2 Literature review and Analytical Framework
3.4 Sources of Data Collection and Methods of Data Analysis
- review of relevant literature;
- analysis of relevant policy documents (including parliamentary records, policy reports, government publications);
- analysis of semi-structured interviews.
It triangulates between the information from those different sources, which is a recommended strategy for collecting evidence in case study research (Yin 1994; George and Bennett 2005). In the following sections these three sources of data will be discussed in more detail.
3.4.1 Relevant literature
The relevant literature includes publications on energy, technology, and environmental and innovation policy in the Netherlands and the UK. The literature review was based on a systematic search through the Web of Science using keywords such as energy, innovation, sustainability, transition, discourse, UK, Netherlands, etc. The IEA national energy policy reports for the UK and the Netherlands have also been a valuable source (IEA 2000; IEA 2002; IEA 2004a; IEA 2007). Important for the Dutch case study especially are publications which have looked at the implementation of the transition management approach in the energy domain (Hofman 2005; Kemp and Loorbach 2005; Kemp, Rotmans et al. 2007; Loorbach 2007; Nooteboom 2007; Dietz, Brouwer et al. 2008; Hendriks 2008;
Loorbach and Kemp 2008) because they can help to triangulate the findings of this case study.
My involvement in a study of the Dutch energy transition led by Adrian Smith not only funded my fieldwork, but also helped shape my initial thinking and led to two co-authored publications (Kern and Smith 2008; Smith and Kern 2009)8. The joint project showed that
8 The empirical material for these publications was gathered by the author of this thesis and he has been active in the detailed research design, was responsible for all the practical research effort, undertook the key preliminary analysis of research materials (documents and interviews) and co-authored two papers with Adrian Smith. The division of labour in the project was largely identical to the supervisor-research student relationship. The main difference was that the project was initially framed by Adrian Smith. The analysis of the ‘Energy Transition’ presented in this thesis is substantially different from the Smith and Kern 2009 paper.
The analysis presented in this thesis tries to overcome some of the weaknesses of this paper, which did not
existing institutions as well as interests were important factors in explaining the policy processes and thus led me to draw extensively on the institutionalist literature in creating the framework presented above and to pay more explicit attention to the conceptualisation of the relationship between discourses and interests.
The UK literature on the subject is helpful in providing insights into, for example, the dominant framing of UK energy policy and the historical evolution of this policy area, but there are no publications directly analysing the Carbon Trust. The relevant literature has been thoroughly and systematically reviewed to inform the interviews and to triangulate conclusions derived from the analysis of the interviews and policy documents.
3.4.2 Relevant documents
The analysis of relevant government policy documents offers insights into how these policy initiatives are justified and explained. Policy goals and details of the implementation of the initiatives and subsequent institutional changes are provided. This includes reports from the respective ministries in both countries (mainly Environment, Economy, and Research), and the International Energy Agency, as well as from public organisations like the Carbon Trust, the National Audit Office in the UK and the Energy Transition Taskforce, SenterNovem, and government advisory councils such as the General Energy Council or the VROM council in the Netherlands.
For the UK case study the parliamentary records were searched for debates on the Carbon Trust (key word search in Hansard). The analysis of the Dutch case study could not make use of parliamentary records due to language difficulties. However, since the case has been investigated by several Dutch researchers the availability of these secondary sources
explicitly differentiate between the content (ideas) and the processes of the discursive policy construction and neglected Hajer’s distinction between discourse structuration and institutionalisation. In contrast, the analysis presented in this thesis relates the emerging storyline more explicitly to the dominant discourse; makes use of more empirical detail, especially about the institutionalisation, to see which practices changed and why; and uses a broader framework which combines attention to discourses (Hajer) with attention to the institutional context (Schmidt) because the initial paper had shown the difficulties of institutionalising this new storyline.
This thesis also pays explicit attention to the discursive mechanisms which shaped the policy initiative.
compensates for the lack of access to primary information. Relevant documents also include policy evaluations by public authorities in English or independent organisations as well as documents authored by stakeholders (such as NGOs, industrial associations, etc.) It is important not to privilege official discourses by limiting the analysis to official policy documents only, but also to look at opposition and resistance and their rationalities (Oels 2005: 193). Similarly, Dryzek has argued that in assessing the effects of a discourse it is important to look at the arguments of the critics of a discourse (1997: 20). The analysis therefore relies not only on official documents but also includes documents from stakeholders and critical voices. The relevant documents have been thoroughly and systematically reviewed to inform and triangulate the expert interviews, identify storylines and gain insights into how the discourses have or have not shaped changes in policies and institutions.
3.4.3 Semi-structured interviews
Both case studies also make use of semi-structured interviews. A snowballing sampling method was used (see Arksey and Knight 1999; Tansey 2007). Tansey has argued that because the goal of process tracing is to
“obtain information about well-defined and specific events and processes, the most appropriate sampling procedures are thus those that identify the key political actors – those who had the most involvement with the processes of interest” (Tansey 2007: 765).
Instead of random sampling the aim is to include the most important actors who were involved in the process to be studied. Generalisation from the actors interviewed to a large population of actors is not what is aimed for. Rather, the research was interested in their personal accounts of these processes and what role they played in them. To be able to include first-hand accounts of actors involved in the processes under investigation is one of the main advantages of elite interviews compared to the analysis of the official version of events found in documents (Tansey 2007). Interviews with key actors can shed light on the informal processes underlying decision-making processes (George and Bennett 2005). Also, Hajer, Schmidt and Radaelli point to the importance of interviews for exploring causal
sequences (Schmidt and Radaelli 2004: 205). Interviews are particularly useful to get a better understanding of the meaning that interviewees attach to particular events (Kvale 1996: 105).
The snowball approach used for the purposes of this thesis started from existing contacts with researchers and civil servants in both countries who helped to identify key actors involved in the particular policy processes being studied. Interviewees were then asked for further suggestions for other important actors involved in the area under research, and these were then contacted if two or more interviewees put forward their names. This process was stopped when interviewees began to repeat the names of interviewees already contacted. In parallel to this, documents were screened for names of actors mentioned as important. To avoid the selection being skewed by interviewees suggesting actors with similar characteristics and views, and in order to interview a diverse set of actors, an attempt was made to create a balance based on two characteristics. Balance can be sought by deliberately seeking opposing positions (Sharp and Richardson 2001), so by including
‘insider’ and critical ‘outsider’ views on the two initiatives it is hoped that a balance in the representation of the policy processes can be achieved. Secondly, balance was sought by seeking actors from different domains of society: government, private firms, academia and non-governmental organisations (NGOs) such as environmental or other lobby groups.
Between January and May 2006, 27 semi-structured interviews were conducted in the Netherlands with actors involved in the implementation of the ‘energy transition’ project, or those who were familiar with it but were not directly involved, from government, industry, civil society and research organisations. 26 semi-structured interviews were conducted for the UK case study between October 2007 and March 2008. These were with current and former employees of the Carbon Trust and key stakeholders from government, research, business and environmental groups. For a list of interviews conducted, see Appendix A.
The interviews were conducted as semi-structured, exploratory interviews to allow for openness, while maintaining a focus on the key issues to be addressed. Exploratory
interviews introduce an issue and then follow up on the interviewees’ answers, seeking new information about and new angles on the topic (Kvale 1996: 97). Key issues were:
- how the interviewees explained the rationale for the new policy initiative and what policy problem they were hoping to tackle;
- why the particular approach was chosen;
- how the new initiative came about and which key actors were involved;
- experiences in the implementation of the policy initiatives;
- reflections on the successes and problems of the initiatives,
- and potential policy and institutional changes that followed from it.
The interviews were based on a flexible schedule depending on the role of the interviewee (for an example of this, see Appendix B), which included a sequence of themes to be covered as well as suggested questions. At the same time, however, there was openness to changes of sequence and form of questions in order to follow up on the answers given (see Kvale 1996). Most of the interviews took place face to face but some were conducted over the phone. All interviews have been recorded and transcribed. All interviewees were assured that they would remain anonymous to enable them to speak openly.
3.4.4 Methods of data analysis
In accordance with Hajer’s suggestion the steps detailed below were followed in conducting the analysis. In the document analysis the research looked for structuring concepts, ideas and categorisations as well as the use of storylines, metaphors, etc. This analysis yields “a basic notion of the process of events as well as the sites of discursive production” (Hajer 2000). The analysis of the interview data can be used to generate more information on causal chains and to gain a better understanding of the meaning of particular events for the interviewees which, for example, have led to a re-framing. This can lead to the identification of key events which are essential for understanding the discursive dynamics.
If meaning is central to the study of policy processes, this has implications for the methods of empirical analysis because “as meanings are not directly observable, the realm of meaning has to be approached through reflection and interpretive analysis” (Fischer 2003:
139). This information on meaning and causal mechanisms is useful to reconstruct the discourses which actors are drawing on and to analyse particular cognitive shifts (reframing) which are of interest for the analysis. To this end, the researcher has to try “as it were, to get inside the heads of the particular players in an effort to figure out the thinking behind the actions at issue” (Fischer 2003: 141).
The analysis is therefore based on meaning interpretation inspired by hermeneutical philosophy. Kvale describes this approach as follows:
“The researcher has a perspective on what is investigated and interprets the interviews from this perspective. The interpreter goes beyond what is directly said to work out structures and relations of meaning not immediately apparent in a text. This requires a certain distance from what is said, which is achieved by a methodological or theoretical stance, recontextualizing what is said in a specific conceptual context” (Kvale 1996: 201).
The context for this interpretation is provided by the main analytical concepts of this thesis:
storylines, processes of coordinative and communicative discourse, institutional rules and norms and the structuration and implementation of the discourse. To structure the rich empirical data of the policy documents, secondary sources and interview transcripts in line with these key concepts, a manual coding technique was used.
3.5 Operationalisation of the main concepts of the analytical