The field research experience described in this article arose from an involvement in a safe drinking water programme in rural Bangladesh since 2005. The projects set out to establish water supplies in several marginalised and arsenic-affected communities – a matter of grave urgency since the discovery of arsenic in the groundwater more than a decade ago. Millions of users are exposed and the implementation of solutions is generally inadequate. The research objectives were (1) to better understand the situation, and (2) to search for more adequate approaches. The research was predominantly informed by project activities on the ground and, in turn, this learning experience was fed back into the programme to help guide new activities. Essentially, the researcher acted both as an observer and practitioner. To fulfil this double role, Participatory Action Research (PAR) provides a potentially useful methodology. This paper presents the experiences with PAR in the field, and reflects on its participative, qualitative and cyclic characteristics. It discusses implications of PAR for governance, as well as for academic research.

NOTE: This self-archived version is the original submission before review. For the final publication see: Rammelt, C. F. (2014). Participatory Action Research in Marginalised Communities: safe drinking water in rural Bangladesh. Systemic Practice and Action Research, 27(3), 195–210.

Introduction

The following is perhaps a simplistic generalisation, but on one hand there is the social researcher, submerged for a period of time in a rural setting trying to understand the development, power structure and functioning of a community; on the other hand there is the practitioner or technical ‘expert’ who hurriedly works on fixing an urgent problem. With the former, villagers can grow tired of being the subject of yet another study that does not bring them any practical improvement. With the latter, they can grow frustrated with having yet another solution that does not fit culturally or economically. A balance must be struck between such opposites, and it becomes important to define one’s double-role as both an observer and practitioner. Participatory Action Research (PAR) provides a potentially useful approach to achieve this.

This article reflects on the application of PAR to a particular drinking water problem in rural Bangladesh. In the last four decades, Bangladesh succeeded in providing more than 97% of its population with access to groundwater for their drinking water needs (Ahmed 2005). However, halfway through the 1990s it was discovered that two-thirds of the groundwater tube-wells yield water contaminated with arsenic, exposing up to 77 million people to the risk of arsenic-related diseases (Smith et al. 2000). Arsenic occurs naturally in certain aquifers, and prolonged low-level exposure can lead to cancers and neurological disorders (World Health Organisation 2000). With low income and limited access to services, the rural poor are the most vulnerable. Given a poor nutrition and health status, they are also physically most susceptible to the poisoning (Atkins et al. 2007).

Despite the availability of a range of technological solutions and a considerable amount of funds from UNICEF, the World Bank and others, practical results have been meagre (Ahmed et al. 2006; Atkins et al. 2007). Relatively few projects have been implemented and even fewer have managed to sustain by developing suitable operation and maintenance arrangements. They have also largely failed to bypass existing power relations and extend equitable services to the very poor (Hanchett 2006; Rammelt and Boes 2008). Since it’s discovery, the problem has been extensively researched, although mainly focusing on the geological and medical aspects. Research on implementation on the other hand has been seriously neglected.

With this in mind, the Arsenic Mitigation and Research Foundation (AMRF), a partnership between Dutch and Bangladeshi academics and NGO practitioners, set out to establish safe drinking water supplies and health care schemes in several arsenic-affected communities, and to record the learning experience (Rammelt and Boes 2005). The author has been involved both as a co-founder of AMRF and as a doctoral researcher on the topic of implementation of drinking water infrastructures in marginalised communities (Rammelt 2009). The research objectives were (1) to better understand the situation, and (2) to search for more adequate approaches. The research was predominantly informed by project activities on the ground and, in turn, this learning experience was fed back into the programme to help guide new activities. Essentially, the researcher acted both as an observer and practitioner.

Some of the methodological challenges in addressing this particular problem are reflected more generally in the PAR literature. PAR has origins in the work of Kurt Lewin, pioneer in social and applied psychology; Paulo Freire, theorist in critical pedagogy; and Orlando Fals-Borda, sociologist and one of the founders of the movement. The approach has increasingly been utilised in development programmes with marginalised communities in both high- and low-income countries. While it grew in popularity, it was also criticised for lacking the methodological rigor of scholarly research (Frideres 1992). Supporters started conforming PAR to academic standards and may well have turned it into a ‘respected’ intellectual movement (Hagey 1997). As will be discussed, this process threatens to corrupt some its original strength by separating the respective learning processes of external and local participants.

In its roots, however, the Participative, Qualitative and Cyclic nature of PAR is still very relevant. This paper reflects on these characteristics and presents our experiences with PAR in the field. It then discusses implications of PAR for governance, as well as for academic research.

Participative

The common failure of top-down approaches, whereby exogenous knowledge is transferred into local communities and cultures, has often been to ignore and undermine endogenous knowledge^[1]. These are criticised for being presumptuous and taking the main concerns and solutions of a dominant minority as a development blueprint for the rest of the population that live in totally different socio-economic conditions with diverse cultural values and rationales. Top-down approaches are also challenged for purely pragmatic reasons. In the face of complexity, the outcome of externally defined interventions is uncertain, and it is an illusion to think that implementation processes can successfully be controlled from the outside (or from the top).

This realisation produced a shift in ‘development practice’ towards approaches based on dialogue with local groups in order to facilitate developments that better reflect their priorities. While recognising the importance of people’s own reality as a starting point in their own development, bottom-up strategies might in many cases have underestimated the tension between both types of knowledge. Under the existing structures of power, exogenous ‘expert’ knowledge is generally favoured, which can contribute to the marginalisation of the poor and their endogenous knowledge base – even unintentionally. This tension needs to be acknowledged and will require appropriate ways of interacting.

Research that draws merely or mostly from the dominant stock of knowledge will not only reinforce the marginalisation of people’s own expertise; it may also be considered unscientific because it gives a one-sided representation of the situation. The research effort itself must become participatory. Before discussing further the meaning and implications of this, we need to understand the notion of participatory development, which has been extensively debated over the past 40 years.

Participatory development

‘Participation’ can mean different things. It sometimes amounts to nothing more than a slogan. In its weakest form, people take part in consultations without the opportunity to influence the process, as the priorities, problems and solutions are defined externally. Slightly more elaborate forms of participation, classified as ‘Rapid Rural Appraisals’ (RRA), emerged in the 1970s as a cost-effective way of learning, involving brief rural visits with multi-disciplinary teams. While RRA is intended to provide information to outsiders, ‘Participatory Rural Appraisal’ (PRA) seeks to empower resource-poor people. In the 1980s, mainly in India, PRA grew based on the philosophy was that knowledge is power, and when shared the monopoly of information is broken (Chambers 1992; Pretty et al. 1995; Mukherjee and Chambers 2004).

It has been argued that PRA, for all its emphasis on capacity building, ownership and empowerment, is still fundamentally controlled by external institutions, e.g., funding agencies or development organisations. Often, local groups are formed in order to legitimise and meet predetermined objectives. At best, people participate in a more-or-less joint analysis and are trained to eventually take control over local decision-making. As an Australian Aboriginal once said: “we’ve had things done to us, things done for us, but never done with us.”^[2] Such relations are based on what Paulo Freire calls pseudo-participation because they are oriented towards concerns that are external (Freire 1972).

Ideally, the control currently vested in donors and project managers is transferred to the poor; their participation becomes self-mobilisation. This cannot happen overnight and relies on catalytic and supportive activities for which project staffs participate as facilitators or animators (Rahman 1993). This has been one of the ideological foundations of ‘Participatory Action Research’ (PRA), which aims to facilitate action and learning through dialogue^[3]. It reflects Paulo Freire’s ‘Pedagogy of the Oppressed’, which must be forged with, not for, the oppressed (Freire 1972).

However, the objectives of outside and local participants, as well as their respective processes of learning, are not always in harmony. It becomes essential for local marginalised groups to develop their own endogenous knowledge. This is perhaps the most important challenge for PAR; the biggest threat to its movement is its transformation into a formal academic research tool (Rahman 1993). The task of systematising people’s knowledge is not one to be organised by researchers from the outside. This does not mean there is no place for external inputs in the learning process.

AMRF defined its role in the drinking water program in Bangladesh as a ‘facilitating’ and ‘learning’ organisation. Among the different technical solutions for the arsenic problem, there is an overwhelming popular preference for deep tube-wells that draw water from underneath the contaminated shallow groundwater aquifers (Ahmed et al. 2006). In the AMRF projects, they provide two benefits: they produce clean drinking water and they create a practical basis upon which people have started building so-called Community-Based Organisations. These are crucial for operation and maintenance of the water supply, but they also help to bring about a dialogue with local communities regarding a range of other urgent activities, such as health care for example. Shifting to safe water is generally not sufficient to detoxify the blood and organs affected by years of gradual poisoning. A safe water supply must go hand in hand with medical support for existing patients, improved primary health care and food security (Rammelt et al. 2011).

As a starting point, rural communities are assisted in resolving the arsenic problem. Development being multifaceted, there must be an opportunity for all participants to learn as new unanticipated problems emerge in the process. To achieve this goal, AMRF functions as:

1. a facilitating organisation that provides a favourable environment for learning about the implementation process.
2. a learning organisation that feeds research experiences back into the AMRF strategy, thus contributing to the formulation of subsequent steps in the projects. This in turn opens the programme up to a new range of experiences, problems and challenges.

While these ideas may sound appealing, even with the best of intentions there can be practical barriers that make it difficult to bring about dialogue and consequently to build upon the priorities of marginalised social groups. Two of those barriers are discussed: bounded rationality and power relations.

First barrier to participation: Bounded rationality

A person’s cognitive capacity, the quality of information and amount of time at hand confines his or her ability to reason (Simon 1956). This ‘bounded rationality’ poses an important challenge in PAR initiatives. To deal with complexity and uncertainty, we (implicitly or explicitly) create concepts and models to help us understand systems and be less surprised by their behaviour. These intellectual constructs are simpler than reality, but may be checked against it. Those concepts that survive severe testing may become models (such as Newton’s or Einstein’s), but they are still not ‘reality’.

How we look at a system depends not only on cognitive and conceptual capacity, but also on the place of the observer in the hierarchy of a system. For example, a farmer can be seen as part of an elite, a landowner capitalising on the abundance of cheap labour during cropping season. The same farmer might also be seen as part of the poor masses, cheated by middlemen and vulnerable to the whims of the global market. Different points of view will highlight different fragments of the objective world. These perspectives are pushed further apart by the different individual mental models of the observers. The challenge is therefore to approach a problem from different angles and embrace different sets of concepts (Flood 2000).

For this, it is essential to acknowledge one’s own bounded rationality and recognise the validity of other rationalities through dialogue. This precondition is often not met. This has long been detected, perhaps starting with Paulo Freire’s criticism on the patronising character of western development institutions (Freire 1972). He suggested that the institutions have brought about relations of teacher and pupil. The teacher tries to keep control because he truly believes he knows best, and the pupil internalises this ‘truth’. It is possible that the ‘expert’ misjudges the quality of a development process entirely because his or her perspective is not challenged.

This is worsened by the relatively short time span of the outsider’s involvement. Short-term successes (or failures) do not automatically endure and cannot be taken for granted in the long-term. Social changes might not be discernible to the observer involved in relatively brief field visits, which then fail to challenge his or her preconceived world-views. The submersion in a rural village for a couple of months, or even a couple of years, can help to create a basis of trust and to sensitise the researcher to local rural living conditions. Nonetheless, with the option to ‘re-emerge’ from the situation at any stage, the world-views of marginalised social groups can never be fully appreciated. The intention here is not to solve this barrier, but merely to acknowledge it, and to suggest that exposure, even if only for a few months, is crucial.

Second barrier to participation: Power relations

The previous section attempted to make it clear that the challenges are to accept our own bounded rationality, engage in a dialogue and embrace multiple perspectives – ideally over an extended period of time. However, regardless of one’s appreciation of and respect for other world-views, we can still be faced with blind spots.

PAR claims to work directly with local capacities to bring about changes in power relations. If effective, this process may generate or worsen local conflicts of interest. The image that needs to be de-mystified is that of the harmonious community; the marginalised are not members of a uniform mass of “poor” (or more recently “hard core poor”). The blind spots arise when the exact informal, cultural, social and professional relationships in a community are not openly discussed with outsiders. For example, an agricultural day-labourer will not openly criticise the landowner on which he depends for his family’s survival. In general, there are fundamental barriers to reaching an open dialogue in a context where basic needs are not fully met.

There is a structural imbalance not just in human living conditions, but also in the power to decide over those living conditions (Galtung 1971). In such a situation, there is a strong resistance against change. First, it is not in the interest of the elite to divert from the status quo. Second, marginalised nations, groups, families and individuals are caught up in dependencies, which means that they are not in a position to act in their own long-term interests. Instead, on the short-term, they are resigned to the structures of power, and may in fact act in the interests of those that monopolises control over decision-making.

The earlier “teacher-pupil” relation can also develop between local and external partners within an organisation. Although this relation can be quite subtle, it can have direct financial or managerial consequences. With financial control in the hands of the foreign “experts”, the local staff learns that it is best to keep them happy. We have observed these problems in the drinking water case as well.

The research took place in some of the poorest villages in Bangladesh. It is clear that the poor in these communities cannot afford the capital costs and even sometimes the maintenance of most arsenic mitigation technologies – at least not individually. As far as we have observed, various drinking water projects have not established the necessary village institutions to share the costs of collective water supplies; Community-Based Organisations may exist only on paper. As a result many water supplies are mismanaged or have broken down. Many other technologies have been distributed directly to the local elite, or are controlled by them; the poor are almost always neglected. The wealthier social class relies on its connections with local (Non) Governmental Organisations to channel for its own benefit the distributed arsenic mitigation technologies. It also relies on its dominance in village institutions to monopolise benefits from new water supplies.

This does not imply that all members of this group are corrupt. One must keep in mind that we are dealing with relative wealth and power; well-connected villagers, local NGO staff or government officials are generally still part of a rural periphery with valid concerns about their own welfare. Also, one must consider the manner in which (inter)national agencies implement their programmes, which can be instrumental to this inequitable outcome, as will be discussed later.

Concluding comment

These barriers to participation (bounded rationality and power relations) raise questions about the origin of change, about how to facilitate developments reflecting the priorities of marginalised social groups – assuming these can be adequately identified. Recent PAR literature gives credence to some of the barriers, but does not seem to follow the argument through to its logical conclusion: If participation is to be taken seriously, the methodological tools to observe and learn about implementation and action must be developed during the process and not before. As researchers, we cannot and should not try to predict the implementation process, nor select in advance the specific tools that will be relevant to learning from it. Instead, researchers need to be flexible and able to mix and adapt a range of tools. The following section explores how certain analytical tools might lend themselves better than others to be used in a collective learning process.

Qualitative

Action research is generally qualitative (Dick 2000), as opposed to what has been referred to as the ‘Hard Systems’ or ‘System Engineering’ approaches, which steer systems to achieve stated objectives and tackle well-defined problems (Ferris and Cook 2006). The argument is that applying ‘hard’ information-intensive analytical tools fails to identify more complex problems accurately (Couprie et al. 2001).

Quantitative tools have also been applied to the planning drinking water infrastructures in rural Bangladesh. The World Bank undertook a quantitative cost-benefit analysis where the risk factors of different drinking water supplies were weighed in monetary terms. Its calculations included: the capital, operation and maintenance costs of different community-based water technologies; the direct health benefits gained from an improvement in the quantity and/or quality of drinking water, which the World Bank translated into greater work capacity, increased production and economic growth; the costs as a household spends time, labour and money to produce safe water; the costs for medical treatment; and the loss in production output due to people becoming unable to work or dying from unclean water (World Bank Water and Sanitation Program 2005a; 2005b). People’s interest in cost sharing schemes for different water supplies was measured through ‘affordability’ or ‘willingness-to-pay’ studies. This supposedly gave the agency insight in how much people intend to spend on different infrastructures, and allowed it to recommend solutions for different types of villages.

The World Bank argued that a single unit of measurement (money) is required; otherwise a “rational” comparison between options and their risks is impossible (World Bank Water and Sanitation Program 2005b). An implication is that those social and cultural concerns that cannot be quantified are overlooked. For example, there are major differences between people’s expressed willingness-to-pay and their capacity to influence procedures. In theory, the community is given the right to choose the technology and its place in the village but, in practice, decisions are taken in such a way that the water supply ends up in, or very near the homestead of influential villagers.

To blame this on government corruption would be simplistic. The implementation of new drinking water supplies is biased towards those who hold powerful positions within the community, who are familiar with public institutions and who are confident negotiators. We found instances where wealthy families collected signatures for deep tube-well application forms from their neighbours by offering to pay the full share of the contribution. In return, the community ‘agreed’ to have the deep tube-well installed on the property of the individual family.

The problem is compounded by organisations in a rush to meet targets set by donors or international development agencies. The sense of urgency associated with the arsenic calamity does not help either. Fast implementation implies working with those households that can quickly contribute larger sums to cost-sharing schemes or possess a site for a deep tube-well installation.

This quantitative method of economic modelling is part of a long history of infrastructure design and planning, which has been widely criticised (Ruitenbeek and Cartier 2001). Policies and incentives that look just at monetary or other quantifiable indicators may miss the mark entirely. Albert Einstein once said; “Not everything that can be counted counts, and not everything that counts can be counted”. Power, class, gender and ethnicity are expunged from analysis because they only seem to muddy the picture of how best to reach a technical agreement (Wilkin 2000). These are, however, the major causes for programs to fail to implement lasting and equitable water supplies.

Actor Network Theory (Law 1991; Callon 2001; Latour 2005) attempts to bring these factors into semi-quantitative analyses. People’s decisions are modelled as sets of elaborate if-then-else decision-making protocols (Crawford and Ostrom 1995), e.g., if the labourer criticises the landholder, he risks his job and therefore acts in the interests of the landholder. The problem is that such rules are not universal, nor are they always conscious or openly shared. Actor Network Theories assume that all “actors” can be equally well mapped. The approach sees power as emerging from the actions of “actors” within the network, but does not take into account pre-existing power structures (and other barriers to participation), which marginalise certain groups from the very process of “mapping”. The word “actor” implies the potential for active participation, but as stated earlier, marginalised groups might (unconsciously or not) act in the interests of the elite (Galtung 1971).

Even if people could theoretically express their views free of dominant control and structures of inequality, there is still a difference between what people say or plan and what they actually do^[4]. Psychologists have criticised models of decision-making that describe people as if they are rational and utilise all of the information presented to them (Tversky and Kahneman 1974). Psychologist Robert Leahy (Leahy 2004: p120) concludes that the “’true’ cost-benefit ratio is determined by what he [the subject] observes in his actions, not by what he says about what he might do”. A relevant lesson and guiding principle for PAR might therefore be to focus more on action and observation and less on explicit public deliberations, which inevitably face the barriers of bounded rationality and power relations.

In defence of any type of (semi-)quantitative modelling, however, “it does not matter that the model will ultimately prove to be inadequate, or at best reliable within only a limited domain. Rather, we derive insight and ideas from the learning process associated with building, evaluating, discarding, and revising models” (Gunderson and Holling 2002: p409). While the argument is legitimate, we need to find appropriate ways to openly discuss these models. Brian Wynne rightly points out that approaches such as the drinking water sector program of the World Bank in Bangladesh do not enlarge institutional processes that could encourage more constructive participation (Wynne 1989).

PAR has adopted a different set of tools to those proposed by conventional experimental methods, such as “Soft Systems” methodologies. These are non-numerical approaches to diagnosis and intervention. Multiple perspectives on a specific problem are collected and used to help structure a dialogue aimed at defining desirable changes (Checkland 1981; Flood 2000). A primary concern is that these qualitative tools cannot wriggle out of the aforementioned barriers to participatory approaches from which data are drawn (Flood 2000).

The dialogue that PAR aims to facilitate through “Soft Systems” and other tools is more than a conversation; its ultimate aim is shared learning from the development process in order to improve it and to study the failures and successes along the way. This suggests the need for a cyclic approach.

Cyclic

There are variations on this, but basically action research has been described as a learning cycle: plan, act, observe, reflect; then, in the light of this, plan for the next cycle (Lewin 1946; Argyris 1976; Kolb 1984). Reflection on action yields insights that may support an appreciation and improvement of the action. Reflection can also be directed at a deeper level, towards readjusting conceptual frameworks or paradigms (Altrichter et al. 2002). “[A]ny theory and set of practices is dogmatic which is not based upon critical examination of its own underlying principles” (Dewey 1963: p22). The divide between action and research is thus narrowed (Dick 2000; Flood 2000).

An important concern is about who reflects and who plans the strategy, i.e., how people’s participation is organised in the learning cycle. In contrast to the quantitative monitoring of physical ‘targets’ (e.g. kilometres of rural roads built, numbers of water supplies installed, hectares of irrigated land rehabilitated), the reflection on social development is more difficult because it is largely qualitative and long-term. External evaluators often resort to using tangible, yet inappropriate, indicators as measures of success (Pretty et al. 1995). This generally gives a distorted approach to development. “Imagine the development of a human child to be assessed in term only of indices such as his or her physical health, grades in school, and eventually his or her financial earnings. A ‘well-developed’ person in such terms may very well be a social nuisance and/or a very miserable person” (Rahman 1989: p1-2).

PAR proposes a collaborative learning process between researchers and people in the situation. For programmes to better reflect their priorities and rationalities, marginalised groups must be involved in all stages of the learning cycle, and especially in Participatory Monitoring and Evaluation (PME). There can be a large difference between a typical evaluation based on externally defined benchmarks (e.g. a Logical Framework Approach), and one that finds its quality in the internal dynamics of the development process based on local priorities and criteria. Much work has been done in this area (Neggers 1998; Vernooy 2005); but the ideas at times seem self-defeating, as the PME systems and procedures are not designed locally, but by social scientists and development specialists.

AMRF is faced with similar challenges in the context of its drinking water program. Through monitoring and evaluation, the reasons why projects fail often turn out to be quite straightforward: if people do not react to public awareness programmes, there is a chance that they can’t identify with the messages; if it is difficult for the poor to contribute financially to the installation of a water supply, they might not have been given sufficient time to save money. Moreover, solutions often follow rather logically from the evaluation of obstacles: if people cannot afford to buy vitamins, they could be assisted to set up their own small vegetable gardens; if the owner of the land monopolises the water supply, collective land could be registered; and so on. This does not mean that all problems are easily identified or resolved, nor that short-term success will last. Although initial results in the AMRF program seem promising, their sustainability is not guaranteed. Both ‘successes’ and ‘failures’ are to be monitored over a longer period of time.

Catalytic and supportive activities must remain relevant to local priorities, particularly those of the poor. This ‘reality check’ can be provided through a Participatory Monitoring and Evaluation (PME) system. Ideas for how to do this are currently being developed. Based on people’s valuation of the implementation process, qualitative and quantitative criteria will be chosen, and activities monitored (Rammelt and Boes 2005). The aim is to phase out external inputs, gradually replacing them with community inputs (Chambers 1997; Neggers 1998). In our view, project activities will become more appropriate as the institutional strength of the Community-Based Organisations and their participation in monitoring and evaluation grows.

Catalytic and supportive activities, such as the installation of a deep tube-well, must be chosen and directed carefully. Managers must give up control – or, rather, the illusion of control. All they can really do is work to create an environment in which learning can occur. To develop a collaborative learning cycle, there is a role for external participants, but the outsider must consider himself/herself a participant in someone else’s development, and success must be assessed from the perspective of local communities (Checkland 1981). Development is a lengthy process that has to come from within and interfering from without may be tactless, inappropriate and perhaps even dangerous.

Conclusions and implications

The participatory, qualitative and cyclic nature of PAR has implications beyond the local level; an action researcher can be affected by funding and academic structures at odds with the open-ended and subjective character of the work.

Implications for management and governance

The adoption of a PME system within PAR initiatives implies that more flexible funding structures are needed so that – depending on what comes up in practice – organisations can abandon certain undesirable activities and make room for new ones better geared towards endogenous risk perceptions, knowledge and priorities.

For Gunderson and Holling (2002), the main question is whether we can find ways to distinguish between what is predictable, uncertain and inherently unpredictable – and then, whether we can define adaptive responses. The dismissive answer is that the complexity of real problems is beyond our capacity to act (Rescher 1998). Kurt Lewin turns this paralysing reasoning around and argues that if you truly want to understand something, try to change it (Lewin 1946). Under the umbrella of “adaptive management and governance”, tools have been developed to help structure an iterative process of decision-making in the face of uncertainty (Gunderson and Holling 2002). While some still adhere to the view that science will eventually fill any knowledge gap, others suggest that policy initiatives should be undertaken as “experiments” that acknowledge indeterminacy (Dovers 2005).

The concept of adaptive management is influenced by the idea of the “learning organization”. Donald Schön, for example, believed that “[w]e must become able not only to transform our institutions, in response to changing situations and requirements; we must invent and develop institutions which are ‘learning systems’, that is to say, systems capable of bringing about their own continuing transformation.” (Schön 1973: p28) This opportunity for learning is primarily discovered at the periphery and not in the nexus of official policies. The centre must function not as society’s trainer, but as its facilitator (Dewey 1963; Illich 1971; Freire 1972).

From our observations in the Bangladesh water sector, this adaptive capacity is lacking. The World Bank argued that despite the limitations, a quantitative cost-benefit analysis is the only way to rationally plan water supply developments (World Bank Water Supply Program 2005b). In an evaluation of the programme, the World Bank acknowledge that the distribution of deep tube-wells in some communities may have been uneven, because some people were able to influence decisions regarding the installation sites (World Bank 2007). It however refuses to take any blame: “It is entirely possible that those contributing more than their share were able to influence the location of wells; however, there is no evidence to support the position that the project promoted this outcome” (World Bank 2007: p7). In our experience at the micro-level, the evidence is there; the question is whether there is sufficient engagement in monitoring, learning from failures and openly discussing them (Rammelt 2009).

The National Policy for Arsenic Mitigation (NPAM) mentions that “[t]he policy shall be reviewed and updated depending on the implementation feedback, if and when such need arises” (Government of Bangladesh 2003a: p6), but it does not explain how this will be achieved. In another section, the policy clearly does not leave much room for initiatives to emerge from the bottom-up: “Different projects and agencies engaged in arsenic mitigation in different parts the country shall be required to accommodate their mitigation approach to this implementation plan. If any project becomes unable to accommodate the national action plan or implement the national emergency water supply program, DPHE shall implement the emergency program in the area” (Government of Bangladesh 2003b: p7).

Implications for academic research

Action research is the creation of a perturbation in a system and the observation of its resulting behaviour (Lewin 1946; Argyris et al. 1982). This approach challenges the view based on an ideal of objectivity in scientific inquiry where researchers must make every effort to neutralise whatever influence they might exert on the object of study. By actually triggering or facilitating certain changes and activities, a researcher becomes a practitioner and thus a subject in his or her own research. This, it is argued, puts the researcher’s objectivity at risk (Frideres 1992).

A counter-argument is that, in a real-world setting, even plain data collection can easily become a value-laden operation. “Suppose that a systems model of a chemical system, say a model of the kinetics of a chemical reaction, does not match the (repeatable) measured kinetics: then the fault must lie with the model builder. But when a model of a human activity system does not match observed human activity the fault might be the model builder’s, but it might also be due to the autonomous real-world behaviour of human beings” (Checkland 1981: p249). Put differently, it is very likely that the research set-up will drastically change during the “experiment”. Under conventional methods the data collected up to that point would have had to be abandoned because the odds under the null hypothesis changed^[5] (Dick 2000).

Almost every academic field has developed its own specific rules of evidence and guidelines for objectivity. Validation of research rests on collective verification that makes individual subjective statements and findings objective (Rahman 1993). An “objective fact” remains true everywhere, independent of human thoughts or feelings. A “subjective fact” is only true at certain times, places or people. Objectivity in this sense requires a transition from the individual to the collective. All well-established schools have a verification system of such a nature, explicitly or implicitly^[6].

A top-down policy-making process draws from research and tools that are both quantitative and based on large-scale surveys and statistical analysis. These tools are rigorous, but only as part of a verification system that itself is inadequate because it is closed to other verification systems. In other words, the knowledge generated by this type of research is not a mix from different knowledge bases. As argued, incorporating the collective views of the local communities through participatory approaches is complicated by the cultural context where people will tend to say what they think you want to hear or what others impose on them. Broader verification will therefore probably not come out of formal interviews and surveys, and until a long-term relationship of trust develops, verification has to rely more on informal meetings and observations.

This article did not attempt to present a wide-scale study on the failures of top-down approaches; the intention was to try to understand the reasons behind the failures that we have directly observed in and around our working areas. Likewise, this research did not look for empirical evidence to demonstrate the merit of the AMRF approach; only time will make this determination. The objective was to begin understanding the different relevant factors for the implementation of safe, durable and equitable drinking water supplies, and PAR provided an appropriate approach. The outcome could very well be that the AMRF approach is only partially successful – as mentioned earlier, this success is defined by criteria of poor social groups – and knowing this would be very valuable. A less gratifying conclusion, but a possibility nonetheless, might be that the AMRF approach is not working at all, which perhaps requires addressing the problem in a different manner.

Acknowledgements

The ideas presented in this chapter reflect the author’s collaboration with Dr Masud Md. Zahed and Mr Palash Torfder, respectively of AITAM and PRIDE, as well as with Mr Jan Boes of AMRF. The chapter also draws on research under the supervision of Dr John Merson and Dr Phillip Crisp. The author would like to acknowledge their contributions.

References

Ahmed, K. M. 2005. Management of the groundwater arsenic disaster in bangladesh. In J. Bundschuh et al. (Eds.), Natural arsenic in groundwater: Occurrence, remediation and management. London: Balkema.

Ahmed, M. F., Ahuja, S., Alauddin, M., Hug, S. J., Lloyd, J. R., Pfaff, A., et al. 2006. “Epidemiology: Ensuring safe drinking water in Bangladesh”. Science 314 (5806): 1687-1688.

Argyris, C. 1976. “Single-Loop and double-loop models in research on decision making”. Administrative Science Quarterly 21 (3): 363-375.

Atkins, P., Hassan, M., & Dunn, C. 2007. “Poisons, pragmatic governance and deliberative democracy: The arsenic crisis in Bangladesh”. Geoforum 38 (1): 155-170.

Callon, M. 2001. Actor-Network theory. In International Encyclopedia of the Social and Behavioral Sciences. Ed. N. J. Smelser and P. B. Baltes. Amsterdam: Elsevier.

Chambers, R. 1992. Rural Appraisal: Rapid, Relaxed and Participatory

Checkland, P. 1981. Systems Thinking, Systems Practice. Chichester: Wiley.

Couprie, D., Goodbrand, A., Li, B., & Zhu, D. 2001. Soft systems methodology. http://sern.ucalgary.ca/courses/seng/613/F97/grp4/ssmfinal.html#Stage2 (accessed May 9, 2011).

Crawford, S. E. S. & Ostrom, E. 1995. “A grammar of institutions”. The American Political Science Review 89 (3): 582-600.

Dewey, J. 1963. Experience and Education. Kappa Delta Pi lecture series. New York: Collier.

Dick, B. 2000. A Beginner’s Guide to Action Research, no. 20/06/2007.

Dovers, S. 2005. Environment and Sustainability Policy: Creation, Implementation, Evaluation. Annandale: Federation Press.

Ferris, T. L. & Cook, S. C. 2006. The bounds of systems engineering. In R. Attwater & J. Merson (Eds.), 12^th ANZSYS conference – sustaining our social and natural capital. Katoomba: ISCE Publishing.

Flood, R. L. 2000. “A brief review of Peter Checkland’s contribution to systemic thinking”. Systemic Practice and Action Research 13 (6): 723-731.

Freire, P. 1972. Pedagogy of the Oppressed. Penguin Education. Harmondsworth: Penguin Education.

Frideres, J. S. 1992. Participatory research: An illusionary perspective. In A World of Communities: Participatory Research Perspectives. Ed. J. S. Frideres. North York, Ontario: Captus University Publications.

Galtung, J. 1971. “Structural theory of imperialism”. Journal of Peace Research 8 (2): 81-117.

Government of Bangladesh. 2003a. National policy for arsenic mitigation.

Government of Bangladesh. 2003b. Implementation plan for arsenic mitigation in bangladesh.

Gunderson, L. H. & Holling, C. S. 2002. Panarchy : Understanding transformations in human and natural systems. London: Island Press.

Hagey, R. S. 1997. “Guest editorial: The use and abuse of participatory action research”. Chronic Diseases in Canada 18 (1).

Hanchett, S. 2006. Social aspects of the arsenic contamination of drinking water: A review of knowledge and practice in bangladesh and west bengal. Dhaka: Arsenic Policy Support Unit.

Illich, I. 1971. Deschooling Society. New York: Harper & Row.

Kolb, D. A. 1984. Experiential Learning: Experience As the Source of Learning and Development. London: Prentice-Hall.

Latour, B. 2005. Reassembling the Social: An Introduction to Actor-Network-Theory. Clarendon lectures in management studies. Oxford: Oxford University Press.

Law, J. 1991. A Sociology of Monsters: Essays on Power, Technology, and Domination. London: Routledge.

Leahy, R. L. 2004. Decision making and psychopathology. In Contemporary Cognitive Therapy: Theory, Research, and Practice. Ed. R. L. Leahy. New York: Guilford Press.

Lewin, K. 1946. “Action research and minority problems”. Journal of Social Issues 2 (4): 34-46.

Mukherjee, A. & Chambers, R. 2004. Participatory rural appraisal: Methods and applications in rural planning: Essays in honour of Robert Chambers. New Delhi: Concept Pub. House.

Neggers, J. C. 1998. Programming, monitoring and evaluation system (PMES), a background paper. In Black Sea NGO Forum Workshop. Varna, Bulgaria.

Pretty, J. N., Guijit, I., Scoones, I., & Thompson, J. 1995. A trainer’s guide for participatory learning and action. London: IIED.

Rahman, M. A. 1989. Qualitative Dimensions of Social Development Evaluation. Dhaka.

Rahman, M. A. 1993. People’s Self Development: Perspectives on Participatory Action Research. London, Dhaka: Zed Books; University Press.

Rammelt, C. F. & Boes, J. 2005. Implementation of safe drinking water supplies in bangladesh. In J. Bundschuh et al. (Eds.), Natural arsenic in groundwater: Occurrence, remediation and management. London: Balkema – Taylor & Francis.

Rammelt, C. F. & Boes, J. 2008. The autonomy of local drinking water institutions in rural bangladesh. In P. Bhattacharya et al. (Eds.), Groundwater for sustainable development – problems, perspectives and challenges. London: CRC Press Balkema – Taylor & Francis.

Rammelt, C. F., Masud, Z. M., Masud, F., & Boes, J. 2011. “Beyond medical treatment, arsenic poisoning in rural Bangladesh”. Social Medicine 6 (1): 22-30.

Rammelt, C. F. 2009. Development and infrastructure in marginalised communities. Thesis, Sydney: UNSW.

Rescher, N. 1998. Complexity: A Philosophical Overview. Science and technology studies. New Brunswick: Transaction Publishers.

Ruitenbeek, J. & Cartier, C. 2001. “The invisible wand: Adaptive co-management as an emergent strategy in complex bio-economic systems”. CIFOR (34), 51.

Schön, D. A. 1973. Beyond the Stable State. New York: Norton.

Simon, H. A. 1956. “Rational choice and the structure of the environment”. Psychological Review 63:129-138.

Smith, A. H., Lingas, E. O., & Rahman, M. 2000. “Contamination of drinking-water by arsenic in Bangladesh: A public health emergency”. Bulletin of the World Health Organization 78(9): 1093-1103.

Tversky, A. & Kahneman, D. 1974. “Judgment under uncertainty: Heuristics and biases”. Science 185 (4157): 1124-1131.

Vernooy, R. 2005. Participatory monitoring and evaluation, Readings and Resources. Rural Poverty and Environment program initiative, IDRC

Wilkin, P. 2000. “Solidarity in a global age – seattle and beyond”. World-Systems Research 6 (1): 19-64.

World Bank. 2007. Implementation completion and results report on a credit in the amount of SDR 24.2 million (US$ 44.4 million equivalent) to Bangladesh for arsenic mitigation water supply. World Bank, South Asia Region.

World Bank Water and Sanitation Program. 2005a. Local government-based community WSS services. Washington: World Bank.

World Bank Water and Sanitation Program. 2005b. Towards a more effective operational response, arsenic contamination of groundwater in south and east asian countries. Washington: World Bank.

World Health Organisation. 2000. Towards an assessment of the socioeconomic impact of arsenic poisoning in Bangladesh

Wynne, B. 1989. Building public concern into risk management. In J. Brown (Eds.), Environmental threats: Perception, analysis, and management. London: Belhaven Press.