International organizations (IOs) appear to be increasingly contested. Britain's decision to leave the European Union (EU), disillusionment with UN climate negotiations, pushback against the handling of COVID-19 by the World Health Organization (WHO), recurring criticism of the International Monetary Fund (IMF), and the general rise of anti-globalist populism all signal substantial discontent with IOs.Footnote 1
This development in world politics has led to a wave of new research on the popular legitimacy of global governance—that is, the extent to which citizens consider an IO to exercise authority appropriately.Footnote 2 Legitimacy beliefs are a subset of general public opinion on an IO that capture citizens’ foundational support for the organization, independent of short-term satisfaction with its outputs.Footnote 3 A key issue in this literature pertains to the sources of such beliefs; research has found support for a variety of individual,Footnote 4 institutional,Footnote 5 and communicative drivers.Footnote 6
Yet one core expectation has failed to attract significant support, namely, that citizens’ legitimacy beliefs toward IOs would be influenced by their ideological orientations—that is, their “beliefs about the proper order of society and how it can be achieved,”Footnote 7 ordered on ideological dimensions such as the classic left–right dimension and the more recent GAL–TAN (green, alternative, liberal versus traditional, authoritarian, nationalist) dimension. Studies have typically found weak and inconsistent evidence for a general relationship between ideological orientations and legitimacy beliefs toward IOs.Footnote 8 Expectations that people on the left would systematically perceive IOs as more legitimate, or that people with traditionalist orientations would generally regard IOs more negatively, are usually not borne out by the data.Footnote 9
The weak support for this relationship is puzzling in several respects. It conflicts with findings in American and comparative politics that ideological orientations are of fundamental importance for people's attitudes toward political issues and institutions.Footnote 10 It deviates from research on public opinion toward US foreign policy, which typically identifies clear partisan differences.Footnote 11 It questions recent accounts that present ideology as central to the new wave of contestation over IOs around the world.Footnote 12 And it challenges the common observation that anti-globalist populism in contemporary politics is fueled by right-wing and nationalist movements.Footnote 13
Establishing whether and how ideology shapes legitimacy beliefs toward IOs is of critical importance. Popular legitimacy is a key resource for IOs, affecting whether they can stay relevant as political arenas, acquire necessary operational capacities, gain support for new policies, and command compliance with international rules.Footnote 14 When national governments make decisions on IOs’ powers and policies, the foundational support for these organizations in domestic populations matters, especially in light of the growing politicization of international cooperation.Footnote 15
In this article we resolve the puzzle by providing a new understanding of how political ideology matters for legitimacy beliefs toward IOs. We argue that citizens’ ideological orientations and legitimacy beliefs are linked in more complex ways than previous research has considered. Instead of expecting an unconditional and uniform association between ideological orientations and legitimacy beliefs, we establish that this relationship is moderated by a critical, overlooked factor: perceptions of the ideological profile of IOs. How citizens are located on certain ideological dimensions is insufficient to predict their attitudes toward IOs; what is also required is an appreciation of how they perceive the ideological profiles of IOs. We develop this new understanding by way of two contributions.
First, we advance a novel theoretical argument about the moderating impact of IOs’ perceived ideological profiles. Although IOs are commonly presented in international relations (IR) theory as political institutions performing non-ideological functions, we ground our argument in the assumption that citizens perceive IOs as having certain ideological profiles. IOs often pursue policy goals that have an ideological dimension, impact people in society differently, and become politicized in domestic politics, making it likely that citizens associate these organizations with particular ideological profiles. Based on this assumption, we theorize how perceptions of the ideological profiles of IOs shape the relationship between citizens’ own ideological orientations and their legitimacy beliefs toward IOs. Specifically, we develop the expectation that citizens accord IOs greater legitimacy when they perceive these organizations as ideologically more congruent with their own orientations.
Second, we test this expectation empirically by combining observational and experimental analyses based on new survey data.Footnote 16 The observational analysis evaluates our argument by relying on measures of citizens’ perceptions of the ideological profiles of four real-world IOs active in different issue areas, which might invoke varying ideological assumptions: the IMF, World Bank, WHO, and UN. The experimental analysis offers a complementary causal assessment by examining the effect of treatments that vary the ideological profiles of hypothetical IOs. For both analyses, we use original survey data from nationally representative samples in Brazil, Germany, Indonesia, and the United States. We select these four countries on the basis of an analysis of World Values Survey (WVS) data, which show inconsistent relationships between citizens’ ideological orientations and legitimacy beliefs in these countries at the aggregate level. If our analyses reveal that these varying aggregate patterns are underpinned by the very same dynamic at the individual level, then we can suggest a generic way in which political ideology matters for IO legitimacy beliefs.
Our key findings are twofold. First, respondents indeed tend to perceive existing IOs as having particular ideological profiles. On average, they rate some IOs as more left leaning and GAL and other IOs as more right leaning and TAN, with striking consistency. The exception is respondents in Indonesia, who, on average, do not appear to differentiate between the assessed IOs on the left–right dimension. This overall pattern supports our assumption that citizens associate IOs with certain ideological profiles.
Second, as expected, respondents’ perceptions of an IO's ideological profile shape the relationship between their own ideological orientations and their IO legitimacy beliefs. The observational analysis shows that the perceived ideological profiles of real-world IOs moderate the association between respondents’ ideological orientations and legitimacy beliefs. This finding is consistent across all four IOs and both ideological dimensions. In addition, the experimental analysis yields strong evidence of a causal effect of the ideological profiles of hypothetical IOs on the relationship between ideological orientations and legitimacy beliefs when respondents are treated with information about right- or TAN-leaning IOs. It provides partial evidence of such an effect when respondents are treated with information about left- or GAL-leaning IOs. We attribute this heterogeneity in experimental findings to respondents’ cognitive priors about the ideological profiles of hypothetical IOs.
Our findings have several broader implications for research. First, they contribute to scholarship on the sources of legitimacy in global governance by unraveling the puzzle of how political ideology matters for legitimacy beliefs toward IOs. Second, they speak to the literature on political ideology by demonstrating how ideology matters more broadly in world politics than previously understood. Third, they suggest that influential scholarship in IR has underestimated the extent to which IOs are perceived in ideological terms, calling for further research into the construction of such ideological perceptions. Fourth, they engage the large literature on contestation in global governance, suggesting that ideology is a powerful driver of whether citizens endorse or reject IOs.
Puzzle: Ideological Orientations and IO Legitimacy Beliefs
The expectation that ideological orientations influence people's beliefs regarding the legitimacy of IOs builds on a rich literature in comparative politics and IR that highlights the role of political ideology in attitude formation. Ideologies are shared sets of “beliefs about the proper order of society and how it can be achieved.”Footnote 17 The distinction between the political “left” and “right” is a central ideological dimension. The “left-wing” end of this dimension is typically associated with support for a more egalitarian distribution of income and greater government intervention in the economy, whereas the “right-wing” end is usually associated with the belief that inequality is a natural condition and with support for a more laissez-faire approach to politics.Footnote 18 The ordering of people's ideological orientations along the left–right dimension has been related to deep-seated societal cleavages,Footnote 19 to the dominant role of socialist and conservative parties in many party systems,Footnote 20 and to individual attitudes toward welfare and government spending.Footnote 21
Recent years have seen growing scholarly attention to another ideological dimension, which distinguishes between GAL and TAN orientations.Footnote 22 This dimension tends to be orthogonal to the left–right dimension, theoretically as well as empirically.Footnote 23 The GAL–TAN dimension captures attitudes on a range of social, cultural, and environmental issues that fit poorly on the left–right dimension but that have become more visible in contemporary politics, such as immigration, gender equality, ecological concerns, and national sovereignty. Other efforts to capture this ideological space beyond left versus right distinguish between materialist and postmaterialist values and between libertarian and authoritarian orientations.Footnote 24 Studies show that growing contestation along the GAL–TAN dimension has contributed to a restructuring of national party systems, as manifested in the rising importance of green parties and new nationalist parties.Footnote 25
A growing body of research in IR suggests that these ideological dimensions structure attitudes toward international politics as well. People on the left tend to be more positively disposed toward globalization,Footnote 26 international cooperation,Footnote 27 and foreign trade than people on the right.Footnote 28 These findings are consistent with ideas sometimes referred to as “left internationalism.”Footnote 29 The weight of the left–right dimension in international issues appears to be particularly strong in the US, where numerous studies of public opinion on US foreign policy and multilateralism find significant differences between Democrats and Republicans.Footnote 30 The evidence in Europe is more mixed.Footnote 31 Similarly, the GAL–TAN dimension has been found to shape attitudes toward international issues, especially when they concern policies on immigration, environment, and trade.Footnote 32 In the European context, this dimension appears to have grown in importance as regional integration has deepened, invoking issues related to state sovereignty.Footnote 33
Inspired by this literature, research on legitimacy in global governance has recently turned to political ideology as a promising explanation. The core expectation is that citizens’ ideological orientations shape their beliefs in the legitimacy of IOs.Footnote 34 Such an explanation would be consistent with earlier research on the role of political ideology in attitude formation, but also with observations that the legitimacy of IOs appears to be increasingly contested along ideological lines.Footnote 35 It would show that legitimacy beliefs toward IOs are ultimately ideological and not only, or even primarily, driven by other individual, institutional, and communicative factors.Footnote 36
Scholarship on legitimacy is distinct from more general research on public opinion in focusing specifically on the extent to which citizens extend foundational support to an IO, often captured through measures such as confidence or trust.Footnote 37 It builds on a sociological conceptualization of legitimacy as beliefs or perceptions that a political institution exercises its authority appropriately.Footnote 38
However, the well-founded expectation that people's legitimacy beliefs toward IOs would be related to their ideological orientations has found surprisingly limited empirical support. The evidence tends to be weak and contradictory across studies. Edwards analyzes attitudes in twenty-four developing countries and finds that left-leaning people are more critical of the IMF, the World Bank, and the World Trade Organization (WTO) than right-leaning people.Footnote 39 Similarly, Lee and Prather show in a survey-experimental study that left-leaning citizens in Australia and the US are less likely to support international law enforcement.Footnote 40 By contrast, Weßels and Strijbis find that left-leaning citizens are more supportive of IO authority when examining attitudes toward the UN in five countries (Germany, Mexico, Poland, Turkey, and the US).Footnote 41 Bearce and Jolliff Scott analyze attitudes toward IOs in thirty-two countries and find that voters of center and left-wing parties support IOs more than those of right-wing parties.Footnote 42
Other studies report no or inconsistent relationships. For instance, Torgler analyzes a broad sample of thirty-eight countries and does not find any evidence that left–right orientations matter for attitudes toward the UN.Footnote 43 Anderson, Bernauer, and Kachi find that legitimacy beliefs toward global climate governance are related to left–right ideology in the US but not in Germany.Footnote 44 Wratil and Wäckerle do not find left–right orientations to consistently moderate cueing effects on legitimacy beliefs toward the EU in five member states (France, Germany, Italy, Poland, and Spain).Footnote 45
The most comprehensive assessment so far was undertaken by Dellmuth, Scholte, Tallberg, and Verhaegen, who examine whether legitimacy beliefs toward six major IOs (International Criminal Court (ICC), IMF, UN, World Bank, WHO, and WTO) are systematically related to left–right and GAL–TAN orientations.Footnote 46 A pooled analysis of data from the seventh wave of the WVS (WVS7) in Brazil, Germany, the Philippines, Russia, and the US suggests that political ideology matters. However, country analyses show that these findings are driven exclusively by the US, where more left-leaning and GAL-oriented respondents have greater confidence in these IOs than right-leaning and TAN-oriented respondents. No association is found in the other four countries. The authors conclude: “For all the talk that value shifts in mass publics would drive a backlash against IOs, our study finds the values-based explanation to matter the least.”Footnote 47
To further substantiate this puzzle, we examine WVS7 data from all thirty-eight countries for which data are available on the determinants of citizens’ legitimacy beliefs toward six prominent IOs—the ICC, IMF, UN, World Bank, WHO, and WTO.Footnote 48 To operationalize legitimacy beliefs, we use the responses to a question about IO confidence, as discussed in the research design section. Respondents were asked to indicate their confidence in each of these six IOs on a four-point scale from none at all (coded 0), through not very much (1) and quite a lot (2), to a great deal (3). We use these items to create an index by summing up the scores for all six IOs and dividing by six to get the mean score. We then estimate the association between ideological orientation and IO confidence in a multilevel random-coefficient model, which allows us to examine how this association might differ across countries (see Appendix A, in the online supplement, for details on model specification).Footnote 49
Figures 1 and 2 summarize the results from this regression analysis (see also Tables A1 and A2 in the online supplement). Each figure shows the relationship for all thirty-eight countries pooled (left) and for a selection of four countries—Brazil, Germany, Indonesia, and the US—which illustrate the diversity of relationships at the country level and subsequently are chosen as our focal countries for the observational and experimental analyses (right). Figure 1 indicates a positive but insignificant association between left–right orientation and IO confidence when pooling all countries. Further analyses indicate that this relationship varies extensively across country samples: left-leaning citizens tend to have more confidence in IOs than right-leaning citizens in some countries (such as Japan, Turkey, and the US), while the relationship is the reverse in other countries (such as Chile, Indonesia, and Ukraine), and entirely absent in a third set of countries (such as Brazil, Germany, and Russia).
Figure 1. Association between left–right orientation and IO confidence
Figure 2. Association between GAL–TAN orientation and IO confidence
A similar picture arises in relation to GAL–TAN (Figure 2). Whereas a GAL–TAN orientation is associated with IO confidence in the pooled analysis, the nature of this relationship, too, varies across countries. While GAL-oriented citizens tend to have more confidence in IOs than TAN-oriented citizens in some countries (such as Japan, Indonesia, and the US), we see no relationship in most other countries (such as Brazil, Germany, and South Korea).
In sum: contrary to common expectations, previous studies as well as our own analysis provide only weak and inconsistent evidence of a general relationship between citizens’ ideological orientations and their legitimacy beliefs toward IOs. Next, we develop an argument that seeks to resolve this puzzle by privileging a hitherto overlooked factor: perceptions of the specific ideological profiles of IOs.
Argument: The Moderating Impact of IOs’ Ideological Profiles
Our argument introduces the ideological profile of IOs as a critical factor shaping the relationship between citizens’ own ideological orientations and their legitimacy beliefs toward IOs. When citizens perceive an IO as ideologically more congruent with their own orientations, they tend to regard it as more legitimate. This expectation rests on the assumption that citizens perceive IOs as having particular ideological profiles. This assumption sets our argument apart from previous research, leading to new expectations about how political ideology matters for IO legitimacy beliefs. In the following, we outline the logic of this argument and derive testable hypotheses.
In IR theory, IOs are oftentimes presented as apolitical institutions that perform non-ideological functions, such as coordinating expectations, lowering transaction costs, and monitoring noncompliance.Footnote 50 When states create and empower IOs, it is to help solve problems impeding cooperation, such as commitment problems, distributional problems, and enforcement problems. While IOs make decisions with distinct political implications in the execution of these functions, IOs themselves do not represent any particular ideological position. Much like legislatures at the domestic level, IOs are neutral arenas for resolving conflict between actors with competing interests.Footnote 51 From this perspective, there is nothing inherently ideological about IOs.
Our assumption moves away from this conception of IOs. We regard it as more plausible that citizens see IOs as ideological by nature rather than as apolitical. This perspective on IOs as ideological constructs has previously featured most prominently in constructivist,Footnote 52 critical,Footnote 53 and postcolonial scholarship,Footnote 54 but recently also made inroads into rationalist research with a focus on states’ ideological struggles.Footnote 55
To start with, IOs have often been intended to advance certain policy goals rather than others: free trade (WTO), human rights (UN), poverty alleviation (World Bank), labor protection (International Labor Organization), democracy promotion (Organization for Security and Co-operation in Europe), macroeconomic stability (IMF), and so on. These policy goals are usually hardwired into IOs through formal treaties and informal understandings, making them part of the organizational DNA. Because of the way IOs actively protect and promote certain political ideals, they have been described as “norm teachers”Footnote 56 and “moral authorities”Footnote 57 vis-à-vis states and societies.
Some observers describe the policy goals of IOs as generally associated with liberal political values, leading them to characterize the global governance system established by Western countries after World War II as a “liberal international order.”Footnote 58 Other observers speak of how IOs such as the UN typically advance a form of “welfare internationalism,” characterized by strong leftist ambitions to “accommodate the poor and disempowered.”Footnote 59 Yet others take an in-between position, describing the postwar order as an ideological compromise between free-market ideals and state interventionism.Footnote 60 Developments in recent years have also led observers to characterize some IOs, such as the Shanghai Cooperation Organization, as authoritarian in ideological profile because of the nonliberal goals they promote.Footnote 61
The political nature of the policy goals promoted by IOs invites citizens to associate IOs with particular ideological positions, even if IOs often seek to present themselves as apolitical.Footnote 62 Promoting free trade, defending labor rights, ensuring macroeconomic stability, and combatting poverty may not be regarded by citizens as neutral goals as much as efforts to further certain political ideals rather than others. Often, the goals of IOs link to dimensions of ideological contestation in politics, such as left versus right. For instance, free trade and deregulation are conventionally associated with market liberalism, while redistribution and social rights are usually associated with socialism or social democracy.
However, the exact ways in which citizens construct the ideological profiles of IOs are bound to vary. An IO's ideological position is ultimately in the eye of the beholder. It is not uncommon that the very same IO is criticized from competing political angles. Consider the EU, which oftentimes is debated on left–right lines of market intervention versus market liberalization.Footnote 63 Many left-wing critics portray European integration as a right-wing project to undermine social welfare provisions, while many right-wing critics regard the EU as a left-wing project for supranational market regulation. On other occasions, however, left-wing supporters see the EU as a way to tame global capitalism, while right-wing supporters see the EU as a way to liberalize markets that they regard as overly regulated at the national level.
Similar dynamics are at play in relation to global organizations. While the UN is often seen as a guardian of human rights, the organization has also been accused of violating those very same rights in its peace-building missions.Footnote 64 While many have praised the efforts of NATO to protect liberal democracy, critics have also condemned the organization for neo-imperialist interventionism.Footnote 65 While the World Bank is appreciated by many for promoting development in the Global South, it is also regarded by others as a neoliberal organization, representing a “Washington consensus” that has aggravated inequalities within and between societies.Footnote 66
Why individual citizens perceive IOs as having certain ideological profiles is an intriguing question that requires further research. Since our focus here is on the implications of such perceptions for legitimacy beliefs, we restrict ourselves to noting a variety of factors that may be at play next to the policy goals of these organizations. Citizens may interpret the ideological profiles of IOs differently depending on their own political leanings. Or they may form different perceptions of IOs because of varying exposure to the policies of these organizations.Footnote 67 Finally, how citizens perceive IOs ideologically may be shaped by how these organizations are politicized domestically, among political parties and social movements.Footnote 68
The assumption that citizens perceive IOs as having certain ideological profiles leads to novel expectations about how ideology matters for legitimacy beliefs. It suggests that citizens form attitudes toward IOs by jointly considering their own ideological orientations and their perceptions of the ideological profiles of IOs. The more they perceive the ideological profile of an IO as congruent with their own ideological orientation, the more likely they are to regard this organization as legitimate. The perceived ideological orientation of an IO thus functions as a moderating factor that conditions the relationship between citizens’ ideological orientations and legitimacy beliefs toward IOs.
This expectation extends theorizing on the role of ideology in politics, but also develops intuitions present in prior research on sources of legitimacy. Scott, for instance, speaks of how the legitimacy of an institution may derive primarily from “societal evaluations of organizational goals.”Footnote 69 Focusing specifically on IOs, Nielson, Hyde, and Kelley write that “actors may assess organizations not merely on how they operate and whether they accomplish their goals, but on what the goals themselves are and whether these are normatively desirable.”Footnote 70 Similarly, Dellmuth and Tallberg suggest that people's legitimacy beliefs toward IOs may be influenced by the social purposes of these organizations.Footnote 71 Research in American politics also offers support for this intuition, indicating that citizens’ trust in the US Supreme Court depends on the degree to which its rulings match their principled beliefs.Footnote 72
We formulate two hypotheses based on this expectation—one for each of the two key dimensions of ideological contention. Research has found varying support for left–right and GAL–TAN as ideological dimensions structuring the international attitudes of citizens.Footnote 73 Similarly, our empirical illustration in the previous section suggests that left–right and GAL–TAN are varyingly related to legitimacy beliefs toward IOs in different countries. Providing two hypotheses, therefore, allows a more nuanced and precise assessment of our core claim.
H1 The more citizens perceive an IO as ideologically congruent with their own ideological orientation on the left–right dimension, the more they regard this IO as legitimate.
H2 The more citizens perceive an IO as ideologically congruent with their own ideological orientation on the GAL–TAN dimension, the more they regard this IO as legitimate.
Research Design
To test the hypotheses, we collected cross-national survey data, which we analyze by combining observational and experimental methods. These two methods are used in a complementary fashion. The observational analysis evaluates the expectations based on citizen perceptions of the ideological profiles of real-world IOs, and the experimental analysis offers a causal assessment of the same relationship in the context of hypothetical IOs. In this section, we describe the design of the overall survey and discuss the operationalization of the hypotheses in the observational and experimental parts.
Survey Design
We chose to conduct the survey in multiple countries since we know from earlier studies and our own analysis that the association between ideological orientations and IO legitimacy beliefs varies greatly across contexts. We selected four countries that display considerable variation in the nature of this association at the aggregate country level, as we saw in Figures 1 and 2: Brazil, Germany, Indonesia, and the US (Table 1). With this selection, we aim to draw generalizable conclusions about the importance of ideological congruence for IO legitimacy beliefs. If our analyses indicate that these varying patterns at the country level are underpinned by the same dynamic at the individual level, then we will have likely discovered a generic way in which political ideology matters for IO legitimacy beliefs.
Table 1. Country selection
This selection has three additional advantages. First, these countries are prominent actors in IOs, which means that our findings on drivers of legitimacy beliefs in these IOs may carry particular importance for global governance. Second, all four are liberal or electoral democracies,Footnote 74 which reduces the risk that legitimacy might be interpreted differently across countries.Footnote 75 Third, all four have high levels of Internet penetration, which strengthens our confidence in the external validity of the survey data.Footnote 76
The survey was fielded online by Bilendi and Respondi between mid-October and mid-December 2022. Bilendi and Respondi used targeted quota sampling for age and gender, where quotas matched the age–sex distributions in each country (see Tables B3a–B5b in the online supplement for a discussion of the integrity of the sample). The final sample sizes are 3,246 for Brazil, 3,221 for Germany, 3,070 for Indonesia, and 3,284 for the US.Footnote 77
The questionnaire began with a range of questions on respondents’ political interests and ideological orientations on both the left–right and GAL–TAN dimensions. It then presented the experiment in the form of vignettes about hypothetical IOs, each followed by a question capturing the main outcome of interest—IO legitimacy beliefs—and a manipulation check. Finally, it asked about people's age, gender, partisanship, and perceptions of the ideological profiles of various existing IOs. We assessed respondents’ level of attention when filling out the survey questions by means of an instructional manipulation check.Footnote 78
In both the observational and experimental studies, we operationalized legitimacy beliefs through a question about the respondent's confidence in a specific IO, on a scale from 0 (no confidence) to 10 (complete confidence). The confidence measure has two main advantages. It aligns well with our understanding of legitimacy as the belief that an IO exercises its authority appropriately; and it allows us to evaluate our findings in the context of other studies that have used the confidence indicator.Footnote 79 Other measures are described in the next two subsections (see Appendix C for more details).
Observational Design
The observational part has two purposes. The first is to examine the extent to which people perceive IOs as having distinct ideological profiles on the left–right and GAL–TAN dimensions—a key assumption in our argument. The second is to test our hypotheses by investigating their main observational implications, namely, that respondents’ perceptions of the ideological profiles of IOs moderate the association between their own ideological orientations and their confidence in IOs.
H1 predicts that more right-leaning respondents have more confidence in an IO, the more they perceive this organization as having a right-leaning ideological profile, and less confidence, the more they perceive it as having a left-leaning profile. Conversely, more left-leaning respondents should have less confidence in an IO, the more they perceive this organization as having a right-leaning ideological profile, and more confidence, the more they perceive it as having a left-leaning profile. H2 generates equivalent observable implications for TAN- or GAL-leaning respondents in relation to confidence in IOs with TAN- or GAL-leaning ideological profiles. In the empirical analysis, we investigate these observable implications by examining the marginal effects of respondents’ ideological orientations on IO confidence with different individual perceptions of the ideological profiles of IOs.
To measure respondents’ ideological orientations, we rely on the responses to two survey questions. One asked for a self-placement on a quasi-continuous left–right orientation scale from 0 (left) to 10 (right). The other asked for self-placement on a gal–tan orientation scale.Footnote 80 This latter scale warrants more attention, given that there is, to date, no scholarly consensus on the measurement of GAL–TAN ideology among citizens. For our measure of GAL–TAN orientation, we include three key features of such attitudes according to the literature: environmental protection, migration, and cultural change.Footnote 81 The survey asked respondents to place their views on an eleven-point scale with endpoints labeled “environmental protection, free and safe migration, and freedom to choose gender identity, sexuality, and family relationships should be political priorities” (for GAL) and “economic growth, restricted migration, and protecting the traditional roles of women and men should be political priorities” (for TAN).Footnote 82
To measure perceptions of IOs’ ideological profiles, we used a question about where respondents would place four existing IOs—the IMF, World Bank, WHO, and UN—on the left–right and GAL–TAN dimensions. We selected these IOs as they work in different issue areas, potentially invoking varying ideological associations among respondents. Moreover, this selection has the advantage of varying the scope of IO mandates. The UN is a general-purpose organization, while the other three IOs are task-specific. Compared to the task-specific IOs, the UN likely deals with a broader range of political issues that citizens may perceive as relevant on both the left–right and GAL–TAN dimensions. As we will report, however, the findings from the observational analysis do not vary with the scope of IO mandates, which underlines the generalizability of our argument.
All four of these IOs have substantial authority in their respective domains, which makes it more likely that citizens have developed opinions on them.Footnote 83 The questions on perceptions of specific IO profiles (left–right io profile, gal–tan io profile) matched the formulation of items on respondents’ own ideological orientations (supplemental Table B1).
We also consider a range of potentially confounding factors that research has shown to matter for IO legitimacy beliefs: socioeconomic status (education and financial household satisfaction); geographical identification (with the country and the world); domestic institutional trust (confidence in domestic government and political satisfaction); demographics (age, gender); and general attitudes toward international cooperation.Footnote 84
Experimental Design
The experimental study complements the observational part by providing a causal analysis of how IOs’ ideological profiles moderate the association between citizens’ ideological orientation and confidence in IOs. Instead of leveraging respondents’ perceptions of existing IOs, as in the observational study, we used vignettes that systematically varied the ideological profiles of hypothetical IOs. While real-world information about IOs could have increased the credibility of the vignettes, hypothetical cases enabled us to vary the relevant theoretical factors with greater precision and to test them without having respondents think about specific IOs.Footnote 85
To improve generalizability, we conducted the experiment in four rounds, each covering a different issue area: migration, trade, climate change, and peace-building. The four experimental rounds were block-randomized to avoid priming effects.Footnote 86 This means that each respondent participated in four experiments, one per issue area, in randomized order. In each of the four rounds, respondents were also randomly assigned to either a treatment group or the control group to ensure that estimated treatment effects do not depend on potentially uncontrolled influences.Footnote 87 Also, the random assignment of respondents to treatment groups and a control group made it more likely that subjects received information about different ideological profiles (io = left, io = right, io = gal, io = tan) across rounds, which increased the chances of keeping respondents attentive and engaged. Balance tests suggest that the randomization was successful (supplemental Table D1).
Table 2 provides an overview of the vignettes. Each vignette began with a brief statement that introduced the respective issue area and mentioned a discussion among governments about how to address it. The statement referred to an ongoing discussion, signaling the topic's relevance and increasing respondents’ interest in the information.Footnote 88 After reading about the problem and the potential new IO, respondents were allocated to a control group or to one of four treatment groups: io = left, io = right, io = gal, and io = tan. The respondents in the four treatment groups received systematically varied information about the ideological profile of this organization, as expressed in its principal strategy for addressing the policy problem in question; respondents in the control group received no such information. All respondents were then asked to indicate how much confidence they would have in the new IO if it were created.
Table 2. Vignettes
A concern here could be that respondents’ sensitivity to treatments about the creation of new IOs would depend on their awareness of real-world IOs in the four issue areas. While we cannot rule out this possibility, we note that all four issue areas are populated by prominent IOs: the International Organization for Migration in migration, the UN in peace-building, the United Nations Framework Convention on Climate Change (UNFCCC) in climate, and the WTO in trade. We will return to this issue when discussing the interpretation of the results across issue areas.
Our baseline expectation is that respondents in the treatment groups, after receiving a vignette about a hypothetical IO that is more (or less) congruent with their own ideological orientation, will indicate more (or less) confidence in that IO, compared to respondents in the control group. We examine two observable implications, one for each hypothesis. H1 predicts that as respondents lean more to the right (or left), their confidence in the new potential IO will increase when learning that it has a right-leaning (or left-leaning) ideological profile, and decrease when discovering that it has a left-leaning (or right-leaning) profile. H2 generates an equivalent observable implication for TAN- and GAL-leaning respondents in relation to confidence in a new IO with a TAN or GAL profile. In the empirical analysis, we investigate these observable implications by examining the marginal effects of respondents’ ideological orientations on IO confidence. More specifically, we compare such effects for respondents in each treatment group (io = left, io = right, io = gal, io = tan) to those in the control group.
We provide validity checks in which we change the reference group from the control to the “opposing” treatment group, such that we compare the effects of the io = left and io = right treatments, and the io = gal and io = tan treatments. The analysis will provide further support for H1 if the association between respondents’ degree of leaning right and IO confidence is more positive when the new IO has a right-leaning profile than when it has a left-leaning profile—and equivalently for H2 and the GAL–TAN dimension.
Ideological orientation was measured before treatment, and IO confidence after treatment. In this way, the treatments did not prime respondents inappropriately to adjust their answers to the questions on ideological orientation. At the same time, pretreatment questions can affect how an experimental stimulus works.Footnote 89 However, in our case, it was crucial for the success of both the observational and the experimental analyses to measure unbiased ideological orientation, which led us to place these measures first.
Empirical Results
We begin by presenting the results from the observational analysis and then turn to the findings from the experimental analysis.
Observational Analysis
Our theoretical argument starts from the assumption that people perceive IOs as having distinct ideological profiles. We find support for this assumption in Figure 3, which shows how respondents locate the four existing IOs on the left–right and GAL–TAN dimensions. The upper panel is based on pooled data from all four focal countries. On average, respondents locate the UN and the WHO more on the left and GAL sides of the respective ideological dimensions, while they perceive the IMF and the World Bank as more right and TAN. This finding is underpinned by a series of t-tests, which suggest that these differences in mean ratings are statistically significant (supplemental Table C3a).Footnote 90
Figure 3. Average perception of IOs’ ideological profiles, pooled sample (upper panel) and country-specific samples (lower panels)
The lower panels disaggregate these results by depicting country-specific means. The results are fully robust across three country samples but deviate in Indonesia. On average, respondents from Brazil, Germany, and the US perceive the UN and the WHO as more left and GAL than the IMF and the World Bank, consistent with the pattern in the pooled analysis (Tables C3b–d). Indonesian respondents, however, tend to differentiate less between the four IOs on the left–right and GAL–TAN scales than respondents in the other three countries. On average, Indonesian respondents see the WHO as more GAL and the IMF as more left than the other three IOs. The remaining mean ratings do not differ significantly from each other (Table C3e).
Country-specific analyses yield additional interesting observations. Most notably, Indonesian respondents tend to locate all four IOs as right and TAN, while US respondents, on average, locate all four IOs as left and GAL. In contrast, respondents from Brazil and Germany tend to locate these IOs further apart and on different sides of these two ideological spectra. These patterns suggest that the perceived ideological profiles of IOs vary systematically across countries, which points to these organizations being politicized and debated differently in domestic political arenas.
Next, we test the hypotheses by seeing whether the perceived ideological profile of an IO moderates the association between respondents’ own ideological orientation and confidence in that IO. For these purposes, we regress confidence in an IO on indicators of individual ideological orientation, the perceived ideological profile of the IO, an interaction term between the two, and control variables.Footnote 91 The more respondents perceive an IO as right or TAN, the more positive we would expect the marginal effect of respondents’ ideological orientation as right or TAN on confidence in that IO to be.
The regression analysis supports our hypotheses. Figure 4 plots the marginal effect of respondents’ ideological orientation on their confidence in a particular IO against the perceived ideological profile of that IO. Each panel shows results from a separate regression model (see supplemental Table C4 for detailed results). The effect increases in size in the expected direction as the perceived IO profile varies, and these results are consistent across all four IOs and both ideological dimensions. Thus, we find clear observational evidence that ideological congruence explains levels of confidence in existing IOs.Footnote 92
Figure 4. Marginal effect of ideological orientation on IO confidence, plotted against perceived IO profile
Two further observations are in order. First, the effect of left–right orientation is larger than the corresponding effect of GAL–TAN orientation (Figure 4). And this difference is greater for the IMF and the World Bank. This pattern suggests that citizens’ left–right orientation matters more than their GAL–TAN orientation for confidence in IOs, especially economic organizations. But this might be due, at least in part, to how we measured GAL–TAN attitudes. We exclude nationalist attitudes to avoid overlap with geographical identification—an important alternative explanation of IO legitimacy.Footnote 93 To the extent that nationalist attitudes drive legitimacy beliefs toward IOs, including such attitudes in our measure of GAL–TAN orientation would likely have strengthened the effect of this ideological dimension.
Second, we do not find systematic differences between a general-purpose IO, such as the UN, and task-specific IOs, such as the IMF, WHO, and World Bank (Figures 3 and 4). Thus, the scope of an IO's mandate does not appear to matter for how it is perceived in ideological terms or for how those perceptions shape respondents’ confidence in the organization.
These findings are robust across different model specifications. Results remain mostly unchanged when adding weights (supplemental Table C6), when including only one ideological dimension at a time (Table C7), and when excluding those respondents who failed the attention check (Table C8). Moreover, the results are largely robust when replicating the models in Table C4 in country-specific regression analyses (Tables/Figures C9–12). In Brazil, Germany, and the US, estimates for all interaction effects are statistically significant and in the expected direction. By contrast, in Indonesia, we find the expected results only with regard to the UN. We rerun the analyses by pooling the data across IOs, and we see positive and significant interaction effects across all countries except Indonesia. In Indonesia, we find a weaker interaction effect for left–right and no significant effect for GAL–TAN (Table/Figure C13). Part of the explanation may be found in the context of Indonesian domestic politics, where political parties tend not to be ordered on a left–right scale as much as on a secular–Islamist scale.Footnote 94 By implication, Indonesians might be less likely to categorize themselves or political institutions on a left–right—or a GAL–TAN—scale than respondents in other countries.Footnote 95
Experimental Analysis
The experiment assesses whether the ideological profiles of hypothetical IOs, as described in vignettes, moderate the association between respondents’ ideological orientations and confidence in these IOs. For these purposes, we regress IO confidence on an interaction between ideological orientation and a treatment dummy (1 = treated with a specific IO profile; 0 = control group). There are four treatment dummies, one for each condition (io = right, io = left, io = tan, and io = gal). For each of the four issue areas covered by the experiment, we estimate one regression model that includes interactions between respondents’ left–right orientation and the two treatment dummies for this first dimension (io = left and io = right), and one regression model that includes interactions between respondents’ gal–tan orientation and the treatment dummies for this second dimension (io = gal and io = tan). This amounts to eight models in total, each testing one of our hypotheses per issue area. All models include fixed effects for countries and a range of control variables to reduce potential omitted variable bias arising from the inclusion of ideological orientation in the interaction terms. The analysis is based on the responses from those respondents who mastered the manipulation check for each round of the experiment.Footnote 96 Because there are no strong theoretical priors about country-specific differences in how IO ideological profiles matter, we pool the data across countries and discuss country-specific analyses in the robustness checks.Footnote 97
Our expectation is that the marginal effects of ideological orientation on IO confidence are statistically different among respondents in the groups treated with IO ideological profiles versus respondents in the control group, who did not receive any information about the ideological profile of the hypothetical IO. Figure 5 presents the main results from the regression analyses (see supplemental Table D2, and unconditional treatment effects in supplemental Figure D3).Footnote 98 Each of the eight panels in Figure 5 depicts the marginal effects of ideological orientation on IO confidence for two contrasting treatment groups and the control group in a given issue area. The evidence is in line with our expectations if the marginal effects of ideological orientation for io = right and io = tan are more positive than the marginal effects in the control group and if the marginal effects of ideological orientation for io = left and io = gal are more negative than the marginal effects in the control group.
Figure 5. Marginal effects of ideological orientation on IO confidence for each IO ideological profile treatment
The results, illustrated in Figure 5, provide mixed support for our expectations. We begin by discussing the results for H1 on the left–right dimension. The four upper panels of Figure 5 show that all four io = right treatments had the expected moderating effect, but none of the four io = left treatments did. In all issue areas, the marginal effect of left–right orientation on IO confidence is significantly more positive among respondents treated with the io = right condition than among respondents in the control group. Note that the moderating effect of io = right is statistically significant in the cases of migration and trade (Table D2), despite the overlap between the confidence intervals for the marginal effects in the respective control and io = right groups.
In essence, as respondents find out that the new organization leans right ideologically, their confidence in it grows the more they lean right and diminishes the more they lean left. But the io = left treatments did not yield the expected moderating effect. In none of the four issue areas is the marginal effect of left–right orientation on IO confidence significantly different among respondents treated with the io = left condition compared to respondents in the control group. Put simply, we do not find any evidence that the relationship between respondents’ ideological orientation and IO confidence is affected when learning that the new organization would have a left-leaning profile.
Turning to H2 on the GAL–TAN dimension, the four lower panels provide consistent support for our expectations with respect to io = tan treatments and more variegated support with regard to the io = gal treatments. In all four issue areas, the marginal effect of ideological orientation on IO confidence is greater among respondents treated with io = tan than among respondents in the control group. As respondents learn that the new organization would have a TAN profile, their confidence in it increases with stronger TAN leanings and decreases with stronger GAL leanings. For the io = gal treatments, the results vary across issue areas: while we find the expected moderation effect in the context of migration and peace, we observe no such moderation effect in relation to trade and climate.
Taken together, these results provide experimental support for the expectation that IOs’ ideological profiles moderate the relationship between individuals’ own ideological orientation and their confidence in IOs. Yet the results are not equally supportive across the two ideological dimensions and four issue areas, which motivates two interpretations. Both point to the role of political priors in shaping the effects of framed information on internationalist attitudes.Footnote 99
First, while all treatments of IOs as right- or TAN-leaning had the expected effects, treating them as left-leaning had no effects, and treating them as GAL-leaning had mixed effects. This pattern suggests that respondents, when reading that a new IO might be established, already assumed that this organization would have a left or GAL profile, such that, in most cases, our io = left and io = gal treatments provided only confirming information. This interpretation is supported by the results for the control groups, which show a negative marginal effect of ideological orientation on IO confidence in all eight models. Thus, even in the absence of information about an IO's ideological profile, respondents reported less confidence in it if they were more right- or TAN-leaning, likely because they took it for granted that it would have a profile incongruent with their own orientation.Footnote 100
Second, the io = gal treatments had the expected moderating effect in the areas of migration and peace but not in climate and trade. This pattern suggests that people have stronger priors about IOs having GAL profiles in climate and trade, making the io = gal treatments in these areas ineffective. By contrast, in migration and peace, people appear to have weaker priors about IOs having GAL profiles, making the io = gal treatments more effective. In these two areas, our io = gal treatments seem to have trumped any prior assumptions about the IO and affected the relationship between respondents’ ideological orientation and confidence in the organization.
An alternative interpretation could be that people are less responsive to treatments about hypothetical IOs in issue areas where well-known real-world IOs already exist. This interpretation would be in line with findings that cueing and framing effects tend to be weaker when people already have well-developed opinions about issues and organizations.Footnote 101 Yet this interpretation does not appear to fit the pattern of findings across issue areas: the UN's global role in peace-building and migration is likely just as well known as the WTO's role in trade and the UNFCCC's role in climate.Footnote 102
We performed several robustness and validity checks. First, we added weights to all models to account for deviations from the original sampling frame with regard to age and gender. The findings are robust to the addition of weights (Table D5).
Second, we excluded the respective control variables for how respondents place themselves on the alternative ideological dimension. The results effectively stay the same (Table D6).
Third, we included all respondents—not only those who mastered the manipulation check of the respective treatment, as in the main analysis. The results change marginally but still support our main findings (Table D7).
Fourth, we ran country-specific models, which confirm our main conclusions but also show interesting differences across countries (Tables D8–D11). Respondents are most sensitive to the ideological profiles of IOs in Germany, slightly less sensitive in Brazil and the US, and least sensitive in Indonesia. This pattern may reflect a greater prominence of contestation along the left–right and GAL–TAN dimensions in Germany than in Indonesia. This interpretation is corroborated by the manipulation checks, which show that Indonesian respondents had greater problems correctly recalling the treatments than German respondents (Figure D2), as well as our earlier observation that the left–right dimension serves a weaker structuring role in Indonesian domestic politics than in the other countries.
Finally, to check the validity of our experimental evidence, we reran the regression models by changing the reference group from the control group to the “opposing” treatment group. This offers an additional test of our hypotheses. More concretely, we estimated the difference in marginal effects between opposing treatment groups (io = right versus io = left, io = tan versus io = gal). This allowed us to test whether the association between respondents’ degree of right-leaning and IO confidence is more positive if the new organization has a right-leaning profile, compared to when the new IO has a left-leaning profile (and equivalent for the GAL–TAN dimension). We find consistent support for our expectations in three issue areas (migration, climate, and peace-building) but not for trade (Table D12). These results further underpin our argument.
Conclusion
Do citizens’ ideological orientations matter for their legitimacy beliefs toward IOs? While the literature is rich in expectations on this topic, it is poor in consistent findings. Despite widespread assumptions that political ideology structures people's attitudes toward international cooperation, studies of IO legitimacy beliefs have typically found weak or contradictory relationships.
This article puts forward a new understanding of how political ideology affects beliefs regarding the legitimacy of IOs. Instead of expecting an unconditional and uniform relationship, we explain why and how ideology and legitimacy are linked in more complex ways. We have advanced this understanding in two steps. First, we have developed a theoretical argument about the importance of citizens’ perceptions of IOs as ideological objects, suggesting that citizens accord IOs greater legitimacy when they perceive them as ideologically closer to their own political orientations. Second, we have examined this expectation empirically through observational and experimental analyses of new comparative survey evidence from four diverse countries.
Our key findings are twofold. First, citizens indeed tend to perceive major IOs as having particular ideological profiles, associating some IOs more with left and GAL positions and other IOs more with right and TAN positions. Second, citizens’ legitimacy beliefs toward IOs are moderated by their perceptions of IOs’ ideological profiles. When they perceive an IO as ideologically more congruent with their own political orientation, they tend to regard it as more legitimate.
While these findings are based on an ambitious multi-method and comparative design, we should also note the study's limitations and how future research might address them. First, while we have focused on four prominent IOs and analyzed data from four diverse countries, future research could assess the further generalizability of our findings by extending the study to other IOs and countries. Second, while we have examined the consequences of citizens’ perceiving IOs as having different ideological profiles, future research could explore why people come to view IOs in specific ideological terms.
For now, our findings carry four important implications. First, they demonstrate that the sources of beliefs about the legitimacy of IOs are richer than previously understood. While previous studies have found support for a variety of individual, institutional, and communicative drivers of legitimacy beliefs, they have not been able to identify a systematic association with political ideology.Footnote 103 We find that political ideology, after all, presents a key explanation of IO legitimacy beliefs, next to other individual-level factors such as socioeconomic status,Footnote 104 geographical identification,Footnote 105 and domestic institutional trust.Footnote 106 This establishment of ideology's contribution to IO legitimacy matters, since popular legitimacy constitutes a critical resource for IOs, shaping their standing and effectiveness in world politics.Footnote 107
Second, we find that political ideology has greater explanatory reach than previously established. While extensive research in American and comparative politics shows that ideology structures people's attitudes toward issues and institutions,Footnote 108 the evidence in IR has been more scattered and mainly drawn from the polarized US context.Footnote 109 Our findings indicate that political ideology matters more broadly for attitudes toward international issues and institutions than earlier understood. The causal importance of political ideology extends to legitimacy beliefs toward IOs and also applies to a varied set of countries beyond the US, such as Brazil and Germany, though Indonesia deviates from this picture. While the impact of political ideology is more complex in the context of global governance than in the domestic setting, these findings once again illustrate the commonalities and interactions between domestic and global arenas.Footnote 110
Third, our findings suggest that influential scholarship in IR has underestimated the extent to which IOs are perceived in ideological terms. While constructivist, critical, and postcolonial theorists have certainly underlined the ideological nature of IOs,Footnote 111 influential mainstream scholarship tends to conceive of IOs as apolitical institutions performing non-ideological functions, such as solving commitment and enforcement problems.Footnote 112 Much like domestic legislatures, IOs are regarded as neutral arenas for resolving political conflicts, which do not by themselves represent certain ideological positions. Our results indicate that this theoretical understanding of IOs poorly matches how IOs are perceived by people in general, who readily assign ideological profiles to IOs, irrespective of issue area and mandate scope. These findings call for a reconceptualization of IOs as ideological constructs and point to a promising research agenda on the sources, expressions, and implications of these conceptions.
Fourth, our findings contribute to a richer understanding of the contemporary contestation of IOs. Scholarship typically attributes the contestation of IOs around the world to the growing authority of these organizations,Footnote 113 the global power shift,Footnote 114 and the rise of authoritarian populism.Footnote 115 Our results support the notion that it is fueled by political ideology.Footnote 116 Ideological perceptions of IOs appear to be a driver of whether citizens endorse or reject these organizations as legitimate political institutions. As international cooperation becomes increasingly politicized,Footnote 117 ideological constructions of IOs are pushed to the fore, contributing to value-based contestation of global governance. This development is at odds with IOs’ conventional way of cloaking themselves in technocratic efficiency, which no longer seems to shield them from criticism.Footnote 118 A key issue going forward is how IOs will respond to such value-based contestation—by insisting on their non-ideological orientation, by adjusting their policy agendas to accommodate political opposition, or by embracing a role as fair and inclusive spaces for resolving ideological conflicts.
Data Availability Statement
Replication files for this article may be found at <https://doi.org/10.7910/DVN/DBFLYD>.
Supplementary Material
Supplementary material for this article is available at <https://doi.org/10.1017/S0020818324000304>.
Acknowledgments
We presented earlier versions of this article at the workshop “Legitimacy and Legitimisation of Global Governance in Crisis,” Durham, 3–4 November 2022; the European Consortium for Political Research General Conference, Innsbruck, 22–26 August 2022; the annual meeting of the American Political Science Association, Montreal, 15–18 September 2022; the workshop “New Research on IO Authority,” Lüneburg, 9–10 February 2023; the annual conference of Political Economy of International Organization, San Diego, 4–6 May 2023, and at seminars at Stockholm University and the University of St. Gallen. For helpful comments and suggestions, we are particularly grateful to Ryan Brutger, Klaus Dingwerth, Tina Freyburg, Farsan Ghassim, Liesbet Hooghe, David Lake, Christoph Mikulaschek, Michal Parízek, Henning Schmidtke, Sujeong Shim, Hannah Smidt, Jan Teorell, Stefanie Walter, Joonseok Yang, and the anonymous reviewers for IO.
Funding
We acknowledge the generous funding provided by Riksbankens Jubileumsfond (M15-0048:1), the Swedish Research Council (2021-01047; 2023-01429), the European Research Council (101097437), and the German Federal Ministry of Education and Research (01UK1810).
Open access
