Public support for referendums in Europe: A cross-national
comparison in 21 countries
Andreas R.T. Schuck*
, Claes H. de Vreese
Amsterdam School of Communication Research, Department of Communication Science, University of Amsterdam, Kloverniersburgwal
48, 1012 CX Amsterdam, The Netherlands
a r t i c l e i n f o
Article history:
Available online 24 February 2015
Keywords:
Referendum support
Political cynicism
Civic duty
EU
Cognitive mobilization
Political disaffection
a b s t r a c t
Previous research is unclear about who supports the use of referendums and why. One line
of research suggests that people with greater cognitive resources are more supportive of
referendums. Another line claims that referendums are supported by citizens who feel
disconnected from the political process. We integrate both perspectives, include civic duty
and political cynicism as key explanatory variables, and offer a model explaining referendum
support across Europe drawing on both individual and contextual factors. Our study is
based on a survey conducted in 21 EU member states (N ¼ 22,806). Results show support
for both perspectives and for our new indicators, suggesting that referendum support is
highest among citizens who are critical of traditional party politics but committed to
democratic practices.
© 2015 Published by Elsevier Ltd.
1. Introduction
The use of referendums has become increasingly popular
in recent years and public support for direct democracy
is generally high. However, previous research has
yielded inconclusive results as to who supports referendums
and why. On the one hand, the cognitive mobilization
perspective claims that people with greater cognitive resources,
such as education or knowledge, are more likely to
make use of the referendum opportunity since they are
more motivated to participate in politics. On the other
hand, the political disaffection perspective assumes that
referendums are supported by those citizens who feel
disconnected from traditional party politics and who are
particularly critical of the political process. Thus, whereas
one perspective assumes that public support for referendums
is rooted in a general and active commitment to
democratic practices, of which referendums are just one
example, the other suspects that it is frustration with the
way politics is run that makes citizens see referendums as a
particularly effective tool to have an alternative say and to
exert actual inﬂuence.
The present study integrates these two perspectives. In
addition, we introduce new factors for each of these, i.e.
focussing on the role of political cynicism and civic duty as
additional explanations for referendum support. Finally,
since we investigate referendum support with regard to
European Union (EU) integration issues and across 21 EU
member states, we pay special attention to the role of EUrelated
attitudes and contextual factors in determining
support.
In sum, our study builds on, and extends, previous
research by investigating the role of new explanations for
both perspectives (e.g., civic duty, habitual voting, political
cynicism, EU attitudes) and also tests for the inﬂuence of
contextual factors (e.g., number of past referendums in a
country, new vs. old EU member states). We investigate this
using a unique cross-nationally comparative study design,
namely a survey conducted in 21 EU member states. Results
lend support to extant explanations in the literature but
* Corresponding author. Tel.: þ31 20 5253283.
E-mail addresses: A.R.T.Schuck@uva.nl (A.R.T. Schuck), C.H.deVreese@
uva.nl (C.H. de Vreese).
Contents lists available at ScienceDirect
Electoral Studies
journal homepage: www.elsevier.com/locate/electstud
http://dx.doi.org/10.1016/j.electstud.2015.02.012
0261-3794/© 2015 Published by Elsevier Ltd.
Electoral Studies 38 (2015) 149e158
also emphasize the need to consider new explanations and
overall suggest that referendum support is highest among
citizens who are critical of the political process as such but
feel committed to democratic practices.
2. Public support for referendums
The use of referendums is on the increase (e.g., de
Vreese, 2007; de Vreese and Schuck, 2014). This development
squares well with citizens' wishes: in general, citizens
in Western representative democracies are supportive of
referendums as a supplement to a system in which representatives
are elected. While a lot of efforts have been
devoted to theorizing about the potential threats and
beneﬁts of direct democracy (e.g., Barber, 1984; Bowler and
Donovan, 2002; Budge, 1996; Christin et al., 2002; Fishkin,
1995; LeDuc, 2003; Resnick, 1997; Sartori, 1987) and support
for direct democracy is generally high among the
public (e.g., Bowler and Donovan, 1998; Dalton et al., 2001;
Mendelsohn and Parkin, 2001), it is also clear that different
groups of citizens are more supportive than others. One
line of research suggests that the referendum opportunity
is more embraced by those who are already more politically
interested and involved (e.g., Donovan and Karp, 2006),
also dubbed the cognitive mobilization hypothesis. ‘Persons
who are interested in one form of election are also interested
in the other; almost all of those who are very or
somewhat interested in one type of election are very or
somewhat interested in the other [ … ]’ (Magleby, 1989:
99). A second line suggests that referendums are rather
supported by citizens at the margins of the political process
and those who feel disaffected with traditional party-based
politics who see referendums as an effective alternative
(e.g., Dalton et al., 2001; Gilljam et al., 1998), also dubbed
the political disaffection hypothesis.
3. The cognitive mobilization hypothesis e the role of
civic duty
The cognitive mobilization hypothesis posits that people
with more political interest and higher engagement are
more supportive of referendums. They possess greater
cognitive resources, such as education or knowledge, and
consequently are more inclined to embrace the referendum
opportunity because they are more motivated to participate
in politics. This perspective is linked to the observation
that political skills among citizens have increased over time
and, as a result, citizens demand to have more of a say and
are less willing to leave political decision-making up to
political elites (e.g., Dalton, 1984; Inglehart, 1990). Thus, in
this perspective, citizens who are more interested and
engaged in politics and possess greater cognitive skills are
seen to be the driving force behind public support for referendums
because these offer an additional opportunity for
political expression (see Schuck and de Vreese, 2011).1
However, the picture is more nuanced and Anderson
and Goodyear-Grant (2010) argue that highly informed
citizens are sometimes even sceptical of referendums.
They note that ‘an intriguing tension presents itself in
that those citizens who may be best equipped for referendum
voting, the highly informed and politically sophisticated,
are the least likely to support the use of this
tool of direct democracy’ (p. 227). This view resonates
with the concern that referendums are prone to populist
tendencies and inﬂuence (e.g., Dalton et al., 2001) which
under certain circumstances might create scepticism
among those who are most politically engaged and who
worry about an ‘uninformed’ outcome. However, this
concern might be more situational and tied to particular
referendum contexts, e.g. when these are very polarized
and/or on sensitive subjects, and not be of general nature.
Returning to the mobilization hypothesis, Rose and Borz
(2013) stress the need to see political participation as a
socialization process. They suggest that the more inclined
one is to participate in politics, the more one favours
referendums due to a self-reinforcing socialization process.
Empirically, this argument takes the shape of
including political interest in their analysis which yields a
positive and signiﬁcant effect on referendum support. In
this article, we build on this research and suggest that
reinforcing processes of political interest and previous
electoral experiences are part of generating what we may
refer to as a sense of ‘civic duty’ (see also Bowler and
Donovan, 2013).
A self-perceived sense of civic duty has been shown in
previous studies to be a key predictor of turnout (e.g., Blais,
2006; Clarke et al., 2004). Indeed this also applies to the
referendum situation. Bowler and Donovan (2013) argue
that individuals are socialized to politics differently and
this results in a baseline level of ‘duty’. For example, those
who were socialized to develop an interest in politics also
become more engaged and see politics in a more positive
light. They also attach a greater sense of duty to the act of
voting, more so than those with less political interest, and
value political engagement up to the extent of seeing it as
an obligation. As Bowler and Donovan (2013) show, civic
duty boosted turnout in the 2011 British electoral system
referendum. They also took a step further and examined
what drives that sense of duty. Here they ﬁnd that assessments
of civic duty are structured by evaluations of politics
and politicians. Low interest in politics, low political efﬁcacy,
and low regard for politicians correspond with less
support for the idea that citizens have a duty to vote on
referendums and at other elections. Some of these propositions
are also closely connected to the underlying assumptions
of the political disaffection perspective
discussed below. In the current study, we assume that civic
duty is an important predictor not just for political participation
(i.e., turnout) but also for referendum support as
referendums provide an additional opportunity to have a
say and become politically engaged and thus perform
citizenship.
Regarding the cognitive mobilization hypothesis we
propose civic duty as a new relevant factor explaining referendum
support next to other more established indicators
(speciﬁed in method section below) which have been
1
In a previous publication (Schuck and de Vreese, 2011) in West European
Politics we outlined the key perspectives. In this article we build
on this while incorporating recent research in these areas.
A.R.T. Schuck, C.H. de Vreese / Electoral Studies 38 (2015) 149e158150
tested by previous research and which we also control for
in our model.
4. The disaffection hypothesis e cynicism, the missing
link?
The political disaffection hypothesis claims that referendums
are especially supported by those citizens who
feel disconnected from traditional party politics and who
are at the margins of the political process. The assumption
is that for these citizens a referendum presents an alternative
way of expressing themselves politically and one
which they favour because they have lost conﬁdence in
the way politics is run. Previous research has shown that
support for referendums indeed often stems from citizens
with critical or sometimes even cynical attitudes towards
politics (e.g., Inglehart, 1999; Norris, 1999). Thus, in this
perspective citizens who are dissatisﬁed and disillusioned
with governments and party politics and who are at the
periphery of politics are seen to be the driving force
behind public support for referendums because these
offer an alternative and more attractive opportunity for
political expression (see Dalton et al., 2001; Donovan
et al., 2009). In sum, and different to the cognitive mobilization
perspective, support for direct democracy here is
not seen as linked to a desire for more political participation
or responsibility but instead to the dissatisfaction
with traditional political representation (e.g., Hibbing and
Theiss-Morse, 2001).
Rose and Borz (2013) address this aspect of political
disaffection in terms of political performance, mainly
grounded in performance evaluations. Such evaluations
can take different shapes. Webb (2013) distinguished between
two types of political disaffection e ‘dissatisﬁed
democratic’ and ‘stealth democratic’. Both types share a
low trust in political elites, but the dissatisﬁed democrat is
politically interested, efﬁcacious and desires greater political
participation, while the contrary is generally true of the
stealth democrat. As Webb (2013) shows, however, stealth
democrats favour direct democracy more, which he interprets
as a result of the populist nature of stealth democratic
attitudes.
We concur that evaluations of politics are indeed very
relevant. We also posit that this is quite a sophisticated
process whereby performance is evaluated by yardsticks
(see also Desmet et al., 2012), and that these evaluations
then contribute to (the presence or absence of) support for
referendums. However, we also believe it is important to
broaden the scope of evaluations we look at: the thrust of
the dissatisﬁed citizen perspective is aversion to current
politics and in recent years, this has been investigated
looking at the concept of political cynicism, however, political
cynicism so far has not been operationalized as a key
predictor for referendum support.
Political cynicism refers to a (perceived) gap between
voters and their political representatives. It ‘ … implies that
the self-interest of political actors is their primary goal and
that the common interest is secondary at best or played out
only for its political advantage’ (Cappella and Jamieson,
1997: 142). At the aggregate level it has increased over
the past decades (e.g., Jackson, 2011; Rhee, 1997). Previous
research has mostly focused on cynicism as an outcome
variable, e.g., the impact of election news coverage on
levels of political cynicism (e.g., Schuck et al., 2013;
Valentino et al., 2001). Few studies also considered cynicism
as a mediator, e.g. for the effect of strategic news
coverage on vote choice (Elenbaas and de Vreese, 2008).
The assumption underlying most of these studies is the
same - that higher levels of cynicism can alienate people
from politics, reduce learning, erode civic engagement and
result in lower levels of participation in the political process
(see e.g., Ansolabehere and Iyengar, 1995; Patterson, 1993,
2002; Valentino et al., 2001) although this view has been
challenged.
Recent research suggests that cynicism does not
necessarily need to be related to disengagement but can
rather be an indication of a kind of critical citizenship
which does not inevitably let people turn away from
politics (de Vreese and Semetko, 2002; de Vreese, 2005).
Thus, it can mean that citizens are critical or even lost
conﬁdence in traditional party politics but remain
committed to democratic practices e or are even especially
supportive of alternative democratic means such as
referendums. Picking up both these perspectives from
previous research, i.e. lower levels of cynicism being
related to higher levels of engagement and cynicism being
a form of critical citizenship, we argue that cynicism is
likely to lead to more support for the use of referendums.
Citizens who feel cynical about traditional party politics
may be more inclined to vote in referendums because for
them it represents an opportunity to have an unﬁltered
and more direct say with a concrete outcome, i.e.
bypassing those standard political practices they are
particularly critical of.
Regarding the political disaffection hypothesis we thus
propose political cynicism as an important factor explaining
referendum support alongside other established indicators
tapping into the characteristics of those at the
‘periphery of politics’ (speciﬁed in method section below).
5. Hypotheses
Based on the above considerations we propose that
above and beyond all relevant key predictors of both the
cognitive mobilization and the political disaffection hypotheses
and other EU-speciﬁc and contextual controls
civic duty and political cynicism are important predictors of
referendum support.
(H1). Higher levels of civic duty contribute to higher levels of
support for referendums [cognitive mobilization
hypothesis].
(H2). Higher levels of political cynicism contribute to higher
levels of support for referendums [political disaffection
hypothesis].
6. Study context
We test our hypotheses in the context of popular support
for referendums on issues of European integration.
Referendums on European integration issues are amongst
A.R.T. Schuck, C.H. de Vreese / Electoral Studies 38 (2015) 149e158 151
the most common in the Western world and the EU and its
treaties, currency, policies, and membership are among the
most voted on issues in the world (e.g., Hobolt, 2009). European
referendums have often been described as ‘secondorder’
elections in which domestic considerations are more
important than EU issue-related attitudes (Franklin et al.,
1994). However, this view has also been challenged
(Svensson, 2002) and EU attitudes have been shown to
become more important over time in line with the
increasing decision-making power of the EU and had a
major impact on recent referendum outcomes (Garry et al.,
2005; Glencross and Trechsel, 2011) and even mattered for
national elections (De Vries, 2007). In the current study,
this means we also include several EU-speciﬁc factors in
our analysis (e.g., general EU support, satisfaction with
government handling of EU integration issues etc.) in order
to account for the fact that voting on European issues is
becoming increasingly inﬂuenced by actual EU-speciﬁc
attitudes and considerations. Furthermore, as our investigation
spans 21 countries across Europe we also control for
the inﬂuence of relevant contextual factors. However, as
shown below (see Table 1 and Fig. 1), cross-country variation
in referendum support is rather low and we are limited
in the number of contextual factors we can include in our
analysis.
7. Data & methods
The data for this study stem from the 2009 European
Election Campaign Study.2
It is based on a two-wave panel
survey which was carried out in 21 EU member states.3
Respondents were interviewed about one month prior to
the EP elections and immediately afterwards. Fieldwork
dates were 6e18th of May and 8e19th of June 2009. The
survey was conducted using Computer Assisted Web
Interviewing (CAWI).4
The dependent variable we use is
referendum support measured in wave 2 of this survey. The
speciﬁc wording of all variables listed below can be found
in appendix A.
Country sample: The ﬁeldwork was coordinated by TNS
Opinion in Brussels and involved TNS subsidiaries in each
country. All subsidiaries comply with ESOMAR guidelines
for survey research. A total of 22,806 respondents participated
in the second wave of the survey.5
On average, 1086
respondents per country completed the questionnaire,
varying from 1001 in Austria to 2000 in Belgium. In
Belgium, 1000 Flemish respondents and 1000 Walloon respondents
completed both waves of the survey. In each
country, a sample was drawn from TNS databases. These
databases rely on multiple recruitment strategies,
including telephone, face-to-face, and online recruitment.
Each database consists of between 3600 (Slovakia) and
339,000 (the UK) individuals. Quotas (on age, gender, and
education) were enforced in sampling from the database.
The average response rate was 31% in wave 1 and the recontact
rate was on average 80% in wave 2.6
The samples
show appropriate distributions in terms of gender, age and
education compared to census data. As we are mostly
interested in the underlying relationships between variables,
we consider the deviations in the sample vis-a-vis
the adult population less problematic and we exert
Table 1
Multilevel regression model explaining EU referendum support.
Fixed effects model
B SE
Gender À0.091*** 0.024
Age À0.004*** 0.001
Education À0.031* 0.013
Political interest (EU) 0.119*** 0.008
Civic duty 0.083*** 0.008
Habitual voting 0.019 0.010
Political ideology (right) 0.071* 0.028
Political ideology (left) 0.130*** 0.030
Political efﬁcacy À0.080*** 0.011
Satisfaction w/domestic democracy 0.070*** 0.010
Satisfaction w/EU democracy 0.070*** 0.011
Political cynicism 0.198*** 0.011
Satisfaction w/government handling of EU
integration issue
À0.070*** 0.009
EU support À0.031*** 0.008
Number of past referendums 0.038 0.040
Eastern Europe À0.315* 0.159
Constant 3.362*** 0.157
Variance country level 0.111
Variance individual level 2.642
Log restricted-likelihood À37236.167
Note. Bs are unstandardized coefﬁcients from a ﬁxed-effects multilevel
model.
*p < .05; **p < .01; ***p < .001 (two-tailed); N ¼ 19,503.
2
The study was funded by the Dutch National Science Foundation (VICI
grant) and additional grants from the Danish Science Foundation, the
University of Amsterdam, and the Swedish Riksbanken Foundation.
3
The countries were the UK, France, Italy, Germany, Spain, Sweden,
Denmark, Greece, Hungary, Czech Republic, Poland, Ireland, Austria,
Portugal, Belgium (Flanders and Wallonia), Netherlands, Finland,
Lithuania, Latvia, Slovakia & Bulgaria. The country selection includes
larger and smaller member states, countries from North, South, East and
West, and long term and new members to the Union. The country selection
was ﬁnalized based on feasibility.
4
Fieldwork started on May 6 in all countries. In the UK and Ireland
data collection ﬁnished on May 11, in France, Italy, Germany, Spain,
Sweden, Greece, Czech Republic, Austria, Portugal, Netherlands, Finland
and Slovakia on May 12, in Hungary, Poland and Latvia on May 13, in
Denmark and Belgium on May 14, in Lithuania on May 15 and in Bulgaria
on May 18. In Slovakia data collection ﬁnished on June 11, in Italy, Germany,
Sweden, Czech Republic, Lithuania, and the Netherlands on June
12, in Ireland and the UK on June 13, in France, Poland and Austria on June
14, in Spain, Denmark, Greece, Hungary, Portugal, Finland, and Latvia on
June 15 and Belgium on June 19.
5
The lower number of respondents in our reported model is due to
missing data on some of the variables included in the analysis.
6
The response rates vary from 19% (Denmark) to 63% (Lithuania) in
wave 1 and the re-contact rate between 67% (Latvia) and 89%
(Hungary). An analysis of the non-participation (i.e. respondents who
were invited but did not participate or not complete the interview)
showed that non-respondents were younger, included more men
compared to women in the UK, Sweden and Denmark and more
women in Hungary, Italy, the Netherlands, Portugal and Austria. Concerning
education, the pool of non-respondents was signiﬁcantly lower
educated in six countries (Spain, Denmark, Czech Republic, Poland, the
Netherlands and Finland).
A.R.T. Schuck, C.H. de Vreese / Electoral Studies 38 (2015) 149e158152
appropriate caution when making inferences about absolute
values.7
Questionnaire and translations: The questionnaire was
developed in English and translated into the different
national languages. It was then translated back into English
as an additional check of the accuracy of the translations.
The translation was supervised by the research
team and it was carried out by TNS (which also executes
and translates the Eurobarometer surveys). Irregularities
and problems arising from this process were resolved by
deliberation.8
7.1. Measures
7.1.1. Dependent variable
Respondents were asked to what extent they are against
or in favour of holding referendums on matters of European
integration (1-strongly against EU referendums, 7-strongly
in favour of EU referendums) (M ¼ 4.95, SD ¼ 1.72).
7.1.2. Independent variables
Civic duty is assessed with four items on 7-point Likert
scales asking respondents to what extent they disagree (1)
or agree (7) that it is a citizens' duty to participate in
elections, feel a sense of satisfaction when they vote, would
feel guilty if they do not vote and think they would neglect
their duty as citizen if they would not vote (M ¼ 4.56,
SD ¼ 1.81, Cronbach's alpha ¼ 0.87).
Political cynicism was measured with four items on
7-point Likert scales on which higher scores represent
stronger levels of cynicism, asking respondents to what
extent they disagreed or agreed that politicians are in
politics for what they can get out of it personally, will sell
out on their ideals and break their promises if it will increase
their power, are truthful with voters (reverse coding)
and are dedicated and should be thanked for their
work (reverse coding) (M ¼ 5.33, SD ¼ 1.24, Cronbach's
alpha ¼ 0.78).9
7.2. Individual controls
To be in line with extant research we control for age
(M ¼ 39.30, SD ¼ 13.21), gender (53.3% female), and education
(three categories from 1-low to 3-high based on
obtainable degrees in a country) in our model. We also
control for factors at the country level, however, variation
on this level was limited (see below).
In line with the cognitive mobilization perspective and
previous research, younger age and higher education are
expected to be positively related to referendum support
(see Schuck and de Vreese, 2011). Furthermore, we include
political interest which is expected to be positively related
to referendum support as well (see Donovan and Karp,
2006; Rose and Borz, 2013). Interest in the current study
is measured with one item on a 7-point Likert scale asking
respondents how interested they are in political issues
Fig. 1. EU referendum support across 21 countries. Note. Bars indicate average levels of EU referendum support in the respective member states with higher
numbers indicating stronger support (M ¼ 4.95, SD ¼ 1.72); N ¼ 22,806.
7
An overview of the composition of our sample vis-a-vis census data
per country showed no differences between the adult population and the
sample in terms of gender in Austria, Ireland, Slovakia and Spain. The
mean difference between the population and the sample was 2.76%
(SD ¼ 3.43%). Small deviations occurred (0e8%), with sometimes women
overrepresented and sometimes men. One substantial overrepresentation
(of women) by 15% occurred in Latvia. Young citizens were generally
slightly overrepresented in the samples. The share of young citizens
(under 35) deviated 9.62% on average (SD ¼ 8.95%), with a minimum of
1% absolute deviation (overrepresentation) in Italy and Sweden, and a
maximum of 34% (overrepresentation) in Greece. The largest underrepresentation
of younger citizens was found in Latvia (14%). The share of
older citizens (55þ) (M ¼ 16.57%, SD ¼ 9.91%) deviated from 1% in France
and the UK (overrepresentations) to 33% (underrepresentation in Greece).
Most countries had slight underrepresentation of older citizens. The
German sample had the largest overrepresentation (2%). In terms of education
(collapsed in three categories, following the European Social
Survey), the sample reﬂected the population in Ireland and Spain, while
deviations (M ¼ 8.12%, SD ¼ 8.37%) were found in other countries with
higher educated citizens being overrepresented in the samples. Underrepresentations
were found in Greece (1%), France (6%), Slovakia (5%)
and Sweden (11%). For more information, see de Vreese et al., 2010.
8
For more information, see de Vreese et al., 2009.
9
Political cynicism is conceptually and empirically distinct from political
efﬁcacy and both concepts correlate only weakly with another
(r ¼ À0.27).
A.R.T. Schuck, C.H. de Vreese / Electoral Studies 38 (2015) 149e158 153
regarding the European Union (EU) with higher scores
representing stronger interest (M ¼ 4.37, SD ¼ 1.59).
Habitual voting is included to provide a more conservative
test of our key independent variable civic duty which is
conceptually and empirically distinct, albeit related to
some extent, in its focus on motivational considerations
rather than habitual behaviour. Habitual voting in the
current study is assessed with one item asking respondents
how often they vote considering all elections they are
eligible to participate in (1-never, 7-every election)
(M ¼ 5.83, SD ¼ 1.60).10
In line with the political disaffection perspective and in
line with previous research, we include political efﬁcacy in
our model, which was assessed with four standard items
on 7-point Likert scales on which higher scores represent
higher levels of efﬁcacy comprising both internal and
external efﬁcacy with regard to EU politics (M ¼ 3.52,
SD ¼ 1.25, Cronbach's alpha ¼ 0.68), and which is expected
to be strongly and negatively related to referendum support
(see Dalton et al., 2001; Schuck and de Vreese, 2011).
Furthermore, we control for left and right political ideology
which is expected to be related to stronger referendum
support (see Donovan and Karp, 2006). Tapping political
ideology on a 11-point scale (1-left, 11-right) we recoded
this variable into two dummy variables, one for left political
ideology (1e4) (26.4%) and one for right political
ideology (8e11) (32.7%) with centre political ideology as
the reference category in our model. Satisfaction with democracy
is expected to be negatively related to referendum
support (Rose and Borz, 2013) and was measured
with two single items, asking respondents how satisﬁed
they were with the way democracy works in their country
(M ¼ 3.72, SD ¼ 1.69) and in the EU (M ¼ 3.77, SD ¼ 1.41).11
Furthermore, given that in our study we aim to explain
support for referendums on EU integration issues (i.e. not
domestic issues) we also include several new EU-speciﬁc
indicators as part of the political disaffection perspective,
i.e. satisfaction with the handling of the EU integration
issue by the national government and general
support for the EU. According to the political disaffection
hypothesis and previous research including other EUrelated
indicators these factors should be expected to be
negatively related to referendum support (see Rose and
Borz, 2013). Satisfaction with government handling EU
integration is measured with one item asking respondents
how well they think their own national government is
handling the issue of EU integration, with higher scores
indicating greater satisfaction (M ¼ 3.46, SD ¼ 1.54). EU
support was measured with one standard Eurobarometer
question on a 7-point Likert scale asking respondents to
what extent they disagree (1) or agree (7) that EU membership
of their country was a good thing (M ¼ 4.61,
SD ¼ 1.70).
7.3. Contextual controls
In our model we control for the number of past referendums
on EU integration issues in a country (ranging from
0 to 8, M ¼ 1.26, SD ¼ 1.83) as well as for old Western vs.
new Eastern European member states (i.e. those which
joined the EU after 2004). For the latter variable we
included a dummy variable indicating geographic afﬁliation
to Eastern Europe (seven countries in our sample).
7.4. Data analysis
We report multilevel regression results from a ﬁxedeffects
multilevel model estimated with restricted
maximum likelihood (REML) which allows us to assess
both individual and contextual factors simultaneously (see
Gelman and Hill, 2007). Referendum support is our
dependent variable and we include socio-demographics
and the two sets of cognitive mobilization and political
disaffection indicators including civic duty and political
cynicism as independent variables in these models.
8. Results
As Fig. 1 illustrates, the degree of individual level referendum
support varies across countries, however, not a lot
(M ¼ 4.95, SD ¼ 1.72).12
It is generally high and highest in
Bulgaria (M ¼ 5.73, SD ¼ 1.53), Ireland (M ¼ 5.45,
SD ¼ 1.73), Portugal (M ¼ 5.43, SD ¼ 1.61) and France
(M ¼ 5.38, SD ¼ 1.69) and lowest in Latvia (M ¼ 4.13,
SD ¼ 1.81) and Hungary (M ¼ 4.19, SD ¼ 1.71).
Turning to our hypotheses and looking at some bivariate
descriptives ﬁrst, we see that referendum support is indeed
higher on average among respondents with high levels of
civic duty (M ¼ 5.16, SD ¼ 1.71) and high levels of political
cynicism (M ¼ 5.10, SD ¼ 1.81) compared to respondents
with low levels of civic duty (M ¼ 4.73, SD ¼ 1.72) and low
levels of political cynicism (M ¼ 4.81, SD ¼ 1.63) (see Fig. 2).
Fig. 2. EU referendum support by levels of civic duty and political cynicism.
Note. Bars indicate average levels of EU referendum support among respondents
with low (M ¼ 4.73, SD ¼ 1.72) and high (M ¼ 5.16, SD ¼ 1.71)
civic duty and low (M ¼ 4.81, SD ¼ 1.63) and high (M ¼ 5.10, SD ¼ 1.81)
political cynicism (median split).
10
Habitual voting and civic duty are only moderately correlated
(r ¼ 0.50).
11
Both items have been included as in previous models explaining
support for referendums (Schuck and de Vreese, 2011) and correlate only
moderately with each other (r < 0.50). 12
Ranging from 4.13 in Latvia to 5.73 in Bulgaria.
A.R.T. Schuck, C.H. de Vreese / Electoral Studies 38 (2015) 149e158154
We now turn to the results of our multivariate analysis.
Based on the empty model the intraclass correlation is .050,
which means that about 5 percent of the variance in our
dependent variable is on the country level. This is not a lot
but still worth considering.13
Turning to civic duty and the
cognitive mobilization hypothesis ﬁrst, we see in Table 1
above that a stronger sense of civic duty is indeed related
to stronger referendum support, providing support for H1.
Habitual voting is unrelated to referendum support. Of the
other standard predictors, age and political interest have
the assumed impact. Younger citizens and those more
interested in EU politics are more supportive. However,
education has a marginally negative impact, i.e. support is
stronger among the lower educated compared to the
higher educated, which already points to the generally
stronger support for the political disaffection hypothesis
which we will discuss next.
Regarding political cynicism and the political disaffection
hypothesis we see a strong effect of cynicism on referendum
support. Thus, citizens who are cynical about traditional
party politics and who do not believe that politicians
are reliable show stronger support for referendums, supporting
H2. Furthermore, most of the other standard predictors
of the political disaffection hypothesis show the
assumed impact. In line with previous research we ﬁnd
support for the expectation that people with more extreme
political ideology are more supportive of referendums, both
on the right as well as especially on the left side of the
political spectrum. As also shown in previous research,
political efﬁcacy is a key variable in explaining referendum
support and has a strong negative effect. Thus, it is especially
those who feel inefﬁcacious when it comes to traditional
party-based politics, i.e. those feeling that their voice
usually does not count and that government is not
responsive, who embrace referendums as an alternative
opportunity to have their say. Related to this observation,
our analysis also shows that EU-related attitudes and policy
evaluations indeed do matter for referendum support in
Europe. Low levels of EU support and low levels of satisfaction
with how the national government is handling the
EU integration issue lead to stronger support for referendums.
At the same time, satisfaction with domestic and EU
democracy in general are both positively related to referendum
support.
Thus, overall we ﬁnd support for both of our new proposed
indicators, civic duty and political cynicism. Both of
them matter for referendum support and add to the
explanation who supports referendums and why. Furthermore,
we ﬁnd general support for both perspectives,
cognitive mobilization and political disaffection, on the
individual level. Overall, support is stronger for the political
disaffection perspective, however, there are connections
between both perspectives which we will further discuss
below.
On the contextual level, the variance is rather low, as
was also already indicated by Fig.1 above. As we can see the
number of past referendums plays no role but support is
overall slightly higher in the old Western European member
states compared to the new Eastern European member
states.14
9. Discussion
It is a paradoxical situation that while referendums are
increasingly popular, they are also increasingly contested.
The discussion is waged in both the scholarly literature and
in societal debates. Can we trust that referendum outcomes
are well informed and carried by a qualiﬁed majority or are
they prone to populist tendencies, hijacked by particular
interests and/or turned into punishment traps by an angry
citizenry frustrated with traditional party politics? We still
have only limited knowledge about which citizens support
the use of referendums and why. These questions are
essential since the answers help us understand the dynamics
of such direct democratic means and they are
telling about the quality of our democracies. At the core of
extant research we may distil two major schools of
thoughts, one suggesting that it is primarily those who are
most politically engaged who support in referendums, and
the other suggesting that it is rather those at the periphery
of politics that exhibit support.
In this study, based on a large-scale survey spanning 21
European countries, we introduce civic duty and political
cynicism as two, thus far neglected but important factors
determining public support for referendums. We hereby
add to two rivalling perspectives on this topic according to
which either cognitive mobilization or public disaffection
are the key driving force behind popular support for referendums.
Identifying and theorizing about additional
factors which help explain referendum support is important
given that almost all extant models suffer from very
low explained variance. The ﬁrst of the factors we identiﬁed,
civic duty, relates to the cognitive mobilization
perspective in the literature, which posits that referendum
support is strongest among those already most engaged in
politics (e.g., Clarke et al., 2004). We ﬁnd support for this
notion which we interpret to be an indication of the motivational
aspects underlying the sense of duty (i.e., including
a sense of satisfaction and self-assurance).
The second factor, political cynicism, is well established
in the literature (e.g., Patterson,1993, 2002; Valentino et al.,
2001) and has most often been accused of turning citizens
off (Cappella and Jamieson, 1997). However, another
perspective sees cynicism as a form of critical citizenship
and claims that citizens can well be cynical and engaged at
the same time (de Vreese and Semetko, 2002). Our ﬁndings
13
Empty model (not reported here): Variance country level: .147;
variance individual level: 2.779.
14
Compared to the empty model the ﬁnal overall model reported in
Table 1 performs signiﬁcantly better (decrease in LL ¼ 444,582 with
df¼16). The model explains 24.4 percent of the variance at the country
level and 4.9 percent of the variance at the individual level. This is not a
lot but in line with previous studies explaining public support for referendums
(also see discussion section). We also tested a range of other
possible contextual controls which we do not report here and which
could not all be included in the model due to the restriction on the
number of possible country level variables given the number of countries
included in the analysis. Among other contextual variables the length of
EU membership (in years) and quality of democracy (index) showed to
not affect referendum support.
A.R.T. Schuck, C.H. de Vreese / Electoral Studies 38 (2015) 149e158 155
show that, in line with the political disaffection perspective,
political cynicism is indeed a strong predictor for referendum
support. Those frustrated with and cynical about
traditional party-based politics embrace the referendum
opportunity as an alternative way of having a more unﬁltered
say regarding a particular policy issue. That said, our
ﬁndings suggest that the story is less straightforward.
Indeed, support for referendums is driven by dissatisfaction
with the political process and politicians, a lack of
perceived self-efﬁcacy, and negative policy evaluations.
However, support is also higher among citizens with
greater interest and satisfaction with democracy.
The overall picture which emerges from our analysis
suggests that both leading perspectives, cognitive mobilization
and political disaffection, receive support, however,
political disaffection matters more overall. In general, we
do not believe that these two perspectives should be seen
as mutually excluding explanations but rather as complementary
to each other. In this respect political cynicism
plays an interesting role. Cynicism is often seen as problematic,
but our ﬁndings suggest that support is strongest
among cynical voters who are critical towards party-based,
representative democracy and who feel that usually their
voice is not being heard or EU politics is going the wrong
way but who at the same time remain committed to democratic
principles. These ‘critical citizens’ embrace referendums
as a more reliable tool to exert democratic
inﬂuence and as an alternative way of having a more direct
say which for them seems ‘worth it’ to engage in. In this
perspective, referendum support would not be driven by
sore ‘losers’ in the political process or generally disengaged
outsiders but by critical and concerned citizens who are
dissatisﬁed with the way politics is run but care about the
whole.
Aside from civic duty and political cynicism and other
established relevant factors we also considered several EUspeciﬁc
and performance-related evaluations as additional
explanatory factors and these also showed to affect referendum
support. In more general terms, EU-issue voting is
gaining in prominence and citizens vote more based on
actual EU-speciﬁc considerations in European Parliamentary
elections and referendums (e.g., Hobolt, 2009; van
Spanje and de Vreese, 2011) and the EU even matters
more in national elections (De Vries, 2010), especially in
recent years and due to the prominence of the ﬁnancial
crisis in the EU. This stresses the need to indeed step
beyond the current standard set of indicators when
explaining popular support for referendums and consider
factors more closely related to the actual context in which
referendums take place, such as the EU (see also Rose and
Borz, 2013). Next to the topical context (EU integration),
we also explored the inﬂuence of country-speciﬁc characteristics
and how such type of context affects referendum
support. However, there appears to be little variation across
countries and we merely control for certain possible inﬂuences
on this level and our ﬁndings in this regard are not
too conclusive and further work needs to be done in terms
of theorizing which factors should be considered here.
Previous research has identiﬁed the media as relevant
factor in changing individual referendum support, e.g. over
the course of a referendum campaign (Schuck and de
Vreese, 2011), which has not been the focus of the current
study. We do believe that referendum support is not a
stable construct and is susceptible to change, especially in
light of changes in the information environment. An
agenda for future research will be to identify the content
features in media coverage which exert an inﬂuence on
changes in individual-level referendum support. This does
not necessarily have to be evaluations of the referendum
instrument as such but could also be more indirect and
expand to other aspects, such as, for example, the quality of
debate during a referendum campaign, or evaluations of
democratic performance within the context the referendum
takes place in (such as the EU or other), or evaluations
attached to the referendum topic, or any content features
with the potential to inﬂuence the factors relevant to
explain referendum support (such as political cynicism
etc.). In addition, future research should also consider the
role of social media. This role may be double-edged, i.e.
social media may on the one hand affect how citizens look
at referendums as such while they may of course also play a
role in mobilizing and affecting vote preferences (see this
discussion in Quinlan et al. this issue). Furthermore, the
impact of traditional mass media as well as social media is
likely to be conditional on other contextual changes. More
work is still needed to disentangle these dynamics. These
challenges and open questions notwithstanding we believe
the current study has made an important contribution in
identifying, theorizing, and testing civic duty and political
cynicism as two additional factors helping us to better
understand who supports referendums and why.
Acknowledgements
The authors thank Rens Vliegenthart for valuable suggestions
and comments regarding the analysis presented in
this paper.
Appendix A. Overview of variables
Referendum support: One item on a 7-point Likert
scale: ‘In referendums, people can vote against or in favour
of, for example, a proposal or a treaty. Some people think
that it is a good idea to hold referendums on matters of
European integration, others do not think that this is a good
idea. What is your position on this issue?’ (1-strongly
against EU referendums, 7-strongly in favour of EU referendums)
(M ¼ 4.95, SD ¼ 1.72).
Political cynicism: Four items on 7-point Likert scales:
(1) ‘Almost all politicians will sell out their ideals or break
their promises if it will increase their power’; (2) ‘Most
politicians are in politics for what they can get out of it
personally’; (3) ‘Most politicians are truthful with the
voters’ (reverse coding); (4) ‘Most politicians are dedicated
and we should be grateful to them for the work they do’
(reverse coding) (M ¼ 5.33, SD ¼ 1.24, Cronbach's
alpha ¼ 0.78).
Civic duty: Four items on 7-point Likert scales: (1) ‘It's
every citizen's duty to vote in an election’; (2) ‘I would be
seriously neglecting my duty as a citizen if I didn't vote’; (3)
‘I feel a sense of satisfaction when I vote’; (4) ‘I would feel
very guilty if I didn't vote in an election’; (1-strongly
A.R.T. Schuck, C.H. de Vreese / Electoral Studies 38 (2015) 149e158156
disagree, 7-strongly agree) (M ¼ 4.56, SD ¼ 1.81, Cronbach's
alpha ¼ 0.87).
Gender: Male ¼ 0; female ¼ 1 (53.3%).
Age: Measured in years (M ¼ 39.30, SD ¼ 13.21).
Education: Measured with country-speciﬁc lists indicating
obtainable educational degrees and recoded into
three categories comparable across countries from lowest
to highest: (1) low (51.4%); (2) medium (9.3%), (3) high
(39.3%).
Political interest: One item on a 7-point Likert scale:
‘How interested are you in European Union (EU) issues?’;
(1-not at all interested, 7-very interested) (M ¼ 4.37,
SD ¼ 1.59).
Habitual voting: One item: ‘Some people vote in every
election, other people never vote. Of all elections in which
you were eligible to vote, how many times have you actually
voted?’; (1-Never, 2-Hardly ever, 3-Slightly less than
half of the times, 4-About half of the times, 5-Slightly more
than half of the times, 6-Almost every time, 7-Every time)
(M ¼ 5.83, SD ¼ 1.60).
Political ideology: Respondents were asked to indicate
their political orientation on an 11-point lefteright scale.
Two dummy variables were built representing respondents
with left political ideology (1e4) (26.4%) or right political
ideology (8e11) (32.7%) with centrist orientation (5e7) as
the reference category.
Political efﬁcacy: Four items on 7-point Likert scales:
(1) ‘People like me don't have any say about what the EU
does’ (reverse coding); (2) ‘I don't think the EU cares much
what people like me think’ (reverse coding); (3) ‘Having
European Parliamentary elections makes the EU pay
attention to what people think’; (4) ‘Over the years, the EU
pays a lot of attention to what people think when it decides
what to do’; (1-strongly disagree, 7-strongly agree)
(M ¼ 3.52, SD ¼ 1.25, Cronbach's alpha ¼ 0.68).
Satisfaction with democracy: Two separate items
focussing on satisfaction with domestic democracy and EU
democracy on 7-point Likert scales: (1) ‘Regardless of who
is in government, on the whole, how satisﬁed or dissatisﬁed
are you with the way democracy works in (COUNTRY)?’
(M ¼ 3.72, SD ¼ 1.69); (2) ‘How satisﬁed or
dissatisﬁed are you with the way democracy works in the
European Union?’ (M ¼ 3.77, SD ¼ 1.41); (1-very dissatisﬁed,
7-very satisﬁed).
Satisfaction with government handling of EU integration:
One item on a 7-point Likert scale: ‘How well do you
think the government is handling the issue of European
integration?’; (1-very poorly, 7-very well) (M ¼ 3.46,
SD ¼ 1.54).
EU support: One item on a 7-point Likert scale:
‘(COUNTRY)’s membership of the European Union is a good
thing’ (1-strongly disagree, 7-strongly agree) (M ¼ 4.61,
SD ¼ 1.70).
Number of past referendums: The number of past referendums
on EU integration issues in a country, ranging
from 0 to 8 in our sample (M ¼ 1.26, SD ¼ 1.83).
Eastern Europe: Old Western vs. new Eastern European
member states (joined the EU after 2004). Dummy variable
indicating geographic afﬁliation to Eastern Europe (seven
countries in our sample).
References
Anderson, C., Goodyear-Grant, E., 2010. Why are highly informed citizens
skeptical of referenda. Elect. Stud. 29, 227e238.
Ansolabehere, S., Iyengar, S., 1995. Going Negative: How Political Ads
Shrink and Polarize the Electorate. The Free Press, New York.
Barber, B., 1984. Strong Democracy: Participatory Politics for a New Age.
University of California Press, Berkeley.
Blais, A., 2006. What affects voter turnout? Annu. Rev. Polit. Sci. 9,
111e125.
Bowler, S., Donovan, T., 2002. Democracy, institutions and attitudes about
citizen inﬂuence on government. Br. J. Polit. Sci. 32, 371e390.
Bowler, S., Donovan, T., 2013. Civic duty and turnout on the UK referendum
on AV: what shapes the duty to vote? Elect. Stud. 32, 265e273.
Bowler, S., Donovan, T., 1998. Demanding Choices: Opinion, Voting, and
Direct Democracy. University of Michigan Press, Ann Arbor, MI.
Budge, I., 1996. The New Challenge of Direct Democracy. Polity Press,
Cambridge.
Cappella, J., Jamieson, K., 1997. Spiral of Cynicism. The Press and the Public
Good. Oxford University Press, New York.
Christin, T., Hug, S., Sciarini, P., 2002. Interests and information in referendumvoting:
an analysis of Swiss voters. Eur. J. Polit. Res. 41, 759e776.
Clarke, H., Sanders, S., Stewart, M., Whiteley, P., 2004. Political Choice in
Britain. Oxford University Press, Oxford.
Dalton, R.J., Buerklin, W., Drummond, A., 2001. Public opinion and direct
democracy. J. Democr. 12, 141e153.
Dalton, R.J., 1984. Cognitive mobilization and partisan dealignment in
advanced industrial democracies. J. Polit. 46, 264e284.
de Vreese, C.H., Semetko, H., 2002. Cynical and engaged: strategic
campaign coverage, public opinion and mobilization in a referendum.
Commun. Res. 29, 615e641.
de Vreese, C.H., 2005. The spiral of cynicism reconsidered. Eur. J. Commun.
20, 283e301.
de Vreese, C.H., 2007. The Dynamics of Referendum Campaigns in International
Perspective. Palgrave, London.
de Vreese, C.H., van Spanje, J., Schuck, A.R.T., Boomgaarden, H.,
Vliegenthart, R., Azrout, R., Elenbaas, M., 2009. European Election
Campaign Study, Scenario I: Data and Documentation. University of
Amsterdam, Amsterdam, the Netherlands.
de Vreese, C.H., Schuck, A.R.T., 2014. Political communication in referendums.
In: Reinemann, C. (Ed.), Political Communication. Handbook of
Communication Science. De Gruyter, Berlin, pp. 129e150.
De Vries, C.E., 2010. EU issue voting: asset or liability? How European integration
affects parties' electoral fortunes. Eur. Union Polit. 11, 89e117.
De Vries, C.E., 2007. Sleeping Giant: fact or fairytale? How European
integration affects national elections. Eur. Union Polit. 8, 363e385.
Desmet, P., van Spanje, J., de Vreese, C.H., 2012. ‘Second-order’ institutions:
national institutional quality as a yardstick for EU evaluation.
J. Eur. Public Policy 19, 1071e1088.
Donovan, T., Karp, J., 2006. Popular support for direct democracy. Party
Polit. 12, 671e688.
Donovan, T., Tolbert, C., Smith, D., 2009. Political engagement, mobilization,
and direct democracy. Public Opin. Q. 73, 98e118.
Elenbaas, M., de Vreese, C.H., 2008. The effects of strategic news coverage
on political cynicism and vote choice among young voters. J. Commun.
58, 550e567.
Fishkin, J., 1995. The Voice of the People: Public Opinion and Democracy.
Yale University Press, New Haven, CT.
Franklin, M., Marsh, M., Wlezien, C., 1994. Attitudes towards Europe and
referendum votes: a response to Siune and Svensson. Elect. Stud. 13,
117e121.
Garry, J., Marsh, M., Sinnott, R., 2005. ‘Second-order’ versus ‘issue-voting’
effects in EU referendums: evidence from the Irish nice treaty referendums.
Eur. Union Polit. 6, 201e221.
Glencross, A., Trechsel, A., 2011. First or second order referendums? Understanding
the votes on the EU Constitutional Treaty in four EU
member states. West Eur. Polit. 34, 755e772.
Gelman, A., Hill, J., 2007. Data Analysis Using Regression and Multilevel/
Hierarchical Models. Cambridge University Press, Cambridge.
Gilljam, M., Pesonen, P., Listhaug, O., 1998. The referendum in representative
democracies. In: Jenssen, A.D., Pesonen, P., Gilljam, M. (Eds.), To
Join or Not to Join: Three Nordic Referendums on Membership in the
European Union. Scandinavian University Press, Oslo, pp. 284e306.
Hibbing, J., Theiss-Morse, E., 2001. Process preferences and American
politics: what the people want government to be. Am. Polit. Sci. Rev.
95, 145e153.
Hobolt, S., 2009. Europe in Question: Referendums on European Integration.
Oxford University Press, Oxford.
A.R.T. Schuck, C.H. de Vreese / Electoral Studies 38 (2015) 149e158 157
Inglehart, R., 1990. Culture Shift in Advanced Industrial Society. Princeton
University Press, Princeton, NJ.
Inglehart, R., 1999. Postmodernization erodes respect for authority, but
increases support for democracy. In: Norris, P. (Ed.), Critical Citizens:
Global Support for Democratic Governance. Oxford University Press,
Oxford, pp. 236e256.
Jackson, D., 2011. Strategic media, cynical public? Examining the contingent
effects of strategic news frames on political cynicism in the
United Kingdom. Int. J. Press/Polit. 16, 75e101.
LeDuc, L., 2003. The Politics of Direct Democracy: Referendums in Global
Perspective. Broadview Press, Peterborough, ON.
Magleby, D., 1989. Opinion formation and opinion change in ballot
proposition campaigns. In: Margolis, M., Mauser, G. (Eds.), Manipulating
Public Opinion. Brooks/Cole Publishing Company, Belmont.
Mendelsohn, M., Parkin, A., 2001. Referendum Democracy: Citizens, Elites,
and Deliberation in Referendum Campaigns. Palgrave, New York.
Norris, P., 1999. Critical Citizens: Global Support for Democratic Governance.
Oxford University Press, Oxford.
Patterson, T., 1993. Out of Order. Knopf, New York.
Patterson, T., 2002. The Vanishing Voter: Public Involvement in an Age of
Uncertainty. Alfred Knopf, New York.
Resnick, P., 1997. Twenty-ﬁrst Century Democracy. McGill-Queen’s,
Montreal.
Rhee, J., 1997. Strategy and issue frames in election campaign coverage: a
social cognitive account of framing effects. J. Commun. 47, 26e48.
Rose, R., Borz, G., 2013. What determines demand for European Union
referendums? J. Eur. Integr. 35, 619e633.
Sartori, G., 1987. The Theory of Democracy Revisited. Chatham House,
Chatham, NJ.
Schuck, A.R.T., de Vreese, C.H., 2011. Public support for referendums: the
role of the Media. West Eur. Polit. 34, 181e207.
Schuck, A.R.T., Boomgaarden, H., de Vreese, C.H., 2013. Cynics all around?
The impact of election news on political cynicism in comparative
perspective. J. Commun. 63, 287e311.
Svensson, P., 2002. Five Danish referendums on the European Community
and European Union: a critical assessment of the Franklin thesis. Eur.
J. Polit. Res. 41, 733e750.
Valentino, N., Beckmann, M., Buhr, A., 2001. A spiral of cynicism for some:
the contingent effects of campaign news frames on participation and
conﬁdence in government. Polit. Commun. 18, 347e367.
Van Spanje, J., de Vreese, C.H., 2011. So what's wrong with the EU? Motivations
underlying the Eurosceptic vote in the 2009 European
elections. Eur. Union Polit. 12, 405e429.
Webb, P., 2013. Who is willing to participate? Dissatisﬁed democrats,
stealth democrats and populists in the United Kingdom. Eur. J. Polit.
Res. 52, 747e772.
A.R.T. Schuck, C.H. de Vreese / Electoral Studies 38 (2015) 149e158158