Chapter 3: Methodology
3.3. Organising data and analysis
78
79 between data and to compare these with emerging themes in order to determine whether the data confirmed or refuted these themes. Comparisons are important as it allows constant identification of similarities and differences between emerging themes and the identification of sub-themes (Dey, 1999). This, in turn, ensures recognition and capture of complexity and diversity in data. Further, negative cases22 were identified and reflected upon. This process of negative case analysis is intended to facilitate qualification and elaboration of the emerging themes.
Throughout these processes23 I reflected on the emerging themes in a research journal. This journal was a written record of reflections, analysis and interactions with theory and was intended to reflect on both the research process and substantive findings. The journal traced emerging themes, justifications of descriptive labels, relationships and linkages between categories, integration of higher and lower-level categories and reflections on the research questions and research direction.
This process of immersion and interacting with the data provided me with an opportunity to reflect on and prioritise further data collection. This process is referred to as theoretical sampling. In brief, it involved collecting further data on the basis of themes that emerged during earlier stages of data analysis. It further enables one to reality-check emerging themes against new data collected in the field. Data collection decisions were thus refined according to initial descriptive themes that emerged. This process enabled a movement from a descriptive level to an analytic level, referred to as theoretical sensitivity.
Following subsequent data collection, selective coding was undertaken. This stage allowed me to filter and code data deemed more relevant to the emerging themes. The key messages or themes of data from interviews, observations and documents were thus selected and coded. This process facilitated the identification and integration of lower-level (descriptive) themes into higher level (analytic units) themes. It involved interpreting rather than simply describing instances of a phenomenon. The process of interpreting themes is important, as it allowed me to become aware of the links and relationships between and amongst themes and helped to arrange these themes in a meaningful way. Comparisons and reflections in my journal were also done in this process. In brief, selective coding facilitated the discovery and identification of analytic themes. Thus, several analytic themes were identified within the four broad themes of this study. These are presented in the Table 3.1. These broadly correlate with the manner in which the substantive chapters (Chapter 5 - 8) are organised.
22 i.e. data that did not fit within emerging categories
23 Open-coding, comparative analysis and identification of negative analysis
80 Table 3.1: Themes and categories emerging from coding
Main themes
Context (Global South)
Scale (urban) Regulation Organisation
Analytic Sub-themes
Poverty and access Constraints to urban transitions
Barriers and constraints
Organisational functions Inequality and
differentiation in relation to infrastructure
Securitisation, marginalisation and bounding
Tensions between national and local government
Organisational structure and cultures
Cross-subsidy systems
National cross- subsidy systems
Impacts of top- down regulation on sub-systems
Risk, compliance and innovation
Resource constraints and competition
Implications of urban autonomy
Divergent interpretations
Decision-making, policy and politics
Core theme Contested values (environment &
development)
Contested values (urban & national securities)
Contested values in regulation
Contested values in organisations
Through analysis of documents, interviews and observations it became evident that most participants in the study expressed some notion or concern related to values and contestation underpinning the reproduction and reconfiguration of the City’s electricity regime. Notably, participants placed importance on a wide range of contested values related to, inter alia, equity, environmental sustainability, climate change, poverty alleviation and redistribution. This cut across most of the emerging sub-themes. This to some extent was regarded as a core theme.
Once this core theme began to emerge, data collection and analysis became more focused on this core theme (regime values). Interviews and observations were thus focused on further exploration of this theme. This was a shorter period of data collection as it appeared that by and large no new sub-themes could be identified from additional data. This was an indication that theoretical saturation had been reached, where dense themes had emerged. Although it is always possible to refine and add to these themes through new data, theoretical saturation is a practical signal to end field work.
Finally, I undertook the last stage of coding referred to as theoretical coding in order to develop the conceptual contributions of this study. In this process, the core theme and sub-themes of this study were sorted, reflected upon and cross referenced with literature. Through theoretical coding, I reflected upon the emergent themes and sought to develop the conceptual relationships between
81 themes and their relevance (Melia, 1996) to transitions in general and the MLP in particular.
Theoretical codes assisted in recognising patterns and linkages and supported the process of theorising on the ways in which sustainable transitions are constrained or supported by values, scale, regulatory systems, organisational conditions and policy contexts. In this way, theoretical coding was a process of knitting fractured themes together in order to provide a broader picture.
This process of theoretical coding further ensured consistency and objectivity in my analysis (Strauss and Corbin, 1998).
This three-stage process of open-ended coding, selective coding and theoretical coding more or less coincided with the three-year period of embeddedness. In the first year of embeddedness I cast a wide net on the data collected and coded according to descriptive themes. In the second year I was able to be more selective in data collection and coding was undertaken with a focus more on higher-level analysis. In the final year coding was far more focused on the core theme that cut across data collection and the implications for the MLP and sustainable transition theory.