During the problem-solving phase, we followed the cyclic process of diagnosing, action planning, action taking, evaluating and specifying learning (Susman and Evered 1978). Accordingly, we used a diagnostic mapping technique suggested by Lanzara and Mathiassen (1985) to analyze the qualitative data from workshops, meetings, and field-observations, and quantitative data from EMR reports, questionnaires, and other sources with the goal of diagnosing EMC’s problems relating to the revenue cycle (see description of the technique in Chapter 7, and outcome of the analysis in Appendix A). This analysis enabled planning and execution of appropriate interventions (Chapter 10), evaluation and interpretation of the interventions (Chapter 11), and specifying the contributions (Chapter 12). The analysis also helped to illustrate the theoretical framework that emerged from the study.
As discussed in Chapter 6, the goal of data analysis during the research phase of our engagement at EMC was to develop new theory. Towards that goal, we followed data analysis procedures suggested by Miles and Huberman (1994) for qualitative case data. They suggest three concurrent flows of activity: data reduction, data display, and conclusion drawing and verification. Furthermore, all three activities take place not only after data collection is finished but continuously throughout the data collection process. In fact, we continued to collect data even as we developed our theoretical framework. For example, the most recent visit in June 2011 provided further data about the heterogeneity of technology configurations at EMC (see Chapter 9 for details and Chapter 10 for illustrative examples). This additional data provided support for a key component of our framework, and facilitated empirical analysis of the case. This
Singh | Dissertation | DATA COLLECTION AND ANALYSIS 104 concurrent and interactive pattern of data collection and analysis helped determine subsequent data collection choices for theory development and facilitated iterative development of the proposed theoretical framework in this research.
Figure 8.2-1 represents the data analysis strategy during the research cycle. We leveraged the outcomes of the problem-solving phase to develop and evaluate our theoretical framework (McKay and Marshall 2001). Evaluation of the outcomes of the problem-solving phase also led to reinforcement and modifications of the theoretical framework (Baskerville and Pries-Heje 1999).
Figure 8.2-1 Data Analysis Strategy for Theory Building
Data Analysis Strategy – Adapted from Miles and Huberman (1994, p12)
8.2.1 Data Reduction
Miles and Huberman describe data reduction as “the process of selecting, focusing, simplifying, abstracting, and transforming the data that appear in written-up field notes or transcriptions” (1994, p10). They assert that data reduction occurs continuously throughout the life of any qualitatively oriented research project. It can start even before fieldwork commences—through initial research questions and conceptual framework from which the researcher operates, and by the site selection and initial data collection choices made by the researcher. As data collection
Data reduction Data collection
Data display Drawing conclusions
Analyze Data Collect Data
Singh | Dissertation | DATA COLLECTION AND ANALYSIS 105 proceeds, further episodes of data reduction occur through writing summaries, coding, teasing out themes, and writing memos.
Following this strategy, the process of data reduction and transforming began when we were working with a group of rural hospitals and community health centers in Georgia in 2007 and selected EMC for a deeper collaboration. During our two-year engagement at EMC, data reduction occurred through presentations for bi-monthly steering committee meetings, summaries of monthly problem-solving team meetings, and communications (such as in Table 7.3-1). Further reduction occurred through identifying problems relating to EMC’s revenue cycle, teasing out practical themes (for example, complexity of revenue cycle), as well as through the evolving application of theoretical frameworks—the transactional approach to information management, multiple centers of decision making, and context-dependent governance—to make sense of the observations and experiences at EMC.
8.2.2 Data Display
Data display refers to “an organized, compressed assembly of information that permits conclusion drawing and action” (Miles and Huberman 1994, p11). Data displays may include matrices, graphs, charts, and networks—all designed to assemble organized information into immediately accessible form. Like data reduction, the development of data displays is an iterative process that occurs during the data collection process as well as after its completion. Accordingly, we created appropriate displays, including tables, graphs, and flowcharts (such as Figure 2.1-1 and Table 2.1-1), that helped us to understand the complexity of a hospital revenue cycle and framed our understanding of the overall workflow and individual activities related to revenue cycle. Table 2.4-1 helped to understand the role of IT in a hospital revenue cycle and to appreciate the heterogeneous technologies prevalent in any hospital. Another example of data display is the diagnostic map (refer A.1 in Appendix A), which helped in identifying opportunities for solving problems related to information management in EMC’s revenue cycle. These data displays developed iteratively, based on the improved understanding of the research team during the two-year engagement and feedback from key stakeholders.
Singh | Dissertation | DATA COLLECTION AND ANALYSIS 106 8.2.3 Drawing Conclusions and Verification
Drawing conclusions includes identifying regularities, patterns, explanations, possible configurations, causal flows, and propositions from available data (Miles and Huberman 1994, p11). As data collection progresses, these conclusions gradually become more explicit and grounded (Glaser and Strauss 1967), and “final” conclusions may not appear until data collection is over (Miles and Huberman 1994, p11). Miles and Huberman (1994) point out that it is important to iterate between drawing conclusions and verifying those conclusions in an ongoing process to maximize the validity of the study’s findings.
The conclusion drawing and verification phase of data analysis occurred during both the problem-solving cycle and the research cycle. During the problem-solving cycle, a thorough diagnostic mapping exercise (refer A.1 in Appendix A) provided a framework to identify key problems relating to each stage of EMC’s revenue cycle by asking (and answering) the four questions: “what happened,” “why,” “what are the consequences,” and “what can be done.” By adopting this structured problem identification exercise, the research team made sense of the problem situation at EMC and was able to provide an initial diagnosis and propose interventions to improve EMC’s revenue problems (see Table 7.3-1). These recommendations reflected the conclusions drawn from the interactions during the initial workshop in March 2008 and subsequent meetings with key stakeholders during visits to EMC. During the planning and execution of each intervention (A through J, see Figure 10.2-1), the research team collected additional data, conducted data analyses using content analysis of transcribed interviews and workshops; analysis of field notes, e-mail communications, archived electronic and paper-based documents; and by using statistics, tables, graphs, and numbers (Denzin and Lincoln 2005, p7). These data analyses confirmed and contextualized the problem situation and helped to fine-tune each intervention based on feedback and review of initial outcomes.
Based on ongoing reflective discussions and the de-briefing sessions between the two researchers about the observations at EMC (“what does it mean?”), we began to make more sense of the context in which EMC’s revenue cycle operated. In addition, following the recommendations of Eisenhardt (1989), Boyatzis (1998), and Yin (2003), we repeatedly read the transcripts, interview notes, and other material to identify key themes relating to the challenges of information management in EMC’s revenue cycle. I also did write-ups for each intervention (A through J),
Singh | Dissertation | DATA COLLECTION AND ANALYSIS 107 provided detailed contextual information about the problem situation, explained how the research team approached possible solutions, and described the outcomes of the intervention. Based on this iterative, within-case analysis (Eisenhardt 1989), we developed a preliminary list of themes (“premises,” see section 9.1) that represented underlying patterns across the interventions. These four premises helped us to tie the case data directly to the study’s research questions and provided a foundation for developing theoretical concepts (Miles and Huberman 1994, p70). Based on these premises, we iteratively identified related components of the new theory (see section 9.2 and Table 9.2-1), while ensuring that the case data provided sufficient evidence for each component. Thus, conclusion drawing in the research cycle led to theory generation, which I discuss in Chapter 9.
In Chapter 10, I verify the applicability of the components of the theoretical framework in the context of a health delivery transaction at EMC. I also verify the applicability of the framework to our interventions to improve information management in EMC’s revenue cycle (see Section 10.2).
Singh | Dissertation | ►PART D: THEORY DEVELOPMENT 108
►PART D: THEORY DEVELOPMENT
This section describes the development and application of the theory and draws on the research team’s efforts to improve information management at EMC.
• Information Polycentricity Framework (Chapter 9): This chapter draws on Polycentricity Theory, Transaction Cost Theory, and existing theories of information management to develop an initial conceptualization—Information Polycentricity Framework (IPF)—that can help to explain the challenges of information management in complex organizations and lay the foundation for further theoretical development. The chapter discusses the premises and components of IPF.
• Application of IPF (Chapter 10): This chapter illustrates the detailed workings of IPF by applying its four components to information management in the revenue cycle at EMC. Next, it provides a detailed contextual account of ten interventions (grouped into four cases) to improve EMC’s revenue cycle as part of the action research study. For each case, I interpret the findings based on IPF.
Singh | Dissertation | INFORMATION POLYCENTRICITY FRAMEWORK 109
9
INFORMATION POLYCENTRICITY FRAMEWORK
In this chapter, I develop a new conceptual framework—the Information Polycentricity Framework (IPF)—to explain the challenges of information management in complex organizations and to lay the foundation for further theoretical development. First, I draw on Polycentricity Theory, Transaction Cost Theory, and existing theories of information management to present the premises of IPF. Subsequently, I discuss the key components of IPF.