Regional inequalities and political trust in a global context

ABSTRACT Climate change, health pandemics, structural decline, and more – the challenges of solving political problems are daunting, particularly when the political institutions addressing them are not trusted. This article tests the economic theory that residents of high-income regions are more likely to trust political institutions, given their positive experiences with services and opportunities, against the rival argument that predicts a negative effect of regional disadvantage within a country on political trust. Using European Values Study and World Values Survey (2017–2020) data, combined with socioeconomic data for 606 regions in 42 countries, this paper analyses samples of regions both in and outside the EU. The results suggest that people living in wealthy EU regions – both in absolute and relative terms – trust national government more and the EU less. In the global sample, the evidence is more variegated and corroborates economic theory only in democracies. The article sketches implications for regional inequality, political trust, and legitimacy research.


Introduction
What effect, if any, does regional income have upon trust in political institutions in Europe and other continents?In view of pressing policy challenges confronting contemporary societies, we need a firm evidence base on why individuals trust political institutions, and how such institutions can become more trustworthy.Public trust in government in Europe has declined since the Euro-crisis in 2009 (Foster & Frieden, 2017) and support for populist and Eurosceptic political parties has increased in many regions (Ejrnaes et al., 2023).Taking a global perspective, people worldwide hold only moderate levels of political confidence in the European Union (EU) and other supranational institutions, such as the African Union (AU) or the Southern Common Market (Mercosur), in recent decades (Dellmuth & Tallberg, 2023, ch. 2).
These dynamics in public perceptions have a spatial dimension.In the EU, people living in economically distressed regions have a higher propensity of casting a vote for populist and anti-globalist parties (Broz et al., 2021) and of distrusting political institutions (Lipps & Schraff, 2021).This spatial-economic logic has been front and centre in the literature on regional inequalities.'Regional inequalities' denote income differences between regions at one point in time, be they across or within countries, but also differences in regional income trajectories over time.Extant research in EU studies has debated why regional inequalities might affect populist attitudes and voting.Within standard economic logic, people in poorer regions should feel 'left behind' and become alienated from mainstream politics (Rodríguez-Pose, 2018).
However, the evidence for this logic is mixed.While some studies find that people in poorer regions are more prone to Euroskeptic voting (Nicoli & Reinl, 2020), others find that richer places vote more for anti-EU parties than poorer ones (Dijkstra et al., 2020).Yet others find that the effects of regional inequalities are variegated when studying political trust in national and EU institutions as outcomes (Lipps & Schraff, 2021;Mayne & Katsanidou, 2023; see Hobolt &De Vries, 2016 andEjrnaes et al., 2023, for overviews).
This article provides a theorisation and empirical inquiry into the effects of regional inequality on political trust.It makes two principal contributions.First, it draws from economic theory to identify and elaborate two arguments about how regional inequality may affect political trust at the individual level.Absolute income theory predicts that residents of poorer regions do not have access to adequate labour market opportunities and public services, which weakens their political trust.Relative income theory assumes people to base their utilitarian considerations on how well other regions are doing, which is why regional disadvantage within a country is expected to depress political trust.These arguments are inspired by broader literatures on political trust (e.g., Harteveld et al., 2013;Lipps & Schraff, 2021) and regional inequality (e.g., Broz et al., 2021;Rodríguez-Pose, 2018).I derive hypotheses on how regional inequalities might shape political trust.
Second, this article uses a dataset with unprecedented global reach in the study of regional inequality effects on attitudes toward national and supranational institutions.Using a measure of political confidence (see 'Methodology'), it analyses a series of multilevel regression models on the basis of data from the European Values Study 5 and the World Values Survey 7 (EVS/WVS, 2022), merged to contextual data for 606 regions in 42 countries worldwide.This dataset is used to compare regional inequality effects in the EU to a global sample excluding EU regions and a pooled sample including all observed regions.To date, regional inequality research is mostly focused on Europe and the United States.In this current study, another essential point of comparison is across the effects of absolute and relative regional income, which are still understudied (see Ejrnaes et al., 2023).
The results from the sample of EU regions suggest that residents in highincome regions tend to trust government more and the EU less.In the global sample of regions outside the EU, people in poor regions tend to trust political institutions more, be they national or supranational.These findings underline that absolute and relative income matter for political trust in government in the EU, but the evidence for economic theory outside the EU is mixed.This depends at least partially on the level of democracy in a country.When examining only the democracies in the global sample, the results are more in line with economic theory, indicating that regional income strengthens political trust.This indicates that democracy needs further studying as a moderator shaping effects of regional inequalities on political trust.These insights have important implications for ongoing debates about regional inequalities, political trust, and the legitimacy of national and international institutions.

Regional inequalities and political trust
The regional inequality literature is firmly embedded in a broader literature on the economic and social consequences of globalisation.Some studies emphasise the economic consequences of globalisation, pointing to the attitudinal consequences of the division of society into winners and losers (e.g., Rodrik, 2018).Other studies stress cultural consequences of globalisation, such as antiimmigration concerns and value shifts, as sources of growing discontent (e.g., Norris & Inglehart, 2019).While some have set up economic and cultural explanations as universal, there is an emerging consensus that both economic and noneconomic concerns, while analytically distinct, can simultaneously contribute to populist and anti-globalist attitudes (Dellmuth et al., 2022;Ejrnaes & Jensen, 2019;Gidron & Hall, 2020;Nicoli & Reinl, 2020;Walter, 2021).
Drawing from economic theory, this section elaborates two distinct arguments about how political trust might affect national and supranational institutions, privileging absolute income and relative income as explanatory factors, respectively.Both logics assume that people are at least partially aware of regional inequalities, either because they experience their own region lagging behind others, often economically, or because of repeated experiences with lacking opportunities.Even if people rarely know enough about domestic or international politics to form independent opinions about political institutions and correctly attribute responsibility, they tend to draw on readily available information.Such information can stem from experiences with the immediate socioeconomic regional environment when developing attitudes toward national or supranational political institutions (Lipps & Schraff, 2021;Nicoli & Reinl, 2020).
The outcome of interest in this article is political trust in national and supranational institutions.Political trust refers to 'citizens' assessments of the core institutions of the polity and entails a positive evaluation of the most relevant attributes that make each political institution trustworthy' (Zmerli, 2014, p. 4887).In the words of Easton (1965, p. 447): 'The presence of trust would mean that members would feel that their own interests would be attended to even if the authorities were exposed to little supervision or scrutiny.' Economic theory assumes people to form opinions based on utilitarian or cost-benefit assessments (Broz et al., 2021;Hooghe & Marks, 2005).Moreover, individuals draw on their position in the economy when developing attitudes toward their government or supranational institutions (Foster & Frieden, 2017;Rodrik, 2018).The environment assumed by economic approaches to regional geography is socially thin.Trade, offshoring, automation, and economic shocks are the dominant environmental characteristics driving the formation of political trust in government.
Inspired by this literature, this paper distinguishes between effects of absolute and relative regional income.The explanatory logic privileging absolute income assumes that people make specific experiences with authorities in high-income regions, which in turn shapes their political attitudes.It raises the expectation that residents of high-income regions are more likely to trust political institutions (Dellmuth, 2021;Lipps & Schraff, 2021).Individuals engage in 'socio-tropic' reasoning, by which they associate regional wealth with adequate labour market opportunities and well-functioning governmental institutions, but also with supranational institutions which have benefited one's region by way of their moral standing, norm diffusion, or technical assistance.
Repeated positive, even if short-lived, experiences with one's region's economic performance continuously nurture feelings of trust.Relatedly, research has shown that government performance and associated economic success (or failure) foster (or discourage) sentiments of political trust (Newton et al., 2018; see also Easton, 1965).From these considerations follows that: Hypothesis 1 (H1): The higher the regional income, the stronger political trust in the region.
An alternative view has it that people perceive regional income in terms of how their own region fares in relation to other regions within the same country.More precisely, individuals living in poorer areas might compare their region's socioeconomic outlook to richer areas, and might thus be more likely to feel left behind and neglected by their government (Rodríguez-Pose, 2018).In turn, regional socioeconomic disadvantage might fuel frustrations with the mainstream political system and weaken political trust in both domestic and supranational institutions (Lipps & Schraff, 2021;Mayne & Katsanidou, 2023).
From this vantage point, individuals are expected to evaluate the relative economic gains and losses of their region.This is in line with research showing that people tend to care about the fairness of economic distribution when developing political attitudes (Brutger & Rathbun, 2021;Fehr & Schmidt, 1999).That perceptions of relative gains matter for opinion formation has also been shown in the context of attitudes towards European integration (Gabel, 1998;Hooghe & Marks, 2005).People usually compare themselves socially both upwards and downwards in the income distribution (Condon & Wichowsky, 2020).
This logic translates into the expectation that residents of relatively poor regions trust political institutions less than residents of richer regions.What is meant by 'relatively poor' lies in the eye of the beholder.It is likely that people evaluating their own region's performance use a typical region in a country as a point of comparison (cf.Mayne & Katsanidou, 2023;McKay et al., 2023).These considerations lead to the expectation that: Hypothesis 2 (H2): The higher the regional income compared to other regions in the same country, the stronger political trust in the region.

Methodology
The hypotheses are assessed by examining a sample of countries in the EU and in other continents, thereby maximising the amount of variation in context.This approach allows for establishing the scope conditions of the argument outside the EU, which has been the predominant testing ground for the attitudinal dimension of regional inequality to date, aside from the US.The analysis is comparative, presenting results separately for the EU and the global sample, and then for the pooled sample.
The analyses are presented separately for national and supranational institutions.While there are no strong theoretical priors as to how regional inequality might matter differentially for trust in national and supranational institutions, it is important to include both measures.Not only are the EU and other supranational institutions increasingly being politicised (Dingwerth et al., 2019).It is also one of the most robust findings in public opinion research that people draw from their trust in national institutions when evaluating institutions they know less about, such as the EU (e.g., Armingeon & Ceka, 2014) or other international institutions (Dellmuth et al., 2022).
The evidence comes from the EVS5 and the WVS7 combined dataset (EVS/ WVS, 2022) and a range of datasets providing contextual variables (see Table 1 for descriptive statistics and Appendix A1 for question wording and operationalisation).The measure of the dependent variable, political trust, relies on the responses to a question about whether people have 'no confidence at all' (0), 'not very much confidence' (1), 'quite a lot of confidence' (2), or 'a great deal of confidence' (3) in a certain political institution.To be sure, this confidence measure has measurement error across samples, as people in different cultures tend to have different understandings of what confidence in political institutions entails (Schneider, 2017).However, the EVS/ WVS (2022) currently is the most encompassing dataset available serving the main purpose of this article, which is the comparison of the effects of regional inequalities on political trust in Europe and on other continents.
In this study, the confidence measure is dichotomised as the theory predicts the likelihood of trusting political institutions, meaning the discrete switch between no confidence (=0) and confidence (1).The items used for the analysis are 'The government' and 'Major regional organisations' (hereinafter referred to as 'supranational institutions').The following supranational institutions are covered: the African Union (AU), the Arab League, the Asia-Pacific Economic Cooperation (APEC), the Association of Southeast Asian Nations (ASEAN), the EU, and Mercosur.
Figures 1 and 2 illustrate the variation in confidence in national and supranational institutions.This first visual inspection shows that there is considerable variation in average regional-level confidence. 1 Discussing the patterns in the EU in more depth, Figure 1 shows that during the observed period 2017-2022, citizens on average have higher confidence in their national government in western Slovakia, northern Hungary, and northern and western Sweden, when compared to the rest of the EU.By contrast, Greece, southern Italy, as well as northern Portugal and Spain, display the lowest levels of confidence in government.With regard to confidence in the EU, Figure 2 suggests a different pattern.In Latvia, Lithuania, Portugal, western Slovakia, and parts  of western France and middle Spain, and Portugal, we see the highest levels of confidence.Conversely, the Czech Republic, Croatia, east Germany, Greece, Slovenia, and northern Sweden and Finland, exhibit the lowest levels of confidence.These findings largely mirror observations using European Social Survey 8 data for trust in national and European parliament (cf.Ejrnaes et al., 2023).
Next, only those countries are retained for which there is data on both political confidence and regional context.Confidence in supranational institutions is measured in these institutions' member states.In non-member states supranational institutions might not be relevant in the same fashion as in member states, implying that including them might undermine the principle of functional equivalence.This selection yields a sample of 42 countries (Table 2).Accordingly, the results for the EU and the global samples in the ensuing section are interpreted for the observed countries only.
The contextual variables are coded at levels 1 or 2 in the Global Administrative Area Database (GADM), i.e., provinces in between the municipal and the country level, amounting to data for 606 regions in the 42 countries (see Appendix A2 for a list of the regional level in each country).These regions are historically grown administrative entities, which tend to be known reference points for people's experiences.All regional data merged to this individual-level survey dataset are measured one year prior to the measurement year in the survey.
To operationalise H1, I include a logarithmic measure regional income, which captures gross national income (GNI) per capita in 1000 USD at purchasing power parity (PPP).Expressing income at PPP is advantageous since it captures people's everyday experiences of what they can afford given their earnings.Log transforming the measure is appropriate in order to treat values at the lower end and upper end of the income per capita distribution in the same way.To operationalise H2, regional disadvantage is captured by taking the percentage difference between the average regional GNI per capita in a country and regional GNI per capita in same country.This measure is expressed in percentages as income levels might vary greatly between regions in the entire sample, implying that regional income differences might be limited in their comparability.The higher the value, the more disadvantaged the region (GDL, 2021).Both income measures are included in the regression models, as different logics behind the effects of regional inequalities might be simultaneously present (cf.Lipps & Schraff, 2021).The analysis includes a set of controls (see Appendix A2 for variable descriptions).Civic skills and socioeconomic status are known to be positively associated with political trust (Newton et al., 2018).Civic skills are proxied by two measures indicating the opportunity to practice such skills (Brady et al., 1995): a dummy variable capturing membership in organisations is coded 1 if a person is a member of a voluntary association and 0 otherwise.Unemployment (1 = unemployed and 0 otherwise) captures negative life experiences which might depress political trust (see also Foster & Frieden, 2017;Newton et al., 2018).Socioeconomic status is measured through education on a three-point scale (cf.Gabel & Scheve, 2007) and through household income by using survey respondents' self-assessment on a 10-point scale (cf.Brady et al., 1995).
Other controls are internal self-efficacy (cf.Bromme & Rothmund, 2021), captured by a quasi-continuous measure of perceived control over one's life, ranging from (0) 'none at all' to (9) 'a great deal'.Urban area measures whether people live in a city above 20,000 inhabitants (1) or not (0) (cf.McKay et al., 2023).Age is a continuous variable and gender is 1 for males and 0 for females.Finally, the analysis recognises that citizens tend to draw from their attitudes toward national institutions when evaluating supranational institutions (de Vries, 2018), by including a measure of the part of confidence in government left unexplained by the model (residual confidence in government).

Empirical analysis
Table 3 presents the results from a series of multilevel logistic regression models. 2Models 1 and 2 analyse confidence in government and in the EU in the European sample.Models 3 and 4 show the results for the global sample.Finally, models 5 and 6 use the entire, pooled sample.It is important to consider these results separately, as the institutions that are the object of political trust differ in terms of their authority and functions.Whereas national governments are functionally similar, this does not equally apply to the supranational institutions.In particular, the EU is more powerful than other supranational institutions (Hooghe & Marks, 2015).There are no strong theoretical priors as to how results might differ across institutions but confidence might  follow different logics in relation to different institutions.The analysis accounts for that by including fixed effects for supranational institutions in Models 4 and 6.The results are discussed for national government first, and then for supranational institutions.
In line with H1, regional income is positively associated with confidence in government in the European sample (model 1).Table 3 reports average marginal effects, which express the likely effects of the independent variables on the probability to have confidence.The probability of having confidence increases by 0.191 as regional income increases by one unit.In other words, the predicted probabilities of having confidence and their 95 per cent confidence intervals are 0.018 [0.002;0.033] in the poorest regions and 0.491 [0.433;0.548] in the richest regions.This finding underlines previous economic theories of regional inequalities and political discontent emphasising the importance of regional income (e.g., Lipps & Schraff, 2021;Mayne & Katsanidou, 2023;Schraff, 2019).
Counter to the expectation in H1, however, regional income is negatively associated with confidence in supranational institutions (model 2).If a region is doing well, respondents might be more skeptical of any potential supranational interference in regional development that is perceived as already working (cf.Dijkstra et al., 2020).What is more, in the global sample that excludes the EU countries, regional income is negatively associated with confidence in both national and supranational institutions (models 3 and 4).This finding might have to do with GNI per capita having a wider range in the pooled sample.From a global perspective, poorer regions are characterised by more confidence in both national and supranational institutions.This latter insight is corroborated in the pooled samples (models 5 and 6), which provide a stronger test than the estimations in the split samples.
Consistent with H2, regional disadvantage is negatively associated with confidence in government in the EU sample (model 1).For every percentage increase in the income difference between a region's income and the average regional income in a country, the probability of having confidence increases by 0.170.Put differently, the predicted probabilities of having confidence are 0.389 [0.346;0.432] in the regional income decile that is farthest away from the average regional income in a country, and 0.023 [−0.001;0.048]for the regions that have or are very close to the average income in the country.This finding from the EU sample indicates that if one's own region is economically lagging, compared to the average regional income in one's country, then that leads to political discontent, here expressed in lower trust in government (e.g., Rodríguez-Pose, 2018).
Again, the sign of the coefficient changes in relation to trust in the EU, which is not in line with the expectation in H2 (model 2).This indicates that people in more disadvantaged regions tend to trust the EU more.Disappointment in government might lead people to put more confidence in the EU (Lipps & Schraff, 2021).This finding is corroborated in the global (model 4) and pooled (model 6) samples.
Discussing the controls, membership in voluntary organisations is positively associated with trust across the board (cf.Newton et al., 2018).Unemployment only appears to depress trust in government in the EU sample (cf.Foster & Frieden, 2017).Gender appears to matter only in relation to trust in the EU, where men trust the EU less than women.Education does not appear to consistently matter.Regarding age, younger people appear to trust government less and supranational institutions more.Feeling internal self-efficacy is positively associated with confidence across the board (cf.Lipps & Schraff, 2021).Urbanisation only appears to matter in the global sample, where urban residents tend to distrust government (cf.McKay et al., 2023).Residual confidence in government is consistently positively related to confidence in supranational institutions (cf.Armingeon & Ceka, 2014).
These results in Table 3 were subjected to a number of robustness checks.To begin with, it was tested whether the effects of regional income and regional disadvantage are conditional upon education.Higher-skilled people are more likely to have a global outlook as they belong to the winners from globalisation (Broz et al., 2021).However, education does not appear to consistently moderate the effects of regional income or disadvantage (Appendix C1 and C2).
Moreover, closed autocracies -Jordan, Myanmar, Thailand, and Vietnam (see Boese et al., 2022 and Appendix A2)were excluded.Under authoritarian rule people might have other rationales when developing legitimacy beliefs than under democratic rule (Zeng, 2015) and people might selfcensor when answering survey questions about political institutions.However, this analysis yields similar results (Appendix C3).When excluding all thirteen autocraciesimplying that Hungary is dropped from the EU samplethe coefficient of regional income in model 3 turns positively significant, which underpins H1 (Appendix C4).This indicates that residents of highincome regions in democracies both within and outside the EU tend to trust government more, but that this effect does not hold under autocratic rule.
Regional income and regional disadvantage were included separately in the regression models.While it is warranted to test if both logics co-occur, it might be that their correlation at r = 0.280 (N = 59103) causes multicollinearity.While the effect of regional income remains unchanged (Appendix C5), the coefficients of regional disadvantage lose statistical significance, implying that those should be interpreted with caution (Appendix C6).Unless we factor in how rich a region is, we cannot consistently estimate whether people, irrespective of absolute income, also engage in comparisons with the economic performance of other regions.However, the concern regarding multicollinearity is weak given that the Variance Inflation Factor is low at 1.22 on average and below 2 for all covariates, which is very low (Wooldridge, 2015, ch. 3).This confirms that a regression model of trust is underspecified if it does not include both absolute and relative income (see also Lipps & Schraff, 2021).
A final robustness check concerns the use of an alternative model specification using regional disadvantage captured by taking the percentage difference between the richest (not the average, as reported in Table 3) region's GNI per capita in a country and regional GNI per capita in same country.The higher the value of this alternative measure, the more disadvantaged is a region when compared to the best-off region (GDL, 2021).This measure recognises that people might know the most well-off region in their country and will use it as a reference point to develop opinions (cf.Lipps & Schraff, 2021).The results are robust in the global and pooled samples, but not in the EU sample (Table C7).This indicates that in the EU, regional disadvantage in relation to the typical or average region matters but not disadvantage relative to the richest region.
In summary, three main findings stand out.First, the evidence underlines that people living in richer regionsboth in absolute and in relative termsin the EU sample trust their national government more and the EU less.This finding from the sample of EU regions is in line with both H1 and H2.Second, in the global sample, people living in poor regions trust their government more, which is corroborated in the pooled sample.These findings run counter to H1 and H2.Third, this deviating result in the global sample is at least partially due to the inclusion of autocracies in the sample; when excluding autocratic countries, regional income is positively associated with trust in government, which is in line with H1.

Discussion
Moving beyond previous research, this article has analysed the effects of regional inequalities on political trust in a global context.Using individualand regional-level data for 606 regions in 42 countries in and outside the EU, the results are evidence for economic theory only in the context of trust in government in the sample of 22 EU countries.By contrast, the results are more variegated in the examined 20 countries outside the EU.
Discussing the results in more detail, both absolute and relative regional income measures are found to matter for trust in government in the EU sample, in line with H1 and H2, thereby corroborating previous research on the economic determinants of political discontent in Europe (e.g., Broz et al., 2021;Mayne & Katsanidou, 2023;Rodríguez-Pose, 2018).These findings also speak to ongoing discussions about the consequences of European integration for regional inequalities (Beckfield, 2019) and the well-being of residents in lagging regions (Dellmuth, 2021).Regional inequalities are pervasive and have been a source of political discontent in the EU.The complex relationships between regional inequalities, quality of living, and public opinion make regional inequalities an important policy challenge.
However, the evidence for both H1 and H2 is mixed in the global sample.Indeed, the spatial dynamics in global public opinion samples might be fundamentally different from Europe (McKay et al., 2023).In the global sample of 20 countries studied here, people living in poorer regions appear to trust both national and supranational institutions more.However, when examining only the democracies included in the global sample, regional income isas expected in H1positively related to trust in government.Future research could usefully theorise and examine how the effect of regional inequalities on political attitudes might be conditioned by democracy.
These results have three broader implications.First, they speak to the debate on how trust in domestic political institutions relates to trust in supranational or international institutions (e.g., Armingeon & Ceka, 2014;de Vries, 2018).A growing body of work suggests that trust in domestic and European institutions are positively related, as people know too little about the EU to evaluate it independently from national institutions (cf.Muñoz et al., 2011;Harteveld et al., 2013;Schlipphak, 2015).By contrast, macro-level data suggest that regional inequality contributes to decreasing trust in government while increasing trust in the EU (Lipps & Schraff, 2021).This might be in part explained by people having hopes that the EU could do more in the future, even in economically difficult times (Simpson & Loveless, 2017).In line with this, the results in this current study also suggest that regional disadvantage decreases trust in government while increasing trust in the EU, but only in the EU sample.
Second, on the dependent variable side, the question arises how regional inequalities are relevant not only for political trust but also for perceptions of political legitimacy.When individuals perceive a political institution to be legitimate, they assume that its actions 'are desirable, proper, or appropriate within some socially constructed system of norms, values, beliefs, and definitions' (Suchman, 1995, p. 574).While political trust and legitimacy are not synonymous and their complex relationship has been discussed, there is much overlap (cf.Dellmuth & Tallberg, 2023, ch. 2;Thomassen et al., 2017).Substantial literatures on legitimacy beliefs in cognitive psychology and International Relations, which have used political trust or confidence as a dependent variable, have examined the effects of socioeconomic context on legitimacy perceptions (see Tyler, 2006, andTallberg &Zürn, 2019, for overviews).For example, previous studies have focused on how domestic social structures such as cultural norms shape legitimacy beliefs (e.g., Bernstein, 2011;Scholte, 2018).However, scholars have yet to examine how regional inequality shapes legitimacy beliefs towards domestic and international institutions.
Third and finally, this article has privileged a global-scale inquiry using data from a specific survey wave to test two theories in the global context.Future research could usefully generate global-scale panel data, although this might imply compromises on which covariates can be included (McKay et al., 2023).More generally, socioeconomic change rather than levels of income are important for political attitudes (Häusermann et al., 2023;Vasilopoulou & Talving, 2023).A longitudinal analysis would allow for getting at period or trend effects shaping the linkages between regional or income inequalities and political discontent, as previous research has demonstrated in the European context (Dijkstra et al., 2020) and in the context of industrialised democracies (Engler & Weisstanner, 2021).This would also help to gain traction on the complex processes involved when individuals in different parts of the world interpret the regional context to develop political trust.

Notes
1.Both measures of confidence in government and confidence in supranational institutions overlap to some extent but are empirically distinct (Pearson's r = 0.357; N = 60,549).2. The main model is written as follows: Pr(y ijk = 1 | u jk ) = H(X ijk β+u jk +z k ), which expresses the probability that respondent i living in region j in country k has political confidence.X are vectors for the independent variables, conditional on a set of random effects u jk for regions and z k for countries.The choice to include random intercepts at the country level is underlined by estimations of the intra-class correlation (ICC) in the pooled sample (ICC = 0.309 [0.274;0.346]for national government and ICC = 0.244 [0.215;0.277]for supranational institutions, as these indicate a moderate correlation between predicted confidence levels for individuals residing in the same region and country.

Figure 1 .
Figure 1.Confidence in national government.Note: Data from EVS/WVS (2022).Regional post-stratification weights calculated by using census data provided by the Minnesota Population Center (2020) using the methodology inDellmuth and Jonsson (2023) (see also Appendix B).

Figure 2 .
Figure 2. Confidence in major supranational institutions.Note: Data from EVS/WVS (2022).Regional post-stratification weights calculated by using census data provided by the Minnesota Population Center (2020) using the methodology inDellmuth and Jonsson (2023) (see also Appendix B).

Table 2 .
Overview of supranational institutions and countries.

Table 3 .
Logistic regression models of confidence.
Note: Dependent variable is confidence (1 = a great deal or quite a lot, 0 = none or not very much).Models 4 and 6 include fixed effects for supranational institutions.Constant included but not reported for the sake of brevity.Weighted using the 1000-equalisation weight in order not to arbitrarily give large samples greater weight (EVS/WVS, 2022).Average marginal effects with robust standard errors in parentheses.*p < .05,**p < .01,***p < .001.