bring about positive changes. Thus, their group was both essential to the Merging of Knowledge process, whilst at the same time, to the extent that they worked on behalf of other persons experiencing poverty beyond their own immediate interests, they also had the most to lose from leaving the process.

POWER-TO
According to Amy Allen (1998), 'power-to' is 'the ability of an individual actor to attain an end or a series of ends'. The unequal distribution of financial and time resources is an objective measurable factor that should be taken into consideration when analysing this ability. In the Merging of Knowledge process, some of the participants were involved in the process on their own time, while others were engaged in the research professionally and could therefore include the time spent in the workshops in their working hours. Moreover, the financial situation of the practitioners and academics was overall significantly more comfortable than that of the activists' group, some of whom were still dealing with very difficult material and social conditions, such as homelessness and/or poor living conditions, unemployment, single parenthood, etc. It should be noted that, while their transportation and meals were reimbursed, the activists were not financially compensated for their involvement in the project. However, ATD Fourth World continuously provided other forms of support before, during and after the study was completed, ensuring a supportive, long-term relationship with the activists. Despite these differences, the methodology sought to involve all participants on an equal footing. The approach required each person to participate fully in the study, including in the various exercises, such as photo voice and forum theatre. By taking part in exercises that forced them to abandon their usual roles and comfort zones, the participants encountered each other on a more level playing field. As one academic put it: Since we participated in the forum theater, we stepped out of our comfortable role as intellectuals and we were obliged to position ourselves among the practitioners and the activists.
Moreover, the use of spokespersons and the active role played by the facilitators sought to balance the power relations within the peer groups. Indeed, the rotating function of spokespersons enabled each person within the peer groups to experiment with speaking on behalf of their group in plenary sessions and taking on a leadership role. However, this form of power-sharing among the participants of different peer groups should be nuanced. Indeed, within peer groups, some power differentials may have remained. For example, as mentioned before, in the academics' peer group, four of the participants were doctoral researchers, while one was a tenured professor. As one academic remarked in an ex-post interview: Even among researchers, we sometimes had trouble understanding one another: we speak different languages … I think the other groups did not notice, but there are differences between where we are at in our trajectories …

POWER-WITH
Amy Allen (1998) defines 'power-with' as the 'ability of a collectivity to act together for the attainment of a common or shared end or series of ends'. Although the research project has been alluded to as a negotiation process in this article, the participants were also encouraged to attain common goals from the very beginning of the project. Indeed, the first two transdisciplinary workshops were devoted to the collaborative definition of the research question, with every member of the research team participating in this phase. The collaborative definition of the research question was conducted as follows: first, the 'peer groups' brainstormed potential questions concerning the food assistance system and food aid initiatives. In a plenary session, these questions were presented to the other peer groups. In a second phase, the coordinating team reorganised these questions, categorising them as responding to the questions of 'Who?' (i.e. questions concerning the beneficiaries of food assistance), 'What?' (i.e. the content of food assistance programs), 'How?' (i.e. the practical organisation of food assistance), 'What effects?' (i.e. the effects of this assistance on beneficiaries), and 'What alternatives?' (i.e. potential alternatives to existing food assistance programs). After a collective discussion about these questions, the peer groups met once again to propose a single research question to the other groups. In a final phase, the groups reconvened in a plenary session to negotiate and agree on a research question together. The objective was clear: the different groups were challenged to go beyond their differences and to find common ground in order to agree on the question that would guide the remainder of the research project.
Collective empowerment was also sought by building trust among participants, who gradually transitioned from working in 'peer' groups to working in mixed groups, requiring interaction with the members of the other groups. As one participant put it: I don't know if I'm over-interpreting it, but I had the impression that there was more distrust … at the beginning, and then it became warmer and more understanding at the end.
Collective empowerment also seems to have been one of the key outcomes of the research process, based on ex-ante and ex-post interviews conducted with all of the participants (Osinski 2020). Indeed, one of the questions in the interviews asked participants whether they felt that they had the power or means to make a change in the way food assistance is distributed in Belgium. Although at the beginning of the research process many of the participants interpreted this question as relating to their individual power or means to make a change, by the end of the research process, they answered the question by referring to collective efforts. For example, in the ex-ante interview, one participant answered:

Yes, through my motivation and my desire. For me, it's really through practice, through putting in place initiatives, and through work …
At the end of the process, the same practitioner answered as follows: … In this framework, clearly, I do not feel that I can change anything in terms of public policy. But what has changed since this Merging of Knowledge, is that I have a better understanding of the actors and I think we should create a lobby, to put ourselves together, to create a collective in order to have more impact and be invited to these types of events and political discussions … Collectively, I think we could have an impact …

Discussion
Using Amy Allen's three notions of power, the analysis above shows that power relations as observed in the 'Food Assistance: What Alternatives?' project were complex and dynamic. While no single individual appeared to dominate the entire process, different groups of actors constrained the choices of others throughout the project, exercised their ability to act or reach certain objectives and worked together to achieve a common purpose. Unlike in other analyses of power in participatory research, it does not appear that the group of academics unambiguously exerted power over others; instead, depending on the perspective taken, the notion of power examined and the stage of the process, both the practitioners and the activists seemed to hold certain forms of power.
Paradoxically, perhaps, the coordinating team, whose role was -among other things -to attenuate the power relations that existed and emerged among the various participants, seemed to hold the most power over the process. Its members held key responsibilities -and thus also control -in defining the topic for research, selecting the participants, the methodological approach and the specific tools to be used. The coordinating team also facilitated the peer groups and the plenary sessions, which gave its members considerable power-over the direction taken in the workshops, the distribution of speaking time and the rhythm of the sessions.
Power was also exercised by other actor groups. All three peer groups were empowered to participate fully at all stages of the research; moreover, collective empowerment was achieved through the negotiation of the research question by all participants involved, as well as through the trust that was built both within and across peer groups. I argue that the reason for the shift from an individual understanding of power ('powerto') to collective empowerment ('power-with') might be explained by the fact that, while participants may have initially been optimistic about their capacity to change or improve the current food assistance system, the research process demonstrated the complexity of the problem at hand, leading them to re-evaluate their individual agency in this regard. The intensive cooperation between different actor groups over a relatively long period (six months) was likely also to have convinced several participants of the value of working together to attain common goals.
While the Merging of Knowledge seeks specifically to attenuate power differentials that exist between academic researchers, persons experiencing poverty and practitioners, the guidelines do not provide insight or guidance on handling power relations within peer groups. Indeed, while the presence of an experienced facilitator is considered to be key to moderating the peer groups as well as the plenary sessions, little has been written about the challenges and potential solutions for resolving issues of power among persons experiencing poverty, among practitioners or among researchers, as if these groups are considered to be homogeneous in terms of power distribution. This is unlikely to be the case in reality: power relations can exist between members of different disciplines in the academic world, between different levels of power in organisations for practitioners and between different individuals for activists. For example, race and gender may play an important role in the power that individuals hold in a collaborative process. Thus, further research could focus on how to handle such power differentials within the peer groups themselves.

Concluding Thoughts and Recommendations
Power pervades research in a variety of ways and forms. While methodological approaches like the Merging of Knowledge, developed by ATD Fourth World in the 1990s, seek to attenuate power dynamics between groups of participants, the unequal distribution of financial and time resources, differences in bargaining power and BATNAs (best alternatives to a negotiated agreement) and different positions in the research process are -to a certain extent -inevitable. In the research process of the 'Food Assistance: What Alternatives?' project, the considerable power held by the coordinating team shaped the process through its selection of the participants, the research topic, the methodological approach and the tools used. Moreover, the analysis of power dynamics conducted in this article sheds light on the different -and sometimes unexpected -ways in which actor groups have 'power-over' others and the ways in which participants are empowered, both individually and collectively. The interviews conducted before and after the process generally show that, while individual feelings of empowerment remained stable or waned, collective empowerment grew as a result of the collaborative research process and the realisation of the complexity of the problem at hand. In other words, participants realised that, on their own, triggering a change in policies related to food assistance was difficult; together, however, they may be able to achieve more.