2. RESEARCH METHODOLOGY
2.2 R ESEARCH PROCESS
2.2.5 Analyzing data
Analyzing data is the heart of building theory from case studies (Eisenhardt, 1989), but it is the most difficult and the least codified part of the process. Since published studies generally describe research types and data collection methods but give little space to discussion of analysis, a huge chasm often separates data from conclusions. As Miles and Huberman wrote:
‘One cannot ordinarily follow how a researcher got from 3600 pages of field notes to the final conclusions, sprinkled with vivid quotes though they may be.’ (Miles, et al., 1994, p.
16)
However, a few key features of analysis can be identified. One key step is the within-case analysis; the other is searching for cross-case patterns. Within-case analysis typically involves detailed case study write-ups for each site. These write-ups are often simply pure descriptions, but they are central to the generation of insight (Gersick, 1988; Pettigrew, 1988) because they help researchers to cope with the often enormous volume of data early in the analysis process. However, there is no standard format for such analysis. Different scholars have used different processes. Quinn (1980) developed teaching cases for each.
7 In addition to the added rigor and internal validity, one of the main benefits of taping and transcribing interviews is that the interviewer can concentrate on what is being said, rather than being continuously distracted by note-taking.
8 Interviewees received copies of the transcripts with requests for approval. If they objected to certain parts of the transcripts they were asked to mark the parts, which were then omitted from the final transcript. This occurred in three instances where a few sentences were omitted by request of the interviewee. Interviewees were also asked to make additions or clarifications, which were then integrated into the final transcript version. Such additions were made in two transcripts. One interviewee submitted a clarification for a single term. With the exception of this clarification, which was transmitted via telephone, the three other requests for changes were transmitted via e-mail and were marked by the interviewees directly in the original transcript data file.
Mintzberg and McHugh (1985) compiled a case history from their set of cases, Leonard-Barton (1988) used tabular displays and graphs of information about each case, and Abbott (1988) suggested using sequence analysis to organize longitudinal data. However, the overall idea is to become intimately familiar with each case as a stand-alone entity. This process allows the unique patterns of each case to emerge before investigators push to generalize patterns across cases. In addition, it gives investigators a rich familiarity with each case, which, in turn, accelerates cross-case comparison.
Coupled with within-case analysis is cross-case search for patterns. The tactics here are driven by the reality that people are notoriously poor processors of information. Different scholars have pointed out their weaknesses:
• leaping to conclusions based on limited data (Kahnemann, et al., 1973)
• being overly influenced by vividness (Nisbett, et al., 1980), or
• by more elite respondents (Miles, et al., 1994)
• ignoring basic statistical properties (Kahnemann, et al., 1973), or
• dropping disconfirming evidence (Nisbett, et al., 1980)
Thus, the key to good cross-case comparison is counteracting these tendencies by looking at the data in many divergent ways. One tactic is to select categories or dimensions, and then to look at within group similarities coupled with intergroup differences. A second tactic is to select pairs of cases and list the similarities and differences between each pair. This tactic forces researchers to look for the subtle similarities and differences between cases. The juxtaposition of seemingly similar cases can break overly simplistic frames. An extension of this tactic is to group cases into threes or fours for comparison (cluster analyses). The third strategy is to divide the data according to data sources. This tactic exploits the unique insights possible from different types of data collection. When a pattern from one data source is corroborated by evidence from another, the finding is stronger and better grounded.
When evidence conflicts, the researcher can sometimes reconcile the evidence through deeper probing of the meaning of the differences.
Overall, the idea behind these cross-case searching tactics is to force investigators to go beyond initial impressions by using structured and diverse lenses of accurate and reliable theory⎯that is, a theory with a close fit to the data. Also, cross-case searching tactics enhance the probability that the investigators will capture novel findings that may exist in the data.
The data analysis in this study is based on three methods: within case analysis, cross-case analysis within the three segments, and cross-segment analysis. The two cross-case analysis
methods build on the within-case analysis framework. This framework is a mixture of the processes proposed by Leonard-Barton (1988) and Abbott (1988). Following the recommendations of Leonard-Barton, the case findings in the three chapters (firm development, alliance portfolio, and alliance processes) were categorized in tabular displays and the quantitative data were graphed. Following the recommendations of Abbott, the longitudinal data were allocated to development stages. These stages were displayed as sequential events.
To note is the network analysis, because the alliance data was broad, more-dimensional, and at the heart of this research study. Starting point for the network analysis was an alliance database, which was filled out during the data collection. This database comprises data on the case study, its partners, and the characteristics of their cooperation over time. Due to the complex data structure, a special software tools was employed. Pajek9 (de Nooy, et al., 2003) was the software package of choice for the representation of network dynamics; a selection, other scholars recently did as well (i.e. Powell, et al., 2002; Uzzi, et al., 2002).
Pajek allows to analyze extended networks and to identify subsets such as multi-connected component and clusters (White, et al., 2001). In addition, Pajek can expose the network’s emergent structure as organizations enter and exist with one another over time (Uzzi, et al., 2002). The visualized networks are presented in chapter 3 to highlight both the process by which new ties and organizations are added to the network and how the network structure evolved. A detailed description how the networks were graphed and which algorithms were used is provided for comprehensibility reasons in section 3.2 together with the first network graphs. This proceeding contradicts the classical division of methodological aspects from the empirical findings but⎯hopefully⎯supports the readability of this study through providing methodological details where the subsequent application illustrates the methodology.
Alliances: research methodology, the network analysis is based on two layers. Its general structure is analyzed mainly based on quantitative data as the network size (number of partners), network quality (intensity of links) and its center of gravity (characteristic of partners). Qualitative data⎯especially narratives on partnerships⎯help to understand how ties are changing and how the process of partnership formation, intensification, restructuring, and termination work.
9 Pajek is a software package for large network analyses. Further information at http://vlado.fmf.uni-lj.si/pub/networks/pajek/
This study is based on three different performance dimensions: growth, profitability, and innovation. Several indicators measure these dimensions. These indicators are weighted and integrated into one single balanced performance index, which is depicted in table 14. This sub-section motivates the different dimension and explains the different indicators and its weighting.
All case studies started as technology growth ventures. The archetypes for these companies are Microsoft, Sun Microsystems and Cisco Systems. The value of these companies is determined through their growth. Therefore, growth is a key performance indicator. Growth can refer to organizational growth and economic growth, which are normally strongly correlated. Economic growth is measured in absolute sales of their last fiscal year10 and the growth rate of the last years, which is used as a short to mid-term growth perspective. The organizational growth is measured analogous in number of employees at the beginning of 2002 and the growth rate of employees. Additionally, reorganizations and lay-offs have been accounted for negatively.
The profitability is the next dimension. Its importance grew constantly over the evolution of that industry. After the capital markets turned bad, all case studies had to focus on internal growth. Profitability secures survival and further growth. The profitability is measured by two indicators: the bottom line profit and the date the break-even point has been reached or is planned to be reached. In finance literature, many profitability measures have been developed. Cash flow based performance measures, as the CFROI11, are more accurate to measure performance, because they are not distorted by depreciation. But they have not been used for simplicity and data availability reasons.
Innovation is the third performance dimension. It measures the distinctiveness of the companies’ technology. The case studies’ growth perspective mainly depends on their unique technologies and services. The innovation is measured in awards obtained. Relevant awards are multi-media, start-up, and new media awards.
From the within-case analysis plus the two cross-case findings and overall impressions, tentative themes, concepts, and relationships between variables begin to emerge. The next step of this iterative process is to compare systematically the emergent frame with the evidence from each case in order to assess how well or poorly it fits with case data.
10 In most cases the year 2001
11 Cash flow based return on investment