The dangers of writing up: a cautionary tale from Bangladesh

The dangers of writing up: a cautionary tale from Bangladesh

By Harriet Matsaert, Zahir Ahmed, Faruqe Hussain and Noushin Islam

In this paper, we reflect on our experiences of writing up for an 'action research' project in Bangladesh. In this work we wanted both to produce high quality anthropological analysis, and also to meet development aims, which depended on effective partnerships with those we were studying. In writing up we found that these aims can be conflicting. Our experience highlights the need for applied anthropologists to be acutely aware of their role and status and to be prepared for the kind of reaction their text may provoke. Some practical lessons include a recommendation to keep the text a 'work in progress' for as long as possible, using this time to share and revise it with others and to build consensus before publication. We also found it useful to move from a critical to an appreciative stance in our analysis. We believe that this change of focus does not lessen the rigour of analysis but presents information in a more constructive way which is conducive to partnerships and action.


Picture this scene. You are an anthropologist working with a small non-governmental organization (NGO) in rural Bangladesh. You have been hired to look at productive social networks on isolated river islands, with the aim of finding people, processes or events that the NGO can support to increase benefits from agricultural production. You are excited by this opportunity. You hope your methods, drawn from anthropological fieldwork, can give your team members an insight into the importance of the social relationships underlying agricultural production. You hope to encourage them to look at indigenous systems and build on existing processes, rather than reinventing the wheel, as so many well-meaning outsiders do.

The Agricultural Extension Department is, potentially, a key player in providing services to the islands. You and your team are meeting with them now to see how they can be part of a plan to build better links with the river islands. Your research has found that at present their staff have no contact with the islands. As your meeting begins, a copy of your research findings is passed around the table. A very user-friendly visual network diagram illustrates graphically the absence of agricultural extension staff on the island. As the paper reaches the hands of the local head of office, he flicks through it and stops as the diagram catches his eye. His face clouds over. He is furious. The Agricultural Extension Department pride themselves on being the government ministry that 'reach every corner' of Bangladesh. Your findings contradict this.

While both you and the agricultural extension officer know that his staff do not reach the islands, it is totally unacceptable to him to have this fact written down in print and in this way made 'official'. Of course if you were a powerful World Bank team presenting this information, the situation would be different. But you are not. You are from a small and powerless Bangladeshi NGO. The meeting, which has been cordial up to this point, deteriorates as the head officer begins to undermine the validity of your work. Your choice of research area is not representative, your staff are unreliable, this information is simply not true. 'You are wasting our time!' At one point you think that you are actually going to be thrown out of the office altogether. Only the sweet-talking of your more politically astute colleague - 'of course this is only one incident, at one particular time, possibly not representative', etc. - manages to restore relations.
Writing up for applied work is fraught with dangers!

Firstly, as applied anthropologists we are usually in a situation - unusual to traditional anthropologists (though surely becoming more commonplace) - where those we are writing about will read, have comments on, and may well disagree with what we have written. Secondly, our work by its very nature has practical aims. It is designed to change behaviour or guide policy. It is therefore political and this, as our experience shows, can be very problematic. Of course, despite the 'dangers' of our new audience, we usually do want our text to be read. This in itself raises a third issue. How can we make our writing accessible and useful? Are there ways in which we can present our findings so that they are concise and readable? Can we make them useable by non-literate clients? And how about groups? 

Finally, how do we balance these demands of writing up with our desire to produce high-quality rigorous anthropological research? Mosse (2001:179) sums up this dilemma: The more that the study of project processes is independent, critical and unrestricted by organization concerns (the more ethnographic it manages to be?), the greater will be its loss of legitimacy and practicality, while the more the analysis is instrumentally focused and tied to project concerns the less interpretative power it will have. Finally this tension finds a parallel in the distinction between participatory and critical research perspectives. The more 'participatory' the analysis of communities and institutions is, the less it is likely to reveal anything about the social dynamics of the participants themselves.

These are some of the questions we have pondered during our work in Bangladesh. In this paper we would like to share with you some experiences, some lessons and some insights we have had along the way.

The project

The work we are going to tell you about took place in Jamalpur district of Bangladesh. This is one of the poorest and least developed districts in the country. Jamalpur is bisected by large rivers, which flood for several months of the year. With each year's flood areas of riverbank are washed away, displacing whole villages. The floods, which bring down silt from the Himalayas, also cause the formation and destruction of river islands, the chars, where many of Jamalpur's citizens live.

Islands come and go but char communities stick together, sending satellite households to new islands as the old chars begin to erode away. These communities are primarily Muslim and conservative. Women hardly leave their homesteads and are not permitted to attend public meetings. Religious leaders play an influential role. Life is tough as many chars are submerged for several months of the year. At this time families hoist their wooden beds up onto the house rafters and sleep suspended over the water. Cattle and goats are tethered on rafts made from banana stems. Young boys dive under the water to cut grass for the animals from the flood plain. Despite what seem to outsiders to be incredible hardships, many char dwellers claim that they prefer this life to being on the mainland, which they describe as too crowded. Chars do have space (a premium in Bangladesh), and when the floods recede, new opportunities emerge. The deposited silt can make land very fertile. At this time, families can make a good income from growing crops and particularly horticultural produce. The chars are famous for their high-grade chilli, and for the bulls reared on the new grass and fattened for sacrifice at the national Eid ul-Azha festival.

These are the chars: unstable and unpredictable, but also a place of opportunities. It is in this context that the small NGO that we mentioned in the introduction works. Its name is Development Wheel, and it is an organization that works with char dwellers to provide emergency relief and small loans, and to help develop income-generating opportunities. In 2002-2004, with funding from the UK Department for International Development (DFID), Development Wheel initiated a research project that looked at the key social networks around agricultural production on the chars. These networks (i.e. how people get access to information, inputs, new technology and market contacts), underlie the agricultural 'innovation system' (i.e. how people adapt to challenges and opportunities). Because of their remote location, strengthening innovation systems on the chars presented a particular challenge. But their very isolation made our network analysis more interesting. In the absence of mainstream services, we hoped to identify alternative, informal, perhaps traditional channels through which innovation was occurring, and which Development Wheel could work with.

The research was conducted by a group of anthropologists (the authors), working together with Development Wheel staff, a business development advisor, and local government agricultural staff. One of the anthropologists, Dr Zahir Ahmed, is an academic based at Jahangirnagar University in Dhaka, Bangladesh, where he is now head of faculty. Harriet Matsaert, in contrast, hasn't been near an anthropology department for years. She is an applied anthropologist with ten years' experience of working with natural resource scientists. Noushin Islam and Faruque Hussain, who joined as research assistants, have recently graduated with MAs in Social Anthropology. Enthusiastic and hard working, they were keen to prove themselves in the field. While our anthropologists loved the research and theoretical discussion, our business development advisor, Munzure Aziz, an agricultural economist, is a practical man who had his eye firmly on the final outcomes of the research. Shah Abdus Salam, Director of Development Wheel, is a 'people manager': politically astute, he managed and balanced the needs of the team, and negotiated space for our work with the local community.

The team's aim was to adapt some anthropological methods into user-friendly tools to visually represent and analyse social networks. These tools could be used to collect information, but also to analyse and 'write up' - in the sense of putting information into a form that could be shared with non-anthropologists and non-literate people, and that could be used for group discussion. In short, the anthropological methods were adapted for 'applied' use. Since they focused on key people or 'actors' in the social networks, we call these 'actor-oriented tools'. Box 1 gives a brief description. (The tools are described in detail in Matsaert et al 2005.)

Our aim was to work not only on the chars but also to follow networks through to the regional and national level. Our intention was to involve key actors, as we identified them, in the research and in the process of building stronger and more effective networks. It was to be an action research process.
Box 1. Some actor-oriented tools Actor linkage map: Here key actors are listed on a piece of paper, and links are drawn between them with arrows. This simple technique is a useful starting point for discussing relationships and flows of information in an innovation system. We often used it for looking at a single actor's linkages with other key actors in the system. Actor linkage matrix: This matrix, which can easily be set up on a spreadsheet, can be used to summarise and store information on linkages in a system. It is useful for combining individual actor linkage maps, which tend to get too complex if you stick to the maps. On the matrix, cross-cutting linkages, gaps and linkage trends become evident. By highlighting particular cells or linkages in the matrix, you can prioritise areas for intervention and monitor your impact. Determinants diagram: This is a 'thinking tool' similar to the 'problem tree' used in Participatory Rapid Appraisal (PRA) methods, where the different causes leading up to a problem are discussed and drawn up in a diagram. It can be used as a group discussion tool to analyse the nature of a particular linkage, including its strengths, weaknesses and possible interventions.

The actors

Ministry of Agriculture staff, both those involved in research and in extension, at the key interface with farmers, were potentially central players in agricultural innovation. We involved them from the start by inviting them to join our preliminary survey. On the chars themselves we found communication between neighbours and relatives, and from char to char through kinship and transport networks, the most common source of innovation. Local business people played an important role in bringing information and technology from the outside world. Thus much of the practical information was based on the dealer's own knowledge and product guidelines, but also on advertisements and new promotions. Local healers were the main health providers on the char, bringing a mixture of traditional treatment (of debated efficacy) and new ideas. These again were taken from private sector interactions with pharmacies rather than with the government, because government staff would not recognize or associate with healers. Women, discouraged from leaving their homes (and usually living away from their maternal homes), had minimal networks, though they were the acknowledged experts on livestock health in the household. We soon realized the critical importance of local NGOs who lend money to women. They not only provide vital cash loans, but also, importantly, lend legitimacy to women who go out and meet together as a group, or with outsiders. At the regional and national level, private sector actors play an important role, with potential for further developments. There are opportunities here for contract buying of produce, private extension, credit and much more.

In this context, how did we do our work, and how did we write up?

Imagine a cool, misty morning in December. It's the start of our research period. Our team - four anthropologists, Development Wheel team members, business development specialist, and agricultural researcher - well wrapped and carrying marker pens, flip charts, pieces of coloured card and 'post its', set off to Char Kulpal. We travel from Development Wheel's riverside office in rickshaw vans: three wheelers pedaled by cycle power, many owned and pedaled by char dwellers from the place we will be visiting. The rickshaws take us to the small dock where we board the ferryboat across to the char. It is a long, shallow, wooden boat, which fits about 20 people, a couple of goats and a few sacks of fertiliser. A young man paddles it. Most people stand and the boat wobbles precariously as it moves across the water. Most of the passengers are men; a couple of young girls are returning from selling onions in the market. The water is low at this time of year and the crossing takes us only a few minutes. During the monsoon, the journey can be hazardous as the waters rise and large waves can cover the boat. But now it is winter and the chars are magnificent. There are no roads and no electricity, so it is completely peaceful. The only sounds are bird calls and the children playing. Along the water's edge are great stretches of lush green grassland on which cattle and goats are grazing. The homesteads, modest structures made of corrugated iron that can quickly be taken down when people need to move, are surrounded by small fields of chilli, onion, aubergine, and the yellow mustard sarisha from which the sub-district Sarishabari is named. Now that we are on the char we begin to meet women. They are dressed in brightly-coloured saris: reds, yellows and greens, gold bangles and nose rings. These women care for their many children and manage the homestead and livestock while husbands migrate to other areas for seasonal work. In terms of innovation (i.e. adapting to challenges and opportunities), their lack of mobility gives them enormous logistical problems, as it is not socially acceptable for them to visit the market or consult with service providers (who are all male).

In Kulpal, we gather in a homestead courtyard and meet with a group of Development Wheel's members, all women, in a shady spot under a pumpkin vine. A second group, made up of men, meet with some other team members in an adjoining courtyard. Sitting cross-legged around our flip chart we begin to draw an 'actor linkage map' around livestock production, an area that women have told us is critical to their livelihoods. We ask them to tell us about their networks: Where do you get information about medicines? Where can you buy medicines? Who gives treatment? Who buys milk? Eggs? Cattle? Where, when? We use thick lines to represent strong important linkages, thin for less regular, and a different colour to show new links that people would like to make. Through the women we learn about a new actor - an itinerant egg buyer who travels from house to house every day, paying a couple of taka (one or two pence) for an egg. Following up this link we later discover a complex system whereby the women's single eggs are collected by local middlemen and become boxes of thousands of eggs shipped by rail every evening to the port city of Chittagong. These small free range or deshi eggs are valued above factory farm-produced eggs, and sell at a premium in Bangladesh's large cities. The women would like to form their own egg-marketing network so that they can add a few taka to their profits by going directly to the middleman on the mainland. The women find it harder than the men to concentrate on map-making. Nearly everyone has a baby at the breast or a toddler or grandchild on her knee. They constantly have to get up and go to deal with domestic dramas. The men's groups are more focused. Their networks are more complex and they engage in lively conversation. Today we learn how the Dhaka traders come every year to buy livestock before the great Eid cattle slaughter. Last year the villagers put their livestock together in one big corral and bargained for a better price. It is initiatives like this that Development Wheel and their partners want to build on to support char dwellers in getting the maximum advantage from their enterprises.

Maps like Figure 1 (see below) are generated with groups of char dwellers: men and women separately, and separately with farmers of different financial status. Richer households tend to have more effective networks or better terms of engagement. We learn for example that resource-poor farmers tend to sell their chilli green, i.e. early in the season, to meet pressing financial needs. Once green chilli is cut it has a limited shelf life and must be sold quickly. Buyers dictate prices and terms. Richer farmers, by contrast, can wait to harvest their chilli red. Since red chilli can be stored, these farmers can sell their chilli when the price goes up, a few months after harvesting.

After the initial mappings, our two research assistants continue to build our understanding of key networks by working with case study families over a 12-month period. Each time they visit they look at significant interactions over the last month. They also monitor the behaviour of key actors at the district, regional and national level. Some are invited to the Development Wheel office to do their own network mapping exercises. We hold a series of meetings with local business people during Ramadan month, mapping out their networks and discussing their aspirations, as we all wait in anticipation for the call from the mosque which tells us that we can break our fast. Then we treat our guests to dates and fresh coconut water, followed by the traditional iftar (end of fasting) feast of vegetable bhajis and puffed rice. We find the business people helpful and enthusiastic. They are intrigued and a little complemented to be involved in a development project and to find that their role in providing information to farmers is recognized.

We compile our findings into actor linkage matrices (described in Box 1), since this data storage method will allow us to pinpoint areas that we would like to support and to monitor changes over time. We find this tool hard to share in a group situation or for presentation purposes, because it takes people some time to understand the matrix. However, for team purposes, the tool is a useful way to condense and share information.

Writing up, stage one: 'anti-social anthropology'

It was during this mid-research period that, blissfully unaware, we made the dangerous move of writing up and distributing our interim findings. The results were described at the beginning of this paper. To balance the negative outcome, we should say that the document in question met with great support from the char dwellers that we were working with. At a meeting with government staff from the Agricultural Extension Department, held on the chars, char dwellers used the report to berate the government staff for their poor performance. For them, the network mapping provided a useful way to monitor government performance. The printed version of the map they had created with us legitimised the way that they felt about the service they were receiving. Interestingly, char dwellers were allowed to wield the document and chastise the government staff without reprisal. In fact, they used the document successfully, as the meeting ended with the District Agricultural Officer committing himself to providing a better service. Shortly afterwards an extension officer was appointed to the area. However, the fact that Development Wheel, a small NGO, had wielded this document (and had been responsible for it getting into the hands of the char dwellers) was another matter. The damage to the team's tenuous relationship with government staff was almost irreparable.

Unlike 'traditional' anthropological research, which often focuses on rural communities and powerless groups, our analysis of actors included members of large formal organizations (such as the Ministry of Agriculture) and powerful officials. In his account of carrying out ethnographic research on a development project in India, Mosse (2001:176) observes that:

"contrary to the tenets of academic research, in organizational settings information is rarely viewed as a 'public good'. Research findings which contradict an organization's self image and the way it represents itself are met with hostility. Where the organization is a powerful one, unwanted findings may even lead to the legitimacy of the research being undermined."

An institutional study of organizations involved in aquaculture development, carried out in Bangladesh by Lewis (Lewis 1998), was for example terminated when findings were found to contradict the self image of the powerful organizations involved. As Mosse (2001:177) notes, 'ethnographic work in organizations makes the link between power and knowledge unusually clear'.

In this research activity, we anthropologists were working as 'insiders'. We were an integral part of the coalition team and had an interest in our research resulting in productive development interventions. As such, we had a responsibility to maintain good relations and to support partnership-building between key actors. Yet the very nature of our work was threatening to produce what Mosse (2006) aptly calls 'anti-social anthropology'!

A point to note when considering this first stage of writing up is that the dramas over the first document at least show that we had succeeded in writing something accessible and useable. The visual diagrams provided a means of writing up - albeit in a very condensed form - that made our research relevant and interesting to a wide group of people. Farmers and extension workers could both relate to this text.

Writing up, stage two: socially aware social anthropology

Though our first piece of writing up was used by char dwellers to attain a positive development outcome, our bumpy ride with government staff encouraged us to revise our approach to writing up in the second stage of the project. We did this in three ways. 

The first was to work harder to build consensus around findings before writing up took place. As a team we initiated regular sessions of what we called 'proactive monitoring'. Here we recorded changes to our planned project outcomes and action needed. One of our planned project outcomes was 'development of research and action coalitions', so changes in relationships or potential dangers were discussed and necessary action planned. For example, when our research assistants observed that there was some negative gossip about our activities on the chars, we planned meetings with local religious leaders to update them on what we were doing. When we had completed the case study monitoring, we held group meetings with all our case study members to discuss, verify and analyse our findings. After drafting our final social network maps and drawing out implications, we invited all stakeholders to a big meeting where maps (still not in print, but drawn onto flip charts) could be debated, discussed and modified if necessary. This final meeting was held in the Higher Secondary College across a paddy field from Development Wheel's office. Extension staff and government officials arrived by car and motorcycle, char dwellers from Kulpal arrived by rickshaw, and a boat was chartered to collect interested people from Nolshunda char, which is further from the mainland. In the Development Wheel courtyard, staff were busy preparing hundreds of lunch boxes for the end of the meeting. Everyone gathered in one of the larger classrooms before moving into separate rooms to look more closely at the livestock or chilli maps, depending on their interest.

While our findings, as they were presented on the maps, were not palatable to all, the beauty of this meeting was that it provided an opportunity for those who had scored badly to commit themselves to improving (and thus seeing a new red line on the map showing a commitment). We had several positive commitments made over this period: agricultural researchers promised to bring new seed varieties to farmers' groups, extension staff set up a stakeholder coordination group for the chars, Development Wheel committed itself to setting up an NGO coordination body, a private sector buyer announced that they would be buying chilli from the area in future, and a chemical manufacturer offered free training sessions to farmers' groups. 'Writing up' was occurring in the meeting. The delay in committing to print was allowing the group to construct something closer to the reality they would like, rather than quibble over the way we had represented reality in a printed - and thus somehow 'finished' - document.

Our second strategy was to change our focus from failure to achievement, following the 'appreciative inquiry' approach (see Watkins and Mohr 2001). This approach, which is used extensively in business development, moves away from a focus on problems towards identifying and sharing strengths and successes instead. Although it was easy to fault the performance of networks, and particularly to criticize government employees, we found that it was not constructive (particularly when our status did not allow us to give criticism). A focus on weakness did not take us forward in terms of our project goals; on the contrary, it led those criticised to behave defensively or with hostility, which was a barrier to partnership building. As the year progressed we found ourselves focusing on identifying the strengths and opportunities that our various actor groups had to offer, and looking for complementarity and potential for partnerships. As can be seen from the offers made at the final meeting, this did generate a spirit of mutual appreciation, willingness, and a 'can do' mentality.

A third strategy was to find local allies who would support our work and present it in a favourable way to their 'community'. We did this by inviting members of these communities to join us for field visits. Field visits, with their long journeys, slight hardships and adventures, are bonding activities in themselves. They also gave us a chance to get to know each other, make friends, and have long talks about our work. After an initially very bumpy start with the Department of Livestock Research (the Director does not favour NGOs), we persuaded a staff member to join us on a field trip.

At the end of the trip we had a firm friend in the department. Our idea of working with traditional healers had initially been rejected outright by the Livestock department: 'These people are quacks'. Discussing the idea with our new ally in the department, we found that it might be possible to incorporate it into an ongoing research programme on indigenous healing. A similar field visit with a staff member from a commercial chilli production firm resulted in him making a recommendation to his company to come and buy from char dwellers in the project area.

But what implications did our strategies have for the quality and rigour of our research? It is true that our final analysis was not as critical as it might have been if we had had no concern for our audience's reaction (if it had been a purely academic work, perhaps - or if we had been a World Bank team!). However, for all that, we do not feel that we sacrificed on quality. The main difference was the interpretation used in the analysis. Our slant, which might have been critical, was instead, appreciative. We focused on achievements and offered constructive ways forward. 

For example, our first report had berated the fact that no extension staff reached the chars. Following this line of argument can be frustrating. Extension staff do not come to the chars because transport is difficult and they are not motivated. They are not motivated because they are not paid enough and expect incentives. This in turn reflects on the type of government in place, which is something we are not in a position to change in the short term. Our second report, by contrast, noted that while extension staff do not get to the chars, there are cases where char dwellers, farmers and business actors take the initiative to go to the extension and research staff themselves, and successfully obtain the information or inputs they want. Now this is a fruitful area to reflect on. Immediately there are potentially achievable options to be followed up: How could you build on this process? Could groups of farmers regularly send representatives to the extension offices, could Development Wheel perhaps support this? Could we build on the relationships formed in these initial visits? Not only is this appreciative inquiry approach more effective in generating action from research, it is also surely a more constructive and positive approach which anthropologists could consider using more generally.

When we talk of our writing up, we talk mainly of the discussion that took place at the last meeting. In comparison to this, the final report that we wrote, too large and long to be accessible to most of our key actors, was almost an irrelevance. The visual tools and the short interim two page summaries were more accessible and more powerful writing up instruments. This made us think that perhaps in the end the process was more important than the written product. In the applied context there are advantages to keeping the text as a 'work in progress' for as long as possible.


What have we learned as we have mapped and talked, argued and planned with the people of Char Kulpal and Char Nolshunda and the many characters whom they interact with and are affected by?

Firstly, that as applied anthropologists we need to be acutely aware of our own role and status in the development context we are working in. As we write we need to consider and be prepared for the type of reaction our text may provoke. With this awareness we can give our writing the best possible chance of being effective, using strategies such as building up consensus around our arguments, finding allies, pre-releasing results, and adopting a more constructive, appreciative approach in our analysis.

Secondly, our experiences have confirmed our belief that visual representation can be enormously useful in making 'writing up' more accessible and more powerful. Actor linkage maps are just one possibility. Photos, film and oral histories could also be useful. The Participatory Learning and Action informal journal, published online by IIED (, can provide many ideas here.

Thirdly, as far as the issue of academic rigour is concerned, we would argue that it is possible to carry out rigorous analysis but to make it palatable by using an appreciative inquiry approach. This has been our experience; others may disagree! In either case there is little doubt that applied anthropologists need a forum where they can discuss issues such as this. In Bangladesh there is currently no network of applied anthropologists, despite their frequent and significant involvement in ongoing development programmes.

A final word. We entitled this paper 'The dangers of writing up: a cautionary tale'. However, our aim is not to frighten, but to encourage. After all, a little danger adds excitement to any journey. We certainly believe and hope that our writing up difficulties have made us more reflective, thoughtful and innovative social anthropologists.


Lewis, D.J. 1998. 'Partnership as process: building an institutional ethnography of an inter-agency aquaculture project in Bangladesh.' In Development as Process: Concepts and Methods for Working with Complexity (eds.) D. Mosse, J. Farrington and A. Rew, 99-115. London and New York: Routledge and the Overseas Development Institute.

Matsaert, H., Z. Ahmed, N. Islam and F. Hussain. 2005. 'Using actor-oriented tools to analyse innovation systems: experiences in Bangladesh.' Participatory Learning and Action 51, 100-110 (,).

Mosse, D. 2001. 'Social research in rural development projects'. In Inside Organizations, Anthropologists at Work (eds.) D. Gellner and E. Hirsch, 157-182. Oxford: Berg.

Mosse, D. 2006. 'Anti-social anthropology: objectivity, objection and the ethnography of public policy and professional communities.' Journal of the Royal Anthropological Institute 12, 935-956.

Watkins, J. and B. Mohr. 2001. Appreciative Inquiry: Change at the Speed of Imagination. San Francisco: Jossy-Bass/Pfeiffer.

About the author: 

Harriet Matsaert is currently based in Nairobi, Kenya. She can be contacted by clicking here
Zahir Ahmed is Head of the Social Anthropology Department of Jahangirnagar University, Dhaka. 

Faruqe Hussain and Noushin Islam are at the International Centre for Diarrhoeal Disease Research, Bangladesh (ICDDR,B). 

No comments: