Introduction to Coding and Thematic Analysis
Coding was based on a thematic analysis, which refers to the process of coding the data and grouping connected codes into themes.
This process is common in qualitative research (Maykut and Morehouse, 1994; Attride-Stirling, 2001; Mason, 2002; Patton, 2002; Ryan and Bernard, 2003; Roulston, 2009; Marshall and Rossman, 2010).
The first step was to apply codes to selected parts of text from interviews, textual data and notes taken from watching the videos of dance.
As Roulston (2009, p. 151) suggests, ‘codes are labels that researchers apply to sections of data . . . that represent some aspect of the data’.
The first category of codes was generated from the research questions, the theory and the literature; as I analysed the data though, new codes started to emerge.
The former type of codes have been referred to as ‘theory-generated codes’ (Marshall and Rossman, 2010, p. 209) or ‘sensitising concepts’ which ‘have their origins in social science theory, the research literature, or evaluation issues identified at the beginning of a study’ (Patton, 2002, p. 456).
Emic vs Etic Codes
The latter codes, emerging from the data, were emic codes. These, according to Patton (2002, p. 454), are ‘key phrases, terms and practices that are special to the people in the setting studied. . . . Anthropologists call this emic analysis and distinguish it from etic analysis, which refers to labels imposed by the researcher’.
Hence, the codes I applied to the data were a combination of etic and emic and the analysis I adopted was both inductive and deductive, as it was based on the continuous interaction between theory and what was emerging from the data.
As Patton (2002, p. 453) explains, ‘inductive analysis involves discovering patterns, themes, and categories in one’s data . . . in contrast to deductive analysis where the data are analyzed according to an existing framework’.
This connects with the stance I highlighted at the beginning of this chapter (4.2) of my use of a sensitizing conceptual framework, which guided my gaze, while the analysis was guided by the data as they emerged.
The codes applied to the dance video analysis were slightly different from those applied to the rest of the data, as dance videos contained a lot more references to the types of movements which were seen in the dance, rather than on discursive elements.
Coding Dance Feelings, Settings, Transmission, Costumes and Props
However, some of the codes applied to all sets of data and these included, for example, codes referring to feelings, settings for the dance, sociocultural values, transmission or the use of costumes and props.
At the same time as coding the data, I carried out what Ryan and Bernard (2003, p. 94) call ‘cutting and sorting’, which ‘involves identifying quotes or expressions that seem somehow important and then arranging the quotes/expressions into piles of things that go together’.
In order to ‘cut’ the important quotes, I used the reference manager programme called Zotero.
Using Zotero in Qualitative Research
I copied and pasted relevant quotes into notes (a feature in Zotero) and then I tagged these notes with the relevant codes.
I then created Zotero reports from these tags, which I saved as independent files and grouped in folders according to the themes they went under.
The process of arranging codes into themes was more physical and it involved the use of paper and scissors.
I first listed all the codes (labelling the codes according to whether they were emic or etic) using an Excel spreadsheet.
I then printed the spreadsheet, cut all the codes in the cells into individual pieces of paper, laid these pieces of paper on the floor, and started grouping them together into themes on A2 paper sheets.
As codes were grouped together, themes emerged.
Finally, I created a new Excel sheet with the themes in the first row and, in the columns under each theme, the codes that went under it.
The process of using paper and scissors to process qualitative data is documented in the literature.
As well as Ryan and Bernard (2003) cited above, for example, Maykut and Morehouse (1994) mention the practice of photocopying interview transcripts, field notes or documents, coding on the margins of the pieces of paper and then cutting them and rearranging them on index cards.
I did this part electronically, using Zotero to code and arrange the meaningful text.
However, I resorted to paper and scissors at a later stage to cluster codes into themes.
As Ryan and Bernard (2003, p. 95) suggest, ‘investigators . . . will need to address the issue of which themes are the most important and worthy of further analysis’.
The process of writing the dance analysis chapter, helped me to do this. As the dance analysis formed the core of this research, I decided to write it first, in the form of a history of raqs sharqi in which formal, physical and sociocultural elements of dance interact.
Process of Re-Coding and Re-Analysis
Going through the dance analysis helped me highlight the most relevant themes, so I then searched the data again, in particular from interviews, to see if there was anything that could support, contradict or complement my findings.
During this process, I connected with the wider sociological and heritage issues, which emerged both from the literature and the data.
During the analysis, in order not to lose focus, I always kept an eye on my research questions, which I had printed out and placed on my desk, as Mason (2002, p. 160) suggests, ‘if you literally keep your research questions nearby . . . you can make sure that you are constantly cross-checking between them and your data in the process of developing and applying categories’.
Next Page >> Presentation of the Dance Research Data