CHAPTER 4: METHODOLOGY
4.7 Data Interpretation
This section describes the interpretive approach which this research adopts and discusses the steps which were followed in each stage of the interpretation process. This interpretive approach is based on Dey’s (1993) ‘omelette’ approach. This approach consists of three stages which are shown in Figure 4.1. The three stages are description, classification, and connection.
Figure 4.1: Description, classification and connection (Adapted from Dey 1993: 31)
The description stage of data analysis is central to any study and can be seen as the initial portrayal of the data in a form that can be easily read and interpreted (Crang 2005; Kitchin and Tate 2000; Dey 1993). The description stage consists of two steps; transcription and annotation. In the transcription step the interviews are transcribed into text, while in the annotation step notes or comments are added to the transcripts by the researcher through a process of reflection on the possible meaning of the initial text. In this research, the tape recordings collected from the interviews were transcribed after each interview. This was done so that the transcripts would be as accurate as possible and that any notes or annotations which were made during the interview or further reflection could also be added (Valentine 2005; Kitchin and Tate 2000). As the methodological approach of this research focuses on language, the transcriptions did not need to include nuances or changes in speech patterns. This is the reason that the transcriptions can be seen as verbatim copies of what ‘was’ said and not ‘how’ it was said.
In the classification stage, the interpretation moved beyond data description to the first stage of interpretation where an attempt was made to make sense of the data (Kitchin and Tate 2000; Dey 1993). In order to interpret the data, the data had to be broken up into its constituent parts and these
Description of Data
• Transcription
• Annotation
Classification of Data
• Categorising
• Splitting & splicing
Connection of Data
• Linking & connecting
• Corroborating evidence
parts placed into similar categories or classes. Using Hajer’s (2006; 2005a) analytical model as a framework, each of the transcripts was interpreted in order to identify the discursive and performative concepts of the policy documents and interviews.
The first phase of classification focuses on the discursive concepts in the SEA policy processes. A deductive approach was adopted in order to develop four broad themes. The themes were based on the overall aim and objectives of this research. The four themes represent the master categories which were identified. The master categories or themes were ‘what is an SEA?’, the ‘need for a SEA’, the
‘need for consultants’, and the ‘role of actors in a SEA’. Within each of these master categories, statements from the transcripts were cut and pasted into the appropriate category in a new document.
The statements were then grouped based on similarity into several sub-categories within the master categories. This process is referred to as splitting or ‘axial coding’ (Crang 2005; Kitchin and Tate 2000; Dey 1993). For example, the master category of ‘what is an SEA?’ included the sub-categories the ‘SEA as a decision making tool’, the ‘carrying capacity of the environment for development’, the
‘need to balance brown and green issues’, and the ‘need to be strategic’. These sub-categories were identified as the story lines which the respondents employed. It is important to note that during the classification stage all the narratives were coded so as not to lose or sever the link with its original source. The purpose of coding the data is not to develop an explanatory framework, but to organise the data in such a way that the relationships between the data can be examined at a later stage.
The second step of this phase is to identify the policy vocabularies which policy makers conspicuously use in the KwaDukuza and Rustenburg SEAs. Within each SEA, particular concepts that featured prominently, such as ‘sustainable’, ‘ecological’ or ‘development’ were identified and highlighted. The number of times that they were used within the SEAs was also recorded. All the derivatives of the concept were also included into this count. For example, the count of the concept of
‘development’ would include all its derivatives of the term development, such as the terms; ‘develop’,
‘developer’ or ‘developed’. The context in which the concepts were used was also highlighted in the transcripts so that the interpretation could identify the most common ways in which that particular concept had been used. For example, the most common way in which the ‘development’ concept was used is in the terms ‘development plan’, ‘development framework’, and ‘sustainable development’.
The concepts were then grouped into master categories based on their similarity. The master categories were identified as the policy vocabularies which the public officials, environmental consultants, and key stakeholders used in the KwaDukuza and Rustenburg SEAs. Three policy
vocabularies were identified. These are the ‘developmental’ policy vocabulary, ‘management’ policy vocabulary, and ‘environmental’ policy vocabulary.
Based on the story lines which were identified in the oral data collected from the interviews and policy vocabularies embedded in the three SEAs, a single discourse was identified. The ‘ecological modernisation’ discourse was found to frame both the KwaDukuza and Rustenburg SEAs.
The second phase of classification focuses on the performative dimension of the SEA policy process.
The performative dimension consists of four concepts; scripting, staging, setting, and performance.
These concepts were used to form the master categories. Within the interview transcripts, statements which relate to the master categories were highlighted, cut, and pasted into the appropriate category in a new document. The statements within the master categories were then grouped based on similarity to form several sub-categories. For example, the scripting master category was split into two sub categories; ‘scripts’, and ‘actors’. All the statements were coded so that the source and context was not lost.
‘Connection’ is the final stage of the Dey (1993) qualitative approach and focuses on identifying and understanding the relationships and associations between the different statements which were identified and classified in the ‘description’ and ‘classification’ stages (Kitchin and Tate 2000; Dey 1993). Unlike the classification stage, the focus here is not on similarities and differences between statements, but the interactions between the various concepts. The stage focuses on the way in which the concepts within and between the discursive and performative dimensions of policy making interact with one another.
The connections between the discourses, story lines, and policy vocabularies of the discursive dimension of policy making are complex. Discourse represents the mix of ideas or concepts which actors use to shape and interpret the environmental problems which they are faced with. In the context of this research, story lines represent a condensed set of statements which actors use to frame a discussion within a specific discourse. Policy vocabularies can be seen as a group of concepts which actors conspicuously use in policy to frame the policy within a specific discourse. However, these discursive concepts of policy making do not take place within a vacuum, but within a particular set of practices (Hajer 2006; 2005a; 1995). There are several connections which were identified and interpreted between the discursive concepts of the SEA policy processes and the performances through which these processes take place. This is because the four concepts of the performative
dimension of policy making, namely scripting, staging, setting, and performance, affect the way in which the discourses of the SEA policy processes are produced, reproduced, and transformed.
Due to the qualitative nature of this research several challenges were encountered in collecting and interpreting the data. Although some of these challenges have already been discussed in terms of the methodological approaches, the following section outlines the major limitations that were encountered.