CHAPTER 3 METHODOLOGY
3.5 Strategy for Processing of Empirical Data
In this section the processing of the interviews will be described. First the methodological implications in the use of transcription will be explained. Then there will be a description of the strategy for the coding of the interviews, followed by a discussion of the consequences of such an analysis of the empirical material.
Given the amount of empirical material and the normative approach of this study it is essential to secure the complexity in the material and that the study is true to what the
interviewees express. This has given rise to the use of a thematic analysis of the data, yet this approach to qualitative data analysis can be conducted in many different ways (Bryman 2012:578). However, the method used in this study draws on the approach described by Kristensen (2010), which is elaborated in the following sections.
3.5.1 Strategy of Transcription
We have transcribed the interviews on the basis of a wish to maintain the complexity of the empirical data (Bryman 2012:482). They have been transcribed in full since we want to minimize the risk of immediate analysis of the data in the selection process of a partial transcript (Kristensen 2010:289). The selection process always constitutes the danger of leaving out important details that is not necessarily included in the themes on which the partial transcription is based (Kristensen 2010:289). Emotions and mumbling has been left out of the transcript since the aim of this project is not to make a discourse analysis but to investigate the structures. The interviews in Danish have been translated to English when quoted in this study. The concrete method applied for transcribing is attached as Appendix 7.
3.5.2 Thematic Analysis
This study’s approach to analysis of the data is build upon an idea that there are themes that cut across the interviews (Kristensen 2010:289). The analysis is focused on selected themes, which are compared across the transcribed interviews. It is based upon an idea that it is necessary to understand the part in relation to the unity (Kristensen 2010:289).
The part, being each quote and extraction, will be analyzed in connection to the interview as a whole, but also in connection to the other quotes and extractions on the same theme from the other interviews. The importance of the dialectic relationship is seen in the focus of the cross readings of the data, and the goal with this analysis is to create an understanding of the entity of every theme.
This kind of thematic analyzes consists of four distinct yet overlapping phases: 1) coding, 2) meaning condensations of each theme in every interview, 3) comparisons and summaries a cross the interviews and 4) the final analysis and answer to the research question (Kristensen 2010:290). According to Kristensen the first phases in the analysis are based on the interviewees’ point of view (Kristensen 2010:290). This implicates a need to keep the interviewees’ perspectives and meanings about the content. The complexity of the empirical data is thus secured, which makes the critical investigation
of the complex and hidden structures in the social reality possible without simplifying the data to fit the theoretical point of view. The final analysis starts when the data has been thoroughly processed (Kristensen 2010:290). This fits the critical theoretical position of this study where the social world needs to be illuminated in order to see where and how it differs from the normative theoretical ideal of democracy.
3.5.2.1 Strategy for Coding
The first step in the thematic analysis is coding the interviews. The codes used in this study were developed in respect to an abductive reasoning (Bryman 2012:401). They are founded in three parts: in the general understanding of this study’s field as it is presented in the study’s first two chapters, the topics of the interview guides, and in the primary knowledge gained in the interview situations. From this we reached the following codes:
• Stakeholders
• Resources and capacity
• Fair procedures
• Communication
• Agenda-setting
• Informal/institutionalized contact (Appendix 14)
This kind of coding enhances the possibility of gaining a structure and an overview of the entire empirical data (Kvale 1996:199). This structuring is especially relevant given the in-depth and semi-structured nature of the interviews where the conversations often exceeded the interview guide, cf. section 3.3.1.
The use of coding in analyzing the collected empirical data has often been criticized for simplifying and fragmenting the empirical data resulting in a simplification of the analysis (Bryman 2012:578). The use of coding is in this particular study not based upon an idea of quantifying and counting the themes seen in the data (Kvale 1996:199) in order to simplify the complexity of the social reality. Rather the aim is to keep the interviews as “thick descriptions” (Torfing 2004:73), i.e. maintaining their complexity.
The themes are therefore processed according to their relevance for answering the research question and not for their quantity. The quantity of representations of a theme is of no importance, whereas the identifiable similarities and differences in the interviews concerning each code are of real importance and is what will be analyzed upon.
According to Torfing (2004) the process of coding is closely linked to interpretation (Torfing 2004:77). It is not possible to code a document without making an immediate interpretation of the relevance of the specific quote. The process of coding thus entails threats to the validity of the analysis of the empirical data.
These critiques of the use of coding are sought countered in this study by keeping the codes rather broad and inclusive. Furthermore, the focus on the interviewees’ point of view has been sought kept throughout the coding, and there has been emphasis on the possibility of new themes to appear from reading the material. According to Kristensen it is likely that there might appear themes that are not included in the codes but still seem to have relevance for answering the research question (Kristensen 2010:291).
When coding, the focus is on themes that ‘stand aside’, seem important, and which cannot be fitted into any of the former codes. The relevance of these themes can be investigated through cross readings of the empirical data, while coding after a new theme. This process can enhance the precision and the nuances in the empirical data (Kristensen 2010:291). In this study it has manifested itself in the development of the code Resources and Capacity (Appendix 14).
The codes have been explained, making a review of the analysis possible, which increases the reliability of the study (Kvale 1996:209). The internal consistency of the study can be enhanced by this explanation since it makes mutual examination possible despite different interpretations, when multiple persons are coding the empirical data (Kvale 1996:208). Disagreements have been sought resolved through debate of the interpretation and the understanding of the different codes, which can be seen in the Appendix 14.
3.5.2.2 Meaning Condensations
The coding process is a complex work, which can result in the loss of context and the interviewees’ own sense of meaning (Kristensen 2010:291). To prevent this, there have been conducted meaning condensations of all codes in all interviews. According to Kvale (1996) the advantages of using meaning condensations when coding qualitative data, is that they ensure a certain loyalty in the exposition of the interviewees understanding of self and surroundings (Kvale 1996:194). These condensations are descriptive and founded in the interviewees’ point of view and we have aimed not to interpret, by staying faithful to the interviewees’ narratives.
Thereby the use of meaning condensation can be seen as a control mechanism ensuring that the complexity in the interviews and in the codes is not lost (Kristensen 2010:291).
There has thus been written meaning condensations of every code of every interview.
By only condensing the meaning of each interview the complexity of the themes would be lost, while condensing the meaning of each code but across all interviews would result in loss of complexity in the meanings concerning each code.