Tentative Research Questions
4.15 Open, Axial and Selective Coding
Open, axial and selective coding procedures were carried out, progressively, during the course of the research. Open coding was associated with open sampling which aimed to uncover as many potentially relevant categories as possible. This included the sampling of persons, places and situations with a view to gathering the most relevant data during the early stages of Phase I.
Axial coding followed open coding. The properties and dimensions of the data which had been broken down into categories during open coding were reassembled in new ways during axial coding using relational and variational sampling procedures. The phenomenon under investigation, the causal conditions associated with it, and the particular context in which it was embedded became much clearer during this stage. This resulted from the uncovering and validation of relationships between subcategories, categories and the phenomenon. This was possible as a result of answering the questions repetitively raised on the data gathered and using the appropriate purposeful action/interaction strategies deployed in handling the data already available and pursuing further data.
22xhe protocol was used for data analysis purposes. Using the protocol for the purpose of presenting quotations in this thesis was carefully considered and tried at first. However, this was changed. It was thought that it would be both unnecessary and confusing. Instead, the appropriate case's serial number only would be cited for reference purposes at the end of a quotation.
Action/interaction strategies were affected, however, by sets of intervening conditions, or "the broader structural context pertaining to a phenomenon the broader and general conditions bearing upon action and interaction strategies" (Strauss and Corbin,
1990: 103). These were not properties of the phenomenon but, rather, conditions outside of the phenomenon itself that nevertheless affected the way it was handled in either a positive or negative manner. Therefore, the researcher had to do his best in contending with the intervening conditions to secure collecting the best quality of data in the interest of the study, particularly during the axial and selective coding phases of analysis.
Four distinct analytical steps were simultaneously used during axial coding. These were:
1) Relating subcategories to a category by hypothesising relationships between them and the phenomenon.
2) Verifying resultant hypotheses against actual data.
3) Resuming the search for properties and dimensions of categories and subcategories. 4) Exploring variation on a preliminary basis in preparation for selective coding.
Selective coding followed axial coding. It was concerned with integrating the categories uncovered to from a set of grounded theoretical propositions. Five main steps were involved at this stage, but not necessarily followed linearly (see Strauss and Corbin,
1990: 117-118). These were: 1) Explicating the story line.
2) Relating subsidiary categories around the core category. 3) Relating categories at the dimensional level.
4) Validating those relationships against data.
5) Filling in categories that needed further refinement and/or development.
Once the core categoiy was identified, the next step involved determining its properties and dimensions and relating the other categories to it. Hence, the other categories would become subsidiary categories. Resultant propositions would then be checked or validated against the data which they fitted sometimes but not always. When they fitted the data they were maintained. When they failed to fit the data, however, further theoretical sampling was required in order to identify the reasons for the lack of fit with the data, with a view to validation of the resulting theory.
4.16 Summary
This chapter described the details of how the field research was conducted. It began with a description of the basis on which the companies researched were selected.
distinguished between primary and secondary companies, and explained about the progress of the research process and the main activities involved in it.
It also clarified important points regarding the interview schedules used and explained the measures taken to insure the validity and reliability of the main interview schedule. It gave a particular attention to the methods and procedures followed in data collection, coding, analysis and interpretation. Coding was described both in terms of data about the cases (i.e. concerning companies and individual actors) and data from the cases (i.e. as required to answer the tentative research questions and develop theoretical proposition through open, axial and selective coding).
Field work activities, interviewing in particular, proved to be an extended learning opportunity to the researcher. A great deal was learned about the ten companies, the change processes concerned, the organisational structures and processes that affected change in positive and negative ways, and the actions and interactions involved as well as the contextual boundaries surrounding them. It would be difficult to imagine that such a rich picture could be achieved using any other methodology.
One important thing should be said, however, about the field work methodology followed: it was a labour-intensive approach indeed in terms of the frequency of visits, transcription of interview materials, and coding and analysis of the resulting data which was voluminous and difficult to handle. The longer the time spent on a study like this, the more likely that the researcher would refine his/her understanding of the phenomenon under investigation and develop a more tested theory.
One important practical advantage of this methodology, nevertheless, became abundantly clear at the end. The researcher could afford to appear less intrusive than in the case of action research or participant observation by building trust and continuity into the research process progressively with time. The accumulation of such trust and continuity was very important in helping the researcher gain a greater degree of access to "protected" data that could further improve the quality of his understanding and thus that of the resulting theory.