|
?:linkContext
|
-
Given tensions between the need for representative democracy and the growing use of referendums , and tensions between representative democracy and the use of referendums, it is important that we better understand popular attitudes about democracy as related to referendums. We use multilevel models to estimate responses across 25 countries to questions about referendums and democracy that were included in the European Social Survey (ESS) in 2012-2013. We examine popular preferences for the use of referendums. We pay particular attention to the gap between how important people see referendum use in principle, and how they see it actually practiced. We test competing explanations of support for referendums. One sees support for referendum democracy as a function of cognitive mobilization and a sign of engaged citizens. A different argument locates support for referendums in the views of disaffected people who are at the periphery of the regular political process. We find more support for the idea that it is this latter group, individuals who are disaffected, who are most supportive of referendums as a democratic principle and who are more disappointed with the practice of referendum democracy in their country. This article makes several contributions. Our crossnational approach allows us to test for national context as well as test individual-level hypotheses. We incorporate national level contextual factors such as income inequality and political corruption alongside individuallevel attributes as we model support for referendum use and "referendum disappointment." We thus raise two points of wider relevance. First, for most people in the countries we examine, majority rule (via referendums) is a component of their conception of democracy in principle. Second, the contrast between how referendums are experienced in practice and how they are expected in principle suggests many citizens are disappointed with democracy in practice as it relates to using referendums. Use of referendums falls short, sometimes far short, of this popular expectation about democracy. In other words, attitudes about referendum use provide us with a window into wider concerns people have about their political system and the workings of democracy. The use of referendums varies substantially across the globe (see , for discussion on the range and variety of direct democracy; . Switzerland represents the extreme case: the popular veto was adopted as early as 1831 and nearly 600 national referendums and citizen initiatives have appeared on Swiss ballot since 1848 . Denmark, Italy, Ireland and Slovenia also provide examples of regular use of referendums , and Iceland, Great Britain, France, Spain, Slovakia, and Latvia all have histories that include the occasional use of national referendums. Referendum use is rare in countries such as the Czech Republic, Belgium, Finland, Norway and the Netherlands, however recurring use of referendums in neighboring countries on European integration and other matters, and referendum use at the subnational level likely means that even residents of places where national referendums are not used nonetheless have a sense of what referendums are. illustrates responses to two ESS 6 questions, one asking about how important it is to "democracy in general" that people have a say through referendums, and another that asks their perception of the extent to which people in their country actually have a say through referendums. Responses to the first item suggest that people in numerous countries included in Round 6 of the ESS (see placed substantial value on referendums as being an important principle of democracy. When asked to rate on a 0-10 scale how important it was to "democracy in general…that citizens have the final say on the most important political issues by voting on them directly in referendums," the modal response was 10 in nearly every country where the ESS was conducted (see Appendix for full question wording; . Attitudes about referendums being important for democracy (in principle), and perceptions of having a final say in their country on important issues via referendums (in practice). Source: ESS Round 6 (2012). for a study using related World Values Survey items measuring perceptions of how essential referendums are to democracy, see . The country-level averages for attitudes about the importance of referendums as a general principle of democracy are plotted with the shaded bars in . This level of support is also found elsewhere. Survey data from Canada ), Australia, New Zealand , and the United States also show high levels of popular support for using referendums. Responses to the second item suggest much greater variation in perceptions that referendums give people in their country a say. People were asked to rate on a 0-10 scale, "to what extent do you think…citizens in have the final say by voting…directly in referendums." These responses are displayed in with the dark bars. The use of referendums in practice varies quite widely across these countries, and so to do perceptions that people feel they have the final say via referendums. Thus, while expectations of having the 'final say' by referendum is high, and has little variance across these countries, it appears many people do not perceive they experience the referendum democracy they expect. What do popular expectations about referendums and democracy actually reflect, and who are those that expect referendums but are disappointed? One set of arguments sees support for referendum use as consistent with the cognitive mobilization thesis . defines cognitive mobilization as the spread of education, access to mass media and low-cost information in advanced democracies, where people are increasing politically engaged. From this perspective, more people are now capable of dealing "with the complexities of politics" on their own . This process is somewhat similar to Norris' concept of the "critical citizen" . Signing petitions, boycotting, demonstrating, and protesting are forms of participation that links to a politically engaged (rather than dutybased) form of citizenship. have found some evidence from New Zealand, Canada, and Switzerland consistent with the idea that use of referendums is supported by those more politically engaged. It follows that one might view citizens' expectations for referendums as a healthy sign of political engagement, or even the reflection of people being interested in participating more actively, and directly, in setting policy. We have less clear expectations about how the more politically engaged or interested may perceive referendum use in practice, as this may be contingent both on how interested a person is in politics and how much referendums are actually used in a respondent's country. If engagement and interest drive expectations that referendums are important, a person with a great interest in politics may evaluate practice more positively where referendums are used more frequently. Conversely, if the po-litically engaged expect referendums and perceive their use is not sufficient, they may be more disappointed with referendum democracy. An alternative view is that popular support for referendums indicates disillusion with social conditions and with established parties and representative democracy. Findings from suggest that support for direct democracy in Germany was strongest among those who felt excluded from establishment politics and those at the 'periphery' of politics. Right-wing populist parties that exploit dissatisfaction with status quo democracy (e.g., UKIP in Britain or the FN in France) champion the use of referendums to give people a greater voice, and to provide an end run around establishment parties and politicians . found that the call for referendum use motivated some people to vote for populist parties in Belgium, Germany and the Netherlands. Given this, we expect that right-populist voters will be more supportive of giving citizens a direct say via referendums. We might also expect that the expectation of having a say through referendums could be more appealing to people who reside in countries where institutions do a poor job processing conflict and allocating resources. We thus expect poor governance, poor economic performance, and tensions associated with income inequality to affect expectations for referendums and perceptions of their use. Hibbing and Theiss-Morse (2002) offer a view of support for direct democracy related to this latter argument (also see . Support for referendums as an important democratic principle may thus reflect public disdain for the perceived inappropriate influence that "special" interests have over elected officials in a representative democracy. From this perspective, enthusiasm for referendums may reflect a mistrust of incumbent parties and establishment politics, rather than greater political engagement. Relatedly, perceptions that referendums are not used in practice could reflect discontent among the disaffected who expect another avenue of voice than is available via representative processes they may not trust. Much existing literature considers support for referendum use in general terms, without modeling how interest in referendums relates to a person's expectations about how democracy should work or perceptions of how it does work. Nonetheless, there are studies that examine support for using referendums, mostly focusing on individual countries. found greater support for direct democracy in Finland among people with less education, less information, and among those who felt unrepresented. found greater support for referendums among the politically disaffected and those exposed to tabloid-style news in the Netherlands, and emphasized the role of cynicism in their subsequent work . also found lower educated Dutch respondents were more likely to prefer direct and stealth democracy over representative democracy. concluded that the greatest support for direct democracy in Germany can be found among those "at the periphery of politics," including the less informed, those less interested, and supporters of extremist parties. Anderson and Goodyear-Grant (2010) also found greater support for direct democracy among less-informed Canadians, which they explained as reflecting that "political sophisticates" have greater confidence in government and are better able to identify the dangers that referendums may present. These findings are consistent with the idea that those who are more disillusioned, disaffected, and less politically engaged may be more likely to view referendums as important to democracy. In contrast, found those more interested in politics were more supportive of referendums in three of six countries studied, and those participating in elections were more supportive in two of those. Another study from the early 2000s across 16 countries found that in nearly all countries, support for using referendums was greatest among those who thought it was important for people to have more opportunities to participate, and, in many countries, among those who were more politically engaged. These results are consistent with the idea that those more engaged are more likely to see referendums as important to democracy. The literature provides us these two contrasting expectations about how people assess the importance of having referendums, each finding some support in previous research. We suggest we might find a deeper understanding about how people view referendums and democracy by simultaneously considering their attitudes about the importance of referendums to democracy, and their perceptions about whether referendums are giving people a say. We are interested in the importance that people place on using referendums (again the shaded bars in ), the extent to which they perceive that referendums give people a say in their country (represented by the dark bars in ), and the difference between these. This difference, we suggest, represents how much a person who expects having a say via referendums might be disappointed by perceiving those opportunities are limited in their country. In the next section, we develop hypotheses that consider country-level factors and individual-level factors that may explain variation in: 1) attitudes about how important people view referendums as part of democracy; 2) perceptions of the extent that referendums give people a say in their country; and 3) the difference between how important people view referendums as part of democracy and their perceptions that referendums give people in their country a say. In the analysis that follows, we consider country-level factors and individuallevel factors that might affect attitudes about, and perceptions of, referendum democracy. We expect attitudes about democracy and referendums, and perceptions of having a say via referendums, are shaped both by characteristics of individuals, and by the social and political context of the country in which the individual resides. We test hypotheses about: 1) factors that structure attitudes about referendums as a general principle of democracy; 2) factors structuring perceptions of referendum use in practice; and 3) factors that explain the gap (or disappointment) between respondents' expectations about referendums as a principle and their perceptions of referendum use in practice. We thus estimate three models, one estimating attitudes about referendums and democracy, one estimating perceptions of referendum use, and one estimating the gap between these first two measures. Popular expectations about the role of referendums in democracy, and perceptions of the use of referendums, are not likely to occur in a vacuum. Rather, discontent and disaffection with social and political conditions might make referendums a seemingly appealing alternative to status quo representative democracy. We expect people in countries with greater corruption, higher unemployment, and greater income inequality to be more disappointed with referendum democracy, while people in countries using referendums more frequently are expected to be less disappointed. The former factors may reflect a context of disaffection that drives expectations for use of referendums, and disillusionment about political practices generally. Use of referendums, we expect, may mitigate some of this. There are few studies that examine country-level factors that might affect attitudes about referendum use (for an exception see Schuck & de Vreese, 2015, on support for referendums on EU integration). Our expectation here is that having a say through referendums could be more appealing to people who reside in countries where institutions do a poor job processing conflict and allocating resources. One difficulty in operationalizing this argument comes in selecting a measure of governance. A key factor here is public corruption, which we see as the antithesis of good governance. It has been shown to erode political trust, and it is associated with pessimism about the performance of democracy in a country . We expect higher levels of (perceived) corruption to correspond with greater expectations for referendums. Our models are specified with the Transparency International measure of perceptions of public corruption. Poor economic performance may breed similar discontent, creating additional demands for referendum use. We expect higher unemployment to be associated weaker support for established institutions and thus with heightened expectations for referendums. Economic inequality, likewise, where a wealthy elite has disproportionate influence over a country's political institutions, may also increase support for using referendums. provides evidence that greater income inequality produces greater political inequality, and Dotti contend that inequality is linked to political cynicism in Europe. argue that inequality can breed resentment toward traditional democratic practices, undermine democratic attitudes, and decrease satisfaction with democracy. Higher levels of economic inequality, then, could lead to greater support for the use of referendums. Unemployment is measured with World Bank data, and inequality is represented by a CIA gini index. We account for party system disproportionality as a national-level control, since this may also affect perceptions of democratic institutions . Further, we expect that people in countries that used referendums more frequently will be more familiar with the process, and thus more likely to accept and expect referendums as a regular feature of democracy. Likewise, we expect people in countries with greater referendum use to be more likely to perceive that citizens in their country have a say via referendums. Our measure of referendum use is a count of national-level referendums used in a country, calculated from Qvortrup (2014). These nations differ in terms of institutional features that affect how consequential various referendums may be on policy . Our measure does not account for this qualitative aspect of referendum use. Rather, our count measure acts as a proxy for experiences respondents' might have with referendums, rather than an indicator of policy consequences of referendums. The engaged citizen thesis as related to referendum democracy has us expect that those scoring higher on participation, education, and interest are more likely to expect referendums as an important part of democracy. Their perceptions of referendum use, and thus potential for disappointment, may depend on how often referendums are used. Highly interested respondents may find their expectations for referendum use met where referendums are used more. This hypothesis flows from the idea that the politically interested have a greater expectation of regularly influencing policy via referendums, while having the capacity to do so. We expect those with greater political interest to be more likely to expect having a say via referendums, and to be disappointed if they perceive they are not able to. The engaged citizen idea also leads us to expect people with greater education, and those who are generally engaged with participating in politics, to be more likely to expect to decide mat-ters via referendums. Interest is measured with a four category self-reported item ranking interest in politics. We use a battery of ESS questions to build a six-item index of political participation (or engagement) that reflects working for a party, for an organization, displaying a badge or sticker, signing petitions, protesting, and boycotting. Education is measured with a 7-category ordinal measure. The rival disaffected citizen thesis as related to referendum democracy has us expect that those who distrust politicians, those who did not supported the main party in government, and populist party supporters would be more inclined to agree that referendums were important. Trust is measured on a 0-10 scale where 10 is most trusting of politicians. Supporters of a governing party, and populist party supporters, are represented with dichotomous measures coded by the authors. We may also see broader kinds of social disaffection. It could be those who perceive themselves as social "have-nots" see referendums as way in which their voice may be heard. The ESS included two items that tap such sentiments. One asked respondents if they felt their "place in society" was at the bottom (versus the top), and another asked people if they felt they were part of a group "discriminated against in this country." 1 We expect people who see themselves in these terms may be more likely to view referendums as important to democracy. Social place is a self-reported 0-10 scale where 0 is "bottom" and 10 is "top of society." Perceptions of discrimination is a dichotomous measure. As noted, the ESS contained questions about referendums that measured two different concepts: attitudes about referendums as a principle of democracy, and attitudes about how much the respondent perceived that people in the respondent's country had a say via referendums in practice. Given the second item asks about perceptions, we expect a somewhat different pattern of results with the item asking about referendum use in practice than with the item asking about expecting referendums as a democratic principle. Specifically, we expect that people who might be more sanguine about status quo politics to be less likely to perceive that people in their country do not have a say in general, and thus less likely to agree that people do not have a say via referendums. As such, people who supported the main parties in government, those who trusted politicians, and those who viewed themselves in the upper echelon of society, could be more likely to respond positively when asked if people have a "final say" by referendums. We expand these hypotheses to considering the gap between attitudes about referendums as a principle of democracy, and perceptions or referendum use (or, referendum disappointment). Country-level factors that we expect to correspond with disaffection (e.g., corruption, inequality) are expected to predict greater disappointment. We expect more disappointment among people who live in a country where referendums are used infrequently. We also expect greater referendum disappointment among those who we hypothesized would be more likely to view referendums as an important democratic principle. As noted in the previous section, a potential effect of referendum use may be contingent on how interested a person is in politics. If interest is associated with placing more importance on referendums as part of democracy, we may find less referendum disappointment among those with high interest in countries where referendums are used more. We examine this with postestimation analysis in Stata. Finally, we include age, gender, and frequency of attending religious services as individual-level control variables. We also control for ideology, a 0-10 scale where high scores are self-reported "right" ideology. The Appendix provides details on the variables and codings. The ESS 6 (2012) covered 29 countries, but we omit four cases (Albania, Kosovo, Russia, and Ukraine) that were arguably less than fully democratic. The ESS 6 included two items asking about referendum democracy; one asked respondents to consider how important referendums were for "democracy in general," and a second asked about how the respondent perceived people in the respondent's country had a say through referendums. This provides us three dependent variables. The first measures a respondent's view of how important it was to democracy in general that citizens have the "final say" on important matters of policy "by voting on them directly in referendums." This is our first dependent variablethe perceived importance of referendums as a general principle of democracy. This item ranges from 0 to 10 with the highest scores reflecting the attitude that referendums are extremely important for democracy (the mean is 8.2). The second ESS question asked respondent's their perception of whether or not people had the "final say" on important issues in their country "by voting on them directly in referendums." This is our second dependent variable, also ranging from 0-10, with higher scores reflecting a person thought citizens in their country had the final say in practice via referendums. The mean for perceptions of referendum use in practice is 5.0, much lower than the mean for attitudes about referendums being an important principle of democracy. These two items are only modestly correlated (0.14). We use these two items to construct a third dependent variable that represents a respondent's disappointment with referendum democracy, as related their expectations and perceptions about the use of referendums. It is created by subtracting a person's score on the first item (perceptions of referendums as a principle of democracy) from their score on the second item (perceptions of referendums in practice). A respondent scoring high on this referendum disappointment measure would be someone who thought it was important for people to have a say via referendums, but viewed their country as a place where this was unlikely. This measure ranges from −10 to 10 (or 0 to 10, depending on specification, see Appendix). This third dependent variable is the gap between what a person views as a general principle of democracy and what they see as its practice, more than simply a measure of support for using referendums. We employed multi-level models to estimate these attitudes about referendums and democracy. Baseline random intercepts models were also estimated to calculate the proportion of variance explained by country-level versus individual-level factors . reports results of a model estimating attitudes about referendums being an important democratic principle, and a model estimating perceptions of referendum use in practice. reports models estimating referendum disappointed (the gap between expectations about referendums as a democratic principle, and perceptions of how referendums are used). We can make a number of points based on the results. First, the major source of variation in attitudes about referendums as a democratic principle is at the individual rather than the country level. In some ways this is not much of a surprise given what we have seen in : there is little variation in opinions cross-nationally when it comes to judging the importance of referendums. The Intra-class correlation (ICC) calculated for a baseline version of Model 1 in estimates that just 4.5% of variance in attitudes about referendums being a general principle of democracy is explained by variance across countries. In contrast, a baseline version of Model 2 in illustrates that a modest amount of variation in perceptions of referendum use in practice (15.6%) is explained by country-level differences. Second, at the individual-level we find the politically and socially disaffected-those who voted for right populist parties, who had low trust in politicians, and who perceived they were discriminated against-were more likely to agree that having a say via referendums was important as a democratic principle, and were more likely to see citizens not having a say via referendums in practice. Those who viewed themselves at the bottom of society were also more likely to say referendums were not used enough in practice. Supporters of governing parties, conversely, were less likely to see referendums as important. Third, we find some mixed support for the expectation that engaged citizens view referendums as important to democracy. Citizens who were more engaged in terms of political participation were more likely to view referendums as important in principle, and lacking in practice. Furthermore, people more interested in politics were also more likely to see referendums as important to .320 .581 Notes: ** = significant at p. < .01; * = at p. < .05. (two-tail). DV in Model 1 is response to question asking how important it is for democracy that citizens have final say via referendums (0-10, with 10 = extremely important). DV in Model 2 is response to question asking how much citizens in r's country have the final say via referendums (0-10, with 10 = agree completely). Estimated with weights. Standard errors in parentheses. R 2 are . .677 .704 Notes: ** = significant at p. < .01; * = at p. < .05. (two-tail). High scores on DVs represent larger gap between r's report of how important it is for democracy that citizens have final say via referendums, and r's report of how much citizens in r's country have the final say via referendums. DV in Model 1 includes negative values (−10 to 10, with 10 = most disappointed). DV in Model 2 sets negative values at 0. Estimated with weights. Standard errors in parentheses. R 2 are . Source: ESS Round 6 (2012). democracy, and to potentially find them lacking in practice. However, support for referendums in principle was greater among those with lower levels of education, and those with more education were less likely to think people were having a say via referendum use in practice. As noted above, 15.6% of variance in perceptions of how often referendums are used in practice is due to country-level differences, and our country-level variables explain 58% of this variance. Our measures of corruption and inequality are both associated with the perception that citizens were not having "a final say" via referendums. We find unemployment associated with expectations for referendum use in principle, but not in practice. Further, the number of referendums used in a country had a positive effect on whether citizens viewed referendums as an important part of democracy, and also had a positive effect on perceptions that people actually had a say via referendums. The proportionality of the party system was not associated with support for referendums as a democratic principle or views of its use in practice. reports estimates of our measure of disappointment with referendum democracy, in terms of the gap between perceptions of referendums being important for democracy, and perceptions of how much they are used in the respondent's country. Higher values on this dependent variable reflect greater disappointment over not having a say via referendums, relative to the importance a respondent places on referendums. Model 3A includes negative values on our disappointment measure, while Model 3B has all negative values set at zero. The mean value for this measure across the sample on the former measure is 3.3, and 3.1 on the latter, implying some disappointment is present. Moreover, 25% of the sample had a 5-point difference or greater between their attitudes about the importance of referendums as a democratic principle and their perceptions of referendum use in practice on the measure bounded at 0-10. A 5-point difference suggests a sizable amount of disappointment. documents that those we assume to be disaffected were more disappointed in democracy as it relates to the use of referendums. Right-populist voters, those who did not support a governing party, those who did not trust politicians, those who felt discriminated against, and hose who rated themselves at the "bottom" of society were each more disappointed-in that they viewed referendum use as important while also being less likely to perceive having a say via referendums practice. Tests using the disappointment measure also produce some results that may be seen as consistent with the engaged citizen thesis; those who scoring higher on our participation measure, the higher educated, and those more interested in politics, respectively, scored higher on this form of referendum disappointment. These respondents may have expected more from referendums relative to what they perceived as occurring in practice. Women, those on the political left, and those who with lower (or no) attendance at religious services (included in the models as controls) were also more disappointed. The ICC for models in demonstrates a modest amount of variation in referendum disappointment (12.6% for Model 3A, 13.7% for Model 3B) is explained by country-level factors. Inequality and corruption are key here. Income inequality corresponded with higher levels of disappointment, and we find greater disappointment in countries with higher levels of public corruption. plots postestimation predictions of referendum disappointment from Model 3B for key county-level variables. It illustrates that a respondent in a country with the highest values on corruption (Bulgaria and Italy) would score 1.5 points higher on the disappointment measure than a demographically identical respondent in the least corrupt countries (Finland and Denmark). Inequality has a similar, but slightly smaller effect. A person in a country with high income inequality (e.g., Portugal) would score a full point higher than someone in the least unequal country (Sweden). also illustrates all else equal, a respondent in a country with the highest use of national level referendums (Switzerland) would score 1.1 points less disappointed than a person in a country that never had held more than one national referendum (e.g., . and display the estimated substantive effects of our key individual-level variables, with plotting estimates for variables associated with the disaffection hypotheses, and plotting effects for variables associated with the engaged citizen hypotheses. These figures suggest the substantive magnitude of the relationship between some variables representing disaffection, and referendum disappointment , is more substantial than the relationships between most of the engagement variables ( ) and referendum disappointment. This is particularly evident with trust. All else equal, a respondent with no trust in politicians (nearly one-fifth of respondents) is estimated to score nearly 3 points higher on referendum disappointment than a respondent who completely trusted politicians. Supporters of right-populist parties ("rwp voter" in the figure) score nearly a point higher on disappointment than supporters of a party in government. Considering that these are additive models, the additive effects of disaffection-right-wing populism, distrust, and rating one's self at the bottom of society-are substantial. Some results here are consistent with the idea that this referendum disappointment could reflect engaged citizens who desire greater opportunities to participate in policymaking. However, the only noteworthy substantive effect in is for our six-item index of participation. demonstrates that the ranks of those scoring significantly higher on referendum disappointment include those with more education, more political interest, and those who participate more. On this latter measure, a respondent scoring 6 (highest) is estimated to be nearly 1.5 points higher on disappointment than someone scoring at 0, however only 20% of respondents score higher than 1 on this measure. The estimated effect of political interest illustrated in , moreover, is less than half a point. We did consider that effects of engagement may be contingent on how often referendums were used. illustrates that those with more political interest placed value on having referendums as part of democracy, while also perceiving that people were not having a say via referendums. In a model not reported here, and in postestimation analysis of Model 3B, we found a significant interaction between political interest and referendum use, where those with more political interest had less referendum disappointment if they lived where more referendums were used. However, a highly interested person only had significantly less referendum disappointment than a person with low interest when referendum use was set at a level equal to Switzerland. Much research on attitudes about referendum use has been grounded in single country studies which do not account for context. Many previous studies have examined general attitudes about the use of referendums (on a specific policy matter) rather than examining attitudes about referendums in terms of expectations people have about democracy. We found broad support for the idea that referendums are important to democracy. Limited variation in this expectation was best explained by how people were oriented to the political system. Those people we assume as politically disaffected-people who supported populist parties, those who did not support a party in government, those who distrusted politicians, and those who felt themselves on the bottom of society-were most likely to say that referendums were important to their conception of democracy. By contrast, those who supported a party in government, and people more trusting of politicians, felt less strongly that democracy requires voters have the final say via referendums. Although we find some evidence that markers of political engagement were associated with viewing referendums as important to democracy, indicators of political disaffection appear more consequential. That noted, our main goal was to examine the gap between this widely held expectation that democracy involves voters having a say through referendums, and their perceptions of how referendum democracy was giving people a say in practice. We find a substantial amount of disappointment regarding the use of referendums. It is not only individual-level markers of political disaffection-right-wing populist voting, distrust of politicians, and seeing one's self at the bottom of society-that most substantially predict this form of referendum disappointment. Country-level forces such as income inequality and political corruption also corresponded with people who viewed referendums as important while perceiving that people were not being given a say through referendums. These results suggest that experiences with inequality and corruption may be factors that lead some people to look beyond established parties and representative democracy for referendums as alternative modes of political influence. Particularly in countries where things were not going as well, respondents were more likely to report a gap between their expectation that democracy should provide a say through referendums, and their perceptions of having opportunities to actually have that say. For many people who are disaffected, democracy in practice was not living up to this widely held democratic principle of allowing people to have a direct say via referendums. This gap between the widely held democratic expectations of referendums, and variable perceptions of not having a say via referendums in practice, may provide space that allows political entrepreneurs to exploit frustrations disaffected people have with representative democracy. Political consequences of this may be seen in growing support for populist parties, in demands for the Brexit referendum, for a second Brexit vote, and in calls for additional referendums that have occurred since. All of this suggests that "outsider" movements may tap into disaffection, and potentially gain additional support, with calls for increased use of referendums-calls that may not simply be calculated to set policy, but to manufacture a sense that people will finally have their say. That said, we must end with some caveats. Referendums are frequently "top down," in that they are placed on the ballot by incumbent politicians. Although we find corruption and income inequality associated with perceptions that people are not having a say though referendums, and with our referendum disappointment measure, these factors could depress trust in democratic institutions generally, including trust in top down referendums. Additionally, we cannot tell from these data which types of referendums it is that people value, nor which types of referendums they perceive as providing a meaningful say. Populists in power Political instability and economic growth Journal of Economic Growth Why are highly informed citizens skeptical of referenda? Electoral Studies Political institutions and satisfaction with democracy American Political Science Review Corruption, political allegiances, and attitudes toward government in contemporary democracies American Journal of Political Science Direct democracy and its critics: Support for direct democracy and 'stealth democracy' in Finland Right-wing populist party supporters: Dissatisfied but not direct democrats European Journal of Political Research Enraged or engaged? Preferences for direct citizen participation in affluent democracies Political Research Quarterly Education and support for representative, direct and stealth democracy Electoral Studies Direct democracy: The politics of initiative, referendum, and recall Cognitive mobilization and partisan dealignment in advanced industrial democracies The Journal of Politics Partisan mobilization, cognitive mobilization, and the changing American electorate Electoral Studies The good citizen: How a younger generation is reshaping American politics Public opinion and direct democracy Journal of Democracy Popular support for direct democracy Party Politics Increasingly unequal? The economic crisis, social inequalities and trust in the European Parliament in 20 European countries European Journal of Political Research ESS round 6: European Social Survey round 6 data European Social Survey How Europeans view and evaluate democracy Conceptualizing and measuring the quality of democracy: The citizens' perspective Politics and Governance The referendum in representative democracies To join or not to join: Three Nordic referendums on membership in the European Union Economic inequality, winner-loser gap, and satisfaction with democracy Electoral Studies Stealth democracy: Americans' beliefs about how government should work When Europe matters: The impact of political information on voting behavior in EU referendums Journal of Elections, Public Opinion and Parties Occurrence and policy consequences of referendums: A theoretical model and empirical evidence Journal of Theoretical Politics Cognitive mobilization and European identity Comparative Politics The politics of direct democracy: Referendums in global perspective The democratic effect of direct democracy America Political Science Review Referendum democracy: Citizens, elites and deliberation in referendum campaigns The populist zeitgeist. Government and Opposition Populist radical right parties in Europe Critical citizens: Global support for democratic government Populism in Western Europe: Comparing Belgium, Germany and The Netherlands Referendums around the world: The continued growth of direct democracy Inflation, unemployment, and government collapse Comparative Political Studies Parties and the expansion of direct democracy: Who benefits? Party Politics Public support for referendums: The role of the media. West European Politics Public support for referendums in Europe: A cross-national comparison in 21 countries Electoral Studies Referendums around the world: The continued growth of direct democracy Modeled variance in two-level models Sociological Methods & Research Economic inequality and democratic political engagement American Journal of Political Science Modeling multilevel data structures American Journal of Political Science The missing dimension of democracy: Institutional patterns in 25 EU Member States between European Union Politics Who is willing to participate? Dissatisfied democrats, stealth democrats and populists in the United Kingdom European Journal of Political Research To what extend you think…the following statement applies in [country]. Citizens in [country] have the final say on the most important political issues by voting on them directly in referendums Calculated as votedir-votedirc. −10 through 10, high scores reflect greater gap between greater regard for referendums as a democratic principle and perceptions of referendum use in practice Disappointment with referendums ESS round 6: European Social Survey round 6 data European Social Survey Modeled variance in two-level models Sociological Methods & Research Referendums around the world: The continued growth of direct democracy Perceptions of Referendums and Democracy: The Referendum Disappointment Gap 27 June 2019 Shaun Bowler shaun.bowler@ucr.edu Department of Political Science University of California 92521 Riverside CA USA Todd Donovan todd.donovan@wwu.edu Department of Political Science Western Washington University 98225 Bellingham WA USA Perceptions of Referendums and Democracy: The Referendum Disappointment Gap 2183-2463 27 June 2019 10.17645/pag.v7i2.1874 Submitted: 12 December 2018 | Accepted: 18 March 2019 GROBID - A machine learning software for extracting information from scholarly documents democracy direct democracy inequality political disaffection populism public opinion referendums We examine the gap between perceptions of seeing referendums as an important democratic principle, versus perceiving how referendums are used in practice. We term this the "referendum disappointment" gap. We find support for referendums as a democratic principle is strongest among those most disaffected from the political system, and that the disaffected are more likely to perceive they are not given a say via referendums. We also find context-specific effects. Disappointment was greater in countries with higher corruption and income inequality. We also find higher disappointment among right-populist voters, those who distrusted politicians, and among people who viewed themselves at the bottom of society. Overall, these patterns reflect disappointment with democracy among sections of society who have a sense of not being heard that conflicts with how they expect democracy should work in principle. Introduction Given tensions between the need for representative democracy and the growing use of referendums , and tensions between representative democracy and the use of referendums, it is important that we better understand popular attitudes about democracy as related to referendums. We use multilevel models to estimate responses across 25 countries to questions about referendums and democracy that were included in the European Social Survey (ESS) in 2012-2013. We examine popular preferences for the use of referendums. We pay particular attention to the gap between how important people see referendum use in principle, and how they see it actually practiced. We test competing explanations of support for referendums. One sees support for referendum democracy as a function of cognitive mobilization and a sign of engaged citizens. A different argument locates support for referendums in the views of disaffected people who are at the periphery of the regular political process. We find more support for the idea that it is this latter group, individuals who are disaffected, who are most supportive of referendums as a democratic principle and who are more disappointed with the practice of referendum democracy in their country. This article makes several contributions. Our crossnational approach allows us to test for national context as well as test individual-level hypotheses. We incorporate national level contextual factors such as income inequality and political corruption alongside individuallevel attributes as we model support for referendum use and "referendum disappointment." We thus raise two points of wider relevance. First, for most people in the countries we examine, majority rule (via referendums) is a component of their conception of democracy in principle. Second, the contrast between how referendums are experienced in practice and how they are expected in principle suggests many citizens are disappointed with democracy in practice as it relates to using referendums. Use of referendums falls short, sometimes far short, of this popular expectation about democracy. In other words, attitudes about referendum use provide us with a window into wider concerns people have about their political system and the workings of democracy. Popular Support for Referendums The use of referendums varies substantially across the globe (see LeDuc, 2003 , for discussion on the range and variety of direct democracy; Vatter & Bernauer, 2009) . Switzerland represents the extreme case: the popular veto was adopted as early as 1831 and nearly 600 national referendums and citizen initiatives have appeared on Swiss ballot since 1848 (Serdült, 2013) . Denmark, Italy, Ireland and Slovenia also provide examples of regular use of referendums , and Iceland, Great Britain, France, Spain, Slovakia, and Latvia all have histories that include the occasional use of national referendums. Referendum use is rare in countries such as the Czech Republic, Belgium, Finland, Norway and the Netherlands, however recurring use of referendums in neighboring countries on European integration (Hobolt, 2005) and other matters, and referendum use at the subnational level (Scarrow, 1999) likely means that even residents of places where national referendums are not used nonetheless have a sense of what referendums are. Figure 1 illustrates responses to two ESS 6 questions, one asking about how important it is to "democracy in general" that people have a say through referendums, and another that asks their perception of the extent to which people in their country actually have a say through referendums. Responses to the first item suggest that people in numerous countries included in Round 6 of the ESS (see Ferrin & Kriesi, 2016) placed substantial value on referendums as being an important principle of democracy. When asked to rate on a 0-10 scale how important it was to "democracy in general…that citizens have the final say on the most important political issues by voting on them directly in referendums," the modal response was 10 in nearly every country where the ESS was conducted (see Appendix for full question wording; Figure 1 . Attitudes about referendums being important for democracy (in principle), and perceptions of having a final say in their country on important issues via referendums (in practice). Source: ESS Round 6 (2012). for a study using related World Values Survey items measuring perceptions of how essential referendums are to democracy, see Fuchs & Roller, 2018) . The country-level averages for attitudes about the importance of referendums as a general principle of democracy are plotted with the shaded bars in Figure 1 . This level of support is also found elsewhere. Survey data from Canada (Mendelsohn & Parkin, 2001 ), Australia, New Zealand (Donovan & Karp, 2006) , and the United States (Cronin, 1989; Gilljam, Pesonen, & Listhaug, 1998) also show high levels of popular support for using referendums. Responses to the second item suggest much greater variation in perceptions that referendums give people in their country a say. People were asked to rate on a 0-10 scale, "to what extent do you think…citizens in [country] have the final say by voting…directly in referendums." These responses are displayed in Figure 1 with the dark bars. The use of referendums in practice varies quite widely across these countries, and so to do perceptions that people feel they have the final say via referendums. Thus, while expectations of having the 'final say' by referendum is high, and has little variance across these countries, it appears many people do not perceive they experience the referendum democracy they expect. What do popular expectations about referendums and democracy actually reflect, and who are those that expect referendums but are disappointed? One set of arguments (Donovan & Karp, 2006) sees support for referendum use as consistent with the cognitive mobilization thesis (Dalton, 1984; Inglehart, 1970) . Dalton (1984) defines cognitive mobilization as the spread of education, access to mass media and low-cost information in advanced democracies, where people are increasing politically engaged. From this perspective, more people are now capable of dealing "with the complexities of politics" on their own (Dalton, 2007, p. 276) . This process is somewhat similar to Norris' concept of the "critical citizen" (Norris, 1999) . Signing petitions, boycotting, demonstrating, and protesting are forms of participation that Dalton (2008) links to a politically engaged (rather than dutybased) form of citizenship. Donovan and Karp (2006, p. 679) have found some evidence from New Zealand, Canada, and Switzerland consistent with the idea that use of referendums is supported by those more politically engaged. It follows that one might view citizens' expectations for referendums as a healthy sign of political engagement, or even the reflection of people being interested in participating more actively, and directly, in setting policy. We have less clear expectations about how the more politically engaged or interested may perceive referendum use in practice, as this may be contingent both on how interested a person is in politics and how much referendums are actually used in a respondent's country. If engagement and interest drive expectations that referendums are important, a person with a great interest in politics may evaluate practice more positively where referendums are used more frequently. Conversely, if the po-litically engaged expect referendums and perceive their use is not sufficient, they may be more disappointed with referendum democracy. An alternative view is that popular support for referendums indicates disillusion with social conditions and with established parties and representative democracy. Findings from Dalton, Burklin and Drummond (2001) suggest that support for direct democracy in Germany was strongest among those who felt excluded from establishment politics and those at the 'periphery' of politics. Right-wing populist parties that exploit dissatisfaction with status quo democracy (e.g., UKIP in Britain or the FN in France) champion the use of referendums to give people a greater voice, and to provide an end run around establishment parties and politicians (Albertazzi & McDonnell, 2015; Bowler, Denemark, Donovan, & McDonnell, 2017; Mudde, 2004 Mudde, , 2007 . Pauwels (2014, p. 159) found that the call for referendum use motivated some people to vote for populist parties in Belgium, Germany and the Netherlands. Given this, we expect that right-populist voters will be more supportive of giving citizens a direct say via referendums. We might also expect that the expectation of having a say through referendums could be more appealing to people who reside in countries where institutions do a poor job processing conflict and allocating resources. We thus expect poor governance, poor economic performance, and tensions associated with income inequality to affect expectations for referendums and perceptions of their use. Hibbing and Theiss-Morse (2002) offer a view of support for direct democracy related to this latter argument (also see Webb, 2013) . Support for referendums as an important democratic principle may thus reflect public disdain for the perceived inappropriate influence that "special" interests have over elected officials in a representative democracy. From this perspective, enthusiasm for referendums may reflect a mistrust of incumbent parties and establishment politics, rather than greater political engagement. Relatedly, perceptions that referendums are not used in practice could reflect discontent among the disaffected who expect another avenue of voice than is available via representative processes they may not trust. Much existing literature considers support for referendum use in general terms, without modeling how interest in referendums relates to a person's expectations about how democracy should work or perceptions of how it does work. Nonetheless, there are studies that examine support for using referendums, mostly focusing on individual countries. Bengtsson and Mattila (2009) found greater support for direct democracy in Finland among people with less education, less information, and among those who felt unrepresented. Schuck and De Vreese (2011) found greater support for referendums among the politically disaffected and those exposed to tabloid-style news in the Netherlands, and emphasized the role of cynicism in their subsequent work (Schuck & De Vreese, 2015) . Coffé and Michels (2014) also found lower educated Dutch respondents were more likely to prefer direct and stealth democracy over representative democracy. Dalton et al. (2001) concluded that the greatest support for direct democracy in Germany can be found among those "at the periphery of politics," including the less informed, those less interested, and supporters of extremist parties. Anderson and Goodyear-Grant (2010) also found greater support for direct democracy among less-informed Canadians, which they explained as reflecting that "political sophisticates" have greater confidence in government and are better able to identify the dangers that referendums may present. These findings are consistent with the idea that those who are more disillusioned, disaffected, and less politically engaged may be more likely to view referendums as important to democracy. In contrast, Donovan and Karp (2006) found those more interested in politics were more supportive of referendums in three of six countries studied, and those participating in elections were more supportive in two of those. Another study from the early 2000s across 16 countries (Bowler, Donovan, & Karp, 2007) found that in nearly all countries, support for using referendums was greatest among those who thought it was important for people to have more opportunities to participate, and, in many countries, among those who were more politically engaged. These results are consistent with the idea that those more engaged are more likely to see referendums as important to democracy. The literature provides us these two contrasting expectations about how people assess the importance of having referendums, each finding some support in previous research. We suggest we might find a deeper understanding about how people view referendums and democracy by simultaneously considering their attitudes about the importance of referendums to democracy, and their perceptions about whether referendums are giving people a say. We are interested in the importance that people place on using referendums (again the shaded bars in Figure 1 ), the extent to which they perceive that referendums give people a say in their country (represented by the dark bars in Figure 1 ), and the difference between these. This difference, we suggest, represents how much a person who expects having a say via referendums might be disappointed by perceiving those opportunities are limited in their country. In the next section, we develop hypotheses that consider country-level factors and individual-level factors that may explain variation in: 1) attitudes about how important people view referendums as part of democracy; 2) perceptions of the extent that referendums give people a say in their country; and 3) the difference between how important people view referendums as part of democracy and their perceptions that referendums give people in their country a say. In the analysis that follows, we consider country-level factors and individuallevel factors that might affect attitudes about, and perceptions of, referendum democracy. Hypothesis and Model Specification We expect attitudes about democracy and referendums, and perceptions of having a say via referendums, are shaped both by characteristics of individuals, and by the social and political context of the country in which the individual resides. We test hypotheses about: 1) factors that structure attitudes about referendums as a general principle of democracy; 2) factors structuring perceptions of referendum use in practice; and 3) factors that explain the gap (or disappointment) between respondents' expectations about referendums as a principle and their perceptions of referendum use in practice. We thus estimate three models, one estimating attitudes about referendums and democracy, one estimating perceptions of referendum use, and one estimating the gap between these first two measures. Country-Level Hypotheses Popular expectations about the role of referendums in democracy, and perceptions of the use of referendums, are not likely to occur in a vacuum. Rather, discontent and disaffection with social and political conditions might make referendums a seemingly appealing alternative to status quo representative democracy. We expect people in countries with greater corruption, higher unemployment, and greater income inequality to be more disappointed with referendum democracy, while people in countries using referendums more frequently are expected to be less disappointed. The former factors may reflect a context of disaffection that drives expectations for use of referendums, and disillusionment about political practices generally. Use of referendums, we expect, may mitigate some of this. There are few studies that examine country-level factors that might affect attitudes about referendum use (for an exception see Schuck & de Vreese, 2015, on support for referendums on EU integration). Our expectation here is that having a say through referendums could be more appealing to people who reside in countries where institutions do a poor job processing conflict and allocating resources. One difficulty in operationalizing this argument comes in selecting a measure of governance. A key factor here is public corruption, which we see as the antithesis of good governance. It has been shown to erode political trust, and it is associated with pessimism about the performance of democracy in a country (Anderson & Tverdova, 2003) . We expect higher levels of (perceived) corruption to correspond with greater expectations for referendums. Our models are specified with the Transparency International measure of perceptions of public corruption. Poor economic performance may breed similar discontent, creating additional demands for referendum use. We expect higher unemployment to be associated weaker support for established institutions (Alesina, Özler, Roubini, & Swagel, 1996; Robertson, 1983) and thus with heightened expectations for referendums. Economic inequality, likewise, where a wealthy elite has disproportionate influence over a country's political institutions, may also increase support for using referendums. Solt (2008) provides evidence that greater income inequality produces greater political inequality, and Dotti Sani and Magistro (2016) contend that inequality is linked to political cynicism in Europe. Han and Chang (2016, p. 94) argue that inequality can breed resentment toward traditional democratic practices, undermine democratic attitudes, and decrease satisfaction with democracy. Higher levels of economic inequality, then, could lead to greater support for the use of referendums. Unemployment is measured with World Bank data, and inequality is represented by a CIA gini index. We account for party system disproportionality as a national-level control, since this may also affect perceptions of democratic institutions (Anderson & Guillory, 1997) . Further, we expect that people in countries that used referendums more frequently will be more familiar with the process, and thus more likely to accept and expect referendums as a regular feature of democracy. Likewise, we expect people in countries with greater referendum use to be more likely to perceive that citizens in their country have a say via referendums. Our measure of referendum use is a count of national-level referendums used in a country, calculated from Qvortrup (2014). These nations differ in terms of institutional features that affect how consequential various referendums may be on policy (Hug, 2004; Leemann & Wasserfallen, 2016) . Our measure does not account for this qualitative aspect of referendum use. Rather, our count measure acts as a proxy for experiences respondents' might have with referendums, rather than an indicator of policy consequences of referendums. Individual-Level Hypotheses The engaged citizen thesis as related to referendum democracy has us expect that those scoring higher on participation, education, and interest are more likely to expect referendums as an important part of democracy. Their perceptions of referendum use, and thus potential for disappointment, may depend on how often referendums are used. Highly interested respondents may find their expectations for referendum use met where referendums are used more. This hypothesis flows from the idea that the politically interested have a greater expectation of regularly influencing policy via referendums, while having the capacity to do so. We expect those with greater political interest to be more likely to expect having a say via referendums, and to be disappointed if they perceive they are not able to. The engaged citizen idea also leads us to expect people with greater education, and those who are generally engaged with participating in politics, to be more likely to expect to decide mat-ters via referendums. Interest is measured with a four category self-reported item ranking interest in politics. We use a battery of ESS questions to build a six-item index of political participation (or engagement) that reflects working for a party, for an organization, displaying a badge or sticker, signing petitions, protesting, and boycotting. Education is measured with a 7-category ordinal measure. The rival disaffected citizen thesis as related to referendum democracy has us expect that those who distrust politicians, those who did not supported the main party in government, and populist party supporters would be more inclined to agree that referendums were important. Trust is measured on a 0-10 scale where 10 is most trusting of politicians. Supporters of a governing party, and populist party supporters, are represented with dichotomous measures coded by the authors. We may also see broader kinds of social disaffection. It could be those who perceive themselves as social "have-nots" see referendums as way in which their voice may be heard. The ESS included two items that tap such sentiments. One asked respondents if they felt their "place in society" was at the bottom (versus the top), and another asked people if they felt they were part of a group "discriminated against in this country." 1 We expect people who see themselves in these terms may be more likely to view referendums as important to democracy. Social place is a self-reported 0-10 scale where 0 is "bottom" and 10 is "top of society." Perceptions of discrimination is a dichotomous measure. As noted, the ESS contained questions about referendums that measured two different concepts: attitudes about referendums as a principle of democracy, and attitudes about how much the respondent perceived that people in the respondent's country had a say via referendums in practice. Given the second item asks about perceptions, we expect a somewhat different pattern of results with the item asking about referendum use in practice than with the item asking about expecting referendums as a democratic principle. Specifically, we expect that people who might be more sanguine about status quo politics to be less likely to perceive that people in their country do not have a say in general, and thus less likely to agree that people do not have a say via referendums. As such, people who supported the main parties in government, those who trusted politicians, and those who viewed themselves in the upper echelon of society, could be more likely to respond positively when asked if people have a "final say" by referendums. We expand these hypotheses to considering the gap between attitudes about referendums as a principle of democracy, and perceptions or referendum use (or, referendum disappointment). Country-level factors that we expect to correspond with disaffection (e.g., corruption, inequality) are expected to predict greater disappointment. We expect more disappointment among people who live in a country where referendums are used infrequently. We also expect greater referendum disappointment among those who we hypothesized would be more likely to view referendums as an important democratic principle. As noted in the previous section, a potential effect of referendum use may be contingent on how interested a person is in politics. If interest is associated with placing more importance on referendums as part of democracy, we may find less referendum disappointment among those with high interest in countries where referendums are used more. We examine this with postestimation analysis in Stata. Finally, we include age, gender, and frequency of attending religious services as individual-level control variables. We also control for ideology, a 0-10 scale where high scores are self-reported "right" ideology. The Appendix provides details on the variables and codings. Measuring Attitudes about Referendums in Principle and in Practice The ESS 6 (2012) covered 29 countries, but we omit four cases (Albania, Kosovo, Russia, and Ukraine) that were arguably less than fully democratic. The ESS 6 included two items asking about referendum democracy; one asked respondents to consider how important referendums were for "democracy in general," and a second asked about how the respondent perceived people in the respondent's country had a say through referendums. This provides us three dependent variables. The first measures a respondent's view of how important it was to democracy in general that citizens have the "final say" on important matters of policy "by voting on them directly in referendums." This is our first dependent variablethe perceived importance of referendums as a general principle of democracy. This item ranges from 0 to 10 with the highest scores reflecting the attitude that referendums are extremely important for democracy (the mean is 8.2). The second ESS question asked respondent's their perception of whether or not people had the "final say" on important issues in their country "by voting on them directly in referendums." This is our second dependent variable, also ranging from 0-10, with higher scores reflecting a person thought citizens in their country had the final say in practice via referendums. The mean for perceptions of referendum use in practice is 5.0, much lower than the mean for attitudes about referendums being an important principle of democracy. These two items are only modestly correlated (0.14). We use these two items to construct a third dependent variable that represents a respondent's disappointment with referendum democracy, as related their expectations and perceptions about the use of referendums. It is created by subtracting a person's score on the first item (perceptions of referendums as a principle of democracy) from their score on the second item (perceptions of referendums in practice). A respondent scoring high on this referendum disappointment measure would be someone who thought it was important for people to have a say via referendums, but viewed their country as a place where this was unlikely. This measure ranges from −10 to 10 (or 0 to 10, depending on specification, see Appendix). This third dependent variable is the gap between what a person views as a general principle of democracy and what they see as its practice, more than simply a measure of support for using referendums. We employed multi-level models to estimate these attitudes about referendums and democracy. Baseline random intercepts models were also estimated to calculate the proportion of variance explained by country-level versus individual-level factors (Steenbergen & Jones, 2002) . Table 1 reports results of a model estimating attitudes about referendums being an important democratic principle, and a model estimating perceptions of referendum use in practice. Table 2 reports models estimating referendum disappointed (the gap between expectations about referendums as a democratic principle, and perceptions of how referendums are used). Results We can make a number of points based on the results. First, the major source of variation in attitudes about referendums as a democratic principle is at the individual rather than the country level. In some ways this is not much of a surprise given what we have seen in Figure 1 : there is little variation in opinions cross-nationally when it comes to judging the importance of referendums. The Intra-class correlation (ICC) calculated for a baseline version of Model 1 in Table 1 estimates that just 4.5% of variance in attitudes about referendums being a general principle of democracy is explained by variance across countries. In contrast, a baseline version of Model 2 in Table 1 illustrates that a modest amount of variation in perceptions of referendum use in practice (15.6%) is explained by country-level differences. Second, at the individual-level we find the politically and socially disaffected-those who voted for right populist parties, who had low trust in politicians, and who perceived they were discriminated against-were more likely to agree that having a say via referendums was important as a democratic principle, and were more likely to see citizens not having a say via referendums in practice. Those who viewed themselves at the bottom of society were also more likely to say referendums were not used enough in practice. Supporters of governing parties, conversely, were less likely to see referendums as important. Third, we find some mixed support for the expectation that engaged citizens view referendums as important to democracy. Citizens who were more engaged in terms of political participation were more likely to view referendums as important in principle, and lacking in practice. Furthermore, people more interested in politics were also more likely to see referendums as important to .320 .581 Notes: ** = significant at p. < .01; * = at p. < .05. (two-tail). DV in Model 1 is response to question asking how important it is for democracy that citizens have final say via referendums (0-10, with 10 = extremely important). DV in Model 2 is response to question asking how much citizens in r's country have the final say via referendums (0-10, with 10 = agree completely). Estimated with weights. Standard errors in parentheses. R 2 are . .677 .704 Notes: ** = significant at p. < .01; * = at p. < .05. (two-tail). High scores on DVs represent larger gap between r's report of how important it is for democracy that citizens have final say via referendums, and r's report of how much citizens in r's country have the final say via referendums. DV in Model 1 includes negative values (−10 to 10, with 10 = most disappointed). DV in Model 2 sets negative values at 0. Estimated with weights. Standard errors in parentheses. R 2 are Snijders & Boske (1994) . Source: ESS Round 6 (2012). democracy, and to potentially find them lacking in practice. However, support for referendums in principle was greater among those with lower levels of education, and those with more education were less likely to think people were having a say via referendum use in practice. As noted above, 15.6% of variance in perceptions of how often referendums are used in practice is due to country-level differences, and our country-level variables explain 58% of this variance. Our measures of corruption and inequality are both associated with the perception that citizens were not having "a final say" via referendums. We find unemployment associated with expectations for referendum use in principle, but not in practice. Further, the number of referendums used in a country had a positive effect on whether citizens viewed referendums as an important part of democracy, and also had a positive effect on perceptions that people actually had a say via referendums. The proportionality of the party system was not associated with support for referendums as a democratic principle or views of its use in practice. Table 2 reports estimates of our measure of disappointment with referendum democracy, in terms of the gap between perceptions of referendums being important for democracy, and perceptions of how much they are used in the respondent's country. Higher values on this dependent variable reflect greater disappointment over not having a say via referendums, relative to the importance a respondent places on referendums. Model 3A includes negative values on our disappointment measure, while Model 3B has all negative values set at zero. The mean value for this measure across the sample on the former measure is 3.3, and 3.1 on the latter, implying some disappointment is present. Moreover, 25% of the sample had a 5-point difference or greater between their attitudes about the importance of referendums as a democratic principle and their perceptions of referendum use in practice on the measure bounded at 0-10. A 5-point difference suggests a sizable amount of disappointment. Table 2 documents that those we assume to be disaffected were more disappointed in democracy as it relates to the use of referendums. Right-populist voters, those who did not support a governing party, those who did not trust politicians, those who felt discriminated against, and hose who rated themselves at the "bottom" of society were each more disappointed-in that they viewed referendum use as important while also being less likely to perceive having a say via referendums practice. Tests using the disappointment measure also produce some results that may be seen as consistent with the engaged citizen thesis; those who scoring higher on our participation measure, the higher educated, and those more interested in politics, respectively, scored higher on this form of referendum disappointment. These respondents may have expected more from referendums relative to what they perceived as occurring in practice. Women, those on the political left, and those who with lower (or no) attendance at religious services (included in the models as controls) were also more disappointed. The ICC for models in Table 2 demonstrates a modest amount of variation in referendum disappointment (12.6% for Model 3A, 13.7% for Model 3B) is explained by country-level factors. Inequality and corruption are key here. Income inequality corresponded with higher levels of disappointment, and we find greater disappointment in countries with higher levels of public corruption. Figure 2 plots postestimation predictions of referendum disappointment from Model 3B (Table 2) for key county-level variables. It illustrates that a respondent in a country with the highest values on corruption (Bulgaria and Italy) would score 1.5 points higher on the disappointment measure than a demographically identical respondent in the least corrupt countries (Finland and Denmark). Inequality has a similar, but slightly smaller effect. A person in a country with high income inequality (e.g., Portugal) would score a full point higher than someone in the least unequal country (Sweden). Figure 2 also illustrates all else equal, a respondent in a country with the highest use of national level referendums (Switzerland) would score 1.1 points less disappointed than a person in a country that never had held more than one national referendum (e.g., Belgium, Finland, Israel) . Figure 3 and Figure 4 display the estimated substantive effects of our key individual-level variables, with Figure 3 plotting estimates for variables associated with the disaffection hypotheses, and Figure 4 plotting effects for variables associated with the engaged citizen hypotheses. These figures suggest the substantive magnitude of the relationship between some variables representing disaffection, and referendum disappointment (Figure 3) , is more substantial than the relationships between most of the engagement variables ( Figure 4 ) and referendum disappointment. This is particularly evident with trust. All else equal, a respondent with no trust in politicians (nearly one-fifth of respondents) is estimated to score nearly 3 points higher on referendum disappointment than a respondent who completely trusted politicians. Supporters of right-populist parties ("rwp voter" in the figure) score nearly a point higher on disappointment than supporters of a party in government. Considering that these are additive models, the additive effects of disaffection-right-wing populism, distrust, and rating one's self at the bottom of society-are substantial. Some results here are consistent with the idea that this referendum disappointment could reflect engaged citizens who desire greater opportunities to participate in policymaking. However, the only noteworthy substantive effect in Figure 4 is for our six-item index of participation. Table 2 demonstrates that the ranks of those scoring significantly higher on referendum disappointment include those with more education, more political interest, and those who participate more. On this latter measure, a respondent scoring 6 (highest) is estimated to be nearly 1.5 points higher on disappointment than someone scoring at 0, however only 20% of respondents score higher than 1 on this measure. The estimated effect of political interest illustrated in Figure 4 , moreover, is less than half a point. We did consider that effects of engagement may be contingent on how often referendums were used. Table 1 illustrates that those with more political interest placed value on having referendums as part of democracy, while also perceiving that people were not having a say via referendums. In a model not reported here, and in postestimation analysis of Model 3B, we found a significant interaction between political interest and referendum use, where those with more political interest had less referendum disappointment if they lived where more referendums were used. However, a highly interested person only had significantly less referendum disappointment than a person with low interest when referendum use was set at a level equal to Switzerland. Conclusion Much research on attitudes about referendum use has been grounded in single country studies which do not account for context. Many previous studies have examined general attitudes about the use of referendums (on a specific policy matter) rather than examining attitudes about referendums in terms of expectations people have about democracy. We found broad support for the idea that referendums are important to democracy. Limited variation in this expectation was best explained by how people were oriented to the political system. Those people we assume as politically disaffected-people who supported populist parties, those who did not support a party in government, those who distrusted politicians, and those who felt themselves on the bottom of society-were most likely to say that referendums were important to their conception of democracy. By contrast, those who supported a party in government, and people more trusting of politicians, felt less strongly that democracy requires voters have the final say via referendums. Although we find some evidence that markers of political engagement were associated with viewing referendums as important to democracy, indicators of political disaffection appear more consequential. That noted, our main goal was to examine the gap between this widely held expectation that democracy involves voters having a say through referendums, and their perceptions of how referendum democracy was giving people a say in practice. We find a substantial amount of disappointment regarding the use of referendums. It is not only individual-level markers of political disaffection-right-wing populist voting, distrust of politicians, and seeing one's self at the bottom of society-that most substantially predict this form of referendum disappointment. Country-level forces such as income inequality and political corruption also corresponded with people who viewed referendums as important while perceiving that people were not being given a say through referendums. These results suggest that experiences with inequality and corruption may be factors that lead some people to look beyond established parties and representative democracy for referendums as alternative modes of political influence. Particularly in countries where things were not going as well, respondents were more likely to report a gap between their expectation that democracy should provide a say through referendums, and their perceptions of having opportunities to actually have that say. For many people who are disaffected, democracy in practice was not living up to this widely held democratic principle of allowing people to have a direct say via referendums. This gap between the widely held democratic expectations of referendums, and variable perceptions of not having a say via referendums in practice, may provide space that allows political entrepreneurs to exploit frustrations disaffected people have with representative democracy. Political consequences of this may be seen in growing support for populist parties, in demands for the Brexit referendum, for a second Brexit vote, and in calls for additional referendums that have occurred since. All of this suggests that "outsider" movements may tap into disaffection, and potentially gain additional support, with calls for increased use of referendums-calls that may not simply be calculated to set policy, but to manufacture a sense that people will finally have their say. That said, we must end with some caveats. Referendums are frequently "top down," in that they are placed on the ballot by incumbent politicians. Although we find corruption and income inequality associated with perceptions that people are not having a say though referendums, and with our referendum disappointment measure, these factors could depress trust in democratic institutions generally, including trust in top down referendums. Additionally, we cannot tell from these data which types of referendums it is that people value, nor which types of referendums they perceive as providing a meaningful say. Figure 2 . 2 Referendum disappointment: Post-estimation predicted magnitude of country-level variables. Notes: From Model 3B. Bars represent 95% confidence intervals. Range is 0-10. Figure 3 . 3 Referendum disappointment: Post-estimation predicted magnitude of individual-level disaffection variables. Notes: From Model 3B. Bars represent 95% confidence intervals. Range is 0-10. Figure 4 . 4 Referendum disappointment: Post-estimation predicted magnitude of individual-level engagement variables. Notes: From Model 3B. Bars represent 95% confidence intervals. Range is 0-10. Table 1 . 1 Attitudes about referendums as a principle of democracy, and perceptions of referendum use in practice. Model 1 Model 2 Refs as dem. principle Ref use in practice Table 2 . 2 Disappointment with democracy, as related to referendum use. Model 1 Model 2 Disappointment Disappointment Politics and Governance, 2019, Volume 7, Issue 2, Pages 227-241 This included discrimination based on race, nationality, religion, language, ethnicity, ethnic group, age, sexual orientation, disability, and (the modal category) "other grounds." Conflict of Interests The authors declare no conflict of interests. About the Authors Shaun Bowler is Professor of Political Science at University of California Riverside, and Graduate Division Dean. His research interests include comparative electoral systems and voting behavior. His work examines the relationship between institutional arrangements and voter choice in a variety of settings ranging from the Republic of Ireland to California's initiative process. He is the author of Demanding Choices: Opinion Voting and Direct Democracy with Todd Donovan and has published in the American Political Science Review, American Journal of Political Science, and Journal of Politics. Todd Donovan is Professor of Political Science at Western Washington University. His research examines political behavior, representation, and electoral institutions, as well as public opinion, and direct democracy. He has authored, co-authored, and co-edited a dozen books, and has also advised media and state and federal courts on election matters. Appendix Note on robustness As demonstrated in Table 2 , our results are not contingent on how negative values of the disappointment measure are coded. Model 1 in Table 2 estimates disappointment where values range from -10 to 10. As a robustness check, Model 2 estimates the same model with all negative values set at 0. We consider negative values as reflecting a lack of disappointment. Any value at 0 or lower may also reflect a lack of disappointment and/or indifference. Model 2 in Table 2 demonstrates that the results are not affected by how negative values are coded. Moreover, in results not reported here we re-estimated models reported in Table 1 and Table 2 with cases limited to European Union nations, with a different measure of income inequality (from the European Union), and with a different measure of government performance (a World Bank governance measure, rather than the TI corruption rating). Regardless of the permutation of cases and measures, our substantive results were unchanged. Results are also consistent whether or not Switzerland (an outlier on referendum use) is included or excluded from the analysis. Our findings are also robust across a range of different model specifications including case selection (e.g., excluding Switzerland), excluding some individual level measures (e.g., religious attendance-which we include as a control), and controlling for "former Communist" states. Participation. 1 point each if respondent worked for "a party or action group" (wrkprty), "another organization" (wrkorg), displayed a badge or sticker (badge), signed a petition (sgnptit), boycotted product (bctprd), or took part in a lawful protest (pbldmn) in the last 12 months. 0-6 scale. 0 = no participation, 6 = all forms of participation. R's place in society (plinsoc). Respondent's self-placement when asked "Your place in society?" 1-10 scale; 0 = "bottom of our society"…10 = "top of our society." Age (agea). Age of respondent in years. Education (eisced). Highest level of education, ES-ISCED. 1-7 scale; 1 = less than lower secondary…7 = higher tertiary education. Interest (polintr). Interest in politics, recoded 1 = low, 4 = high. Female (gndr). 1 = female, 0 = male. Experienced discrimination (dscrgrp). Respondent self-reported perception of being "member of a group discriminated against in this country." 1 = yes, 0 = no. Attend religious services (rlgatnd). How often attend religious services apart from special occasions? 1-7 scale, 1 = every day…7 = never. Country level variables Number of referendums. Calculated from reports in Party System Disproportionality. Gallagher index for recent election result. Ranges from .73 (Denmark) to 17.6 (France). Unemployment. 2012 levels, from World Bank. Ranges from 3.2 (Norway) to 25.2 (Spain). Gini index (inequality). From CIA database. Ranges from .23 (Sweden) to .45 (Bulgaria). Corruption. Transparency International Corruption Perception Index. Scale ranging from 41 (Bulgaria), 44 (Italy) to 90 (Finland, Denmark). Transposed so higher scores reflect greater corruption. Dependent variables (original question wording) Referendums in principle (votedir). "Thinking generally rather than about [country], how important do you think it is for democracy in general that citizens have the final say on the most important political issues by voting on they directly in referendums" 0 (not at all important for democracy in general) through 10 (extremely important for democracy in general). Populists in power D Albertazzi D Mcdonnell Routledge New York, NY Political instability and economic growth A Alesina S Özler N Roubini P Swagel Journal of Economic Growth 1 2 Why are highly informed citizens skeptical of referenda? Electoral C Anderson E Goodyear-Grant Studies 29 2 Political institutions and satisfaction with democracy C Anderson C Guillory American Political Science Review 91 1 Corruption, political allegiances, and attitudes toward government in contemporary democracies C Anderson Y Tverdova American Journal of Political Science 47 1 Direct democracy and its critics: Support for direct democracy and 'stealth democracy' in Finland A Bengtsson M Mattila 32 West European Politics Right-wing populist party supporters: Dissatisfied but not direct democrats S Bowler D Denemark T Donovan D Mcdonnell European Journal of Political Research 56 1 Enraged or engaged? Preferences for direct citizen participation in affluent democracies S Bowler T Donovan J A Karp Political Research Quarterly 60 3 Education and support for representative, direct and stealth democracy H Coffé A Michels Electoral Studies 35 Direct democracy: The politics of initiative, referendum, and recall T E Cronin Harvard University Press Cambridge, MA Cognitive mobilization and partisan dealignment in advanced industrial democracies R J Dalton The Journal of Politics 46 1 Partisan mobilization, cognitive mobilization, and the changing American electorate R J Dalton Electoral Studies 26 2 The good citizen: How a younger generation is reshaping American politics R J Dalton CQ Press Washington, DC Public opinion and direct democracy R J Dalton W P Burklin A Drummond Journal of Democracy 12 4 Popular support for direct democracy T Donovan J A Karp Party Politics 12 5 Increasingly unequal? The economic crisis, social inequalities and trust in the European Parliament in 20 European countries Dotti Sani G M Magistro B European Journal of Political Research 55 2 ESS round 6: European Social Survey round 6 data 10.21338/NSD-ESS6-2012 European Social Survey European Social Survey How Europeans view and evaluate democracy M Ferrin H Kriesi Oxford University Press Oxford Conceptualizing and measuring the quality of democracy: The citizens' perspective D Fuchs E Roller Politics and Governance 6 1 The referendum in representative democracies M Gilljam P Pesonen O Listhaug To join or not to join: Three Nordic referendums on membership in the European Union A. Jenssen, P. Pesonon, & M. Gilljam Oslo Scandinavian University Press Economic inequality, winner-loser gap, and satisfaction with democracy S M Han E C C Chang Electoral Studies 44 J R Hibbing E Theiss-Morse Stealth democracy: Americans' beliefs about how government should work Cambridge Cambridge University Press When Europe matters: The impact of political information on voting behavior in EU referendums S Hobolt Journal of Elections, Public Opinion and Parties 15 1 Occurrence and policy consequences of referendums: A theoretical model and empirical evidence S Hug Journal of Theoretical Politics 16 3 Cognitive mobilization and European identity R Inglehart Comparative Politics 3 1 The politics of direct democracy: Referendums in global perspective L Leduc Broadview Press Peterborough The democratic effect of direct democracy L Leemann F Wasserfallen America Political Science Review 110 4 Referendum democracy: Citizens, elites and deliberation in referendum campaigns M Mendelsohn A Parkin Springer New York, NY The populist zeitgeist. Government and Opposition C Mudde 39 C Mudde Populist radical right parties in Europe Cambridge Cambridge University Press Critical citizens: Global support for democratic government P Norris Oxford University Press Oxford Populism in Western Europe: Comparing Belgium, Germany and The Netherlands T Pauwels Routledge New York, NY Referendums around the world: The continued growth of direct democracy M Qvortrup M. Qvortrup Palgrave Macmillan London Referendums in Western Europe Inflation, unemployment, and government collapse J Robertson Comparative Political Studies 15 4 Parties and the expansion of direct democracy: Who benefits? Party Politics S E Scarrow 5 Public support for referendums: The role of the media. West European A R Schuck C H De Vreese Politics 34 2 Public support for referendums in Europe: A cross-national comparison in 21 countries A R Schuck C H De Vreese Electoral Studies 38 Referendums around the world: The continued growth of direct democracy U Serdült M. Qvortrup Springer New York, NY Referendums in Switzerland Modeled variance in two-level models T A B Snijders R J Bosker Sociological Methods & Research 22 3 Economic inequality and democratic political engagement F Solt American Journal of Political Science 52 1 Modeling multilevel data structures M Steenbergen B S Jones American Journal of Political Science 46 1 The missing dimension of democracy: Institutional patterns in 25 EU Member States between A Vatter J Bernauer European Union Politics 10 3 Who is willing to participate? Dissatisfied democrats, stealth democrats and populists in the United Kingdom P Webb European Journal of Political Research 52 6 To what extend you think…the following statement applies in [country]. Citizens in [country] have the final say on the most important political issues by voting on them directly in referendums Referendums in practice (votedirc. does not apply at all) through 10 (applies completely Calculated as votedir-votedirc. −10 through 10, high scores reflect greater gap between greater regard for referendums as a democratic principle and perceptions of referendum use in practice Disappointment with referendums ESS round 6: European Social Survey round 6 data 10.21338/NSD-ESS6-2012 European Social Survey European Social Survey Modeled variance in two-level models T A B Snijders R J Bosker Sociological Methods & Research 22 3 Referendums around the world: The continued growth of direct democracy M Qvortrup M. Qvortrup Palgrave Macmillan London Referendums in Western Europe
(xsd:string)
|
![[This page as RDF]](https://data.gesis.org/gesiskg/static/rdf-icon.gif){kind=icon}