Classification and analysis of the data

This chapter explains the procedure for classifying and analyzing the interview data (known as the case study protocol, as Yin 1994, calls it). In the first phase this was done with the memos that were written down during the interviews. In the second phase the data was classified and analyzed after transcribing the interviews.

The following list includes the steps taken in the process of classifying the data gained in the interviews and in the introductory sessions:

1) A general overview of the interview data (making notes online during the interviews, listening to the tape, and updating the notes). The interview memos were written and sent to each interviewee for verification.

2) Reading through the interview memos: marginal remarks were made in order to highlight the topics of discussion. The interview coding was created (see

Appendix 3). It was based on i) the contextual factors of the cases, and ii) the nature of information sharing. The governance of information sharing (practices, success factors, lessons learned) was included in the contextual factors to emphasize their appearance in the specific program/project.

3) Preparing the interview summary: the counting of the events (according to [Miles & Huberman 1994, 69] this is called the first-level coding) and their classification according to the main categories per R&D program (business context, information sharing specific issues, R&D supplier and program management and other governance practices and lessons learned). Referring to Miles & Huberman (1994, 69), this is pattern coding.

Another part of the classification was done later after the transcription of the interview data. The procedure was similar to the one mentioned above in that the process and interview coding remained the same. However, at this time the interview summaries were done differently: the summaries of each main research issue were generated into Excel worksheets (altogether 27 worksheets) according to the case contexts, content, media, and challenges of information sharing, among other things.

After classifying the interview data, the analysis process began. It was first done by utilizing the interview summaries of the Excel worksheets. This phase helped in creating a deeper understanding of the big picture, and it was easier to find out the most emphasized issues. To find out relations/dependencies and explanations between the program contexts and the nature of information sharing, for example, mind maps were drawn. Appendix 4 includes examples of the interviewees’

comments on the interrelations between different factors (codes and sub-codes).

Attention was paid to the issues that were most highlighted in the interviews. The analysis of the data did not concentrate on cross-case comparison as such, but on comparison of the contextual factors inherent in the Sub-Cases. Therefore, the research results reported in Chapter 5 include extra notifications, if one phenomenon is highly emphasized in one or two particular Sub-Cases. However, it was found that

evaluating differences between the programs often proved difficult: as an example, it was difficult to define which of the products in the R&D programs was the most complex one (instead, the complexity of the parts of the products was easier to evaluate).

The transcription of the interviews resulted in more detailed interview summaries than was possible when using the interview memos as the main source of data. After generating the research summary for the Case Company, the analysis went further.

Firstly, the iteration of the company-specific research results was done by preparing a case report, and the thesis advisors and other competent persons they nominated in the Case Company could check the consistency of the interview results. Secondly, theories and new literature references were studied in order to either find support to or explain the interview results.

The analysis phase included evaluating the research quality, and the sources of data and the representativeness of the data had to be taken into account as well. The interviewees had a strong and wide background of the business in question. Before each interview the interviewees were asked of their previous and current duties. This helped in analyzing the relevance of each interviewee’s opinion (if the person had been only a short time in their position, the data would not have been that valuable).

However, careful planning in the selection of the interviewees turned out to be fruitful and was noticed here: all the interviewees had been at least six years with the Case Company and most of them in the same business unit and in the R&D programs in specific. Nevertheless, it is worth mentioning that a couple of the interviewees were no longer in the position or in the same business unit the interview handled, but they spoke about their experiences in the R&D program and the relationship being involved in.

Also, the selection of the programs based on their time of occurrence proved to be the right decision, although some interviewees still had memory retrieval problems (especially in terms of exact dates of the program milestones). This problem was

eliminated by asking the same questions of different interviewees, which helped in collecting a comprehensive and coherent view of each problem area. Access to other program information and documents also led to a detailed view of the program, and the memory retrieval issues could be verified easily.

Appendix 5 summarizes the data collection process as well as the analysis of the data.