«Researchers have endeavored to increase understanding of the relationships between investments in information systems (IS), competitive advantage, ...»
Node positions within the network structure are calculated through measures of network centrality including degree, closeness and betweenness. Degree is the number of direct ties a node possesses in the network. Closeness is a measure of the length of path for a given node to other nodes in the network structure. The node with the most efficient pattern of direct and indirect ties will have the lowest closeness score in a network.
Betweenness is a measure of the connectivity of a node between disparate points in a
network in acting as a bridge or broker between nodes. A node with high betweenness has the most control over information flows within the network structure as a whole. It should be noted that centrality measures are node-level measures, not group-level measures (Hanneman & Riddle, 2005).
The social network analysis in this dissertation used data gathered through managerial interviews conducted in the context of competitive dynamics. Netdraw 2.089 Graph Visualization software was used to process the data to generate the graphic representations of managerial one-mode social networks and two-mode IT-mediated networks as well as centrality measures presented later in the discussion of findings. The following sections provide an overview of the steps followed in this research to conduct one-mode and two-mode social network analysis.
One-mode Social Network Analysis. Following Borgatti and Cross (2003), the first step was to identify the organization and its population of interest. The organization was again the focal firm, FCI, and the population of interest was the managerial team.
While Borgatti and Cross (2004) suggest that surveys are often used in gathering social network data, such methods have been criticized in extant literature, as it can be difficult for respondents when asked to accurately recall individuals in their social network on their own and the context in which such interactions were meaningful (Bernard et al.
1982). Thus, this limitation somewhat addressed by utilizing data from both semistructured and structured interviews and presenting managers with the context of interest:
stages of the competitive dynamics process were identified from the data.
Interview data was examined to determine if a tie between managers exists in a given stage of competitive action. If a tie was determined to exist, then the strength of the tie was evaluated based upon frequency of communication between the two managers (Hatala, 2006). Based upon an examination of the data, one-mode social network matrices were constructed (Hanneman & Riddle, 2005). One-mode social network analysis seeks to examine ties within one data set. For example, in this study, information was sought on ties between individual managers. Once social network matrices had been constructed, Netdraw 2.089 Graph Visualization software was used to derive visual depictions of each managerial social network at each stage of the competitive dynamics process, and to calculate three measures of network centrality (degree, betweenness, closeness) (Hanneman & Riddle, 2005).
Table 3 lists the job titles of managers identified as either having ties to them, from them, or bidirectional ties in the context of New Product Development.
Table 3 Managerial Positions of Managers in Study II with Strong and Weak Ties Chief Executive Officer Chief Financial Officer Chief Operating Officer Corporate Controller Director of Engineering Director of Global Sales Director of Human Resources Director of Information Technology
A detailed discussion and images of the one-mode managerial social networks are presented in chapter eight in the Discussion of Findings and Research Implications later in this dissertation.
Two-mode Social Network Analysis. It has been noted that little is understood about how individuals interact with technology in the organizational setting beyond the single user-system relationship (Lamb & Kling, 2003). Futhermore, current knowledge is largely limited to how organizational participants use a single information system, when most individuals employ multiple systems (Kane & Alavi, 2005). In this dissertation, efforts were made to recognize any media used by the managerial team in the context of communicating within each stage of the competitive dynamics process: conceiving, enacting, executing, and firm performance.
Interview data was searched for various media identified by members of the managerial team when communicating in the context of the competitive dynamics process. The following media were identified from the data: informal face-to-face
landline telephone, ERP system, Blackberry email, Blackberry cellular telephone, SMS text messaging, instant messaging, Skype, video conferencing, digital knowledge repository, and China cellular telephone (China Phone).
Semi-structured and structured interview data was used to develop two-mode social network matrices (Hanneman & Riddle, 2005). Two-mode social network analysis seeks to identify patterns relating different types of data. For example, in this study, information was sought on the use of a particular media in a relational tie between two managers, i.e., is Blackberry email used in the relational tie between Manager A and Manager B, and if so, how frequently is it used. The strength of use of a particular media was evaluated by its frequency of use (Hatala, 2006) at a given stage in the competitive dynamics process.
Once the two-mode social network matrices had been constructed, Netdraw 2.089 Graph Visualization software was used to derive visual depictions of the two-mode social network at each stage of the competitive dynamics process. A detailed discussion and images of the two-mode managerial social networks are presented in chapter eight in the Discussion of Findings and Research Implications later in this dissertation.
Validation of Social Network Analysis. Validation of social network analysis was conducted in two ways, by the employment of an ‗insider‘ and ‗outsider‘ researcher (Evered & Louis, 1981), and validation by an organizational participant. In the first stage of validation, to corroborate the findings of the insider researcher, a second, objective researcher reviewed each stage of the data analysis. Ideally, a second researcher who has
seldom employed). This outside researcher takes on the role of a more detached investigator who analyzes the data ‗objectively‘ (Gioa & Chittipeddi, 1991). The inside researcher was a bona fide participant who conducted the interviews, and an outside objective researcher was consulted to ensure that the software-based objective analysis was conducted correctly and that proper procedure as outlined in the literature has been correctly followed. Secondly, one manager participating in the study was asked to review each one-mode and two-mode social network to provide his view with regard to the legitimacy of the findings.
7.3.2. Centering Resonance Analysis. Study I of this dissertation is now under its third and final review in the Special Issue on Digital Systems and Competition in Information Systems Research. In the comments received from the Senior Editors of the Special Issue it was pointed out that ―Your decision-making team is composed of a network of managers.‖ Centering Resonance Analysis was recommended by one Senior Editor as a novel and viable method to investigate discourse among managerial social network participants using a text analysis-based approach and to ―establish convergence across managers for a given decision process.‖ This study has conceptualized social computing and communications technologies as mechanisms for enabling and facilitating interactions among managerial participants in the context of conceiving, enacting and executing competitive actions and responses toward firm performance. Collective discourse is an important consideration with regard to studies of organizational communication (Corman, Kuhn, McPhee & Dooley, 2002).
Fairhurst, 2001). Without effective discourse, the ability to organize is compromised, as discourse shapes organizational activities (McPhee, Corman, & Dooley, 2002). Tulin (1997) suggests that ―Organizations are processes of communication and discourse analysis is the means to discovering the interactive bases of organizational phenomena‖ (p. 101).
Discourse analysis is defined as the study of words, symbols, patterns, language structure and interpretation of discursive practices. Unique patterns of discourse emerge in the organizational setting, evolving from organizational culture and structure, organizational groups, power structures and so forth, where individuals and the organization affect the composition of discourse in a bi-directional association (Fairhurst & Putnam, 1998).
An emerging tool for the analysis of discourse in complex social systems is Centering Resonance Analysis (CRA). ―Centering resonance analysis (CRA) is a method of network text analysis that is designed for the study of complex discourse systems‖ (Brandes & Corman, 2003, p. 41). CRA uses computational linguistics to present transcribed text as a network representation. CRA locates, links, and maps concepts within either transcribed conversations or online messages such as email (Corman et al.
2002) and can compare mappings across various groups. CRA can be used in organizational problem solving, including ―…experts being organized into research teams or between clients describing problems and the experts whose discourse shows they can solve those problems‖ (McPhee et al., 2002, p. 275).
theory of local discourse coherence and salience used in studies of languages and language structure. Centering theory incorporates a set of rules and constraints that govern the relationship between the subject of discourse, syntax, and the salience (prominence or importance) of noun, pronoun, etc., in the sentence structure. Centering theory is described as a way to model the centers of discourse (Walker, Joshi, & Prince, 1998).
CRA incorporates several basic steps. The initial stage is the selection of noun phrase elements or the focal words that point to the center of discussion. In the second stage words are linked into a network indicating their sequence inside sentences. The next stage is indexing. Two indices are created, influence and resonance. Influence measures the betweenness centrality of a word, which is the likelihood of that word having the shortest path in the network connecting any two other words. Influential words are those which facilitate connections of meaning across different words and different parts of the word network. The influence concept inherent in CRA is significant, as other text analysis methods rely on word counts or frequencies, which does not take into account the relationships between words and meanings across words (Canary & Jennings, 2008). Word networks that are similar in terms of influential words and phrases are said to resonate with each other. Thus, resonance measures the similarity of any two networks having the same influential words. The last stage is concept mapping, where the most influential words and their relationships are displayed visually as a network. CRA can be used to generate networks based upon single ―speakers‖ or
―a sophisticated discourse analysis approach, sensitive to conceptual linkages expressed in a single sentence, yet able to generate networks describing vast stretches of discourse.‖ Corman et al. (2002) suggest that CRA is well suited toward studies of communication in complex collectives (Perrow, 1967) such as organizations, where members produce vast quantities of discourse. Corman et al. (2002) describe the difficulty in examining communication within complex systems such as organizations where existing research methods are incapable of handling both the volume and multifariousness of communication data. They describe Browning and Beyer‘s (1998) grounded theory investigation of the development of standards within an organization, where the authors analyzed tremendous volumes of interview data in an effort to examine communication. Corman et al. (2002) believe that the Browning and Beyer (1998) study made valuable contributions to the study of communication as derived through grounded theory. However, they also suggest that while grounded theory was able to identify specific insights and linkages, it was not up to the task of handling the volumes of data in such a way as to identify specific patterns and microcosms of communication throughout the organization. Corman et al. (2002) point out that this situation is not unique to the Browning and Beyer study. They suggest that Barley‘s (1986) qualitative study of technological change in two hospitals could be significantly strengthened by a comprehensive examination of discourse patterns inherent in these organizations.
―Detecting and describing complex patterns spread out over a vast field of discourse may well be too difficult a task for informants, or for human analysis of accounts and residual
practices of social discourse can we come to understand collective level social constructions.
Thus, following the logic presented above, Study II of this dissertation has incorporated the use of Centering Resonance Analysis to extend the grounded theoretical findings in Study I by examining organizational discourse in the context of social computing, competitive dynamics and firm performance within the focal firm, FCI.
Study I of this dissertation identified four specific categories and explained how information systems are being used by top level managers in engaging in competitive actions and responses toward firm performance. Additionally, Study I identified the importance of collectivity in terms of information flow and decision-making in the organizational context. However, grounded theory as a methodology is not equipped to identify minute patterns among words, or to give precise and calculated information about the significance of concepts at varying levels of significance. This study will be the first to create a synthesis between grounded theory, social network analysis and centering resonance analysis in the information systems or competitive dynamics streams of literature.