Keywords

1.1 Introduction

The main aim of the chapter is to discuss the case study method. We shall begin by confronting its definition. It is quite a challenge, as researchers representing various paradigms embark on this type of research project. These paradigms define the way we perceive the explored reality, our chances of understanding/cognizing it, and the acceptable research methods. As a consequence, not only is the case study subject to various definitions, but it is also employed to achieve manifold goals (Hassard and Kelemen 2010). Despite these differences, we can point out a number of characteristics that distinguish case study method; they shall be the focus of our discussion. As much as possible, we shall take into account the variety of perspectives in case study-based research, or recommend to readers the sources where they can find more detailed information on a particular issue. In this chapter, the presentation of premises and types of case studies will be followed by a manual, guiding readers in their endeavor to design their own research using the method discussed. For greater clarity, the manual is organized into sections, each providing answers to the following questions:

  • Step one: What do we want to find out?

  • Step two: Where shall we look for data sources?

  • Step three: How is data collected and selected?

  • Step four: How should empirical data be analyzed?

  • Step five: How are research conclusions formulated and how should we approach writing a research report?

Each of these steps is illustrated by an example from case study-based research in business and management.

1.2 Specific Character of the Case Study as Compared to Other Research Methods

The case study strategy Footnote 1 requires an in-depth and comprehensive analysis of a case within its context. A group, an individual, an organization, a processes, or social relationships can all be considered “cases” and, as such, be subject to research. It is advisable to begin by imagining the case as an example of a social or a theoretical phenomenon. A detailed description, together with a thorough analysis, should contribute to understanding the case and formulating several theoretical conclusions.

When emphasizing the comprehensive approach to analysis, some authors go as far as claiming that it is not owing to methodological reasons that researchers have recourse to the case study method, but rather because of their interest in a particular case and the desire to thoroughly examine and comprehend it (Stake 2005). The method allowing them to attain this goal is secondary. Therefore, the selection of techniques and sources is purely pragmatic. The defining feature of the case study strategy is, therefore, a wide range of research techniques combined with various types of data used (Creswell 2007; Gerring 2007; Stake 2005; Yin 2003a).

This variety of tools and data types can hardly be considered typical of the case study strategy, as nearly all qualitative studies combine different techniques, such as interviews, observations, content analysis, and various sources, that is, individuals (both as interviewees and as objects of observation), documents, films, and photographs. The same is true for research questions regarded as typical of case studies, that is, why and how a particular thing happens (Yin 2003b). We ask them when we are interested in processes, interactions, and dynamics (Miles and Huberman 1994; Hijmans and Wester 2010) rather than in “snapshots of social life” (Kostera 2008). Undoubtedly, taking the former (i.e. processes, interactions, and dynamics) into account is one of the advantages of case study, although yet again, it is not uncommon for qualitative methods. Another feature that is characteristic but not specific of the case study is the observation of events in their natural environment and context (Gerring 2007; Yin 2003b; Hijmans and Wester 2010).

Given how difficult it is to discern the features that distinguish the case study from other qualitative approaches, the task is sometimes referred to as the “definitional morass” (Gerring 2007, p. 17). We shall, nevertheless, attempt to identify such characteristic features.

It seems that the difference that defines the case study can be grasped through reference to the aim set by researchers who adopt other approaches to research. For it is the aim that dictates which techniques and data sources should be used and where to put emphasis in the research process . In the ethnographic method, emphasis is on the reconstruction of the cultural context within which the examined group functions; all tool design and data collection procedures are subordinated to this aim. The grounded theory method seeks to create a theory that “fits” the observed reality. In the case study method, the primary aim is a comprehensive description and understanding of the case and of its context. Then, depending on the researcher’s paradigmatic affiliations, the following alternative options are possible:

  • using the obtained results to create abstract (theoretical) general concepts that can be used to describe and explain the examined phenomenon (Stake 2005; Creswell 2007);

  • developing theories that expound social reality within a delimited area (Mills et al. 2010; Eisenhardt and Graebner 2007);

  • modifying or supplementing existing theories (Burawoy 1998; Mills et al. 2010; Wadham and Warren 2014; Yin 2003a);

  • referring to a wider class of similar phenomena (Seawright and Gerring 2008; Stake 2005);

  • providing practical solutions to specific types of problems, for example, organizational issues or problems with the evaluation of various social interventions (Hassard and Kelemen 2010).

Another distinguishing feature of the case study method is the way in which units of analysis are treated. The case and its context are often described as a system or a bounded phenomenon (Stake 2005; Creswell 2007; Gerring 2004), which is to emphasize that the investigated unit of analysis (organization, person, process) is defined together with its numerous aspects and within a broad network of social, political, institutional, ethical, and aesthetic phenomena and meanings (Creswell 2007; Mills et al. 2010; Stake 2005).

By now, it should be clear how difficult it is to formulate an adequate definition of the case study research method. Let us quote the definition suggested by John W. Creswell, as it appears to comprise all important elements mentioned thus far, yet remains general enough not to exclude the numerous variants of relationships between theory and research that are typical of the method in question. Creswell (2007, p. 73) defines the case study as:

a qualitative approach in which the investigator explores a bounded system (a case) or multiple bounded systems (cases) over time , through detailed, in-depth data collection involving multiple sources of information (e.g., observations, interviews, audiovisual material, and documents and reports), and reports a case description and case-based themes.

Nothing should prevent an investigator from having recourse to quantitative techniques in the case study methodology if it is deemed justified and advantageous for the research question (Creswell 2007; Mills et al. 2010). It is important, however, to ensure consistency with the adopted theoretical (and methodological) approach and to properly select research patterns. As mentioned above, the case study method is, after all, described and applied by representatives of various paradigmatic traditions.

1.3 Premises of the Case Study Strategy

Despite divergences in understanding the case study method, a number of basic traits distinguish it from other methods. They relate to several key issues, such as the understanding of the concept of “case”, priority attributed to the description of a single case, the importance of theory and generalization, the way in which reality is understood and expounded. The above are approached in various manners, depending on the research tradition within which the case study is defined. We shall now attempt to illustrate and explain this diversity.

Case as a “System”

As mentioned above, by applying the case study method, we choose to understand the case as a system within its broadly defined contextual framework, which often poses the problem of case boundaries. If we study an organization in the context of its relationships with the cultural, social, and institutional environment, we may wonder where the organization ends and its surroundings start. As researchers, however, we are obliged to trace case boundaries at an early stage of the research project. This implies the need to make fundamental decisions about the object of our research endeavors. We should specify the timeframe and the extent of the phenomena we intend to research within the case.

Focus on the Specific Character of a Given Case

Case study method is founded on the principle that each case should be regarded as a complete and unique phenomenon if we are to understand its internal dynamics. Regardless of the number of cases analyzed, any study based on this method requires a thorough knowledge of each case (Stake 2005). This has implications for analytical procedures: if we examine several cases, we need to prepare a separate report for each of them, and then carry out a comparative analysis. As Miles and Huberman (1994, p. 207) write:

It is crucial to have understood the dynamics of each particular case before proceeding to cross-case explanations. Without that, superficiality sets.

The above authors point out that cases do not form fully comparable data sets: each of them is governed by its own rules and has its own narrative , and therefore it would be a mistake to dismiss the context and content oneself with an inventory of similarities and differences between individual cases. They recommend “marrying” the variable-oriented approach with the case-oriented approach. Case-oriented analysis is “an analysis that aims to understand a particular case or several cases by looking closely at the all the details of each”, while the variable-oriented analysis means “an analysis that describes and/or explains a particular variable” (Babbie 2016, p. 383).

The Place of Theory in Case Studies

Divergences in understanding the role of theory in case study research are an excellent example of the impact that the individual author’s philosophical outlook has on his/her research assumption. Many post-positivist authors (i.e. those looking for mechanisms governing social behavior, Sharma 2010) use case studies to achieve two goals at the same time . The first is to describe and explain a given phenomenon within the framework of a specific theoretical perspective. The second is to fill in any gaps in a specific theory or to modify it. The latter can be attained through a flexible approach to categories set within the theory and the researcher’s openness to empirical data. This approach is adopted, for example, by Michael Burawoy and Robert Yin (Burawoy refers to it as the extended case method, see Burawoy 1998). Somewhat in line with this approach is Miles and Huberman’s suggestion that case study method, when compared to ethnography, is more dependent on specific theoretical models both in the initial phase of research and during the analysis of the collected empirical material. As these authors point out, in the case study strategy the emphasis is placed on the unification of data collection procedures and on a more systematic approach to the selection and use of analytical tools (Miles and Huberman 1994; Hijmans and Wester 2010).

Proponents of qualitative research that represent the interpretative tradition criticize this approach and argue that through imposing explicit theoretical frameworks or resulting hypotheses , researchers become less sensitive to signals from the field. They suggest that researchers should specify the general theoretical perspective they adopt in the study (e.g. critical or feminist theory). It is referred to as the “abstract dimension” or the “working theory” (Stake 2005) and provides certain sensitizing concepts (Charmaz 2005). Examples of these abstract dimensions include criminology, conflict resolution, resources, hegemony, or domination. It is also possible to have recourse to theory during the final stage of the research, when results need to be interpreted (Creswell 2007; Lincoln and Guba 1985).

Numerous authors highlight the usefulness of case studies in exploratory research that results in generating hypotheses and—at a later stage—developing theories (Eisenhardt 1991; Eisenhardt and Graebner 2007; Flyvbjerg 2006; Gerring 2007; Silverman 2005; Stake 2005; Hijmans and Wester 2010). In this situation, the first case provides initial concepts and hypotheses whose pertinence is subsequently verified through the analysis of subsequent cases selected with a view to verifying the generated interdependences (Aaltio and Heilmann 2010).

Another option is to relinquish theoretical references or references to broader phenomena of a particular type. This means focusing only on the description of a given case and explaining it within its own categories identified in an analytical process (Creswell 2007; Stake 2005). Students who decide to apply this approach should be aware that this type of case study may raise objections, as it does not involve attempts at drawing theoretical and/or practical conclusions. As David Silverman (2005, p. 127) writes: “If all you aim to do is simply to ‘describe a case’, you may rightly get the response: ‘so what?’”. However, a number of authors, for example, Robert Stake (2005) and Bent Flyvbjerg (2006), argue in favor of this strategy (see also “Generalization Based on Cases”).

In practical terms, the diversity of paradigmatic affiliations of researchers using and describing the case study method requires a certain level of alertness when literature sources are selected. For example, when planning an interpretive research, one should not choose Robert Yin’s work relaying on realist and objectivist presuppositions (Haverland and Yanow 2012, p. 403).

Idiographic Bases of Understanding and Explaining

The etymology of the adjective “idiographic” indicates the nature of this explanation: idio (Gr. special, separate) indicates that the explanatory procedure rests on a comprehensive description that facilitates the understanding of a given phenomenon. A case study provides a detailed and “pure” description with references to the case’s history, or a chronology of events (Stake 2005). The case is only systematically analyzed in the subsequent step. Some authors argue that the analysis ought to be based on references to the concurrence, sequentiality, and contextuality of events, claiming that social reality is unique and random (Stake 2005). Other authors underline the potential of qualitative research —and of the case study in particular—in reconstructing causal relationships understood as causal mechanisms (Gerring 2007; Miles and Huberman 1994). They claim that the potential of qualitative analyses stems from the fact that they are embedded in the local reality and take into account the diversity of social phenomena, their sequential and processual nature. In addition, they propose different analytical procedures in order to detect these interdependences (see Step Four: How Should Empirical Data Be Analyzed?).

Generalizations Based on Cases

Case study researchers do not have the same approach to generalization. Again, the paradigmatic approach is a distinguishing factor in this respect. We shall discuss different generalization possibilities afforded by research based on case study methods: first, those related mainly to the realist perspective, and subsequently those representing the anti-realist approach. In research based on case studies, Robert Yin represents the first group, while Robert Stake is associated with the second (Moriceau 2010).

As highlighted at the beginning, cases may be selected for research to exemplify broader phenomena or theoretical constructs . Consequently, any conclusions drawn from case observation are based on the knowledge of these phenomena or theories. In light of the above, Burawoy posits that case studies ought to be used to modify and supplement theories. Conclusions drawn from a case study can be theoretical in the sense that they can serve the purpose of developing a typology or enriching theoretical knowledge of a specific kind of phenomena. Such generalizations are called analytical (Glaser and Strauss 1967; Yin 2003a). This is why particular attention is paid to the selection of cases.Footnote 2 The selected cases should enable us to observe the mechanisms or phenomena of our interest. Very often, cases are selected for research precisely because they are atypical and unique, or they call into question generally accepted theories and stereotypes. The value of this kind of analysis goes beyond its capacity to satisfy our curiosity that difference naturally arouses. Contrary to conventional wisdom, exceptions do not confirm the rule, but enable us to discern it.

Case studies often focus on “natural experiments” (Hammersley and Atkinson 1992), that is, events that test a particular theory and the interdependences it suggests. Selecting cases that are significant from the theoretical point of view and lend themselves to analytical generalization is called theoretical sampling (Silverman 2005). More generally, we can refer to this kind of sampling as strategic (Flyvbjerg 2006), because the categories governing the selection are not always theoretical: they may relate, for instance, to our expectations about the case’s informative content.

Many researchers wish to juxtapose their research findings with abstract interdependences, but also to give them a certain prognostic value in relation to other cases within a specific population type. The following strategies may be applied for this purpose (Hammersley and Atkinson 1992; Silverman 2005):

  • assessing the typicality of the case (obtaining information, for example, statistical data about relevant aspects of the population of cases and comparing our case to them),

  • combining in-depth case analysis with survey research on a random sample of cases,

  • coordinating several qualitative studies,

  • gathering a relatively large number of similar cases (relatively rare but we may ensure that the sampling covers the entire population, for example, all companies operating within a given industry),

  • referring to the analytic model, according to which each case lends itself to generalization (e.g. a study of language use within the field of sociolinguistics or conversation analysis).

The last generalization option is different from those mentioned above. It is not a formal generalization (does not refer neither to theory not to population), but hermeneutical (based on interpretation ). An investigator who presents a case in a detailed and naturalistic manner provides the readers with a vicarious experience. When faced with a fine case study, readers have the impression of a firsthand observation of the events described, they can make generalizations when encountering a similar case and, subsequently, confirm or modify any conclusions drawn from it. Stake and Trumbull (1982) term this process “naturalistic generalization”, while Lincoln and Guba (1985) relinquish any generalization and suggest replacing it with the terms “transferability” and “fittingness” (where a hypothesis developed in one context “fits” or can be transferred in another). This method of generalizing concerns first and foremost—but not exclusively—intrinsic case studies (devoid of direct references to theories or wider phenomena).

1.4 Types of Case Studies

Depending on the needs of researchers and on research aims, different types of case studies can be distinguished. Bent Flyvbjerg (2006, p. 230) proposes a typology of cases on the basis of the criterion of case suitability from the point of view of its information content. The typology of case studies developed by Flyvbjerg is presented in Table 1.1.

Table 1.1 Flyvbjerg’s typology of cases
Full size table

Flyvbjerg claims that the choice of the case does not necessarily mean that only one particular type among the above is possible. A case study can be extreme, critical , and paradigmatic at the same time (Flyvbjerg 2006, p. 233). However, maximum variation cases and critical cases tend to be favored by post-positivist researchers, as they imply a certain regularity of social reality that lends itself to theoretical explanations with varying degrees of generality. In turn, Robert Yin founded his typology of cases on the criterion of research aims. A detailed description of selected case study types identified by Yin is presented in Table 1.2.

Table 1.2 Yin’s typology of cases
Full size table

Yin proposes yet another manner of differentiating case study strategies, based on the number of cases. Depending on the number of cases examined, he distinguishes between single case studies and multiple case studies (Yin 2003a).

Multiple case studies have the logic of an experiment that produces either similar or contrasting results, depending on the value of the independent variable. The criterion for the selection of cases is the presence of certain determinants, and the purpose of the study is to ascertain whether their occurrence actually produces the predicted effects (Yin 2003a). This approach to the case study reflects the logic behind maximum variation cases in Flyvbjerg’s typology and is representative of the post-positivist approach, where experimental logic is considered the main form of scientific reasoning.

Yin (2003a, pp. 40–42) argues in favor of a single case study when the case in question is:

  • a critical case study—that is, serves to test an existing theory; when all assumptions of a given theory are satisfied by a single case, this case provides a sufficient test for the theory’s validity ;

  • an example of rare or unique circumstances—the study is carried out with a view to obtaining new information on existing mechanisms when they concur with an unusual situation that may reveal an aspect that has yet to be explored. An atypical situation may also provide the foundation for creating a new typology (e.g. description of a new, unknown disease);

  • a typical case—when the study serves to examine typical conditions or common situations in order to refer research findings to other cases that belong to a given group (e.g. exploring a single, “typical” firm representing an industry in order to draw conclusions pertaining to other companies from the same sector);

  • a novel case—when it is possible to explore phenomena that have been out of the researchers’ reach or their area of interest;

  • a long-term case study—when it is possible to examine a case over different periods, that is, two or more points in time .

1.5 How to Use Case Study Method in Practice: A Step-by-Step Guide

At the beginning of our discussion on the case study, we noted that it is used in various ways, depending on the researcher’s paradigmatic beliefs . However, if we are to develop a practical guide on designing research, it is necessary to opt for a specific approach, that is, an approach embedded in a particular paradigm. Our focus here is on the search for processes and structures providing both description and causal explanation of events (Miles and Huberman 1994)—hence the emphasis on the importance of theoretical analysis of the subject matter. Given the above features, the present guide represents primarily the post-positivist trend, recognizing nevertheless the complexity of human behavior, the social nature of reality, and problems in distinguishing cause-effect relationships (Sharma 2010). Wherever we considered it possible, we point to solutions typical of the case study method that are adopted in other paradigms.

1.5.1 Designing Research as Creating a Chain of Evidence

Now, we shall present the subsequent steps of the research procedure. They are logically linked and form a coherent sequence of decisions, initiated with the researcher’s decision to study a particular phenomenon from a given conceptual/theoretical perspective and with asking specific research questions . If we decide to study a particular phenomenon, we must choose an appropriate case to examine and, within this case, look for events, attitudes, and ideas pertaining to the explored subject. Therefore, we need to select the tools and decide where we will look for particular aspects of the phenomenon we explore.

Once we have gathered all relevant data (i.e. related to the subject of our interest), we need to organize and analyze it in order to answer the research question . As the case study is either inductive or partially inductive and partially deductive, the analysis may be interwoven with data collection. Qualitative research is flexible. The data collection phase may lead to the modification of the research question or to supplementing it with new questions that better reflect the observed phenomena.

Then, on the basis of the organized and analyzed data, we should be able to draw definite conclusions in response to the research question and in reference to its theoretical framework. Individual steps form a closely linked sequence of events that Robert Yin calls the “chain of evidence”: from research questions to final conclusions. The coherence of these steps is also called methodological congruence (Morse and Richards 2002; Creswell 2007) (Fig. 1.1).

Fig. 1.1
figure 1

Designing research as a process of creating a chain of evidenceSource: authors’ own

Full size image

At each of these stages, we shall encounter problems compelling us to make ethical decisions. Qualitative research is carried out “close to people” and it involves research that interferes—to a greater or lesser extent—with their world. Negotiating access to a particular community, involving its members in the research process , and gathering data that is often personal or emotional, or related to individual or group interests, are among particularly sensitive tasks. When conducting research and analyzing its results, and in particular when planning their publication, we must consider whether our decisions will not adversely affect those who take part in our project (more on this subject, see, e.g. Creswell 2007; Silverman 2001).

Step One: What Do We Want to Find Out?

When planning research, we already have a general idea about the kind of phenomenon we intend to explore. In the first step, we should structure our research plans. The process of structuring research process begins with asserting the problem that shall be analyzed, and with expounding the reasons for undertaking research within the chosen area. Potential reasons include the author’s willingness to fill gaps in the extant literature, to acquire knowledge about unexplored or insufficiently explored areas of social life, or to grasp the social significance of the problem ; there may be practical reasons, such as the researcher’s wish to address a particular problem, for instance, an organizational issue. To properly ground our intentions , we review the extant literature and other research on the topic, searching for suitable theoretical approaches, discussing the aim of our research project with the supervisor, project participants, or students.

At this point in time , we can answer the following question: what exactly do we want to find out? What kind of explanatory factors will we seek? At this stage of research process , the degree to which we refine the theory providing us with a framework for problem analysis may vary depending on the type of case study that we choose to perform.

If the selected topic has yet to be explored, we will probably opt for the exploratory case study. We may also simply be intrigued by a particular case and treat theoretical interest as secondary—in this situation a single case study design will be our choice. Nevertheless, even then we have to build our research on a certain theoretical concept or concepts (more on this subject, see Silverman 2001, 2005). The conceptual or theoretical perspective may be a general outlook that affects our perception of a phenomenon (e.g. critical theory, feminist theory, organizational culture metaphor ), or a more specific theoretical system, such as new institutionalism or resource dependency theory. If our goal is to develop a middle range theory on a given subject (“nested” in a wider theoretical system), we are more likely to choose a multiple case study design. If the goal is to test or modify an existing theory, we will rather opt for a critical case study, or for extended case method.

Even if it is going to change at the later stage of the inquiry, the object and the purpose of the study must be precisely defined, which requires a certain theoretical effort. Example 1.1 illustrates how theoretical interests can be linked with the choice of both research method and research object, together with its context.

Example 1.1 Linking Theoretical Interests with Research Method and Object. Source: Gawer and Philipps (2013)

In the research of a computer industry and particularly of Intel Corporation, authors were guided by a theoretical aim—to better understand the link between institutional logics and institutional work. They have asked following research questions : What kinds of institutional work do organizations perform as they attempt to influence the institutional logic that characterizes their field; and what kinds of institutional work do organizations perform in response to the resulting logic shift?”

A single case study was chosen as a method because “the links between institutional work and institutional logics are not well understood; because we are interested in the worldviews of organizational members; and because our study is exploratory and aimed at theory building” (2013, p. 1040).

Intel Corporation situated in the computer industry became a case. Authors focused on a specific timeframe (between 1980 and 2000) as, during this period, the computer industry underwent a profound transformation underpinned by a shift in institutional logics. Similarly, to other organizations in the industry, Intel Corporation had to adapt to this changing logic. But additionally, during this time Intel became an influential actor that played a central role in the changes that occurred in the industry. Hence, as authors of the research argue, “This case therefore offers a unique opportunity to examine the institutional work performed by an organization within a field as the field undergoes a shift in institutional logic” (p. 1040).

Step Two: Where Shall We Look for Data Sources?

The next step is to answer the following question: how do we want to find out everything that we need in order to answer the research question ? If we are interested in a particular phenomenon, we should ask ourselves: where should I look for data that will allow me to answer the research question ? In the case study strategy , the answer to the latter question is, in a way, two-level. The first level refers to the selection of the case as an example of a problem that we wish to explore. The second level requires sampling within the examined case.

Cases are selected through purposive sampling. We can do it through adopting such criteria as uniqueness or atypicality of the case, or, on the contrary, select “ordinary” cases (Creswell 2007). It is worth remembering that uniqueness or typicality are assessed against other phenomena or against a theoretical background. For instance, the case of a trade union that does not follow oligarchization trends is unique in the context of other trade unions (Lipset et al. 1956) and against the background of theories that explain the oligarchization process. This fact points to the relevance of theory for the selection of cases; we therefore argue that purposive sampling equals, essentially, theoretical sampling (Silverman 2005). Theory points to important variables that differentiate social phenomena or issues playing a significant role in social processes and predicts the course of events. Consequently, theory guides us towards those places where we will be most likely to find problems of interest to us. At this point, we should also decide whether we will explore a single case or several cases and provide a clear definition of the case in our research. Definition indicates the unit of analysis, that is, what or who will be explored: organization, process, relation, or person. Then, we have to consider what exactly will be researched as a case. The process of defining case, its boundaries, and the unit of analysis is facilitated by the exploration of the research question (i.e. reflection on what it is exactly that we wish to learn from the study) and the theoretical perspective (What kinds of relationships are of interest to us? What are our key concepts?). Example 1.2 illustrates the interplay between choosing core theoretical concepts (collaboration between business units), choosing research design (multiple cases), and defining cases and units of analysis (organization as a unit and collaboration as a subunit of analysis).

Example 1.2 Defining Cases and Units of Analysis. Source: Martin and Eisenhardt (2010)

In their theory building research, Martin and Eisenhardt used so-called embedded case studies, involving more than one unit of analysis; in this kind of research, subunits are also taken into consideration (Yin 2003b). Here, the analysis involved the firm (unit) and the cross-business-unit collaboration (subunit).

Authors justify their theoretical goals flowingly: “Given limited theory about how executives create high-performing cross-BU [Business Units] collaborations, we relied on inductive theory building using embedded, multiple cases. Multiple cases are likely to yield more generalizable, robust, parsimonious theory than single cases” (p. 268). Further they explain choosing of the software industry as a proper research setting. First, software industry is knowledge-based industry for which it is typical to have many opportunities for cross-BU collaboration. Second in the case of software industry, this kind of collaboration is strategic and widespread. Hence, new cross-BU collaborations are likely to be frequent.

Six publicly held software firms were selected for the research, each of them with multiple BUs. The selection was diversified according to industry segments (consumer, enterprise, and infrastructure), age (founding dates extending from 1967 to 1995). The combination was aimed at improving the robustness and generalizability of the results.

Based on the literature and informants’ perceptions, researchers used several criteria to define a business unit. “First, a BU was defined as a distinct and separable organizational entity with authority over key BU-level strategic decisions, including resource allocations. Second, it sold distinct products that customers could purchase independently of those offered by other BUs in the same firm. Third, it was managed by a GM with an executive team. Fourth, its firm evaluated it using profit measures such as return on investment (ROI) and return on sales (ROS)” (p. 269).

Polar sampling was used for selecting recent cross-business-unit collaborations in each firm (i.e. one collaboration was high-performing and one, low-performing). This kind of sampling was used as it is a particularly effective theoretical sampling approach, “making the emergent constructs and theoretical relationships ‘transparently observable’” (p. 269).

The second level, where we select sources from which empirical data is obtained, is the sampling within the examined case. In qualitative case studies, it usually involves purposive sampling based on the criterion of “suitability” of the source as the subject/object, which is to provide the data used in order to answer the research question . Therefore, the sampling method depends on the research question and on the adopted theoretical perspective.

Although sampling within the case is determined by the nature of the research question , it is important to note three general categories that will likely be relevant to the explored phenomenon and should, therefore, be included in the sampling. These are time and its impact on behavior, such as the beginning and the end of the working day; people and their perspective—contingent on the position occupied within a community; and the context that influences behavior, for example, employees’ behavior during office hours versus a corporate team building trip (Hammersley and Atkinson 1992, pp. 56–63). It is important to take these differences into account in the process of sampling.

Step Three: How Is Data Collected and Selected?

Suppose we have already developed a preliminary theoretical perspective, we have selected a number of cases, and that we know who or whatFootnote 3 we intend to explore. Now, we have to answer the following question: how can we collect and select relevant data? We decide on the type of research tool we will use and we work on its design . The specific design of the research tool allows us to select information, that is, decide which among the great variety of data available in the study will help us answer the research question .

Let us first discuss the choice of techniques. It requires plenty of information: what kind of data is best acquired through interviews and which types of information lend themselves best to observation? Once we have chosen the type of tool that we need, we have to think about its design. If we decide to conduct interviews, what ought to be included in their scenarios? If we opt for document analysis, what should be its angle? We must consider which elements of reality are important from the point of view of the research problem and then include them in our research tools (in the interview scenario, observation plan, or content analysis plan). This is a tool that will help us “capture” relevant empirical data from the explored reality. We can look for signs and expressions of these important issues in empirical data—both induced (e.g. interviews) and existing independently from researchers’ intervention (e.g. documents). Example 1.3 illustrates the way in which issues significant for research are detected and subsequently included in the research tool .

Example 1.3 Choosing Research Tools and Data Sources. Source: Belz and Binder (2017)

In their paper based on a multiple case study, Belz and Binder (2017) attempted to explore sustainable entrepreneurial processes (SEP) understood as recognition, development, and exploitation of an entrepreneurial opportunity with a view to balancing social, environmental, and economic goals. They used the following data collection procedures. First, face-to-face interviews with the (co-)founders. Semi-structured, open interviews were a major source of information allowing for an in-depth understanding of decision-making processes and for gaining relevant background information from key informants. The interview scenario consisted of three sections: (1) personal background of the (co-)founder; (2) entrepreneurial process from the first idea to the market entry (including contextual influences of political, economic, and social nature); and (3) weighting of economic, social, and ecological goals. The questions evolved around activities, events, and outcomes, rather than hearsay—a measure reducing the potential for retrospective bias. Before and after the face-to-face interviews with the (co-)founders, archival data from internal and external sources were gathered (e.g. the websites of the sustainable enterprises, blogs of the (co-)founders, and press releases, social and print media dealing with the SEP). The latter kind of data was real-time archival data, which allowed a triangulation with the personal account of the sustainable entrepreneurial journey as told by the (co-)founders in the interviews. Using this measure authors increased the internal validity of the study and reduced the potential for retrospective bias. To enhance the external validity of the study they asked the co-founders to review drafts of the case study report.

The case study method is characterized by the multiplicity of data sources used and research techniques applied. Triangulation , which means combining various techniques (methodological triangulation) and data sources (data triangulation) within a complex research process , allows for a more complete description and comprehensive understanding of the case, thus increasing the relevance of findings . When designing a study, we should take several research tools into consideration.

It is important to remain open to data coming from the field. The process of designing a qualitative case study combines inductive and deductive strategies, which means adopting at the very outset a particular theoretical framework, which will subsequently be modified or supplemented. The openness of the qualitative research method allows us to observe phenomena that we have not foreseen in the initial approach (serendipity, Glaser and Strauss 1967); they allow us to formulate a new hypothesis , modify and supplement the initial theoretical approach with new, empirically observed elements. In the course of the research, it may transpire that the research problem evolves (Stake 2005), and that the research question itself requires modification, because its original version fails to address the most important phenomena shaping the situation. If this is the case, then we must modify our approach; the modification ought to be documented and justified in the research report. After all, discovering that reality is different from what we have anticipated is one of our research findings (see the next subsection.)

The research protocol and the database (Yin 2003a, 2010; Davis 2010) form part of the phase described in this subsection and are, at the same time , typical of the case study strategy. Both tools increase the reliability and credibility of the study. The protocol is a collection of research instructions and is built on a case-by-case basis. It includes questions that the researcher should answer on the basis of the collected data, the likely sources of information, reminders pertaining to data collection procedures (e.g. interview scenarios, observation plans) and organizational information such as the timeframe, credentials, information that can be given to a person interested in the project (e.g. aims and auspices). Tips on writing a case study report may be an additional element (for an example of a report plan, see Stake 2005). The database, in turn, contains structured empirical data, for example, interviews, observation notes, and documents used in the analysis (for more information on registering data, see Creswell 2007, p. 138).

Step Four: How Should Empirical Data Be Analyzed?

The next step in the evidence chain—from the research question to the answer—is data analysis. If we have opted for an exploratory case study, that is, one whose aim is to describe the phenomenon, or one that is to generate hypotheses and develop theories, then analytic categories will emerge primarily during the analysis of the collected empirical material (Creswell 2007). The concepts that we used at the beginning of the project as “sensitizing concepts” should be at this stage formulated as a problem, which means that we need to bestow them with meaning in the experience of the group subject to research and in their understanding, while avoiding textbook definitions (Charmaz 2005). In the course of the research we ascertain, inter alia, “if, when, how, to what extent, and under which conditions these concepts become relevant to the study” (Charmaz 2005, p. 512).

This type of analysis is primarily based on comparisons (in terms of differences and similarities) and on a shift from specific to increasingly abstract categories , which takes place through the reduction of the number of codes . The process is called open coding and it is rooted in grounded theory and is followed by axial coding (relating codes to each other). If the purpose of our research is to establish certain explanatory mechanisms or theories, we could use such strategies as analytic induction, which is aimed at systematically developing causal explanations, search among elements that overturn these explanations, adjust explanations to make sure they take into account any observed exceptions until explanations match the observed phenomena (Hammersley and Atkinson 1992). The analytical induction strategy comprises two basic techniques: the constant comparative method and the deviant-caseFootnote 4 analysis (Silverman 2001). Excellent descriptions of coding and data analysis in inductive research can be found in ethnographic studies (Hammersley and Atkinson 1992, pp. 210–244; Silverman 2001) and studies of the grounded theory method in its both interpretive and post-positivist versions (Charmaz 2005, 2006; Glaser and Strauss 1967).

If, on the other hand, we rely on an existing theoretical framework, for instance, to carry out a critical case study, or when we use an extended case method, the way in which we organize the data largely depends on the research question and on the adopted theoretical perspective. This means that they must be referred to in the process of designing analytical codes (i.e. categories according to which we organize data). Coding is, in such cases, prefigured to a certain extent, which means they need to correspond to the adopted theoretical framework (Creswell 2007; Wadham and Warren 2014). When analyzing the collected evidence, we should pay close attention to the data that does not match our codes . Such unclassified data will serve as the basis for the creation of open codes , based on the analysis of data that does not match the scheme. The data and its interpretation form a background for rebuilding and enriching the original theory. In order to stay faithful to the data (Maxwell 1996), we may be forced to reorganize the research perspective, which often entails modifying the research question , or even the aim of the study. Such incidents are quite typical of case study researchers, who often admit that “their preconceived views, assumptions, concepts, and hypotheses were wrong and that the case material compelled them to revise their hypotheses on essential points” (Flyvbjerg 2006, p. 235).

Both during and in parallel to the coding process, we can apply a variety of analytical techniques (Yin 2003a).Footnote 5 First, we can resort to pattern matching; it involves the matching of patterns based on empirical data to those foreseen in the preceding theoretical analysis. If we are looking for cause-effect explanations, we must remember that the mere conformity of observations with theoretical predictions—even if it confirms the internal validity of the analysis—does not yet guarantee the existence of a cause-effect relationship. We should verify whether the observed conformity is not due to factors different from those identified in the theoretical perspective. The verification technique is based on testing alternative explanations. Criteria for assessing an idiographic explanation (valid for the case study research) comprise, first, the logic, that is, establishing how credible it is and, second, the demonstration that alternative explanations have been seriously considered and found wanting.

The problem of cause-effect relationship or causal mechanisms (Gerring 2007) is also solved by another analytical technique called explanation building (Yin 2003a), which involves distinguishing a pattern of cause-effect relationships. In the case of multiple case studies, the aim is to build a general explanation that fits each individual case, even if certain details differ. Building such an explanation is based on the same logic that applies to pattern matching, yet in this case any comparisons with the theory are only the beginning of the process. This technique consists in performing a series of repetitions: comparisons of observed phenomena with theoretical predictions, their potential revision, and further comparisons with case data. It often compels the researcher to return to the field in order to gather additional data. This is adequately reflected in the aforementioned statement that the case study method is partially inductive and partially deductive.

Time series analysis or time-ordered display is another technique that can be applied (Miles and Huberman 1994; Denis et al. 2001; Yin 2003a). The chronological organization of events does not need to be merely a descriptive tool. It may have an analytical purpose, which is to establish cause-effect relationships, as causes and effects may not be reversed in time (see Example 1.4).

Example 1.4 Time Series Analysis as an Analytical Tool. Source: Denis et al. (2001)

Denis et al. (2001) used time series analysis to develop a process theory of strategic change in pluralistic settings characterized by diffuse power and divergent objectives. They draw on five case studies in health care organizations. The analysis involved decomposing the chronological data for each case into successive discrete time periods. The periods (or phases) became comparative units of analysis. Phases were defined so that “there is continuity in the context and actions being pursued within them, but discontinuities at their frontiers” (p. 815). Here, the boundaries of the periods were designed either by changes in the key people involved (the leadership constellation) or by a major change in the environment.

Following techniques suggested by Miles and Huberman (1994) researchers conducted the comparative analysis of periods within and between cases. The three most important categories for the analysis within each period included “(1) the characteristics of the leadership constellation during that period (who the important members were, what roles they played, what their degree of complementarity was), (2) the actions of the leadership group (what was done, what kinds of tactics were used, and (3) the effects of these actions and tactics (symbolic, substantive , and political)” (p. 815). With the advancement of the analysis authors developed new theoretical ideas and refined categories for analysis.

Another technique is cross-case synthesis, which consists in organizing data from different cases according to the same pattern (e.g. according to the same code categories ). The purpose is not only to analyze individual characteristics but also any similarities and differences between cases, which may result in building a typology or increase the accuracy of predictions about the existence of causal links.

Finally, it is important to emphasize one very important thing: during the analysis, special attention should be paid to any observed exceptions, phenomena that deviate from patterns and do not suit the formulated explanations. Exceptions have a particular potential for improving and modifying accepted explanations or descriptions. Even if we fail to incorporate them into the constructed models, principles governing research oblige us to record them.

Step Five: How Are Research Conclusions Formulated and How Should We Approach Writing a Research Report?

The last step in the chain of evidence is the writing of research conclusions. They are an extended version of answers to the research question and refer to specific (cases) and general issues (theories) contained in it. We should therefore describe and explain the phenomena observed in the case under consideration and juxtapose them with the a priori adopted theoretical framework or present findings through the lens of a theory emerging from the empirical data. Research-based conclusions will form part of the final report, whose remaining sections are devoted to literature and methodology . In the literature part, research conducted thus far and relevant theories are discussed; in the latter, we describe the adopted research strategy (type of case study, case selection and sampling, research tools used, analysis of the collected empirical data). It is important to prepare drafts of these parts in advance, as it constrains us to consciously plan subsequent research stages, while providing a sense of support and a reference point for further steps, including the writing of research conclusions. Usually, when we prepare a journal paper, a monograph, or a thesis, both literature and methodological part are (repeatedly) rewritten in order to embrace the changing conclusions from iterative moving between data gathering, analysis, and theory.

The core of what will be included first in the case study report and later in research conclusions is determined by the research question formulated both at the beginning of the project and possibly also, it its modified version, in the course of the research. It provides us with categories that serve as a basis on which we decide whether a given thread should be included in final research conclusions. Ignoring the research question at the stage of writing the study report and the final report results in inconsistency and chaos in the study, which is, after all, the final outcome of the research.

The form in which research findings are presented is another important issue that must be addressed. Research conducted using the qualitative case study method must take the form of extensive, descriptive reports. Compiling results is worth additional effort, as it allows us to communicate them more effectively and helps readers and authors alike to understand them better. Matrices, graphs, diagrams, and network diagrams (representations, Miles and Huberman 1994) are tools that can be used for this purpose, highlighting the most important factors and relationships between them. At the same time , they are the fruit of an analytical process, during which the collected empirical data is organized. In this sense, working on the representation forms part of the analytical process: not only is it a way of presenting results but also of generating them. Another formal aspect concerns the composition of the final report as a whole. It largely depends on the decisions we made at the stage of research design. For example, if we choose a strategy based on the design of rival explanations (Yin 2003a), that is, if we apply rival theories in order to analyze a selected case or cases, the structure of our report ought to be comparative. Therefore, each case must be presented several times , always using a different description or explanation scheme. This allows us to ascertain which explanations best suit the data collected during the research procedure.

If we opt for a strategy based on theoretical assumptions, our report will likely follow a classical linear-analytical structure. It begins with a review of the extant literature, which is followed by a methodological section and the presentation of results obtained on the basis of the collected data. We may begin by presenting case descriptions, and then move on to a comparative analysis, or begin with a comparative analysis and include case descriptions in an annex. Empirical and theoretical and possibly also practical conclusions form the final part of the study. This pattern is typical of scientific studies, such as theses or academic papers (see Example 1.5). Description of other proposals for the formulation of case-based research reports can be found in Yin (2003a) or Hammersley and Atkinson (1992).

Example 1.5 Communicating Results: Strategy Based on Theoretical Assumptions. Source: Rządca and Strumińska-Kutra (2016)

In our paper on local governance learning, we used the extended case method, that is, we built on a preexisting theory with a view to modifying it on the basis of empirical research). The structure of the paper is linear-analytical.

We started with two theoretical approaches: organizational learning and institutional theory. The first enabled us to conceptualize the phenomenon of learning new patterns of rule connected to the concept and practice of governance. The latter was used to analyze conditions influencing the process of governance learning.

In the second step, we provided a case study of local governance practices undertaken by a public agency (city administration) in the process of public dispute resolution. Case description and analysis was preceded by a methodological note explaining why the extended case method had been chosen and how data collection and analysis had been conducted.

In the third step, the case study was used to develop an initial conceptualization of the governance learning phenomenon into a richer, empirically informed framework. As a result, our analysis contributed to advancing the knowledge of governance learning through (1) specifying different types of governance learning, which are linked to the structure of learning and not to its motivation, (2) linking the micro level of local governance practices with the mezzo level of organizational structures and with the institutions regulating governance at the macro level, and (3) explicating the difference between learning and institutional change. Based on our research, we have introduced a new theoretical category of astonishment, which we treat as a prerequisite for governance learning. It is defined as a cognitive state caused by a disruption of institutionalized patterns of thinking and behavior deployed by a (public) organization to deal with a specific (social) problem.

In the final part, “Discussion and further research”, we referred to the limitations of our approach and proposed further research that might advance the field of governance learning.

In addition to structure, when writing a report, we must also consider the question of style and rhetoric. As stated by Hammersley and Atkinson (1992, p. 192), language is an analytical tool, not a transparent medium of communication (see also Miles and Huberman 1994). Hence the choice of style should be deliberate, as it is also an interpretative tool. This argument refers in particular to the so-called traditional writing of research reports, which create the impression that an objective connection between the analysis and the field exists . As Denzin and Lincoln note (2005, p. 3), “Experimental, reflective ways of writing first-person ethnographic texts are now commonplace”.

1.6 Conclusions

The so-called didactic case study is used for teaching purposes in numerous fields (such as management, law, medicine), as it allows for the simulation of participation in specific situations, and thus helps students understand certain issues and their relevance for the aspects of reality that they study. At the epistemological level, the case study used in research fulfills the same function; it enables contextual , embedded in real life, observations of the phenomena that are the subject of our interest. Knowledge gained through analyzing specific cases allows us to understand social processes and the role of factors considered important in many theories (Gerring 2007).

Understanding the specificity and diversity of social life in its various aspects would not be possible without the exploration of specific cases. Bent Flyvbjerg argues that a researcher having recourse to the case study method gains specific experience and context-dependent knowledge that supplements rule-based knowledge (Flyvbjerg 2006). The proximity of the examined reality and its careful observation make us aware of the multifaceted nature of relations within the social world and compel us to constantly revise our knowledge and beliefs . The mere knowledge of theoretical interdependences is not enough to fulfill any social role, in particular the role of a scientist, expert, or researcher. Although the case study, as evidenced in this chapter, can be used to build or modify theoretical knowledge, its uniqueness lies in its ability to provide in-depth descriptions that accurately represent the explored phenomenon.