classes are far from each other. The most important outcome of partitioning the actors into clusters is to gain interpretability of the clusters. Clustering thus depend on the definition of similar (Wichern & Johnson, 2013).
In simplicity, the number of clusters can be chosen by looking at the overall shape of the dendrogram and the plot of the within-group variance gains (Husson et al., 2010).
theories. Constructivists are less interested in the general laws of explanations, but instead seek to map the different forms of explanations that reveals patterns. Therefore, the overall research design is more of exploratory nature, despite using a confirmatory approach in this process, and leans more towards constructivist philosophy of science, because it seeks to identify socially constructed patterns, realizing that they are context bound and based on the precept that each actor participates in the construction of the world:
“We cannot trust inductive reasoning to produce general statements that are true; (…) because observed events can never embrace all possible objects/events of the world” (Moses & Knutsen, 2010).
The two methods employed in this thesis – network analysis and correspondence analysis are both quantitative methods, which in each way can contribute to identifying socially constructed patterns. A network analysis is necessary, because actors are related, and these relationships can be complex. The social network analysis can help to reveal the structure and dynamics of such relationships. Instead of perceiving complex relationships as nothing more than complex, network analysis can provide empirical insights into the extent of which actors are related and how. The focus of this thesis is on the explanatory power of ties, and how they can channel information flow and norms to facilitate contagion. The network analysis therefore relies on an understanding of the social world in which ties are a proxy of holding power. A correspondence analysis is necessary to detect and uncover the underlying structures in the data set. It can analyze complex and large amounts of data and visualize patterns and reflections of such data in a simplistic way. Therefore, these two methods serve as aids to discuss and uncover other patterns in the data, which might explain their homogeneity, other than what the linear regression can answer.
6 Data and processing
The analyses of this thesis are based on two types of data. The network analysis uses relational data extracted from the EU’s database for lobbying organizations, the EU’s Transparency Register. Categorical survey data for the correspondence analysis have been based on an online questionnaire survey issued by the European Commission. This section will go into details on how these types of data have been collected and processed in the thesis.
The European Commission is the institutional organ responsible for preparing and updating legislation in the EU. In the preparatory stages of a legislative proposal, it is standard procedure for the European Commission to conduct consultations of external stakeholders. The main purpose of the external consultations is to provide the Commission with information it would not otherwise be able to obtain. Subsequently, these consultations provide opportunities for external stakeholders to formally engage with the Commission and to influence the policy draft outcome of a legislative proposal (Hardacre & Akse, 2015, p. 29). The external consultations can take the form of either online surveys, open hearings, bilateral meetings, workshops and/or events. The
Commission has multiple tools available to conduct these consultations, and they are often used in combination depending on the Commission’s needs. A majority of the formal types of consultations (i.e. survey and open hearings) are made accessible to the public. These provide great insights into external stakeholders’ opinion on a delimited and specific political issue, and furthermore provide a unique insight into the types of external actors having invested lobbying-resources in the given issue. The policy-making process of the EU is a rather transparent and delimited procedure in the EU, where information often is made accessible to the public.
Furthermore, it is a process that (at least on paper) is mostly free from influence stemming from other EU institutions.
The European Commission issued an online survey available in the period 20th of July till 28th of October 2016 as a part of its consultations of external stakeholders in preparations for the legislative proposal. The Commission identified eight groups of stakeholders as subjects for the external consultation. Among these were automobile manufacturers, component and material suppliers, energy suppliers, drivers associations, environmental, transport and consumer organizations, social partners and member states (European Commission, 2017). In addition to the online survey, the Commission’s consultations of external stakeholders included workshops, bilateral meetings with industry associations and member states. However, due to the limited accessibility of data from these events, the online survey has been selected as the subject of analysis in this thesis. The other types of consultations will not be touched further upon.
The survey was divided into two main sections, the first, which asked questions of a general nature, including the need and objectives for EU action relating to the topic, and the second of more technical nature asking questions related to the policy design intended for well-informed respondents. It was voluntary whether respondents wanted to complete only the first section or both. The survey contained a maximum of 54 questions depending on the individual responses of the respondents, i.e. some of the questions contained a number of sub-questions depending on the previous answer of the respondent. For each question, a number of choices were given to the respondent, for example: “In your view, how important is the following action?”
(Question 1) of which the respondents were given the following choices: “Very important”, “Important”,
“Somewhat important”, “Not important”, or “I don’t know”. Another example: “In comparison to the current reduction rates, do you think new targets for the period after 2020 should be set at levels which require:”
(Question 5) of which the respondent were given the following choices: “A rate of reduction less than that required under the current regulations”, “a similar rate of reduction to that required under the current regulations”, or “a higher rate of reduction than that required under the current regulations”. For some of the questions, respondents were also given the to opportunity to answer the question with their own remarks. All questions can be found in appendix 3.1. In general, the survey addressed the following five key issues:
1. The need and objectives for setting CO2 emission targets for cars and vans after 2020
2. Technology specific requirements
3. Division of responsibility among different types of actors 4. Incentivizing low- and zero-emission vehicles
5. Modalities (eco-innovations and derogations)
Aside from the technical questions, respondents were asked to state their 1) name, 2) transparency register number (elaborated in the next section), 3) country of residence, and 4) to select which of the following group they belonged to: individual/private person, public authority, academic/research institution, international organization, civil society organization, professional organization, private enterprise or ‘other’. The online survey was fully accessible for all interested stakeholders inside and outside the European Union during the period of the survey.
The Commission received a total of 205 responses, distributed as follows: 82 of these were individuals or private persons, 33 civil society organizations, 31 professional organizations, 28 private enterprises, 11 public authorities, 6 academic institutions, 4 international organizations, and 10 classified as ‘other’. Responses originated from Belgium (34), Germany (26), Netherlands and Denmark (17), France (15) and Hungary (13).
A few stakeholders originated from countries outside the EU, including Japan (4), Norway (1) and ‘global’
(1). Of the 205 respondents, 106 car manufacturing companies have been identified by this thesis, see Table 2.
Car manufacturers
1 Bayerische Motoren Werke Aktiengesellschaft (BMW Group) 2 Honda Motor Europe
3 Mazda Motor Europe
4 Mitsubishi Electric Europe B.V. Brussels Representative Office 5 Nissan International
6 Groupe Renault 7 Skoda Auto
8 Suzuki Motor Corporation 9 Toyota Motor Corporation
10 Volkswagen Aktiengesellschaft (Volkswagen Group)
Table 2: The 10 car manufacturers involved in the survey questionnaire of the Commission.
From now, just referred to as BMW, Honda, Mazda, Mitsubishi, Nissan, Renault, Skoda, Suzuki, Toyota and Volkswagen (VW). These car manufacturers are what constitute ‘the automobile industry’ in this thesis and
6 Toyota has submitted two identical responses in the survey, one from “Toyota Motor Corporation” and one from
“Toyota Motor Europe”. Since the two responses are identical, this thesis has only included the response from “Toyota Motor Europe” in the analyses. The same applies for Volkswagen, and therefore only the response from “Volkswagen Aktiengesellschaft” has been included.
they are the subject of analysis both in terms of the network analysis and as a part of the correspondence analysis.