'Trust me, I'm a think tank': A survey experiment on the trustworthiness of think tank advice in the Netherlands

(1)

A thesis:

„Trust me, I‟m a think tank‟: A survey experiment on the trustworthiness of think tank advice in the Netherlands

By Mechteld Visser

s1537784

Leiden University

Thesis supervisor: Dr. Valérie Pattyn

(2)

One year of researching, discussing, reading, writing, and learning has culminated in the document that presented to you. This thesis, produced as the final piece of the Master Public Administration: International & European Governance, has only been one part of the educational process that keeps transforming the way I look at the world and the mechanisms that keep it in place. Even though I have not yet reached the end of this process, I would like to express my gratitude to some of the wonderful people who have brought me this far. My special thanks go out to my thesis supervisor, Dr. Valérie Pattyn, who always believed in this project, even at times when I did not. Her feedback and encouragement were indispensible in completing my thesis. I would like to thank my dear friend Stephanie Bijl for enlightening me in the ways of quantitative research, which was a new and challenging experience for me. My gratitude goes out to all my fellow students, especially those who took the time and effort to take part in the survey experiment that is at the heart of this thesis. I would like to extend my thanks to the teachers at Leiden University in all their efforts to inspire and educate students like me. As always, I would like to thank my parents, who gave me the opportunity to pursue an education. Lastly, I wish to remember my beloved grandfather, who most eagerly anticipated seeing me graduate, yet never would.

Voorburg, August 2018

(4)

Abstract

Due to the externalization of policy advice in recent decades, the study of trust in non-governmental policy actors has become increasingly relevant, yet think tanks remain largely unaddressed in the literature. Traditionally, think tanks have mostly been approached as political organizations, equated to other policy actors such as lobbyists and NGO‟s. Only scholars adhering to the institutionalist perspective consider think tanks as a category of organizations with an internal range from scientific institutions to political advocacy groups. A survey experiment among 121 Dutch students proves the significant differences between academic and non-academic think tanks in terms of perceived trustworthiness. Respondents indicated that policy advice supplied by non-academic think tanks was significantly less trustworthy than the publications by both academic think tanks and government ministries and it was not desirable for these non-academic think tanks to be involved in the policy process. Since trustworthiness was shown to be higher when institutions were perceived as possessing a higher level of autonomy and scientific expertise, think tanks and other policy actors are advised to strengthen these academic credentials if they wish to increase and/or maintain their trustworthiness as a provider of policy advice.

(5)

1. Introduction

In the early weeks of 2015, Dutch research institute and think tank Clingendael proudly announced its position in McGann‟s yearly Global Go To Think Tank Index Report. The institute ranked 34th on the list of the best think tanks worldwide, while taking eleventh place among its peers in Western Europe. Clingendael defines itself as an „influential and autonomous think tank in the field of international relations‟, facilitating the exchange of research and knowledge through their own research, consultancy, and trainings (Clingendael, 2015). As Clingendael was enjoying their status as the best think tank in the Netherlands, the Dutch think tank community welcomed a new member the very same year. In reaction to a perceived swing to the left in Dutch society, a group of young conservative intellectuals started to arrange regular seminars on political issues, mostly relating to the EU. The resulting think tank took on the name „Forum voor Democratie‟, and presented itself as a platform for a new intellectual outlook on these issues. This think tank, however, employed a very different model of „intellectual exchange‟: the think tank required participants to pay a fee to attend a discussion, which would then always champion the participants‟ political preferences. In summary, the sponsors decided which position the speakers would take in the debates. Their claims did not require scientific evidence (Aalberts, 2015). Two years after its establishment, the think tank transformed into a political party and managed to win two seats in the Dutch national elections. And Clingendael? It managed to rank 33th on the best think tank list, continuing their research as usual.

This brief illustration of two very different think tanks in the Netherlands demonstrates the fluidity of the think tank concept. While some established institutes strive for the exchange and dissemination of policy relevant knowledge, others consist of groups of advocates working towards the advancement of political opinions and ideologies by igniting debates. Both under the common denominator of think tank, these types of organizations have become relevant players in policymaking, despite their varying level of scientific expertise and autonomy (Abelson, 2009).

From a scholarly perspective, there has already been an abundance of attention for the role of expertise in policymaking (Hoppe, 2005; Weible, 2008; Griffin, 2009; Carmel, 2016; Brostrom & McKelvey, 2018). In this debate, there is little to no mention of think tanks. While think tanks have been studied over the past decades, the traditional perspectives merely take think tanks as

(6)

political interest groups, often equating them with other interest groups such as lobbyists or NGO‟s (see for example Cater, 1965; Freeman, 1965). These perspectives overlook the scientific credentials of think tanks such as the Clingendael Institute, which produce scientific research in their own right. Considering the think tank involvement in watershed policy decisions in the modern age (Medvetz, 2012a, p. 5) and the growing number of think tanks in the last few decades (Kelstrup, 2016, p. 2), it would be fruitful to study think tanks as a wider category, in which organizations vary between science and advocacy.

The study of think tanks as both scientific and advocacy institutes is not only interesting from a scholarly perspective, but also crucial for everyday bureaucrats and politicians seeking to make the best policy decisions possible. As a government official in a democratic state, one strives to make policy decisions approved by the largest proportion of constituents possible. When confronted with policy advice from various sources, it remains unclear to what extent this advice is to be trusted. Especially in the case of think tanks, trustworthiness may be difficult to determine due to the range of organizations under the umbrella of this concept. To this end, it is useful to ascertain to what extent policy advice provided by think tanks can be perceived as trustworthy in the Dutch context. Is policy advice provided by think tanks perceived as more or less trustworthy in comparison to the advice of government ministries? Does the perceived trustworthiness of think tank advice depend to some extent on the level of scientific expertise through research and the level of autonomy? In this thesis, a survey experiment among Dutch students at Leiden University will provide answers to these questions. The student population is appropriate for this research due to the lack of exposure to a work environment and the networked political context in which they grew up. In such networked environments, think tanks are not novel or uncommon political actors, which makes students more receptive to the idea of think tank involvement in the policymaking process.

In order to establish whether there is a difference in trustworthiness between types of think tanks based on their footing in either the scientific or the political field, the thesis is divided in five sections. In the second chapter, I will outline how the literature on trust in political institutions has been almost exclusively focused on governments, with recent excursions into non-traditional policy actors such as NGO‟s. Such recent studies have become increasingly relevant due to the externalization of policy advice. Whereas the state used to act monopolistically in the policy

(7)

process, it now relies on external actors to provide input and advice. Despite this development – which has also occurred in the Netherlands – there has not yet been a study on citizen‟s trust in think tanks.

In the section 2.2, I will analyze the concept of a think tank along the lines of Kelstrup‟s definition of a think tank as „an organization which claims autonomy and attempts to influence

public policy by mobilizing research’ (2016, p. 10). I break down this definition into four parts.

Firstly, by referring to think tanks as organizations, they differ from networks such as epistemic communities. It also allows us to better understand the position of think tanks as one actor in the broader contexts of knowledge regimes. Secondly, autonomy in think tanks may vary both in terms of funding and ideological allegiance. Third, the level of influence also varies between different think tanks, which has lead scholars to attempt to establish the determinants for think tanks success. Fourth, the extent to which think tanks mobilize research determines their position on the science-politics continuum, which is assessed differently by many scholars throughout the years.

The debate on the mobilization of research is expanded upon in section 2.3, where I provide an overview of the four main theoretical perspectives on think tanks that have dominated the literature over the years. Scholars from elite theory, pluralist and statist schools have all underrepresented or overlooked the differentiation among think tanks as either providers of scientific expertise or advocacy institutions. The hypotheses resulting from these perspectives therefore all predict a non-significant difference in perceived trustworthiness between the different types of think tanks. The institutionalist approach to think tanks is the only main paradigm elaborating on the hybrid conceptualization of think tanks. According to this approach, think tanks function as bridges between the scientific and political community. The insights from this perspective result in a different hypothesis, predicting a significant difference in perceived trustworthiness. All the hypotheses can be found in chapter 3.

While the fourth chapter elaborates on the research design employed in in this thesis, the fifth chapter presents the results and analysis of the survey experiment, which confirm the significant difference in perceived trustworthiness between the three think tank categories: academic think tanks („universities without students‟), semi-academic think tanks (government contracted research institutions), and non-academic think tanks (advocacy organizations). Not only do these

(8)

think tanks differ in trustworthiness, the analysis also proves that scientific expertise and autonomy are significant predictors of the level of trustworthiness received.

In the sixth chapter, the implications and limitations of the survey experiment are further discussed, concluded by possible future developments for the relatively young field of think tank research in the Netherlands. The research results have implications for each perspective on think tanks, especially for the institutionalist perspective, since it confirms earlier institutionalist works arguing that think tanks become less trustworthy when they move further away from the scientific sphere and move closer to the political sphere. This has practical implications for bureaucrats and policymakers, as it encourages them to evaluate policy advice on its scientific credentials in order for the policy process to remain trustworthy.

(9)

2. Theoretical Framework

2.1. Trust & Policymakers

Little scholarly work deals with citizen‟s trust in think tanks specifically. Rather, the trust literature tends to focus on other political actors, especially governments (see i.a. Nye et al., 1997; Veal & Clark, 2011). „Trust‟ implies a relationship in which one person believes that another will act consistently with expectations of good behavior (OECD, 2017, p. 42). This „other‟ could be a person, another regular citizen. In such cases, we speak of „interpersonal trust‟, which constitutes an integral part of any functioning society (Putnam, 2001). However, this type of trust, while constituting an important topic for further research, is not relevant when studying think tanks. Instead, „institutional trust‟ is better fit to suit this purpose.

Institutional trust is the umbrella term for the level of trust citizens have in political institutions, law and order institutions or non-governmental institutions. There are two ways to conceptualize the trustworthiness of these institutions. On the one hand, one can measure trust according to the „logic of consequences‟, which entails that the level of trust is causally dependent on the outcomes that an institution is able to produce. On the other hand, trust can also be measured as the extent to which an institution conducts its business in accordance with common standards of integrity, fairness, and transparency. This is referred to as the „logic of appropriateness‟ (OECD, 2017, Bouckaert, 2012).

The bulk of institutional trust literature elaborates on the trustworthiness of political institutions, specifically the government. (Carnevale, 1995; Powell, 1996; Nye et al., 1997; Citrin & Muste, 1999; Warren, 1999; Chanley, 2000; Maxwell, 2010; Grimmelikhuijsen, 2013; Nicholls & Picou, 2013). In all democracies, policymakers directly or indirectly serve the wider public. It is therefore important that they, as representatives of the people, enjoy a basic level of goodwill. In other words, “democracy may not be viable without at least some trust in its representatives and institutions” (Lahno, 2011, p. 676). Consequently, establishing a level of trust is a main priority among democratically elected governments. In the scholarly context, the idea that the trustworthiness of governments used to have important implications for the trustworthiness of produced policies explains the attention for the topic. Bureaucratic administrations, headed by

(10)

the government officials, were seen as the dominant force in the policymaking process, thus determining the eventual direction of policy development.

However, in recent years scholars have expanded their research to include non-governmental actors, arguing that many democratic governments allow for the interference of other actors in the policy process. Increased participatory options and use of external expertise by government actors have led to the notion that policy advice is more „externalized‟ in the contemporary political landscape (Craft & Howlett, 2013, p. 189).

According to a study by Van den Berg (2017), the policy process in the Netherlands has been externalized as well, due to three major historical pressures on the policy advice system. First of all, the process of depillarization brought down the stark divisions between socio-economic groups that had characterized Dutch society for decades. In pre-war Dutch society citizens relied on the organizations within their own „pillar‟ for most of their needs. After the Second World War, however, more and more service delivery and finance responsibilities shifted to the government. To facilitate the new tasks, a system of advisory bodies and councils was put in place to provide policy input, leading to partial externalization of policy advice. In the 1970s and 1980s, New Public Management theories gained foothold in the Dutch administration, spreading the ideas of decreased government control and privatization of government services. Consequently, private consultants flourished and external contractors partially replaced public servants, in line with the new managerial style. In the late 1990s, the development of mass media, competition in the political arena, expanded auditing, and political polarization added pressure on the government to perform in all sectors. Higher expectations would lead government officials to draw their information and advice from sources that are easy to reach and control, such as ad-hoc committees and affiliated organization who will not contest the official‟s decisions (Van den Berg, 2017). All these developments have provided ample opportunity for non-traditional policy actors to enter the realm of policy advice.

Externalization of policy advice to non-traditional policy actors such as think tanks – but also NGOs, lobbyists, and possibly less formalized sources – increases the importance of a closer examination of trust in think tank advice. Decades ago, policy advice produced by think tanks traditionally was marginal in comparison to that of the government. Traditional machineries of governance tended to disregard think tanks as somewhat of a nuisance because they would

(11)

question their preferred policy choices (Dror, 1984, p. 207). In the current climate of externalization, however, “think tanks are unquestionably a major purveyor of policy research” (Rich, 2005, p. 25). This development shows that it is now much more relevant to assess the level of general trust in the policy advice produced by think tanks. The entire policy process needs to be trustworthy, also when this includes external actors. If citizens do not trust the policy advice supplied by think tanks, they may disapprove of its increased inclusion in policymaking. Not only does this decrease the trustworthiness of the policy process as a whole, it may also have implications for future involvement of think tanks advice in the policy process.

The increased importance of non-traditional policy actors due to the externalization of policy advice is reflected in the recent studies and debates exploring trust in these organizations. NGOs, for example, are frequent subjects of scientific studies in the context of trustworthiness. In a 2010 study, Lee et al. analyzed the relation between media independence and trust in NGOs in post-communist countries. Terwel et al. (2009) empirically demonstrated how Dutch citizens have more trust in environmental NGOs in comparison to industrial organizations in the domain of climate change. Furthermore, the Edelman Public Relations yearly Trust Barometer, includes the category „NGO‟ next to businesses, media, and government. In this survey, NGOs score continuously higher on trust when compared to the other categories (Ries et al., 2018). The topic does not lack scholarly debate either, as demonstrated by a study by Keating and Thrandardottir (2016) which disputes these high levels of trust. Instead, the authors elaborate on the alleged „crisis of trustworthiness‟ among NGOs.

Despite the externalization of policy advice and the presence of a scholarly debate on other policy actors, think tanks have yet to be studied from this perspective. One study that approximates the topic is a survey experiment executed among policy analysts in British Columbia. The experiment showed that scientific research is perceived as more credible than research produced by think tanks or advocacy organizations (Doberstein, 2017). However, it is not evident how credibility would relate to trust and if these results are applicable to the general population, or to other national context. Moreover, in the article, think tanks are distinguished both from scientific research and advocacy organizations, while this distinction often does not hold in reality, as we will see in the next section. Overall, it is time for an exploratory study into

(12)

trust in think tanks, preferably in a non-Anglophone context. The Dutch context offers an excellent opportunity, given that it has also been subject to externalization of policy advice.

2.2. Think Tanks: Definitions & Debates

Even though most studies of think tanks emerged in the last decades, the concept has been around for a while longer. The notion of a „think tank‟ dates back to the Second World War, when small rooms used for strategy planning received the term „think tank‟ as a nickname (Kelstrup, 2016, p.5). In the 1960s, think tanks referred to private non-profit research organizations. Most scholars in these early years considered think tanks as American phenomena, where they could thrive due to the United States‟ unique political system in combination with the common practice of private funding for research organizations. Think tanks first spread to other Anglophone countries, but have established themselves in most modern political systems today (Ladi, 2011). Today, still 27% of all think tanks in the world hail from the United States (McGann, 2017). Unsurprisingly, the bulk of literature on think tanks is focused on the U.S. In the Netherlands, most of the existing think tanks are related to the Dutch government due to the difficulties in securing non-governmental funding (Van Beek, n.d.), with the exception of some international think tanks situated in The Hague. Currently, the country ranks 14th in the list of countries with the largest amount of think tanks, tied with Switzerland at 76 think tanks (McGann, 2017).

The exact definition of a „think tank‟ remains a topic of contention, reflecting the main divides still present in think tanks literature today. Out of many definitions, this thesis employs the definition put forward by Kelstrup (2016), because it adequately integrates the debates in the literature. Using this definition as a basis, I will elaborate on the most important discussions on think tanks and what they are supposed to represent. In his book, Kelstrup defines a think tank as „an organization which claims autonomy and attempts to influence public policy by mobilizing

research‟ (2016, p. 10).

The first characteristic captured in Kelstrup‟s definition is that of a think tank‟s organizational qualities. This presupposes a more concrete structure in comparison to other related concepts. It

(13)

distinguishes the think tank phenomenon from that of epistemic communities and knowledge regimes. Epistemic communities are understood as networks of experts that exercise scientific authority through the policy-relevant knowledge they possess over a certain domain (Haas, 1992, p. 3). As for their relation to epistemic communities, think tanks are an organizational expression of these loose networks of experts (Stone, 1996). Moreover, think tanks distinguish themselves from epistemic communities by their organizational qualities and less strict demands on scientific expertise of their employees. Campbell and Pedersen (2011) conceptualized knowledge regimes as the accumulation of actors and organizations producing and disseminating policy-related ideas which influence how policy-making systems are operated and organized. These are not clearly defined (groups of) organizations, rather, they form the context in which more organized networks of actors operate. Think tank-affiliated actors need not necessarily be engaged in a coherent profession or be equipped with specific credentials or forms of expertise (Medvetz, 2012a, p. 15). While taking on their multiple functions, think tanks may also constitute parts of existing knowledge regimes and legitimize these as well. In their work, Campbell and Pedersen criticize the think tank literature for being overly descriptive and ignoring their role in maintaining knowledge regimes (2011, p. 170). However, since their publication, various scholars have started researching the interaction of think tanks with the knowledge regimes in which they operate (Nachiappan, 2013; López Aymes, 2014; Hustedt & Veit, 2017).

The second characteristic captured in Kelstrup‟s definition – autonomy – hits at the core of the think tank literature. Moreover, it is one of the defining characteristics employed in think tank typologies. His formulation already betrays the subjectivity involved in labeling think tanks „autonomous‟. It is challenging to determine when an organization is genuinely autonomous. Securing autonomy is important to think tanks because they do not have members and do not represent the ideas of a specific group of constituents. Therefore, they lack democratic accountability and can only claim authority based on their high quality research. Since researcher autonomy is the cornerstone of modern academics, protecting their autonomy should be a think tank‟s first priority (Fraussen & Halpin, 2017). Any academic think thank should therefore possess a certain degree of autonomy in addition to scientific expertise.

(14)

Nevertheless, the degree of autonomy varies between think tanks. Most typologies of think tanks take the extent to which they produce autonomous and scientific research as a main defining quality. While there are many different typologies in circulation, most draw from Kent Weaver‟s 1989 classic article. In this article, Kent Weaver divides think tanks into policy research institutions, government contractors or specialists, and advocacy think tanks (Kent Weaver, 1989; Abelson, 2011, p. 1646). Whereas the autonomy of the first category is usually incontrovertible, the autonomy of the latter two may be dubious depending on their score in two measurement categories: funding and ideology (Kelstrup, 2016, p. 11).

The first way to measure a think tank‟s autonomy is to assess its material funding structures. The source of a think tank‟s funding significantly determines a think tank‟s discretion. If a think tank receives its funding from a select number of sources it will be difficult to safeguard organizational independence. Only last year, a U.S. based think tank called the „New America Foundation‟ saw its reputation become severely damaged because of autonomy concerns. Since 1999, this think tank developed public policies for the digital age. Allegedly, a research project into monopolies was closed down because its head researcher had criticized the monopoly position of Google in the digital market. Critics condemned the foundation for this decision after it was revealed that Google, as well as its former CEO, are major sponsors to the think tank (Overby, 2017). Reduced funding for independent institutions in the 1970s in the U.S. seems to have stimulated corporate and partisan influences, as think tanks were forced to rely more and more on private sponsors (Rich, 2005). Since corporate and individual sponsors are often interested in a particular goal or policy area, think tanks have developed accordingly. Specialized think tanks now operate in the political arena, actively advocating in their niche. This change is reflected in the most recent scholarship on think tanks, which has specialized as well. New research emerged, examining the role of think tanks in specific policy areas such as environment policy (Dunlap & Jacques 2013), education policy (McDonald, 2013), and health policy (Shaw et al., 2014).

The second way to measure autonomy is to assess allegiance in the realm of ideas. Similar to universities, think tanks must “claim intellectual independence to retain credibility” (Kelstrup, 2016, p. 11). Nevertheless, many think tanks (especially those with a clear advocacy role) are grounded in specific ideologies. When scholars conduct research with the aim of legitimizing a

(15)

certain belief, however, they are vulnerable to teleological reasoning. Their ideology may tempt them to steer the results in the direction of a desired outcome. The predicate of „think tank‟ may even become a legitimizing mechanism for predetermined ideologies. This development has also received attention in the think tank literature, as Smith and Marden (2008) unmasked how the conservative movement in Australia kept up an expertly crafted façade of intellectual autonomy while secretly using their expert image to spread political values in the media. Evidently, a think tank‟s ideological focus may result from the preferences of (private) sponsors, so these measurements are interlinked.

The third part of Kelstrup‟s definition pertains to the attempt to influence public policy. This characteristic forms a field of interest for many scholars. They attempt to establish whether think tanks also succeed in exerting the influence they desire and what characteristics would lead to a successful outcome. A prominent analysis of think tanks in the United States and Canada by Abelson (2009) demonstrates that think tanks can exercise effect on the direction of political debate, as well as alter policy preferences and choices of decision-makers, albeit in different stages of policy formulation and in different ways (p. 4). However, it remains challenging to accurately measure „influence‟ in every context. The extent of think tank influence may depend on its academic pedigree, size, policy domain, level of specialization, and this list is far from exhaustive. (Weidenbaum, 2010). Nevertheless, the measurement problem has not disheartened scholars from investigating the topic (see o.a. Lindquist, 1993; Parmar, 2004; Xufeng, 2009; McNutt & Marchildon, 2009). Despite the numerous studies into the influence of think tanks on policy formulation, the debate lacks attention for the normative implications of such involvement. It is yet to be determined to what extent it is desirable that think tanks influence the policy process.

The last part of Kelstrup‟s definition relates to the core activity, reoccurring in every think tank. „Mobilizing research‟ refers to the collection and accumulation of knowledge from which think tanks may derive truth claims about the policies they wish to advance. However, the wording leaves room for interpretation: it is undisclosed whether a think tank produces knowledge itself by scientific research, relies on pre-existing or contracted scholarship, or uses scientific knowledge at all to navigate the political debate. Over the years, scholars have debated to what extent think tanks relate to both the scientific community and/or the political community. In the

(16)

next section, I will outline the perspectives on the scientific and political role of think tanks and to what extent scholars have deemed their influence in politics desirable.

2.3. Perspectives on the Position of Think Tanks: Between Science & Politics

The first group of scholars exploring think tanks and their relation to science and politics derives its arguments from elite theory. Scholars supporting elite theory see think tanks as the intellectual ammunition for groups of powerful corporate and political elites. Taking on an instrumentalist view, think tanks supposedly exist solely by the grace of promoting the agenda of the ruling class (Medvetz, 2012a, p. 8). The elite theory perspective rose in the 1960s and 1970s in reaction to Great Society politics and the Vietnam War. Managerial and policy experts were accused of taking over political decisions and experts were seen as the intellectual engineers of the Democratic Party‟s liberal policies. Marxist and other radical readings of politics formed a great source of inspiration for elite theorists. One of the first to address think tanks in this perspective was William Domhoff, who argued in his 1967 book „Who Rules America‟ that think tanks have important and direct influences on the policy process through their large variety of strategies, facilitated by the power elite. This Marxist inspiration is also clearly captured in Gouldner‟s 1970 work „The coming crisis of Western sociology‟, where he asserts that “The liberal technologue … produces information and theories that serve to bind the poor and the working classes to the state apparatus and the political machinery of the Democratic party” (Gouldner, 1970, p. 500 as cited in Fischer, 1991, p. 2). Other radical theorist works belonging to this stream are Chomsky (1971), Steinfels (1979), and Gross (1980). The same line of thought reoccurred in neoconservative rhetoric in later years. According to these neoconservatives, policy experts are a threat to representative democracy (Fischer, 1991). More recently, Smith (1991) tackled this subject by thematically describing how expertise and politics have been intertwined through history. Tracing this relation through American history, he argues that think tanks are the most prolific organizational form used by experts to engage in politics. Moreover, the experts employed in think tanks are part of the „new policy elite‟, who claim ownership of the policy process, thereby effectively excluding other groups from participation in policy

(17)

formulation. Even though Smith recognizes that expertise is indispensible in devising sound policy, he warns for such exclusion as it may not produce “wise democratic deliberations” (p. 239). Dye (2001) has further brought elite theory into 21st century consciousness, as he argues that policy is constructed „top-down‟ by elite actors in exclusive negotiations behind closed doors. This does not only include think tanks, but also other interest groups and lobbyists. All in all, adherents to the elite theory place think tanks predominantly on the political spectrum, taking a skeptical stance towards the scientific quality of their research – if there is any to speak of.

The elite perspective has garnered quite some criticism for its mechanical and functionalist reading of the think tank phenomenon. While many scholars focused on the relation between elite politicians and think tank personnel, it remained unclear how this would translate into substantial policy influence. Moreover, the theory fails to account for cases of successful think tanks that do not have clear ties to the ruling elite or even advocate against the status quo (Medvetz, 2012, p. 9; Abelson, 2006, p. 22). Conversely, it would be unable to explain how other think tanks supported by elitist groupings can fail in promoting their policy agenda.

The pluralist school addresses some of the issues involved with the elite theory perspective. Scholars in this tradition regard think tanks as one of the many groups of organization involved in the competition over public policy. In this view, policymaking is a constant interplay of organizations with their own supporters, resources, goals and strategies (Medvetz, 2012a, p. 8). This perspective evolved in tandem with the elite theory perspective in the 1960s and 1970s as scholars elaborated on the increasing influence of interest groups and other societal actors on the policy process. For example, Cater (1965) analyzed this influence through the lens of „subgovernments‟, which he defines as dense networks of actors with similar preferences and the possibility to change policy (Cater, 1965, 17). Freeman (1965) conducted a similar research, establishing the relation in and between subsystems, which he defined as the network of actors involved in policymaking within a specific domain. He argued that policy influence is dispersed and contained within these subsystems through the arrangement between interest groups (such as think tanks), executive bureaus and congressional committees. (Freedman, 2006, p. 2). Newsom (1996) later applied the pluralist perspective to foreign policy, arguing that the policy formulation in this domain is “at the same time confidential and public” (p. ix), through the involvement of think thank and other actors. Newsom distinguishes think tanks from the

(18)

scientific world in his analysis, which illustrates the lack of attention for the scientific character of think tanks in the pluralist perspective.

Pluralist scholars thus argue that think tanks are a part of the political process. The multitude of actors battling over policy influence justifies any influence think tanks may have on policymaking. Even though pluralist scholars diverge on the range of actors, they all assert that there is not one group dominating the entire process (Freedman, 2006). Nevertheless, they do admit that some groups may have greater access to power through larger resources or other advantages, which results in a larger impact on policy outcomes (Liao, 2006, p. 8). The pluralist model is further subdivided in two sub-models. The arbiter model places the government as a moderator in the conflict over policy, making sure that all groups are equally represented. The government is expected to maintain a level playing field by determining the rules of the game (Abelson, 2006, p. 16). The arena model places politicians and other participants on equal footing, both involved in negotiations and bargaining over policy proposals. In this sub-model, government control is limited, as they depend on certain groups to implement the eventual policy (Liao, 2006, p. 9). In this perspective, neither the scientific credentials nor the trustworthiness of think tanks is specifically addressed. Since think tanks are just one of many interest groups, this is deemed irrelevant.

Even though the pluralist school is better equipped in explaining why think tanks have taken on a more important role in policymaking, it still suffers some shortcomings. For example, it is not explicit which think tank characteristics determine the scope of its policy influence (Abelson, 2006, p. 23). This probably explains the volume of studies on think tank influence mentioned in the previous section. Still, if think tanks are only one part of a large fight over policy, why are some think tanks significantly more successful in reaching their goal in comparison to other interest groups? Another shortcoming relating to the arbiter sub-model is the presupposed neutrality of government officials. It is highly likely that politicians favor a certain policy, which puts think tanks advocating for that specific option at an advantage.

A third perspective plays down the influence of think tanks, arguing that it is the state in the end who decides over public policy. Despite the influence of other actors in the policy process, the statist paradigm reasons that the state is the ultimate actor determining and implementing policy. This reduces think tanks to potential advisory actors. In his book „Defending the national

(19)

interest‟, Stephen Krasner reacts to the elite theory and pluralist stream by asserting that the state is an independent actor following the national interest and that state behavior can therefore not be explained by the interests of groups or classes (1978). His arguments were reiterated by Skocpol (1985), who argued that research on public policy had previously underestimated the role of the state and the power of government.

Arguably, a higher scientific standard could increase a think tank‟s chances of being noticed by a controlling government in a larger pool of policy advisors. In this case, the scientific standard of think tanks would indeed matter. However, this paradigm does not explicitly differentiate between types of think tanks. Moreover, it is not a given that every state selects advice based on the intellectual standard or autonomy of those providing it. Government officials may still choose to follow up on advice that fits their preferred interests already.

On the one hand, this paradigm helps to explain why think tanks are not always successful, even if they possess powerful resources. On the other hand, it is still unable to fully explain cases in which a think tank is successful without them having a visible foothold within the state. It is also unable to account for the cases in which statist actors have consulted and negotiated with interest groups, given that they do not strictly need to consult other parties (Abelson, 2009).

The final perspective diverges from the previous three, as it approaches think tanks as a subject in their own right. Instead of remaining part of larger groups of interest groups, the institutionalist perspective follows an approach in which think tanks are studied in their own structural environments, with the rules and norms that shape their behavior and the developments to which they must respond (Medvetz, 2012a, p. 12). Some scholars along this line have studied think tanks at a micro-level, focusing on specific institutions and their position on the science-politics nexus within their given context. Others have studied think tanks as part of the wider group of epistemic communities (Abelson, 2009).

The most important distinction of the institutionalist perspective with regard to internal differences in the think tank category is that it links the scientific character of think tanks with their political role, for its scholars assert that think tanks have their roots in the scientific community, and that think tanks are formed by scholars who wish to enter the political realm. Stone (1996) illustrates this „bridge building‟ function by developing a notion of think tanks as

(20)

media for intellectual activists to transform knowledge in to policy action. In her book, she defines as organizations that move ideas into politics while remaining outside of formal political arenas (Stone, 1996). Medvetz (2012b) more recently utilized think tanks in his discussion on „boundary organizations‟: “organizations specialized in mediating relationships among more established fields” (Medvetz, 2012b, p. 114). A „field‟ is a concept originally devised by Pierre Bourdieu describing the space of relations in which participants attempt to monopolize a particular socially valued resource. In other words, it is the sector in which people attempt to establish themselves as the highest authority, such as lead scholar in biochemistry or best harp player in classical music. Medvetz raises his idea of boundary organizations as he argues that think tanks cannot fully be understood as members of a single field, overlapping multiple fields, or constituting a field of their own (Medvetz 2012b). Also in the Dutch context, think tanks have been studied as scientific venues that play an important role in the policy process. Timmermans and Scholten (2006) argue that think tanks may constitute an important formal institution enabling interaction between scientific experts and policymakers, which contributes to the policy process by providing clear problem definitions and bringing the potential to depoliticize controversial issues. However, it may also provide „knowledge ammunition‟ to politicians, thereby fueling policy disputes. (Timmermans & Scholten, 2006). Lastly, one of the most important studies of this „bridging position‟ of think tanks is executed by Rich (2005), who demonstrates how think tanks can have a dual role and relate both to the scientific and the political realm and how this relation changed over time. He argues that think tanks are understood on the one hand as credible providers of expertise, and on the other hand as players in the ongoing debate over ideas. Moreover, his analysis of the American think tank landscape reveals that the latter understanding has become more pervasive over the years. Whereas most of the early think tanks adhered to the strict expert function by scientifically assessing policy options, most of the think tanks active today have clear ideological motivation.

With regards to the scientific nature of think thanks, Rich‟s 2005 study has also found that think tanks with less identifiable ideological motivation and those who refrain from marketing strategies enjoy higher credibility under government officials. Three quarters of the respondents in the research regarded advocacy groups as less credible than think tanks with a stronger scientific base (Rich 2005; Doberstein 2017). The earlier mentioned study among Canadian policy analysts showed similar results, confirming Rich‟s findings (Doberstein, 2017). From

(21)

these analyses, it seems that the more think tanks abandon their „bridge building‟ function between politics and science, the less they are able to hold up a credible image. However, it is not yet clear if this expectation holds for regular citizens and to what extent credibility relates to trustworthiness.

There are two gaps in the current research. First of all, it is too American-centered. Almost all (empirical) analyses originate from the United States, with some rare excursions to the wider Anglophone world. This has the undesired consequence of embedding a U.S. bias into the discussion (Kelstrup, 2016, p. 4). This is problematic, for the European and other non-American conceptions of think tanks differ significantly from the American one. They perform different tasks and relate differently to the various political systems. For example, Germany is home to many influential think tanks that largely tie to either universities or political parties. In France, think tanks flock around the government in Paris and have an adversary relationship with French political parties, rivaling political decisions with their research. In other states, think tanks only started to emerge after the establishment of democratic governments. (Ladi, 2011). Only in recent years has the study of think tanks departed from the Anglo-American understanding.

Second, the current body of work mostly addresses diverging theories on the position of think tanks in the political sphere. Moreover, almost all perspectives treat think tanks as political organizations, putting them in the same category as NGOs or lobbyists, thereby trivializing their scientific credentials. This is a big shortcoming, as many think tanks have a scientific focus and take the production of academically sound research as their main mission. Only scholars in the institutionalist perspective properly address the scientific character of many think tanks. Whether think tanks act more as political organizations or scientific institutions may have great implications on how their policy products are perceived and whether this makes a think tank‟s policy products more trustworthy or not. Also, these perspectives assess whether it is desirable for think tanks to influence policy based on the configuration of the political context and power distribution amongst policy players. It lacks an assessment of whether these theories on the desirability of think tank influence are shared under the wider public. Again, citizens might not feel the same kind of approval for the involvement of academic think tanks in comparison to non-academic ones. It is these gaps that this research aims to address.

(22)

3. Conceptual Framework

3.1. Concepts

The main research question is whether there is a difference in trustworthiness between the policy advice produced by academic think tanks in comparison to semi-academic and non-academic think tanks as perceived by the Dutch population, in this case measured by a student sample. Trustworthiness, as the dependent variable, is measured according to the logic of consequences, reflecting the extent to which a person believes that the policy advice generated by an institution, whether it be think tank or government, is consistent with outcomes of expected positive

behavior (OECD, 2017, p. 42). Since it is easier for the respondents to assess the products of an institution in comparison to the intents, it is appropriate to measure the level of trust in the policy advice produced by think tanks instead of measuring the trust in think tanks themselves.

In order to establish the hypothesized difference, the research includes the trust in policy advice produced by the government ministry as a control category. A government ministry is a

specialized government department presided by a minister. Government policy advice is the most relevant as a control variable, since it was for a long time considered to be the most important advice in the policymaking process. Only after the externalization of policy advice to other sources think tanks started to gain prominence in the academic literature. „Think tanks‟ are defined according to Kelstrup‟s conceptualization, explaining the phenomenon as “an

organization which claims autonomy and attempts to influence policy by mobilizing research” (2016, p. 10).

Dutch students form the target population for this research. The student population is especially relevant for three reasons. First, a significant portion of these students will find employment in government (Nordlohne et al, 2017). Those students are the policymakers of tomorrow, so it is useful to have an insight into how they assess think tanks and their trustworthiness. Second, students have not yet been subject to immersion in a working environment and will still be able to approach the concept of think tanks in a critical and relatively unbiased way. Third, these students are likely to be born in the 1990s, during the last phase of externalization of policy advice (Van den Berg, 2017). This means that they grew up in a political climate in which it is

(23)

not unusual to consult non-governmental sources when in need of policy advice. The

incorporation of think tank advice in the policy process is not an uncommon encroachment on an exclusively governmental affair for these students.

The research will attempt to establish whether citizens trust the policy advice provided by think tanks and if its trustworthiness is higher or lower than that of the government. In doing so, the position of think tanks vis-à-vis the political sphere will be clarified. Moreover, the research will add to the literature by distinguishing between different types of think tanks in measuring their trustworthiness. The think tanks are divided in three groups based on an adapted version of the main typology used in the literature, which was originally devised by Weaver (1989). Since Lagendijk & Needham already utilized a slightly adapted version of Weaver‟s original typology in their study of a Dutch think tank (2012), we assume this typology to match the Dutch context. According to their version of Weaver‟s typology, think tanks can be divided in three groups, which range from science-based to politically based organizations:

1. Academic think tanks (policy research institutions / „universities without students‟)

= autonomous organizations producing scientific output with the purpose of understanding social phenomena (Abelson, 2000, p. 19). This type is closest to the scientific sphere.

2. Semi-academic think tanks (government contractors) = non-autonomous

organizations, commissioned or established by the government. The goal of these think tanks is still to provide scientifically sound advice (These are typically referred to as „planbureaus‟ in Dutch). This type is still based in the scientific sphere, but moves into the political sphere.

3. Non-academic think tanks (advocacy think tanks) = non-autonomous organizations,

their primary interest is to advance a certain ideology (Lagendijk & Needham, 2012, p. 478). Non-autonomy may be caused by reliance on external funding and the produced policy advice is not necessarily determined by scientific research. This type is closest related to the political sphere.

(24)

Even though this typology covers most of the current think tank landscape, it is important to realize that the stark contrast presented between the categories does not necessarily correspond with reality. Think tanks are not immune to change and may adapt to the political climate of the day and age. Moreover, a researcher might not classify a think tank in accordance with the image the think tank itself wishes to uphold. For example, the think tanks presented in the study by Smith and Marden (2008) demonstrate great effort to appear as a policy research institute while implicitly running as an advocacy think tank. The way a think tank presents itself may have implications for the extent to which a citizen is prepared to place trust in the organization. This research excludes the effects of framing through the experimental setting, in which the

organizational information of the think tank in question is formulated as objectively as possible by the researcher and not by any institution itself.

The third and final part of this thesis is dedicated to explaining the hypothesized difference in trust between the think tanks. The three types differ in the two main independent variables:

autonomy and scientific expertise. Autonomy refers to both spheres of sponsorship and ideology.

Scientific expertise refers to the extent to which the policy advice was formulated based on either own scientific research or previously published scientific work. In this part, I aim to uncover whether there is a causal relationship between autonomy and scientific expertise and the level of trust in think tanks.

(25)

3.2. Hypotheses

In this thesis, I argue that the institutionalist perspective best explains the level of trust citizens have in think tanks. It is the only perspective with regard for the various degrees in which think tanks take on either a scientific or a political role in the policy debate in the Dutch context. Such internal difference in the think tank category more accurately reflects the think tank landscape in the Netherlands and is therefore most likely to generate correct hypotheses on the trustworthiness of think tanks. In the Netherlands, non-academic think tanks move more towards the political sphere, as their main concern is lobbying and advocacy. For example, the „TeldersStichting‟, the think tank of the liberal Dutch governmental party VVD, claims to be a „political-scientific institution‟. However, their annual report shows that they do not engage in producing research themselves. Rather, the think tank hires external experts and adds liberal policy advice to the publications. Apart from these publications, the think tank mainly engages in publishing works on liberal history and organizing symposia and debates on liberal issues and principles (Van Schie, 2016). In the case of the TeldersStichting, it is clear that the political function trumps the scientific character of the think tank. In cases of academic think tanks, such as the Clingendael institute mentioned in the introduction, the scientific research is pivotal in all its activities. These think tanks move more to the scientific sphere.

Out of the four perspectives provided in the theoretical framework, only the institutionalist perspective would expect a difference in trust levels between the types of think tanks, which may range from credible providers of scientific expertise to thinly veiled lobbyist groups. The difference in think tanks is expected to translate into a difference in trust for two reasons. First of all, empirical studies in the institutionalist perspective by Rich (2005) and Doberstein (2017) already suggest that academic think tanks are perceived as more credible than their non-academic counterparts. There is a close relation between the concepts of credibility and trustworthiness, for credibility is regarded as one of the determinants to achieve trustworthiness (Graneheim & Lundman, 2004). Consequently, if a population perceives policy advice provided by academic think tanks as more credible, it is plausible that this perception will translate to trustworthiness.

(26)

Second, recent research has pointed out that the Dutch population has significant more trust in science compared to media, politics, or businesses (Dijstelbloem et al., 2013). Surveys taken in 2012 and 2015 by the Rathenau Instituut display similar results. A representative sample of the Dutch population scored their trust in several institutions on a scale from one to ten. Trust in science ranked by far the highest in both years, as evidenced in Figure 2.

Since science is more trustworthy than politics in the eyes of the Dutch citizens, it is likely that they would trust scientific institutions over political organizations. It can therefore be expected that policy advice provided by academic think tanks will be perceived as a more trustworthy factor in policy formulation. Therefore the main hypotheses of the research will be the following:

H1a: Policy advice provided by academic think tanks will be perceived as more trustworthy in comparison to policy advice provided by non-academic think tanks.

H1b: The difference in perceived trustworthiness can be explained by a higher level of scientific expertise and autonomy of academic think tanks as opposed to non-academic think tanks.

The earlier three perspectives result in other hypotheses. These will also be tested in the survey experiment to measure the extent to which their expectations hold. Elite theorists are radical in their position, arguing that think tanks are an instrument of the ruling elite to justify the social order. According to this Marxist perspective, both the government and think tanks are tools in the oppression of the masses. Both institutions are expected as a rule not to act in accordance

0 10 20 30 40 50 60 70 80 Source: De Jonge, 2015

Figure 2: Citizen's Trust in Institutions in the Netherlands

(2012 & 2015)

2012 2015

(27)

with expectations of positive behavior. Therefore, we would expect low levels of

trustworthiness for both governmental and think tanks policy advice, irrespective of their scientific or autonomous characteristics. Elite theory would thus result in the following hypothesis:

H2a: Given that the elite theory perspective holds, trustworthiness of policy advice provided by think tanks as well as trustworthiness of policy advice provided by a government ministry will be low

Also, the scientific credentials of a think tank would not matter for the level of trust, since all think tanks are seen to be orchestrated by the socio-economic elite. Consequently, a second hypothesis can be considered:

H2b: Given that the elite theory perspective holds, trustworthiness of policy advice provided by academic think tanks will not significantly differ from trustworthiness of policy advice provided by non-academic think tanks

Pluralist theorists have a less monopolistic outlook on the policy process, arguing that all policy advice is produced in part of the competition over policy formulation. Think tanks are

conceptualized as one of many actors in a larger network attempting to steer policy. Given that pluralists pay little attention to think tanks specifically, there is no clear expectation of the level of trust in think tank output vis-à-vis that of the government. Furthermore, think tanks are not differentiated according to their academic stature. Rather, they are all similar players in the policy competition. If they are truly equal players, we would not expect citizens to make a clear distinction between them either. This results in the same expectation as in hypothesis H2b:

H3: Given that the pluralist perspective holds, trustworthiness of policy advice provided by academic think tanks will not significantly differ from trustworthiness of policy advice provided by non-academic think tanks

The statist perspective does not make specific assertions when it comes to trustworthiness of think tank advice. Similar to the previous two perspectives, think tanks do not receive specific attention in studying the policy process. Therefore, there is no apparent expectation to be deduced from the literature on the level of trust in think tanks advice when compared to advice

(28)

coming from governmental actors. However, the literature inspired by the statist paradigm does iterate the strong expectation that the influence of think tank advice is minimal in comparison to the advice produced by the government. If in rare cases think tanks do succeed in having their policy advice taken into consideration by the government, it should not make a difference whether this advice comes from an academic or a non-academic think tank, since the state always decides on policy guided by the national interest. A distinction between types of think tanks seems not to be relevant. Therefore we expect no real difference in the levels of

trustworthiness of think tank advice either. This leads us again to a similar hypothesis:

H4: Given that the statist perspective holds, trustworthiness of policy advice provided by academic think tanks will not significantly differ from trustworthiness of policy advice provided by non-academic think tanks.

Apart from the hypotheses resulting from the think tank literature, there are some additional expectations from the data with regard to the political profile of the respondents. First of all, it is plausible that respondents who are more involved in public policymaking and/or display greater interest for political affairs assign higher levels of trustworthiness to the products of policy actors due to their more extensive knowledge on the policy process (see for example Christensen & Lægreid, 2014 for the effect of political involvement on trust in government). Alternatively, these respondents might also know more about the flaws and inherent problems of the process of public policymaking and of those involved. Either way, it would be enlightening to measure whether these factors have a significant effect.

Second, the political affiliation of respondents may have an effect on the extent to which they are likely to trust any policy actor. Earlier research into political identity and trust in the United States found that trust levels varied between conservative and liberal minded citizens, especially regarding trust in science (Rudolph & Evans, 2005; Gauchet, 2012). However, the same effect was not found in studies conducted in the Netherlands (Tiemeijer & De Jonge, 2013, p. 61). This contradiction highlights the importance of diversifying the think tank literature to non-American contexts, since the same assumptions may not hold outside of the U.S. It would therefore be interesting to see whether the results show an effect of political affiliation on the trustworthiness of policy actors, even though no singular hypothesis can be devised.

(29)

4. Methodology

4.1. Survey Experiment

In this research, I will employ a survey experiment to test the hypotheses. The survey experiment performs very well in establishing causal mechanisms, for the researcher is able to control over the environment. He/she is able to provide different stimuli to randomized groups, varying on the factor of interest to see whether there are significant changes in outcome (Toshkov, 2012, p. 167). In a survey form, the experiment is less costly to perform and can be spread more widely. Furthermore, employing the experimental method enables the concretization of earlier observational findings concerning think tanks while simultaneously introducing the trustworthiness factor.

4.2. Survey Design

The experiment was conducted among Dutch students at Leiden University in May 2018 (N= 121). Some other students had volunteered in advance to test the survey. These pre-tests allowed for further improvement on the terminology, extent of information given, and other stylistic shortcomings. Moreover, they already served to show whether the answers conformed to the hypothesized outcomes. Of about ten pre-tests, a majority conformed to the expected pattern. After the implementation of the improvements, the survey was distributed via email, social media, and leaflets with QR-codes and anonymous links, which were spread at multiple faculties and the library of Leiden University. Distributing the survey via email is preferable over the other methods, for it allows the researcher to track the target audience, response rate and to select participants. However, after a week of email distribution, it was clear that response rates did not develop rapidly enough to reach a sizeable participant group within an acceptable timeframe. By enlarging the reach through leaflets and social media, a proper sample could be gathered by the end of the month. In the end, 70% of the participants entered the survey via the anonymous link, 20% via email, and 10% via QR-code.

(30)

In the experiment, all the respondents encountered the same general biographical questions. In the second group of four blocks, the respondents faced a scenario of a policy decision. This decision displayed the same, non-salient issue in order not to distort the results. Respondents may already have a strong opinion on salient issues, such as migration or medical-ethical issues. If they would be presented with such an issue, they will probably not alter their opinion and disregard the trustworthiness of the policy advice that led to the decision. To overcome established preferences, the scenario describes how the minister of transport gets to spend his extra budget. He decides to spend this on improvement of lighting on the highways. This scenario is unlikely to arouse strong reactions.

The blocks differed only on the type of organization offering the policy advice leading to the policy decision. The blocks were randomly assigned to the respondents by Qualtrics software. In the control survey, it was a memo by the governmental ministry. In the three other blocks, the policy advice was distributed by CENI, a fictional think tank which was described in one block as a policy research institution, in another as a government contracted think tank, and in the last as an advocacy think tank. In order to relate these concepts to the Dutch context, the nature of the latter two types required some additional information. Along with the description of the government contracted think tank, the survey mentions the title of „planbureau‟. According to a study by Lagendijk & Needham (2012), planbureaus are a Dutch version of government contracted think tanks. Ministries establish planbureaus in order to provide scientific research in a specific field. Since this name is more likely to result in recognition by the respondents, it was added as a synonymous title. In the case of advocacy think tank, the vignette described a think tank affiliated to an undisclosed political party. The aim of this fictional think tank was to advocate and disperse the party‟s vision. Research may be utilized in achieving their mission, but it is not a strict necessity. The party affiliated think tank is the best example of an advocacy think tank in the Dutch context. Although internal differences exist, party affiliated think tanks are first and foremost concerned with advocating the party‟s ideological underpinnings. The Dutch government offers all political parties a subsidy for establishing their own think tank. Almost all established political parties make use of this opportunity, with the exception of the Party for Freedom (PVV) (Van Beek, n.d.).

(31)

A few multiple-choice questions followed the vignette with answering options distributed on a 5 point Likert scale. The answer options range from 1 (strongly disagree) to 5 (strongly agree). These questions inquired into the perceived trustworthiness of the policy advice, its perceived suitability as a general base for policy decisions, the extent to which its influence is desirable, its perceived impact on policy decisions and the perceived autonomy and scientific expertise in its establishment. The full vignettes can be found in Appendix A. After completing the survey, the students had a chance at winning a €20 gift card.

All theories are tested in the last multiple-choice questions. The first questions on trustworthiness test the main hypotheses and are meant to validate the institutional approach. If there is no significant difference in trust levels between think tanks or between think tanks and the government ministry, one of the earlier theories may offer a better explanation of the results. If overall scores on trustworthiness are low, the lack of difference may be accounted for by elite theory. The additional questions of impact and desirability of influence test the assumptions of pluralist and statist theories. If impact scores of think tanks are low in comparison to those of the government ministry, the statist theories may offer a better explanation of the results. Statists argue that think tank‟s impact is limited due to the state‟s ultimate decision power. However, the statist theory would expect that the influence of think tanks is not very desirable, for they have no business in state policy. Pluralists encourage as many actors to engage in politics as possible, to ensure a democratic process. Therefore, if the pluralist theory holds, scores on desirability would be high. In the end, the questions included in the survey test all predictions based on the theories, as summarized in Table 1.

Table 1 Elite theory Pluralist Statist Institutionalist

Difference in trustworthiness? X Overall low trustworthiness? X Low perceived impact? X