Knowledge, context and problemsheds: a critical realist method for interdisciplinary water studies

ABSTRACT

Understanding water issues as problemsheds addresses the narrowly water-centred framing of watershed and basin-focused water research and policy. In a critical realist approach problemshed also serves to identify the context-specificity of water knowledge, by navigating between the extremes of positivist generalization and interpretivist local specificity and bridging the divide between academic and applied research by identifying the structural similarity in their problem framing. Problemshed is operationalized by situating it in critical realism’s structures-mechanisms-events ontology, and by drawing on realist evaluation’s context-mechanism-outcome configurations. I use large-scale canal irrigation in India to illustrate how this is done.

KEYWORDS:

• Interdisciplinary water studies
• problemshed
• CMO configuration
• critical realism
• contextuality of knowledge

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1. With ‘interdisciplinary water studies’ I refer to the area of scholarship that looks at freshwater use, management and governance with a broader ‘social’ perspective than the longer-established technical water sciences (including hydrology, hydraulics and other specializations). Within this, ‘critical’ water studies explicitly address the social relations of power inherent in water use, management and governance, and have an explicit normative stance that seeks to address equity, justice, sustainability and other problems in water situations. ‘Mainstream’ water studies do not explicitly address social relations of power and construe science as politically neutral or objective. There is a lot of more-or-less and different kinds in between. I write ‘water studies/research’ for short after the introductory section. For an effort to describe critical water studies as a field, see Mollinga (2008).

2. Ontology is concerned with identifying the kinds of things that exist; epistemology is concerned with the nature of knowledge: its possibility, scope and general basis (https://www.webpages.uidaho.edu/engl257/classical/ontology_and_epistemology.htm, accessed 10 January 2020). For a more elaborate treatment of these two categories see https://plato.stanford.edu/entries/epistemology/and https://plato.stanford.edu/entries/logic-ontology/ (accessed 16 June 2020).

3. See Burawoy (1998) for these three ‘levels’ of method.

4. A fascinating illustration of the ‘worlds apart’ in much of such discussion, and the emotion invested in them, is ‘An exchange between Ben Crow and Yoram Eckstein on the global water crisis’, Water International 39(5): 774–784 (July 2014).

5. I use here Grix’s (2004) threefold division of philosophy of science schools into positivism, critical realism and interpretivism. ‘Critical realism situates itself as an alternative paradigm both to scientistic forms of positivism concerned with regularities, regression-based variables models, and the quest for law-like forms; and also to the strong interpretivist or postmodern turn which denied explanation in favour of interpretation, with a focus on hermeneutics and description at the cost of causation’ (Archer et al., 2016). Critical realism (Sayer, 1984/1992) distinguishes itself through a ‘depth ontology’ in which there are structures, mechanisms and events (reality as an open system exhibiting nonlinearity, equifinality and emergence); a notion of generative causality as against successionist or configurationist views of causality (Pawson, 2008); an epistemology that emphasizes the inherent fallibility of knowledge while maintaining a notion of testability and ‘practical adequacy’ in relation to empirics and research purpose, and retroductive reasoning as the basis for identifying causal relations for generating explanation.

6. Interdisciplinary water studies as a field has clear regional and sector-wise subfields or clusters, among other divisions.

7. I use the term ‘public action’ to refer to different forms of policy and advocacy in a generic sense (cf. Mackintosh, 1992).

8. Except in water and health studies, randomized control trials, the ‘gold standard’ of objectivist research for some (see Sampson, 2010, for discussion), are not used in the type of water studies discussed in this article. Systematic reviews are increasingly used in water studies (cf. Hooper, 2015; Plummer et al., 2012). For a critical realist critique of systematic review as a method, see Clegg (2005).

9. There is a wide-ranging literature on (environmental) policy, knowledge and power, discussion of which is outside the scope of this article; see for instance Keeley and Scoones (2000).

10. See for instance Mollinga (2009), Uhlenbrook and de Jong (2012), and Richard-Ferroudji (2014).

11. My own exposure includes policy-related workshops on irrigation reform in India and internationally, and global water meetings like global water forums and weeks.

12. Also see e.g. Bennett and Herzog (2000); Cohen and Davidson (2011); Duram and Brown (1999); El-Ahraf et al. (1999); and Gerlak (2006).The oldest published sources using ‘problemshed’ (also written as ‘problem shed’ and ‘problem-shed’) I have found are Fisher (1967) and Kneese (1968); both discuss US environmental management. The notion is also used outside the sphere of environmental/water management; see for instance Pillemer (1988), Griswold et al. (2013), and Bartel (2015).

13. Such logic is strongly sedimented in the common-sense understanding of both disciplinary water academics and water practitioners. While for the hydrologist this may be a case of analytical reductionism informed by disciplinary framing, for the water resources professional it may be a strategic essentialism that helps keep the water bureaucracy at centre stage and professionalism fenced and guarded. Disciplinary scientists and resource governors may form powerful alliances. On basins and power, see Molle (2007, 2009).

14. Discussing the issue of scale in greater depth is outside the scope of this article. I find Edwards (2003) on micro-meso-macro scales useful for studying canal irrigation as a large sociotechnical system. For this article it is worth emphasizing that ‘research objects’ and ‘research questions’ are not necessarily confined to a single scale-level, even when methodological issues are often most easily presented in that manner. In fact, the context specificity of water knowledge is to a large extent about the multiscalar nature of water resources processes (cf. Hoogesteger et al., 2016; Hüesker & Moss, 2015).

15. This gets institutionally translated, for instance, as the moment at which research projects are submitted for funding and forms the basis on which they are assessed, and as the moment PhD students get research proposals approved that allow them to start their fieldwork.

16. I use the term as in feminist standpoint theory. See particularly (Haraway, 1991) on ‘situated knowledge’; also see Furlong (2015).

17. Such casual description of positionality can be deepened by ‘exploring how the affective is related to the cognitive in individual lives of … scientists’ (Shah, 2018, preface). She argues that ‘conceptual theories are isomorphic with the world emotionally and existentially desired’, thus aiming to show how the ‘affective self is profoundly constitutive of the method and philosophy of science’ (chapter 1).

18. A more sophisticated way of engaging with this issue is found in analyses of the relationship between (Western) science and indigenous knowledge systems (cf. Goodall, 2008; Wohling, 2009).

19. It is possible that my result is a product of framing the role of caste in a way that is not relevant to the situation studied – my knowledge remains fallible (for instance, I focused on caste difference rather than within-caste relationships). The result does not imply that caste is not important in Indian irrigation in general. Its importance (causal efficacy) is well documented for tank irrigation (Mosse, 2003). In the strongly caste-stratified society of North Indian Uttar Pradesh it may also be pertinent (Pant, 1984), and it is relevant in groundwater markets in West India (Naz, 2015; Prakash, 2005). One reason caste may not have been pertinent to water distribution in this case is the overwhelming importance of the presence of migrant settler farmers in addition to local, original inhabitant, farmers. The ‘mixed up’ caste profile of agricultural production that resulted may explain why the positions of employer and/or moneylender were decisive. This all illustrates the contextual nature of knowledge.

20. The analysis was first developed based on data on three secondary canals (out of over 80 in total in the irrigation system). It was subsequently tested in a number of large distributary canals in the same irrigation system, and in a number of other irrigation systems in the same state. It was found to be robust (this additional research has yet to be published). It is likely that the structure and socio-logic would not hold in other parts of India where water distribution has been designed in qualitatively different ways.

21. Based on unpublished fieldwork. One finding of these repeated visits is that unequal distribution geographically concentrates slowly over longer periods of time, possibly reducing the intensity of everyday water struggles, tendentially changing some of the ‘core logic’.

22. Put differently, critical realism has a ‘generative’ rather than a ‘successionist’ understanding of causality (Harré, 1985, p. 116). Also see Pawson (2008).

23. For rainfall this can be easily shown, as it behaves stochastically across years, and thus varies naturally. Effects can be relatively easily observed, for instance in the influence of very low or very high rainfall in a given year on the prevalence water distribution conflicts. Establishing the causal relevance of other conditions of possibility may be less straightforward.

24. As Denyer et al. (2008, p. 399) summarize: ‘Pawson (2006) argues that the key to synthesising research for the purpose of informing practice lies in the developing an understanding of the underlying generative mechanisms (M).’

25. Frequently cited in the realist evaluation literature is Tilley’s (1993) listing of nine different potential mechanisms explaining how the installation of CCTV (closed-circuit television) near car parks reduces car break-ins and theft (the observed outcome).

26. Pawson and Tilley (1997) use the term ‘programme theory’ for the causality (chain) assumed in a policy or programme. Realist evaluation is about testing that assumed causality against observed outcomes by identifying actually existing mechanisms and conditions.

27. Denyer et al.’s (2008) term for the assumed/desired causality is ‘design proposition’. They distinguish between ‘explanatory science’ and ‘design science’. ‘The mission of a design science is a quest for improving the human condition by developing knowledge to solve field problems, i.e. problematic situations in reality’ (p. 394). ‘Design propositions created [by applying the CIMO logic] … contain information on what to do, in which situations, to produce what effect and offer some understanding of why this happens’ (p. 396).

28. Like Pawson and other realist evaluators, for Cartwright only the steps/sequence in the middle are the ‘mechanism’, and the things outside this are ‘context’. Cartwright understands mechanisms as ‘Elster type mechanisms’, being dispositions/behaviours that are common in individuals or institutions (for more detailed specification see https://www.lshtm.ac.uk/newsevents/events/evidence-action-new-settings-importance-middle-level-theory, accessed 18 June 2020; on ‘Elster mechanisms’ see Persson, 2012; Chuang, 2012). In terms of the critical realist understanding of ‘mechanism’ there is thus no perfect match, as Cowen and Cartwright themselves also observe. From a critical realist perspective context is nothing but a set of relevant other mechanisms (as indicated in Figure 5 by using ‘mechanism’ also in the context boxes). Context is defined in relation to the direct causal process of interest, but not ontologically different from the specific process that is the research focus.

29. Such unpacking through specification helps avoid correlational and structuralist approaches to how context shapes outcomes. Both correlational and structuralist approaches look at the C–O connection without specifying the M (and I).

30. For further exploration see Halkier (2011) and Danermark et al. (2002).

31. Moreover, these conditions are partly recursively related to the ‘mechanisms at work’: social process can be transformative. Archer’s (1995) ‘morphogenetic approach’ to social change captures this: episodes of social interaction effect ‘structural elaboration’.

32. More precisely: upland, protectively designed canal irrigation in South India.

33. This may be a too India-specific and suggest too negative an appreciation of existing water policy processes in the global South. Meijerink and Huitema (2010) is a more optimistic effort at comparative assessment of water sector policy processes; the literature on policy diffusion & transfer and sociotechnical transitions (Geels & Schot, 2007; Marsh & Sharman, 2009) may provide theoretical (and practical) inspiration, though it is largely focused on the global North, possibly implying context-specificity.

34. Ostrom’s (1993) ‘design principles’ for community-based natural resource management are of a similar kind.

35. The search for ‘lessons learnt’, ‘best practices’ and ‘toolboxes’ in international water policy, mentioned above, can be understood as a hope or claim that singular mechanisms can be elevated to general validity.