CHAPTER 3: THE CODING AND CATEGORIZATION OF THE 18-CASE SAMPLE (FIRST ANALYSIS)
3.4 The stabilization of the classification structure and the conceptual saturation
The first 6 iterations consisted of analyzing the content of the 18 cases progressively (3 cases in every iteration) through open and axial coding consisting of: (1) the creation of new themes, (2) the merging and grouping of detailed themes, (3) the splitting of general themes, and (4) the redistribution of themes among the large categories created (the 4 blocks in the model and the Strategic Motivations and Free Nodes blocks).
While most themes in the final model were created at the end of Iteration 1 (in the analysis of the first 3 cases), a large number of adjustments and restructurings were done subsequently throughout the other 4 iterations to stabilize and simplify the model (axial coding).
A lexicon of themes was developed and maintained throughout the analysis process to constantly be reminded of the definition of the themes (categories) used and their conceptual boundaries.
References to the literature were also needed to clarify the definitions.
To simplify the analysis process, the material was modularized in four chunks and addressed progressively through four iterations. Each chunk contained the central interviews of three cases and was analyzed in a separate iteration.
Even though the selection of the cases grouped in each chunk was not a critical task in the process, the groups were not completely random. Instead, the cases picked for the first iterations were those felt to provide more pertinence, richness, and nuance.
It is important to note that the numbers in the tables below do not represent the occurrences of themes in the material. Instead, they represent the number of themes (sub-categories) in each major category of the model. In this critical qualitative analysis phase (Phase 4), the number of occurrences of the themes is not important, only the meaning and quality of the themes is analyzed, and whether the themes exist in each case or not. The number of times the theme is
found in each case is not important at this stage. This table shows the number of themes in the classification structure before and after the adjustments in Iteration 1:
Table 3-4 Number of themes before and after Iteration 1 adjustments
Iteration 1 Themes Before Adjustments After Adjustments
Theme blocks 1A 1B 1C TOTAL 1A 1B 1C TOTAL
The columns show the number of themes created during the analysis of each one of the first three cases (1A, 1B and 1C) for every block in the model. After Iteration 1 adjustments (columns on the right), the total number of themes was brought down to 59 from 70 and many themes (categories) were merged to create more inclusive and relevant ones. A total of eleven themes were removed and their content re-coded with other more general themes. Below are a few examples of the adjustments that led to the reduction of the number of themes in Iteration 1:
Table 3-5 Examples of Iteration 1 adjustments
Theme Iteration Blocks impacted Adjustments
1. Technical Difficulty - B: BANK - Project Governance
- Removed
- Content coded in Changes and Complexity
- Content coded in Team Selection Process and IT Roles and Competences 3. Offshore
- Content coded in Externalization (Outsourcing)
- Content coded in Contract Management
- Content coded in Project Team 6. Partner Selection
- Content coded in Consultant Selection Process
For all adjustments in Iteration 1 see Appendix 6. Some themes were renamed and some moved from a block to another. The other nine interviews were analyzed in the subsequent three
iterations and triggered other important changes in the classification structure / conceptual model. In Iteration 2, the number of themes increased to 70 before adjustments and then decreased to 65 after adjustments. In Iteration 3, themes reached 67 before adjustments and fell to 64 after adjustments. In Iteration 4, one theme was added and one removed maintaining the total number of themes at 64. In Iterations 5 and 6 the classification structure remained unchanged and no theme was added or removed. This table shows the evolution of the number of themes throughout the first 4 iterations in the qualitative analysis process:
Table 3-6 Evolution of the theme structure from Iterations 1 to 6 Evolution of the theme
structure in Iterations 1 to 4 Iteration 1 Iteration 2 Iteration 3 Iteration 4 Iteration 5 Iteration 6
Theme blocks
Before After Before After Before After Before After Before After Before After
1: Project Governance 23 17 18 16 18 18 19 18 18 18 18 18
2: Intra-Organizational Governance
14 13 19 19 19 16 16 16 16 16 16 16
3: Inter-Organizational Governance
25 21 25 24 24 24 24 24 24 24 24 24
4: Innovation 8 8 8 6 6 6 6 6 6 6 6 6
TOTAL 70 59 70 65 67 64 65 64 64 64 64 64
Note that the following types of adjustments are not obvious in the above table:
- Renaming (and redefining) themes
- Moving themes from one block to another
- Re-coding previously analyzed content with new themes
The following graph illustrates the conceptual saturation process in the first six iterations.
Figure 3-1 Change in the number of themes in the first six iterations
The addition of the six last cases did not affect the themes in the model at all. In other words, no new concept was added to the model with the analysis of the last six cases. The progressive decrease in the number of adjustments (themes created or removed) is an indication of the stabilization and saturation of the conceptual model and in turn of the comparability of the cases analyzed. In parallel to the stabilization of the classification structure (and the conceptual saturation), the number of references to the themes increased in a linear way. This indicates that the themes exist throughout the cases and reinforces their comparability. This graph illustrates the variation in the number of references per block at the end of Iteration 6:
Figure 3-2 Reference variation per block at the end of Iteration 6
The average number of references per case at the end of Iteration 6 was 143 and the cumulative number of references for the 18 interviews was 2575.