Legitimacy beliefs towards global governance institutions: a research agenda

ABSTRACT Legitimacy is central to the functioning of global governance institutions (GGIs) such as the European Union (EU) and the United Nations. There is a vibrant debate about legitimacy in International Relations, and a burgeoning literature in comparative politics on public attitudes towards the EU. Yet, these literatures rarely speak to each other, which has resulted in missed opportunities for theoretical advancements on the sources and consequences of citizens’ legitimacy beliefs vis-à-vis GGIs. To assist researchers in advancing on this state of the art, this research note develops a conceptualization of popular legitimacy as a multidimensional belief system including both moral convictions and self-interest. A statistical analysis of public attitudes towards the EU from 1973 to 2012 suggests that commonly used survey measures capture self-interest rather than moral beliefs. This note concludes by suggesting a research agenda intended to push theory and survey research on legitimacy beliefs towards GGIs forward.


Introduction
How legitimate are global governance institutions (GGIs) such as the European Union (EU), United Nations (UN) or the International Monetary Fund (IMF)? Why do some citizens believe in the legitimacy of GGIs while others do not? These questions are relevant to large literatures in comparative politics and International Relations (IR) on legitimacy, compliance, and effectiveness. It matters significantly for the viability of a political body that citizens believe that its authority is appropriately exercised (Suchman 1995;Weber 1922Weber /1978. Contemporary history shows that this general point for governance arrangements is applicable also to GGIs (Zürn 2018).
In IR, scholars have long debated the normative legitimacy of GGIs, focusing on the philosophical standardssuch as fairness and democracythat institutions ought to live up to in order to be legitimate (Buchanan and Keohane 2006). In the context of the EU, there has been a vibrant debate about what is the adequate standard to judge the legitimacy of the EU. At the core of this debate is whether the EU's legitimacy should be understood on the basis of the same or reassessed democratic standards as those applicable in the context of nation states (Moravcsik 2002;Scharpf 1999;Schmidt 2013).
More recently, researchers have turned to questions of why GGIs enjoy legitimacy in practice. From a sociological perspective, scholars focus on the processes through which those who are subject to rules confer legitimacy to GGIs or withhold legitimacy from them. Audiences of GGIs include both state and societal actors that may be in or outside a GGI's jurisdiction . Delegitimation and legitimation are processes of contestation and endorsement of GGIs with the aim of affecting audience beliefs in a GGI's legitimacy (Schmidtke 2018;Zaum 2013).
In this research note, we focus on the beliefs of a central audience in GGIs' rightful exercise of authority: citizens. While states have long been regarded as the principal audience of GGIs in IR (Binder and Heupel 2015;Hurd 2007), there is a burgeoning literature directing attention to popular legitimacy, i.e., the degree to which ordinary citizens believe in GGIs' legitimacy (e.g., Dellmuth and Tallberg 2015; Karp et al. 2003;Schlipphak 2015).
This research note discusses an extraordinary gap in existing research: despite the acknowledged importance of popular legitimacy for GGIs, we still know little about how to conceptualize it in the context of public opinion research, and how to measure it appropriately through survey formats and survey indicators. Previous public opinion studies (for a recent overview, see Hobolt and de Vries 2016) and assessments of the dimensionality of EU attitudes (Boomgaarden et al. 2011;Fuchs 2011;Tausendpfund and Schäfer 2018;Westle 2007) have conceptualized attitudes in relation to 'specific support' and 'diffuse support' (Easton 1975) and not in relation to popular legitimacy as appropriated in normative or sociological approaches. As we discuss in the ensuing section, we propose an alternative conceptualization of legitimacy beliefs as a multidimensional cognitive belief system. This belief system is based on two analytical dimensions of legitimacy discussed in IR: moral beliefs and self-interest. It is directed towards an institution as a whole, i.e., both towards specific and diffuse objects related to the institution. As discussed in greater detail below, while Easton argues that legitimacy (diffuse support) refers exclusively to moral beliefs, we argue that legitimacy as a belief system links both moral convictions and self-interest. Moral convictions are a necessary condition for legitimacy beliefs. Self-interest is not enough in order for legitimacy beliefs to be present, but is likely linked to moral beliefs in the legitimacy belief system. In this vein, our argument adds analytical depth to Easton's conceptualization of support that will be useful in pushing survey research on legitimacy beliefs forward. In this respect, this research note seeks to bridge IR scholarship on legitimacy and existing literature in comparative politics on public attitudes toward GGIs.
The remainder of this research note proceeds as follows. First, we develop an argument that popular legitimacy is best conceptualized as a multidimensional belief system to allow for precise measurement through survey indicators. Second, using the extended Eurobarometer Trend File , we illustrate that commonly used survey indicators to analyse EU attitudes capture self-interested calculations rather than moral beliefs. We focus on the EU, as availability of different survey indicators over time and across bodies is best compared to other GGIs (Dellmuth 2018). We conclude by sketching a research agenda for comparative politics and IR scholars for further conceptual and empirical assessment of legitimacy beliefs vis-à-vis GGIs, using novel ways to gather data on legitimacy beliefs and their priors.

Conceptualizing legitimacy beliefs
Popular legitimacy is commonly referred to as the degree to which citizens believe GGIs to have the right to rule (Suchman 1995). The conventional wisdom holds that legitimacy beliefs are moral beliefs (Agné 2018;Buchanan and Keohane 2006;Gilley 2006;Hurd 2007: 30). From this vantage point, citizens evaluate legitimacy based on varying criteria that are not grounded in habitual obedience, fear of sanctions, or entirely instrumental cost-benefit calculations (Suchman 1995). Legitimacy encompasses acceptance of the rules and requirements of political institutions when a person sees them 'as conforming to his [sic!] own moral principles, his own sense of what is right and proper in the political sphere' (Easton 1975: 451;Suchman 1995: 574).
Legitimacy as moral beliefs may be shaped by normative debates about moral issues, although sociological legitimacy and normative legitimacy are analytically distinct (Buchanan 2002). For example, when making judgments about appropriate authority, citizens may be influenced by the academic-philosophical thinking in their context , by normative debates about the efficiency and fairness of institutional output (Buchanan and Keohane 2006), or by debates about whether political institutions have the right to rule if they can impose political obligations on citizens, or if they also need to fulfil moral criteria (Buchanan 2002).
By contrast, a number of researchers assume legitimacy beliefs to include at least partially utilitarian self-interest. From this perspective, institutions are judged on the basis of whether they are appropriate instruments to yield a specific payoff, which tends to be a function of the performance of an institution (Dellmuth and Tallberg 2015;Scharpf 1999). This payoff may be either calculated based on estimated impact on personal wellbeing ('egotropic' considerations) or on the economic performance of citizens' subnational region or country ('socio-tropic' considerations). Moral and self-interested calculations may go hand in hand when forming legitimacy beliefs, as 'the collectively valuable institutional functions that principles of political legitimacy support will rarely be wholly moral in character' (Macdonald 2015: 419).
Thus, although legitimacy beliefs may focus on the public utility of governance, such moral beliefs are likely associated with self-interested calculations. This begs the question of whether self-interest is enough in order for legitimacy beliefs to be present. If we allow for a conceptualization of legitimacy beliefs limited to self-interest, we would open up to justifications of political power divorced from considerations of public interest, which is normatively undesirable for convincing reasons discussed in depth by Buchanan and Keohane (2006). We therefore argue that in order for legitimacy beliefs to be present, they need to be deep seated in moral considerations about the public utility of governance, but that such moral beliefs are linked to self-interested calculations in a complex belief system.
These insights are in line with previous views on legitimacy as supportive attitudes that include self-interested calculations in moral beliefs. For example, Scharpf (2009) puts forward that legitimacy beliefs indicate that individuals are more likely to accept GGI rules or policies that are perceived as undesirable to them. In this logic, moral beliefs in a GGI's right to rule are present if citizens at the same time accept that the GGI may have undesired impacts on them personally.
Our argument is based on the assumption that citizens' legitimacy beliefs are based on pre-existing knowledge or socialization, as is the case for citizen attitudes towards domestic political institutions. However, most citizens are not socialized into supranational or global political institutions. In the context of regional and global governance, long-term moral beliefs may have to be newly established over time, based on the grounds of repeated experiences with the institutions, as has been argued for citizens' legitimacy beliefs toward institutions in newly developed democracies (Fuchs 2011).
Our conceptualization of legitimacy as a multidimensional belief system complements previous scholarship based on Easton's works on political support (1965;1975). A large literature on public attitudes towards the EU and other GGIs draws on the distinction between 'diffuse support', i.e., support vis-à-vis a political institution as a whole, and 'specific support', i.e., evaluations of specific politicians or civil servants (e.g., Fuchs 2011; Tausendpfund and Schäfer 2018; Westle 2007). Diffuse and specific support are conceptualized based on citizens' relationships with institutions and specific actors, and thus refer to a vertical link between citizens and the objects they evaluate. Our conceptualization is relevant to both diffuse and specific support inasmuch as moral convictions and self-interest can form in relation to both diffuse and specific objects. Citizens can hold moral beliefs about whether both political institutions as a whole or specific politicians or civil servants are in the interest of society at large. Self-interested calculations may respond to experiences with GGIs as a whole, or to specific representatives of these organizations. By contrast, our argument implies that legitimacy beliefs link moral convictions and utilitarian self-interest, and emphasize the moral component as a necessary condition in order for legitimacy beliefs to be present. Thus, our conceptualization enables future scholarship to explore the cognitive dimensions of legitimacy as a belief system, both in relation to the IR and the comparative politics literature, providing a more fine-grained understanding of legitimacy.

Measuring legitimacy beliefs using existing survey data
To assess legitimacy beliefs toward GGIs, we would ideally need cross-national survey data, as legitimacy beliefs are formed in country-specific social contexts (Suchman 1995), and repeated measurements over time to analyse the stability or volatility of legitimacy beliefs.
To identify the most commonly used indicators in the literature on public attitudes toward the EU, we collected data on existing journal articles on EU attitudes in five highly ranked journals in which the debate about EU attitudes and Euroskepticism mostly takes place: European Journal of Political Research, European Political Science Review, European Union Politics, Journal of Common Market Studies and Journal of European Public Policy. We cover the period of the last ten years, which we consider sufficiently long to gain a good impression on the most commonly used legitimacy indicators. This selection process yields 34 articles dealing with attitudes toward the EU among EU Notes: N = Number of articles. Rows may add up to more than N as some articles contain more than one measure. Authors' collection of cross-national public opinion studies of EU legitimacy over the past 10 years (see Appendix A). a This category includes measures tapping into specific conceptions of or feelings towards the EU, and were used mostly in addition to either trust or support measures.
citizens comparatively across countries (Appendix A). This overview demonstrates that support and trust measures are most central (Table 1). Indeed, much of the debate on EU public opinion revolves around studies of public support for the EU (e.g., Braun and Tausendpfund 2014;Hooghe and Marks 2005;Rohrschneider 2002), and trust in EU institutions (e.g., Harteveld et al. 2013;Muñoz et al. 2011; for an overview, see Hobolt and de Vries 2016). Support measures typically rely on the Eurobarometer question of whether people think that membership in the EU is good, bad, or neither good nor bad, or the question of the degree to which one's country benefitted from EU membership (Appendix B). Support measures capture instrumental concerns and private interests by making people think about what benefit one's country has from EU membership. Similarly, by presenting the question about the degree to which one's country benefitted from EU membership, respondents are asked directly to monitor the payoff of EU membership for their country, and by implication, for themselves. With regard to trust, existing studies rely on metrics of trust either in the EU or specific bodies of the EU, such as the European Commission, the European Court of Justice, and the European Parliament. Some scholars argue that trust or confidence measures are appropriate measures of legitimacy as moral beliefs, while others open up for the possibility that they may capture self-interest to some extent (Dellmuth and Tallberg 2015). Taken together, this discussion suggests that support indicators likely omit moral beliefs while trust indicators might or might not contain moral beliefs.
To illustrate this point, we use the Extended Eurobarometer Trendfile , which is the most encompassing dataset to date containing several support and trust measures over a long time period. 1 We (re)coded all of the variables in binary measures, so that 1 indicates respondent's trust into or support of the EU and its institutions, and 0 the absence of trust or support (Appendix B).
Our analysis proceeds in three steps. First, bivariate (tetrachoric) correlation analyses between the binary survey items shows that all support and trust items correlate strongly (Table C1). 2 The coefficients vary between .7 and .9, and thus suggest very strong bivariate correlations.
Second, we perform an exploratory factor analysis to identify whether the items load on a single or multiple factors. Explanatory factor analysis is a useful tool to identify whether different variables measure the same concept or dimension (factor) or if the items are in fact measuring two or more dimensions (factors). If the items were to yield multiple factors, then this would indicate multiple dimensions of legitimacy beliefs, making it possible to examine the degree to which legitimacy beliefs are moral values or rather considerations pertaining to self-interest. However, all items load on one factor that captures more than 90 percent of the variance of citizen perceptions of the EU, across all configurations and time points (Table 2).
Third, these results are corroborated by a test of scale reliability between the items. Scale reliability refers to the degree to which the indicators that are theoretically considered to describe the same concept actually measure the same conceptthat is, if they strongly correlate with one another. If indicators were to capture the difference between the interest-based and moralbased dimension, we should see a low scale reliability or low internal homogeneity. Results of a series of Omega tests suggest a very high degree of internal homogeneity of the common dimension of the five or three variables (Table 2). 3 These results are robust in a series of robustness checks (Appendices C and D).
Taken together, the results suggest that the main indicators used in the literature cover only one dimension. This dimension measures favourability toward GGIs based on self-interest, implying that existing survey measures thus do not lend themselves to make inferences to the belief system on legitimacy.

A research agenda on legitimacy beliefs towards global governance institutions
Never has there been a more exciting time for research on the popular legitimacy of GGIs. GGIs are today more publicly contested than ever (Zürn 2018), and methodological advances enable us to examine the beliefs of individuals in GGIs' legitimacy, and the analytical dimensions of these beliefs. This research note is intended to help us think afresh about how to conceptualize legitimacy beliefs in ways that lend themselves to appropriate measurement at the individual level. The burgeoning literature on public support and trust towards GGIs, especially the EU, has heightened the need for conceptual clarity and for parsing out measures of support, confidence, and trust that  , 1970-2012. are sometimes used to make inferences to legitimacy. Although the evidence in this research note comes from the EU, we argue that our results indicate important challenges in existing empirical assessments of legitimacy beliefs vis-à-vis GGIs more generally, as data coverage is best in the context of the EU compared to other GGIs.
Bringing public opinion research on attitudes towards GGIs and legitimacy in comparative politics and IR research closer together presents us with a research agenda that seeks to meet the conceptual, theoretical, and empirical challenges in the assessment of GGIs' popular legitimacy and its sources. This is all the more important as it helps to more strongly connect this literature on citizens' legitimacy beliefs to the literature analysing legitimation strategies of domestic and international actors (Ecker-Ehrhardt 2018, Schmidtke 2018). Formulated another way: By proposing a research agenda focusing on a more elaborate way to conceptualize and measure citizens' legitimacy beliefs toward GGIs, we suggest improving the analysis of the causes and effects of such legitimacy beliefs. We propose that this agenda consists of three parts.
First, there is a need for future research to examine the dimensionality of legitimacy beliefs conceptually. In this research note, we propose that citizens hold belief systems on legitimacy, which link moral beliefs to self-interested considerations about GGIs and their activities. Our empirical assessment shows that existing survey measures do not adequately capture individual beliefs in an institution's moral right to rule, but rather tap into citizens' self-interested calculations about the personal utility gained from GGIs. Hence, the understanding and explanation of GGI legitimacy would gain from future research building on our conceptualization of legitimacy beliefs, drawing from both comparative politics and IR. Although survey researchers sometimes make inferences to legitimacy theories, legitimacy research is an active and multivocal area of research where conceptualizations of legitimacy beliefs are debated. Inferences to legitimacy beliefs from studies of attitudes without a systematic conceptualization anchored in IR legitimacy research risks fostering the impression that all supportive attitudes are relevant to legitimacy beliefs. Furthermore, future research could usefully empirically explore the dimensionality of legitimacy beliefs across GGIs. For instance, do citizens' legitimacy beliefs mirror moral-based and interest-based dimensions equally across different GGIs? Or are moral-based beliefs stronger in the context of some GGIs than others? Several arguments may be plausibly developed and tested to answer these questions, pertaining, for instance, to varying levels of authority or issue-area.
Second, we are convinced that conceptualizing legitimacy beliefs systematically results in scholarship that will be able to find novel ways to measure legitimacy beliefs. Methodological insights from social psychology and comparative politics on survey experiments and conjoint analyses may be helpful in this endeavour (Chong andDruckman 2007, Hainmueller et al. 2014). In fact, there are survey experiments trying to identify the influence of a policy being in favour of or disadvantageous to the respondents' country on respondents' acceptance of that policy (Mutz and Kim 2017). Such endeavours can be considered to be a first step to measure the dimensionality of legitimacy beliefs, e.g., by linking attitudes to behavioural indicators of legitimacy, such as a willingness to voluntarily comply in the face of undesirable rules, policies or decisions. Future public opinion research may also benefit from using qualitative methods, such as in-depth interviews or focus group discussions with citizens from different strata of society, to gain insights about the dimensionality of legitimacy beliefs.
Third, starting from better measures, researchers will be able to develop more compelling research designs to study why, when, and how legitimacy beliefs are formed, maintained, or undermined. That moral beliefs and selfinterest related to a GGI may influence each other is a common claim in the literature (Suchman 1995;Tallberg et al. 2018). However, rather than assuming that this will be the case, the claim should be subjected to empirical research. Questionnaires that contain several measures tapping into a legitimacy belief systempreferably over timemay help us to disentangle the correlation between moral beliefs and self-interest, and provide us with more information about the factors influencing both of them.
In this respect, legitimacy beliefs vis-à-vis GGIs may be shaped by different individual and context factors. Future research could help unravelling these associations in detail. Given that legitimacy beliefs are grounded in moral convictions, we may expect factors such as religion, identity conceptions and personal predispositions related to e.g., authoritarianism to play a stronger role for moral beliefs than for self-interested beliefs. Moreover, emotions may affect the formation of legitimacy beliefs. While the study of emotions has in recent years received increasing scholarly attention in public opinion research, legitimacy beliefs have so far not been connected to emotions. This is surprising in an era of increasing legitimation and delegitimation struggles in global governance (Zürn 2018), where populist rhetoric might mobilize peoples' anxieties or threat perceptions (Hameleers et al. 2017). Indeed, previous research on emotions in public opinion formation (Gross et al. 2009) and recent evidence on EU sympathy (Garry and Tilley 2015) provides useful insights as to how to conceptualize and measure emotions through survey items. In this respect, systematic tests of the effects of legitimation and delegitimationfor example through populist rhetoricin the media or in parliamentary debates (Schmidtke 2018) on legitimacy beliefs become highly relevant. Such tests could ask questions relevant to both IR and comparative politics, such as: What types of elite communication might have the largest effect on citizens' legitimacy beliefs toward GGIs: cueing, framing, or priming? What characteristics make a claim contesting or defending a GGI's legitimacy more or less credible to the public? What types of delegitimation or legitimation attempts make citizens most emotional about GGIs, with potential effects on legitimacy beliefs? In addressing such questions, future research could draw from previous research on heuristics (Dellmuth and Tallberg 2018;Schlipphak 2015) to examine whether (de-)legitimation practices are more (or less) effective in swaying legitimacy beliefs among more (or less) politically knowledgeable citizens.
To conclude, our hope in raising these issues is that scholars in comparative politics and IR join in promising research co-operations. We recognize that contributing to both comparative politics and IR is not an easy task, but we believe the payoffs for theory development can be substantial. Notes 1. As support questions have not been asked together with trust questions in the same Eurobarometer surveys since 2012, we cannot extend the timeline. As trust variables are not included in the same questionnaire, we can only use either trust in the EU or trust in specific EU institutions. Data are available at http://www. lisadellmuth.se (accessed 12 February 2019). 2. Commonly used Pearson's correlation coefficients, factor analysis, and Cronbach's alpha are inappropriate in the context of binary measures. We therefore use tetrachoric correlation matrices (Bonett & Price 2005) and McDonald's Omega to test the reliability of the single factor solution (Dunn et al. 2014: 406). 3. The literature on psychological measurement recommends McDonald's Omega as the more appropriate instrument than Cronbach's alpha (Dunn et al. 2014).