4 Methodology and research design
4.7 Introduction to data analysis
4.7.2 A coding-process
Above I referred to interpretations of collected data as crucial in the gradual development of the case in my study and processing of data. In- terpretation was highlighted already in Section 4.1 where I argued for working as an interpretive researcher studying and supporting develop- ment in mathematics teaching. Interpretation is by Flick (1998) denoted as theoretical coding:
Interpretation is the anchoring point for making decisions about which data or cases to integrate next in the analysis and how or with which methods they should be collected (Flick, 1998, p. 179).
The mentioned maps of data were sorted in different ways, on schools and teachers, kinds of events, such as computer lab visits and workshop sessions, and the kinds of tools that were used such as Excel, Cabri and compasses. Eventually these maps also included what I consider as being codes, at the start some brief headlines indicating the content of the events. These first codes evolved based on the content of the events and the content’s interplay with the research questions. An example: Re-
search Question 2 emphasises characteristics of teachers’ orchestrations of Cabri use. When the teachers reported from their teaching with Cabri in workshops, they expressed several times how they experienced a need for assisting their students to use Cabri accurately. Observations in Ja- kob and Frode’s lessons exemplified how the teachers, typically in a step by step way, guided their students to accomplish a construction. Appen- dix 6b gives an overview of some codes I had developed until June 2006, where teachers’ emphasis in the example above was coded as tool skills.
Now I briefly want to relate my coding process to Flick’s (1998) de- scription of theoretical coding where he argues that interpretation is a three stage process: open coding; axial coding; and selective coding. Based on his description, open coding was typical for my initial work with data and transcripts where expressions in the events were classified based on the content. I developed a number of disconnected codes rele- vant to the research questions. I noticed teachers’ comments about issues like time, own lack of skills and experience in using computer software, concerns about students’ inaccurate use of Cabri, and expressed impor- tance of designing tests and have available good tasks. Three of my codes were time, lack of access and students’ tool skills. All the codes, numbered 1-16, are presented in Appendix 6b. The codes are not merely expression in the events but codes which reflected my research ques- tions.
Adjustment of codes in new categories, like my categorisation of teachers’ comments on students’ inaccurate use of Cabri as a focus on students’ tool skills, seems close to what Flick denotes as the axial cod- ing. In what I regard as my axial coding process, categories of data were sorted for the analysis, and I considered more extensively how the data and the codes could contribute to address the research questions. Data coded time and lack of access were categorised as highlighting issues in teachers’ implementation process, and they are indicated as codes in the category named a) in Appendix 6b. This categorisation process with codes also stimulated further development and refinement of the research questions.
In accordance with my focus on teachers’ implementation and or- chestration of Cabri-use, I developed two main categories of codes named a) and b) (see Appendix 6b. The two categories are dominated by respectively teachers’ concerns about issues in the implementation proc- ess, and students’ lack of utilising the potential of Cabri and how the teachers responded to this. I also considered relationships between these two categories of codes since teachers in sessions often paid attention to both implementation and their use of Cabri in teaching. Utterances close in time or the very same utterance often contained data which was coded in both categories, such as Jakob and Frode’s design and use of Cabri-
tests in all classes at Grade 8 which worked as a criterion for implemen- tation of Cabri and energised their orchestration of Cabri-use in teaching. The development of new categories of codes and considering relations, are according to Flick also typical for the selective coding process as a further step from the axial coding.
The development of codes, evolving based on study of my earliest made transcripts, in fact helped me afterwards when I made data reduc- tions of new events, in selecting which parts I wanted to transcribe and in orienting the coding of new transcripts. In the data reductions, the right column with brief description of the content includes use of some of the codes indicated in Appendix 6b. In Chapters 5 and 6, altogether 15 episodes from this data reduced session are considered in the analyses. One of these episodes is presented already in the coming section (4.7.3) to illustrate how I present transcripts in this thesis.
An obvious limitation with my coding process is that the interpreta- tion process with development of codes obviously could have been done in many ways. The same data could “produce” many different emphases based on different aims of the research. In this thesis I try to remain open to the reader and include descriptions and short comments to each of the events in the case study and how I processed with the data (Appendices 1 and 6). The maps and codes were very helpful in the early phase of de- velopment of the analysis chapter. However, the coding went on along- side the writing where analytical findings helped me in reorganising sec- tions and chapters. Consequently, the end of the coding process is im- plicitly visible in the design of Chapters 5 and 6 and their different sec- tions.