Governance Diaries: An Approach to Researching Marginalized People’s Lived Experiences in Difficult Settings

Abstract

How do chronically poor and marginalized citizens interact with and make claims to the different public authorities that exist in fragile, conflict and violence-affected contexts? In other words, how does governance from below look like in difficult settings? Given the centrality of the ‘leave no one behind’ agenda, an understanding of how such populations meet their governance needs can help identify the constraints to achieving development for all in these challenging settings. We wanted to research these questions comparatively, to see if there were common features of response in different contexts, with the presence of various kinds of non-state actors, diverse histories of colonialism and authoritarianism, and widely different social norms. In this article we describe the governance diaries approach, an iterative alternative to large-n surveys and multi-sited ethnographies we developed in the process of answering these questions. Governance diaries, working as a qualitative panel data, are a suitable approach for researching complex behavior that changes over time as large-n surveys are insufficiently dynamic to trace the processes behind change (lacking sensitivity) and ethnographic studies often have limited generalizability (lacking comparability). We describe here how this approach works and the challenges and opportunities it offers for research.

Introduction

What does governance look like for those governed in settings affected by fragility, conflict, and violence? How can we find out? This article describes an approach that we evolved in the process of answering this question— governance diaries. We wanted to understand what empowerment and accountability meant to chronically poor and marginalized people living in these settings. How do people navigate the formal and informal institutions that govern their lives? What historical and socio-political influences shape their behaviors? We wanted to research these questions comparatively, to see if there were common features of response in contexts with different types and levels of conflict, the presence of various kinds of non-state actors, diverse histories of colonialism and authoritarianism, and widely different social norms.

Most of our understandings of empowerment and accountability come from places with relatively stable and effective states. Historically, accountability gains have emerged through broad-based movements for socio-economic rights, yet lessons from these places have limited relevance in fragile contexts where fear and trauma born of experiences of repression and violence make social and political action rare. Fragile contexts are marked by weak and fragmented state institutions which lack legitimacy, whilst simultaneously other non-state actors control territory, provide public services and claim to represent the population. This makes claim-making complex, as it is unclear who is accountable for what, and on what basis. These features, combined with low levels of trust, suggest that informal networks carry greater weight. Yet, researching marginalized groups’ experiences in fragile settings is difficult: challenges around gaining permissions, accessing populations, developing trust and ensuring safety of both respondents and researchers, mean conventional research methods are either time-intensive or infeasible. While developing this research, we needed a methodology that allowed us to capture poor and marginalized households’ diversity of experiences with a medium ‘n’ that allowed us to make some mid-level generalizable observations across contexts about how governance takes place and what it means for accountability.

This article focuses on the approach we developed for meeting these needs –governance diaries. ¹ Governance diaries involve following the same set of households/individuals over a long period with regular visits, enabling in-depth research questions, following household events as they unfold. We used governance diaries in Mozambique, Myanmar, and Pakistan from 2017–2021, in two phases with an interim analysis and reflection period in between. In the first phase we focused the diaries on 164 poor and marginalized households to enquire how they interacted with and made claims on the diversity of public authorities present in their context, and whether these interactions advanced empowerment or accountability. Phase 1’s key finding was that poor and marginalized households rarely engage directly with state or non-state authorities; instead, they do nothing, self-provide, or most commonly, go through intermediaries. Households were usually unable to choose the intermediaries to approach – these were fixed for the group, or issue, or location. The discovery of the intermediaries’ importance led us to use governance diaries with them in phase 2 to understand their role and how they navigated different public authorities.

The rest of the article is devoted to the approach rather than the substantive findings. ² The next section focuses on the challenges in exploring the lives of hard-to-reach populations and concludes showing how governance diaries were able to overcome some constraints. The third section details our research design and operationalization of key concepts tailored to our country contexts. In the fourth section we elaborate on field implementation and the fifth section describes some of the tools we used. In the sixth and seventh sections we offer some reflections on the approach, as well as on the insights that the approach allowed us to surface. Finally, we conclude with some ideas on the relevance of governance diaries to other research, and to adaptive practice.

Researching Grassroots Experiences in Difficult Contexts

Gaining a deep understanding of marginalized groups’ lived experiences is a daunting task, particularly so in fragile, conflict and violence-affected contexts (Atkinson and Flint, 2001; Clark, 2006; Cohen and Arieli, 2011; Jacobsen and Landau, 2003; Khan Mohmand et al., 2017). Key among the challenges are:

• high levels of insecurity leading to an atmosphere of fear and distrust, restricting the open or easy flow of information between researchers and respondents.

• cultural, social, and economic constraints, from language barriers to never having engaged with research, to relative positions in local hierarchies.

• ideological, religious, and political affiliation related barriers, particularly when belonging to minority or powerless groups.

• technical, administrative, and legal obstacles, such as mobility limitations and state regulatory frameworks.

Overcoming these challenges to get at the questions we were interested in required extensive of fieldwork, for building trust and uncovering the real meaning of processes we heard about or observed. The challenges also make it more difficult to operationalize abstract concepts such as ‘accountability’ which may mean different things in different places, and for different groups (Fox, 2022). Moreover, comparing processes and indicators across contexts with varying levels and combinations of fragility, violence and conflict required an iterative process of data collection, analysis, reflection, and adaptation (Khan Mohmand et al., 2017).

Conventionally, there are two main approaches to such research: ethnographies and large-n surveys. Ethnographic work is better placed to explore more sensitive aspects of everyday life, as time spent embedded in local communities allows researchers to build relationships of trust with the respondents while observing as much as possible of their everyday lives. While ethnographic studies provide rich, descriptive information and insights about target populations, their small sample sizes, and purposive sampling limits generalizability (Lyberg et al., 2014). This is particularly problematic if a research aim is to compare cases across locations: the attributes of different settings, as well as differences in the nature and prevalence of hard-to-reach populations, lead to variations in problems making comparison difficult (Smith, 2014). Thus, while ethnographies are useful in concept building, they are less valuable in theory testing. By contrast, large-n surveys provide one-off snapshots that can present generalizable, comparative pictures across a larger set of cases suitable for theory testing. Large-n comparative studies help eliminate competing hypotheses and isolate the explanatory, most often by having a control group (Jacobsen & Landau, 2003). Yet, such surveys are insufficiently dynamic to trace the processes behind change: they do not allow for trust building or permit sufficient time for researchers to observe ‘the field’.

We developed the governance diary approach drawing on the advantages of both approaches outlined above. ³ It involves researchers interviewing the same households (or other actors) regularly over several months to follow specific questions/issues while tracing events as they unfold in real time. The idea is to generate a qualitative medium-n panel data set that allows for in-depth probing of certain themes in a comparative fashion, while enabling trust-building with respondents. We find that governance diaries are suitable for researching complex behavior that changes over time. They appear particularly suitable for exploring issues that people are uncomfortable talking about to outsiders. Regular follow-up breaks down respondent reticence over time, thus reducing misinterpretation. The more entries in the ‘diary,’ the more trust between researchers and respondents and a deeper emerging understanding of issues being explored. In Myanmar for instance, discussion themes changed from less sensitive health-related issues in the first couple of visits, to more sensitive conflict-related stories in subsequent ones. In Pakistan, a few households that mentioned health expenses in the first visits disclosed later that these expenses were actually money spent on drugs by some household members. Besides building trust, capturing people’s experiences over time provided us with new insights into how state and non-state institutions shift in relevance, legitimacy, and trust; and why these perceptions change over time.

Designing Research and Operationalizing Concepts

We started by defining and operationalizing key concepts and identifying our units of analysis. Here we highlight our key choices, our rationale, and how these changed along the way.

Empowerment and Accountability

We used Eyben, et al.’s (2008: 6) work as a starting point for conceptualizing empowerment, ‘when individuals and organized groups are able to imagine their world differently and to realize that vision by changing the relations of power that have been keeping them in poverty’. We saw empowerment as both an action – the act of gaining control over decisions and resources that affects one’s life – as well as a state of being (objective and subjective), where people have a greater voice over decision making, allowing them to expand their choices (as well as the possibility of making those choices), and eventually giving them increased control over their own lives. For accountability we used Schedler’s (1999) conceptualization as a broad two-way relationship between two parties, that incorporates both answerability (obligation to inform and justify actions) and enforcement (rendering judgements with attached sanctions or rewards). These initial definitions were refined during the research as it became clear that accountability in the strict sense of answerability and enforcement did not quite capture the processes we observed; rather attention to sources of expectations and obligations on the one hand, and processes of scrutiny and judgements on the other were of greater importance (Anderson forthcoming).

Public Authority

We defined public authorities as formal and informal institutions which ‘can undertake core governance functions: protection from external threats and managing external relations; peaceful resolution of internal conflicts; and providing or facilitating the provision of a range of collective goods and services’ (Unsworth, 2010: 9). Using this definition allowed us to focus on functions rather than on form, while being more neutral about the processes and actors involved – inside or outside the formal state This broad definition had to be redefined when faced with examples from the field. How were we to assess the healthcare claims that a woman heading a household made to a relative who seemed to have authority over her actions, as he was both a respected member of the community and a relative? After much discussion we agreed to exclude familial sources of authority unless they were accepted as responsible for public goods more broadly and were coincidentally related to households under study. Operationally we defined public authority as people, organizations, or institutions who households considered responsible for the provision of particular public goods or services delivered to a wide range of people within the community – e.g., traditional local governance institutions, armed groups, local state institutions, religious leaders – who had the legitimacy and capacity to carry out their functions. We did not hold a priori assumptions about the role that any of these played, as they played different functions across the three countries in our diaries.

Public Goods

Poor and marginalized households engage with public authorities on a range of governance issues. To choose the issues to follow, we began by focusing on core state functions that people might expect authorities to provide. Our choice of functions was steered by Stewart and Brown’s (2009) operational concept of state fragility involving three dimensions, namely: authority failures; service failures; and legitimacy failures. We chose a core function from each dimension: security; health provision; and revenue collection. ⁴ Over the course of the research, we reframed security into security, justice, and conflict resolution, as we could see an interconnection in the households’ stories of accessing security and justice and the existing conflict resolution mechanisms availed by them. We also added themes which emerged as important, specifically social protection, employment, and poverty, accessing resources and (other) services, and legal documentation. ⁵

Household Selection

Our unit of analysis in phase 1 was the household, as governance issues are usually experienced by and responded to by households, rather than individuals. ⁶ We collectively defined household as a group of people living together and sharing meals. We selected chronically poor households using Collins et al. (2009: 190, 195)’s definition:

households that display evidence of deprivation of basic human needs that had existed over a long period of time (many months and often years) … such households are poor or “at the bottom end”.

In this definition, we tested for local interpretations of households considered marginalized within each research community, either because of ascriptive identities, religion, or household characteristics. To account for intersectionality and contextual specificities (Crenshaw, 1989; Lyberg et al., 2014) our selection process took three stages. We started with a typology based on our previous knowledge of poverty and marginalization in each country: household composition; social status; assets and occupation; and type of dwelling. We then discussed and validated this categorization with country field researchers and refined the list at country level. For instance, while ethnicity was a key factor in Mozambique and Myanmar, it did not play a role in household marginalization in any of the sites in Pakistan. Instead, being a religious minority did, as did belonging to lower-ranked tribes. Not owning land was an important marker in most of Pakistan while not so much in Mozambique or Myanmar, as in these countries’ land arrangements revolve around user rights rather than ownership. Finally, we validated the categorization with local communities and purposely selected 10 to 15 households in each location during our first two visits. In this selection we aimed to have a significant number of female-headed households, a particularly marginalized population group in any country. Anticipating that some households were likely to drop out, we added a few additional households in each location during the second and third visits through snowball sampling, ⁷ ended up with a total of 164 households across the three countries of which 40% (in Pakistan) to 54% (in Myanmar) were female-headed. ⁸

Intermediary Selection

In phase 2, we followed individuals or organizations acting as governance intermediaries as identified by our respondents from phase 1. ⁹ Intermediaries were the first link in the governance chain that households reached out to for addressing an issue outside of the family. Thus, our set of intermediaries were identified by marginalized households rather than those that might have been identified by from the top down. In other words, these were the real brokers in mediating between the poor and public authorities. Overwhelmingly male, the intermediaries varied widely across the three countries and included village elders and respected individuals, political party workers, religious leaders, and members of social movements. As in phase 1, anticipating that intermediaries would drop out, we started with about 20–30 intermediaries roughly divided across four locations in each of the three countries. In the end we followed 81 intermediaries across the three countries.

We started interviewing intermediaries in person; however, Covid-19 forced the countries into lockdown starting in late March and April 2020. We consequently switched physical visits to phone and voice over IP services. This was problematic not only because intermediaries were often busy with Covid-19 issues and did not have the time or patience to speak with us, but they were also uncomfortable with the perceived insecurity of phone/online conversations (IDS, 2020). Yet the pandemic gave us an opportunity to witness how intermediaries across different locations dealt with the same crisis: how they interacted with citizens and public authorities, their roles, practices and strategies, and how they made themselves essential to the functioning of local governance systems.

Implementing Governance Diaries

Selecting Researchers to Build Trust: Gaining access and Gatekeepers

Access and trust are key challenges of doing research with the most marginalized. Distrust is heightened by fragility, violence, and conflict –speaking to the wrong people can lead to violence from powerholders, whether local armed groups or powerful political interests. We took the strategic decision of choosing field researchers ¹⁰ who were closely linked to the locations, to reduce distrust and facilitate access, as well as to ensure that the interviews could be conducted in local languages. This was achieved differently in each country. In one country, researchers were graduate students at a national university who returned home for the interviews. In another, we recruited junior faculty from regional universities who relocated to the research sites for substantial periods. In the third country, where ongoing ethnic conflict was an issue, we trained local civil society organization staff from the same ethnic background and language of each location.

Having field researchers with existing connections to the locations meant our teams could gain access more easily both through local gatekeepers but also directly with households and intermediaries – in several cases the field team acted both as field researchers and gatekeepers. The researchers who were local worked through several local gatekeepers with whom they had previous connections with or formed connections with during the first research stage. The researchers who relocated for the study duration took time to establish networks and relationships to support the work before beginning interviews. Finally, the country team working through local civil society organizations chose organizations well-known locally for service provision. Building and strengthening local networks and relationships improved access and increased our chances that gatekeepers would trust us. In turn, this trust between local researchers and gatekeepers increased respondent willingness to take part in both phases: most households and intermediaries opened-up as they came to recognize the field researchers as people who understood their lived experiences.

Working with and through local field researchers was also key to avoiding public authorities considering our activities suspicious. Research in fragile, conflict and violence-affected settings (FCVAS) can pose potential risks not only to respondents, but also to researchers (Campbell, 2017). It is crucial to be mindful of cultural, social, and political sensitivities and power relations, being aware of whom to ask what, as well as how to ask it (Goodhand, 2000). Conducting this research through a partnership of research institutes and an international NGO, and with local field researchers allowed us to have in place some of the tools and approaches aid agencies have developed for staff security (Mazurana and Gale, 2013), as well as the familiarity and experience of the settings to which Goodhand (2000) refers. ¹¹

Building Conversational Communities Among Researchers

From the outset, the research plan involved periods of reflection and analysis between field visits. The period between the monthly field visits was used to transcribe and clean interviews, sense check what was being heard, identify issues for further probing, discuss new tools and start the analysis process. Every three months, there was a more intense period of reflection. Finally, there was a gap between the two phases for analysis and rethinking of strategy. The reflections took place through ‘conversational communities’ (Gudeman and Rivera, 1990). Throughout our research, we created three layers of conversational communities: a local one, where the field researchers were paired to share descriptive and reflective notes as a unit; a national one, where country research teams would reflect on their monthly observations with the principal investigators and each other, and plan subsequent visits; and the international one, where principal investigators shared their observations and challenges and tried to keep a comparative element to the whole process.

As well as using conversations to improve data quality, a(nother) key activity of our conversational communities centered on unpacking the words and expressions respondents used to tell their stories, and the meanings behind them. To ensure we were questioning and measuring the same concepts across locations, we used a similar iterative procedure to that we had used for household selection. Keeping true to the intended meaning of our concepts entailed seeking agreement within the different possible interpretations by validating translations across a series of in-country actors, namely the households and communities, the intermediaries, the research teams, academics, think-tanks and research institutes, activists, and development actors.

A Menu of Field Tools

Conducting interviews in these settings, even by local researchers, poses several challenges. A general atmosphere of suspicion in FCVAS influences the interviewing process, shaping not only the kind of questions we can ask, but also the strategies we must adopt in asking these questions (Goldstein, 2014). Interviewing in FCVAS requires patience, subtlety, and above all flexibility (Kovats-Bernat, 2002). The multitude of unpredictable parameters which could restrain the interviewing process (particularly during an active conflict) forced us to adopt a reflective/adaptive approach to fieldwork. In these situations, rigid questionnaires can appear threatening and inappropriate (Barakat and Ellis, 1997), so it is best to let respondents lead the discussion, as the more comfortable they feel, the more likely they will be put at ease resulting in more detail (Goldstein, 2014). We used conversations rather than formal interviews. The aim was not to simply interview different households and observe their everyday activities, but to engage in long, relaxed conversations. Conversations were particularly valuable to help build trust and informality between the field researchers and households; it allowed researchers and respondents to be at ease.

As part of the method, we used several tools from an array of methods suitable for research on social and political action in FCVAS (Khan Mohmand et al., 2017). The aim was to create a menu which field teams could use iteratively (Chambers, 1994) and allowed us to probe deeper into respondents’ answers, and them to reflect on their answers, while simultaneously reducing respondent fatigue (Höglund, 2011). Among these tools three were key for our research: ethnographic tools, institutional mapping, and anchoring vignettes.

Ethnographic Tools

We largely used three ethnographic tools: participant observation; in-depth interviews with individuals and focus groups (including life histories); and thick description. Participant observation allowed for a holistic awareness of events as they unfolded, giving a more comprehensive understanding of what really matters to respondents. Central to this reliance on observation was the need to maintain an ‘anthropological eye’, that is, a sensibility to local culture. Being both insiders and outsiders enabled researchers to ask questions and make observations that local inhabitants would not usually think of. Positionality though, can become an issue when researchers are also gatekeepers in their own research. As most of our field research teams belonged to the study locations, there was a risk they would take local everyday life for granted and not record observations. Therefore, a significant element of the training focused on getting the field teams to become detached observers. We asked them to pretend they had landed from Mars; everything is new so worth recording and questioning. Having an anthropological eye was not only useful for observing how people reacted to special events, but also to ordinary ones. For instance, in the third visit to one of the households, during a moment of silence when the household had already hinted there was nothing new to report, one of the researchers noticed a cow outside the house. When asked if the cow was theirs, the household head replied it was not, but that she oversaw it. Probing deeper, we uncovered a local informal social protection arrangement through which poorer households rear non-poor households’ livestock in exchange for a cut when the cattle is sold.

The iterative nature of the diaries meant that incomplete or unexpected observations shared and discussed amongst us in a previous visit, became follow-up questions to probe in the next. These follow-up questions, along with issues discussed across the three countries became semi-structured thematic mental prompt lists for the research teams. Having a prompt list in their minds allowed the field researchers to bring respondents back to the main issue when they moved away from the theme. ¹² Quite a few of these unstructured and semi-structured interviews developed into focus groups, as other respondents within the households joined the conversation with their opinions and stories. Life histories are useful tools for capturing processes of change. Life history interviews and life event analysis allowed for individuals to discuss not only themselves and their lives, but also the social, economic, and political spaces that they inhabit. We adapted life histories to our needs, and in particular life event analysis, by focusing not on the life of the respondent – be it the household head, the entire household, or the intermediary – but on the life of the events themselves.

The third ethnographic tool we used was thick description (Geertz, 1973), where we not only explain the behavior of individuals and groups within a society, but also its context, as understanding the local context is crucial to fill in certain blanks. ¹³ Thick description was composed not only of facts but also of commentary and interpretation: what we find out as researchers is inherently connected with how we find it out (Emerson et al., 2011). Thus, it was critical that the field teams documented their own activities, circumstances, and emotional responses to fieldwork, as these shaped the process of observing and recording others’ lives (Emerson et al., 2011). Thick description also allowed us to engage with narrative analysis as an analytical method, to interpret social meaning in respondents’ everyday lives and interactions with different public authorities by analyzing their stories. It allowed us a richer thematic approach when analyzing the data.

Institutional Mapping

As the research unfolded, we felt we needed to bring in other tools not only to probe deeper into the nature of the relationships between households, intermediaries, and public authorities, but also to counteract survey fatigue. We decided to introduce institutional mapping as a participatory tool that could deal with these two issues. ¹⁴ Institutional mapping allowed us to see the perceived importance, accessibility, and impact of different public authorities to different households, as well as the key institutions existing in their communities, and how they related both to each other and to external agencies involved in service delivery and the administration of programs.

In each location we selected which households’ stories were most noteworthy and could help us gain a deeper knowledge of people’s perceptions of their local institutions – including different kinds of formal and informal public authorities – and their access (or lack of) to them. The main objective was to see which actors helped households solve problems. The field teams started by helping households identify key actors and their relative power in helping them solve problems (the more powerful the actor, the bigger the circle representing it), while simultaneously noting the basis that households used for determining power. Then the field teams asked how accessible these actors were to the households, using distance from the center – household – as a measure of accessibility. Part of the exercise was to probe the map, to ask how and why households chose to identify those actors, the criteria they used to determine power, and the level of relationship between different actors.

Anchoring Vignettes and Hypothetical Scenarios

Studies based on small samples and in-depth interviews rarely allow for conducting longitudinal or geographic comparisons unless very carefully selected (Jacobsen and Landau, 2003). We used anchoring vignettes with households in phase 1 to measure empowering and disempowering situations when holding public authorities accountable, and with intermediaries in phase 2 to ascertain levels of answerability. ¹⁵ These, we felt, would allow for cross-location analysis. Measuring empowerment and disempowerment is difficult not only technically, but also conceptually (Masset, 2015). As such, we developed a set of hypothetical scenarios against which we could calibrate respondents’ self-assessments (King et al., 2004). Each vignette described a situation related to an empowering/disempowering moment of a fictitious character. We used an amalgamation of real stories to construct them locally, making them plausible within each context. We chose a health-related event as this was the theme brought out by most households. After each vignette the field teams probed the rationale for the respondents’ answers. Following on from this reflection, the field teams used the exercise to inquire about respondents’ claim-making abilities with existing public authorities. Similarly, we prepared anchoring vignettes to measure perceptions of accountability components (information, justification, and enforcement) among intermediaries with two scenarios: one mediating a land dispute resolution for rural locations, the other mediating citizen access to public security for urban locations.

In Myanmar, due to scheduling issues of the project, ¹⁶ we did not manage to use anchoring vignettes; however, we used similar vignettes within 18 focus group discussions to unpack the role of trust. These hypothetical scenarios – based on real stories that had emerged from the governance diaries – allowed us to create a conversation around how people would act if the hypothetical situation were to happen in their location. This enabled a slightly broader triangulation of what respondents had said in the household interviews, while also allowing for a reality check of the proposed actions of others from within their community to try to counter potential response bias. These hypothetical scenario vignettes allowed us to ask questions about trust in various governance actors indirectly and understand people’s attitudes and behavior in situations in which authorities were potentially involved. Asking directly whether participants trust certain authorities would have been too sensitive, making participants uncomfortable about speaking openly; this way, the groups were able to have these conversations whilst allowing them to save face and not be speaking about specific individuals.

Reflections on the Research Process

The research process encountered several challenges, from the operationalization of key concepts across different contexts and languages, to respondent fatigue, to the amount of time required for training and accompaniment of the field teams due to the iterative nature of the approach and the need to carry out analysis as one went along. There were twelve languages used across the project, each having words for key concepts that either had additional implications, or several words for the concept with slightly different nuances. In Myanmar, for example, the term ‘authority’ often relates to ‘legitimacy’, while in Mozambique people use the term to refer to the police. In Myanmar, the literal word ‘trust’ in local languages has more spiritual meanings, and so we needed to find other ways to get at the concept without using the word. The definition for who could be considered public authority was also tricky – we were asked multiple times why God was not on our list of governance actors.

The iterative nature of the governance diaries allowed us to come back from the field, reflect on the stories, and think of other ways of enquiry: sometimes because we had reached a methodological impasse and wanted to probe deeper; other times to look at issues from a different angle or epistemological framework; or to reformulate an abstract query into something more concrete to allow respondents to reflect on their actions. Yet this iteration also posed challenges. Every three months the country teams would bring together field researchers and principal investigators to reflect on and undertake a preliminary analysis of trends that were coming out. However, due to skill sets, methodological issues, and confusion over roles, these were often spent on troubleshooting methodologies rather than analysis: whatever time we planned for training, it was never enough. We chose local field teams based on the need for local expertise, but as the visits were monthly, we did not employ them full-time– and naturally, they had other commitments. The local researchers were keen on the research and really wanted to learn; however, it was a gradual process to develop their confidence and create ownership of the research.

At the end of phase 1, there were three key process-related lessons: first, the need to collectively develop and agree definitions, concepts and meanings, units of study, and targets right from the beginning. Second, there will always be a need for more training and more accompaniment than planned, due to the unpredictability of FCVAS and the fact that often we are doing remote research dependent on gatekeepers who may not have the same level of understanding of the research process. Third, while it is important to have an initial plan/design, it is imperative to keep the whole process flexible both in terms of tools and methodologies used and critical reflection in and between countries – be ready to embrace change.

Our phase 2 research with intermediaries posed even greater challenges. To start, intermediaries were more assertive than households, largely because their role demanded it of them. This meant they had less patience with repetitive questions as well as being more demanding of some form of compensation for their time. After the initial few visits Covid-19 struck, which meant we could no longer interview in person. As mentioned earlier, this complicated fieldwork as intermediaries became even busier solving local problems and did not want to talk on the phone or online for fear of surveillance. As a result, we lost several intermediaries and had to come up with alternative ways to keep in contact during Covid-19 restrictions. ¹⁷

A concern we had from the beginning was that the research should not be extractive. The households have had long experience of subjugation by different configurations of the state (repressive colonial, repressive post-colonial, authoritarian, neoliberal), in which local formal and informal public authorities played a key role, and we wanted the research to be a positive and reflective process for them. All participating households and intermediaries knew they would not receive payment for participating in the research, that there were no benefits, and that they could withdraw at any point. ¹⁸ And yet, most households – particularly those in rural areas – kept talking to us, as did quite a few intermediaries even during the pandemic. Three key factors played a role: there was someone there to listen to them – both as catharsis as well as the feeling that someone cared about them; we could potentially help with their lives; ¹⁹ and the different tools employed avoided monotonous repetitiveness and provided a welcome distraction in their everyday routines.

Trust started developing between respondents and field teams from the third visit onwards. The idea of communicating with people who came from provincial capitals, places not easily accessible for those living in the most remote districts, served as a stimulus for the respondents to keep speaking with us. The fact that these highly educated university and NGO researchers came to their homes to hear about their lives made many feel important. ²⁰ Just the fact that we made recurring visits showed that we were worried about their situation, they said; even those who were initially reluctant to provide details became more receptive after the third visit. Some would even initiate the conversations themselves, as they already knew the nature of the research. One of the Pakistan researcher’s notes from a third visit reflects this:

We have started feeling that people are becoming more comfortable in talking to us as time passes. They discuss their issues more openly now. We felt this in the case of Mr. X, as he talked about his father’s murder in great length. He told us all the events that happened before, during and after the murder. He even shared the responses of the womenfolk in his family after the murder. Men in this region usually don’t share the stories of their women and the events that happen within the family, except to the closest of their friends. So, in a way, we are becoming his friends. By the third visit, we felt he was waiting for our arrival and was desperate to share the story. Governance diaries, the interviews, have become a source of catharsis for him.

The regular visits took on a cathartic tone, including for women. As one field researcher noted:

In the first two visits, [women] kept their answers very short; but now they have started opening up about their issues. One of the major problems with interviewing women in this region is that generally they don’t speak to males except those within the family. Women here are often kept aside from public social and political life: since their childhood they are taught to respect and obey their male members, starting with their brothers and fathers. At every point in their lives, they are told to recognize their inferior status in local society as compared to their male counterparts.

In phase 2 we continued fieldwork in 12 locations out of the original 20 from phase 1. Households in these 12 locations played a key role helping us identify intermediaries at the start of the second phase and many kept interacting informally with field researchers when they visited these locations. While the interaction between field researchers and most households from the other eight locations stopped once phase 1 ended, there were still a few that infrequently would contact field researchers for assistance or to mark festive occasions such as Eid or Christmas. If not friends, they at least became good acquaintances and someone to network with.

Reflections on Findings

We have outlined above some of the opportunities and challenges presented by governance diaries. Yet, what evidence do we have that the approach generates useful insights that would not be generated otherwise? While it is beyond the scope of this article to elaborate on the results of the research, here we offer a brief glimpse into the insights and their implications. ²¹ Traditional understandings of change from below suggest that marginalized groups: (a) express voice on a given issue or grievance; (b) mobilize (usually publicly) to hold appropriate authorities to account; and then (c) authorities will respond in a positive or negative fashion. Our research using the governance diaries in FCVAS challenged these assumptions on several counts and showed the nuances and ambiguities of how the process of claim-making works for marginalized groups.

First, in FCVAS, histories of violence and fear mean that it is hard to change internalized norms and deep emotions of powerlessness that affect the possibility of voice. The overwhelming majority of the households felt disempowered and had low expectations of public authorities. There were, however, interesting ways of coping with this marginalization – of rationalizing their situation and relative impotence. In all three countries, when people had no access to healthcare they chose deliberate ignorance, choosing not to find out about their illnesses as they could not afford treatment. In one location in Mozambique, when crimes were committed and criminals were known, people would resign themselves by saying, ‘[The criminals] are being resourceful; they are also getting by’. Fate was often invoked as an explanation, and many turned to God or prayer as a positive action to solve problems.

Secondly, unsurprisingly, we found rapidly closing civil society spaces and a history of authoritarian regimes making collective action rare and risky. In these settings it is difficult for marginalized people to collectivize or even find a common identity. In our 164 households across the three countries, we only found two clear instances of collective action to either substitute public authorities or to hold them accountable. Understanding these instances of positive outliers seems to offer a promising start to unlocking constraints to accountability action from below.

Thirdly, and what triggered phase 2 of our research, marginalized people rarely reach state or non-state authorities directly but work through intermediaries who mediate their claims. Intermediaries navigate these diverse sources of authority, working across formal and informal local governance systems, and being the deciders themselves at times as they too exercise significant authority. There are strong pressures within the local governance system involving intermediaries to resolve things locally. Higher authorities decline to take on problems till serious efforts at local levels have failed. These authorities impose hierarchies and punish ‘skipping levels’. As one intermediary from Myanmar noted, ‘I always try the best to make the big crime small, and the small crime disappear’. Self-provision is common, with low expectations of higher-level authorities. This does not mean that people see self-provision as the best solution; it is often the least bad.

Overall, we see local governance systems as a web of networks that take different shapes depending on the location. Standard structures rarely apply, even within one region. Across all three countries we see a diversity of actors and institutions key for local-level decision making and governance needs. These public authorities are neither inherently good nor bad: just because poor and marginalized people identify them as legitimate public authorities does not mean they deliver, are just, or accountable.

Conclusions

The governance diaries approach uses a basket of methods and sits between a medium ‘n’ survey which gathers information in a single snapshot, and a detailed ethnography taking place in several locations simultaneously over an extended period. This hybridity offers several advantages. Repeated interactions with the same respondents combined with adapting to use different methods for different needs – for both data collection and analysis – allows for a relatively open-ended agenda in FCVAS, and to slowly ‘tighten the net’ by focusing on particular stories. Thus, the approach probes deeper into households’ access to different public authorities while limiting respondent fatigue. In the process, households can also visualize and reflect on their position vis-à-vis different public authorities. Similarly, understanding how intermediaries deal with issues presented to them enables a triangulation of the household findings, as well as looking upwards to see how the intermediaries fit within broader governance networks.

Analysis of the diaries gave us pause to think about how people at the margins imagine the state and public authority. In all our focus countries people thought of state and public authority in ways different from our conceptions, variously invoking the state as an absent father who is remiss in providing as in Mozambique, or as an arm of repression for people who want to be left alone. Alternative forms of public authority were perceived to have different degrees of legitimacy and credibility in terms of delivering on public goods.

While we used this approach to understand governance issues in FCVAS, it can be used for other questions in different settings where researchers are interested in a ground-up view of how particular services, institutions or discourses are experienced and perceived. The approach is particularly valuable in places where: (a) there is likely to be limited trust between populations of interest and outsiders, particularly around research; (b) where the environment is rapidly changing requiring shifting responses from people; or (c) where the subject matter is particularly sensitive and one-shot approaches are unlikely to generate accurate data. One could imagine using the approach with other hard-to-reach populations, such as internally displaced people (IDPs) and refugees in camps, those living with stigmatized illnesses, unemployed youth in industrialized countries, or even with frontline workers or mid-level administrators in hierarchical bureaucracies.

In addition to these, we can imagine the governance diaries approach used as an independent, real-time, accompaniment for development programs, to sense test whether theories of change and assumptions work, or whether they need adaptation. Such a strategy can strengthen monitoring, evaluation and learning tools. Given the current emphasis on adaptive approaches, including ‘thinking and working politically’; ‘doing development differently’; and ‘program driven iterative adaptation’; governance diaries can offer an additional source of evidence on what is working, what is not, and what the reasons might be for the observed interim results. We are already seeing governance diary adaptations to explore the extreme poor’s urban livelihood strategies in Bangladesh pre-Covid 19 (Devereux and Shahan, 2020) and the new poor’s during and post-Covid 19 (Durdiner diaries, under the CLEAR project), the intersectionality of gender and health in urban informal settlements in Sierra Leone (Conteh et al., 2021), household financial governance and coping strategies in D.R.C. (Stys et al., 2021), women’s struggles against backlash in South Asia (struggle diaries), peoples’ struggles during the military coup in Myanmar (emergency diaries), and the impact of Covid-19 on young people’s lives in Nepal and Indonesia (livelihood diaries).

The urgent need to understand how public policies and programmes affect the most marginalized make the search for appropriate tools and methods that can offer insights a priority for development practitioners and researchers. Governance diaries offers a powerful practical approach to meet this need – it is relatively cost-effective, offers real-time qualitative data, and can be used in comparative research to produce mid-range generalizable insights that have wide implications.