4. Research Design and Methodology
4.4. Triangulation Analysis
4.4.1. Sample Case Examination
In this part, the analysis steps for the case descriptions will be introduced and how the result from the quantitative questionnaire and qualitative interview are interpreted and aggregated commutatively to attain the sample result.
4.4.1.1Quantitative Scoring of Questionnaire
As mentioned in the data collection part, the quantitative questionnaire follows the Likert-Scale between 1 (strongly disagree) and 5 (strongly agree) for each characteristic, except financial resources (slack) with the scale from 1 (strongly disagree) to 7 (strongly agree). After inquiring the questionnaire, the overall score has been calculated. For example, if a company reveals the highest level of global mindset, the score might be 40 – the maximum score for each characteristic is 40 (global mindset), 100 (cultural intelligence), 105 (learning orientation), 110 (network competence) and 28 (financial resource). These maximum scores are considered as 100% that a company could reach, consequently, the given scores in absolute number are converted into relative number of percentage. Similar to the qualitative interview, some of the questions are not answered due to the inter- organisational security reason. In this case, the maximum score of the question – 5 in the Likert-Scale – has been subtracted from the overall maximum score of the affected characteristic. For instance, the case of having one non- answered question in the global mindset questionnaire, the maximum score of global mindset is 35 instead of 40, so 35 is considered as 100% in the conversion.
4.4.1.2Qualitative Coding of Interview
Different from the quantitative scoring of the questionnaire, the qualitative coding on interview demonstrates higher complexity. The usual way of coding the qualitative interview might offer more freedom in analysis, however, because the modified fsQCA in form of combination set analysis is applied in this thesis, all qualitative results should be quantified as well and that are highly challenged for the qualitative coding of interview.
Because a code is a word, phrase, or sentence that represent aspects of a data and capture the essence or features of a data (Saldaña, 2015), the meaning of appropriate code selection should be considered. The given thesis follows the the descriptive coding concept which summarises the primary topic of the excerpt, assigns topics to aspects of the data and normally consists of nouns as codes (Miles, Huberman, & Saldana, 2013; Saldaña, 2015). The codes, that are used for the qualitative coding of interview, are based on the literature evidence, thus, these already cover an abundant range of the feature of each characteristic. However, when a new code is remarkable and illustrate the vital meaning during the coding process, these were integrated in the codes-pool as well. This kind of coding process is called directed content analysis; the study starts with theory, the codes are defined before and during data analysis and codes are derived from theory or relevant research findings (Hsieh & Shannon, 2005), as the given case descriptions demonstrates. In this way, the qualitative evidence from the interview are able to be coded as much as possible and the code-pool effectively covers the research boundaries with possible evidence.
The coding process was supported by the coding software Atlas.ti. In the end of the coding process, 218 codes are used for the coding process – global mindeset (48 codes) (see Appendix 1.1 Table 2), cultural intelligence (27 codes) (see Appendix 1.2 Table 4), learning orientation (32 codes) (see Appendix 1.3 Table 6), network competence (78 codes) (see in Appendix 1.4 Table 8), financial resources (16 codes) (see in Appendix 1.5 Table 10), focused internationalisation (13 codes) (see in Appendix 1.6 Table 11) and unfocused internationalisation (4 codes) (see in Appendix 1.6 Table 12). Each interview transcription was coded through the given 218 codes.
As mentioned briefly before, after coding, the codes are divided into code-groups which indicate the five characteristics, focused and unfocused internationalisation strategies. By analysing the codes, the absolute number of the codes are converted into the relative number. In other words, the current implementation provides a percentage relative to the total number of quotations for the selected code or code group. Because the codes cover highly abundant range of each organisational characteristics and highly differentiated at the same time, the result of conversion into relative percentage was highly reliable.
4.4.1.3Aggregation of Quantitative und Qualitative Research
After the conversion of qualitative coding of interviews, two percentages for each characteristic – one from the quantitative questionnaire analysis and one from the qualitative coding of the interviews, should be aggregated (see Appendix 4.3 Table 18). To set the equal weight on the meaning of qualitative and quantitative result, both are equally regarded – 50% each – into the aggregated research result. The given portion of both analyses also clearly represent the equal meaning of both methods in this thesis which follows the triangulation method.
In general, if the aggregated percentage of one characteristic (X) shows X 0,50, the characteristic is considered as significant. Due to the average conversion, the sensibility of the aggregated percentage has been dramatically increased, in other words, even though the number 0,50 and 0,48 do have the numeric difference of 0,02, the real level of each characteristic is huge, hence, if the aggregate result of one characteristic was lower than 0,50, it was interpreted as low level of the certain characteristic.
Having said this, compared to other characteristics, the characteristic financial resources should be considered differently. Because both questionnaire and interview questions were designed to investigate the existence of financial slack, the answers in the interview are relatively short, which means there was a lower number of quotes in general that might influence the calculation of code appearance in the previous coding stage. At the same time, because it has a more differentiated range of Likert-Scale between 1 and 7, it seems to be much harder to reach the full score in the analysis. Also, the fact that the talking about financial resources is considered as a sensitive topic and relates to the inter-organisational security issues, the interviewees generally did not express themselves much about the financial resources. Consequently, the number of quotes that might be able to demonstrate the evidence in terms of financial resources are also automatically reduced. Due to the given obstacles, the characteristic financial resource (slack) has been considered as significant, if X 0,40. In other words, if the aggregated percentage is over 0,40, a company is considered as the one which demonstrates high level of financial resources (slack).
Based on the aggregated percentage of each characteristics and focused or unfocused sequencing strategy, the case desciptions are formulated in the qualitative and verbal way. As the result of the case descriptions, the combination of vital characteristics for the sequencing strategy will be introduced and analysed for each case description, hence, the thesis will provide 10 different written case descriptions.