Case study as a research method

The description of the research process in the next section is structured according to the steps for building theories from case studies (Figure 7). The process presented by Eisenhardt (1989) has a positivist view of research, as it is directed toward the development of testable propositions and theory which can be generalised across settings. Although this study does not aim at such generalisation, it was decided to describe the conducted procedures through the stages of the theory building research due to the clarity and elaborateness of this roadmap.

Reaching

Figure 7: Process of building theory from case study research (adapted from Eisenhardt, 1989)

Definition of the research question

Ideally, in the beginning of a theory-building research, there should be no theory under consideration and no assumptions to test, in order not to bias or limit the findings by preordained theoretical perspectives. Instead, one should formulate a research problem and possibly specify some potentially important variables, with some reference to existing literature. An initial, tentative definition of the research question enables a researcher to specify the kind of organisation to be approached and the kind of data to be gathered.

(Eisenhardt, 1989) Selecting cases

The population defines the set of entities from which the research sample is to be drawn.

Furthermore, the selection of an appropriate population controls extraneous variation and helps to define the limits for generalisation of the findings (Eisenhardt, 1989). The selection of cases can aim at replication of previous cases, extending emerging theory, filling theoretical categories, or providing example of polar types (Ibid.). Another point of consideration is the replication logic. Cases should be selected so that each of them either predicts similar results (literal replication) or produces contrasting results but for predictable reasons (theoretical replication) (Yin, 1994).

Crafting instruments and protocols

Case studies typically combine such data collection methods as documents, archives, interviews, questionnaires, observations and physical artefacts (Yin, 1994). Flexible and opportunistic data collection methods allow the investigator to take advantage of emergent themes and unique case features (Eisenhardt, 1989). However, in small business research, it has been noticed that the key decision makers show disinclination for completing questionnaires (Bell et al., 2004). Furthermore, making an enquiry is typically complicated by a lack of published information (e.g. shareholder reports, commercial analyses) and poor recording of internal data (Carson et al., 1995). It was noticed early in the empirical study that the data would be mostly limited to the interviews, because of scarce availability of written material. Observations were excluded because no on-going cooperation decision process was available for inspection at that moment. As a consequence, the study was chosen to be conducted qualitatively through in-depth interviews with the representatives of the selected case companies. The Interview guide approach (Patton, 1990) was chosen over standardised open-ended interviews or questionnaires to ensure richness of material. The goal was not to gather quantifiable data, but to construct a holistic picture of the phenomenon.

Entering the field

Marshall and Rossman (1999) distinguish elite interviewing as one specialised form of interviews. It focuses on a particular type of interviewee – those considered to be influential,

prominent, or well-informed people in an organisation. Elites can usually provide an overall view of an organisation or its relationship to other organisations (Ibid.). Furthermore, they are able to report and elaborate on an organisation’s policies, past histories, and future plans from a particular perspective. In theory-building research, data collection frequently overlaps with data analysis, which allows researchers to make adjustments during the data collection process (Eisenhardt, 1989). The researcher is guided by initial concepts and developing understandings, but they are modified alongside the data collection and analysis (Marshall and Rossman, 1999). This can mean addition of questions to an interview protocol or addition of data sources, to take advantage of emerging opportunities. In order to maximally exploit the interviewees’ expertise and experience, the topical guides were adjusted during the data collection.

Analysing the data

In qualitative studies, data collection and analysis go hand in hand to build a coherent interpretation of the data (Marshall and Rossman, 1999). According to Eisenhardt (1989), the first stage is within-case analysis. It involves detailed case study write-ups for each site. The idea of this stage is to become familiar with the data and to set up preliminary theory generation (Ibid.). Each case is examined as a stand-alone entity before looking for patterns across cases. The analysis is continued with cross-case comparison. This means looking for the presence of constructs across multiple cases and examining whether similar themes emerge in multiple settings (Miles and Huberman, 1984; Eisenhardt, 1989). The process of category generation involves noting patterns expressed by the interviewees (Marshall and Rossman, 1999). For editing and immersion strategies, the categories are generated through prolonged engagement with the data in text form (Ibid.). The segments of text are then placed into these categories.

Shaping propositions

The purpose of this stage is to sharpen constructs and theory, and verify relationships.

Internal validity is built through search for evidence for “why” behind the relationships. The emergent frame should be iteratively compared with the evidence from different cases.

(Eisenhardt, 1989) Enfolding the literature

The emergent concepts and theory need to be compared with the existing literature.

Comparison with conflicting literature increases the confidence in the findings and builds internal validity, whereas comparison with similar literature enables wider generalisability and higher conceptual level. This stage is particularly important if the findings are based on a very limited number of cases. (Eisenhardt, 1989)

Reaching closure

Ideally, a research should be continued until it reaches the point of theoretical saturation, but it is not uncommon to plan the number of cases in advance due to the availability of resources and time constraints. Eisenhardt (1989) recommends conducting 4 to 10 cases in order to have a manageable, but convincing level of complexity. The iteration between theory and data has reached its saturation point when the incremental improvement to theory is minimal. The outcome of the process can be concepts, a conceptual framework, propositions, or middle range theory (Ibid.). The latter is a solution to a problem that contains a limited number of assumptions and considerable accuracy and detail in the problem specification (Weick, 1989).