Chapter 3: Theoretical framework and methodology
3.6 Data analysis
Analysis of qualitative data refers to searching for meaning through interpreting the views and behaviours of the participants. However, it is challenging to record the process thoroughly. As Bryman put it, “unlike the analysis of quantitative data, there are few generally agreed rules of thumb for the analysis of qualitative material” (1989, p. 166). In general, Miles and Huberman (1994) suggest three types of activity in data analysis: (1) data reduction, (2) data display, and (3) conclusion drawing and verification. Miles and Huberman portray qualitative data analysis as an iterative process, consisting of the action of data collection and the above-mentioned three forms of activities in data analysis. Creswell presents the process of data analysis for the case study, including several phases as below (2007, pp. 156–157):
Data managing: Create and organise files for data;
Reading, memoing: Read through text, make margin notes, form initial codes;
Describing: Describe the case and its context;
Classifying: Use categorical aggregation to establish themes or patterns; Interpreting: Use direct interpretation; develop naturalistic
generalisations;
Representing, visualising: Present in-depth picture of the case (or cases) using narrative, tables, and figures.
The data analysis process for this study followed a similar set of steps to those recommended by Creswell (2007). First, the interviews were recorded and transcribed.
Data of each company, including interview transcript and documentary data, were put into an individual file folder (both electronic files and printed hard copies).
Second, initial data analysis included reading the interview transcripts and related documents, and sorting out the data. The researcher made margin notes, while noticing particular themes. The data analysis was facilitated by a coding list (See Appendix 3). The coding of this study involved three main steps: generating initial codes; collating data relevant to each code, revising the codes or creating new codes; and searching for themes. The initial coding categories were developed based on the theoretical framework and the three research questions. The codes were divided into three tiers. While reviewing the data collected, the researcher started to place the collected data into general and, subsequently, into more specific categories. For example, one of the questions posed to all participants was the sources of competitive advantage of the company. The transcripts were initially coded under the tier one category, ‘Source of competitive advantage’. When all the relevant transcripts of the participants had been placed under this category, they were further analysed. This process was to determine what sub-categories (tier two) might be identified from this broad category, for example, ‘Resource advantage’. Furthermore, the transcripts in this category were further placed into the tier three categories—‘Markets’, ‘Internal’, ‘Inter-firm’ and ‘Others’ respectively—as the characteristics had been distinguished from the data. While re-examining the tier three category, ‘Markets’, the themes emerged were ‘human resources’ and ‘financial capital’. During the process of coding, the transcripts were coded by using the initial set of codes at the beginning. Then, a new code might be created, or an existing code modified, if needed. The coding list was finalised when all the transcripts were coded by the tier three codes.
Third, the next step was to analyse the case companies individually. As Stake put it, “our first obligation is to understand this one case” (1995, p. 4). The attention of within-case analysis is put on the particularities and complexity of each case. To achieve this goal, a case summary of each case company of this study was made. During the process of data analysis, the most important thing was to identify any emergent theme that could be linked to the research questions, or could potentially contribute new insights to the subject area.
Fourth, the following work involved categorising and organising emerged themes. For example, Alpha’s engagement in customers, by continuous efforts in product development and customer services, was a key factor of its source of competitive advantage, which could differentiate itself from its rivals. The huge capital commitments by Epsilon and Zeta created significant cost advantages, in terms of economies of scale, of the production of large TFT-LCD panels.
Fifth, as this thesis is a qualitative, multiple-case study, each individual case was a part of the whole study. Accordingly, the subsequent step was to draw cross-case conclusions. The conclusions drawn from each case would then be considered as the base of supporting evidence for replication in other cases. To search for cross-case patterns, Eisenhardt (1989) recommends three strategies: (1) the aspects suggested by the theoretical framework or current research questions were identified and cross-case similarities and differences were acknowledged; (2) similarities and differences of selected cases were displayed; (3) data gathered from different sources had been compared and prioritised in order to determine which patterns were more significant than others. For example, the isolating mechanisms that could preserve competitive advantage, including time compression diseconomies, causal ambiguity, social
complexity, and transaction costs (e.g., Dierickx & Cool, 1989; Jones, 1995; Reed & Defillippi, 1990) were identified from the literature. In the cross-case analyses, three themes emerged: technological and manufacturing capacities, environmental investments, and internationalisation, which were more significant than other themes and could be used to elaborate on and support time compression diseconomies.
The last step was to present the arguments of the research by using tables or figures. The purpose of quantitative analysis is to identify or discover conceptualisations of pattern, structure and meaning from the empirical data (Patton, 2002; Strauss & Corbin, 2008). Thus, in the key chapters of this study (Chapters 4, 5, 6 and 7), tables or figures were used to summarise and illuminate the important themes or concepts resulted from the empirical findings.