CHAPTER 5: METHODOLOGY
5.5 Data analysis
The purpose of data analysis in the study was to draw out important themes or variables that would enable the construction of the simulation. The transcripts of the interviews, observation notes, and document analysis were accordingly analysed. Qualitative data analysis can be laborious; therefore, to assist with the data, the researcher concurrently engaged in data analysis and interpretation, and collection and theory development (Irvine & Gaffikin 2006; Creswell
1994). I also endeavoured to balance objectivity and sensitivity in the data analysis, considered by Strauss and Corbin (1998) to be essential. Objectivity ensures that the researcher feels confident that the findings are presented in a neutral and reasonable manner, whereas sensitivity enables creativity, resulting in novel theory. This was of the utmost importance in the study context.
The following principles of qualitative analysis, as presented by Tesch (1990: 95) were followed in this study:
• Analysis is not the last phase in the research process; it is concurrent with data collection or cyclic.
• The analysis process is systematic and comprehensive, but not rigid.
• Attending to data includes a reflective activity that results in a set of analytical notes that guide the process.
• Data are ‘segmented’, i.e., divided into relevant and meaningful ‘units’, yet the connection to the whole is maintained.
• The data segments are categorised according to an organising system that is predominantly derived from the data themselves.
• The main intellectual tool is comparison.
• Categories for sorting segments are tentative and preliminary in the beginning; they remain flexible.
• Manipulating qualitative data during analysis is an eclectic activity; there is no one ‘right’
way.
• The result of the analysis is some type of higher-level synthesis.
The above process is briefly elaborated on.
Data analysis involved data organising and data interpretation (Tesch 1990). This meant that the text had to be separated into segments and consequently sorted into groups. Text segments were cut out of their context, such that the meanings were retained, and also that those segments had a likely link to the study. I gained a holistic understanding by carefully reading all documents, and noted any thoughts that came to mind (Tesch 1990; Luna-Reyes & Andersen 2003).
Organising took shape through a combination of the theoretical framework and research questions, and the data (Tesch 1990). As Hannabuss (1996) notes, the information can be organised and analysed in a deductive and inductive manner. The deductive manner implies
verification of known common principles, whereas the inductive manner involves offering proof from which to infer the reality of such principles. Generated theory described themes or concepts that emerged from comparing the texts (Luna-Reyes & Andersen 2003).
As recommended by Sofaer (2002) I was careful in distinguishing between actual observations and likewise interviews, as opposed to my interpretations thereof. Short descriptions of the topic, which refers to what is discussed or written, and not to the content which is about the essence of the message, were made next to the relevant text (Tesch 1990).
To identify the topics, I commenced with a few documents and made up a list to draw out the topics (Tesch 1990). Refining occurred as the analysis proceeded, and this resulted in the materialisation of categories. Data sorting occurred through tagging text segments with information regarding the category. This simply meant that I assigned abbreviations or codes to the categories. Thereafter, things that fell into a certain category were assembled there in a way which made sense.
Patterns, definitions, narratives and messages were searched for in the various interviews during the analysis (Luna-Reyes & Andersen 2003). Themes or concepts also surfaced as I became familiar with the data, and began formulating logical links to the interview questions, while also considering important details from the literature review (Bowen 2005). Reviewing the secondary data also contributed to a better understanding.
I followed practical advice, provided by Ryan and Bernard (2003) to identify the themes or concepts. I searched for repetition in the text by noting when a certain idea appeared more than once. This was considered most likely to be a theme. I also looked for terms which were not well-known or which were utilised in a less obvious way. I was also alert to the use of metaphors and analogies, as this was a way for the respondents to express their feelings.
Themes also emerged from natural shifts in content. Text was also analysed in terms of similarities and differences that were detected. I also noted causal relation when respondents employed words such as ‘because’, ‘since’, and ‘as a result’ and kept an eye out for missing data. Finally, I also considered theory-related material but attempted to not allow prior theorising to hinder the development of fresh perspectives.
I essentially thus compiled each of the interview transcripts into one document in the order in which the interviews occurred. I then carefully read and re-read first through each interview and thereafter through the whole document, and added in ‘comments’ which were essentially themes.
The results of the analysis of the interviews are presented in Chapter 6.
I then focused on the construction of the simulation, once all the data from the interviews had been analysed to determine the issues.