Chapter 3: Research design
In Chapter 2, the themes that underpin this research were summarised and the reader was orientated to the nature and significance of the research problem through the conducting of a literature review. Briefly, the research problem suggests that the current BI capability in preoperative care does not provide anaesthetists with sufficient actionable insights relevant to their own clinical settings in support of risk-benefit assessment before elective surgery. This chapter is concerned with the research design that was employed in an attempt to solve this problem. This chapter is structured as follows. First, I describe the dominant research paradigms and the rationale for situating this thesis in the interpretive and pragmatist paradigms. This section is followed by an overview of action research (AR), and more specifically participatory action research (PAR) as the strategy of enquiry. Next, I describe the techniques for data collection and analysis used within the PAR process before addressing the ethical considerations inherent in this research. This chapter concludes with a discussion of the limitations of AR and the steps I took to ensure a rigorous approach to this research.
All research is based on a set of beliefs about knowledge and how to acquire it, and about the physical and social world [454, 455]. Chua [456 p. 604] formulates three sets of beliefs that “delineate a way of seeing and researching the world:” (1) beliefs about the notion of knowledge, (2) beliefs about the phenomenon or “object” of study, and (3) beliefs about the relationship between knowledge and the empirical world. Different positions in relation to these beliefs are seen to constitute the worldview or research paradigm that social scientists adopt in the conduct of their research . The researcher must be clear about these underlying beliefs if the research is to be conducted appropriately and evaluated accordingly .
The underlying paradigms for this research are interpretivism and pragmatism. In the subsections that follow, I describe the interpretive and pragmatic paradigms and provide the rationale for combining the two. Chua  and subsequently Orlikowski and Baroudi  and Lincoln and Guba  have provided valuable descriptions of the critical, positivist and post-positivist paradigms as alternative worldviews for IS research.
Interpretive studies assume that our “knowledge of reality is gained only through social constructions such as language, consciousness, shared meanings, documents, tools, and other artefacts” [460 p. 69]. People derive meaning from their interactions with the world around them  and therefore the goals of interpretive research are to understand phenomena through the meanings that people assign to them . Interpretive researchers reject all forms of absolutism, adopting instead a non-deterministic perspective to obtain a relativistic understanding of phenomena through human sense-making [455, 457]. In IS, interpretive studies aim to produce “an understanding of the context of the information system, and the process whereby the information system influences and is influenced by the context” [462 p. 4-5]. Hence, interpretive researchers seek “to understand the context of a phenomenon, since the context is what defines the situation and makes it what it is” [455 p.39]. Other characteristics of interpretive studies include evidence that the phenomenon of interest is examined in its natural setting through prolonged fieldwork and from the perspective of the individuals under study [455, 463]. The interpretive researcher seeks to gain an overview of the research context that is “holistic” (systemic, encompassing, integrated) comprising of its logic, its arrangements and its explicit and implicit rules . More extensive discussions of the nature of interpretive research in IS can be found in Walsham [462, 464-466], Myers  and Klein and Myers .
As an alternative to positivist and anti-positivist positions in the literature (see Baskerville , Goles and Hirschheim  and Wicks and Freeman ), pragmatism holds that both the meaning and truth of an idea is a function of lived experience and practical outcome . It follows that the pragmatist is concerned with “asking the right questions, and getting empirical answers to those questions” [471p. 331]. Pragmatism gives the researcher licence to use “whatever philosophical and/or methodological approach works best for the particular research program under study” [472 p. 5]. Contingent as the approach may be, it does not mean that anything goes. According to Goles and Hirschheim , what works best or is most useful is that which is instrumental in producing desired or anticipated results. Wicks and Freeman elaborate on this point with their definition of usefulness in the pragmatic sense:
Usefulness simply requires those engaged in research or decision-making to scrutinize the practical relevance of a set of ideas as defined by their purposes and those shared by their community (e.g. within a country, a corporation, a research stream). [469 p. 15]
Goldkuhl [473, 474] illustrates the meaning of pragmatism for the field of IS research. He identifies five features of pragmatist studies in IS: (1) there is a focus on actions in empirical enquiries, but also (2) the broader practice context is considered; (3) abstractions that are not grounded in practical reality are avoided; (4) the aim is to develop knowledge that improves actions and practices, requiring that researchers take part in changing and improving the world rather than acting as disinterested observers; and (5) that interpretation is necessary but not sufficient – studies must use pragmatic instruments to evaluate which actions worked and which did not.
Baskerville and Myers , in their introduction to the 2004 special issue on AR in MIS Quarterly, identify pragmatism as the philosophy shared by most forms of AR in IS. Drawing on the works of Peirce , James , Dewey [477, 478] and Mead , they describe the four key AR premises that arise from pragmatist philosophy:
Peirce’s tenet that all human concepts are defined by their consequences; James’ tenet that trust is embodied in practical outcome; Dewey’s logic of controlled inquiry, in which rational thought is interspersed with action and Mead’s tenet that human action is contextualized socially, and human conceptualization is also a social reflection. [471p. 331]
The authors elicit the characteristics of AR in reference to these four premises, further emphasising the pragmatist nature of AR in the field of IS.
More extensive discussions on the subject of pragmatism can be found in Howe , Cherryholmes , Reichardt and Rallis , Murphy  and Rorty . The literature on mixed methods research provides an especially valuable resource (see, e.g., Tashakkori and Teddlie [472, 485]). Specific to IS research, Hirschheim  takes an historical perspective to IS epistemologies, briefly describing the influence of James, Peirce and Mead. The treatment of pragmatism in IS is further described in Goldkuhl  and the combination of interpretivism and pragmatism in IS research is discussed in Marshall .
Rationale for combining interpretivism and pragmatism
Research is a particular form of enquiry, concerned with seeking solutions to problems and answers to questions. According to Gillham [489 p.3], “the raw material of research is evidence, which then has to be made sense of”. Assumptions about what constitutes evidence and how this evidence can be known are represented in the researcher’s choice of paradigm. It follows that the decision to study a topic in a particular way always involves some kind of philosophical stance about what is important .
The choice of research paradigm for this thesis reflected the need to appropriately address the research problem . The research process was emergent and the research problem (and subsequently methods) evolved with the research process (discussed later). As a consequence, the research problem required a balance between intervention and understanding obtained through enquiry, and field study to yield knowledge that was both interesting and constructive (see Goldkuhl ). This thesis contributed to local improvements through intervention, requiring continual exploration and learning with the aim of generating constructive knowledge for everyday clinical practice. However, it also required an appreciation of participants’ meaning-universes and professional languages, as well as the interpretation of social constructs (again see Goldkuhl ). Accordingly, this research is couched in the pragmatist and interpretivist research paradigms.
(Functional and methodological) pragmatism (see Goldkuhl  for a discussion of the types of pragmatism in IS research) is adopted as the base paradigm and elements from the interpretive paradigm are used in a supportive fashion. I draw heavily on Goldkuhl  for much of the rationale for this approach. In his paper on the combination of pragmatism and interpretivism in IS research, Goldkuhl  argues that pragmatist thinking has played an important, though largely implicit, role in ISD research (see, e.g., Braa and Vidgen [492, 493]). The research process is undoubtedly concerned with change, action and reflection, and the knowledge that results from the interplay between these three aspects. As an intervention, ISD is instrumental in co-constructing knowledge that may be useful for local practices as well as general ones. The knowledge forms produced by the research process are not restricted to explanations (as in positivism) or understanding (as in interpretivism). Rather, ISD produces diverse forms of prescriptive (as in guidelines), normative (exhibiting values) and prospective knowledge (suggesting possibilities) . Indeed, the act of developing an IS embodies the very essence of the pragmatist philosophy – to be helpful to the world. Goldkuhl  goes on to identify pragmatism as an appropriate basis for research processes that intervene in the world, rather than merely observe it. It follows that this research – which involved developing a BI prototype to gain actionable insights into perioperative outcomes at the request of local stakeholders – was (appropriately) grounded in pragmatist thinking.
However, IS do not exist in a vacuum; their development shapes and is simultaneously shaped by the social and organisational context . Modern healthcare systems are incredibly complex social and organisational entities. They are characterised by “a high degree of professional specialization, marked division of labour, and an interdependence among relatively autonomous healthcare providers” [495 p.53]. I reasoned that the design and development of the BI prototype would be no less complex because ambiguity and uncertainty are salient aspects the development process . Adopting an interpretive approach towards this research also permitted the discovery, and ultimately revelation, of those organisational, technical, social, professional and therapeutic aspects that shape and are shaped by the course of ISD (and more specifically, BI) in healthcare . This allowed for a focus on the real-life “messy” reality of healthcare work and BI development in action .
At this point it is fitting to state the reasons for not situating this research problem in the (post-)positivist traditions of natural science. I draw on Susman and Evered’s  work for this rationale. The authors summarise the differences between positivist science and AR before considering the question of which approach is better. They identify the positivist paradigm as a poor choice when the unit of analysis is a “self-reflecting subject, when relationships between subjects (actors) are influenced by definitions of the situation, or when the reason for undertaking the research is to solve a problem which the actors have helped to define” [499 p. 600]. This research problem met elements of each of these criteria because it deals with groups of people within an organisation whose characteristics, ideas, strategies and behaviour are complex and difficult, if not impossible, to predict . It follows that the (post-)positivist paradigm was not well suited to the research problem.
The relationship between research problem, objective and methods is an important one. Consideration of this relationship allows the researcher to come to an informed decision about the research design; to decide which method(s) are appropriate for a piece of research; and to anticipate the constraints and limitations that may impinge on any subsequent findings . According to Myers , a research method is a strategy of enquiry that represents a way of finding empirical data about the world. Denzin and Lincoln  provide a comprehensive pedagogy for qualitative research that is consistent with this viewpoint. AR, for example, is regarded as a strategy of enquiry alongside others such as ethnography. Different techniques for data collection and approaches to analysis are available within this method, including the interview, direct observation, analysis of artefacts or documents and personal experience.
I have adopted this view of a research method throughout the thesis. In the following sections, AR, and more specifically PAR, is described as the research method. The rationale for selecting this method is provided. The focus of the chapter then turns to data collection and analysis techniques as part of the PAR process. The chapter concludes with a discussion of rigour in AR and the limitations of the AR method.
AR aims to “solve current practical problems while expanding scientific knowledge” [458 p. 59]. As a general strategy for institutional change, AR has been practised since the 1920s . Development of the AR method is often credited to Kurt Lewin [76, 77], a social and experimental psychologist who combined theory, practice and change to study social psychology within the framework of field theory in the mid-1940s [78-80]. AR had a parallel but independent development which is attributed to the innovative work of psychiatrists, clinical and social psychologists and anthropologists at the Tavistock Group (later the Tavistock Institute) (see Trist and Murray ). The Tavistock Institute developed a method similar to Lewin in the 1950s and 1960s to study psychological and social disorders among soldiers who served in the Second World War . Lewin later joined the Tavistock Institute and the two developments converged .
Lewin pioneered AR as a way of producing “knowledge about a social system while, at the same time, attempting to change it” [503 p. 121] . In his 1946 paper “Action Research and Minority Problems”, he proposed a spiral of steps, “each of which is composed of a circle of planning, action and fact-finding about the result of the action”. Baskerville and Wood-Harper [80 p. 92] observe that Lewin’s original model of AR included iteration of six phased stages:
analysis, (2) fact finding, (3) conceptualisation, (4) planning, (5) implementation of action, and (6) evaluation. Peters and Robinson [504 p. 115], in their paper on the origins and status of AR, present a synopsis of Lewin’s writings on the method. Amalgamating his two works on the subject of AR [77, 505] with those inferences from his more general theoretical writings on social psychology, they summarised the points Lewin made:
The study of social groups and social problems can yield a set of general laws; and one can express these laws as “if/so” propositions and use them in conjunction with a diagnosis of specific situations to plan how to resolve or improve social conditions. One should, in turn, evaluate these strategies by observing their effects and modifying and re-evaluating the strategies if necessary.
Although Lewin’s model has been much adapted, the essence of AR remains unchanged today. Myers , following Elden and Chisholm , claimed that a shared definition of AR has existed for the most part since Lewin’s seminal work. One of the most widely cited definitions of AR is Rapoport’s:
Action research aims to contribute both to the practical concerns of people in an immediate problematic situation and to the goals of social science by joint collaboration within a mutually acceptable ethical framework.[75, p.499]
This definition emphasises the collaborative aspect of AR and dual contribution to practical problem solving and knowledge.
Although there is great diversity in contemporary AR methods, Elden and Chisholm  suggest a minimum of five elements that must be present for the research to be classified as AR. These are as follows:
Purpose and value choice. AR is used to solve the kinds of problems that yield general knowledge as well as practical solutions. AR researchers are not “value neutral”. AR is change oriented and researchers adopt the method to bring about change that has positive social value.
Contextual focus. AR is described as “context bound-inquiry” . The contextual focus is on the content of the problem as well as its solution. The twin focus is necessary to solve “real-world” practical problems which are defined by the people who experience them.
Change-based data and sense-making. AR is change oriented, so the method requires data to track the after-effects of intended changes. AR requires researchers to interpret and make sense from data that are collected systematically over time.
Participation in the research process. AR requires those who experience or “own” the problem under study to be actively involved with the researcher and research process. As a minimum requirement, participants should be involved in selecting the problem and sanctioning the search for solutions. AR is a collaborative endeavour; active interaction is needed to encourage the ongoing cyclical and emergent nature of the process.
Knowledge diffusion. Classical AR assumes that a good solution to a chosen problem will spread by virtue of the value it adds to the existing stock of knowledge. Therefore, the results of the AR process must be written up according to the cannons of accepted social science practice. The researcher should identify the learnings that arose from the process and relate the topic to the existing literature to specify general knowledge.
Peters and Robinson [504 p.120] distinguish between weak and strong forms of AR. Both forms share three characteristics: being change-focused, organic and collaborative. The weak form is relegated to little more than a strategy for “getting things done”, whereas the strong form emphasises the central importance of the participants’ beliefs, values and intentions. In recognition of participants’ agency – that participants have created their own histories but are capable of changing or transforming reality – the strong form is emancipatory. That AR aims to produce “emancipation” is a defining characteristic of several well-known approaches to the AR method (see Kemmis and McTaggart  and Carr and Kemmis ).
Baskerville and Wood-Harper , in their paper on the diversity of AR methods in the IS field, offer an alternative, more pragmatic set of features. The authors define a boundary for IS AR that is characterised by multivariate social settings; interpretive assumptions about observation; intervention by the researcher in the problem setting; participatory observation; and the study of change in social systems. In their comparison of AR forms, the authors identify four types of characteristics that are conducive to comparative analysis. The authors distinguish between: iterative and linear (process); fluid or rigorous (structure) forms; collaborative, facilitative and expert styles of typical researcher involvement; and the primary goals of organisational development, systems design, individual learning and theory development.
It is worth pointing out that there exists disagreement about what constitutes AR in IS. Authors variously define models of AR that are more or less exclusive. This variation leads to confusion about the IS AR tradition . Recognising that there is not one single, definitive AR method, Baskerville and Wood-Harper  appeal to the “vocational” nature of the discipline in their efforts to construct a more inclusive boundary for AR in IS. The authors contend that the lines between action and AR are not (or should not) be so easily drawn in our field. They call attention to the wider debate about inclusion and exclusion of certain IS practices as valid forms of AR. For example, consulting is widely considered to be a false claimant to the AR method. The inclusion of long-established, action-based IS, such as sociotechnical enquiry, is equally problematic because these approaches often do not explicitly claim to adopt the AR method .
It is apparent that AR defies easy description. The reader who reviews the literature in search of an all-encompassing definition, is instead rewarded with an appreciation of the diverse opinions about what is considered to be AR. Baskerville and Wood-Harper, for example, venture AR as something of a shining example of post-positivist research. “It is empirical, yet interpretive. It is experimental, yet multivariate. It is observational, yet interventionist” [79 p. 236]. Dick  describes the AR method as enigmatic. It is organic and emergent, but does not proceed without a plan. It requires the deliberate intervention and active self-involvement of the researcher, but seeks to uphold the principles and values of the problem owners. It is collaborative and inclusive, though the roles of researcher and participant, subject and co-researcher, and the extent to which each is involved in aspects of the process is highly variable. In the almost half century since the term AR was coined, much has changed in the world and the ways in which researchers study the world have changed in response. “It is likely that the ideas behind Lewin’s work remain viable today only because they are being practiced in new ways, in innovative research designs, and applied to new problems” [503 p.121].
PAR in context
PAR is one of several forms of AR that have emerged since the Lewinian definition and is the form of AR adopted in this thesis. According to Baskerville [78 p. 17], PAR is distinguished from earlier forms by increased client participation, such that “co-researcher status” is accorded to participants belonging to the client organisation . The roles of researcher and participant are emergent, less distinct, and take on more collaborative and synergistic forms This observation is in keeping with Lewin’s proposition that “causal inferences about the behaviour of human beings are more likely to be valid and enactable when the human beings in question participate in building and testing them” [510 p. 86].
In PAR, client participants are “actively engaged in the quest for information and ideas to guide their future actions” [6 p. 20], and share in the responsibility for theorising . To fulfil this responsibility, researchers and clients inform the research process in different ways; each comes with their own body of theoretical knowledge. According to Baskerville  and Elden and Chisholm , researchers draw on their knowledge of the AR method and general theories (in IS) while client participants introduce situated, practical knowledge into the research process. It follows that the social setting is realigned: “free to self-reorganize rather than be artificially determined by the external researchers” [78 p. 17]. By this token, PAR is founded on the assumption that reality is socially constructed  and social systems are self-referencing .
Whyte and colleagues’ work on PAR is noteworthy. The authors argue for the scientific and practical value of PAR. In one of their earliest works, they define PAR as an approach in which:
Some of the people in the organization or community under study participate actively with the professional researcher throughout the research process from the initial design to the final presentation of results and discussion of their action implications. [6, p.20]
They illustrate the PAR process with three case studies: one with Xerox Corporation in New York State, another with the FAGOR group of the Mondragon cooperatives in Spain (see Whyte, Greenwood and Lazes ), and Thorsrud’s work  in the Norwegian shipping industry (see Whyte ). A variety of treatments of the subject of PAR can also be found in Whyte .
Later, Greenwood, Whyte and Harkavy  published on the subject of PAR, both as a social science research method and process, and the goal of such research. The authors described the key features of PAR (Table 14) and presented the West Philadelphia schools, Xerox Corporation and Mondragon cooperative cases to argue for the emergent nature of PAR and its inherent ability to link participation, social action and knowledge generation. Following this, Elden and Chisholm  presented Greenwood and colleagues’ paper among a collection of reports on novel variants on contemporary AR in their introduction to the special issue on AR in Human Relations. In this introduction they identify several refinements Greenwood and colleagues contributed in relation to the idea of participation in AR. First, AR is emergent; none of their three cases in  began as PAR. Hence, all AR should strive to be more participatory because enhanced participation is required to improve the scientific quality of the research results. Perhaps most significantly, Elden and Chisholm draw attention to the authors’ point that participant-managed research ought to be viewed as a matter of degree, not dogma. The nature of participation in AR is a product of the local situation and full participation in all phases of the research is seldom feasible.
In this thesis, therefore, I adopt the description of PAR given by Greenwood, Whyte and Harkavy. According to the authors, PAR is:
a form of AR in which professional social researchers operate as full collaborators with members of organizations in studying and transforming those organizations. It is an on-going organizational learning process, a research approach that emphasizes co-learning, participation, and organizational transformation. [81 p. 177]
This definition was chosen because it is prevalent in the business and management literature, it reflects the need for researchers to become deeply involved with the organisation, and, above all, it represents the working relationship between participants as a partnership. Now that this subsection has identified and defined PAR as the research method, and justified this choice, the subsections that follow move to address the structure of the PAR process as it emerged to address the research problem.
Table of contents
List of tables
List of figures
Glossary of terms
List of abbreviations
Chapter 1 : Introduction
1.1 Setting the scene: Health, healthcare and medical informatics
1.2 Elective surgery in NZ
1.3 Background literature
1.4 Overview of the research problem
1.5 Overview of the research design
1.6 The research context and the action researcher
1.7 The structure of the thesis
Chapter 2 : Literature review
2.1 Preoperative anaesthetic assessment of the surgical patient
2.2 Preoperative risk assessment
2.3 A framework for risk assessment
2.4 Patient risk factors
2.5 Surgical risk factors
2.6 Risk scoring systems
2.7 Evidence-based preoperative risk assessment
2.8 Closing the research-practice gap: Making the case for RCD in the learning healthcare system
2.9 Enabling the LHS through BI
Chapter 3 : Research design
3.2 Research paradigm
3.3 Research method
3.4 PAR and the Multiview methodology
3.5 Data collection techniques
3.6 Data analysis techniques
3.7 Ethical considerations, limitations and rigour
Chapter 4 : Developing a BI prototype
4.2 Organisational analysis
4.3 Information system modelling
4.4 Software development
4.5 Feedback on the BI prototype
Chapter 5 : Exploring the work of risk communication at AAC
Chapter 6 : Perioperative mortality among patients undergoing non-cardiac surgery: A retrospective analysis of RCD
6.3 Materials and methods
Chapter 7 : Conclusions
7.2 Reflecting on the research problem and objectives
7.3 Drawing the findings and conclusions together
7.4 Research limitations
7.5 Contribution and implications
7.6 Concluding remarks
GET THE COMPLETE PROJECT
Towards business intelligence in preoperative care: Choice, chance and communication