Capítulo 2: El Efecto de la Identidad Social en la Acción Colectiva de los Pequeños
2.5. Estudio de caso
2.5.3. Elementos identitarios y su incidencia en la acción colectiva
2.5.3.1. Proceso de reivindicación de los pueblos indígenas en los Altos de Chiapas
5.5.0. Data Analysis
Data analysis has been an ongoing process that started when I began fieldwork and carried on throughout the writing of this thesis. After every interview session I listened and re-listened to the tape-recordings connecting each respondent’s comments while reflecting on the theory of the thesis. Some modifications took place at this stage, with new questions added, reframed, or removed if deemed unnecessary. To a considerable extent, the fieldwork proceeded without a precise beginning or end, but this stage of the research required a dialectic bridging of analytical research themes. Dey (1993) argues that dialectic informs data analysis from the outset through all phases of the research, rendering arguments that generate and transform raw data into new knowledge.
There are several analytic strategies in qualitative research, although most data analysis is based on analytic deduction or induction. Analytic deductive reasoning, rather than inductive reasoning, was adopted for this research because it allows the testing of data, confirming or negating the study hypothesis, with the goal of most accurately representing the situation. I decided not to use inductive reasoning because this study was not intended to generate data for formulating a tentative hypothesis or in any way end up with the development of theory.
The analytical method used in this thesis, as mentioned before, is the Proactive Decision-Making model (PDM) analytic tool developed from the theoretical propositions (such as mnemonic PROACTIVE). PDM facilities a deeper understanding of decisions and non decisions by analysing relationships between various variables such as institutions, actors, power, influence, force, authority, socio-economics, politics, evidence and norms. By studying the mnemonic PROACTIVE and applying it to the 2009 pandemic decisions enabled a four type process of analysis, identifies a problem and evaluates alternatives (a set of possible actions), assumptions (simplifications of reality), assessment (spectrum of possibilities), and performance (beyond the obvious and routine). PDM, as applied to thematic analysis, seeks to deal with data that takes into account the production and utilization of “codes”. In the present research, analysis of interview transcripts, field notes, histories and policy documents use thematic analysis. The “data” being analyzed involves how policymakers formulate or reach policy decisions and how they translate policies into implementation programmes. A thematic approach to data analysis allows cross-site comparisons, using the accounts of participants to create a multi-dimensional picture. This approach is highly connected to the development of theory and readily informed by existing theories. Thematic analysis is similar to grounded theory since its structure is equally designed to support the process for carrying out this type of code-related analysis. Ryan and Bernard (2000) consider thematic coding a process performed in grounded theory, rather than a specific approach in its own right. Braun and Clarke (2006) argue that thematic analysis is an analytical method in its own right, essentially independent of theory and epistemology and able to be applied across a range of theoretical and epistemological approaches.
Grounded theory, originally developed by Glaser and Strauss in the 1960s, is essentially more concerned with developing theory about phenomena of interest using collected data (Glaser et al., 1967). Since the purpose of this study was not to develop theory, grounded theory was not used, although part of its theory on the process of coding was used to complement thematic analysis. Before demonstrating how I carried out thematic analysis, it is important to distinguish it from framework analysis since both employ a simple ordering of data. The latter approach is suited to applied policy research and better adapted to research that has specific questions, a limited time frame, a pre-designed sample (e.g. professional participants) and a priori issues (e.g. organizational and integration issues) that need to be dealt with in a particular setting (Srivastava and Thomson, 2009). The former, as indicated earlier, involves searching through data to identify recurrent patterns. At first the framework
approach seemed an appropriate analytical tool; it was, however, rejected as it requires a short timescale due to the nature of applied research, and there is often a need to link the analysis with quantitative findings (Pope et al., 2000). In addition, data under framework analysis is more structured, and analysis is explicitly and more strongly informed by a priori reasoning.
There are set rules when conducting thematic analysis but the most important are pin-pointed by Dingwall et al. (1998) as being trained to acquire skills and experience in order to be able to conduct good qualitative analysis. It was from this perspective that prior to the transcription of interview data I assumed the task of researcher to familiarize myself with the data, which is key to thematic analysis. Researchers should be very much involved in good thematic analysis, collecting data and transcribing it themselves. Prior to transcribing the interview data myself, I analyzed the audio recording of each interview session each day in order to create superficial themes. This involved jotting down interesting notes into a “data analysis logbook” clearly marked for each interview. This process was repeated after full transcription, during the coding of the data, and again at the time of writing.
To code and analyse the data, Computer Assisted Qualitative Data Analysis Software (CAQDAS) was utilized. NVivo was the preferred package used to code chunks of data and reduce it to themes and categories, partly because I had easy access to the software (Student license with the University of Nottingham). CAQDAS Textbase Manager or programme requires the data to be fully transcribed to rich text to facilitate the thematic coding and development of categories; the user also needs to be proficient in the programme. Some Textbase Data Managers have complex functions and methods of analysis, thereby raising debate over whether word-processing packages are better. Even among CAQDAS itself, there are numerous pros and cons as to what type of programme is best to use. It is argued, however, that word-processing packages do a better job, as opposed to many tasks undertaken by CAQDAS that initially rely on a lot of the researchers’ reading and familiarization with both the data and software. CAQDAS only worked after recorded audio tape interviews were transcribed and field notes analyzed ready for further analysis. The way CAQDAS handles analysis is not different from word processor use, although I was using this package for the first time. My background is more embedded in quantitative methods, where statistical packages, such as STATA and Epi-info, are used to analyze data. This did
not mean I abandoned analysis using CAQDAS, but I took on board comments by Dingwall et al. (1998), who argue for researchers’ experience in data analysis.
Without much experience with CAQDAS, and considering my concern that it would complicate the process of analysis rather than facilitate it (potentially limit the reflexivity and in-depth analysis), in addition, I used a manual approach of data analysis while iteratively working with CAQDAS. This involved re-reading the transcribed data, annotating thoughts in the margin of the script and, where possible, pasting the line references in CAQDAS. Data analysis was kept as simple as possible using word-processing and index text while familiarizing myself with CAQDAS. Applying CAQDAS and manual data analysis was important for rigour and depth. While this promoted a close relationship with the data in terms of meaning and context, I did not distance myself from its advantages, such as coding and retrieval, thereby allowing for comparisons to be made within a vast qualitative dataset.