3. Research design
3.3. Data management and analysis
The collected data constitutes documents, photos, recordings, personal descriptive and reflective notes, and transcripts. I have used the software Nvivo to manage them. In my data analysis, I attempted to concentrate on the narratives of my interviewees: what they say, how, when and in which contexts.
3.3.1. Data management
During the data collection, I took personal notes of my observations and interviews.
These have been typed up or photographed. The focus group discussion and interviews were recorded when participants consented. I did not take written notes during the interviews but noted salient points and key impressions after each interview and discussion with Hoby. I decided to transcribe all the interview recordings to have a comprehensive database. This was done in the three months after the data collection.
77
Hoby transcribed and translated local interviews in French. I transcribed the national- and global-level interviews in the spoken language (French or English). I then used Nvivo to organise and cluster the data under various themes. All data were saved on my computer, external hard drive and in a password-protected cloud service. I have also used a coding system to ensure the anonymity of the interviewees. Daily back-ups have been made of the different chapters during the writing-up stage.
3.3.2. Data analysis
The research questions (structured around conceptions on tenure security, practices of securing tenure and interactions between actors) and the conceptual framings (organised around policy narratives, assemblages and power dynamisms) have informed data analysis (see Table 3.5. below). This analysis has been gradual. I have reflected on and analysed data as I collected it.
I started by building general knowledge of Malagasy land policy process through event-ethnography, document analysis and exploratory interviews. First, this entailed mapping national and international actors involved in the policy process. Second, this meant building a general idea of the policy framings, ideas and practices (Keeley and Scoones 2003). These steps indicated how tenure security is officially conceived and practiced, shed light on how actors refer to global instruments or local experiences, and revealed the nature of links actors have with the policy. The elements emerging from this ground analysis also guided the interviews.
I continued by recognising the way in which individual actors and institutions conceive tenure security and the practices they undertake to secure tenure. I also identified the concrete roles actors play in the policy process, the attitudes and positions they hold towards the policy, and the connections they establish between themselves. I first interviewed regional and local administrators, intermediaries, chiefs, elders and farmers.
By the time of the national- and global-level interviews, I was aware of the key issues on which I invited the interviewees to elaborate.
With the interviews, I tried to allow my participants to tell the story of the policy process from their perspective. I wanted to ground the analysis in the ‘everyday life experiences
78
and understandings’ of actors (Long 1992a, p.5). I have aimed to recognise the multiple realities of actors, be aware of on-going social processes and grasp the real dynamisms at play (Long 1992a; Olivier de Sardan 1995). This has been possible by keeping the interviews as open as possible and by attending to the narratives of interviewees in the data analysis. I consider that through stories interviewees communicate messages and make sense of events and actions (Riessman 1993; Adger et al. 2001). Therefore, aside from the regular discussions, I have taken note of what they said and how, the history of their accounts, the social context in which their accounts were constructed, the way they persuaded the listener of the validity of their stories, and the way they gave voice to actors not commonly represented (Riessman 1993; Gibbs 2007; Gubrium and Holstein 2009). These elements have enabled me to highlight the lived experiences of actors and the dynamics of the policy process.
I then listened to and transcribed the global- and national-level interviews, and read the translated local interviews. This enabled me to refamiliarize myself with their content, who said what and where, and continue writing reflective notes. The organisation of the data in Nvivo was also helpful for analysing key content-related issues. I clustered some data under themes that gradually emerged in this process. These clusters were in turn organised under wider groupings: i) overall meanings of land and views on the land policy process; ii) concepts of tenure security; iii) practices of securing tenure; iv) relations and power plays; and v) new sources of tenure insecurity. I did the clustering and grouping separately for local, national and global data. Furthermore, the clusters informed the organisation of the empirical chapters and enabled me to go back and forth with the thematic content during the write-up.
Finally, drafting the thesis has enabled me to connect content with the conceptual framework and with personal reflections. I have recognised the diverse concepts of tenure security, captured the variety of practices undertaken to secure tenure, examined the translations of the policy on the ground and attended to the interaction between actors. By linking official sources with the accounts of actors and my own observations, I have analysed how actors relate to the different concepts, practices and other actors. I have also been able to identify (dis)connections between levels. This analysis has informed the identification of policy narratives, assemblages of actors and existing power dynamics, wherein some actors are included in, and others excluded from, the
79
policy process. In particular, the stories of the interviewees enabled me to go beyond the factual elements of the policy process and grasp how policy dynamisms manifest on the ground.
In this final process, I have also been able to structure ideas, elaborate key messages and form a coherent narrative. I have described a dominant policy narrative, but also explored counter- and competing stories that alter and contest the dominant narrative (Roe 1989, 1991 and 1994). I have recognised that these narratives belong to ‘complex, multi-participant and multigroup’ assemblages of actors and the different weight given to them depends, for instance, on their access to economic and political power (Roe 1989 and 1994). I have paid attention to the key actors producing, transforming and maintaining these narratives and assemblages as well as how they persuade others of the validity of their ideas (see Li 2007).
Table 3.5. Analysis of key variables
Variable Definition Elements to be considered (examples) Actors People and
approaches; values given to land; references to sources of and conditions for tenure security (e.g. political, social, cultural, legal,
administrative); references to conflicts, tensions, threats; references to practices of securing tenure; new sources of tenure insecurity; etc.
Practices Concrete actions,
vision on policy process; new sources of tenure insecurity; etc.
80 Mechanisms Processes shaping the
consideration of tenure security in the policy process
Policy narratives; assemblages of actors around certain ways of seeing and doing; power in policy and decision-making processes