Bias in stakeholder
representation:
The role of salience
and policy type
Noa van Hunen, S2096862
Leiden University
Thesis submitted for MSc Public Administration –
International and European Governance
Supervisor: Dr. C.H.J.M. Braun
January 20, 2019
Content
• Introduction p.2
• 1. Literature review p.4
o Risk governance in context p.4
o Risk governance and stakeholder consultation p.5 o The added value of legislative lobbying literature p.7
• 2. Theoretical framework p.12
o Bias in stakeholder representation and its dimensions p.12
o Theoretical model p.14
• 3. Research design p.18
o Case selection and data collection p.18
o Dependent and independent variables p.20 o Statistical tests used for data analysis p.21
• 4. Results and discussion p.22
• Conclusion p.36
• Bibliography p.39
• Appendices p.44
o Appendix A: Dataset p.44
Introduction
Stakeholder involvement is an often-discussed topic amongst scholars in different scientific disciplines. When it comes to public administration it is important that societal interests are taken into account. Stakeholders play an important role in representing these various societal interests in decision-making to increase its democratic legitimacy and quality (Greenwood, 2007). Interest groups can consist of any part of society and represent any societal interest. Regarding risk governance inclusion of a large variety of interest groups is crucial, because it allows policymakers to compile a complete overview of all societal interests and views concerning potentials risks in legislation involved (Van Asselt & Renn, 2011). Simply enough, risks cannot properly be assessed and legislated without the input of a wide variety of stakeholders. The European Union (EU) is employing a more inclusive outreach to stakeholders. The use of open consultations is one example. However, in practice it appears that open consultations are often not that inclusive after all. When it comes to the representation of stakeholders, it seems that this can be highly unequal: even in open consultations there is bias in stakeholder representation1 (hereafter referred to as ‘bias in SR’).
The aim of this research is to contribute to finding an explanation for this inequality in stakeholder representation by examining whether certain factors mentioned in academic literature can be associated with this. Recent literature on stakeholder involvement in general points out the potential role of salience and policy type on stakeholder representation (Berkhout et al., 2015; Klüver, Braun, Beyers, 2015; Rasmussen & Carroll, 2014). Therefore this research focuses on these factors. Consequently, the main research question is: Are salience and policy type associated with the level of bias in stakeholder representation in open consultations? To answer this question, two hypotheses are examined: H1: A higher level of salience2 regarding
open consultations can be associated with a more equal distribution of stakeholder representation, a lower level of salience can be associated with a more unequal distribution of stakeholder representation, and H2: Open consultations based on regulatory policies can be associated with a more equal distribution of stakeholder representation, those based on (re)distributive policies can be associated with a more unequal distribution of stakeholder representation.
This research is of academic and societal relevance. With regard to academic relevance, there is a gap in risk governance literature when it comes to stakeholder involvement. In risk governance studies it is mentioned that stakeholder participation in the policymaking process is necessary because to make a balanced decision they need an overview and in-depth knowledge of which possible risks are at play for which groups in society (Renn, 2015; Renn, Klinke & Van Asselt, 2011; Van Asselt & Renn, 2011, p.439-443). Nevertheless, stakeholder engagement is not extensively and sufficiently explored. For instance, it does not examine why certain interest groups are over or underrepresented in (open) consultations. To fill this gap, this study links risk governance literature with legislative lobbying literature. Scholars in the
1 The definition of bias in stakeholder representation is given in the first part of the chapter Theoretical
framework.
2 The definition of salience in open consultations is given in the second part of the chapter ‘Theoretical
field of legislative lobbying have written extensively about such issues. Connecting these two strands of literature leads to a deeper insight into the factors that should be considered when it comes to involving stakeholders in risk governance. Additionally, the empirical findings from this research are complementary to existing risk governance literature. It is useful to know what factors can and cannot be related to the level of bias in SR or what not, to determine how a more equal inclusion of stakeholders in decision-making can be accomplished. Consequently, this is also of societal relevance because policymakers can use this knowledge, to reduce the bias in SR. When they would put out an open consultation, they would already know what factors are at play and how they could influence the outcome. Policymakers would be able to anticipate the outcome in terms of how varied the representation of different groups of stakeholders will be. Ultimately, they might even be able to adjust or change open consultations in such a way that there is less or no bias in SR at all.
To test if salience and policy type are associated with bias in SR a large-N research design is used. For this study 42 open consultations of the European Commission (EC) within the policy area Public Health have been selected. Respondents of each consultation were coded on two aspects, namely group type (citizen groups, professional associations, business associations and trade unions, firms, research institutes, public authorities or other) and geographical orientation (subnational, national, European, international). In this way the equality of the distribution of stakeholder groups can be analyzed for two different dimensions of bias in SR. A more unequal distribution of stakeholder groups indicates a higher level of bias in SR. For both hypothesis the association with the (in)equality of the distribution of stakeholder groups was tested for two dimensions, namely on the distribution of the group type, and on the distribution of the geographical orientation.
The rest of this study includes four chapters. It starts with a literature review. Previous research on risk governance regarding stakeholder involvement is discussed and legislative lobbying literature is connected to risk governance literature. The chapter ends with the finding that there is bias in SR in open consultations, which is further discussed in the chapter on the theoretical findings. Secondly, a theoretical framework is presented, in which the notion of bias in SR is explored and two important dimensions of stakeholder representation are highlighted: group type and geographical orientation. Additionally, the hypotheses are constructed based on two factors that may be associated to the level of bias in SR: salience and policy type. In the third chapter the research design will be described. It reviews the data collection and the methods used to code and analyze the data. Finally, the empirical findings are presented and discussed, and eventually linked back to the framework of risk governance.
1. Literature review
Risk governance literature often addresses the need for stakeholder consultation, but it does not extensively discuss aspects such as the lobbying process, what type of stakeholders are or are not engaged or the complications that can arise during the consultation process. What is lacking in risk governance studies however, is extensively discussed in studies in the field of legislative lobbying. Exploring this gap in literature and establishing a link between these two different strands of literature forms the core of this chapter. Before addressing the connection between stakeholder consultation and risk governance, and how legislative lobbying literature can be of added value to this, first it will be briefly examined what has been written on the concept of risk and good risk governance.
Risk governance in context
The concept of risk governance falls within the broader framework of governance, and in turn the transition from government to governance. Without the emergence of governance, it would of course never have been possible to add the word ‘risk’ to ‘governance’. During the 1980s a large group of scholars found that the world was becoming increasingly global, an interdependent and interconnected environment, and that governing started taking place outside of the government (see for example Cleveland, 1980 and Boyer, 1986 as cited in Boyer, 1990, p.51). Actors that were not part of any governmental agencies started engaging in what we now call ‘governance’. Boyer (1990) stated that there would be “more and more governance, and less and less government” (p.51). There was a lot of debate between scholars on whether the nation-state really was becoming of less importance in favour of this new ‘governance’ which would involve non-governmental actors (see for example Almond, 1988; Evans, Rueschemeyer & Skocpol, 1985). While Keohane and Nye, for instance, supported the notion of governance, they also stated that the nation state would not disappear altogether (Keohane & Nye, 2000, Introduction). However, many did believe in the transition to governance because they realized that transnational issues could often not be handled on a national level by individual governments (Benn, 1971; Boyer, 1990, p.51).
The emergence of risk governance began as a vague notion before it became an important shift in paradigm. Earlier ideas on the meaning of risk originated some time even before the transition of government to governance. Knight (1921) was one of the first scholars to define the notion of risk. For a long time, his views on the concepts were quite dominant. As an economist he described risk in terms of probabilities and effects, in which risk and uncertainty can be measured and an adequate response can be calculated (Knight, 1921, p.197-232). Eventually scholars started using this concept in the context of governance. In his book, Ulrich (1992) writes about risk and how states are more and more preoccupied with risks and how to organise governance in such a way that risks can be dealt with adequately. Although in Ulrich’s book the two concepts are connected (‘risk’ and ‘governance’), risk governance is not explicitly mentioned. The articles of Elliot (2001) and Heriard-Dubreuil (2001) were the first to use the term in their titles, and Hood, Rothstein and Baldwin (2001) were one of the first to directly discuss governance in the context of risk related decision-making in their book. Although they now used the term risk governance as a notion, none of them provided a concrete
definition of risk governance. Nevertheless, risk governance became an important shift in paradigm. Ultimately risk governance became of actual importance was when the International Risk Governance Council (IRGC) was established (Van Asselt & Renn, 2011). A few years later Renn and Walker (2008) were the first scholars to include the concept of risk governance in their work. Nowadays risk governance is an aspect of governance. It can be defined in two ways: 1) as the critical study of complex, interacting networks in which choices and decisions are made around risks” and “as a set of normative principles which can inform all relevant actors of society how to deal responsibly with risks.” (Van Asselt & Renn, 2011, p.443).
Risk governance and stakeholder consultation
What is of more interest however, is the role of stakeholders in risk governance. Van Asselt and Renn (2011) came up with three risk governance principles: 1) “communication and inclusion”, 2) “integration”, and 3) “reflection” (p.439). According to them these principles should play a role in all stages of risk governance decision-making processes (Van Asselt & Renn, 2011, p.439-443). This seems quite logical. If you take a closer look at the first principle, it is about involving all actors that play a part in the governance process. Decision-makers need to have an idea of what these actors perceive as risks to have insight in what the societal demands are (Renn, 2015, p.8). They need this variety in perspectives to get a balanced overview of possible risks. If they were to only have one-sided input, their final decision could also be one-sided and would only benefit a certain part of society. The more perspectives they have, the more balanced the outcome (Renn et al., 2011). As for the second principle, integration, the authors state that risk assessment and risk management should not be two separate things but should form one whole (Van Asselt & Renn, 2011, p.441-442). They are interconnected. For instance, while risk management is conducted, it is beneficial to keep assessing the situation in order to achieve successful risk management. The last principle, reflection, is actually slightly intertwined with the first one. Actors that are somehow affected by the decision-making should also be involved in giving constant feedback and evaluation on the work that is going on (Van Asselt & Renn, 2011, p.442-443). In short, stakeholders should be present in the entire process of risk governance.
Although the principle of communication and inclusion3 resonates within many studies
in the field of risk governance, risk governance studies do not systematically explore and/or explain all aspects of stakeholder consultation. This points to a gap in risk governance literature. Firstly, some risk governance studies have a tendency to only focus on the need for stakeholder consultation. It is often argued that actors that are not included in the policymaking process possess a lot of highly relevant knowledge and information (Kern & Bulkeley, 2009). Decisionmakers could benefit of this knowledge and improve the efficiency of the decision-making process or the decision itself. Additionally, all stakeholders that are affected by the decision-making process view and conceptualize the situation in a different way, and they all try to encourage the decision-makers to listen to their take on the situation (Lidskog, Uggla & Soneryd, 2011). Aven and Renn (2012) also find that in a democratic society, stakeholders simply need to be included in policymaking not just because the different viewpoints of society
3 Hereafter referred to as ‘principle of inclusion’ because the focus of this research lays on inclusion and
should be respected, but mainly because it is necessary in order to have the support of the stakeholders. A more important need to involve stakeholders however, seems to be the need for a balance in how over- or under-protective the measure against the risk in question should be (Klinke & Renn, 2012; Renn et al., 2011). Of course, the risk is in the eye of the beholder. For instance, some stakeholders will foresee many risks whilst others will not, or some will see risk in a certain aspect of the proposed legislation while others see risks in another part. Accordingly, if a lot of stakeholders are involved in a consultation, the decisionmakers get a better overview of all risks identified by various groups on various levels. Naturally, if only a few participate in such a consultation, the balance in what measures are overly cautious or not cautious enough, is off.
Secondly, in risk governance literature it is often only stated to what extend stakeholders should be involved in decision-making. In the articles of Renn, Klinke and Van Asselt (2011) and Klinke and Renn (2012) risks are divided along three characteristics: complexity, uncertainty and ambiguity. The level of stakeholder involvement depends on these characteristics (Klinke & Renn, 2012; Renn et al., 2011). Klinke and Renn (2012) even provide a decision tree of what type of risk management should be applied for when different characteristics of risk are at play (p.289). In case the level of complexity, uncertainty and ambiguity is low, stakeholders can still be included in order to make the decision-making process more efficient because they have the expertise (Klinke &Renn, 2012, p.286). When it comes to the necessity for stakeholders in providing a balance in cautiousness, this is especially important when there is high uncertainty (Klinke & Renn, 2012, p.287). Furthermore, they even go as far to say that when there is a high level of ambiguity, the general public should also participate (Klinke & Renn, 2012, p.287-288). Nevertheless, although all authors point out why it is necessary to include interest groups into the policymaking process and to what extent this could be done according to three elements of risk, they do not mention how policy-makers can involve them successfully. Hence, this points to a gap in the literature.
On the other hand, there are also scholars who, rather than focusing on these three separate characteristics of risk, claim that stakeholders should be involved in all stages of the decision-making process. In their case study on chemicals in the Baltic Sea, Karlsson, Gilek and Udovyk (2011) also examine the role various actors played in the risk governance process. They find that because in this case there were so many scientific uncertainties as well as complicated social and political issues, all stakeholders were arguing about the available missing data (Karlsson et al., 2011, p.144). They would go as far as to try to have an impact on risk governance on a national and international level (Eriksson, Gilek & Rudén, 2010; Karlsson et al., 2011, p.144). If stakeholders have high stakes in the process, then naturally they will try influence politics when they get the chance. However, in the article the authors not only say the stakeholders want to exert their influence, but they also say that objective and scope of risk assessment needs to be established through communication with stakeholders (Aven & Renn, 2012; Karlsson et al., 2011, p.151). Abt, Rodricks, Levy, Zeise and Burke (2010) agree with them and say that in every stage of risk governance interest groups should be included (p.1034). They go even further by saying they should be involved as early as possible (Abt et al., 2010, p.1034). This would allow decisionmakers to take into account the societal concerns that play a role in the matter right from the start. Consequently, societal interests would be taken into
account throughout the entire process, and not just later on. What is implied in the article of Karlsson et al. (2011) is shown more elaborately in the framework drawn up by Abt et al. (2010). Although their article zooms in on risk assessment, their framework shows in the scoping phase, as well as the risk assessment and risk management phase, stakeholders should be involved in the process to have a continuous oversight of all different views in terms concerning the risks (Abt et al., 2010, p.1034). However, once again the authors do not go much further than stating that stakeholders should be consulted.
Lastly, even when it is more explicitly stated how stakeholders should be involved, it is still written from the perspective of the decisionmakers and not the stakeholders. Any information on to what extent such consultations are beneficial for differing groups of stakeholders, for example, is not included. The article of Renn (2015) is one of few that actually discusses how interest groups can be engaged in the decision-making process. In Renn’s (2015) article the three characteristics of risk are still present (uncertainty, complexity, ambiguity). What is quite new here is that for each of these he describes what kind of stakeholder involvement is appropriate. The four formats, epistemic, reflective, participatory and hybrid, correspond to three separate characteristics of risk, respectively: complexity, uncertainty, ambiguity and a combination of these three (Renn, 2015, p.14). These formats provide policymakers with some kind of guideline of how to include interest groups when they are dealing with one or all of the challenges a risk can pose (Renn, 2015). A few examples are: stakeholder hearings, roundtables, public hearings and surveys and focus groups (Renn, 2015, p.15-16). It can and has been argued however, that the three elements of risk are not in fact the three elements of risk. De Vries, Verhoeven and Boeckhout (2012) propose that instead it is more about uncertainty, which can be translated into a risk. An uncertainty in itself does not necessarily have to be a risk. Only when there is some information, but incomplete, an uncertainty can turn into a risk (De Vries et al., 2012). This is a much more plausible explanation for why policymakers would involve stakeholders. When they have identified a risk but are left with many uncertainties, they will want to reduce hose uncertainties by acquiring as much knowledge as possible from external actors. However, the authors also do not go beyond stating in what way stakeholder could be consulted. They find there should be a different system of checks and balances and large companies, for instance, could play an important role because they as well can possess a lot of influence (De Vries et al., 2012, p.496). Accordingly, one component of a company’s risk governance strategy should be to include stakeholders in their strategical discussions (De Vries et al., 2012, p.496). In sum, while these studies propose different strategies of involving interest groups in risk governance, it only provides decisionmakers with minimal knowledge on how to engage them and they do not discuss any other important aspects of stakeholder involvement.
The added value of legislative lobbying literature
As has been pointed out in the previous section, there is a gap in risk governance literature. While risk governance literature claims there should be inclusion of stakeholders during many or all phases of risk governance, it does not reflect on several elements such as: what the real effect of stakeholder involvement is on the policymaking process, to what extent all interests are really included, what determines how successful interest groups can be in finding a way
into the governance process and to what extent stakeholders have access and influence into the policymaking process. However, this has been extensively discussed in legislative lobbying literature. In the review of this strand of literature below it becomes evident that the findings of these articles can be applied in risk governance literature. Linking these two strands of literature leads to a more complete accumulation of scholarly knowledge, that in turn results in a deeper insight into what factors should be considered concerning stakeholder engagement when it comes to risk governance. The following section starts with addressing lobbying, after which it zooms in on lobbying within the EU.
Regarding the principle of inclusion from risk governance literature, studies in the field of legislative lobbying show us that there is inclusion of stakeholders, but that this does not necessarily mean there is always an equal inclusion or representation of stakeholders in consultations of all societal interests involved. First of all, in earlier literature on lobbying it was evident that interest groups have a role to play in politics and that some have a more dominant role that others. Interest groups can represent some sort of freedom to through which the public can urge politicians to listen to their demands, to create a window for change (Baumgartner & Leech, 1998). However, some are better at mobilizing than others, which results in unequal opportunities to influence politics. If this were to be controlled this might also take away this freedom to create change in society (Baumgartner & Leech, 1998). This inequality in influence between different interest groups is evident in numerous articles. Salisbury (1984) noted that institutions, by which he means firms or local authorities, have a more prominent position compared to other stakeholders. He was not the only scholar to notice this. Schlozman (1984) even narrowed this down to just business interest groups. The more important things however, was not just that this group seemed to take on a more important role, but that this was done at the cost of other groups (Schlozman, 1984). Naturally, this leads to an unbalanced system if the interests of the general public and especially those that are less well-off are underrepresented compared to businesses. Of course, while these articles were based on politics in the United States, later articles also focused on the EU (see for example Hull, 1993; Richardson, 2000). All in all, many scholars noted that inclusion of stakeholders is often unequal, which could be problematic because this could lead to decision-making that is not representative of all or most societal interests.
Another way to look at stakeholders that have a more or a less dominant role is by examining the insider/outsider model. On the one hand scholars use this term in the literal sense: to point out that in stakeholder consultations there are those that have direct access to the decision-making process (the insiders), and those that do not (the outsiders) (Broscheid & Coen, 2003). Broscheid and Coen (2003) argue that this is the case in the EC, especially because it relies in highly specialist information from those groups in consults, which are usually private actors such as large companies. On the other hand, this model is not necessarily as simplistic as to say that those stakeholders that are considered to be insiders in the political system are much more powerful than those that are not, the outsiders. Maloney, Grant and Mclaughlin (1994) argue that it is more nuanced than that: outsiders can be those that have radical change in mind rather than incremental and that are thus not committed to long-term involvement within the political apparatus, whilst insiders are committed to long-term involvement. Moreover, they find that it is more important to make a distinction between core
and peripheral insiders, where core insiders have the necessary resources but where peripheral insiders have less influence (Maloney et al., 1994). In whatever way you look at the insider/outsider model however, it seems that some stakeholder groups have a more influential role than others.
Furthermore, when it comes to access to consultations, there are several factors that play a role. Aside from domestic politics, stakeholders also increasingly try to access consultations on a European level and change their strategy accordingly (Beyers, 2002). There are several theories on what determines how successful a stakeholder is in accessing the political system. One of them is about the “supply and demand for access good” in which the supply is the amount of expert information a stakeholder can offer, and the demand is the need of policymakers to obtain this information (Bouwen, 2002b, p.372). The better or more useful information a stakeholder would have, the larger its chance to access the political process. Beyers also adds that it seems that the more specific the interest of an association is, the more success it will have in not only accessing domestic politics but also in accessing politics in the EU (Beyers, 2002). However, how do scholars explain why for example groups with business interests are so highly present in consultations? Resources seem to be an important explanation, as well as the number of (potential) members of an organization, and how large the geographical orientation is that the organization focuses on (Berkhout et al., 2015). Businesses, or groups with business interests, want to provide legislators with their input, but in all likelihood not just because they would like to share their expert knowledge with policymakers. They might be incentivized to participate in stakeholder consultation due to factors such as the possibility to obtain more money (or funding), prestige and/or to gain influence. Then again, how rich and large an interest group is does not have to be the only rationale for why stakeholders can be overrepresented in consultations. Beyers and Braun (2013) argue that resources can be of less importance to an organization than network positions or coalition membership. When an interest group has a good position with its coalition and when, in addition, they have many ties with other groups or coalitions, the chance of obtaining access to the policymaking process would be high (Beyers & Braun, 2013). In sum, depending on various elements and circumstances, interest groups have a lower or higher chance of gaining access to the decision-making process.
Gaining access to the decision-making process is a prerequisite to gaining influence, but it can be quite difficult to measure how powerful stakeholders actually are as regards to influencing the decision-making process. Several single case studies provide some insight on the influence of stakeholders in the EU, and many more hypotheses exist but in most cases they have not been tested empirically (Dür, 2008). Furthermore, studies can be contradictory on what exactly influences how powerful interest groups can be. For example, in Mahoney’s (2004) research on demand-side forces it is claimed that it is the institutions of the EU that creates the demand for stakeholder mobilization. However, a few years later Berkhout et al. (2015) argue that institutions do not create a demand, but that they can influence the supply of interest group activity. Thus, they find that stakeholder mobilization is not caused by demand-side forces, but by supply-demand-side forces. The ability of a stakeholder to gain influence is related to its ability to gain access to the policymaking process. It seems logical that the more a stakeholder can bring to the table, the more leverage it has and the better it can exert its
influence in politics. Naturally, if the EU could choose which stakeholders to involve, it would choose those that can give the most useful information, that allows it to claim they have negated the so-called democratic deficit, and that provides the EU with the most economic influence (Klüver, 2013). The findings of Greenwood (2007), for example, make sense when placed in this context. The author finds that interest groups such as public interest organizations and cross sectoral business associations have more success in securing influence because they are linked to more than one constituency and can thus provide more citizen support, and the opposite would be true for territorial groups (Greenwood, 2007). In the same line of reasoning, those groups that have more influence are also those that have more resources and in turn more economic power (Greenwood, 2007). Groups that have more resources and/or represent a larger population could also provide the EU with more information, but this is merely a loose claim because if the EU needs very specific information this could also come from more local interest groups. Ultimately, how to measure stakeholder influence in various contexts remains a contested topic and it is not clear how to properly measure this (Beyers, Eising & Maloney, 2008; Arras & Braun, 2018). In the end, what is clear is that in order to gain influence, interest groups need to be heard first.
On a different note, during the past years, stakeholder participation in the decision-making process in the EU has evolved. The EU always seeks to improve its democratic legitimacy. It aims to do so via the Better Regulation reforms (European Commission, “Better regulation: why and how”). Through improving transparency the EU wants to increase its democratic legitimacy and regulate the involvement of interest groups in decision-making (Bunea, 2018; Mörth, 2009; Quittkat & Finke, 2008). Obviously, the focus of the Better Regulation reforms is to provide better regulation. To achieve this, the EU has decided it needs more input from a bigger variety of stakeholders and emphasizes the needs for open consultations to reach a wide audience (European Commission, Better regulation). By holding more public consultations the EU obtains input from numerous interest groups, which gives the EU a position of authority because it has access to all this information (Arras & Braun, 2018). Nevertheless, although increasing transparency may serve the needs of the EU, it may not directly serve the needs of the stakeholders. Regarding the Transparency Register, stakeholders perceive the effectiveness and sustainability to be low and are demanding reforms (Bunea & Ibenskas, 2017; Bunea, 2018). As long as it serves their lobbying purposes they support the system, until there is a possibility it can harm their access to and influence in the policymaking process (Bunea, 2018). The reforms that would like to see vary from deregulation and technocratic policymaking to a more participatory form of decision-making (Bunea & Ibenskas, 2017). Therefore, involving stakeholders has its benefits and costs, especially for the EU. On the one hand it provides the EU with expert knowledge and gives it a more powerful position. On the other hand, in cases where the EU is heavily reliant on such expert knowledge, stakeholders could get room to claim more influence, which could result in agency capture (Arras & Braun, 2018, p.15; and see Carpenter & Moss, 2014). Again, this points to 1) the need for a careful balance of stakeholder representation in (public) consultations so that some stakeholder interests do not overshadow others, and 2) the aim of stakeholders to try to gain a higher level of influence.
Moreover, even though public consultations do not pose formal access restrictions, inclusiveness, or: equal representation, proves to be an issue. This is not to say open consultations are not a good way to involve stakeholders. In fact, there are several advantages in using them. Using open consultations can ensure that policymakers are better informed, mainly because with the input they receive comes from a wide array of stakeholders they have a more complete picture of the issue (or risk) at hand (Bunea & Thomson, 2015; Quittkat, 2011). This is especially true for when public consultations are widely used, since it is a useful venue to engage with actors in civil society (Schulz & Newig, 2015). Therefore, this seems an especially convenient tool for the EU because they want to increase democratic legitimacy. This provides the EU with the opportunity to incorporate multiple societal interests in its legislation. Another advantage is that when many stakeholders take part in a consultation, this is also linked to more successful bargaining on the subject of the consultation, even when other legislators are reluctant (Bunea & Thomson, 2015). However, several studies have identified certain issues regarding inclusion and equal representation of stakeholders. Although the consultations are open to all, the inclusion of interest groups is very unequal: there is quite a high degree of bias in representation (Pedersen, Halphin & Rasmussen, 2015; Rasmussen & Carroll, 2014; Quittkat, 2011). Even though other types of interests are welcome to respond, businesses, for instance, are still overrepresented (Rasmussen & Carroll, 2014). When we take this into account, it is evident that the success of open consultations in not ensured. If the aim of the EU is to include a wide range of interests into their policymaking, a clearer understanding of why there is such a bias in the inclusion of interests needs to be provided. As regards to public consultations such explanations are quite lacking. One explanation that has been empirically tested is given by Rasmussen and Carroll (2014): the mobilization of stakeholders is dependent on who has to cover the cost of a certain legislation. Naturally, this cannot be the only explanation for bias in SR, and additional or alternative answers need to be provided.
Regarding the level of bias in SR in open consultations, this is exactly where this research makes a contribution to academic literature. Firstly, in a broader sense, this research provides a link between risk governance and legislative lobbying literature to fill in the gap of risk governance literature so that we have a more detailed exploration of stakeholder involvement in risk governance. There is no need for risk governance scholars to waste time rediscovering what has already been presented in studies in the field of legislative lobbying. When these two fields are connected, we gain much more insight into the matter of stakeholder consultation in risk governance. Secondly, recent studies have called for more research on the representation of stakeholders. Pérez Durán (2018) states that research needs to be done on proportionality of stakeholder involvement (p.251). This study does not focus on setting a normative standard on what the proportionality should be, but on how (un)equal interest representation is. More specifically, this study explores what factors are related to the (in)equality of the representation of various stakeholder groups in open consultations. The factors in question, salience and policy type, are examined in-depth in the next chapter. Overall, this is in line with the statement of Arras and Braun (2018) that the effect of “access instruments [(such as open consultations)] on the number and type of stakeholders” should be researched (p. 15). Subsequently, the main research question: ‘Are salience and policy associated with the level of bias in stakeholder representation in open consultations?’ will be answered within the framework of risk governance.
2. Theoretical framework
After having placed this research within the context of existing literature, this chapter examines the theoretical foundation of this study. The notion of bias in SR is explored in the first section, with a special focus on its dimensions, namely: group type and geographical orientation. Finally, the main theoretical model is laid out, including two hypotheses concerning salience and policy type.
Bias in stakeholder representation and its dimensions
The finding that the risk governance principle of inclusion is still an issue in open consultations seems paradoxical. Bias in access seems to be a barrier to stakeholder representation in consultations. However, when this barrier is removed, there is still inequality in the representation of interest groups. At first glance, it seems likely that if access to consultations is no longer restricted, many different stakeholders would take this opportunity to access consultations. Baumgartner (in Lowery et al., 2015) makes a similar claim: when resources no longer determine whether a stakeholder gains access to the policymaking process, when there are no such inefficiencies, then participation between those that do have many resources and those that do not will be equal (p.1213-1214). Having access to the policymaking process matters as well as having equal access, so that it is not always the same group of stakeholders that can participate in decision-making (Lowery, Halphin & Richardson in Lowery et al., 2015, p.1217, 1222-1223). Despite this, there are no such constraints to access in open consultations. As the name already implies, access is open to everyone, sending in a response is free of charge and all that is necessary is an internet connection. Yet, there is still a form of bias/inequality in the interest group system. The openness of this system may give more opportunity for inclusiveness of a wider array of stakeholder interests, but in practice there are large differences between the representation of stakeholders (Quittkat, 2011). Consequently, the bias in SR/inequality needs to be examined more in-depth.
Regarding this concept of bias in SR, this is not clearly defined but there is a recurring concern for this form of bias. On the one hand, there is the question of what fair representation of stakeholders is. Some scholars are open to the idea that some interest groups should be more represented in consultations, although they struggle with finding an answer to what a suitable balance in such a representation should be (see for example Schlozman in Lowery et al., 2015, p.1224-1225). According to Klüver (in Lowery et al., 2015, p.1219) it is sufficient if an interest is represented by at least one interest group. This is a very radical stance of course and in many instances this could still lead to a certain imbalance when some interests are represented by many groups and some just by one. Nevertheless, if you view representation of stakeholders in the light of risk governance and lobbying literature with the idea that it is crucial to involve many different stakeholders with different viewpoints on the perceived uncertainty and corresponding risks (Hojnacki in Lowery et al., 2015, p.1218; Van Asselt & Renn, 2011), asking ourselves to what extent different stakeholder should be involved is less relevant. What is much more interesting in this context is to accept that some types of interest groups are more numerous than others, and to discover which have a higher representation than others and what could be related to this.
Concerning this research, bias in SR stands for the inequality in representation between interest groups in stakeholder consultations. This study will not attempt to benchmark what (good) stakeholder representation means exactly. If you take the word ‘bias’ and say that it means ‘inclination’ or ‘prejudice for a certain group’, then we can view bias in SR in terms of how under- or over-represented some interest group types are compared to others. The level of bias in SR can tell us how inclusive open consultations are (Quittkat, 2011). Again, this is relevant because risk governance advocates for a highly inclusive form of governance regarding the involvement of stakeholders. Naturally, when making a statement about the representation of stakeholder groups, it is crucial of course to first know how the groups are classified. Different classifications of group types result in different outcomes in terms of representation (Baroni, Carroll, Chalmers, Muñoz & Rasmussen, 2014). Whether the representation between various stakeholders is more or less equal depends on how many group types there are and in which group types they belong according to such a categorization.
The concept of bias in SR can be explored through examining its various dimensions, two of them are emphasized in this research. The first dimension is group type, which stands for stakeholder type and is based on the interests advocated by such stakeholder groups. When it comes to classifying interest groups, Baroni et al. (2014) make a distinction between behavioral and organization characteristics. The authors favour the latter because it shows a strong connection between types of interest groups classified according to organizational characteristics and the background characteristics of the stakeholders (Baroni et al., 2014)4.
Nonetheless, if stakeholders are classified according to organizational characteristics, the stakeholders within each category share similar organizational traits and not similar interests. Since the focus of this study is on the degree of bias in SR, it is much more relevant to classify stakeholders according to group type. When you look at stakeholder groups classified according to group type, such as: firms, professional associations or research institutes, there is a common denominator for each group type. They all represent a certain type of interest, respectively: commercial interests, the interests of a certain group of professionals and the ability to carry out research. When it comes to the variety of interests in the dimension of group type, Berkhout and Lowery (2010) argue that the variation in interest groups has grown during the past few years due to a slow growth in traditional business groups. However, in many studies interest groups with business interests are usually overrepresented, and groups that campaign for, for example, environment or human rights are usually underrepresented (Baumgartner and Leech 2001; Berkhout et al. 2015; Lowery et al., 2015; Schattschneider, 1960; Schlozman, Verba & Brady, 2012; Wonka, Baumgartner, Mahoney & Berkhout, 2010).
The second dimension of stakeholder representation is geographical orientation. As might be expected, the overall representation of the various geographical orientations of stakeholder groups can be unequal. Some interest groups have a broader European or international focus while others are focused on a very small geographical area. In their study on business representation in economic sectors, Berkhout et al. (2015) find that there is a larger amount of interest groups when a sector is globally or European oriented, rather than on markets that are nationally oriented (p.466-467 & 473). Accordingly, this could mean that in
public consultations on future or existing legislation of the EU, interest groups with a broader geographical focus are overrepresented compared to those with a narrower geographical focus. In general, open consultations from the EU do focus on Europe as a whole and not individually on each Member State. In this context, it would seem logical that because of this wider European focus there are many interest groups that have a European focus. Then again, EU legislation eventually also impacts stakeholders that only represent certain interests on a sub-national level. Ideally there would therefore be a balance in representation of these different geographically orientated interests.
In addition to this, multiple studies have claimed that interest groups with a European or international scope have had better access to the EC compared to those oriented at a (sub)national level (Bouwen, 2002a, 2002b; Eising, 2007; Michalowitz, 2004). Rasmussen and Gross (2015) also find that with regard to advisory committees, groups that have a more European and global scope enjoy more access than those that have a much smaller scope (p.349). Again, this is plausible because if a regulation is focused on Europe as a whole, it is more relevant for interest groups that also focus on the EU. Furthermore, this is also a way for the EC to ensure its legitimacy (Rasmussen & Gross, 2015, p.346). The EC wants to ensure that they can take into account all different interests. By inviting interest groups that have an international or European focus, it allows them to cover more interest than when they would work with groups oriented at a national or sub-national level. It is a matter of who can speak on behalf of who? If the EC would not invite global or European oriented stakeholders, they would have to invite a larger number of stakeholders in order to ensure that they can take into account all European interests. It is also much easier to cooperate and negotiate with a smaller number of actors than with numerous interest groups that all have a national or sub-national geographical scope (Rasmussen & Gross, 2015, p.346). Although, the study by Rasmussen and Gross (2015) is on advisory groups, in which case the EC invites interest groups, this outcome can also be used in light of open consultations. Interest groups focused on a (sub)national level can still respond to a call for an open consultation, but their incentive could be lower than for European or international stakeholder groups. Legislation on EU level can appear quite far away or abstract for those groups with a narrow geographical orientation. They could have more interest in, for example, laws made directly by their governments rather than in regulations drawn up by the EU.
In short, in this study the level of bias in SR is viewed in terms of more or less equal representation of stakeholders in (open) consultations. More specifically, this can be done through looking at representation of interest groups in terms of group type or based on geographical orientation.
Theoretical model
Now that the notion of bias in SR has been thoroughly explored, the following part sheds light on the main theoretical model. The question is: can salience and policy type be associated with the equality of stakeholder representation in open consultations? This theoretical model leads to two hypotheses that form the core of this research.
Salience
The first factor that could be related to the level of bias in SR is the degree of salience regarding open consultations (hereafter shortened to ‘salience’). Whether salience is high or low could be related to how equal interest group representation is. The more attention a legislative proposal receives, the more stakeholders mobilize (Toshkov, Lowery, Carroll & Berkhout, 2013, p.50-53). Of course, this could be a snowball effect because when more stakeholders are mobilized, this might in turn lead to more attention in the form of, for instance, media coverage. There is not one clear definition of salience (Klüver & Braun & Beyers, 2015, p.451-452). However, in this study the concept salience stands for how important or noticeable the content of an open consultation is for stakeholders: how aware and informed are stakeholders regarding a policy or policy proposal featured in a public consultation? (Salience, n.d.). This can ultimately be linked to how involved stakeholders are in an open consultation (Klüver & Braun & Beyers, 2015, p.451). When a policy proposal receives little or no attention it is of course unlikely that many actors know about the issue. In this scenario it would be expected that there is not much conflict between various viewpoints of different stakeholders, and that the outcome of the consultation would show that the variety in the different interests that are represented is low. Of course, the opposite can also be true. When there is a high level of salience, there are probably more opposing viewpoints and a higher number of different types of stakeholders that feel the need to express their opinions on the legislation. More actors will be mobilized to respond to other stakeholder’s views (Klüver et al., 2015, p.451-452; Rasmussen & Carroll, 2013, p.448). In the end, this could be related to a more equally distributed representation of interests.
Moreover, when it comes to the level of salience, some actors try to use this to their advantage. By influencing the salience of a certain policy (proposal), the level of awareness on this topic can be influenced as well (Kollman, 1998). In the end, decision-makers can only take into account those preferences that are included in the consultation. The more diverse the responses, the less biased an online consultation will be regarding the representation of stakeholder interests (Rasmussen & Carroll, 2014, p.448). This variety in opinions and risk perceptions are beneficial for policymakers as it gives them a balanced oversight of all interests involved in the matter. However, for a stakeholder it could be much more advantageous if the salience of a policy proposal is low. The reason for this is that it is likely that those stakeholders that are aware of the issue and that do respond to the consultation have a larger chance of shaping the outcome, because their preferences do not have to compete with a great amount of other interest groups with opposing preferences (Klüver et al., 2015, p.452). Ultimately, a correlation between the level of bias in SR and the level of salience is plausible, because if all stakeholders affected were to voice their interests the inclusion of diverging interests is likely to be very high. This leads to the first hypothesis:
H1: A higher level of salience regarding open consultations can be associated with a more equal distribution of stakeholder representation, a lower level of salience regarding open consultations can be associated with a more unequal distribution of stakeholder representation.
Policy type
Another factor that could be associated with the degree of bias in SR is policy type. Klüver et al. (2015) expect that policy type can have an effect on the mobilization of stakeholders and on the way in which they lobby (p.454). As was first defined by Lowi (1964), there are three different policy types: regulatory, distributive and redistributive. Regulatory policies can be about two main things. They can be about behavior and prescribe the way an actor is supposed to act (Klüver et al., p.451; Lowi, 1964, p.690) or they have an impact on costs (Lowi, 1964, p.690). Regulatory policies can put behavioral constraints on an actor or extend its room for maneuver. Additionally, they can increase or decrease certain costs or decide who has the responsibility to fund a certain matter. Regulatory policies can often not be narrowed down to one individual but only to one sector because in order to regulate the behavior of one, a general rule is needed (Lowi, 1964, p.691&695). This type of policy is narrow in the sense that it is not directed to everyone, but it is targeted at a specific group (Lowi, 1964, p.691&695). The second policy type, distributive policies, are those that determine how much of a certain governmental resource should go to a specific part of the population (Klüver et al., p.451; Lowi, 1964, p.690). In contrast to regulatory policies, distributive policies are less general in the sense that specific resources are distributed to a specific societal group (Lowi, 1964, p.690). Finally, as the word ‘re’ plus ‘distributive’ already implies, the third type of policy is about determining which resources should be moved from one societal group to another (Klüver et al., p.451; Lowi, 1964, p.691). The overall aim here is to equalize various social classes (Lowi, 1964, p.691).
Furthermore, researching the level of bias in SR in light of regulatory and (re)distributive policy types can lead to new insights. Using this classification of policy types is quite different from the one used by Rasmussen and Carroll (2014). While in the beginning of their paper they do briefly address the distinction explained above, their research is focused on policy types classified as regulatory, expenditure and administrative (Rasmussen & Carroll, 2014). The authors conclude that in contrast to expenditure policies the mobilization of different interest groups in online consultations based on regulatory policies is less equal (Rasmussen & Carroll, 2014, p.454). When the authors refer to the inequality of stakeholder representation, they refer to how concentrated or diffuse the cost effects of a certain policy are on stakeholders (Rasmussen & Carroll, 2014). If costs are concentrated on a small group, only a small group will be interested in responding to an online consultation on that policy. Therefore, it makes sense to now research the level of bias in SR for regulatory policies in comparison to (re)distributive policies. This could lead to another outcome, while the results presented by Rasmussen and Carroll remain correct.
Whether open consultations are based on regulatory or on (re)distributive legislation can thus be important regarding the equality of interest representation. In contrast to Rasmussen and Carroll’s (2014) findings it is expected that public consultations based on regulatory policies are related to a more equal representation of stakeholders. This type of policy has direct impact on those it regulates and sometimes requires coercion, which makes it highly conflictual and will in all likelihood result in more bargaining between all actors involved (Smith & Larimer, 2009, “Another “Theory” of Public Policy: Policy Typology”). For example, these actors would either argue about who’s behavior should be targeted or about who should pay
(for, for example, certain aspects of a policy that is implemented). When it comes to distributive and redistributive policies, both can be associated with a more unequal representation of stakeholders in open consultations compared regulatory policies. This is because (re)distributive policies do not have such a direct impact on the behavior of stakeholders and consequently do not often require coercion to ensure the regulation (Smith & Larimer, 2009, “Another “Theory””). Aside from this, there are of course differences between distributive and redistributive policies. Distributive policies can allocate resources on a more individual basis whilst redistributive policies focus on a much broader group of the population (Lowi, 1964, p.690-691; Smith & Larimer, 2009, “Another “Theory””). The latter also leads to more bargaining than distributive ones because it involves a larger number of different societal groups (Smith & Larimer, 2009, “Another “Theory””). Since redistributive policies are more controversial, it demands more social rearrangements (Shull, 1997, p.127). Accordingly, for redistributive policies the likelihood that a conflict will arise is also larger (Smith & Larimer, 2009, “Another “Theory””). Yet, the difference between distributive and redistributive policies is less profound than the difference between, regulatory and (re)distributive policies. Subsequently, this study will not emphasize the difference between open consultations based on distributive and redistributive policies. On a more general note however, policy type can be related to the equality of the distribution of different stakeholder groups. This leads to the second hypothesis:
H2: Open consultations based on regulatory policies can be associated with a more equal distribution of stakeholder representation, those based on (re)distributive policies can be associated with a more unequal distribution of stakeholder representation.
To conclude, both salience and policy type can be related to the degree of bias in SR in public consultations. The theoretical model presented in this section provides the basis for the following chapters.
3. Research design
The research design that underlies this study is associational. As shown in the previous chapter, deductive reasoning was used to construct the two hypotheses. It will be measured whether there is a relationship between the level of salience and the level of bias in SR, and between policy type and the level of bias in SR. To test for this, a large-N quantitative (positivistic) analysis is most fitting.
Case selection and data collection
The sample of open consultations used in this study is selected from open consultations of the policy area public health of the EC. This research on consultations within the field of public health is representative for all consultations and risk governance within the EU. Due to the Better Regulations agenda of the EU public consultations that are available online play an increasingly important role in EU policymaking. The EU wants to achieve transparent and inclusive policymaking (European Commission, “Better Regulation: why and how”). Open consultations are used so that the EC can incorporate evidence and societal needs into its policymaking process in a transparent and low-cost manner (European Commission, “Have your say”; European Commission, “Better Regulation”). There is a 12-week period during which the public can sent in responses, and eventually the EC includes all responses and a document stating the outcome of the consultation (Quittkat, 2011). Through such consultations citizens and stakeholders can either evaluate existing policies or comment on new policy proposals with regard to what they find most important of the policy, the scope and the added value of the policy (European Commission, “Have your say”). An important aspect of open consultations is that they are used to assess what the impact of existing or future EU legislation is or would be for society. Therefore, open consultations can be directly connected to risk governance. The EC involves citizens and stakeholder because it needs their input on what the potential risks of a certain EU initiative or policy could be and how it could best regulate them. For the policy area of public health, the risks would obviously concern the health of the public. Although the extent to which they are used differs per policy domain (Rasmussen & Carroll, 2014), the aim of the Better Regulation agenda is to make more use of it. The consultations within the domain public health are however representative for all other consultations of the EU because they are all conducted within the framework of Better Regulation.
Furthermore, this study on open consultations within the policy area of public health is also representative to other open consultation conducted outside the EU but to a lesser extent than those conducted by the EU. The EU has several important features in common with other political systems (Hix & Høyland, 2011). Open consultations on policy proposals and initiatives of policymakers conducted within different political systems are comparable to those conducted in the EU after which they are passed on to the parliaments (Chabanet & Trechsel, 2011; OECD, 2006). This demonstrates that also in political systems other than the EU the input of stakeholders is asked so that policymakers have insight into what the potential risks could be. Again, this shows that in all these consultations risk governance plays a role. Nevertheless, Rasmussen & Carroll (2014) do find that there are some structural differences between the EU and these other political systems in terms of how interests are represented and
how stakeholders lobby (p.449). This is why the degree of bias in SR may differ (Rasmussen & Carroll, 2014, p.449). Therefore, in general this research on consultations in the field of public health of the EC can be representative for all open consultations, but to a lesser degree than for those held by the EU.
Additionally, there is also a more practical reason for choosing a sample of online consultations within the policy area Public Health. The EC has a webpage5 where open
consultations can be found. Yet, this webpage did not provide sufficient consultations or not sufficient information on the available consultations within one policy field, or at least not within the field of Public Health. The webpage on Public Health of the EC however, does provides a link to its own consultations, listing many more consultations. For this research it is important that the consultations all fall within one policy field. When the respondents can be grouped within that one policy area, it allows for better comparison between the consultations than when they do not. The total sample contains 42 consultations, which were selected through starting with the most recent closed consultations and then going back in time. The search was carried out at the end of November 2018. The length of the consultations varies but there is no need to control for this because a previous study has shown that this has no effect on the mobilization of different types of stakeholders (Rasmussen & Carroll, 2013, p.454). The aim was to find 25 consultations based on directives and 25 based on regulations. (More details regarding the reason for this are explained below.) However, it proved impossible to find 25 based on regulations within the same time-span as the 25 based on directives: 2005 – 2017. Even going back up to the year 2003 did not result in any suitable consultations. Thus, N = 17 for open consultations based on regulations. The overall time span of all consultations is between 2005 and 2017.
Some more recently closed consultations or several consultations in general are not included. This is because some or all of the following factors were present: a list with respondents was missing, information on whether the consultations are based on either regulations or directives was not included and/or consultations were based on both regulations and directives. It is not possible to give an exact number of how many consultations were not suitable for this research. This is because the webpage showing the consultations for public health of the EC was often unclear. Firstly, the titles of the consultations that appeared in the list of ‘closed’ open consultations did not clearly show whether it indeed concerned an open consultation or something else such as a preliminary opinion or just a call for an open consultation. Secondly, after clicking on these titles (search hits), the layout of the webpage was completely different. Sometimes the webpage did not actually include a list of respondents or include information on one consultation, but rather a lengthy and confusing overview of the whole policy process concerning a certain legislation topic. Such overviews could mention everything, including impact assessments, policy drafts and multiple consultations. The same consultations would then reappear later on in the search list under a different name. This made it impossible to accurately count those consultations that were not suitable for this study.
Dependent and independent variables
The multidimensional phenomenon of bias in SR has been conceptualized in two dimensions: group type and geographical orientation. For the first one, group type, all respondents (except individual citizens) of each consultation have been coded according to seven different categories: citizen groups, professional associations, business associations and trade unions, firms, research institutes, public authorities and other. (This classification corresponds to the classification used in other studies within this research area. See for example: Rasmussen & Gross, 2015; Bunea, 2017). This classification is sufficiently broad to show there are several different types of interest groups that advocate different interests, but not so numerous that the classification of interest types becomes meaningless because it is spread out amongst too many group types. The category ‘other’ is included because there are some stakeholders that do not fit any of these categories, such as organizations that serve as an overarching network for other networks. Regarding the second dimension, stakeholders were coded according to their geographical orientation. The corresponding four categories are: subnational, national, European and international. This provides sufficient distinction in geographical scope, which allows for the necessary differentiation between stakeholders. It is important to note that the subgroup ‘international’ is a more encompassing category. This subgroup can include stakeholders that are orientated towards several countries within the EU (but not the entire EU), stakeholders that are orientated towards within and outside of the EU, and stakeholders that are orientated towards territory outside of the EU.
These two classifications of stakeholder groups result in two sets of dependent variables: stakeholder distribution according to group type and stakeholder distribution according to geographical orientation. A more equal distribution is seen as a lower level of bias in SR, and a more unequal distribution is seen as a higher level of bias in SR. Hereafter the names of these variables will be shortened to ‘group type’ and ‘geographical orientation’.
The first hypothesis relates to the dependent variable salience. Here salience has to do with how informed stakeholders are about a consultation, and subsequently, about the topic it covers. The assumption is that the amount of newspaper articles about the content of a consultation written in the two and a half years preceding the closure of the consultation, is an expression of salience. This resulted in one frequency number per consultation. The database Factiva was used to acquire this information. It contains numerous articles from newspapers all over the world. It updates daily and the starting year varies between the 1990s and 2004. Per consultation it was determined which search terms to put in the search builder. However, a similar set of search terms has been used for each consultation. Since the search terms were in English only newspaper articles written in English were generated. The first part of all sets of search terms consists of the institutes or organizations that were involved, such as the EU, the EC and DG SANTE. The full form of these acronyms was included as well. In this part OR was used between every search term. The second part is about the content the consultations. On the one hand, this had to be done in such a way that it was broad enough to encompass the entire topic of the consultation. On the other hand, it had to be sufficiently specific to ensure that only news articles about this particular topic of this exact consultation would be generated. The use of AND/OR was subject to this. Of course, sometimes a certain term could be
expressed in a different way such as ‘medicinal products’ and ‘medicines’, and again, sometimes acronyms were used that could also be spelled out. However, only using OR would be too broad, but only using AND would be too restrictive (in most cases). This had to be determined per consultation. Out of all the searches in Factiva, 39 were carried out on 30/12/2018, one on 01/12/2018, and two on 06/01/2019.6 An overview of the set of search
terms used for each consultation can be found in Appendix B: Factiva Search Summaries. The second hypothesis relates to the independent variable policy type. Policy type has been divided into regulatory versus (re)distributive. These are operationalized as consultations based on regulations and consultations based on directives. Regulations are regulatory because Member States of the EU have to directly implement these. In contrast to this, directives are (re)distributive because Member States have the freedom to implement these in whichever way they see fit. Stakeholders can try to exert their influence in this process. In short, the legislation is (re)distributed to the Member States. This is also the reason why approximately half of the selected open consultations are based on regulations and the other half on directives. The information on which policy type a consultation was related to could be found in either the title of the consultation, the information on the webpage of the consultation and/or in a document listed on the webpage of the consultation. For this variable it was important of course that the selected cases were based on either policy type but not both. Nevertheless, as mentioned in the beginning of this chapter, the two groups are not equal: N = 17 for consultations concerning regulations and N = 25 for consultations related to directives. This means that the outcome for the group ‘regulations’ is less strong than for ‘directives’. This inequality is due to limited available information.
Statistical tests used for data analysis
First of all, to be able to carry out any statistical tests using the dependent variables, they need to be calculated first. The construct ‘distribution of stakeholder groups’ has been operationalized in two dependent variables, namely a variable based on group type, and one based on geographical distribution. To construct these variables, the normalized Herfindahl-Hirschman Index (HHI) is used rather than the standard HHI. In general, the HHI is often used to measure how competitive a certain market is. This test is also applicable here because the bias in SR can be expressed by measuring the distribution between the categories for both stakeholder classifications, group type and geographical orientation. In this research it is purely about the number of respondents of a certain category compared to the other categories. The HHI can show how (un)equal the representation within one group (or in this case: consultation) is. However, the standard HHI is based on the total number of the group. This would be impractical since there is a lot of variance in the total number of respondents per consultation, and therefore the standard HHI cannot be used to compare the HHI scores between all 42 consultations. This is why a normalized HHI is used.
6 The reason that three of the 42 searches in total were carried out on a later date was due to the misspelling
