Full Terms & Conditions of access and use can be found at
https://www.tandfonline.com/action/journalInformation?journalCode=ccpo20
Contemporary Politics
ISSN: 1356-9775 (Print) 1469-3631 (Online) Journal homepage: https://www.tandfonline.com/loi/ccpo20
And yet it matters: referendum campaigns and
vote decision in Eastern Europe
Sergiu Gherghina & Nanuli Silagadze
To cite this article: Sergiu Gherghina & Nanuli Silagadze (2019) And yet it matters: referendum
campaigns and vote decision in Eastern Europe, Contemporary Politics, 25:1, 29-46, DOI:
10.1080/13569775.2018.1543790
To link to this article: https://doi.org/10.1080/13569775.2018.1543790
Published online: 08 Nov 2018.
Submit your article to this journal
Article views: 206
View related articles
View Crossmark data
Citing articles: 3 View citing articles
And yet it matters: referendum campaigns and vote decision
in Eastern Europe
Sergiu Gherghina a
and Nanuli Silagadze b
a
Politics, University of Glasgow, Glasgow, UK; b
Political Science, Abo Akademi, Turku, Finland
ABSTRACT
Referendums campaigns are important and earlier research closely
analysed their general functioning, effects on turnout, and the
importance of media and information for voting behaviour.
However, the role of referendum campaigns as such (with all its
components) in shaping voting behaviour was widely neglected.
This article seeks to partially fill this gap in the literature and
argues that referendum campaign is an important predictor of the
voting decision as long as people perceive it as informative and
follow it. We investigate this effect in the context of three
referendums organized in 2015–2016 in Bulgaria, Poland and
Slovakia. The results indicate that these two variables explain the
decision of citizens to support referendums across different
settings. Their effects are consistent and significantly stronger
than alternative explanations employed in the literature such as
the limited effect of campaigns, second-order elections, partisan
cues or amount of information received.
KEYWORDS
Referendum campaigns;
policies; voting; Eastern
Europe
Introduction
Referendum campaigns differ from election campaigns in a number of ways: run longer
and thus increase the likelihood of unforeseen events, are influenced by domestic and
international factors, have higher levels of electoral volatility and uncertainty among
voters, lower perceived importance and involvement, and political parties may be internally
divided over the policy subjected to popular vote (de Vreese & Semetko, 2004; Franklin,
Marsh, & Wlezien, 1994; Hobolt, 2005; LeDuc, 2002; Silagadze and Gherghina 2018). For
all these reasons, referendum campaigns are considered influential and earlier research
closely analysed their general functioning, effects on turnout, and the importance of
media and information for voting behaviour (Christin, Hug, & Sciarini, 2002; de Vreese,
2004; de Vreese & Semetko, 2004; Kriesi, 2006; Schuck & de Vreese, 2009; Siune & Svensson,
1993). In spite of the consensus regarding the importance of referendum campaigns, little
scholarly attention has been devoted to them as explanatory variables for the way in
which people vote in referendums. So far, most approaches followed either the path of
policy and institutional explanations that ignored campaign or the path of intra-campaign
features and developments that left out the campaign as a general process.
© 2018 Informa UK Limited, trading as Taylor & Francis Group
CONTACT Sergiu Gherghina sergiulor@yahoo.com
CONTEMPORARY POLITICS
2019, VOL. 25, NO. 1, 29–46
https://doi.org/10.1080/13569775.2018.1543790
In this sense, a broader line of enquiry sought to explain voting behaviour along the
debate issue voting versus second-order voting with empirical evidence supporting alternatively
both theories (Mendez, Mendez, & Triga, 2014). Issue-voting implies that the citizens’
vote is influenced by their attitudes towards the policy to be decided upon (Siune &
Svensson, 1993; Siune, Svensson, & Tonsgaard, 1994). The second-order voting postulates
that the attitudes towards the national political parties in general and to the incumbent
government, in particular, are decisive for the vote. Referendums thus become secondorder
national elections where considerations about first-order national politics determine
the political behaviour of citizens (Franklin et al., 1994; Franklin, 2002; Hobolt, 2007; LeDuc,
2002). In addition to these views, previous research reveals the importance of political cues
communicated by elites during the referendum campaign especially for voters with little
prior information (Lupia, 1994; Zaller, 1992). The lower the level of information among the
electorate on an issue, the more determining the political cues for their final decision. Party
endorsements serve as heuristics that help voters to arrive at competent decisions despite
their lack of factual knowledge (Font & Rodriguez, 2009; Hobolt, 2007; LeDuc, 2009).
The line of enquiry focusing on intra-campaign features investigated mainly the reasons
for which citizens oppose the referendum initiatives. The findings indicate that voting
decision was influenced by the type and shape of messages – including here media
framing – received during campaign (Hobolt, 2005; Schuck & de Vreese, 2009), the low
level of information that made them oppose the status quo in the presence of a particular
heuristics (Christin et al., 2002; Highley & McAllister, 2002; Kriesi, 2005) or to the risk aversion
that determines voters to prefer safe and known over uncertain options (Hobolt,
2009). In the particular case of European referendums, another feature emerges, namely
the challenge of a two-dimensional political space. The domestic and transnational dimensions
are interlinked, both playing a role in constituting a two-dimensional political space
in which the political parties operate and the campaign takes place (Shu, 2009).
The rich empirical evidence provided along these two lines of enquiry neglected the
role of referendum campaign as such (with all its components) in shaping voting behaviour.
To partially fill this gap in the literature, our paper argues that referendum campaign
is an important predictor of the voting decision as long as people perceived it as informative
and follow it. We investigate this effect in the least likely setting of supporting a referendum
proposal, previous research revealed that it is easier to mobilize people against a
referendum proposal. The explanatory potential of campaign is tested against the usual
suspects described above in the context of three referendums organized in 2015–2016
in different East European countries (Bulgaria, Poland and Slovakia).1
The referendums
under scrutiny shared several features: each of them had three separate questions out
of which at least one was not very related to the other two, none of them was valid
due to low turnout, and none of them was initiated by political parties (the Bulgarian
and Slovak referendums were bottom-up, while the Polish referendum was initiated by
the country president). We also selected these referendums due to their diversity of
topics with the aim to identify whether the explanatory power of campaign is context sensitive
or holds across various topic and countries. The referendums refer to issues of a very
different nature: electoral reform, political funding, same-sex marriage and adoption
rights, education pertaining to sexual behaviour or euthanasia in schools, or positive
interpretation of taxation. We use individual-level data from an online survey conducted
in March–April 2017.
30 S. GHERGHINA AND N. SILAGADZE
The remainder of this paper proceeds as follows. The first section presents the mechanisms
through which a referendum campaign may influence voting preferences. In
doing so, it uses combines findings from research on referendums with ideas inspired
by findings on election campaigns. The second section presents the methodology and
data used in this study, while the third one provides background information about the
cases investigated in this paper. Next, we present the results of descriptive and inferential
quantitative analysis with emphasis on the similarities and differences between countries.
The conclusions summarize the key findings, discuss the implications of this paper and
open the floor to further avenues for research.
How referendum campaigns make a difference
Referendum campaigns influence voters’ preferences towards the policy through a combination
of mechanisms. One of the most straightforward and arguably the most investigated
is the hypodermic model that assumes an effect due to campaign features. Among
these features, the information delivery is prominent due to its potential to decrease the
unknown to ballot proposals (de Vreese & Semetko, 2004; Hobolt, 2005; LeDuc, 2002;
Schuck & de Vreese, 2009). The information allows for a systematic processing that will
be the basis of voters’ calculus. Under these circumstances, citizens are more likely to
make a calculus regarding their alternatives and perceived consequences based on
what they learn and much less on heuristic processing based on risk propensity (Morisi,
2016). In this sense, campaigns generate large quantities of information about different
aspects of the policy under consideration such as costs, benefits, implications, reasons
for change or for the status quo, etc. The longer and more intense the public debates,
the simpler it is for voters to make up their minds as there is a lot of available information
on the issue (Font & Rodriguez, 2009). Since the issue at stake in a referendum is divisive
and polarizing, most of the information is substantive with limited superficial features
encountered in regular elections. Regarding the latter, media is under fire by both academics
and politicians for its sensationalistic and market-based journalism that ignores
the ideological components and issue-solving approaches. Nevertheless, there is empirical
evidence that even under the circumstances news coverage and advertisement during
campaign has a great informational value for citizens (Iyengar & Simon, 2000). Moreover,
campaign has a socializing effect for the audience. Since the issue to be voted upon is
intensely discussed, people become aware about its content, nuances and consequences
and are likely to make an informed decision.
In addition to the campaign features, existing citizens’ attitudes should be factored in to
understand the complexity of campaign effects. An important strand of literature dealing
with elections claims that voters make up their minds before the campaign starts based on
their partisan affiliation or candidates’ sociological characteristics (Berelson, Lazarsfeld, &
McPhee, 1954; Campbell, Converse, Miller, & Stokes, 1960). This is also due to the fact
that voters get socialized to parties, know very well the competitors and voted for
some of them in the past. The studies supporting the minimal-effect hypothesis of campaign
(Finkel, 1993; Gelman & King, 1993) argue that election outcomes can be predicted
without accounting for the campaign. According to this perspective, voters have preferences
in place before the election period and the campaign only helps to activate
those latent preferences.
CONTEMPORARY POLITICS 31
Some of these conclusions from election campaigns can be transferred to referendum
campaigns as long as specific elements such as socialization and experience with competitors
are not included. We are interested in highlighting that predispositions and information
do not operate independently from each other, but rather predispositions
determine what information or third-party opinion a citizen accepts, since people tend
to ignore the information or opinions that do not agree with their prejudices and orientations
(Marcinkowski, 2007; Zaller, 1992). Even when accounting for predispositions
and previous preferences, a referendum campaign may have an effect on the vote
choice. The resonance model explains how messages received during campaign reinforce
voters’ prevailing predispositions or preferences (Iyengar & Simon, 2000). In this case, citizens
who already favour the proposal of the referendum will acquire during campaign
further arguments that may give them confidence that the initial attitude was the right
one; and such a belief is easily translated into a vote choice.
Two inter-related functions of referendum campaigns that transcend the two models
are the agenda setting and simplification of choices. Voters do not pay attention to all
the details surrounding a political issue and instead focus on a few elements that
appear important at the moment. However, what is important is not a matter of objectivity
or subjective choice on the side of citizens, but it is mainly a media-driven decision. Journalists
decide what to cover and what to ignore and thus they model saliency within the
electorate. The framing and priming of events during a campaign influences perceptions
by evoking different images in the minds of particular segments in society, leading to issue
interpretations (de Vreese, 2004). Such effects can be observed both in the presence of
existing opinions (Domke, Shah, & Wackman, 1998) and in their absence. In particular,
priming ensures that the issues considered salient by voters – saliency that can also
emerge during campaign – become the criteria of evaluation in the minds of the electorate
(Iyengar & Simon, 2000). To use a recent example, the discourse of the Leave camp in
the Brexit referendum picked-up on the salient issue of Britain being a net contributor to
the European Union. It played this card heavily during campaign and thus the assessment
of people evolved around that issue set up by one of the camps. In brief, the agendasetting,
framing and priming simplify the choices for people by providing them with
readily available information.
These arguments illustrate how a campaign yields a multi-faceted effect on the voting
behaviour. We argue that such effects are likely to occur especially in the case of voters
who support the referendum proposals. This happens for at least three reasons. First,
the cognitive benefits brought by campaigns reduce the costs of uncertainty and diminish
the risk propensity. Since people are more strongly motivated to avoid costs than to
achieve gains (Tversky & Kahneman, 1981), those who wish to avoid costs and have a predisposition
prior to the campaign, will be rather immune to the content of campaign. Perceived
risk is a motivator of self-protective behaviour (Schuck & de Vreese, 2009) and thus
such voters will be oriented to oppose the referendum proposals. Through its learning,
persuasive and socialization processes a referendum campaign reduce uncertainty
about outcomes and remove the burden from the shoulders of those seeking gains. Moreover,
even those who are risk averse and under normal conditions would oppose a proposal
due to its uncertain consequences (Christin et al., 2002), will face less uncertainty once
exposed to campaign. Second, the efforts made by the camp supporting the referendum
are on average greater than those of the opposing camp. The political elites (in case of top-
32 S. GHERGHINA AND N. SILAGADZE
down) or citizens (for bottom-up) behind the referendum are aware that quite often they
start with a handicap: people have to be persuaded to support their initiative. This challenge
increases when the pursued issue has low saliency in society. Consequently, the promoters
of the policy subjected to referendum are likely to be more active than the
opponents. The most visible result of such a visibility is a greater level of information
and more persuasive ads launched to the public. Third, voters who oppose the referendum
proposals may adopt a strategic behaviour and skip voting. Since many referendums
require a participation quorum for validation, those who oppose the proposals have a
greater say if they do not turn out. This is what could be observed in Italy where political
parties opposing the referendum proposals demobilised voters and encouraged them to
stay away from polls (Uleri, 2002). The existence of such a strategic behaviour indicates
that those who pay attention to campaign are more likely to be open or undecided regarding
a support for the referendum proposals. Following all these arguments, we hypothesize
that:
H1: Citizens who perceive the campaign as informative will support the referendum proposals.
H2: Citizens who follow the campaign will support the referendum proposals.
Controls
We test the explanatory potential of the hypothesized effects against the variables highlighted
in the literature as main predictors for voting behaviour: opinion before the campaign,
support for government, partisan cues, access to information relative to the
referendum issue and socio-demographics (education, age and gender). As previously
explained, an influential body of literature posited that the beliefs formed before campaign
play an important role in the final voting preference. The theory of second-order
considerations claims that people often decide on their vote choice in referendums
with evaluations of the incumbent government in mind. They do so either because
they feel that there is not much at stake (Franklin, 2002) or because they wish to use
the referendum as a punishment tool. Partisan cues were found by earlier studies as
quite important determinants for voter choice. On complex issues, voters use the shortcuts
received from the political parties they support (Font & Rodriguez, 2009; Kriesi, 2005;
LeDuc, 2009). Party endorsements may allow citizens to imply their own position on a
ballot issue without detailed information about it (Hobolt, 2007). Access to information
is a variable meant to capture the amount of information to which individuals have
access during campaign. Our hypothesized effect goes far beyond the amount of information
in campaign and that is why we include the simple exposure to information
related to the referendum (via different media environments) as a control variable.
Finally, we control for socio-demographic variables since they may also have an impact
on the support for the referendum proposals.
Research design
To test the hypotheses, we use individual-level data from an online survey conducted in
March–April 2017 with a total number of 511 valid responses in three countries: Bulgaria
(257), Poland (121) and Slovakia (133). The number of answers in the statistical analysis is
CONTEMPORARY POLITICS 33
lower than 511 because respondents skipped some questions or provided answers that
were not included in the study (see below). We selected the most recent multiple question
referendums in Eastern Europe. The variation in their topics was important because results
are more robust when observing similar patterns across different types of decisions.
The survey had the same questions translated into the national language of every
country. The focus of the analysis is those who voted and a probability representative
sample would have been ideal to generalize findings. However, setting up such a
sample was not possible because there is no official statistics regarding the features of
those who turned out to vote in these referendums (socio-demographics). Instead, we
use a convenience sample with a snowball distribution of the link through Facebook
and e-mail addresses. While the number of answers may not look very high relative to
the turnout in each of these referendums, it is less important in the context of the
current analysis. Our primary goal is to understand the process through which campaigns
influenced voting choice rather than making generalizable statements. We are aware that
there is a self-selection bias in which those with access to Internet and skills to use devices
could fill in the survey. However, all three countries have high levels of Internet access,
ranging between 60 and 80% (World Bank 2017). For analysis, we use a combination of
bivariate statistics and ordinal regression with a pooled and country level models.
Variable operationalization
The dependent variable of our study is the support for referendum proposals. It is
measured as the answer to the following question: ‘How did you vote on the referendum
issues?’. Available answers were coded on a four-point ordinal scale as follows: all yes (1),
two yes and one no (2), one yes and two no (3), and all no (4). A fifth option was also listed
(blank vote) but since this is of no importance in the current context, it was removed from
the analysis. The independent variable informative campaign (H1) is operationalized
through the answers provided to the following question: ‘Thinking about the referendum
campaign, was it informative in providing with all necessary information about the
topics?’. The available answers were coded on a four-point ordinal scale as follows: very
informative (1), quite informative (2), hardly informative (3), not at all informative. The variable
following campaign (H2) is measured through the ordinal answers to the question
‘How much did you follow the referendum campaign?’, coded as follows: very much (1),
much (2), to some extent (3) and not at all (4).
The opinion before the referendum and partisan cues were operationalized as the
answer to the following question: ‘When voting in this referendum, to what extent did
your opinion before the referendum campaign play a role in your decision?’ and ‘When
voting in this referendum, to what extent did the position of the party for which you
voted in the (year closest to referendum) parliamentary elections play a role in your
decision?’. The answers were coded on a four-point ordinal scale as follows: very much
(1), much (2), to some extent (3) and not at all (4). For these two items, a ‘not available’
answer was listed since the individual could miss the election or did not have the right
to vote due to age limitations. All ‘not available’ answers were removed from the analysis
and treated as missing data.
Government support is measured through the answers to the question ‘How satisfied
are you with the way the (country) government does its job?’ and coded as follows: very
34 S. GHERGHINA AND N. SILAGADZE
satisfied (1), quite satisfied (2), hardly satisfied (3) and not at all satisfied (4). The amount to
information during campaign is a six-point scale index cumulated after summing up the
answers to the question: ‘Where did you get information about the referendum topics
from?’. The respondent had a multiple choice with several options and every time an
option was mentioned it received a code of 1: TV, newspapers or radio, online news
portals and Internet forums, Facebook, posters and leaflets, discussions with friends. The
resulting index ranges from a value of 0 reflecting the situation in which a respondent
did not get information from any of the five environments to a maximum of 5 in case
the respondent used all the available environments.
Gender is a dichotomous variable in which male is coded 1 and female 2. Age is an
ordinal variable recorded as the number of completed years at the number of survey,
while education is a five-point ordinal scale ranging from primary (coded 1) to post-university
degree (coded 5). The following section provides an overview of the referendums to
illustrate the topics and the driving forces behind campaigns.
Background information about the referendums
The referendum experience of the three analysed countries differs. Bulgaria organised its
first referendum in 2013 (bottom-up) and since then it has been very active in this field,
Poland organised five referendums in its post-communist history with quite some distance
between them, while Slovakia had eight referendums with only the most recent (included
in our analysis) being a bottom-up procedure.
Bulgaria
The Bulgarian referendum was held on the 6th of November 2016 simultaneously with the
presidential elections. The citizens were asked three major questions: the replacement of
the proportional vote by a majority vote, the introduction of compulsory voting, and the
reduction of the state subsidies given to the political parties – from 11 leva (5.60 €) to 1 lev
(0.50 €) for each vote won. The second question was quite redundant since compulsory
voting had been already introduced a few months earlier (Stoychev, 2017).
Prior to the fall of the communist regime three popular votes were held in the country –
in 1922, 1946 and 1971. In modern Bulgaria, the use of the direct democratic instruments is
mainly regulated by the Law on Direct Participation of Citizens in the State and Local Government.
In compliance with Article 3, a national referendum can be initiated over the
issues which are in the competence of the National Assembly, with the exception of constitutional
arrangements, taxation and budgetary topics. The result of a referendum is
mandatory if the turnout reaches the level of the turnout in the last parliamentary elections
and if more than a half of votes are in favour of the proposed question. If the
turnout criterion is not met but was higher than 20% and if the number of ‘yes’ votes
exceeds 50% the National Assembly is required to discuss the issue (Milanov, 2016).
The referendum in 2016 was initiated not by a political force, but by a very well-known
Bulgarian showman, Stanislav Trifonov, who has been a TV star from the beginning of the
‘90s (Martino, 2016). It was the first successful bottom-up initiative leading to a nationwide
referendum in Bulgaria since 1989. He managed to mobilise the disenfranchised citizens.
A special Facebook page was launched for subscription campaign. Not only a large
number of volunteers was engaged, but also many celebrities publicly supported the
CONTEMPORARY POLITICS 35
campaign. In the end, over 570,000 valid signatures instead of 400,000 required were collected
for six proposed questions. However, president Rosen Plevneliev referred to the
Constitutional Court questioning the legality of three questions (to halve the number of
MPs from 240 to 120, to introduce electronic voting and direct election of police chiefs).
The Court accepted the president’s arguments and blocked these questions (Stoychev,
2017).
The campaign itself was criticised for lacking the substantial debate. The main political
parties were occupied by the simultaneously running presidential election campaign and
paid little attention to the referendum: ‘none of the major parties registered to campaign
for the referendum at the Central Electoral Commission. Only 14 minor parties and
coalitions with less than 1% of electoral support registered, mainly because of the
money for media coverage provided by the state [about 20,000 EUR]’ (Stoychev, 2017,
p. 191). In this situation, the TV show of Trifonov became the primary source of information,
though one-sided, massively agitating for the ‘yes’ vote. This resulted in a high
degree of confusion among people over the asked questions (Cheresheva, 2016).
Those who took part in the poll overwhelmingly said ‘yes’ to all three questions.
However, the result was not binding, since the referendum turnout of 50.8% fell short
of the 2014 parliamentary elections turnout of 51% by a small margin of roughly 13,000
votes. Despite this fact, ‘the 2016 referendum was the most successful popular initiative
in the contemporary history of Bulgaria. Although it did not produce a legally binding
result, it produced a politically legitimate demand for electoral reform’ (Stoychev, 2017,
p. 192).
Poland
The Polish referendum held on 6th of September 2015 was a fifth national referendum in
the history of modern Poland. The reason behind conducting the popular vote was the
outcome of presidential elections held in May 2015. The former president Bronislaw
Komorowski, backed by the ruling centrist Civic Platform (PO) initiated the referendum
‘as a panic move following his shock defeat in the first round of May’s presidential election’
(Szczerbiak, 2015). Despite the prediction of all polls Komorowski came behind his main
opponent Andrzej Duda, the candidate of the right-wing Law and Justice (PiS) party.
The most surprising fact was, however, that a rock musician Pawel Kukiz won around
20% of votes with his promise to implement single-member constituencies in the Sejm
elections. A day after the elections, Komorowski declared to hold a referendum justifying
it as a reaction to the voters’ wish for change – indeed he intended to win the votes of
Kukiz’s supporters by this political move. In the end, Komorowski lost. The referendum,
however, will be remembered as a part of presidential campaign (Hartliński, 2015).
The legal base was the Article 125 allowing the President to call a referendum if the
Senate agrees on it. The results are binding if the turnout is more than 50%. The referendum
was comprised of three questions: on introducing single-member constituencies in
elections to the Sejm, Poland’s more powerful lower chamber of parliament; on maintaining
the current method of financing of political parties from the national budget and introducing
a presumption in favour of the taxpayer in disputes over taxation law. The first
question was the actual reason behind the initiation of the referendum and it caused
also the most of legal doubts, since its implementation contradicted the Article 96 of
36 S. GHERGHINA AND N. SILAGADZE
the Constitution stating that the elections to the Sejm shall be inter alia proportional
(Hartliński, 2015).
The whole referendum campaign was barely visible. The main axis of the campaign was
the issue of single-member constituencies. The PO and Kukiz were clearly in favour of the
issue, the PiS politicians claimed that they would take part in the referendum but their
party did not invest much effort in mobilizing their supporters. Other parties’ position
was against the issue and their representatives declared not to be going to vote. This inevitably
was to influence the participation rates. The second question reflected more or less
the same division and the third one was not a matter of debate since every party articulated
in favour of strengthening the position of citizens. The campaign itself was delegated
from established parties to mainly unknown/minor societies and foundations that enjoyed
the benefit of being present in the media and receiving remuneration. The parties, having
in mind the approaching parliamentary elections, were not actively engaged in the
debate. No true campaigning took place in comparison to the presidential or parliamentary
elections (e.g. no TV commercials, no posters) (Hartliński, 2015). ‘Walking around the
streets of Krakow, talking to its inhabitants, watching the news and reading the national
press, discussion of the issue is notable by its absence. Indeed, many Poles seem to be confused
about the referendum and about the electoral system in general’ (Tilles & Bill, 2015).
The lack of real campaign found its reflection in a very low level of knowledge and interest
among the population and as a consequence in a miniscule turnout of only 7.8%,
though all questions were approved by the voters (Hartliński, 2015). It was the first time
in the Polish modern history that the result of the referendum was not binding due to
the low participation. Undoubtedly the topic wasn’t as important as previous ones –
e.g. on the Constitution in 1997 or joining the EU in 2003 (Hartliński, 2015). One
member of the Polish Electoral Commission described it as ‘one of the most expensive
public opinion polls in Europe’ (Szczerbiak, 2015). However, the way of calling this referendum
as a part of presidential rivalry and the very low-profile campaign did not contribute
to winning support for more direct democracy in Poland. This popular vote was a big
fiasco for the political parties (Hartliński, 2015).
Slovakia
The Slovak same-sex marriage referendum held on 7th of February 2015 was unique in
many terms. The so-called Referendum on Family was the first referendum in the
history of the country that was initiated not by the established political parties, but by
the civil society. In record times over 400,000 signatures were collected (roughly 10% of
the country’s electorate) demonstrating unprecedented civil engagement with over
10,000 volunteers and more than 100 pro-life associations. Furthermore, it was also the
first time in the Slovak history that the issues of sexual education, forms of modern
family and adoption was subjected to public debate. The referendum campaign was
highly polarizing and extremely salient (Rybar & Sovcikova, 2016).
The Alliance for Family, a civic organization established in 2013, was the main organizer of
the popular initiative. Roughly one year before the referendum, in February 2014, the Slovak
Parliament and the governing social-democratic Smer party passed a constitutional amendment
(Art. 41) that clearly defined the marriage as a union between a man and a woman.
This was not ambitious enough for the Alliance demanding not only explicit prohibition
of child adoptions by same-sex couples but also the denial of legal basis for same-sex
CONTEMPORARY POLITICS 37
unions – the latter was proved as unconstitutional by the Court (Kużelewska, 2015; Rybar &
Sovcikova, 2016). ‘The referendum was meant to take a back-step and reiterate that in Slovakia
a family consists of a husband (male) and wife (female) … ’ (Zordova, 2015). This would
make the future attempts to change the legislation even more difficult. The argument was
that the values of the traditional family are under threat since more and more countries liberalise
the rules – e.g. the neighbouring Austria and the Czech Republic allow various forms
of same-sex unions and child adoption by gay couples (The Guardian, 2015).
According to the Article 95 of the Slovak Constitution, the president of the state calls a
referendum either on the basis of the National Council’s resolution or if a petition receives
more than 350,000 signatures. The vote is valid only if turnout exceeds 50% of all eligible
voters. All referendums until this date were initiated in accordance with the first paragraph
and all of them – with the exception of EU membership referendum in 2003 – failed due to
insufficient turnout (Rybar & Sovcikova, 2016).
The referendum campaign was heated, debated and controversial. However, not the
political parties, but the Catholic Church was one of the main players in the campaign, supporting
the Alliance both morally and financially: for instance, petition sheets were available
in the local churches and priests openly encouraged their parishioners to support the
petition, providing theological backing (Rybar & Sovcikova, 2016; Smrek, 2015). Shortly
before the referendum date, Pope Francis addressed the Slovak people: ‘I greet the pilgrims
from Slovakia and, through them, I wish to express my appreciation to the entire
Slovak church, encouraging everyone to continue their efforts in defence of the family,
the vital cell of society’ (Harris, 2015). In contrast, the leading figures from the political
elite refused to give a position on a referendum topic, thus, ‘the overriding silence of
high officials and political parties on the issue’ was one of the main peculiarities of this
popular vote (Kral, 2015).
The advocates framed the campaign in a quite positive manner, stressing the desire to
preserve the status quo in regard to ‘marriage’ definition and to give parents more rights
to decide about the content of their children’s education. The slogan of the Alliance was
‘Three Yeses For Children’ which defined the confrontational tone of the campaign, depicting
the homosexuals ‘as perpetrators who seek to challenge the status quo’ (Smrek, 2015).
Thus, the public debate, concentrated along two lines, namely the attitudes towards homosexuals
and the role of the Catholic Church in public life (Rybar & Sovcikova, 2016). The
opponents – mainly the LGBT minority represented by the non-governmental organisation
Iniciatíva Inakosť campaigned for not taking part in the referendum (Valkovičová, 2017).
In the end, the vote was invalid due to the low turnout of 21.4%, below the average
compared to other referendums. However, over 90% of voters said ‘yes’ to all the three
questions: whether marriage can only be a union of a man and a woman, whether
same-sex couples should be banned from adoptions, and whether children can skip
classes involving education on sex and euthanasia.
Analysis and results
This section starts with a bivariate correlation between the variables to observe the statistical
relationships on a one to one basis. It continues with ordinal regression models
ran for several variables. Figure 1 explores the ways in which people voted at aggregate
level and in each of the three countries, to illustrate the degree of variation on the
38 S. GHERGHINA AND N. SILAGADZE
dependent variable. The graph illustrates that the highest percentage among the total
number of respondents (45% in all three countries) supported all three referendum questions.
This trend is observable in Bulgaria and Slovakia, the exception to the rule being
Poland where the highest percentage of respondents (46%) voted yes to two questions
as opposed to 17% who supported all three questions. One explanation for the massive
vote in this direction is that two out of three questions were for change, while the third
was to maintain the status quo. The latter referred to public party funding and the referendum
results indicated that almost 83% of those who turned out to vote answered negatively
to this question. in Slovakia the large percentage of votes in favour of all three
proposals is due to the quite radical measures proposed in this referendum against
same sex marriage, adoption and non-compulsory sexual and euthanasia education in
schools. Since the three topics are related, the likelihood to support all of them is high.
This is an issue reflected also in the official results when more than 90% of those who
went to the polls casted a positive vote for each of them.
Table 1 includes the bivariate correlation coefficients between the support for referendum
proposals and all other variables. The consistent observation across the pooled
and country level data is that informative campaign is the variable that correlates the
highest with support for the referendum proposals. The relationship goes in the
Figure 1. Voter distribution according to the support for referendum topics (N = 511). (Source: Statistical
Office of the Slovak Republic).
Notes: The wording of referendum questions differed across the examined cases. Bulgaria: (a) Do you support the national
representatives to be elected by majoritarian electoral system with absolute majority in two rounds? (b) Do you support the
introduction of compulsory voting in elections and referendums? (c) Do you support the annual subsidy for financing the
political parties and coalitions to be 1 BGN for every valid vote at the last parliamentary elections? (Source: Central Electoral
Commission). Poland: (a) Are you in favor of the introduction of single-mandate electoral districts in the elections to the
Sejm of the Republic of Poland? (b)Are you in favor of maintaining the current method of financing political parties from
the state budget? (c) Are you in favor of introducing the general rule of doubts about the interpretation of tax law in favor
of the taxpayer? (Source: National Electoral Commission). Slovakia: (a) Do you agree that the concept of marriage denotes
solely the legal bond between one man and one woman, but it must not denote any other cohabitation between persons?
(b) Do you agree that it should not be allowed to couples or groups of persons of the same sex to adopt and subsequently
raise children? (c) Do you agree that schools should not require participation of children in the field of sexual education or
euthanasia if their parents or the children themselves do not agree with the content of such education?
CONTEMPORARY POLITICS 39
hypothesized relationship (H1) with respondents who find the campaign informative
supporting all the proposals. The weakest relationship is in Slovakia, mostly due to
the low variation on the dependent variable. In that particular case, irrespective of
how respondent found the campaign, many of them cast a positive vote to all
three questions. This result is not surprising if we keep in mind that some voters in
Bulgaria and Poland were confused about the topics of the referendums (see the previous
section). Those are the cases in which the referendum campaign can have an
important effect by clarifying issues, educating the public and helping voters decide
whether they support the proposal.
The bivariate correlations provide empirical support for H2 only in the pooled and
Slovak model with respondents who closely follow the campaign being more inclined
to favour the referendum proposals. In the other two cases, there is statistical independence
with no relationship between these two variables. In the pooled model the variables
corresponding to the main effects correlate the strongest (at the highest level of statistical
significance) with the dependent variable. The variable informative campaign correlates in
all but the Slovak model the highest with the support for referendums and that indicates
robust empirical support for the first hypothesis.
Among the controls, only the pre-campaign opinion, access to information and education
correlate with support for referendum proposals. People who cast a vote according
to their attitude before the campaign have a slight tendency to support the proposals as
opposed to those who do not have one. This relationship goes in the same direction only
in Slovakia at country level, in Bulgaria and Poland it is reversed with people who do not
hold pre-campaign opinions being more likely to support all referendum proposals.
People with less access to information are slightly re likely to support the referendum proposals
(–0.07 in the pooled model), a relationship that is quite strong in Slovakia (it is the
highest correlation coefficient, statistically significant at the 0.05 level). Less educated
people are more likely to support referendum proposals, a relationship that holds also
in Slovakia but it is reversed in Poland – where mode educated people support the proposals
– and displays statistical independence in Bulgaria.
Other controls with relevant correlation at country level are satisfaction with government,
party cues and gender. In Poland, citizens who are satisfied with the activity of
the government are more likely to support the referendum proposals (0.21, statistically significant
at the 0.01 level). This observation is quite intuitive since it is the only referendum
Table 1. Bivariate correlation with support for the referendum proposals.
Variable Pooled Bulgaria Poland Slovakia
Informative campaign 0.42*** 0.51*** 0.24** 0.16*
Follow campaign 0.14*** –0.01 0.01 0.16*
Pre-campaign opinion 0.10** –0.03 –0.05 0.11
Satisfaction with government –0.03 –0.06 0.21** –0.04
Party cues 0.01 0.21*** 0.12 –0.10
Access to information –0.07 0.11 –0.04 –0.19**
Gender –0.02 0.01 0.16 –0.15
Age 0.01 0.01 0.11 –0.10
Education –0.09 –0.02 0.07 –0.11
Notes: Correlation coefficients are Spearman’s rank.
N varies between 396 (party cues) and 473 (informative campaign).
***p < .01; **p < .05; *p < .1.
40 S. GHERGHINA AND N. SILAGADZE
among the three investigated that had been initiated top-down, by political elites. In Bulgaria,
those respondents who followed the opinion of the party for which they voted in the
most recent legislative elections prior to the referendum are more likely to support the
three proposals (coefficient 0.21, statistically significant at the 0.01 level). In Poland, the
direction of relationship is similar, while in Slovakia it goes in the opposite direction
with people who ignore their parties being slightly more likely to support the referendum.
The latter can be explained through the ambiguity of party positioning on the referendum
topics. The messages of political parties were not obvious, cues were unclear and thus we
cannot observe a relationship between these and the voting decision. Finally, in Poland
male are more likely to support the referendum proposals, while in Slovakia more
female are among the supporters.
These bivariate correlations are an initial estimation of the relationship between variables.
We will now turn to a regression analysis that seeks to assess causality. We run
two types of ordinal regression models (due to the nature of the dependent variable):
without control variables and with control variables. In the models with controls we
exclude the socio-demographics for the following reasons: (1) the bivariate analysis
showed the existence of poor relationships with the support for referendum proposals,
(2) we ran regression analysis with them and the results confirm the existence of a very
weak effect and (3) the interpretation of results is more parsimonious with fewer controls.
In brief, by not reporting them there is not much substantive information lost.
Table 2 includes the odd-ratios for the two main effects. The reference category for all
the variables are the absence of those features, e.g. for informative campaign, it is the category
of people who voted against all three proposals. At pooled level, the citizens who
consider the referendum campaign as very informative are more than three times more
likely to support the referendum proposals (OR = 3.02, statistically significant at the 0.01
level) as opposed to those who do not find the campaign informative at all. Voters who
followed very much the campaign are 1.24 times (statistically significant at the 0.1 level)
more likely to support the proposals as opposed to those who did not follow at all. At
country level, the effect hypothesized in H1 finds empirical evidence in all countries,
with strength varying from 1.94 time more likely in Slovakia to 3.66 (statistically significant
at the 0.01 level) in Bulgaria. The empirical support for H2 is mixed at country level where
only Slovakia goes in the hypothesized direction: voters who follow campaign very much
are two times (OR = 2.01) more likely to vote for referendum proposals as opposed to
those who did not follow at all. In Bulgaria and Poland, people who follow the campaign
are less likely (OR = 0.83 and OR = 0.90) to support the referendum proposals compared to
those respondents who did not follow at all the campaign.
Table 2. Ordinal regression with support for the referendum proposals as DV (no controls).
Variable Pooled Bulgaria Poland Slovakia
Informative campaign 3.02*** (0.38) 3.66*** (0.62) 2.23*** (0.66) 1.94 (0.87)
Follow campaign 1.24* (0.16) 0.83 (0.15) 0.90 (0.26) 2.01 (0.90)
Pseudo R2
0.08 0.11 0.03 0.06
Log likelihood –543.03 –287.17 –128.49 –44.58
N 472 242 109 121
Notes: Reported coefficients are odds-ratios, standard errors in brackets.
***p < .01; **p < .05; *p < 0.1.
CONTEMPORARY POLITICS 41
The differences in the size of effect between H1 and H2 can be better visualized when
comparing the two marginal effects in Figure 2, modelled at the pooled level. The left side
of the graph depicts the marginal effect of Informative campaigns on the support for referendum
questions. It shows a strong effect where the perception of high or very high informative
campaign makes a substantive difference in the way in which voters support the
referendum proposals. The rights side of the graph depicts the marginal effect of following
the campaign and it is much weaker. The interval on which it has an effect on the dependent
variable (vertical axis) is considerably smaller compared to the effect of informative
campaigns.
Table 3 includes the ordinal regression models with control variables. Informative
campaign remains a very strong predictor in the pooled model where people who perceive
the campaigns as very informative are almost three times more likely (OR = 2.94,
statistically significant at the 0.01 level) to support all three proposals. The effect of following
campaigns (H2) is also positive but considerably weaker than H1. Pre-campaign
opinion has a positive effect on support for proposals, while satisfaction with the government
and party cues have a negative effect. Access to information has no effect on
the voting behaviour. This observation is relevant because it indicates that campaign
means much more than simply information. The empirical evidence in our cases illustrates
that there is no effect of pure information, while the campaign overall – as long
as people consider it informative and follow it – has a consistent positive effect on
people’s behaviour.
Figure 2. The marginal effect of informative campaigns and follow campaigns (pooled model).
Table 3. Ordinal regression with support for the referendum proposals as DV (with controls).
Variable Pooled Bulgaria Poland Slovakia
Informative campaign 2.94*** (0.43) 3.28*** (0.65) 1.80* (0.63) 2.31* (1.11)
Follow campaign 1.13 (0.17) 0.68* (0.15) 1.29 (0.50) 1.38 (0.72)
Pre-campaign opinion 1.15 (0.11) 0.88 (0.13) 0.94 (0.28) 1.36 (0.41)
Satisfaction with government 0.89 (0.12) 0.82 (0.17) 1.57* (0.38) 1.11 (0.66)
Party cues 0.87 (0.10) 1.57** (0.52) 1.11 (0.23) 0.54 (0.22)
Access to information 0.97 (0.10) 1.17 (0.17) 1.14 (0.29) 0.72 (0.27)
Pseudo R2
0.08 0.11 0.04 0.12
Log likelihood –433.12 –212.79 –107.65 –40.93
N 379 180 93 112
Notes: Reported coefficients are odds-ratios, standard errors in brackets.
***p < .01; **p < .05; *p < .1.
42 S. GHERGHINA AND N. SILAGADZE
The country level models indicate a strong and consistent empirical support for the first
main effect (H1). The second main effect (H2) is also quite consistent and with the exception
of Bulgaria in the other countries goes in the hypothesized direction. The rest of variables
have only isolated effect in some countries, e.g. satisfaction with government in
Poland or party cues in Bulgaria. The major conclusion of the statistical analysis is that
the two main variables (H1 and H2) are the only ones with a consistent effect across
the three investigated referendums; this observation is revealed both by the bivariate
and multivariate statistical analysis.
Conclusions
This paper argued and tested the importance of referendum campaign as a predictor for
voting decision by looking at its perceived informative character and the degree to which
people follow it. It analysed this effect in the least likely setting of supporting referendum
proposals on different topics in Bulgaria, Poland and Slovakia. The study relied on individual-level
analysis coming from a web survey conducted in the spring of 2017. The results
indicate that the two variables explain the decision of citizens to support referendums
across different settings. Their effects are consistent and significantly stronger than
alternative explanations employed in the literature such as the limited effect of campaigns,
second-order elections, partisan cues or amount of information received.
The implications of these findings go beyond the comparative study of the three East
European countries and they could have an impact on the broader study of voting
behaviour during referendums. At theoretical level, our study reveals the importance as
campaign as a valuable explanatory variable for the decisions taken by citizens in referendums.
So far, research focused mainly on components of campaigns (information or
media priming/framing) or campaign actors (parties or initiators) but has paid little attention
to the campaign per se. Our results show that campaign is much more than information and
it may yield much stronger relevance than the actor involved. Consequently, it is a determinant
that could be included in further frameworks for analysis. At empirical level, the perception
of campaign and degree to which people follow it appears to have consistent effect
across different types of referendums. These are the only variables (especially the perceived
informative feature of campaign) with such features among the ones considered. They are
not context sensitive and are useful explanations for the way in which people vote.
One limitation of this study is the convenience sample on a fairly limited number of voters
used to observe these behavioural patterns. Further research could address this issue by conducting
a study on a large sample that could also pursue representativeness as long as data
regarding the profile of voters will become accessible. Another avenue for further research is
a more qualitative approach that involves interviews to explain the underlying mechanism.
For example, we showed that the perception of informative campaign matters but we
cannot know what is behind that perception. Focus groups or extended semi-structured
interviews with voters could clarify the meaning and substantiate our findings.
Notes
1. The generic term of Eastern Europe is used to define the broader region to which the three
countries belong. There are several definitions of the region and various ways to label these
CONTEMPORARY POLITICS 43
countries, e.g. two of these countries can be considered Central Europe, while the third lies in
South-Eastern Europe. Since identification issues lie beyond the goal of this article, we follow
the definition used by the United Nations (https://unstats.un.org/unsd/methodology/m49/
#geo-regions) that sees all these three countries as belonging to Eastern Europe.
Acknowledgements
The authors are very grateful for the constructive comments and suggestions received on earlier
drafts of this paper from the participants at the International Workshop Democratic Innovations in
Eastern Europe (Södertörn University Stockholm, 2017), the Seminar Series of the Åbo Akademi
University Finland and at the graduate research seminar at the Autonomous University of Madrid.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes on contributors
Sergiu Gherghina is a Lecturer in Comparative Politics at the Department of Politics, University of
Glasgow. His research interests lie in party politics, legislative and voting behaviour, democratization,
and the use of direct democracy.
Nanuli Silagadze is a PhD candidate at the Department of Political Science, Åbo Akademi University.
Her research interests lie in the fields of direct democracy, public opinion and voting behaviour.
ORCID
Sergiu Gherghina http://orcid.org/0000-0002-6627-5598
Nanuli Silagadze http://orcid.org/0000-0001-6251-7011
References
Berelson, B. R., Lazarsfeld, P. F., & McPhee, W. N. (1954). Voting: A study of public opinion formation in a
presidential campaign. Chicago: University of Chicago Press.
Campbell, A., Converse, P., Miller, W., & Stokes, D. (1960). The American voter. New York, NY: Wiley.
Cheresheva, M. (2016). Referendum on political system confuses Bulgarians, Balkan Insight. Retrieved
from http://www.balkaninsight.com/en/article/upcoming-referendum-raises-concern-in-bulgaria-
11-01-2016
Christin, T., Hug, S., & Sciarini, P. (2002). Interests and information in referendum voting: An analysis of
Swiss voters. European Journal of Political Research, 41(6), 759–776.
de Vreese, C. H. (2004). Primed by the euro: The impact of a referendum campaign on public opinion
and evaluations of government and political leaders. Scandinavian Political Studies, 27(1), 45–64.
de Vreese, C. H., & Semetko, H. A. (2004). Political campaigning in referendums: Framing the referendum
issue. London: Routledge.
Domke, D., Shah, D. V., & Wackman, D. B. (1998). “Moral referendums”: values, News media, and the
process of candidate choice. Political Communication, 15(3), 301–321.
Finkel, S. E. (1993). Reexamining the “minimal effects” model in recent presidential campaigns. The
Journal of Politics, 55(1), 1–21.
Font, J., & Rodriguez, E. (2009). Intense but useless? Public debate and voting factors in two referendums
in Spain. In M. Setala & T. Schiller (Eds.), Referendums and representative democracy:
Responsiveness, accountability and deliberation (pp. 162–185). London: Routledge.
44 S. GHERGHINA AND N. SILAGADZE
Franklin, M., Marsh, M., & Wlezien, C. (1994). Attitudes toward Europe and referendum votes: A
response to Siune and Svensson. Electoral Studies, 13(2), 117–121.
Franklin, M. N. (2002). Learning from the Danish case: A comment on Palle Svensson’s critique of the
franklin thesis. European Journal of Political Research, 41(6), 751–757.
Gelman, A., & King, G. (1993). Why are American presidential election campaign polls so variable
when votes are so predictable?. British Journal of Political Science, 23(4), 409–451.
Harris, E. (2015). Pope Francis backs traditional marriage as Slovakia vote looms, Catholic News Agency.
Retrieved from https://www.catholicnewsagency.com/news/pope-francis-backs-traditional-marriage-
as-slovakia-vote-looms-37102.
Hartliński, M. (2015). The 2015 referendum in Poland: Direct democracy notes. East European
Quarterly, 43(2–3), 235–242.
Highley, J., & McAllister, I. (2002). Elite division and voter confusion: Australia’s republic referendum in
1999. European Journal of Political Research, 41(4), 845–861.
Hobolt, S. B. (2005). When Europe matters: The impact of political information on voting behaviour in
EU referendums. Journal of Elections, Public Opinion and Parties, 15(1), 85–109.
Hobolt, S. B. (2007). Taking cues on Europe? Voter competence and party endorsements in referendums
on European integration. European Journal of Political Research, 46(2), 151–182.
Hobolt, S. B. (2009). Europe in question: Referendums on European integration. Oxford: Oxford
University Press.
Iyengar, S., & Simon, A. F. (2000). New perspectives and evidence on political communication and
campaign effects. Annual Review of Psychology, 51(1), 149–169.
Kral, D. (2015). Despite a controversial referendum on same-sex marriage, democracy remains alive
and well in Slovakia, LSE European Politics and Policy (EUROPP) Blog. Retrieved from http://bit.ly/
1vh9yEe
Kriesi, H. (2005). Direct democratic choice: The Swiss experience. Plymouth: Lexington Books.
Kriesi, H. (2006). Role of the political elite in Swiss direct-democratic votes. Party Politics, 12(5),
599–622.
Kużelewska, E. (2015). How far can citizens influence the decision-making process? Analysis of the
effectiveness of referenda in the Czech Republic, Slovakia and Hungary in 1989–2015. Baltic
Journal of European Studies, 5(2), 171–196.
LeDuc, L. (2002). Referendums and elections: How do campaigns differ? In D. M. Farrell & R. SchmittBeck
(Eds.), Do political campaigns matter? Campaign effects in elections and referendums (pp.
145–162). London: Routledge.
LeDuc, L. (2009). Campaign tactics and outcomes in referendum. In M. Setala & T. Schiller (Eds.),
Referendums and representative democracy (pp. 139–161). London: Routledge.
Lupia, A. (1994). Shortcuts versus encyclopedias: Information and voting behavior in California insurance
reform elections. American Political Science Review, 88(1), 63–76.
Marcinkowski, F. (2007). Beyond information and opinion. The importance of public communication
in the referendum process. In Z. T. Pallinger, B. Kaufmann, W. Marxer, & T. Schiller (Eds.), Direct
democracy in Europe: Developments and prospects (pp. 94–107). Wiesbaden: VS Verlag für
Sozialwissenschaften.
Martino, F. (2016). Bulgaria, amid presidential elections and the referendum, OBC Transeuropa. Retrieved
from www.balcanicaucaso.org/eng/Areas/Bulgaria/Bulgaria-amid-presidential-elections-and-the-
referendum-175104.
Mendez, F., Mendez, M., & Triga, V. (2014). Referendums and the European union: A comparative
inquiry. Cambridge: Cambridge University Press.
Milanov, V. (2016). ‘Direct Democracy in Bulgaria’, Pázmány Law Working Papers, pp. 1–18. Retrieved
from http://www.plwp.jak.ppke.hu/
Morisi, D. (2016). Voting under uncertainty: The effect of information in the Scottish independence
referendum. Journal of Elections, Public Opinion and Parties, 26(3), 354–372.
National Referendum 2015. (2015). National Electoral Commission. Retrieved from http://
referendum2015.pkw.gov.pl/341_Polska
National Referendum 2016. (2016). Central Electoral Commission. Retrieved from http://results.cik.bg/
pvrnr2016/tur1/referendum/index.html
CONTEMPORARY POLITICS 45
Referendum 2015 – General information. (2015). Statistical Office of the Slovak Republic. Retrieved
from http://volby.statistics.sk/ref/ref2015/en/info.html
Rybar, M., & Sovcikova, A. (2016). The 2015 referendum in Slovakia: Direct democracy notes. East
European Quarterly, 44(1–2), 79–88.
Schuck, A. R. T., & de Vreese, C. H. (2009). Reversed mobilization in referendum campaigns: How positive
news framing can mobilize the skeptics. The International Journal of Press/Politics, 14(1), 40–66.
Shu, M. (2009). Coping with a two-dimensional political space: Party mobilisation in referendums on
European integration. European Journal of Political Research, 48(3), 397–431.
Siune, K., & Svensson, P. (1993). The Danes and the Maastricht treaty: The Danish EC referendum of
June 1992. Electoral Studies, 12(2), 99–111.
Siune, K., Svensson, P., & Tonsgaard, O. (1994). The European Union: The Danes said ‘no’ in 1992 but
‘yes’ in 1995: How and why? Electoral Studies, 13(2), 107–116.
Silagadze, N., & Gherghina, S. (2018). When who and how matter: explaining the success of referendums
in Europe. Comparative European Politics, 16(5), 905–922.
Slovak conservatives fail to cement gay marriage ban in referendum. (2015). The Guardian. Retrieved
from https://www.theguardian.com/world/2015/feb/08/slovak-conservatives-fail-gay-marriage-ban
Smrek, M. (2015). The Failed Slovak Referendum on “Family”: Voters’ Apathy and Minority Rights in
Central Europe, Baltic Worlds. Retrieved from http://balticworlds.com/the-failed-slovak-
referendum-on-“family”/
Stoychev, S. P. (2017). The 2016 referendum in Bulgaria. East European Quarterly, 45(3–4), 187–194.
Szczerbiak, A. (2015). Who won Poland’s ‘referendum war’ and how will it affect the October election? LSE
European Politics and Policy (EUROPP) Blog. Retrieved from https://polishpoliticsblog.wordpress.com/
2015/09/11/who-won-polands-referendum-war-and-how-will-it-affect-the-october-election/
Tilles, D., & Bill, S. (2015). Poland’s referendum and election: The big issues no one’s talking about, notes
from Poland. Retrieved from https://notesfrompolandcom/2015/09/04/polands-referendum-and-
election-the-big-issues-no-ones-talking-about/
Tversky, A., & Kahneman, D. (1981). The framing of decisions and the psychology of choice. Science,
211(4481), 453–458.
Uleri, P. V. (2002). On referendum voting in Italy: YES, NO or non-vote? How Italian parties learned to
control referendums. European Journal of Political Research, 41(6), 863–883.
Valkovičová, V. (2017). “Regrettably, it seems that breaking one border causes others to tumble”:
nationalism and homonegativity in the 2015 Slovak referendum. POLITIQUE EUROPÉENNE, 55,
86–115.
World Bank. (2017). Individuals using the Internet (% of the population). Retrieved from https://data.
worldbank.org/indicator/it.net.user.zs
Zaller, J. R. (1992). The nature and origins of mass opinion. Cambridge: Cambridge University Press.
Zordova, D. (2015). A Take on Slovakia’s Same-Sex Referendum, The Central European Journal of
International and Security Studies. eContribution. Retrieved from http://www.cejiss.org/
econtribution/a-take-on-slovakia-s-same-sex-referendum
46 S. GHERGHINA AND N. SILAGADZE