Data analysis and the role of existing theory

Table 6. Overview of data collection.

Ecosystem actors

Data

Vendors x 2 case studies x 6 interviews x Documents x Observations

x Demo system (NAV 2009 RTC)

Partners x 10 case studies x 16 interviews x Documents x Observations Customers x 5 case studies

x 5 interviews x Documents x Observations

x Implementation data of NAV 2009 RTC

ontological theory of Bunge (1977) and is thus often referred to as the Bunge-Wand-Weber (BWW) model or theory (Soffer, Golany, Dori et al. 2001; Rosemann and Green 2002). The BWW theory has been applied extensively in IS and ES research for various purposes including: meta models (Rosemann and Green 2002); process models (Green and Rosemann 2000); IS modeling (Soffer et al. 2001); and ES misfits (Sia and Soh 2007; Strong and Volkoff 2010). The BWW theory comes with an extensive set of ontological constructs for the purpose of evaluating ISs. However, for the purpose of framing the modeling and representation of roles in role-oriented ESs, their general notions of structures may be sufficient. The modeling of organizational roles and the methods used to reflect them in the ES may thus be perceived as deep structures of roles while the subsequent representation of the roles at the interaction level may be viewed as the surface structures of roles.

While the BWW theory was not applied in the appended research papers, it is applied in this cover paper to provide a theoretical lens through which to view the representation of organizational roles at the user interaction level (RQ1).

2.7.2 The role of existing theory in analysis of the case studies

While Case Study research supports a deductive approach where existing theoretical constructs are applied to the case analysis (Eisenhardt 1989), inductive inquiry without commitment to a particular existing theory is equally valid (Flyvbjerg 2006). When conducting case study research to answer the three research questions, two categories of existing theory were applied as “lenses” for analyzing the data: 1) organizational role theory and theory on user models and 2) the Resource-Based View.

The application of organizational role theory and user models was primarily based on the role-related concepts and structures uncovered during the review of previous literature (see section 3.1). Although the work of Katz and Kahn (1966) played a significant role in laying the foundation for the derived concepts and structures of roles in an organizational context, the theoretical concepts and structures, such as role

aggregation, role transition, and role specialization, applied to the analysis of the case studies is based on a synthesis of existing organizational role theory. Existing theories on user models (see section 3.1.4) were likewise applied in the analysis of the case studies. For a detailed description of the applied theoretical concepts see papers III, IV and VI.

2.7.2.1The Resource-Based View

The study on competitive advantage of the Microsoft partner ecosystem (paper I) applied the Resource-Based View (RBV) of organizations as theoretical framework.

The RBV theory was originally proposed as a response to the emphasis on environmental conditions as determinants of the competitiveness of firms in different industries, such as Porters (1985) five forces model, (Barney 1991; Grant 1991). RBV proposes that idiosyncratic attributes of the individual firm may impact its competitiveness (Barney 1991). More specifically RBV suggests resources, in the form of assets and capabilities, as determinants of a firm’s competitive advantage (Barney 1991). The two key assumptions of the theory are that: 1) firms operating in the same industry may be heterogeneous in respect to the strategic resources they control and 2) these resources are not perfectly mobile between firms. Barney (1991) suggested: value, rareness, imperfect imitability, and substitutability as attributes for describing strategic resources. These attributes have, however, been extended in several studies and Wade and Hulland (2004) review and synthesize these extensions and suggest the attributes of: valuable; rare; appropriable; inimitable; imperfectly mobile; and non-substitutable as attributes of resources that lead to competitive advantage.

A resource is considered valuable when it enables the firm to implement strategies that improve its efficiency and effectiveness (Barney 1991). Rarity refers to the condition where the resource is not simultaneously available to other firms (Wade and Hulland 2004). A resource is considered appropriable when it has the potential of generating

rent relative to the appropriation of the particular resource, which is difficult to access (O'Leary 2000). Inimitability prevents competitors from copying the resource (Wade and Hulland 2004). Imperfect mobility is the ability to prevent the transfer or acquisition of a resource between firms (Wade and Hulland 2004). Finally, a resource is considered non-substitutable when there are no strategically equivalent substitutes (Barney 1991). Depending on the degree to which a firm’s resources meet the criteria for the attributes, the firm can obtain different degrees of competitiveness (Barney 1991). The degree of competitive advantage can thus be categorized into: competitive disadvantage, competitive parity, temporary competitive advantage, and sustained competitive advantage, depending on the attributes of a firm’s resources.

The main reason for applying RBV theory in the study of the Microsoft ecosystem was its ability to explain and describe complementarity of resources of the actors in the ecosystem. The RBV thus helped to explain why the ecosystem has been, and according to our analysis continues to be, at a competitive advantage within the ES industry in Denmark. By combining the RBV with theories of network effects and lock-in (Shapiro and Varian 1999) it was also possible to assess Microsoft’s strategy of strengthening the relationship with their ecosystem partners.

2.7.3 The role of existing theory in analysis in the GTM studies

The debate on emergence vs. forcing (see section 2.3.3) has noticeable implications on the role of existing theory, or what Strauss and Corbin (Strauss and Corbin 1990) refer to as “technical literature”, when conducting GTM. Strauss and Corbin (1990) suggest two purposes for existing literature. First, previous research should help identify gaps in current understanding of a given phenomenon. This is not inherently problematic from a GTM perspective. Second, however, Strauss and Corbin suggest that existing literature and previous research “helps the researcher to delineate important variables for study and suggests relationships among them” (1990, p. 49). This is a significant departure from the “original” stance on existing literature, proposed in Glaser and

Strauss (1967), which suggests “literally to ignore the literature of theory and fact on the area under study, in order to assure that the emergence of categories will not be contaminated by concepts more suited to different areas” (Glaser and Strauss 1967, p.

37). Glaser (1992; 2007a), on the other hand, preserves the original stance on existing literature for theoretical integration and development of formal theory, only after the emerging substantive grounded theory has emerged, or as part of the data collection, in which existing theory ranks equally to any other data source.

The two papers in this dissertation that apply the Grounded Theory Method to analysis of data differ slightly in their use of existing theories. In the study of the transition between two versions in the Microsoft ecosystem (paper II), a general understanding of existing theories was present before doing the analysis, but existing literature was not studied in-depth until after the theory of ‘version transitioning’ had emerged. This gives the paper a slight Glaserian flavor, in regard to the role of existing theory. On the other hand, the analysis in the study of the tailoring of predefined roles in NAV 2009 RTC by the Microsoft partner companies (paper V) was carried out with an existing in-depth knowledge of the theoretical constructs of organizational roles from organizational role theory. While the existing theoretical constructs were not “forced”

onto the data, the closer interplay with existing theory in the analysis resembles the Straussian approach, in regard to the use of existing literature.

2.7.4 Analyzing the data for Grounded Theory

Analyzing data in GTM consists of the process of coding data. Again, the disagreements between the two founding fathers of GTM (section 2.3.3) have given rise to some disagreements between researchers on how the process of coding should be conducted. Hence, two approaches to coding data in GTM have been proposed. One approach, proposed by Strauss and Corbin (1990) consists of open, axial, and selective coding. The other consists of open, selective, and theoretical coding (Glaser 1992).

In the GTM approach prescribed by Corbin and Strauss (1990), open coding is the process of breaking down, examining, comparing, conceptualizing, and categorizing data. Axial coding consists of a set of procedures whereby data are put back together in new ways after open coding, by making connections between categories. Strauss and Corbin (1990) propose the use of a conditional matrix as an “analytical aid” in the process of doing axial coding. It is the use of this conditional matrix that has been the center of much debate among GTM researchers, and Strauss and Corbin (2008) have later emphasized that using the matrix is an optional part of their approach coding. The conditional matrix was thus not used in the axial coding phase of GTM. Selective coding is the process of selecting a core category, systematically relating it to other categories, validating those relationships, and filling in categories that need further refinement and development. In Glaser’s (2005) additional coding stage of theoretical coding, only the theoretical categories of the emerging theory is linked to each other.

Although Strauss and Corbin’s (2008) selective coding and Glaser’s (2005) theoretical coding differ conceptually, the selective coding seems to be able to achieve the purpose of theoretical coding and was thus applied.

The concepts and categories of the two GTM studies are described in detail in the two papers that describe the studies (paper II and V), but the following paragraphs offer elaboration on the process of coding the data in the studies.

2.7.4.1 Coding in the study on version transitioning

In the analysis for the study on how the partner companies transition from one version of the NAV system to another (paper II), the process of coding data was aided by the use of the Atlas.ti software, which is designed for the exploratory process of theory building (Muhr 1991). The interview transcripts were thus imported to ATLAS.ti for detailed analysis. In the phase of open coding, the transcripts were initially coded word-by-word. It quickly became apparent that consistent coding at this level was too detailed to yield a useful level of conceptualization. Instead, a combination of coding

word-by-word and line-by-line was applied. As the analysis progressed and concepts became more saturated, multiple lines of interview transcripts were coded at a time.

Coded instances of data were constantly compared to ensure consistent labeling of data. The following phase of axial coding did not apply the conditional matrix proposed by Strauss and Corbin (1990), as mentioned previously. Instead, concepts from the open coding phase were linked together and grouped into categories. This phase involved repeatedly changing concepts and re-coding data. In the phase of selective coding, a single category, ‘version transitioning’, was selected as the focal category for the emerging grounded theory and all other categories and their concepts were related to this category. Again, this phase included repeatedly re-conceptualizing and re-coding data to achieve the best possible “fit” of the categories of the emerging theory. The analysis was thus an iterative process which constantly sought to saturate the emerging concepts and categories of the grounded theory, by collecting and analyzing more data.

2.7.4.2Coding in the study on adaption of the predefined roles

The analysis of data for the study on how the partner companies adapted the predefined roles of NAV 2009 RTC (paper V) generally applied the same approach as the study on version transitioning (paper II). However, the purpose of the study was not to develop a “full” grounded theory, and hence the phase of selective coding was omitted from the analysis cycle. The analysis cycled between the phases of open and axial coding, and no single core category was selected from which to evolve a grounded theory around. Analysis ended with a set of related categories of the introduction of predefined roles, misfits of the predefined roles of NAV 2009 RTC, and strategies for addressing the misfits during implementation of the system.