Impactos de los TLC sobre los países miembros

In order to analyse and present the findings of the study, I adopted a thematic analysis process, which ‘organizes and describes your data set in (rich) detail’ and ‘offers an accessible and theoretical flexible approach to analysing qualitative data’ (Braun and Clarke 2006:79). I followed the six phases of thematic analysis that Braun and Clarke (2006:87) suggest:

1. Familiarizing yourself with your data: Reading and re-reading the data, noting down initial ideas.

2. Generating initial codes: Coding interesting features of the data in a systematic fashion across the entire data set, collating data relevant to each code.

3. Searching for themes: Collating codes into potential themes, gathering all data relevant to each potential theme.

4. Reviewing themes: Checking if the themes work in relation to the coded extracts and the entire data set, generating a thematic ‘map‘ of the analysis. 5. Defining and naming themes: Ongoing analysis to refine the specifics of

each theme, and the overall story the analysis tells, generating clear definitions and names for each theme.

6. Producing the report: The final opportunity for analysis which consists in the selection of vivid, compelling extract examples, final analysis of selected

123

extracts, relating back of the analysis to the research question and literature, producing a scholarly report of the analysis.

As suggested by Braun and Clarke’s phases of thematic analysis (2006:82), before defining the themes that ‘capture something important about the data in relation to the research questions’, it is necessary to start with the coding of data. Walliman (2011:217) states that a valuable step when analysing is to organise the amount of data and clarify the relationships among concepts ‘by identifying differences in it and thereby forming subgroups within [a] general category’. As Richards (2003:273) declares, ‘analysis depends on identifying key features and relationships in the data, something that is difficult if not impossible unless some degree of order is imposed’. Furthermore, ‘all the qualitative coding techniques are aimed at reducing or simplifying the data while highlighting special features of certain data segments in order to link them to broader topics or concepts’ (Dörnyei 2007:250). For this, the development of a coding system is important since it facilitates the organisation of data and the analysis of important aspects. Before continuing, it is worth stating that for the purpose of this work, I considered Themes and Categories as synonyms. I name one or the other depending on the way it is referred by the researcher being quoted.

Due to my limited experience in research and the coding system, and having in mind the importance of analysing and organising my data in order to define the themes and interpret them (Wolcott 1994), I made the decision to be extremely systematic in the coding process and the generation of themes. Hence, I backed up and complemented Braun and Clarke’s (2006) phases with the processes suggested by other researchers. The process that I followed while coding to generate themes is described below (see Appendix 6 for samples of the coding process).

124

Dörnyei (2011) suggests a Pre-coding stage (which coincides with Braun and Clarke’s first phase of the thematic analysis) and as a preparatory move, in order to make sense of data and re-familiarise with it. I read and re-read the journals and transcripts, in order to reflect on them and note down my initial thoughts in my journal and memos (see Appendix 6, Figure A). Even though I had done this (as part of my preliminary analysis during my AR cycles), I believed that it was useful to do it again in order to see data from a different perspective and in a more complete and systematic manner (compared to the way I did it for the preliminary analysis).

The following step was to start the generation of initial codes (Braun and Clarke 2006; Richards 2003), in order to engage with data in a more detailed manner (see Appendix 6, Figure B). I followed an ‘open mind’ and ‘free coding’ of it to avoid ‘premature commitment to particular categories’ (Richards 2003:273). Richards also asserts that this process helps the production of labels, which suggests possible lines of organisations and will derive into future categories. As suggested by this author, I did this initial coding line by line. I highlighted passages relevant for the topic and added an informative label in the margin while I was reading (Dörnyei 2011). It was essential at this stage to keep an open mind and to produce as many new ideas and codes as possible, and it was also necessary to summarise the data (Bryman 2012). The labelling of data included ‘some key words from the actual passage to make the preliminary codes more authentic’ and as explicit as possible (Dörnyei 2011:251) (see Appendix 6, Figure B).

After the initial coding, I started collating codes into potential themes (Braun and Clarke 2006) in order ‘to go beyond a mere descriptive labelling of the relevant data segments’ (Dörnyei’s Second-level coding 2011:252) and to look for patterns and identify more abstract similarities that emerge across the data (see Appendix 6,

125

Figures C, D and E). Selecting potential extracts was done after that. As Dörnyei states,

One way of launching second-level coding is to go through several respondents’ accounts and list all the codes that we have identified in them. There will inevitably be some similar or closely related categories, which can be clustered together under a broader label. At this point we need to look at all the specific extracts that are linked to the newly formed broader category to decide whether the new label applies to all of them or whether some may need to be recoded. If the majority of the extracts fit the new system, this can be as a sign of the validity of the code. […] In some studies this process is iterated more than once.

(Dörnyei 2011:252)

During the process of defining themes, I considered constant comparison, which is described as one of the terms of grounded theory by which connection between data and conceptualisation is maintained (Richards 2013; Bryman 2012). I visualised this process as a ‘zigzag’, which means constantly going back and forth from a selection of data to another selection of data, and from one category to another. This idea made me decide on using the grounded theory coding system (open, axial, and selective) in order to categorise my codes and organise the information to be presented, but mainly to make connections between categories (axial coding) and systematically relate them to other categories (selective coding) (Braun and Clarke’s suggestion for reviewing and naming themes, 2006) (see Appendix 6, Figures F, G, H and I). I agree with Richards (2003:18, quoting Hemmersly 1984:60–62), who states that probably ‘the greatest attraction of grounded theory is that it offers a systematic way of analysing and interpreting the data, normally a messy and frustrating process’. My intention in using the constant comparative approach was to analyse the data collected (from different instruments) and compare them to each other in order to find representations of categories or themes that emerged, as well as relationships.

126

I used various sources for categorisations: memos, notes, and ideas from readings (Richards 2003). Walliman (2011:219) notes that we use memos ‘to explore links between data and to record and develop intuitions and ideas’. Richards (2013:278) adds that memos serve to draw ‘attention to points that might be relevant to the analysis’. While categorising, I kept in mind the essential features of an adequate category described by Richards (2003:276): the categories that emerge need to be analytically useful, empirically relevant, practically applicable, and conceptually coherent. Finally, I made sense of all my data and interpreted them according to the results that emerge during the analysis, but also linked them to the literature and other studies in the area. The whole process helped me name the definite themes and report of findings (Braun and Clarke final phase of thematic analysis 2006) in Chapters five and six.

The techniques that I used during my analysis included printed data in order to highlight relevant information, and written codes and initial categories or themes in the margin, using variously coloured pens. I also used filing cards, written memos, and my research journal. Additionally, I used the computer-aided qualitative data analysis (CAQDAS) NVivo, to help me organise codes and relevant pieces of quotes (extracts) into themes. Using printed data and doing the selection of codes and themes electronically meant double work for me and was time consuming; however, I am the kind of person who works better with printed versions of the information or data. As Wolcott (2001:43) observes, when researchers manipulate data manually rather than electronically, it helps them ‘visualise processes partly hidden by technology’ and they get a ‘physical feel for what they are trying to accomplish’. Hence, I preferred doing it this way first and then using software that helped me organise and store my data, making it easy to identify themes, memos, extracts, and all the information resulting from the analysis. It is worth saying that

127

the themes emerged during the coding process referred to the topics or focus of reflection that the PSTs included in their DJs and the GRs. Based on those themes, I was able to analyse the manner in which the PSTs reflected upon them.

To sum up, my analysis followed a data-driven inductive approach, with a combination of thematic analysis and data analysis processes, such as pre-coding and second-level coding (Dörnyei 2011); initial coding and categorisation (Richards 2003); and open, axial, and selective coding (Strauss and Corbin 1990).