Introduction

Within the literature, various authors have claimed that an expansion of the political participation repertoire that is available to citizens might serve as an ideal manner to counter the alleged loss of legitimacy of representative politics (Saward, 2003; Tormey, 2015). It has been argued that if public policy is expected to correspond to citizens’ preferences, the best way to achieve such a congruence is by involving citizens more directly in the decision-making process (Budge, 1996). Examples that have been mentioned in this literature are the introduction of deliberative democracy, efforts to get citizens involved in local projects or participatory budgeting and a wider use of direct democracy initiatives. Indeed, the number of referendums in liberal democracies has been expanding continuously over the past decades as more governments seem to be confronted with a problem of democratic legitimacy (Qvortrup, 2014). There seems little doubt that in these democracies, there is a demand by at least some groups of the population to expand the opportunities they have at their disposal to get their voice heard in the process of political decision-making (Dalton and Welzel, 2014).

Traditionally, however, it is assumed that increasing participation opportunities might increase patterns of inequality. The ‘participation paradox’ argument as it was introduced by Verba and Nie (1972) claims that more intense forms of political participation will be used mainly by the more privileged groups within society, that already have more opportunities to have an impact on political decision-making. Subsequent research has indeed confirmed that this participation paradox to a large extent determines patterns of social stratification in political activity (Schlozman et al, 2012). Marien et al, (2010) have shown that the social stratification of participation in new and emerging forms of political participation is much stronger than the stratification one typically finds among institutionalized forms of participation. Their analysis suggests that levels of education function as the most important determinant of stratification, as the new forms of participation often have in common that they require a rather high level of cognitive resources, and these resources are not available in an equal manner across the population (Nie et al, 1996). Expanding the opportunities to participate, therefore, would in practice predominantly benefit the well-off (Verba, 2003).

Other authors, however, have made exactly the opposite claim. Kriesi (2004) and Tarrow (1998) start from the notion that the opportunities being offered by the political system to a large extent determine whether or not citizens will participate in the political process. Their assumption is that the more, and the more diverse opportunities there are, the larger the segment of the population that will actually participate. Theoretically, this is in line with the Lijphart (2012) argument that a more open political system, i.e., a political system with more opportunities for power sharing, will lead to a more equal distribution of political power across the population. Given these opposing points of view, it is therefore highly relevant to ascertain what are the effects of the availability of direct democracy procedures within a country.

We know from previous research that the participation paradox is present among the more intensive forms of democratic innovation, like deliberative democracy. In that case, participants indeed need rather high levels of political interest and knowledge (Grönlund et al, 2015). Social conventions on public discourse, too, might narrow the opportunities for specific groups of the population to achieve full equality in deliberative settings (Young, 2001). While proponents of deliberative democracy state that deliberative democracy increases the quality of democracy, other authors have argued that this form of innovation might in fact lead to cumulative forms of inequality (Hooghe, 1999). Obviously, political deliberation can be considered as a time-consuming form of participation that also requires quite some social and discursive skills. The question, however, is whether the participation paradox can also be found for forms of participation that seem more accessible, even to those who have fewer political resources. At first sight, it is less evident that exactly the same mechanism would be present for direct democracy. In most cases, experiences with direct democracy remained limited to a referendum, where a simple yes/no answer is all that is required from participants. This would not require all that much sophistication, while a limited participation act like this is also less likely to lead to democratic fatigue. Nevertheless, some of the research for Switzerland clearly suggests that for referendums too, especially the lower educated refrain from participating (Kriesi, 2008). What is even more troubling is that there might be a spill-over effect: the proliferation of referendums not only leads to lower levels of participation in the referendums themselves, but also spills over into electoral participation (Freitag and Stadelmann-Steffen, 2010). Again, the Swiss case stands out in this respect: while the country boasts itself as the best example of practicing direct democracy, voter turnout levels for parliamentary elections typically are below 50 per cent, which is remarkably low for European standards.

Thus far, however, we do not have sufficient evidence to arrive at a conclusion about the full democratic potential of direct democracy, and the effect it has on the social stratification of political participation in general. On the one hand, the availability of direct democratic instruments provides citizens with an additional entry point to the political system that is in principle available to all citizens, thus leading to more equal participation (Lijphart, 2012). On the other hand, this additional instrument also adds to the complexity of the political system and might induce a participatory fatigue that is most noticeable among those that are least likely to participate in the first place (Verba et al, 1995). In this article, therefore, we analyze the relation between the availability and the use of direct democracy in a country, and the stratification of various political participation acts. We focus on stratification because the alleged goal of representative democracy is to ensure that all groups within the population receive an equal opportunity to have their interests represented in politics. If this ideal is not realized, obviously political systems do not live up to the expectation of providing equal voice to all their citizens (Roemer, 2000).

We aim to contribute to the currently available literature, because the empirical research on the effects of direct democracy thus far has been inconclusive. Dalton and Weldon (2013, p. 59) even state: ‘The estimated effects of direct democracy are typically weak across various citizenship measures. (…) But most effects are not statistically significant and tend to vary across different indicators of national usage of direct democracy’. It is therefore important to contribute solid empirical insights into this debate that, thus far, has been conducted primarily on normative grounds. In this article, we start from the assumption that one of the reasons for the non-findings might be that in previous research effects have not been differentiated. However, we can assume that different groups within the population react differently to the incentives given by the political system. Depending on how these differences work out, these might strengthen or reduce inequalities within the population. This might suggest a trade-off effect. It might be that direct democracy is associated with some positive attitudinal effects, but the downside for this might be a larger degree of inequality in what groups within society get their voices heard within political decision-making.

In this paper, we first review the literature on the effects of direct democracy, before we present our data and methods. Following the results section, we reflect on what this implies for the democratic potential of direct democracy initiatives.

Literature

In the literature, it is routinely assumed that experiences with direct democracy have a positive effect on democratic attitudes (Qvortrup, 2014), although it has to be noted that these effects might be moderated by personality characteristics (Freitag and Ackermann, 2016). There is indeed a strong research line, starting from the notion that experiences with democratic procedures would lead to a subsequent rise in, e.g. political trust or political efficacy. The guiding idea is that participation and power sharing are empowering experiences that should be associated with a more positive attitude towards the political system in general. If citizens have had the experience that they have participated in the decision-making process in a meaningful manner, it is more likely that they will consider the outcome to be legitimate (Tyler, 2011). This claim renders the expansion of direct democratic procedures particularly appealing as a potential remedy for the alleged contemporary crisis of representative democracy. While turnout levels are rapidly declining (Gray and Caul, 2000; Dassonneville and Hooghe, 2016), political parties are losing members (Dalton and Wattenberg, 2002; Mair and van Biezen, 2001; van Biezen et al, 2012; Whiteley, 2011) and citizens become increasingly sceptical towards democratic processes and its institutions (Kaase and Newton, 1995; Klingemann, 1999; Newton and Norris, 2000), proponents of direct democracy argue that increased direct engagement of citizens can contribute to maintain and enhance perceptions of legitimacy and strengthen citizens’ linkage with the state (Cain et al, 2003). Instead of electing public officials, referendums and other similar initiatives allow citizens a direct instrument to decide on how their country should be run.

In previous studies, it is suggested that citizens who live in direct democracies believe more strongly that government is responsive to their demands (Bowler and Donovan, 2002; Hero and Tolbert, 2004; Mendelsohn and Cutler, 2000). Smith and Tolbert (2004) show that the availability of direct democracy initiatives is associated with higher levels of political knowledge, and they assume that citizens put more effort in getting themselves informed if they live in a system where they are asked to be involved in the decision-making process. Some of the research suggests that the availability of direct democracy is associated with higher levels of political knowledge, interest (Smith, 2002) and civic engagement (Smith and Tolbert, 2004; Tolbert et al, 2003). The assumption is that when citizens know it is likely they will have to express their opinions on salient political issues, they will experience this as an incentive to obtain information on these issues. Therefore, strengthening the opportunities for direct democracy is often considered to be a key element of democratic innovation, as it is stated that opening up the possibility to decide for oneself should restore or strengthen democratic legitimacy (Geißel and Newton, 2012).

However, even by proponents of direct democracy, direct democratic instruments are widely considered as a supplement to representative democracy, rather than a replacement (Frey, 1994). Introducing direct democracy would always imply that referendums, in some way or another, will be combined with the indirect, institutionalized and non-institutionalized instruments of representative democracy. Therefore, the question how these new channels of citizen engagement would affect the existing channels of political participation in the representative system is crucial. As no author sees direct democracy as a stand-alone instrument that would solve the entire complex interaction between citizens and the state, its effects should not be investigated in an isolated manner. Rather, it is important to investigate what effect direct democracy has on the whole array of institutionalized and non-institutionalized instruments citizens have at their disposal to get their voice heard, and that is exactly the goal of the current paper. If direct democracy would be associated with, e.g. a lower level of willingness to become engaged in electoral politics, this would counteract much of the democratic gain associated with the referendums themselves. This study therefore focuses on the potential effects of direct democracy on participatory equality in different forms of political participation. If direct democracy would have a negative spill-over effect on other, well-established forms of political participation, this might eliminate any democratic effects the direct procedure itself might have. Participatory equality, which represents a basic goal of democratic systems, is assumed to be reached when in making collective decisions, the interests of all citizens are equally taken into account (Dahl, 1989, 1996; Marien et al, 2010). Citizens do not only communicate their interests, preferences and needs through political participation, they also use the institutionalized and non-institutionalized channels of participation to pressure public officials so that those interests are taken into account. As citizens’ interests and preferences differ, the equal consideration of all interests would seem to require equal participatory input. The quality of a democratic system therefore depends not only on the sheer volume of political participation but also on its distribution across the population. Previous research has shown that political participation is stratified along the measures of socio-economic status, and most notably along levels of income and education (Berinsky, 2002; Stolle and Hooghe, 2011; Verba et al, 1978; Verba et al, 1995). This paper therefore aims to answer the question whether the existence, regulations and use of national referendums and initiatives affect the stratification of political participation along the measures of socio-economic status.

The current literature yields mixed results on the relationship between levels of political participation and direct democracy. While some scholars find that voter turnout is higher in systems where direct democratic rights are available (Tolbert et al, 2001; Tolbert et al, 2003; Tolbert and Smith, 2005), others find that direct democracy has a negative effect (Freitag and Stadelmann-Steffen, 2010) or no effect on electoral turnout (Fatke, 2015) and virtually no effect on non-institutionalized forms of political participation (Dalton and Weldon, 2013). Peters (2016) argues that these inconclusive findings might stem from the fact that different institutions of direct democracy have different effects on political participation. Initiative referendums and agenda initiatives function in a complementary way to institutions of representative democracy, and therefore, they might boost political participation. However, other institutions of direct democracy such as recalls and constitutional and legislative referendums fulfil tasks that are traditionally accomplished by institutions of representative democracy. Therefore, these specific institutions of direct democracy compete with the institutions of representative democracy and function as an alternative channel in the decision-making process. As a consequence, the extensive availability of recalls and constitutional and legislative referendums is argued to depress electoral participation.

In summary, it can be stated that direct democracy does not necessarily lead to more political participation. Moreover, what we also know from the Swiss case is that direct democratic instruments are not used by everybody in the same manner, particularly when turnout is low (Linder, 1994). Fatke (2015), on the other hand, suggests that direct democracy does neither increase nor decrease social bias in the Swiss electorate. A limitation to this study is that it only tests the moderating effect of direct democracy on the relationship between socio-economic status and electoral participation. Less, however, is known about the effect of direct democracy on the stratification of political participation more generally. Two competing theoretically informed scenarios are possible to explain this relation.

First, it could be assumed that direct democracy creates greater equality in political participation. Advocates of participatory democracy expect that participation has a socializing and empowering effect on citizens (Barber, 1984; Pateman, 1970, 2012). It is argued that by participating regularly, citizens become psychologically more engaged and this should hold for all citizens, independent of their socio-economic status. Dalton and Weldon (2013) argue that referendums could function as an ideal training ground for political participation more broadly defined, as typically a referendum boils down to a very easy yes/no option. Theoretically, this should make participation easier than taking part in, e.g. parliamentary elections, where voters are confronted with a myriad of issues, parties and candidates. Offering this kind of low-intensity act, therefore, should lead to a positive spill-over effect on other forms of participation: ‘because of their low involvement costs, referendums can act as a gateway to broader, more intensive types of political participation’ (Dalton and Weldon, 2013, p. 53). If referendums are considered as an ‘easy’ form of participation, they should offer more opportunities to especially the least sophisticated, and therefore, the expectation is that they contribute to political equality. Once the lower educated have the experience that they can cast a meaningful vote in a referendum, it is more likely that they will also take the next step to fully engage in institutionalized and non-institutionalized participation.

An opposing line of the literature, however, expects exactly the opposite spill-over effect. While direct democratic instruments represent more opportunities for citizens to voice their interests and preferences, they also require that citizens make additional decisions which they would not have to make in a purely representative system. For every referendum, this implies that they need to inform themselves about the issues at stake and to obtain additional cues in order to take an informed decision (Donovan and Karp, 2006). Direct democratic initiatives and referendums thus create additional cognitive costs for citizens and they render the system more complex for its citizens. Especially for the citizens with the lowest levels of political sophistication, this might imply an additional barrier for effective participation in general. Those citizens could hence be expected to refrain from political participation all together: They would neither engage in the direct democratic initiatives and referendums nor in the institutionalized and non-institutionalized ways that are characteristic of today’s representative democracies. Given the fact that especially non-institutionalized forms of participation require more effort, the literature on the ‘participation paradox’, would allow us to assume that these acts will be the first ones to suffer from the attention that is being devoted to forms of direct democracy. Previous research has shown that the emergence of new, additional forms of political participation has led to more, rather than less distortion of political participation along the lines of education (Marien et al, 2010; Stolle and Hooghe, 2011). Also, existing studies show that direct democracy fosters voter fatigue among citizens (Bowler et al, 1992). The basic argument here is that citizens are not willing or able to be routinely involved in all kinds of political decision-making. If they experience that the demands placed on their cognitive resources or on their spare time become too intensive, the easiest solution is simply to refrain from all of these institutionalized and non-institutionalized political acts. This line of the research therefore leads to the hypothesis that a proliferation of participation opportunities would lead to a reduced willingness to participate, especially among the population groups with lower levels of political sophistication. These two lines of the literature thus lead to opposite expectations on the spill-over effect of direct democracy on other forms of participation. While quite some of the earlier studies on this topic have been limited to one country, or a limited number of countries, in this paper we will investigate the occurrence of these spill-over effects for a vast array of European countries that have participated in the European Social Survey.

Data, measurement and method

To test the moderating effect of direct democracy on the relationship between socio-economic status and political participation, we rely on cross-national individual-level data from the fifth round of the European Social Survey (ESS Round 5: European Social Survey Round 5 Data, 2010). These population figures are combined with information from the Index of Power Dispersion in Direct Democracy from the Diffusion of Power in 61 Democracies dataset (Vatter and Bernauer, 2015).

This study presents a synchronic comparison of the effect of socio-economic status on political participation in 23 European states in 2010. The European Social Survey has been chosen as data source, because the data is comparable and because the ESS is respected due to its high standards concerning the survey design and the data collection (Lynn, 2003). We have selected the fifth round, because it contains the broadest range of data on political participation, as the question on party membership was no longer included in subsequent rounds of ESS. The fifth round of the ESS which was administered in 2010 contains cross-sectional samples from 28 countries.

For 24 of those countries, we also have information available on the availability of direct democracy for its citizens: Belgium, Bulgaria, Switzerland, Czech Republic, Germany, Denmark, Estonia, Spain, Finland, France, the UK, Greece, Croatia, Hungary, Ireland, Israel, Lithuania, the Netherlands, Norway, Poland, Portugal, Sweden, Slovenia and Slovakia. The units of analysis are individuals, which were interviewed face-to-face. In total, 46,849 respondents participated in the survey.

The ESS Round 5 contains information about eight different forms of political participation. The respondents were asked whether they participated in these following forms of action during the last 12 months: contacting a politician or other government official, working in a political party or political action group, displaying political views by wearing a badge or sticker, signing a petition, taking part in lawful public demonstrations and boycotting certain products. Moreover, the respondents were asked whether they are a member of a political party and whether they had voted in the last national election. The eight variables resulting from these questions were recoded as dichotomous variables that take on 0 for ‘No’ and 1 for ‘Yes’. Respondents who did not give any answer, opted for ‘Don’t know’ or were not eligible to vote in the previous national elections were excluded from the analysis. The frequencies with which respondents indicated that they had participated in a particular form of political participation are presented in Table 1.

Table 1 Frequencies of different participation forms
Full size table

Table 1 shows that voting represents by far the most widely spread form of political participation, followed by signing petitions and boycotting specific products. On the other hand, being a party member, working in a political party and action group do not seem to be very widespread as forms of political participation among European respondents.

In line with the existing literature, going back to the participation typology created by Barnes and Kaase (1979), we distinguish between three forms of political participation: voting, institutionalized and non-institutionalized forms of political participation (Marien et al, 2010; Stolle et al, 2011). First, voting is regarded as a single item, as just because of its sheer volume this act would fully determine any other factor. Working for a political party or action group and being a member of a political party are considered as acts of institutionalized political participation that are related to the electoral process, whereas signing a petition, taking part in a lawful demonstration or boycotting certain products represent non-institutionalized forms of political participation. Wearing a badge or sticker as well as contacting a politician is excluded from the analysis, because they cannot clearly be identified as institutional or non-institutional acts1. As the distribution of participation acts is not normal, both variables were dichotomized (did not participate in any act vs participated in at least one of these acts). Hence, all three participation variables are binary.

The independent variable of interest is socio-economic status which is measured based on two indicators from the ESS (2010): level of education and household income. The level of education was recoded into three broad categories, distinguishing between (1) those who have no formal or only primary education, (2) those who obtained secondary education and (3) all those with tertiary education. Looking at an individual’s income represents another straightforward way to measure socio-economic status. This is done by including the item that captures the respondents’ total available household income2.

Direct democracy which serves as moderating variable is measured with the Index of Power Dispersion in Direct Democracy (Vatter and Bernauer, 2015) that distinguishes between forms of direct democracy that concentrate power and forms of direct democracy that disperse power. The index was developed in an effort to expand on the rather rudimentary dichotomy Lijphart (2012) uses to investigate the role of direct democracy in distinguishing between various styles of democracy. This index has been chosen because it considers not only the existence but also the regulations and use of national referendums and initiatives. In the index, countries are awarded points according to two criteria: first, if they have legislation on direct democratic instruments. Second, countries receive additional points if these direct democratic instruments are actually used. The index therefore does not just express what is the legal status of a referendum, but also how often it is being used. This implies a high score for a country like Switzerland with its strong tradition of referendums and Lithuania, where twelve nation-wide referenda were held since independence in 1990, and a zero score for Belgium, Germany or the UK, countries that have no provision for a national referendum whatsoever.

Finally, we control on the individual level for sex, the age of the respondents in 2010 and for their level of political interest, as high levels of political interest increase the probability to become politically engaged (Neundorf et al, 2013; Verba et al, 1995). Also, we include a categorical variable to account for respondents’ employment status. This variable is recoded so that white-collar workers form the reference category, while the other categories are dummy-coded. These dummies represent (1) blue-collar workers, (2) unemployed respondents and (3) others3. The differentiation between white-collar and blue-collar jobs was made based on the ISCO88 codes (jobs with codes 0, 1, 2, 3, 4 and 5 are considered as white-collar jobs). On the country level, we control for a country’s democratic history, because levels of political participation are still lower in Eastern European countries whose democratic transition is rather recent (Hooghe and Quintelier, 2013). To this end, a dummy is introduced that distinguishes established and more recent European democracies.

All in all, after removing those cases that contain missing information, we obtained a sample of 40,329 respondents who live in 23 European countries. We account for this nested structure by estimating logistic multilevel models.

Results

Our analysis is based on the assumption that the indicators of socio-economic status affect the probability respondents will become engaged in political participation and that these effects differ across the different countries. In Table 2, we test the first of the two assumptions, and we do so separately for every form of political participation we distinguished. First, we estimate null-models, showing there is quite some country-level correlation, ranging from four per cent for institutionalized participation (Model III), over nine per cent for voting (Model I), to 16 per cent for non-institutionalized participation (Model V).

Table 2 The effect of direct democracy on the probability to participate politically
Full size table

If we subsequently add the independent variables, the results show that the level of education represents a strong determinant of stratification for all three types of political participation. Tertiary education (as compared to secondary education) is positively related to all three forms of political participation. Additionally, for citizens with primary education, the probability to engage in non-institutionalized political participation is significantly lower than for citizens with secondary education (Model VI). Income increases the probability to vote (Model II) and the probability to become involved in non-institutionalized political participation (Model VI). However, it does not affect the probability to engage in institutionalized political participation (Model IV). In sum, our findings support the already existing literature that states that citizens’ socio-economic status represents an important source of stratification in political participation. This seems to be particularly true for voting and for non-institutionalized engagement.

When we move to the country-level variables in Table 2, we can first observe that non-institutionalized participation is lower in countries that made their democratic transition following the fall of the authoritarian regimes in Eastern Europe. For the two other forms of participation, we do not find any significant differences with the more established democracies. The presence of direct democracy, furthermore, also does not seem to have an effect on institutionalized an non-institutionalized participation, but we do observe a negative effect on the level of voting. This might serve as a first indication for a form of voter fatigue.

So far, we have only tested the assumption on which our argumentation is built, namely that socio-economic status stratifies political participation. However, based on our reasoning, we are actually interested in whether the effects of socio-economic status on participation change depending on the degree of direct democracy. In order to test this, we allow the indicators of socio-economic status vary—one by one—across the 23 counties and we introduce cross-level interactions in order to investigate whether the degree of direct democracy can explain the changes in these effects.

We start in Table 3 with the analysis of voting. While the effects of tertiary education and income voting vary across the countries, direct democracy cannot explain this variation, as all cross-level interactions are insignificant. Also, with respect to the effect of socio-economic status on institutionalized political participation (Table 4, Models I–III), the degree of direct democracy does not explain the variance that we found across the countries. However, the degree of direct democracy does moderate the effects of primary education, tertiary education and income on the probability to engage in non-institutionalized forms of political participation. Model IV shows that the negative effect of primary education on non-institutionalized participation that is shown in Model VI, Table 2 is enforced in countries with a high degree of direct democracy. Also, the positive effect of tertiary education (compared to secondary education) becomes stronger, the higher the degree of direct democracy. Both findings support the reasoning that in systems with an extensive availability and use of direct democratic instruments, participatory inequality is higher. Also, Model VI is in line with this claim: the positive effect of income on non-institutionalized political participation is enhanced in countries that are characterized by a high degree of direct democracy4.

Table 3 The effect of socio-economic status on the probability to vote moderated by direct democracy
Full size table
Table 4 The effect of socio-economic status on the probability to become active in institutionalized and non-institutionalized political participation moderated by direct democracy
Full size table

Discussion

Despite the ever-increasing popularity of direct democratic instruments, little is known about how these institutions affect the institutions of representative democracy. With this article, we shed light on this interplay by investigating the moderating effect of direct democracy on the relationship between socio-economic status and political participation. We have discussed two possible scenarios. For citizens with a low socio-economic status that are less likely to become politically involved, direct democracy could on the one hand function as ‘gateway’ to other types of political participation. On the other hand, however, it could also render the system more complex and lead to a fatigue that is particularly prominent among citizens with a low socio-economic status. While the first scenario would lead to more participatory equality, the latter will further undermine it. The results of our analysis support the second scenario: in systems with an extensive availability and use of direct democratic instruments, the level of education and income appear to be more important as predictors of non-institutionalized political participation than in countries where direct democratic instruments are less prominent. Not only is participation in referendums and initiatives distorted along the lines of education (Kriesi, 2008; Linder, 1994), these inequalities also seem to spill-over into other forms of non-institutionalized political participation. Apparently, socio-economically disadvantaged citizens, who refrain from using the additional channels that are provided in direct democracy systems, also abstain from participation through other channels offered by the representative system. It seems thus conceivable that in systems with a high degree of direct democracy, the system’s complexity induces a general participatory fatigue among these strata of the population. However, this has been observed only for non-institutionalized forms of political participation. With respect to voting and institutionalized participation, the availability of direct democratic procedures does not have a differentiated effect on the various population groups. Nevertheless, it should be noted that while there might be solid normative arguments in favour of direct democracy, the downside might be that it dampens the enthusiasm of specific groups of the population to participate in forms of non-institutionalized political participation.

Obviously, the goal of the current article was not to challenge the hypothesis that direct democracy leads to a ‘kinder, gentler’ form of democracy, as we did not analyse the effect of direct democracy on various forms of democratic attitudes. Following earlier analysis, we can indeed expect that there is an effect in this regard and we do not want to challenge this assumption. Given the current concern about the (lack of) democratic legitimacy in liberal democracies, this does remain a very important consideration in the debate. What we did show, however, is that there might also be a drawback to this positive effect of direct democracy, as apparently the lower educated do not use the political participation repertoire as intensively as the highly educated in a country where they are routinely asked to provide input by means of a referendum or an initiative. Possibly, more complex systems could counteract this fatigue among the socio-economically disadvantaged strata of society by providing widely accessible information shortcuts (Lupia, 1994).

It has to be noted also that in our sample, the countries that have direct democratic procedures add these to the electoral procedures that are available. The result is that the citizens of these countries have a wide array of participation opportunities. If, theoretically, direct democracy would actually replace some forms of representative democracy, the total number of participation opportunities would not rise, so it be worthwhile to investigate whether in such a context, we would observe similar democratic fatigue effects. It should also be clear that our assessment of the effects of direct democracy is dependent on current practices with regard to frequency and mobilization patterns. One could indeed imagine different forms of direct democracy, with a larger role of mobilization structures and initiatives. Here too, however, we have to note that we are limited by the data that are currently available. As a further limitation, it has to be noted that we only had access to cross-sectional data, and in the future more longitudinal data, especially from political systems with a strong experience in direct democracy, could help us to establish patterns of causality in a more convincing manner.

It falls outside the scope of the current article to investigate why exactly direct democracy has this effect. When we consider Switzerland as an outlier (and in the case of the prevalence of direct democracy, there is every reason to do so), we can observe that direct democracy does not diminish in a significant manner the participation level of lower educated groups within the population. Rather, the highly educated seem to be more eager to actually use all the opportunities that are on offer to express their opinion, and this by itself is sufficient to create a larger gap within the population. Basically, this means that especially the higher educated also use the opportunities created by systems of direct democracy, and they also seem to be able to exploit these opportunities. This finding therefore is perfectly in line with the assumptions of the participation model developed by Verba et al (1995). The more opportunities there are for participation, the easier it will be for citizens with a high level of political resources to use those opportunities to make sure their interests are being taken into account by the political system. Citizens with fewer skills and resources, on the other hand, are much less apt to use all these opportunities, thus indeed leading to a stronger participation paradox. In the quest for a better functioning democracy, apparently a trade-off between opportunities and equality is an element that thus far has been largely neglected in the literature.

Notes

  1. 1

    Wearing a campaign badge or sticker could be a badge or sticker from a political party or candidate and would accordingly fall under institutionalized political participation, but it could also be a badge or sticker from a campaign from, for example an NGO that accuses public officials in which case it would count as non-institutionalized political participation. Contacting a politician could be an individual act of a constituent questioning the position of his or her representative, but it can also be an organized and coordinated action of a protest group.

  2. 2

    Comparable household income data on Portugal is not available in the fifth round of the ESS. Portugal is therefore left out from the multivariate analysis.

  3. 3

    This category consists out of people who are in education, permanently sick or disabled, retired, in community or military service, looking after children or other persons and people who do housework or other activities.

  4. 4

    Due to this strong tradition of direct democratic decision-making, Switzerland can be considered as an extreme case. Therefore, we have repeated the analysis without Switzerland and we found that the direct effect of the index for direct democracy on voting disappeared. However, also in this smaller sample, direct democracy was moderating the relationship between tertiary education and non-institutionalized participation and the effect of income on non-institutionalized political participation. Our main conclusions therefore remain unchanged.