Evaluating and analyzing qualitative data requires a process of organizing, de-composing, synthesizing data, searching for patterns and implication of the data and deciding how it should be conveyed to others (Bogdan & Biklen 1982). The method of analyzing data is dependent on the philosophical paradigms of the study (Myers 1997). For example, in grounded theory paradigm the researcher does not come to the field with a well-defined set of constructs and instruments
with which to measure social reality. Consequently, the researcher is only able to derive the categories from the field through in-depth examination and exposure to the phenomenon under study (Sobh & Perry 2006: 1204). Unlike other qualitative paradigms, realism researchers enter the field with prior theories because it is be-lieved that there is an external reality that other people may have usually re-searched or experienced aspects of that reality beforehand (Sobh & Perry 2006:
1204). Consequently, their perceptions are some of the many “windows” on to that reality deserving some consideration before realism data collection starts (Perry, Riege & Brown 1999: 18).
Qualitative data analysis process follows a formal process of data reduction (Sobh
& Perry 2006: 1204) through the interpretation and examination of textual data from different perspective (Sinkovics et al. 2008: 704). This requires three forms of coding process such as open coding, axial coding and selective coding (Miles
& Huberman 1984; Sinkovics et al. 2008: 704). Open coding refers to the process through which concepts and their properties are identified in the textual data (Sinkovics et al. 2008). It involves the naming and categorization of phenomena throughout the examination of the data. Axial coding refers to the process of es-tablishing relationships between categories and their sub-categories and identify-ing the intersection of related categories. It is also the process where codes from the conceptual framework are allocated to chunks of text. Selective coding refers to the process of refining and integrating categories to build a theory. (ibid) Selec-tive coding is the state where the concepts are established, and their underlying statements are used to explain the phenomena under study. It also involves the process of comparison and interpretations (Sobh & Perry 2006: 1204).
Data reduction process suitable for realism requires that the first phase of data reduction (i.e., open coding) be omitted. This is because only those perceptions relevant to the external reality are worth investigating (Sobh & Perry 2006: 1204).
Consequently, the codes used to reduce data in realism research should be gener-ated from the conceptual framework rather than the data. In contrast, other para-digms, for example, constructivism research argues for open coding as the first phase because every detail of the respondent’s perceptions should be the reality under investigation. Thus, realism research requires the use of the prior estab-lished constructs as the codes followed by axial coding and selective coding.
(ibid)
The potential challenge of this process is that other emerging themes that might not be in the conceptual framework might be omitted. Sobh and Perry (2006:
1204) suggested that this problem can be solved by the constant interaction of the conceptual model and the interview data. This implies a constant interaction
be-tween the conceptual model and data analysis. For this study, this challenge is solved by undergoing open coding, axial coding and selective coding for the rest of the chunk data that did not constitute a part of the prior constructs. A model of the data reduction process for this study is shown in Figure 14.
Figure 14. Data Reduction Process for the Case Study
There are established software for qualitative data analysis (e.g. Nvivo software, Coding Analysis Toolkit and HyperRESEARCH). However, the data analysis for this dissertation was done manually because the aim of realism research is not to track and match every phase of every perception in the textual data subjectively rather to track only those perception about an external objective reality (Sobh &
Perry 2006: 1206). Using qualitative software for data reduction is more relevant to subjective research that focuses on textual-based-theory (e.g., grounded theo-ry). Such software cannot analyze, interpret data nor make deductions and gener-alizations from the data. (ibid)
The first step for the case analysis was axial coding. This was done by reading through the textual data and marking a sentence or group of sentences according to their implied meaning from the conceptual framework. That means inserting a comment in the Microsoft Word document and naming the comment after the conceptual constructs they imply. The second step was selective coding. This was
done by using the established constructs in the first step and its interpretations in explaining their relationship to the choice of acquisition strategies, changes in equity and acquisition performance. The third step followed a process of checking the remaining chunk of data, identifying emerging constructs, naming and catego-rizing the constructs. If the outcome of the third step yields any emerging theme, this theme is then subjected to axial coding in search of related sentences and their implied meaning. Finally, selective coding is done until coding saturation.
All the coding process is done within-case analysis (i.e. within each case) (Patton 1990: 388; Yin 1994; Sobh & Perry 2006: 1203). The within-case analysis was examined using embedded design (Yin 1989). This requires analyzing the within-case analysis at each acquisition within-case. After the within within-case analysis, cross-within-case analysis was done. To ensure cross-case analysis, the summary of key findings in each case acquisitions was further tabulated into a single table often referred to as meta-matrix (cf. Miles & Huberman 1994; Nummela, Saarenketo, Jokela &
Loane 2014). Meta-matrix is shown in Table 20. The meta-matrix facilitates case comparison, contextualization, and generalizability.
The aim of the cross-case analysis is to enhance generalizability, deepen under-standing and explanation of the phenomenon under study by examining the simi-larities and differences between the within-case analysis (Miles & Huberman 1994: 173). The cross-case analysis followed a process of searching for common-alities and contextualized explanation as suggested by realism stance (Easton 2010; Welch et al. 2011). This entails searching for commonalities of causal mechanisms and the contextual conditions under which they work (Welch et al.
2011: 745). For this study, the commonalities and context were done by classify-ing the acquisitions accordclassify-ing to acquisition strategies utilized at the time of entry (i.e. partial, staged and full acquisition). Also, the acquisitions were classified in terms of the year of market entry and host country formal institutional classifica-tion at the year of entry. The host country formal instituclassifica-tional classificaclassifica-tion was based on International Monetary Fund classification (i.e. advanced, emerging and developing economies). Similar classifications or data from IMF (used in its orig-inal form or in combination with other economic institutional reports) are used in IB studies as measures of formal institutional environment (e.g. Arslan 2011;
Arslan & Larimo 2011). The classification of host country institutions based on advanced, developing and emerging is to aid our understanding of how and why these strategies were utilized in these contexts. Contextualization was also achieved by classifying the acquisition based on the motives of the acquisition.
The impact of informal institutions was solely based on the interview data and linked to the construct developed in the theoretical section (contractual complexi-ties).
Reporting Findings: Reporting findings refers to the process of presenting the data to the reader. Analytical reporting style and reflective reporting styles are two ways of reporting the findings of a qualitative research (Dooley 2002). Ana-lytical reporting style follows the sequence- literature review, methodology, re-sults, and discussions. It represents an objective writing style where the research-er’s presence is passive. (ibid) The second form of reporting qualitative research is the reflective reporting style. Unlike analytical reporting style, reflective report-ing requires the researcher’s presence to very noticeable in the report. Analytical reporting style is utilized in this study. In addition to ensuring that researcher’s presence is minimally present in the report, certain realism report guidelines are applied to ensure that the focus of the explanation in the reporting is on contin-gencies, structures and mechanisms through which the external reality is con-strued. In doing so, a conscious process of reflection was undertaken to under-stand what these findings mean, what alternative explanations of such findings existed, what disconfirming evidence are there for these explanations and how many of these findings relates to outcomes of previous research cycles (Cepeda &
Martin 2005: 861). The following guidelines in accordance to Sobh & Perry (2006: 1206) were applied to this study:
• Showing numerical frequencies of empirical experiences to enable readers follow the line of thought.
• Reporting findings should avoid numbers and should concentrate on inter-pretations.
• Every observation should have an explanation of why the observation oc-curred.
• There should be illustrative quotations in support of explanations and should frequently occur in the text with links to the respondent who said it, to provide in depth understanding that realism researcher seeks.
Reporting qualitative research follows the deductive vs. inductive dichotomy, i.e.
reporting with reference to theory and reporting without reference to theory (Ali
& Birley 1999, Thomas 2006). In this study, both deductive and inductive ap-proaches are combined when reporting the phenomenon under study (Ali &
Birley 1999, Thomas 2006). Thus, inductively, all specific data incidents that are not present in the prior conceptual constructs are reported as emerging theories and their underlying context are discussed. Deductively, the existing constructs are subjected to confirmation. Using the integrated approach enhances theory building in realism.