Although democracy is generally regarded as superior to other forms of government, a widespread malaise can be observed in many Western democracies today. More specifically, citizens have exhibited increasing disenchantment with the functioning of representative democracy over the last decades (e.g., Pharr and Putnam, 2000; Pitkin, 2004). This concerning trend manifests itself in various symptoms such as lower participation rates, a growing alienation between citizens and their representatives, decreasing trust in elites and institutions, more frequent electoral successes of populist actors, and even recurring social unrest, all of which ultimately call into question the legitimacy of representative democracy (Leininger, 2015; Altman, 2019; Vatter et al., 2019).

In this context, new participation rights in general and the introduction of direct-democratic institutions in particular are often discussed as promising remedies to overcome the current democratic crisis (e.g., Cain et al., 2003). In line with the theorists of participatory democracy (Pateman, 1970; Barber, 2003), the hope is that, by providing citizens with the opportunity to participate in democratic decision-making, the legitimacy of the political system will be restored. Indeed, direct democracy enjoys great popularity in representative democracies. International comparative surveys have shown that the overwhelming majority of respondents in most countries welcome the introduction of referendums and initiatives (e.g., Bowler and Donovan, 2019; Bessen, 2020). This strong demand for increased participation rights does not seem to have been quenched by the rise in the number of direct-democratic votes across the world over the last decades (Brüggemann et al., 2023).

As an increasing number of citizens have gained experience with direct democracy over recent years, scholars have expressed increased interest in studying whether such devices can deliver on their promises. At the most basic level, academic contributions draw a distinction between (primary) instrumental and (secondary) spillover effects (e.g., Tolbert and Smith, 2006). Whereas the first refers to policy outcomes, the latter is concerned with actor behavior and attitudes that go beyond specific issue-related decisions. As far as the instrumental effects are concerned, empirical studies have primarily focused on economic performance, congruence between citizen preferences and policy outcomes, and protection of minority rights. Except for the latter, a rather consistently positive picture emerges from the state-of-the-art (Lupia and Matsuaka, 2004; Matsusaka, 2005; Maduz, 2010; Vatter et al., 2019). This conclusion is in line with the view that, despite the much-discussed risk of subversion by special interest, direct democracy generally ‘serves the many and not the few’, as Matsusaka (2005: 200) put it.

Regarding the spillover effects, the evidence seems to be much more disputed. This is remarkable since the proponents of direct democracy have long dominated the field, both from theoretical and empirical points of view. Following participatory theories, direct democracy used to be predominantly regarded as a valuable element of democratic citizenship by scholars. The possibility to take part in referendums and initiatives was expected to empower and enlighten citizens by bolstering their civic virtues in democratic life (e.g., Frey, 1997). In accordance with this line of reasoning, numerous studies have found evidence for such desirable ‘educational effects’, above all in terms of participation, efficacy, knowledge, and satisfaction (Smith and Tolbert, 2019). However, due to the emergence of non-significant and negative effects, some serious doubts have been cast on the prevailing view about the positive side effects of direct democracy. Contrary to the classical elitist and realist theorists (Schumpeter, 1942; Sartori, 1987), who emphasized the presumed political disinterest and widespread incompetence of ordinary citizens, these challenging studies basically refute the thesis of educational effects (at least in the case of the United States) by arguing that ballot measures intensify partisan conflict around an issue, since mobilizing actors have an incentive to strongly criticize the jobs made by members of the government and legislators (see Dyck and Lascher, 2019).

This article aims to provide an overview of scholarship on the question of whether direct democracy increases individual civic virtues by relying on a systematic literature review. To that end, it focuses on the effects of direct democracy on those four areas to which scholars have devoted much attention so far: (1) electoral turnout, (2) external and internal efficacy, (3) political knowledge, and (4) subjective wellbeing and satisfaction with democracy. Note that the fourth area includes both individual and collective dimensions, whereas the latter refers to political support and institutional trust in addition to satisfaction with democracy. Based on a selection of 67 studies, it turns out that there is only little positive overall effect of direct democracy on civic virtues, with a great deal of variation. The empirical analysis establishes a negative time trend, indicating that scholars increasingly reported negative findings over the years. The main result of this review thus calls into question the expectations advanced by theorists of participatory democracy and gives some credit to the skeptics of direct democracy.

The remainder of this article is organized as follows: The following section describes the procedure for the selection of studies that are used for this systematic literature review. Thereafter, the empirical results are presented in two steps. After an overall analysis, the effects of direct democracy are studied at the level of four selected civic virtues. Finally, the last section summarizes the main findings and provides interested scholars with some advice for future empirical investigations.

In this section, I shall specify the strategy adopted to search the literature and the criteria used to select the analyzed studies. Given that the objective of this article is to provide an overview of the effects of direct democracy (independent variable) on a selection of civic virtues (dependent variables), I decided to focus the search strings on these two crucial elements. For the independent variable, the strategy consisted in relying on ‘direct democracy’ as well as on two of its most common synonyms: ‘referendums’ and ‘initiatives’ or ‘ballot initiatives’. This resulted in three search terms ‘direct democra*’, ‘referend*’, and ‘ballot’. Asterisks were employed to account for possible singular forms (e.g., ‘referendum’), adjectives (e.g., ‘direct democratic’), and different spellings (e.g., ‘referenda’). Note that ‘initiatives’ were eventually omitted, given that the inclusion of the term ‘initiative*’ would have produced far too many search results, the vast majority being not related to direct democracy.

As to the dependent variables, the search strings refer to each of the selected civic virtues. Regarding electoral turnout, I decided to resort to ‘turnout’ as well as to ‘electoral participation’ and ‘voter participation’, since the more general term ‘participation’ yielded a non-manageable number of results.¹ While I ultimately stick with ‘efficac*’ for external and internal efficacy, I rely on knowledge (‘knowledge*’) and awareness (‘aware*’) for the domain of political knowledge. Finally, the relatively broader area of satisfaction is ultimately covered by four search terms. In addition to the obvious and general ‘satsif*’, ‘happiness’ is meant to cover the individual dimension, whereas the selection of ‘political support’ and ‘trust’ is expected to ensure the inclusion of studies that focus on the collective dimension. ²

The pre-selection required to include at least one of the three search terms on direct democracy as well as at least one of the 10 just mentioned virtue-specific terms. This resulted in the following search query:

This search query was entered into the Web of Science database in August 2023 by making use of two restrictions. First, I used the topic search option, meaning the records only applied to the title, abstract, and keywords fields. Second, I limited the search query to articles published in scientific journals, thus leaving aside other formats such as books, PhD theses, and data studies. In doing so, my pre-selection amounted to 2,016 records.

To select suitable studies for analysis, a couple of inclusion and exclusion criteria were subsequently implemented. Most importantly, from a substantial point of view, I retained those articles that contained at least an empirical analysis that examined the effects of direct democracy on at least one of the four selected civic virtues. More specifically, variations in terms of direct-democratic contexts had to be present on the dependent variable(s). Note that there is a distinction between genuine direct-democratic and non-direct-democratic contexts. For the latter (i.e., contexts outside Switzerland and the subnational level of the United States), I also decided to include longitudinal studies on single direct-democratic votes. Such ‘change analyses’ (i.e., before and after a vote) examine a substantial increase in terms of direct democracy in essentially representative political systems. It also needs to be mentioned that studies had to report the overall effects of direct-democracy variables for being selected, meaning that empirical analyses limited to the effects of a particular segment of the population (e.g., participating voters as opposed to all respondents) were removed.

As to the dependent variables, I specified the inclusion criteria on the level of each civic virtue. Regarding electoral turnout, I restricted the selection to participation rates in official elections and direct-democratic votes. With respect to efficacy, I considered studies on both internal and external efficacy. As to political knowledge, I limited myself to analyses that use factual political knowledge items. ³ Hence, both general and issue-specific knowledge were included. Finally, the studies on satisfaction included subjective wellbeing and happiness at the individual level, along with institutional trust, political support, and satisfaction with democracy at the contextual level.

From a formal point of view, I only included articles written in English. A total of 48 articles met these criteria. In a second step, I replicated this procedure in Scopus, Elsevier’s citation and abstract database. This yielded four additional articles. Finally, I relied on Google Scholar, a web search engine for scholarly literature that can be regarded as a useful complementary resource (Gusenbauer and Haddaway, 2020). This added two articles to my stock of 54 articles in total. ⁴ However, since some articles focused on several dependent variables of interest, the number of selected relationships between direct democracy and the selected civic virtues amounts to 67 for the present analysis.

As to the selected civic virtues, most of these empirically analyzed relationships focused on electoral turnout (42%), followed by satisfaction (28%), efficacy (21%), and knowledge (9%). ⁵ The selection at hand is dominated by subnational studies, with 70% taking full advantage of variations at the regional level (such as US states and Swiss cantons) and 10% at the local level. In addition, all studies were conducted in a quantitative manner. Furthermore, it is noticeable that the selected analyses were published on a rather regular basis between 2000 (oldest contributions) and 2023 (newest ones), with the median year being 2011. Almost four in five analyses refer to the individual level (79%) using representative survey data that were enriched with contextual data, at least about variations in terms of direct democracy, with the remaining ones limiting themselves to the contextual level (21%). As for their general design, 54% of the selected analyses are cross-sectional in nature, 37% are time-series and cross-section analyses, while the remaining 9% are purely longitudinal. In geographical terms, the United States of America proves to be the most studied country context (48%). Next in line is Switzerland (22%), while internationally comparative studies account for 16% and other single-country studies account for 12% of the selection. ⁶

In the context of the current crisis of representative democracy, many citizens strive to decide the fate of major issues themselves through the adoption of direct-democratic institutions. In line with this demand, the number of referendums and initiatives has substantially increased across liberal democracies over recent decades. This has provided scholars with numerous opportunities to study whether these participatory devices have the potential to restore legitimacy. While direct democracy seems to have basically delivered on its instrumental promises related to policy-related outcomes (primary effects), the state-of-the-art on the secondary effects (i.e., normatively desired actor behavior and attitudes beyond policies) can probably best be described as inconclusive.

This article aimed to shed light on the subject by conducting a systematic literature review on the effects of direct democracy on four civic virtues: (1) electoral turnout, (2) external and internal efficacy, (3) political knowledge, and (4) subjective wellbeing and satisfaction with democracy. Based on a selection of 67 studies that empirically analyzed the effects of these associations, this article established only little beneficial effects on average with a great deal of variation. In addition, this review showed that scholars increasingly reported negative findings over time. This tendency applies to all selected civic virtues except for the studies on both internal and external efficacy.

If this review had been published 15 years ago, it almost certainly would have concluded on an optimistic note, stating that direct democracy can be considered a promising institutional device to promote civic virtues. Yet, the strong negative time trend calls into question the key argument put forward by theorists of participatory democracy, according to which referendums and initiatives empower and enlighten citizens. By contrast, it gives some credit to more skeptical views on the merits of direct democracy regarding secondary spillover effects. Alternatively, the negative time trend reported here may be linked to a general decline in terms of democratic quality, as a reviewer observed. This implies that it would be more difficult for direct democracy to deliver on its promises under such deteriorating conditions. This article has brought some further insights to light. Analyses that examined the lower levels of government were found to be more likely to arrive at more positive conclusions. In addition, the empirical contributions on the United States turned out to be significantly more optimistic than the Swiss studies. Yet, most of this discrepancy can be attributed to studies on electoral turnout, which may be a result of diverging practices in these two countries when it comes to the concurrent scheduling of elections and referendums.

In light of the main findings of this review, I would like to suggest some avenues for research. Given that it is not settled whether direct democracy leads to an increase in civic virtues, more research is obviously needed. This is exemplified by the fact that the number of analyzed relationships in this review only amounts to 67. ¹¹ Although scholars have increasingly relied on large-scale international comparisons and more systematic and encompassing analyses (e.g., by including several data sources or various ballots) in recent years, there is still a lot of potential. Perhaps, except for electoral turnout, this article suggests that there is little conclusive evidence at the level of single individual civic virtues. Since this review covered the most studied areas so far, the call for more empirical contributions on other virtues seems all too evident. This would allow complementing existing findings on various domains such as political interest (Ladner and Fiechter, 2012; Freitag and Zumbrunn, 2022), social trust (Dyck, 2012), tax morale (Torgler, 2005; Hug and Spörri, 2011), and forms of political engagement that go beyond electoral participation such as associational engagement (Freitag, 2006; Boehmke and Bowen, 2010; Stadelmann-Steffen and Freitag, 2011) and new forms of deliberative citizen participation (Ladner and Fiechter, 2012).

Based on the main results of this review, it would not be surprising to observe that mixed findings emerge from such additional studies. In the absence of clear-cut evidence, scholars may be well advised to increasingly take into consideration the conditions under which direct democracy exerts a positive or negative effect on a given civic virtue. In other words, they may more systematically look at possible interaction effects. A couple of studies suggest that mediating factors at both the contextual and individual levels can be at play. These may range from the type of direct-democratic institutions (Altman, 2013) to electoral contexts (Childers and Binder, 2012), over-issue characteristics (Biggers, 2011, 2012), population diversity (Dyck, 2012), and personality traits (Freitag and Zumbrunn, 2022).

This review has found that the academic literature is dominated by two country contexts: the United States and Switzerland. With the global increase in direct-democratic votes in recent decades, it would be appropriate for researchers to increasingly devote their attention to other political systems in which direct democracy has been traditionally less established. Following the Swiss and U.S. studies, cross-sectional analyses could be conducted at the subnational level in countries characterized by variations in terms of the existence or the extent of democracy (see Ackermann et al., 2023 for such an analysis on the states of Germany). From a longitudinal perspective, however, it would seem appropriate, from a researcher’s point of view, to focus on those jurisdictions characterized until recently by an (almost complete) absence of referendums and that have experienced a recent significant increase. This would make it possible to conduct change analyses in which the development of citizens’ civic virtues before and after direct democracy took off could be studied. According to the C2D database (available at http://www.c2d.ch), country contexts that meet these criteria include Lithuania, Mexico, the Netherlands (at least up to 2018), Slovakia, Slovenia, and Taiwan, as far as referendums at the national level are concerned.

In addition, it may be promising to focus on comparisons across direct-democratic systems. Indeed, this review has shown that this kind of approach has so far been neglected in the literature (Altman, 2013 being an exception). This is all the more regrettable as comparative studies have shown that direct-democratic institutions and practices can vary substantially across political contexts (Altman, 2019). At the same time, the present article has also revealed that studies on the subnational level in general and on the local level in particular tended to yield rather positive effects. However, empirical research proves to be particularly scant when it comes to studies on municipalities (and equivalents). Hence, the local level can be regarded as another blind spot of the academic literature that would need some enlightenment in the near future when it comes to examining the merits of direct democracy in general and in terms of individual civic virtues in particular.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

LB: Conceptualization, Formal analysis, Investigation, Methodology, Visualization, Writing – original draft.