3.3 Research design and methods
3.3.2 Semi-structured interviews
The semi-structured interviews were the most relevant methods of data collection to approach the information, representations and experience of those who plan, develop, inhabit and contest the garden enclaves. The interviews aimed to capture the agents’ narratives about the enclave development and the actors involved; to get an insight in their subjectivities and perception of everyday life experience; to obtain the agents’
environmental discourses about the enclave ecology and metabolism; and to gather insight on the agents’ perceptions of being political subjects.
The interviews were directed to selected key informants, defined as those who have a specific type of knowledge about the phenomena being researched, through their role in the production, reproduction or contestation of the enclaves.
Study population and sampling
I defined the study population as those active actors of the enclave production and development from the state, the private sector, and the civil society. The selection of specific groups from these three social categories considered those who conceived and planned, approved, designed, built, studied, lived and contested the eco-enclaves. There are of course many other groups that arguably take part in the making, maintaining and living the enclaves. Beyond obvious limitations of time and resources, the justification for prioritising on selecting interviewees among these groups of agents is that apart from their own role in the enclave production, they were also consistently present in my literature review and public debate (as I found in my preliminary review of the press).
Although I considered many subcategories of actors I finally grouped them to produce a sampling framework of seven broad categories described below (Table 3.3): private developers; consultants; state officials; scholars;
activists, residents; and workers. The list of subcategories, institutional
affiliation (or membership) of the interviewees, gender and sampling size can be found in Appendix 4.
Table 3.3 Interviewee actors’ per category. Source: author’s own.
Actor Definition
Developers Owners, managers and sales agents of property development companies.
Representatives of developers association.
Consultants
Professionals and practitioners as architects, landscape architects, urbanists, lawyers, and any other profession involved in Santiago’s enclave projects planning, design, and development. Staff belonging to think tanks of the construction industry in Santiago.
National/regional/local government
Professional or directive staff from ministerial offices, regional government and municipal level departments involved in the Chicureo enclave planning, approval, governance or management.
Scholars University based academics involved in the enclave research and debate.
Activists
Members of ecologist and urban social justice organizations involved in conflicts with Santiago’s urban enclave development. Residents and workers of the enclaves’ hinterland participating in community organizations and associations.
Residents People currently inhabiting a house in a Chicureo enclave, without distinction between owners or tenants.
Workers Manual workers. Employees or contractors of firms operating in the enclaves or enclave residents
To overcome the high heterogeneity of the groups and possible restrictions I developed a purposive or non-probabilistic sampling (Bryman, 2008) directed to represent as many of the case’s facets as possible (Merkens, 2004). Following these principles I set four criteria for the sampling:
i) Represent all the defined groups of enclave agents;
ii) Maximise the variety of the sample members, getting interviewees from different categories, roles, and gender (i.e. interviewees from developers from different companies; as well as interviewees with different social roles and positions, as directives or employees, workers, students or housewives, female and male).
iii) Balance a sample number large enough to obtain enough data density and diversity for the analysis and at the same time feasible and manageable.
iv) Flexibility to select new interviewees during the fieldwork.
Within this general purposive method, I developed different sampling strategies in accordance to each group characteristic and availability. These sampling strategies can be described as personal (when a specific individual is targeted by his or her relevance); institutional (any individual playing a specific role in an institution or organization can be sampled); opportunistic snowballing (names of potential interviewees within an institution or enclave are gathered from other contacts or interviewees) and random (any person playing a specific role in a selected institution or enclave can be selected).
Throughout the development of the interviews I noticed that some of the interviewees could be classified, for the purposes of my research, in more than one social identity, as in the case of consultants that were also scholars, or scholars that had been state officials. In those cases I classified them in my data base by the first identity, that is the one for which they were firstly selected, registering their second one in another field, a knowledge that became useful to understand the links between the actors of the enclaves, as I develop in chapter 7.
Interview contents
About the questions and contents of the interviews, no questionnaire was made but a guide for the interviews was tailored based on the research questions and the preliminary revision of secondary information. As the interviews and my knowledge of the field progressed, I sought particular emphasis in each group and type of informants. Interviews had greater variation than I expected, in length, depth, and specific subjects that were dealt with. Respecting the interviewees` emphasis, an effort was made to consider a common core of themes for all group members in the above listed themes. The description of the themes is displayed below (Table 3.4), and the questions that guided the interview of each category can be found in Appendix 14.
Table 3.4 Thematic guide for interviews and analysis. Source: author’s
- Land, planning and housing policies in Chicureo
- Santiago’s urban sustainability policies and situation
- Santiago growth, splintering and sprawl
- Enclosure and commodification
- Public and private convergence and divergence
- Perception of the enclaves’ impacts
- Perception of the enclaves’ greenness
- Perception of own consciousness
- Neoliberal urban reforms
- Role of market, state and civil society in the greening of the enclaves
- Life experience in the enclaves
- Environmental management of the enclaves
- Environmental discourses
- Enclaves’ planning, design, urbanism, architecture, landscaping
- Urbanisation of Chicureo and enclave development
- Role of actors and conflicts
- Recent or current conflicts, coalitions and strategies
- Governance of Chicureo enclaves and the Colina borough
- Ecology of Chicureo and hinterland
- Urban metabolism and flows in Chicureo
Although most of the questions I asked were contained in these themes, new themes were also raised by the interviewees during the interviews, and once in the analysis phase, the number and organization of the themes did not remain the same.
Data collection
The interviews were conducted in two periods of fieldwork, developed in Santiago de Chile, the first from November 2012 to February 2013, the second from March to May 2013. In total, I conducted 57 interviews, except
4 from a specific condominium3. Also in four cases there was more than one interviewee actively participating in the session, and they are counted as interviewed. In total, I had 62 individuals interviewed in 57 interviews.
Table 3.5 Interviews facts. Source: author’s own.
Contacted Interviewed Interviews Place of interview Site of interview Chicureo Colina town Santiago Workplace Home
130 62 57 19 3 35 37 20
In accordance with each sampling strategy, an email, webpage, telephone or face-to-face contact was established with the selected potential interviewees.
It must be stated that two groups were particularly hard to contact, developers and workers. Many developers simply did not answer my e-mails, telephone calls and messages, or just refused to be interviewed. In the case of workers, it was even harder; as I had to explore several strategies to reach and contact potential interviewees, from leaving leaflets at the bus stops, search for recommendations, to direct approach. It was extremely difficult to contact workers, and most of those contacted straightforward refused any possibility of being interviewed, which limited my final sample of this group. Once I was processing the data and started the analysis some categories revealed problematic, as I found that some interviewed actors belonged to more than one category. For reasons both of data organization and the sake of the analysis I finally merged local, regional and national state officials in one category, State officials; by their role in contesting the enclaves I included Community and organization members in the category of Activists; and by their role as providers of services I included interviewees originally grouped as Think tanks members as part of Consultants (Table 3.6).
3 This was a strategy to interview residents of a very small parcela enclave condominium. In the course of an interview, a resident from this parcela manifested her willingness to contribute with the research, and so I trained her to conduct some interviews to people of her condominium. I agreed to pay for her work as research assistant, also as a way to exert more control over the interview process. After each interview I demanded a report and her preparations for the next interview.
Table 3.6 Population study sample final categories. Source: author’s own.
Category Sample
Developers 5
Consultants 12
State officials 11
Scholars 6
Activists 7
Residents 19
Workers 2
Data processing and analysis
The transcription of the interviews was time and resource consuming, even if for about half of them I relied on a paid transcription service, which did not exempt me from doing a final edition of all the interview transcriptions.
Once transcribed, the text of the interviews4 was coded through the use of the NVivo software for qualitative data analysis (versions 7 and 10). My aim was to develop a discourse analysis through NVivo, but I ended by developing an analysis with the help of NVivo as an organiser and searcher.
Although I did run analysis of the categories emerging from the coding of the text, and used them to develop basic forms of content analysis, for the discourse analyses I relied on my own interpretation of linkages among what I selected as significant chunks of text, the broader narrative of the interviewee, and the wider context about the interviewed agents on the interviewee.
Regarding the discourse analysis, along with looking for more Foucauldian representations of power, desire, fear, ideal models and control mechanisms, I tried to pay attention to the discourse audience; to the categories of context, practices, attitudes and experiences; to identify the persuasive discursive structures that unify and signify the discourses; and to take note of inconsistencies and silences (Waitt, 2010).
4 I left the text body in Spanish and only translated those text extracts to be quoted in the dissertation. All the codes were named in English. All the quotes from the interviews in this dissertation are my translation.
With respect to the use of NVivo, I took coding as the process that allows finding both the manifest and latent meanings over a text (Dunn, 2010), which is segmented into “categorizing and qualifying data” for its interpretation (Cope 2010 p.284) and involves two main levels of codes: the descriptive and the analytical levels (Waitt, 2010). Among the process of coding I organised the codes into clusters by characteristics and relational variables, to facilitate the capture of the connections between themes, in a reflexive process of iterative reviewing of data and re-coding (Cope, 2010).
My strategy was to use a sequence of coding procedures as defined by content analysis and discourse analysis. I first used an open coding to transform the data into concepts, but the results weren’t satisfactory and I ended with a too fragmented and unrelated bulk of 3,737 codes, a mistake in the selection of the coding procedure, possibly more suitable for grounded theory as I realised too late. After struggling with methodologies that in my perspective focused on a too narrow, formalist, microanalysis, I shifted to a constructionist discourse thematic analysis, a more flexible and accessible method (Braun & Clarke, 2006).
Finding regularities and patterns of association between the codes allowed a much clearer and more meaningful categorisation of the interviews (Bazeley, 2007). Within this analysis I first developed 290 basic nodes organised by themes (Appendix 5), then grouping codes in trees and sets and conducting queries and relationship maps to analyse interviews more specifically, as I developed in the following chapters. As previously stated, I avoided to develop grounded theory from interpreting the coded patterns and rather tried -with different degrees of success- to use the text to develop a narrative of the process of urbanisation, the social position of the actors and their actions, and their representations and subjectivities about their environment and their own role in it.
I additionally performed some quantitative content analysis, understood as a research technique for the “quantitative description of the manifest content of communication” (Berelson 1995 cited in Bryman, 2008 p.274), to inform the frequency and distribution of some themes, concepts or keywords,
among the different categories of interviewees, as in the case of those concepts related to greenness as developed in Chapter 7. In those cases I made an effort to contrast the results with a qualitative analysis of the content context to enrich the analysis.