2.3 | Analyzing the data
After collecting the data, our next step was to categorize it according to format (video, photo, audio, text); source (primary, secondary) and type of case (assentamentos, ecovillages and bureaucrats). For the current analysis, we knew it would not be possible to closely analyze the whole data set due to the time and format constraints of the Hertie School of Governance MPP program. Therefore, we prioritized in-depth interviews in video (where most of the content lies) as our main source of information.
Table 2. Overview of data collected (by community and method) (2014)
After transcribing all the relevant interviews, we used QDA (Qualitative Data Analysis) software (Atlas.ti) to start analyzing them. Inspired by Corbin and Strauss (1990) and Mair, Marti, & Ventresca (2012), we pursued an analytical process that consisted in successively “coding” our data in order to make sense of patterns and identify salient topics and questions.
First, we created in-vivo/open codes while reading interview transcripts, selecting quotes (data extracts) and assigning keywords either spoken by interviewees or created by the research team. [1] In this first step, the main goal was to associate “tags” to different quotes in order to facilitate our search and to start to understand what our data was about, while identifying topics about which we could write and areas we could use to sort and synthesize our material. The main goal was to identify pre-existing structures, dynamics, challenges and insights about sustainability in rural communities that could subsequently be placed in broader categories. This task was first developed by individual team members, and later on together.
During this initial coding stage, we elaborated memos and notes that already hinted to insights, ideas and analytical categories that could be used in our argument. Those memos were “partial, tentative and exploratory” and were used to “capture ideas in process and progress”, and as an initial opportunity to “learn about the data rather than just summarizing material”, creating ideas on how to proceed (Charmaz, 2008, p. 166).
After we finished the first round of open coding, we assessed which codes appeared more frequently and were emerging as significant to answer our evolving research question. We then moved to a more focused or selective coding, aiming to sort and synthesize the data and expedite our work by reducing the complexity and number of categories we were working with. This second coding stage involved the creation of second-order codes (or “families” in Atlas.ti) that were used to aggregate different first-order codes, and point out larger topics and patterns emerging from our data.
In a process of constant interaction with our data, we selected the codes and families that could help us provide a description of those clusters and dimensions, identifying negative and positive examples on each one of them and writing down possible insights on relevant common elements. In this process, we started to identify patterns and commonalities on salient challenges for sustainability in assentamentos and possible actions that could be taken to tackle them. In other words, we started to obtain answers to the research question.
Our analytical procedure, inspired by Glaser (1978, p. 57) prompted us to have two main questions in mind when looking at the data: “What is happening here?” and “What theoretical category is this data a study of?” Following Charmaz (2008. p. 161):
As a final result of our analysis, we identified categories of challenges and tools and potential insights that could be used to reflect upon assentamentos under the lens of the broad sustainability concept, as outlined above. Our contribution lies in highlighting the dimensions that should be taken into consideration when designing policies to foster sustainability in Brazilian agrarian reform, and hinting at potential approaches that could serve in working towards this aim.“[...] interrogating their data repeatedly with these two questions, grounded theorists explicate, expedite, and enhance intuitive strategies that other qualitative researchers often invoke on a descriptive level. These strategies include probing beneath the surface: comparing data, checking hunches, refining emerging ideas, and constructing abstract categories from data analysis.”
We acknowledge that our research has limitations due to its grounded-theory-inspired approach, as the coding and categorizing process may generate a certain decontextualization and the chaining and interplay of particular events may sometimes become lost in this process (Langley & Abdallah, 2011). However, we tried to overcome those limitations, as much as we could, by permanently recalling the conditions of visited communities (aided by audiovisual materials) and incorporating them in our analysis.
>>> 2.4 | A Visual Story of Our Journey
[1] According to Charmaz (2008, p. 163), this first step requires a “close reading and interrogation of the data” coupled with “research field analysis of the data”, but we could not properly carry this out in practice, as most of the cases analyzed were located in Brazil and our team was based in Berlin.