Page 139 - Getting of the fence

The relevance and usefulness of the Comprehensive Approach
The videos were coded by three researchers. The coding instructions were first carefully discussed in order to resolve any discrepancies. The three researchers together coded several parts of lessons by different teachers by going through the coding manual step by step. Then a random lesson was selected and coded individually by one researcher. The coding generated a total of 212 time-stamps (units). The Mediacoder output was transferred to an xls file and the codes were deleted leaving only the time stamps. These time stamps were given to the other two researchers who independently coded each of the 212 time-stamps. An appropriate measure of intercoder agreement for more than two raters is Krippendorff ’s alpha, which takes into account the possibility of chance agreement and the magnitude of disagreement (Neuendorf, 2002). This was computed using a macro created for SPSS (Hayes & Krippendorff, 2007), with the analysis revealing a modest degree of reliability (α = .78, n = 212). After this procedure, the three researchers discussed any discrepancies and the coding instructions were amended where necessary.
6.3.3.2 Research question 2: Which changes regarding EFL literature teaching
did teachers perceive after working with the Comprehensive Approach for one
year?
In-depth, semi-structured interviews were conducted with each of the eight teachers 6 at the end of year 2 (between May and September 2017) in order to investigate
which changes the teachers experienced regarding EFL literature teaching after working with the Comprehensive Approach for one year. All interviews were conducted face-to-face in Dutch by the same researcher and lasted on average 45 minutes (range between 34 and 53 minutes). The interviews were recorded and then transcribed verbatim by an independent research assistant.
All transcripts were subjected to a qualitative, inductive analysis in which we followed the coding procedure as defined by Saldaña (2013) as first and second cycle coding methods. During a first reading of the transcripts, emerging impressions were recorded in memos and were used to distinguish specific themes. This was followed by a process of in-vivo coding (first coding cycle) where the data were further analysed to specify these themes, which resulted in initial codes. We selected in-vivo coding because, according to Saldaña (2013), in-vivo coding is appropriate for studies “that prioritize and honour the participants’ voice” (p. 91).
In the second cycle, the data were analysed further by re-examining the initial codes through pattern- and focused coding. Whereas pattern coding “develops the category label that identifies similarly coded data”, focused coding “categorizes
137