CHAPTER 4 – METHODOLOGY
4.3 Methodological soundness
To judge validity and reliability of qualitative research within the realism paradigm, Healy and Perry (2000) draw on the ontological, epistemological and methodological
elements of the paradigm to propose six criteria. Those being; ontological appropriateness, contingent validity, triangulation, methodological trustworthiness, analytic generalisation, and construct validity. The criteria have been strongly supported in the literature as appropriate for the realism paradigm (see Golafshani, 2003; Alam, 2005). This chapter now proceeds with a defence of the realist paradigm used within this study.
4.3.1 Ontological appropriateness
Given the primary objective of the study to investigate the presence of a proposed generative mechanism and to account for the contingent conditions associated with its operation and/or suppression, the world being investigated can be categorized as being Popper’s world 328 (Magee, 1975) in which the world “consists of abstract things that are born of people's minds but exist independently of any one person” (Healy and Perry, 2000, p. 120). Therefore, it is not the researcher’s perceptions that are the explicit focus of the research, but rather their ability to access a reality that lies beyond the researcher’s perceptions (Stake, 1995). As such, the ontological position (or realism) adopted in this study is preferred to positivism (that operates in Popper’s world 1 and requires a more objective approach) and/or constructivism (that operates in Popper’s world 2 using more subjective analysis). Transcendental realism is used to ensure that the potential operation (and/or suppression) of the generative mechanism is not only born from the researcher’s
28 Magee (1975) describes Popper’s three worlds in the following manner; World 1 is objective and
contains material things (therefore it is positivist), World 2 is born from the subjective mind (therefore it is based on constructivism), and World 3 is related to abstract things born from people’s minds, but which
mind (e.g. transcendental idealism), but is also empirically confirmed or disconfirmed. The ontological position adopted is therefore appropriate.
4.3.2 Contingent validity
Pawson and Tilley (1997, p. 150) argue that “unlike the laboratory where the conditions for effective triggering can be created, no such opportunity exists in the social world”. This being especially true when the events under investigation have already occurred, and therefore subject to an outcomes-based explanation (see Mahoney, 2003). Therefore the challenge is to identify the contingent conditions that relate causally to the mechanism under investigation. Through the identification of different types of towns (or environs) and time periods within which the process of Transferred Demand may be activated and/or suppressed, this research has achieved contingent validity.
4.3.3 Triangulation
Healy and Perry (2000, p. 123) argue that “realism relies upon multiple perceptions about a single reality” to access a window of reality that is triangulated. Figure 4.7 illustrates the integration of forms of data (i.e. qualitative and quantitative) and methods of analysis (see page 141). Denzin (1970) identifies four types of triangulation; those being data triangulation, investigator triangulation, theoretical triangulation, and methodological triangulation. Across both phases of this study, data has been gathered in five independent pizza marketplaces across a period of time spanning the early 1970s to 2004. No significant dissimilarities have been observed across the data collected vis-à-vis the process of industry development and the processes of firm survival. The variety of data
gathering strategies employed (see Figure 4.7, over page) provide evidence of data triangulation. Next, whilst the nature of this study precludes the use of more than one researcher to gather and interpret data, care has been taken to consult with colleagues from the broader ecological community to ensure consistent application of ecology theory, and to gain access to the process of ecological thinking as it is germane to the nature of this study. Thus, a degree of investigator triangulation has been achieved.
The process of retroduction requires the researcher to engage in the process of theoretical redescription. Within this study, aspects of the process of firm survival have been viewed from both an organizational studies perspective and also contrasted against multiple ecological perspectives, and therefore, it is claimed that solid theoretical triangulation
has been achieved. Lastly, a variety of data collection methods have been employed to advance the researcher’s capacity to understand what’s happening in between the evolutionary processes of selection and adaptation. Thus, it is also claimed that strong
Figure 4.7 – Triangulation Through Mixed-Methods
4.3.4 Methodological trustworthiness
Quantitative Data Qualitative Data Research Method
Postulate 1 Yellow Pages data
Data from semi-structured and informal interviews and from researcher observations
Noticing-Collecting-Thinking (NCT) and Correlations
Postulate 2 Yellow Pages data Data from semi-structured and informal interviews and from researcher observations
NCT
Postulate 3 Yellow Pages data
Data from semi-structured and informal interviews and from researcher observations
NCT and Canonical Discriminant Analysis (CDA)
Postulate 4 Yellow Pages data and Regional statistics NCT, Diversity Indices and (CDA)
Postulate 6 Yellow Pages data and Regional statistics NCT, Paired Samples T-Test
Postulate 5 Yellow Pages data and Regional statistics NCT, CDA, Community Similarity Index (CSI) and Survival Analysis
Postulate 7 Yellow Pages data and Regional statistics Data from semi-structured and informal interviews and from researcher observations
NCT, Survival Analysis and Advertising Efficiency Index (AEI)
Postulate 8 Yellow Pages data and Regional statistics
Data from semi-structured and informal interviews and from researcher observations
Abundance Ranking, Diversity Indices, and CDA
Postulate 9 Data from semi-structured and informal interviews and from researcher observations
NCT
Postulate 10
Yellow Pages data and
Regional statistics Data from semi-structured and informal interviews and from researcher observations
NCT, Survival Analysis and AEI
Postulate 11
Yellow Pages data NCT, Survival Analysis
Postulate 12 Data from semi-structured and informal interviews and from researcher observations
NCT
Postulate 13
Data from semi-structured and informal interviews and from researcher observations
NCT
Postulate 14 Data from semi-structured and informal interviews and from researcher observations
NCT
Postulate 15
Yellow Pages data and Regional statistics
Data from semi-structured and informal interviews and from researcher observations
NCT, Survival Analysis, CDA and Correlations
Postulate 16 Yellow Pages data and Regional statistics Data from semi-structured and informal interviews and from researcher observations
NCT, Survival Analysis and Diversity Indices
4.3.4 Methodological trustworthiness
Throughout the data collection process in both phases of the study, interviews have been transcribed and notes have been written up from informal conversations (see Appendix 9 for examples). In the first phase of the study, an account of the Hobart pizza industry’s growth and related firm survival was developed (see Appendix 5) and made available to those interviewed. This account has also been published within a journal article related to the first phase of the study (see Jones, 2007). In the second phase this level of feedback was not necessary as the aim was to confirm the similarity of the industry’s development and identify the contingent conditions under which the proposed process of Transferred Demand might (or might not) exist, rather than collect an in-depth account of the industry’s history. Nevertheless, throughout the study, the same methods of data collection and analysis have been used to ensure consistency (Lincoln and Guba, 1985).
4.3.5 Analytic generalization
Given the underlying logic of realism as theory-building rather than theory testing (Healy and Perry, 2000), the emphasis is on analytical generalization, rather than on empirical generalization. The process of case selection is driven by the need to “clarify the necessary and contingent relationships between structures” (Danermark et al., 2002, p. 105). Further, it is not the aim of the study to understand the nature of a specific reality, one that may not allow for prediction (see Danermark, 2002, p. 63). Hence, the particular attention paid to spatial and temporal boundaries to ensure access to the same embryonic processes unfolding at the beginning of the pizza industries in Hobart, North-West and North-East Tasmania, Geelong and within Yorkshire. So the focus within this study has
been upon theory building across several interrelated cases, rather than engaged in any form of activity from which statistical generalization is required. Thus the processes outline above demonstrates analytical generalization.
4.3.6 Construct validity
Within the context of case study research, construct validity is a contentious issue (see Yin, 2003). Yin claims that the use of subjective judgements and weak operational measures are frequently used as reasons to lessen claims of construct validity in case study research. Within this study, the fear of subjective judgements have been nullified by the used of a mix-method approach that has allowed multiple forms of data to wash over the proposed concept of Transferred Demand. Further, a very high degree of field utility, resonance and parsimony (see Gerring, 2001) have been demonstrated for the concept of Transferred Demand throughout this study. That is, with regards to field utility, the only concept that is similar in operation is an economic spillover, but this relates to an externality to which the beneficiaries are difficult to isolate. Whereas, the beneficiaries of Transferred Demand are definable, as are the conditions that relate to its operation.
As the study progressed, formal interviews and informal conversations continually demonstrated resonance for the concept of Transferred Demand. It is intuitively clear how local independent pizza firms benefit from increased consumer demand being transferred to their operations via the behaviour of other (franchised) firms. As such, the term itself achieves a high degree of parsimony in that it clearly describes a specific
process (i.e. the transfer of consumer demand). Lastly, it is worth noting that while this study has made significant progress towards developing construct validity around the concept of Transferred Demand, at this stage, the outcome of the study is the development of a model that requires more empirical testing. Therefore, at this point in time, the model of Transferred Demand is potentially an emergent theory that will develop greater construct validity in its immediate future development.
4.3.7 Summary
The methodology used in this study uses a realist epistemology (i.e. transcendental realism) which emphasises the philosophical importance of theory development. The specific research method used is Bhaskar’s (1979) process or retroduction which was used to allow the researcher to access a reality that is likely to be not directly obvious in the domain of the empirical reality. Cases were selected on the basis of their potential for analytical generalization. A mixed-method approach was used to gain access to more forms of data from which to describe the events under investigation. A combination of quantitative (i.e. Yellow Pages directory data and Regional statistics) and qualitative (i.e. semi-structured interviews, informal conversations and research observations) provided the impetus towards understanding the complexity of events investigated across five different, yet similar landscapes. The data was also analysed using several ecology-based methods. The method is defended on the basis of Healy and Perry (2000) criteria for assessing the validity and reliability of research undertaken within the realism paradigm.