• Tidak ada hasil yang ditemukan

Integration framework

Glossary of Terms

Chapter 4: Methodology

4.3 Integration framework

119 Kiribati has been the target of a number of aid projects (in particular the SAPHE and KAP 2 projects), administered by the Asian Development Bank and the Australian government’s aid organisation AusAID among others. Taken all together, there should be considerable opportunity to gain access to stakeholders, data and reports.

120

• Identify a transition pathway towards a more equitable and better adapted water governance system; and

• Improve the adaptive capacities of local institutions to coordinate flexible programmes of action.

These are considerable challenges that will involve institutional or even societal transformations. As argued by Pahl-Wostl (2002a), this requires social learning and therefore the human dimension is crucial. Building on the experiences of Kurtz and Snowden (2003) within action research; three pervasive assumptions are relaxed about individuals and organisations decision making, that are common in decision making and policy formulation:

The assumption of order: that there are underlying relationships between cause and effect in human interactions and markets, which are capable of discovery and empirical verification. In consequence, it is possible to produce prescriptive and predictive models and design interventions that allow us to achieve goals. This implies that an understanding of the causal links in past behavior allows us to define ‘best practice’ for future behavior. It also implies that there must be a right or ideal way of doing things.

• The assumption of rational choice: that faced with a choice between one or more alternatives, human actors will make a ‘rational’ decision based on minimizing pain or maximizing pleasure; and, in consequence, their individual and collective behavior can be managed by manipulation of pain or pleasure outcomes and through education to make those consequences evident.

• The assumption of intentional capability: that the acquisition of capability indicates an intention to use that capability, and, as a consequence, actions from individuals or communities are always the result of intentional behavior. In effect, it assumes that every

‘blink’ we see is a ‘wink’. In contrast, we assume that people happen to do things by accident. (Kurtz and Snowden 2003: 462-463).

This does not mean that these assumptions are never true; but rather that they are not universally true and that the representation of reality is not simplified by assuming that they are.

121 Furthermore, in this study, a post-normal stance is adopted by acknowledging that in complex situations with high stakes there is a need to accept multiple plausible realities; and incorporate the review and social validation process with affected stakeholders as part of the research process.

In fact, as was found in the AtollGame experiences, the same individual may be acting according to conflicting mental models of reality depending on the context (Dray et al. 2007).

Additionally, like Dray and colleagues (2006a; 2006b; 2007) it is believed that individuals’

representations of reality (mental models) are constructed in interaction with their physical and social environments; and that as per Becu and colleagues (2003), those mental models can be elicited and translated into conceptual models using knowledge engineering. It has also been described how such mental models, including multiple and conflicting mental models, can feed into computational models, i.e. Agent Based Models (ABMs) and Bayesian Networks (BNs).

Based on the relaxation of the described assumptions about order, rational choice and intentional capability, the Cynefin framework prescribes a number of activities for sense- making in what they refer to as the un-ordered domain of complex relationships in which patterns can be perceived but not predicted, and these are:

• Analysis of history as a way to understand systemic properties. This will help us identify and respond to general patterns, and make us prepared for any unexpected patterns;

• Use of narrative techniques. This will help to generate a richer picture of the diversity and will allow us to capture important aspects which can not be easily formalised in numbers and theories;

• Use of multiple perspectives on the nature of the system. In this way it is possible not only to identify the multiple perceptions of reality (i.e. mental models), but also to identify a greater diversity of patterns that occur in the system;

• Exploratory analysis in order to temporarily move to a situation where cause and effect relationships are discoverable. While patterns may not be predictable, they are usually

122 understandable using hindsight. This will help us to understand the range of possibilities, and in some circumstances to probabilistically assign more or less likely future scenarios;

• Use of an adaptive approach, where interventions are designed as probes, and analysed in retrospect. This will help us understand how to promote desirable patterns of behaviour and destabilise undesirable patterns.

It turns out that it is possible to map these sense-making activities against our reviewed methods, as per Table 4-2. This mapping forms the basis for our integration framework.

Table 4-2: Methodological guidance based on the Cynefin framework

Cynefin sense making activity Method

Analysis of history as a way to understand systemic

properties Historical review of previous experiences

Use of narrative techniques Survey/interviews

Use of multiple perspectives on the nature of the system Delphi survey Exploratory analysis in order to temporarily move to a

situation where cause and effect relationships are discoverable

Bayesian Networks Use of an adaptive approach, where interventions are

designed as probes, and analysed in retrospect in order to understand how to promote desirable patterns of

behaviour

Agent Based Modelling

This selection of methodologies addresses the fact that the problem of water supply and sanitation in small towns in PICs in reality is a messy problem, and that it is difficult to establish which of the Cynefin domains that this problem spans; or in other words, to what extent is the problem in the known, knowable, complex or chaotic domains.

In the known domain, literature and simple analysis will suffice in order to provide explanation of the system – this domain is covered by the literature review and Delphi survey to explore expert knowledge, predetermined practice, field guides and manuals etc.

123 In the knowable domain there exist stable cause and effect relationships although they may not be fully known to all actors. Systems thinking on the basis of interviews, historical review and case study review feeding into Agent Based Modelling allow exploration of this domain.

In the complex domain, the domain of complexity theory, patterns emerge and can be perceived but not predicted. These may or may not be stable, and therefore structured methods that seize upon such patterns are likely to eventually become ineffective. In this domain, Bayesian Networks that establish probabilistic causal links based on retrospectively identified patterns have a use. However such modelling needs to be done in acknowledgment that patterns may change (and hence may consider input from people who can sense more current system dynamics).

In the chaotic domain, small and seemingly random stimulus or disturbances may create large changes in systems dynamics. This generates a level of uncertainty and randomness. Whilst interviews may provide some insights into those largely undefined things that may impact on the system, the Bayesian Networks to some extent deals with the uncertainty of this domain.

More importantly, this domain is dealt with by stakeholders using strategies such as control;

uncertainty reduction; or shifting the domain. As a researcher, engaging with stakeholders who are critical actors in the system is a way of acknowledging this domain.

Synthesising the above, the Historical review of the case study water sector will provide the setting, and will allow us to identify patterns which have been observed in the past, providing what is referred to as retrospective coherence. It will also allow for continuity in an on-going learning experience. The interviews will provide an interrogation technique which is not bound by traditional scientific constraints, but will allow for gaining a more integrated and rich understanding of issues and factors. The Delphi survey will allow for formally collecting perspectives and narratives of diverse stakeholders and experienced individuals in the wider Pacific Islands region, and thereby allowing for making wider generalisations. Analysis based on BNs will allow for, at least temporarily, understanding the cause and effect relationships in water development interventions which are often both complex and uncertain. Systems analysis

124 using an ABM will allow for exploring how probes can be designed and explored using scenario analysis and the impacts can be evaluated from a number of different perspectives. This will help the understanding of how to promote desirable patterns of behaviour in the system.

The sequence of these activities also forms a natural sequence in terms of generating the data and understanding required for developing models. In other words, the historical review, interviews and the Delphi survey provide a rich and diverse information base to feed into the modelling exercises. This approach also allows the researcher to develop an understanding of the context in consideration of the steep learning curve that anyone would have that is unfamiliar with the Kiribati culture and context.

Additionally, given the remote location and the practical/financial difficulties in spending long periods of time at the case study location, the Delphi survey will allow knowledge elicitation to be carried out remotely via email. This will also allow for more general information about water supply issues in small towns in the Pacific Islands.

It is also noted that in reality, it is and has been important to iteratively respond to local circumstances and feedback, which means that methodological decisions are made during the research process. In other words, a linear sequence of activities is deceptive in that in reality these events do not follow in quite such a sequential order. The over-arching iterative procedure that is applied in this research is shown in Figure 4-2:

• Reflection: using inductive and deductive reasoning, to identify patterns and understand observations and data. An important part of the reflection is to raise new questions as well as to discuss/debate with colleagues and other relevant individuals.

• Field study: based on the improved understanding and new questions that have emerged, further ideas and information is collected from the field. This also includes evaluation by local stakeholders (i.e. water managers) of existing models or frameworks in terms of their appropriateness in light of field realities.

125

• Modelling: based on new evidence and questions, models are developed to allow further exploration as well as probing in a virtual laboratory. This can also be seen as theory development and hypothesising.

Figure 4-2: Research process as iteration between reflection, field study and modelling

Finally, it is worth noting that the goal of this study is not action but rather to develop a management framework and a transition strategy. Therefore, the problem definition is somewhat diffuse and complex; and implementing solutions are likely to be strongly dependent on political support. Because of this high level of uncertainty, it is deemed that having action on the ground as the aim is a risky strategy because research success would be dependent on political outcomes; and may generate unrealistic expectations and is therefore potentially unethical and therefore poor engagement practice.

126