4 Chapter Four: Analysis of the data
4.3 Data analysis organisation
4.3.1 Coding and categories
All the interviews for this study followed a qualitative semi-structured design. The analysis procedure for each interview was identical. The interview analysis process entailed the following 14 stages post-interview translation and transcription:
• Complete general reading of the interview to gain a sense of the whole.
• First general coding for categories.
• Re-examination of interview to develop and create additional new categories.
• Creation of node graphs for each interview participant and analysis of category responses.
• Clustering of categories into themes and identification of main theme with their sub-themes.
• Revisiting the data for further possible new and unexpected categorisation to emerge.
• Creation of matrices for all main themes with sub-themes and general analysis.
• Creation of classification sheet from the values of the attributes assigned to each node.
• Query runs from data text and creation of diverse graphs for further analysis.
• Cross and comparative analysis of main themes and sub-themes from each of text analysis, matrix analysis and classification sheet. • Creation of diverse matrices and comparative analysis.
• Development of concept maps for interconnecting features. • Summary of all findings.
• Creation of visual summary finding figure/s.
The interview questions for research participants were identical within each specific data set. Sensitive language use concerning non-ambiguity in meaning, clearly formed
questions and the level of researcher–respondent interaction understanding was always carefully addressed (Smyth, 2006). This is necessary if the findings of the comparative analysis are to be considered dependable and to avoid increasing numbers of variations in the type of responses. Nevertheless, expansion of questions was often undertaken and is also a major characteristic of qualitative semi-structured interview procedures enabling intensive probing into participant responses when considered helpful to understanding and clarification for both the researcher and interviewee (Silverman, 2006). In addition, it allowed interviewees ‘the space to talk’ (Rapley, 2004, p. 25) and this led to unexpected information which required further categorisation and hence new node creation directly from the data.
4.3.1.1 Teacher interviews
The first stage of analysis (Figure 4.2, Box 1) involved the advanced coding of the existing broad and subcategories that were derived from the sub-questions, dimensions and protocol questions which had been developed in light of the pilot project findings. The categories were used for the pre-analysis of the interview and questionnaire data before installation of the NVivo software. For example, the category L2 (English) teacher training, which was investigated during the teacher interviews, contained the subcategories English competence and English teaching skills. New additional coded data could be placed into an existing
category where appropriate and new categories were created when necessary. For example, although the subject integration category contained the subcategory time factors, closer examination identified lesson priority as a feature of teacher interview responses concerning lesson planning of English integration into subject content. Hence the new subcategory lesson priority was created from the interview responses. The two separate interview responses involving the same question, ‘When does English integration take place during the week?’, but undertaken on different days and individually with the case study teachers Clara and Helen are displayed in Extracts 4.1 and 4.2.
Extract 4.1
Clara: When there is time, when I have to exclude something
then it is English. Well, I am doing English today now … in
between the lessons … to relax the lessons and give the children a break between subjects and so forth, ten minutes from the
book.
Extract 4.2
Helen: I have two thirty minute lessons weekly … I do thirty
minutes of German and thirty minutes of English, twice weekly. However, when it is inconvenient then I teach it on another day,
it varies … let me think. Naturally when I organize a MOFF
*
it must be good. Half an hour is then not enough time.Sometimes we have more than one hour of English in the week.
(Source: Teacher interviews)
(MOFF*Monats Treff – Monthly Meeting, where class projects are presented to peers and parents. In this interview the teacher talks about English presentations)
Note: The original German text of all extracts displayed in this study can be observed in Appendix E. Tables of each extract source can be viewed in Appendix F.
The interview protocols assisted in the identification and guidance for some prior category node creation for the analysis in advance of the actual data processing. The category ‘tools’ is one example of pre-node creation. Classroom ‘tools’ and teacher perceptions of them is related to teacher perspectives and investigation was undertaken during the data collection of the teacher interviews. An example of nodes coded from the interview data of the case study teacher Helen is shown in Figure 4.3.
Figure 4.3 The coded interview nodes of case study teacher Helen
NVivo allows researchers to create visual graphs of all contents coded at specific nodes for each individual source, which assisted linking of ideas and comparisons between teacher
participant responses for the analysis. Therefore, by clicking on a node, immediate retrieval of all data coded at the node for each individual teacher could be viewed.
After coding the entire teacher interviews further finer coding was undertaken. This entailed revisiting data sources and renaming categories when new or divergent information arose.
Six main categories developed as the analysis proceeded and six parent nodes were created with relating subcategories, as displayed in Table 4.1.
Table 4.1Teacher interviews with the main theme categories and their subcategories