Research strategy

3.3.1 Action research

The iterative approach taken in the project is analogous with the cyclic pattern of action research. In action research, initial reflection on a problem is developed into planning, which must, from a critical realist perspective, emerge from a concept of how the planned action would affect the studied structures and mechanism of the world. At this stage, the plan might be naïve, but putting it into action creates the opportunity for empirical observation and assessment of its effects. These observations can be followed by more informed reflection, which in turn can refine the plan (Figure 6).

Figure 6: Action research cycle, after Zuber-Skerritt (2001)

This action research process should be understood as an ‘orientation to inquiry rather than as a methodology’ (Reason and McArdle, 2004). It has evolved out of the work of C. S. Peirce, John Dewey and Kurt Lewin (Barton et al., 2009), and has been transformed and applied in many different guises; its definition is not static or absolute (Altrichter et al., 2002; Zuber-Skerritt, 2001). The terminology and the way that the sequence of activities are gathered under different headings vary between authors, but in principle, action researchers advocate learning by doing (Winter, 1989). Ideas and hypotheses are tested by ‘change experiments’, preferably in the context that they aim to influence (Barton et al., 2009). The researcher is involved in the subject of study, rather than attempting to act as a neutral bystander, free of values (Perry and Zuber-Skerritt, 1992). There is often a social motivation to an action research strategy, in which

‘researchers work with and for people’, co-producing knowledge, ‘as opposed to simply undertaking research about them’ (Fahy and Davies, 2007). Reason and Bradbury (2001) emphasise the ‘pursuit of practical solutions to issues of pressing concern to people’. These moral aspects may be shaped by social scientists’ suspicion of research into people’s behaviour that is produced exclusively in academic institutions, at a distance from their subjects (and in a dominantly positivist paradigm) (Reason and McArdle, 2004). Such researchers’ goal is for power to be shared between researcher and active participants, rather than held tightly in the hands of the academic.

Since Lewin’s (1946) seminal paper introduced the term ‘action research’ as a means of investigating intergroup social relations, action research has been used in graduate management research (Perry and Zuber-Skerritt, 1992), intra-company problem solving (Zuber-Skerritt, 2001), design practice (Swann, 2002) and in many other fields (Kemmis et al., 2013: 4).

There are some examples in waste research: Fahy and Davies (2007) and Farrelly and Tucker (2014) use action research in addressing household waste minimisation. They note that household waste behaviour has been investigated primarily through quantitative methods, and adopt action research approaches to (a) add depth and nuance to knowledge of waste behaviour, and (b) encourage individuals to take up pro-environmental behaviour, on the basis that providing people with information alone is often ineffective. A similar approach could be envisaged in C&D waste management, collaborating with individual construction workers to understand behaviour and encourage waste minimisation and segregation. However, this may struggle to achieve widespread impact due to structural impediments beyond the individual’s control. In a conference paper reporting on the early stages of an action research project on urban C&D waste management systems, Aid and Brandt (2010) note the conflict between an individual’s agency and large-scale system change. Their project participants proposed a multi-stakeholder process to catalyse change, but unfortunately any subsequent outcomes of this strategy have not been published.

The present project look for ways to improve practice, so it adopts a strategy inspired by the action research connection of practice and theory; but it recognises the limitations of working with individuals on complex systemic problems. Various aspects of the work were carried out with the involvement of practitioners. However, unlike action research undertaken in purer social science settings, the primary intention of this engagement was not to bring about immediate change within an organisation or group of people. The emphasis, instead, was on learning about practice, by oscillating between action and reflection to develop increasingly thorough descriptions of structures, mechanisms and possible interventions (Acaroglu, 2014: 8-9). The goal is to describe a context for wider, longer-term change, outside the scope of the research project; that is, more on the ‘research’ than the ‘action’ (Dick, 1995, cited in Swann, 2002). Towards this end, action research provided, firstly, a process of learning from the way that individual organisations relate to the context and issues of C&D waste management; and

secondly, of investigating how they might relate to proposed interventions. In so doing, it leads to suggestions of how a context for change could be conceived and implemented.

3.3.2 Systems thinking and systems engineering

Frequent reference is made in the thesis to ‘systems’. Systems are understood as cohesive sets of interacting elements, in the line of thinking advanced by Ludwig von Bertalanffy (1968) in biology; but they can be natural or human-made. Depending on the lens through which they are considered, systems can be seen at various scales. A system may be decomposed into a collection of sub-systems, and may in turn be integrated within a higher-level system (INCOSE, 2017). To say that something behaves like a system is not to claim that it can be reduced to this depiction, or deny its interactions with other elements outside of the imposed system boundary (Checkland, 1983). Systems thinking emphasises the complex interactions between systems, and emergent behaviour from these multi-scalar entities (Flood, 2010). This is analogous to the structures and generative mechanisms that produce patterns of behaviour in critical realist analysis. A critical realist paradigm and systems thinking both encourage non-reductionist exploration of the complexity and multi-disciplinarity of open systems (Easton, 2010; Flood, 2010; INCOSE, 2017). Positivist research, by contrast, must create artificially closed systems to achieve the conditions necessary for deductive inference (Barton et al., 2009).

Given the indivisible and indefinite nature of an open systems view of the world, every piece of research requires some demarcation of a boundary around the studied phenomena. In systems thinking, the researcher’s judgement is called upon to define what really comprises a phenomenon and what is relevant in its context (Winter, 1989: 47-48). The related field of systems engineering provided useful tools for organising thought on the demarcation of the urban- and product-systems studied in this project. Systems engineering begins from a problem statement and a set of requirements to be met by the system design (INCOSE, 2017).

Although systems engineers are often concerned with ‘hard’ systems (where a boundary and system requirements can be more definitively expressed, e.g., a piece of software to meet a company’s needs or a transport system to meet urban needs), the model proposed by Martin (2004) is not specific to any one discipline, and its flexibility makes it applicable to this project.

His ‘seven samurai’ model (Figure 7) proposes a means of considering all aspects of system requirements holistically, to minimise the risk of misconceiving needs and aim at reaching the

‘best’ solution to the problem. The ‘intervention system’ is the primary system to be designed, but modelling the six connected systems attempts to ensure the design’s effectiveness in the given context as well as successful adaptation to a changing context.

Figure 7: ‘Seven samurai’ systems engineering model, holistic view, adapted from Martin (2004)

The design of an intervention begins with an understanding of the context system (S1), within which is the identified problem (P1). To bring the intervention system (S2) to bear on the problem requires a realisation system (S3), which consists of all the tangible and intangible resources needed to conceive, develop, produce, test and deploy the intervention. Martin (2012) gives an example: the problem of needing to transport people over long distances, addressed by the intervention of the passenger jet, with a realisation system comprising jet engine manufacturers, airline companies, airports, safety standards, and so on. The realisation system may be in the form of an enterprise, and must be responsive to the nature of the context system.

The goal of the intervention is to solve the identified problem and thus transform the existing context system into the sought context system (S1’). When put into action, the intervention system becomes a deployed system (S4). The deployed system may differ from the design of the intervention system, and it may or may not achieve the intended changes; it may have unintended consequences that manifest as new problems (P2) in a modified context system (S1”). The context may have changed because of the passage of time, or because of its interactions with the deployed system. Martin (2004) did not distinguish between what are here termed S1’ and S1”, but doing so makes the goal of the intervention explicit and contestable, and in the fullness of time, allows the gap between design intent and reality to be assessed.

In most scenarios, a deployed system will not address the problem by itself, but will have associated collaborating systems (S5) that will also interact with the context. To maintain the running of the realisation system requires a sustainment system (S6), the limitations of which

could undermine the running of the deployed system if not considered at the outset. Finally, there may be competing systems (S7) that aim to solve the same problem in a different way, and may conflict with the deployed system. By highlighting the interactions between all of these systems, Martin’s model aims to stimulate systems engineers who focus ‘too much […] on the intervention system’ to acknowledge the complex, adaptive nature of their work (Martin, 2012).

The model is intended to ensure that the systems engineering process is verified (‘building the right system’) and validated (‘building the system right’).

In adopting this framework, it is necessary to qualify its use. In Martin’s work there is an implicit belief that a problem’s context is knowable and that a systems engineer will be able to diagnose needs and respond with ‘correct’ decisions, i.e., that they can grasp the structures and generative mechanisms that lead to observable phenomena and react objectively and impartially. In this project, the fallibility of perception and interpretation is acknowledged.

However, the pragmatist approach to the study accepts Martin’s tool as an effective means to an end: it presents boundaries within which to formulate ideas about an existing and envisaged context, and to describe the various facets of interventions intended to engender a transition.

In summary, action research provided a guiding strategy for the sequences of research activity, and systems engineering provided a strategy for conceptualising the outcomes of this activity and formulating interventions. Both strategies are connected to design processes in their iterations of activity between concept and detail, between thought and action, and between analysis of how things ‘are’ and how they ‘ought to be’ (Hevner et al., 2004). Designers’

knowledge can be said to be primarily of the artificial world – the human-made world of artefacts, and changes or additions to it (Cross, 2001) – this is different to knowledge of the natural world. The normative is more inherent in the artificial; the question of how things ought to be cannot be avoided or considered outside the designer’s remit, and the designer cannot remain a detached observer. The author’s experience as a designer-practitioner implies a

‘designerly’ way of knowing, and a strategy that begins from, and cyclically returns to, a normative vision of a modified construction industry context. The research strategy thus cycles between description of problems within an existing context (observation and reflection), proposal of alternative scenarios in which problems are solved (planning), and provisional testing of intervention and realisation systems needed to achieve the change (action and observation).