campaign effects and voter behaviour in referendums

Our academic experts are ready and waiting to assist with any writing project you may have. From simple essay plans, through to full dissertations, you can guarantee we have a service perfectly matched to your needs.

GET A 40% DISCOUNT ON YOU FIRST ORDER

ORDER NOW DISCOUNT CODE >>>> WELCOME40

European Union Politics
2014, Vol. 15(2) 235–254
! The Author(s) 2013
Reprints and permissions:
sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/1465116513514780
eup.sagepub.com
Article
Emotions and voting in
EU referendums
John Garry
Queen’s University Belfast, UK
Abstract
There is an emerging scholarship on the emotional bases of political opinion and behaviour
and, in particular, the contrasting implications of two distinct negative emotions –
anger and anxiety. I apply the insights in this literature to the previously unresearched
realm of the emotional bases of voting in EU referendums. I hypothesise that anxious
voters rely on substantive EU issues and angry voters rely on second-order factors
relating to domestic politics (partisanship and satisfaction with government). Focusing
on the case of Irish voting in the Fiscal Compact referendum, and using data from a
representative sample of voters, I find support for the hypotheses and discuss the
implications of the findings for our understanding of the emotional conditionality of
EU referendum voting.
Keywords
Emotions, EU referendums, Ireland, voting
Introduction
Much of the research to date on voting behaviour in EU referendums has focused
on the distinction between the ‘second-order national election’ approach and the
‘substantive issues’ approach (Glencross and Trechsel, 2011). According to the
former, originally elaborated by Reif and Schmitt (1980) to explain electoral behaviour
in European Parliament elections, citizens are driven by domestic political
factors when voting on an ostensibly European issue. Thus, citizens may use the
referendum as a vehicle for casting judgement on their national politicians: if
unsatisfied with the domestic government’s performance, they can ‘punish’ the
government by refusing to support the EU treaty it is advocating in the referendum;
if satisfied, they may reward the government by supporting it and voting yes
Corresponding author:
John Garry, Queen’s University Belfast, 25 University Square, Belfast BT7 1PA, UK.
Email: j.garry@qub.ac.uk
at the referendum (Franklin, 2002; Franklin et al., 1994, 1995). Voters may also be
driven by domestic factors in the sense of relying on their partisan attachments.
They may follow the line of their party when voting in the referendum, relying on
party cueing rather than an in-depth reflection upon issues relating to the specific
referendum under consideration (Hobolt, 2006, 2007).
In contrast, the substantive issues approach is more consistent with the notion of
deliberative democracy and assumes that citizens do actually engage with, and
reflect reasonably upon, European issues when voting in an EU referendum
(Siune and Svensson, 1993; Siune et al., 1994; Svensson, 1994, 2002). This may
occur at one of two levels. Voters may rely on their underlying ideological disposition
towards European integration, with pro-integrationists supporting a treaty
and eurosceptics opposing it. Voters may also be driven by their assessment of
aspects of the specific content of the treaty at issue: those who perceive positive
consequences resulting from treaty acceptance will favour the treaty while those
calculating negative consequences will be opposed (on this two-level distinction, see
Garry, 2013).
The implications of the second-order versus substantive issues debate are
important for understanding citizens’ individual level political behaviour in EU
referendums and also for normative debates surrounding the use (or non-use) of
referendums as a means of legitimising regional integration advance. If national
domestic political concerns, unrelated to the actual content of the treaty being
voted upon, drive behaviour then a normative defence of referendums as facilitators
of calm, reasoned and evidence-based reflection and decision making is diffi-
cult (see Moravcsik, 2006). However, the evidence to date suggests that citizens are
indeed more likely to rely on substantive issues than on second-order factors when
reaching judgement in an EU referendum context (see discussion in Garry, 2013).
This article seeks to advance knowledge in this field by investigating the emotional
conditions under which either second-order or substantive issues are relied
upon to drive voting. Previous work on conditionality has focused on the contextual
level (for instance, the more intense the EU referendum campaign, the greater
the role of substantive issue-voting and the less important the role of second-order
factors; Hobolt, 2006, 2007) or the individual level (for instance, it is argued that
issue voting in EU referendums is associated with relatively sophisticated voters
and second-order voting with less sophisticated voters; Hobolt, 2005, 2009). Here,
I focus on the individual level and investigate the previously unresearched role
played by emotions in driving the relative importance of second-order and issue
factors in EU referendum voting. Recent work on the emotional basis of political
behaviour has moved away from comparing positive to negative affect and instead
focuses on the distinct implications of different types of positive and negative
affect – in particular isolating the contrasting implications for political behaviour
of the two negative emotions of anger and anxiety (see Feldman et al., 2012 for an
overview). Notably, research from a neuropsychology perspective (Marcus, 2002;
Marcus et al., 2000) suggests that anxiety and fear are associated with a decrease in
reliance on habitual behaviour and an increase in the use of reason, information
236 European Union Politics 15(2)
search and reflective judgement. In direct contrast to anxiety, ‘anger may lead
people away from the kind of deliberation that many democratic theorists cherish’
(Valentino et al., 2008: 267) and is instead associated with behaviour driven by
long-standing political allegiance (MacKuen et al., 2010) or by the desire to blame
and punish (Petersen, 2010). Research by Druckman and McDermott (2008) also
suggests a direct link between the type of negative emotion and attitudes to ‘risky’
policies: angry citizens are likely to support a risky option while anxious citizens
will oppose it.
I apply these insights to the realm of EU referendum voting. First, I hypothesise
that anxious and fearful citizens are more likely than angry citizens to be issue
voters who learn about, reflect upon and vote on the basis of the implications of the
substantive content of the treaty. Second, I hypothesise that angry citizens are more
likely than anxious and fearful citizens to rely on second-order factors, either using
the referendum as a vehicle to pass judgement on their domestic government
(blaming and punishing their government if they are dissatisfied with its performance)
or relying on habitual partisan-based voting (in line with the recommendation
of their favoured party). Third, I hypothesise a direct effect of anger on
support for the referendum option that is portrayed in the campaign as ‘risky’
and a direct effect of anxiety on support for the particular referendum option
‘framed’ as ‘non-risky’.
I test these general hypotheses in the specific case of recent voting behaviour in
the Irish referendum on the ‘Fiscal Compact’ treaty,1 the only state to have a
referendum on the Treaty agreed to by all member states except the UK and the
Czech Republic in March 2012. Using data from a post-referendum voter survey,
I empirically demonstrate that issue voting (and particularly treaty-specific considerations)
is associated with anxiety while second-order voting (specifically, the
impact of government evaluation) is associated with anger. A strong direct relationship
also emerges between anger and voting for the ‘risky’ No option.
The emotional bases of voting in EU referendums
Emotions and voting
Party identification theory, which assumes that a tendency to vote for a particular
party is the result of an affective attachment to that party, has been the dominant
psychological approach in the electoral behaviour literature. Ever since it was originally
elaborated by Campbell et al. (1960), party identification theory has
informed almost every national and international election study and spawned hundreds
of analyses testing and contesting the approach. Although sometimes conceptualised
and operationalised in terms of both positive and negative affect
(Richardson, 1991), party identification is typically studied in terms of the
relationship between positive affect (positive party attachment) and vote choice.
Other psychological approaches to understanding voting have pitched general positive
affect against general negative affect, for example using thermometer scales
Garry 237
(how warm or cold do voters feel towards specified political actors?) or likeability
scales (to what extent do voters like or dislike certain actors?).
In contrast to these approaches, the influence of particular types of emotions on
vote choice has played – until very recently – a minor role. An influential point of
departure is the Theory of Affective Intelligence (TAI) developed by Marcus and
co-authors (Marcus and MacKuen, 1993; Marcus et al., 2000). The TAI draws
upon a neuro-psychological approach to emotions and specifies a number of emotional
subsystems that govern the extent to which humans rely upon either habit or
reason. The ‘disposition’ system induces enthusiasm that in turn triggers habitoriented
behaviour. In contrast, the ‘surveillance’ system monitors the environment
for potentially threatening stimuli; if such stimuli are detected, anxiety is induced.
Anxiety cuts off reliance on habit-related activity and instead prompts conscious
attention, information gathering and reasoning to deal with the potential threat.
Marcus et al. (2000) demonstrate that in relation to voting in US elections, anxiety
leads to a decrease in reliance on (habit-oriented) partisan-based behaviour and an
increase in reasoning and reliance on substantive political issues. While such findings
have attracted methodological criticism (Ladd and Lenz, 2008, 2011), Marcus
et al. (2011; see also Brader, 2011) have sought to demonstrate their robustness.
Thus, somewhat counter-intuitively given that it is a negative emotion, anxiety
may well enhance the quality of democracy as it is associated with facilitating citizens
to engage in evidence-based consideration of opposing points of view. Accordingly,
Valentino et al. (2008) investigate whether a worried citizen a good citizen. In their
analysis, the authors contrast anxiety with a second distinct negative emotion: anger.
They conclude that ‘concern about fear mongering in politics may be at least somewhat
overblown … outrage, perhaps, is more damaging than fear if we hope to foster
an informed citizenry’ (267). Similarly, MacKuen et al. (2010) argue that anger
(which they refer to as aversion) contrasts with anxiety in that the former operates
within the ‘disposition’ system and is consequently more likely than anxiety to be
associated with habit-oriented behaviour. The authors’ state:
A standard expectation is that when people get emotional about politics, their positions
become entrenched, resolute, and steadfast. Our theory predicts that this will be
true – for aversion but not for anxiety … people can engage either as deliberative or
partisan citizens, where the kind of citizenship practiced will depend upon whether
people react with anxiety or with anger. (MacKuen et al., 2010: 442)
Petersen (2010: 358) also directly contrasts anxiety and anger, suggesting that
the two emotions ‘are not only distinct but operate in distinct domains of political
life’. While, in line with TAI, he suggests that anxiety prompts the consideration of
issues, he draws on evolutionary psychology research to suggest that anger is a
moral emotion and is associated with the desire to attribute blame to (and then
punish) particular actors who are responsible for negative events.
In addition to these moderating (conditional or indirect) effects of negative
emotions on political opinion (i.e. anxiety is associated with issue-based
238 European Union Politics 15(2)
deliberation while anger is associated with habit-based partisanship and blaming/
punishing), it is also likely that there is a direct impact of negative emotion on
opinion formation. Druckman and McDermott (2008) argue that there is a direct
link between anger and support for ‘risky’ policies and a direct link between anxiety
and support for ‘non-risky’ policies. This is due to the different levels of certainty
associated with feelings of anger and anxiety. Drawing on the work of Lerner et al.
(2003) and Smith and Ellsworth (1985), Druckman and McDermott (2008: 301)
suggested that that ‘hostility or anger produces optimism about future outcomes
(due to certainty) and risk-seeking choices… In contrast, distress or fear leads to
pessimistic judgements (due to uncertainty) of the future and greater risk aversion’.
The authors conducted experiments inviting participants to choose from a set of
possible responses to disease outbreak and the need for investment, with the
response options varying in terms of certainty and riskiness. When the emotional
state of the respondents was related to option choice, they found that ‘anger
encourages greater risk seeking while distress encourages a more cautious
approach’ (Druckman and McDermott, 2008: 317).
Emotions and EU referendum voting
These insights map fairly neatly onto the second-order versus issues debate relating
to vote determinants in EU referendums. The ‘issues approach’ suggests that citizens
do rely upon EU related factors when deciding how to vote in an EU referendum
(Siune and Svensson, 1993; Siune et al., 1994; Svensson, 1994, 2002).
However, an important distinction within this approach relates to whether citizens
rely upon underlying ideological disposition or treaty-specific assessments (Garry,
2013). Voters who decide on the basis of underlying ideology may be distinguished
in terms of eurosceptics who are generally opposed to further EU integration and
euro-integrationists who are generally supportive of further unity. The scepticintegrationist
ideological distinction may have roots in a number of sources: economic
factors (Anderson and Reichert, 1996; Eichenberg and Dalton, 2007; Gabel,
1998a, 1998b), culture (de Vreese and Boomgaarden, 2005; McLaren, 2002) or
identity (Carey, 2002; McLaren, 2004). A somewhat different approach focuses
on citizens’ evaluations of the specific treaty under consideration rather than
their more long standing disposition towards the EU project. Citizens are assumed
to be information seekers who engage with and reflect upon debates relating to the
treaty at issue. If, after consideration of the specific content of the treaty, citizens
regard the consequences of acceptance of the treaty as positive, then they are
assumed to want to accept the treaty. Citizens who foresee deleterious consequences
in the event of acceptance are assumed to vote no.
The extent to which perceptions of the consequences of the treaty drive referendum
vote choice may plausibly be conditional upon the inducement of anxiety
during the campaign. The argument that it is citizens’ calculations regarding treaty
consequences that shape their vote choice essentially assumes that the treaty is a
potential threat: citizens’ attention to the treaty is triggered, leading to the search
Garry 239
for treaty-related information and reliance on reason and judgement designed to
minimise the perceived threat relating to either acceptance or rejection of the treaty.
Such information seeking and reasoning is consistent with the consequences of
anxiety induction, as described in the previous section.
The influence of underlying values – the sceptics versus integrationists ideological
distinction – on vote choice is also plausibly conditional upon anxiety.
Reliance upon EU ideological disposition represents citizen engagement with the
substantive domain of EU politics (rather than merely voting in line with your own
party or rewarding/punishing the government) and so is consistent with the induction
of anxiety and the consequent triggering of attention and reason. However, the
conditional effect is likely to be less strong than that relating to treaty-specific issues
given that the sceptic-enthusiast distinction is, to an extent, a general short-cut
heuristic rather than a direct engagement with and reflection upon the specific
issues relating to the treaty. Hence, both issue effects (ideology and treaty specific
considerations) are likely to be conditional upon the induction of anxiety; strongly
so, for treaty-specific issues that are indicative of attention and learning, and less
strongly so for ideology as it is indicative of reliance on an underlying disposition.
As far as ‘second-order’ voting is concerned, this approach has mainly been
studied in the context of European Parliament elections, with analysts investigating
the extent to which domestic political concerns drive voting in what is ostensibly an
‘EU’ election (Reif and Schmitt, 1980; for a recent analysis, see Hix and Marsh,
2011). The approach may also valuably be applied to EU referendums (see discussion
in de Vreese and Semetko, 2004: 701–702) in order to assess whether voting
behaviour is driven by the substantively important political issue (the EU referendum)
or domestic political concerns such as partisanship and evaluation of government
performance. In an EU referendum campaign, citizens in whom the treaty
referendum campaign induces anxiety are unlikely to rely upon second-order factors
when deciding how to vote. Notably, anxious citizens are unlikely to be influenced
by habit-oriented heuristics such as party cueing and partisanship. This is not
to argue that partisan-based voting will not be an important driver of voting,
consistent with previous work (Hobolt, 2006, 2007), but its level of importance
will depend upon emotional disposition. Anxious voters will be focused on the
substantive details of the campaign and unlikely to rely on the partisan heuristic.
Given that, according to TAI, anger (an example of aversion) is associated with the
habit oriented disposition system (rather than the surveillance system with which
anxiety is linked), angry citizens are hypothesised to be more likely than anxious
citizens to rely on partisanship when voting in the referendum (see discussion in
MacKuen et al., 2010: 441–442).
In addition to partisanship, the other aspect of second-order voting relates to the
use of referendums to pass judgement on the performance of the domestic government.
Anger is more likely than anxiety to prompt voters to ignore the substantive
EU issues in the campaign and simply use the referendum as a chance to express
their evaluation of the national government. Building on the work of Petersen
(2010) and Lazarus (1991), anger encourages voters to identify responsible actors
240 European Union Politics 15(2)
and pass judgement. During a referendum campaign, consistent with the secondorder
referendum voting approach, voters are likely to direct their anger at the
national government and to blame the government; they may hold it to account at
the time of referendum voting and punish it by voting against the government’s
preferred position on the referendum. Hence, previous work highlighting the
importance of domestic government evaluation in EU referendum vote choice
(Franklin, 2002; Franklin et al., 1994, 1995) is valid, but is expected to be dependent
upon the emotional make-up of the voter: angry voters are more likely than
anxious voters to operate in the moral domain of holding responsible and passing
judgement on the government.2
H1: Anxious citizens are more likely than angry citizens to be ‘issue’ voters who vote
on the basis of the substantive content of the EU Treaty.
H2: Angry citizens are more likely than anxious citizens to be ‘second-order’ voters
whose voting is driven by their attitude to domestic politics (government approval and
partisanship).
It is worth explicitly emphasising that the conditional nature of the hypotheses just
discussed prompts a consideration of what Petersen (2010: 359) notes as ‘the diffi-
culties associated with analyzing emotions and cognitions’. Petersen highlights
Huddy et al.’s statement (2007: 210) that ‘cognitions and emotions operate in a
recursive loop in which cognitions influence emotions which influence cognitions,
in turn—making it difficult to know whether cognitive appraisals are a cause or
consequence of differing emotional states’. Petersen (2010: 359) suggests that emotions
and cognitions co-occur and thus the aim of empirical analysis should be to
use interactive modelling to demonstrate that the effects of a specified explanatory
factor co-occur with a specified emotional state. Accordingly, the aim of the current
study may be seen as somewhat modest in terms of making causal claims: the
aim is to identify whether certain specified negative emotional states (anger or fear)
are associated with certain specified explanations of referendum voting: is anxiety
linked with issue-based voting and is anger linked with either partisan or blamebased
voting?
In addition to the moderating effects of anger and anxiety, there are also likely to
be direct effects. Druckman and McDermott’s (2008) findings regarding the direct
link between anger and choice of risky policy option (and anxiety and choice of
non-risky policy option) have implications for referendum vote choice. If voting in
a particular way in a referendum (e.g. voting No) is, during the referendum campaign,
strongly ‘framed’ or portrayed as a very risky option then angry voters will
tend to vote No. Similarly, if voting in a particular way (e.g. voting Yes) is characterised
during the campaign as the risk-free option, anxious voters will tend to
vote Yes.
H3: Angry citizens are more likely than anxious citizens to vote for the referendum
option which is characterised in the campaign as the ‘risky’ option.
Garry 241
Whether it is the Yes or No option that is ‘framed’ in the campaign as the risky
option is obviously dependent on the specific campaign. The Irish referendum
campaign is now briefly focused on in order to situate substantively and contextually
the generally applicable hypotheses that have just been explicated.
The case of Ireland and the ‘Fiscal Compact’ treaty
The European Fiscal Compact was designed to implement a set of financial rules in
participating states that would enhance the long-term survival chance of the Euro
currency, essentially placing constraints on states’ borrowing and spending activities.
In return for these accepted constraints, states would have access to a central
fund that would prevent complete state bankruptcy by providing emergency capital
in the event of a state being locked out of international capital markets.
At the time of the referendum, the effectively bankrupt Irish state was being kept
financially alive by a bailout from the combined forces of the IMF and the European
Union. Pro-treaty actors emphasised that an even worse economic catastrophe
would descend in the event of a rejection of the treaty, as rejection would prohibit
access to any further bailout funds. Pro-treaty actors argued that such a bailout was
looking ominously necessary as Ireland remained stubbornly isolated from international
funding markets. Anti-treaty actors emphasised that the main Irish parties
and the EU had generated the crisis and were not deserving of trust or support; they
argued that rejection of the treaty would not place Ireland in a precarious and vulnerable
economic position, but would rather put an end to the wrong-headed austerity
approach to economics pushed by the government and international lenders.
The Fine Gael/Labour coalition government portrayed itself as behaving
responsibly in the crisis but was seen by its detractors as behaving reprehensively,
breaking its general election promises related to obtaining an improved deal for
Ireland and instead imposing austerity measures upon an already suffering population.
Only one party was clearly opposed to the Treaty: Sinn Fe´in. It ran an
energetic anti-Treaty campaign that condemned the anti-growth pro-austerity
treaty and what it regarded as the scaremongering tactics of the government who
portrayed No voting as profoundly risky behaviour in a highly uncertain economic
climate. However, it was the Yes side that achieved victory: on polling day, 31 May
2012, Irish citizens voted yes by a margin 60.3% to 39.7%, with a turnout of 50%
(for a description of the campaign, see Fitzgibbon, 2013).
The Irish referendum on the Fiscal Treaty is arguably a good case to focus on to
demonstrate the effects of anxiety and anger on referendum vote choice. The referendum
was very plausibly a novel and threatening stimulus on the Irish political
environment. Ireland was already in a very precarious economic state and much of
the debate related to the strong possibility of a radical worsening of the situation,
with the spectre of Ireland literally running out of money to keep hospitals and
schools functioning. In such a campaign context, the inducement of anxiety in
many citizens is likely. This referendum represented not just another run of the
mill ordinary election or mundane referendum; the economic survival of the state
242 European Union Politics 15(2)
was at issue. This novel threat was likely to trigger, in many citizens, a discontinuation
of the ‘every day habit-based behaviour’ associated with the disposition
system and its replacement with a focused attention on the threat (the referendum)
and how to react to it in a way that could minimise reduction in well-being.
The nature of the Irish referendum campaign was also likely to incite anger at
those deemed responsible for the country’s disastrous position. Critics argued that
the government was not a passive actor in the Irish crisis; rather it played a key role
in exacerbating the economic difficulties. The government was formed in 2011 on
the back of election promises to renegotiate Ireland’s relationship with its international
lenders to achieve an amelioration of its position. It attracted much criticism
for failing to do so, leaving Labour’s general election campaign slogan ‘It’s
either Labour’s way or Frankfurt’s way’ looking like a slogan the party would
rather forget.
In short, there is good reason to expect many voters to be induced to be either
angry or anxious in the Irish ‘Fiscal Compact’ referendum campaign and this
provides fertile ground for assessing the possible (direct and moderating) effects
of such emotions on referendum vote choice. Indeed, in an interview with an Irish
Times journalist after the referendum, one close observer of the referendum campaign,
Gerard O’Neill of Ama´rach Research, interpreted the result of the referendum
and the nature of the campaign very much through the lens of the fear/anger
distinction:
The great struggle was between fear and anger – and fear won. Anger is about the past
and the present, about what has happened to people’s lives, about cuts and unemployment
and emigration … Fear is about the future – about what might happen to the
euro, to the world, to Ireland, to the economy. So you had that tension between anger
and the past and the present, and the referendum, which is about the future and
whether Ireland will have access to funds and that’s a vision of the future that motivates
fear. (Irish Times, 2 June 2012)
Methods
Experiments are particularly valuable with respect to identifying causal mechanisms
and have been used fruitfully with respect to the study of emotions and voting
(recent examples include Kiss and Hobolt, 2011, 2012; MacKuen et al., 2010). A
downside of experiments is that they are, by their nature, typically abstract and
divorced from the real world of campaigns and voting behaviour. Here, I rely on a
survey designed to capture voting in an actual EU referendum. While limited in
terms of unpacking causal order, survey datasets are useful in terms of identifying
conditional relationships via the use of interaction terms in appropriately structured
regression models (as emphasised by Petersen, 2010: 359) and have the
advantage of capturing important political events, such as voter behaviour in the
Irish ‘Fiscal Treaty’ referendum.
Garry 243
RedC polling company was commissioned to conduct a survey of voters (as
opposed to a survey of all citizens) in the direct aftermath of the referendum. A
representative sample of 1000 voters was achieved via telephone interviewing.3 In
order to generate a measure of voters’ perceptions of the likely economic consequences
of acceptance of, or rejection of, the treaty the following questions were
asked:
Now that Ireland has voted ‘Yes’ in favour of the Fiscal Treaty, I would like you to
think about how much you agree or disagree with each of the following statements
about voting in favour of the treaty. A ‘Yes’ vote will …
… make the Irish economy more stable and secure
… help Ireland avoid complete bankruptcy
(disagree strongly, disagree, agree, agree strongly, don’t know)
If Ireland had voted ‘No’ and rejected the Fiscal Treaty do you think this would have …
… led to great uncertainty about Ireland’s economic future
(disagree strongly, disagree, agree, agree strongly, don’t know)
Responses to these three items were summed and rescaled to run from 0 to 1, with
higher scores indicating strongly negative perceptions of the economic consequences
of treaty rejection. In order to measure citizens’ underlying values regarding
the EU, the following question was asked:
As regards the European Union in general, which of the following statements comes
closest to your view? Ireland should do all it can to unite fully with the European
Union VERSUS Ireland should do all it can to protect its independence from the
European Union
In order to tap voters’ satisfaction with the Fine Gael/Labour government,
respondents were asked:
How satisfied are you with the performance of the Fine/Gael Labour government:
very satisfied, fairly satisfied, not very satisfied or not at all satisfied?
In relation to measuring partisanship, respondents were asked which party they
would give their first preference vote to if a general election were to be held. Three
categories are generated: supporters of parties clearly in favour of the Treaty (Fine
Gael, Labour, Fianna Fa´il), supporters of the party clearly against the Treaty (Sinn
Fe´in),4 and other respondents.
In order to measure whether voters were ‘angry’ or ‘anxious/fearful’, the following
question was asked:
During the referendum campaign, when people were talking about the Fiscal Treaty,
some people said they were angry and other people said they were fearful. What about
you, which did you feel most – anger or fear?5
244 European Union Politics 15(2)
The word ‘fear’ was used rather than ‘anxious’ or ‘anxiety’ in order to make the
question as easy as possible for the respondent to answer. As noted by Bogdan
(2012), while fear and anxiety may not be precisely the same (Rhudi and Meagher,
2000), they are very similar (Davis, 1992). Furthermore, the terms are often used
interchangeably in the literature (e.g. Petersen, 2010: 360, 363; see also the discussion
in Druckman and McDermott, 2008: 317). The forced-choice nature of the
question may be of some concern. Again, the aim was to ask the question in as
simple and colloquial a way as possible. This style of question loses nuance (exactly
how fearful, exactly how angry?), but it does achieve the desired aim of a straightforward
categorisation of voters (see discussion of the measurement of emotions in
Marcus et al., 2006).
A number of questions were also asked to generate control variables that,
although not directly relevant for testing the hypotheses, have been shown to be
predictors of referendum vote choice in previous research. Social class was identified
using the usual social grade categories used in the UK (distinguishing between
professional (abc1) and manual (c2de) groups (as well as farmers)). The sex of the
respondent was noted and age was measured in categorical form. Some authors
also include a general measure of ideology, such as respondents’ self-placement on
a left–right scale, as a control variable. In the questionnaire used in this study, left–
right self-placement was not included. In any event, it may not have served as a
useful control variable given that the meaning of the measure is ambiguous
(Inglehart and Klingemann, 1976; Knutsen, 1997), meaning that it would not be
entirely clear what in fact left–right self-placement would be controlling for.
Results
In order to test the hypotheses regarding the impact of negative emotion on referendum
vote choice, a number of binary logistic regressions were conducted with
vote choice as the dependent variable (voted Yes rather than voted No). Table 1
reports four models, beginning with a regression including all predictors except the
emotion variable, then a regression including the emotion variable, followed by
analyses in which respondents are split by emotion. In model 1 of Table 1, the
‘issues’ variables (EU ideology and perceptions of consequences of the treaty) and
the second-order factors (partisanship and level of satisfaction with the domestic
government) are entered, in addition to controls (age, social class and sex). The
variables predict voting in the expected manner. Pro-integration voters and voters
who perceive negative economic consequences resulting from treaty rejection tend
to vote Yes and Sinn Fe´in supporters and voters who are dissatisfied with the
government tend to vote No. All variables run from 0 to 1 and a comparison of
coefficients indicates that issue-related factors, and in particular perceptions of
treaty consequences, are stronger predictors than second-order factors. In model
2, the variable capturing the distinction between angry and anxious voters is also
entered and is highly statistically significant. Fearful voters are more likely than
angry voters to support the treaty. This simple model confirms H3, which posited
Garry 245
that the option that was characterised in the campaign as ‘risky’ (voting No) was
associated with anger while the option that was characterised as relatively ‘riskfree’
(voting Yes) was associated with anxiety.
A preliminary investigation of the hypothesised conditioning effects of anger
and fear is conducted by re-running model 1 but this time splitting the analysis by
emotion type. In the regression containing only angry voters (model 3), a strong
Table 1. Binary logistic regression predicting Yes (rather than No) vote.
Model 1 Model 2
Model 3
(angry voters)
Model 4
(fearful voters)
Favours EU integration 1.685*** 1.629*** 1.202** 2.292***
(.287) (.293) (.416) (.473)
Perceives positive economic
consequence from acceptance
of the treaty
7.396*** 7.027*** 6.089*** 9.190***
(.669) (.671) (.873) (1.228)
Dissatisfied with the performance
of the government
2.003*** 1.681** 3.045*** .051
(.536) (.548) (.767) (.904)
(Supporter of a pro-Treaty party¼ref)
Sinn Fe´in supporter 1.065* 1.131* 1.545* 1.351+
(.451) (.471) (.768) (.715)
Other response to party support question .510+ .387 .427 .294
(.310) (.321) (.425) (.519)
Under 35 years old .151 .252 .161 .299
(.305) (.316) (.423) (.504)
Female .157 .002 .056 .194
(.290) (.300) (.421) (.468)
(Professional class¼ref)
Manual class .473 .642* 1.080* .238
(.292) (.302) (.436) (.460)
Farmers .249 .256 .093 .339
(.672) (.691) (.921) (1.208)
Fearful 1.121***
(.297)
Constant 3.125*** 3.650*** 2.010+ 5.151***
(.619) (.650) (.797) (1.133)
n 724 724 321 403
Cox & Snell r-square .565 .573 .554 .444
Models 1 and 2 include all voters who indicated that they were either fearful or angry. ‘Favours EU integration’
is a dichotomous variable whereby 1¼favours integration and 0¼opposes integration; ‘Perceives positive
economic consequences of treaty acceptance’ is a 0 to 1 scale, with higher values representing greater
perception of positive consequences’; ‘Dissatisfied with the performance of the government’ is a 0 to 1
scale, with higher values indicating greater dissatisfaction with government performance.
Coefficients are log odds (SE).
Statistical significance level: + .1; * .05; ** .01; *** .001.
246 European Union Politics 15(2)
and highly statistically significant relationship emerges between dissatisfaction with
the government and voting No. In striking contrast, for anxious voters (model 4),
there is no relationship at all between being dissatisfied and voting No. In addition,
the coefficients relating to the issues variables (ideology and perceptions of treaty
consequences) are stronger for anxious voters than for angry voters. While these
preliminary findings are in line with my expectations, the effect of partisanship on
referendum vote choice appears quite similar across emotional type, suggesting no
emotional conditioning effect.6
In order to test the conditioning effects formally, a further binary logistic regression
is conducted in which ‘issues’ variables, second-order variables and control
variables are entered along with the emotion variable and interactions between the
emotion variable and each of the issues and second-order variables (see Table 2).
H1 is supported. Perceptions of the economic consequences of the treaty is a
Table 2. Binary logistic regression predicting Yes (rather than No) vote.
Favours EU integration 1.171** (.402)
Perceives positive economic consequences from acceptance of treaty 6.012*** (.850)
Dissatisfied with the performance of the government 3.033*** (.758)
(Supporter of a pro-Treaty party¼ref)
Sinn Fe´in supporter 1.524* (.738)
Other response to party support question .403 (.418)
Fear 3.298** (1.286)
Fear Favours EU integration 1.066+ (.616)
Fear Perceives positive economic consequence from acceptance of treaty 3.173* (1.471)
Fear Dissatisfied with the performance of the government 3.256** (1.163)
Fear Sinn Fe´in supporter .183 (1.004)
Fear Other response to party support question .147 (.666)
Under 35 years old .204 (.321)
Female .090 (.310)
(Professional class ¼ ref)
Manual class .692* (.311)
Farmers .173 (.730)
Constant 1.940* (.759)
N 724
Cox & Snell r-square .582
Included are all voters who indicated that they were either fearful or angry. ‘Favours EU integration’ is a
dichotomous variable whereby 1¼favours integration and 0¼opposes integration; ‘Perceives positive economic
consequences of treaty acceptance’ is a 0 to 1 scale, with higher values representing greater perception
of positive consequences’; ‘Dissatisfied with the performance of the government’ is a 0 to 1 scale, with higher
values indicating greater dissatisfaction with government performance.
Coefficients are log odds (SE).
Statistical significance level: + .1; * .05; ** .01; *** .001.
Garry 247
stronger predictor of vote choice for fearful people than for angry people and the
interaction (fear perceives positive economic consequence from acceptance of the
treaty) is statistically significant at the .05 level. Similarly, ideological disposition
towards the EU (favouring unity) is a stronger predictor of voting Yes for fearful
people than for angry people. However, the strength of the conditionality, consistent
with expectations, is weaker than was the case for treaty-specific perceptions:
the fear favours EU integration interaction is statistically significant at .1 level
(although unusual to report effects at .1 level, it is deliberately done here as there is
an expectation that there will be an interaction effect for ideology but it will be
weaker than that for perceptions of consequences).
The impact of dissatisfaction with the performance of the government on referendum
vote choice is stronger for angry voters than for anxious voters to a highly
statistically significant extent. However, partisan-based voting is not stronger for
angry voters than for anxious voters. Thus, H2 is partially supported.7
The substantive size of these statistically significant interaction effects is illustrated
in Figure 1.8 The graphs plot the difference, between angry and fearful
voters, of a change in the probability of voting Yes given a shift in the predictor
variable. Figure 1(a) shows a large difference between the two emotional groups in
terms of the effect of dissatisfaction on voting Yes. Moving from a position of one
standard deviation below the mean on the ‘dissatisfied’ scale to a position one
standard deviation above the mean implies no decrease in the probability of
voting Yes for anxious voters (essentially a flat line in Figure 1(a)), but does
imply a large decrease in the probability of voting Yes for angry voters (a decrease
from .73 to .30). An increase in voters’ positive perceptions of the consequences of
acceptance of the Treaty (from one standard deviation below the mean of the scale
to one standard deviation above the mean) results in an increase in the probability
of voting Yes for both angry and anxious voters (see Figure 1(b)), but the increase
is greater for the latter than for the former (an increase of .58 compared to an
increase of .45). Finally, Figure 1(c) illustrates that a movement from a sceptic
position to a pro-integration position is related to an increase in the probability of
voting Yes for both types of emotional voters, but the increase is greater for anxious
voters than for angry voters.
Discussion and conclusions
The EU, and more specifically the Euro-zone, is in a state of prolonged economic
crisis. Many commentators believe that the only way for the Euro to survive is for
the EU to radically change in order to achieve a dramatic and speedy financial,
economic and political integration that would serve to solidify the currency.
Significant moves in an integrative direction would necessitate referendums in
many states and such referendums would likely occur in relatively turbulent
crisis-like conditions. It seems plausible to expect that emotions such as anger
and anxiety would play an important role in such campaigns. The Irish case studied
in this paper can shed some light on how emotions may influence European
248 European Union Politics 15(2)
citizens’ reactions to referendum campaigns and how they vote. Emotions and
cognition are often depicted as opposites, with reason-based thinking and behaviour
compromised by emotional factors. Recent research, however, shows that
emotions govern or regulate cognitive resources and that affect and rationality
(a)
(b)
(c)
_______
———
Probability of
vong Yes
Probability of
vong Yes
anger
fear
0.2
0.4
0.6
0.8
1
less dissafied more dissasfied
0.2
0.4
0.6
0.8
1
less posive more posive
0.2
0.4
0.6
0.8
1
an-integraon pro-integraon
Probability of
vong Yes
Figure 1. (a) Impact on probability of voting Yes of moving one standard deviation below the
mean on the ‘dissatisfied with the government’ scale to one standard deviation above the
mean. (b) Impact on probability of voting Yes of moving one standard deviation below the
mean in the ‘positive economic consequences of treaty acceptance’ scale to one standard
deviation above the mean. (c) Impact on probability of voting Yes of moving from an anti-integration
position to a pro-integration position on the dichotomous integration measure.
Garry 249
work in tandem with each other to aid human decision making, opinion formation
and behaviour (Feldman et al., 2012). The moderating effect of different types of
negative emotions on the determinants of vote choice in EU referendums highlights
an important development of the second-order versus issues debate. Issue voting
(and particularly treaty-specific considerations) is associated with anxiety while
second-order voting (specifically, the impact of government evaluation) is associated
with anger. The central argument in this paper is that emotional heterogeneity
is a conditioning influence on EU referendum vote choice and should be
integrated into future research on EU referendum voting.
The results may also have implications for how we think about the campaign
strategies of parties and governments during referendums. Emotional rhetoric
employed by competing political actors may hinge on possible direct emotional
effects resulting from risk-related framing of referendum options. If actors frame
the option advocated by their competitors as the inherently risky option, this may
prove to be a successful strategic ploy to generate and attract anxious voters.
Emotional rhetoric may also have indirect effects. Actors who are confident that
reasonable voters would agree with them may be incentivised to use campaigning
devices that raise as much anxiety as possible in the minds of voters (who then
calmly and rationally deliberate and arrive at the ‘right’ answer). Other actors may
offset this and seek to tap into an existing well of discontent with the government
by deploying anger-related rhetoric that seeks to highlight the possibility of blame
related voting. Hence, the role of emotions in referendum campaigning may be
deployed (indirectly) to raise (or lower) the salience of different drivers of referendum
vote choice.9
More generally, this study contributes to the emerging field of research on emotional
bases of political beliefs and behaviours (see Feldman et al., 2012 for overview).
The analysis in this article supports recent work that advocates investigating
the distinct implications of different positive and negative emotions, in contrast to
earlier work on emotion that focused on the negative versus positive emotion distinction.
In particular, this study complements previous research that argues that
two specific negative emotions (anger and anxiety) have important distinct implications.
The hypotheses tested here are not drawn from a single theoretical perspective
but are rather derived from work from a neuropsychological perspective,
an evolutionary psychology perspective and a risk-oriented perspective. What these
approaches share, however, is the assumption that anger and anxiety have different
implications for political behaviour. This assumption is empirically supported in
the application to the specific case of EU referendum voting behaviour studied in
this article. Further research, focusing on more cases and more time points, may
shed light on how generalisable these findings are.
Acknowledgements
I would like to thank Josie Knowles for commenting upon an earlier draft of this paper and
would also like to thank the editor and the anonymous reviewers. I would like to acknowledge
that the data used in this paper is part of a wider project with Michael Marsh.
250 European Union Politics 15(2)
Notes
1. The intergovernmental treaty is formally entitled the ‘Treaty on Stability, Coordination
and Governance in the Economic and Monetary Union’ and is typically referred to more
informally as the European Fiscal Compact or European Stability Treaty.
2. Thus, insofar as anger enhances second-order effects, there are two distinct possible roles
it could play: strengthening the effect of either partisan factors or factors relating to
(dis)satisfaction with the government. The aim here is simply to attempt to identify
empirically which particular moderating role, if any, anger plays.
3. RedC conducted eight representative pre-referendum polls that contain both vote intention
at the Fiscal Stability treaty and party vote intentions in the event of a general
election. In each of these surveys, respondents were asked if they were happy to be recontacted
for research by RedC in the future and approximately 65–70 percent of
respondents in the surveys stated that they were. This provided a pool of approximately
5500 people that could be followed up post the referendum. Demographic quotas were set
to ensure a nationally representative sample of voters: the quotas were based on responses
to the question ‘how likely are you to vote in the referendum?’ asked in the last two prereferendum
polls (to ensure the final sample was representative of those who actually
voted, rather than all potential voters). From the pool of 5500 people, 1000 interviews
with people who voted at the Fiscal Stability referendum were completed in the days
following the referendum vote.
4. It may possibly be regarded as a concern that the operationalisation of the party cues
hypothesis essentially compares Sinn Fe´in to the main Irish parties, with the potential
that what is being captured is the peculiarities of Sinn Fe´in rather than party cueing
effects per se. However, the peculiarities of Sinn Fe´in (which arguably currently relate to
its position as a working class protest party) are controlled for via the inclusion of social
class and level of government satisfaction, with the consequence that the ‘party’ variable
likely captures, in large part, party cueing effects.
5. Respondents were not explicitly offered further response categories. In addition to
respondents who stated that they were either angry (33.1%, n¼331) or fearful (42.2%,
n¼422), 22.1% of respondents responded that they were ‘neither’ angry nor fearful. A
small number of respondents (2%) also indicated that they were ‘both’ (and two respondents
were coded as ‘don’t know’). One concern may be that because respondents were not
explicitly offered the ‘neither’ or ‘both’ categories, respondents may have felt forced to
choose either ‘anger’ or ‘fear’, hence undermining the meaningfulness of the responses. A
range of analyses were conducted to assess the ‘neither’ respondents and it emerges that, in
line with expectations, those who chose ‘anger’ or ‘fear’ were the most distinct from each
other and ‘neither’ respondents seemingly situated between the two groups (see below).
6. A separate regression analysis was conducted including only those respondents who
indicated that they were ‘neither’ angry nor fearful (see Online Appendix A). In that
regression, the coefficients for the two issue variables (underlying attitude to EU unity
and perceptions of the economic consequences of the treaty) lie between those for the
‘angry’ and ‘fearful’ groups (compare Online Appendix A to models 3 and 4 in Table 1).
The same is the case for the dissatisfaction with government variable. This suggests that
the ‘neither’ respondents substantively lie between the ‘angry’ and ‘fearful’ respondents.
7. Further analysis was conducted using a different operationalisation of emotion. ‘Angry
versus not angry’ and ‘fearful versus not fearful’ categories were generated. (‘Not angry’
is made up of those who responded that they were ‘fearful’ plus those who said they were
Garry 251
‘neither’ angry nor fearful; similarly for ‘not fearful’). These were then used as interactions
(see results in Online Appendices B and C). Focusing on the two clear substantive
findings of the ‘angry’ versus ‘fearful’ analysis, it emerges that the results are substantively
the same in these new categorisations, although the statistical significance level falls
just outside the .05 cut-off for two of the effects (p¼.055 for the interaction between
‘fearful-not fearful’ and ‘perceptions of economic consequences’ and p¼.064 for the
interaction between angry-not angry and ‘perceptions of economic consequences).
Hence, the direct contrast between anger and anxiety produces similar but somewhat
stronger results than the ‘angry versus not angry’ and ‘fearful versus not fearful’ distinctions.
This is perhaps not surprising given that the literature summarised earlier emphasises
the direct contrast of these two negative emotions.
8. For Figure 1(a): assuming a voter who has a mean score on the perceptions of treaty
consequences measure and who holds a sceptical view on EU integration. For
Figure 1(b): assuming a voter who has a mean score on the dissatisfaction with government
measure and who holds a sceptical view on EU integration. For Figure 1(c): assuming
a voter who has mean scores on the perceptions of treaty consequences measure and
the dissatisfaction with government measure. In all three analyses, the voter is assumed to
be a supporter of a pro-Treaty party and is older, male and middle class.
9. This potentially raises some endogeneity concerns when modelling vote choice. If it is
political parties (via party strategists) that tailor a particular campaign such that certain
emotional reactions of specific groups are successfully triggered, researchers need to
carefully specify the role of ‘party’ and the role of ‘emotion’ in shaping voting behaviour.
References
Anderson C and Reichert S (1996) Economic benefits and support for membership in the
EU: A cross-national analysis. Journal of Public Policy 15(3): 231–249.
Bogdan A (2012) The anxious partisan: A test of affective intelligence theory in
RomaniaAvailable at: http://www.etd.ceu.hu/2012/bogdan_alexandra.pdf (accessed 4
October 2012) (2012, ).
Brader T (2011) The political relevance of emotions: ‘Re-assessing’ revisited. Political
Psychology 32(2): 327–346.
Campbell A, Converse PE, Miller WE, et al. (1960) The American Voter. New York: Wiley.
Carey S (2002) Undivided loyalties: Is national identity an obstacle to European integration?
European Union Politics 3(4): 387–413.
Davis M (1992) The role of the amygdala in fear and anxiety. Annual Reviews of
Neuroscience 15: 353–375.
de Vreese C and Boomgaarden H (2005) Projecting EU referendums: Fear of immigration
and support for European integration. European Union Politics 6(1): 59–82.
de Vreese C and Semetko H (2004) News matters: Influences on the vote in the Danish 2000
Euro referendum campaign. European Journal of Political Research 43(5): 699–722.
Druckman JN and McDermott R (2008) Emotion and the framing of risky choice. Political
Behaviour 30(3): 297–321.
Eichenberg R and Dalton R (2007) Post-Maastricht blues: The transformation of citizen
support for European integration, 1973–2004. Acta Politica 42(1–2): 128–152.
Feldman S, Huddy L and Cassesse E (2012) Emotional underpinnings of political behaviour.
In: Sun R (ed.) Grounding Social Sciences in Cognitive Sciences. Cambridge MA: MIT
Press, pp. 127–156.
252 European Union Politics 15(2)
Fitzgibbon J (2013) Referendum briefing: The referendum on the intergovernmental treaty
on stability, coordination and governance in the the economic and monetary union in
Ireland, 31 May 2012. Representation 49(2): 229–239.
Franklin M, Marsh M and McLaren L (1994) Uncorking the bottle: Popular opposition to
European unification in the wake of Maastricht. Journal of Common Market Studies
32(4): 455–472.
Franklin M, van der Eijk C and Marsh M (1995) Referendum outcomes and trust in government:
Public support for Europe in the wake of Maastricht. West European Politics
18(3): 101–117.
Franklin M (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.
Gabel M (1998a) Public support for European integration: An empirical test of five theories.
Journal of Politics 60(2): 333–354.
Gabel M (1998b) Interests and Integration: Market Liberalization, Public Opinion, and
European Union. Ann Arbor, MI: University of Michigan Press.
Garry J (2013) Direct democracy and regional integration: Citizens’ perceptions of treaty implications
and the Irish reversal on Lisbon. European Journal of Political Research 52(1): 94–118.
Glencross A and Trechsel A (2011) First or second order referendums? Understanding the
votes on the EU Constitutional Treaty in four EU member states. West European Politics
34(4): 755–772.
Hix S and Marsh M (2011) Second-order effects plus pan-European swings: An analysis of
European Parliament elections across time. Electoral Studies 30(1): 4–15.
Hobolt S (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 (2006) How parties affect vote choice in European integration referendums. Party
Politics 12(5): 623–647.
Hobolt S (2007) Taking cues on Europe? Voter competence and party endorsement in referendums
on EU integration. European Journal of Political Research 46(2): 151–182.
Hobolt S (2009) Europe in Question – Referendums on European Integration. Oxford: Oxford
University Press.
Huddy L, Feldman S and Cassese E (2007) On the distinct political effects of anger and
anxiety. In: Neumann R, Marcus G, Crigler A, et al. (eds) The Affect Effect. Cambridge:
Cambridge University Press, pp. 202–230.
Inglehart R and Klingemann HD (1976) Partisan identification, ideological preference and
the left-right dimension among western mass publics. In: Budge I, Crewe I and Fairlie D
(eds) Party Identification and Beyond. London: Wiley, pp. 243–273.
Irish Times (2012) The great struggle was between fear and anger – and fear has won.
Irish Times 2 June. Available at: http://www.irishtimes.com/news/the-great-strugglewas-between-fear-and-anger-and-fear-has-won-1.1063363
(accessed 8 October 2012).
Kiss Z and Hobolt S (2012) Negative campaigning, emotion and political participation. In:
EJECDEM Final Conference, Florence, June 28–30, 2012.
Kiss Z and Hobolt S (2011) The emotional voter: An experimental study of the moderating
effect of emotions on partisan behaviour. In: 6th ECPR General Conference, Reykjavik,
August 25–27, 2011.
Knutsen O (1997) The partisanship and the value-based component of left-right self-placement:
A comparative perspective. International Political Science Review 18(2): 191–225.
Ladd JM and Lenz GS (2008) Reassessing the role of anxiety in vote choice. Political
Psychology 29(2): 275–296.
Garry 253
Ladd JM and Lenz GS (2011) Does anxiety improve voters’ decision making? Political
Psychology 32(2): 347–361.
Lazarus RS (1991) Emotion and Adaptation. Oxford: Oxford University Press.
Lerner J, Gonzalez R, Small D and Fischoff B (2003) Effects of fear and anger on perceived
risk of terrorism: A national field experiment. Psychological Science 14(2): 144–150.
MacKuen M, Wolak J, Keele L and Marcus GE (2010) Civic engagements: Resolute partisanship
or reflective deliberation. American Journal of Political Science 54(2): 440–458.
Marcus GE (2002) The Sentimental Citizen: Emotion in Democratic Politics. University Park,
PA: Pennsylvania State University Press.
Marcus GE, MacKuen M and Neuman WR (2011) Parsimony and complexity: Developing
and testing theories of affective intelligence. Political Psychology 32(2): 323–326.
Marcus GE and MacKuen MB (1993) Anxiety, enthusiasm and the vote: The emotional
underpinnings of learning and involvement during presidential campaigns. American
Political Science Review 87(3): 672–685.
Marcus GE, Neuman WR and MacKuen M (2000) Affective Intelligence and Political
Judgement. Chicago: University of Chicago Press.
Marcus GE, MacKuen M, Wolak J, et al. (2006) The measure and mismeasure of emotion.
In: Redlawsk DP (ed.) Feeling Politics: Emotion in Political Information Processing.
New York: Palgrave Macmillan, pp. 31–47.
McLaren L (2002) Public support for the European Union: Cost/benefit analysis or
perceived cultural threat? Journal of Politics 64(2): 551–566.
McLaren L (2004) Opposition to European integration and fear of loss of national identity:
Debunking a basic assumption regarding hostility to the integration project. European
Journal of Political Research 43(6): 895–912.
Moravcsik A (2006) What can we learn from the collapse of the European Constitutional
Project? Politische Vierteljahresschrift 47(2): 219–241.
Petersen MB (2010) Distinct emotions, distinct domains: Anger, anxiety and perceptions of
intentionality. Journal of Politics 72(2): 357–365.
Reif K and Schmitt H (1980) Nine second order national elections. European Journal of
Political Research 8(1): 3–44.
Richardson B (1991) European party loyalties revisited. American Political Science Review
85: 751–775.
Rhudi JL and Meagher MW (2000) Fear and anxiety: Divergent effects on human pain
thresholds. Pain 84(1): 65–75.
Siune K and 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 and Tonsgaard O (1994) The EU: The Danes said ‘‘No’’ in 1992 and
‘‘Yes’’ in 1993 – How and why? Electoral Studies 13(2): 107–116.
Smith C and Ellsworth P (1985) Patterns of cognitive appraisal in emotion. Journal of
Personality and Social Psychology 48(4): 813–838.
Svensson P (1994) The Danish Yes to Maastricht and Edinburgh: The EC referendum of
May 1993. Scandinavian Political Studies 17(1): 69–82.
Svensson P (2002) Five Danish referendums on the European Community and European
Union: A critical assessment of the Franklin thesis. European Journal of Political
Research 41(6): 733–750.
Valentino NA, Hutchings L, Banks AJ, et al. (2008) Is a worried citizen a good citizen?
Emotions, political information seeking, and learning via the internet. Political
Psychology 29(2): 247–273.
254 European Union Politics 15(2)
Political knowledge and campaign effects in the 2008 Irish
referendum on the Lisbon Treaty
Johan A. Elkink*
, Richard Sinnott
University College Dublin, Belfield, Dublin 4, Ireland
article info
Article history:
Available online 25 February 2015
Keywords:
Referendums
Political knowledge
European integration
Campaigns
Lisbon Treaty
Ireland
abstract
This article makes a distinction between the attitude component of campaigns and the
knowledge component and argues that a campaign that influences knowledge of a proposal
can be quite successful in influencing the vote. On June 12th 2008 Irish voters voted
against ratification of the Lisbon Treaty. Perceptions, and especially misperceptions, of the
Treaty played an important role in the referendum. The campaign focused in particular on
influencing voters’ perceptions of, rather than attitudes towards the Treaty. This article
examines the interaction between knowledge, campaigns, and perceptions of the Treaty, in
the referendum.
© 2015 Elsevier Ltd. All rights reserved.
1. Introduction
The nature of European Union referendums, as distinct
from general elections and from clearly single issue referendums,
provides valuable material with which to further
our understanding of how voters make their decisions on
whether and how to vote. EU referendums are also typically
quite consequential, making it important not only to
understand them from a scientific, but also from a policymaking
point of view.
Voting behaviour in EU referendums is generally
understood either in terms of satisfaction or dissatisfaction
with the government of the day e the so-called secondorder
perspective (Reif and Schmitt, 1980; Franklin et al.,
1994; Franklin, 2002); or in terms of the attitudes of
voters towards European integration (Garry et al., 2005)
and towards European policies (Hix, 2006; Laffan, 2008;
Gamble, 2006); or in terms of the utilitarian benefits one
expects to obtain from European integration (Gabel, 1998;
Ehin, 2001; Van Apeldoorn, 2009); or, finally, in terms of
the effect of knowledge of the issue at hand or of the
European Union in general (Binzer Hobolt, 2005). Previous
studies of the Irish referendums suggest that political
knowledge was crucial in determining the outcome of the
referendums on the Nice and Lisbon treaties (Sinnott, 2001,
2003; Sinnott et al., 2009).
This article provides an in-depth examination of the role
of political knowledge in the 2008 Irish referendum on the
Treaty of Lisbon. It argues that a central feature of the
campaign was the emphasis on perceptions of the contents
of the Treaty, rather than evaluations of these contents.
Although these two components of a referendum campaign
are inseparably linked, they are nevertheless conceptually
distinct.
Referendum campaigns can be distinguished by the
extent to which they emphasise either the knowledge or
the attitude component. On the basis of a statistical analysis
of the results from an opinion survey conducted shortly
after the referendum it is apparent that knowledge played a
crucial role (Sinnott et al., 2009). This research shows that
perceptions of the contents of the Treaty can be separated
into two groups, one corresponding to issues emphasised
by the NO campaign and one to issues emphasised by the
YES campaign. The alignment of what voters perceive to be
the contents of the Treaty with either the NO or the YES
campaign is shown to be an important factor in accounting
* Corresponding author.
E-mail addresses: jos.elkink@ucd.ie (J.A. Elkink), richard.sinnott@ucd.ie
(R. Sinnott).
Contents lists available at ScienceDirect
Electoral Studies
journal homepage: www.elsevier.com/locate/electstud
http://dx.doi.org/10.1016/j.electstud.2015.02.003
0261-3794/© 2015 Elsevier Ltd. All rights reserved.
Electoral Studies 38 (2015) 217e225
for the referendum result, as opposed to the overall level of
knowledge of the Treaty contents or the negative or positive
evaluation of these Treaty contents.
The next section of this article briefly describes the 2008
referendum campaign, thus providing the context in which
the empirical analysis is pursued. Section 3 sets out the
causal mechanisms that we believe connect political
knowledge to voting behaviour. Section 4 introduces the
survey data and the measurement of the key variables.
Section 5 tests the causal claims in a series of regression
analyses. Section 6 summarises the findings and draws some
tentative conclusions. The appendices contain definitions of
the variables and further details on the empirical analysis.
2. The Irish context
The two sides of the campaign in the 2008 referendum
were of very different composition. Whereas the YES
campaign was forced to defend the Treaty in its entirety
and had to have an answer to every possible criticism of the
Treaty, each different actor in the NO campaign could
simply focus on a specific aspect of the Treaty. Quinlan
(2009: 109) succinctly describes the NO campaign:
“As in previous European referendums, the no side was
a diverse coalition[, including] … Sinn Fein [as] the only 
party in the Oireachtas to call for a no vote …, the Peace
and Neutrality Alliance (PANA) … the People’s
Campaign was a broad coalition which included [several
MEPs] … Meanwhile, sustained opposition to the Treaty
on the right was provided by Coir  … But the most vocal
opposition to the Treaty came from the think-tank Libertas.
Founded by Galway businessman Declan Ganley,
Libertas opposed Lisbon because in its view it did not
provide a transparent democratic Europe, weakened
Irish power within the EU and opened the way for Irish
corporate tax rates to be interfered with by Brussels.”
The YES campaign on the other hand consisted of almost
all major political parties (with the Green Party formally
staying neutral), the Irish Alliance for Europe and a number
of business interest groups (Quinlan, 2009: 108e109). In
addition to the protagonists on either side, there was the
role of the Referendum Commission, which was established
by statute to “explain the subject matter to the
public, … promote awareness of the referendum and …
encourage people to vote” (Quinlan, 2009: 109).
While this is at first sight a neutral assignment, in a
referendum in which lack of knowledge of the Treaty
contents is an explicit issue, the task of increasing
“awareness” easily becomes indistinguishable from the
political campaigns.
While the YES campaign was emphasising the increased
efficiency and improved decision-making in the European
Union to be brought about by the Lisbon Treaty, the NO
campaign actively attempted to influence specific perceptions
of the Treaty, often by emphasising items that were
not actually present in the Treaty text, but that are important
to Irish voters. For example, one section of the NO
campaign emphasised the risk that the Lisbon Treaty would
lead to the establishment of a European army with
conscription for Irish citizens (Quinlan, 2009: 111).
3. Knowledge and referendum voting behaviour
Knowledge played a crucial role in determining the
outcome of the Lisbon Treaty referendum. This is clear from
the subjective evaluations of a nationally representative
sample of voters. When NO voters in a post-referendum
poll were asked why they voted NO, 46% of them
mentioned something related to lack of information,
knowledge, or understanding (Sinnott et al., 2009: 13).
Subsequent statistical analyses of the survey data using
appropriate control variables confirms that political
knowledge was a crucial factor in determining the referendum
outcome (Sinnott et al., 2009), which would not
surprise even a casual observer of the campaign.
The idea that knowledge affects voting behaviour is
hardly new (Carpini and Keeter, 1996; see also Bowler, this
issue). In the 2008 Irish referendum, however, the role of
knowledge was a complicated one, involving campaign
effects, perceptions and misperceptions of the contents of
the Treaty, and subjective evaluations of the level of
knowledge of the Treaty. Fig. 1 provides a graphical
depiction of our key causal argument. The difference between
subjective knowledge d the extent to which a voter
feels informed about the issue at hand in the referendum d
and objective knowledge d the actual level of knowledge
of the issue at hand, or of the European Union more
generally d will be a crucial distinction in what follows.
The primary mechanism through which subjective
knowledge can be expected to have an impact on vote
choice (Arrow 4) is through risk aversion. The more uncertain
a respondent feels about the impact of voting for a
referendum proposal, the more likely the voter is to support
the status quo and to vote against the proposal (Suiter
and Reidy, 2013). Binzer Hobolt (2009: 40e57) provides a
formal model elaborating on this idea. In this argument, the
level of objective knowledge is of less importance;
although it can of course be assumed that objectively less
knowledgeable voters are also likely to feel less knowledgeable
(Arrow 3). A feeling of a lack of knowledge was
widespread in the 2008 referendum. Indeed some politicians
exacerbated this feeling by making statements such
as the admission by Taoiseach (prime minister) Brian
Cowen on television that he had not read the Treaty “cover
to cover” or the remark by Irish EU Commissioner Charlie
Fig. 1. Causal model examining the effect of knowledge on the vote (YES or
NO). Objective knowledge is here conceptualised as perceptions that are
correct, hence is part of the perceptions variable in this diagram.
218 J.A. Elkink, R. Sinnott / Electoral Studies 38 (2015) 217e225
McCreevy a week later that any voter who read the Treaty
was “insane” (Quinlan, 2009: 113).
The impact of objective knowledge is likely to be more
indirect, as there is no reason a priori to assume that higher
levels of knowledge lead to stronger support for European
integration. Furthermore, lack of knowledge of politics in
general, of the European Union, or of the Lisbon Treaty does
not necessarily render the voter incapable of casting a
rational choice: “if there are multiple informational pathways
to a competent vote (e.g., if interest-group endorsements,
effective heuristics, etc., lead a voter to cast the
same vote as she would have if better informed), then
voters need not use the same information that political
elites use” (Lupia, 2006: 222; Bowler, this issue).
This reliance on informed sources to make a decision
does depend on the trustworthiness of the source. In other
words, the information can be either correct or incorrect
and therefore such heuristics contain a risk for the voter d
the interests of the sender of information might be in
conflict with those of the receiver (Lupia and McCubbins,
1998). An objectively less knowledgeable voter is thus
relying more on campaign messages from trusted sources
than a more knowledgeable voter, and is more susceptible
to potentially misleading campaign messages about the
referendum issue at hand.
Less knowledgeable voters are therefore more dependent
on high quality information, but are likely to in fact
rely on more easily accessible sources (Bowler and
Donovan, 1994). Hooghe and Marks (2009: 13) make a
similar argument in particular in the context of public
opinion on Europe, which they argue “is particularly susceptible
to construction: i.e. priming (making a consideration
salient), framing (connecting a particular
consideration to a political object) and cueing (instilling a
bias)” (see also, Druckman, 2004; Druckman and Nelson,
2003; Miller and Krosnick, 2000).
The finding that knowledge is important, even when
controlling for well-known effects such as the popularity of
the government of the day or attitudes towards the European
Union in general, has implications for the kind of
campaigns that could be effective in a referendum. If
knowledge matters, campaigns can either focus on changing
the knowledge or perceptions voters have of the issue
at hand or they can concentrate on the attitudes these
voters have towards what they perceive to be the issue. We
are thus making a distinction between the attitude and the
knowledge components of campaigns.1
An obvious criticism of this distinction is that these two
components can never be observed entirely independently.
Any campaign that emphasises how, for example, corporate
taxes are part of the Lisbon Treaty, also implies that this is a
negative thing. In other words, such a campaign influences
perceptions and attitudes simultaneously. It can be argued
that these are still distinct, if generally coinciding, components.
The point becomes clearer when we compare the
Irish referendum on the Lisbon Treaty with the Irish referendum
on abortion in March 2002. In a referendum on
abortion, the issue at hand is clear to voters. Campaigns
focus on influencing voters’ evaluations of whether allowing
abortion is a good or a bad thing, so they attempt to
influence attitudes rather than perceptions. In the referendum
on the Lisbon Treaty, however, the emphasis was
very much on influencing perceptions of what was in the
Treaty rather than the evaluations of these. For example,
the NO campaign would argue that the Treaty could lead to
an increase in Irish corporate tax rates. On both sides of the
campaign, many voters would oppose such an increase, but
it is the perception that this is part of the referendum decision
that the campaign tried to influence, rather than
whether such increase would be a good or bad thing for
Ireland. It was implied that it is a bad thing, but that was
not the contentious issue.
Since ‘perceptions’ and ‘attitudes’ are often confused, it
is crucial that we are specific about what we mean by
these terms. According to the expectancy-value approach
in social psychology, attitudes are “a multiplicative combination
of (a) strength of beliefs that an object has certain
attributes and (b) evaluations of these attributes” (Perloff,
2003: 46; see also Ajzen and Fishbein, 1977). In this
article we will refer to ‘perceptions’ or ‘knowledge’ when
talking about the beliefs that an object, in this case the
Treaty, has certain attributes. In short, whereas ‘perceptions’
refers to what the voters think, rightly or wrongly, is
contained in the Treaty, ‘knowledge’ refers to the extent to
which these perceptions were correct. We refer to ‘attitudes’
when talking about the evaluations of these
perceived attributes.
An example would be the campaign item related to the
possibility of an increase in the Irish corporate tax rates. We
make a clear distinction between the perception that the
Lisbon Treaty does indeed imply future increases in tax
rates and the evaluating attitude the respondent has towards
the idea of such increase.2 There is an extensive
literature in the US context outlining the conditional and
nonlinear nature of issue voting (e.g., Grynaviski and
Corrigan, 2006) and the impact of political knowledge
therein (Campbell et al., 1960; Nie et al., 1979; Palfrey and
Poole, 1987; Krosnick, 1988; Zaller, 1992; Alvarez, 1997;
Basinger and Lavine, 2005; De Vries et al., 2011).
4. Knowledge and the campaign messages
The main data for the following empirical analysis is a
national opinion poll conducted by Millward Brown IMS on
behalf of the Department of Foreign Affairs of the Republic
of Ireland. Data were collected from 2101 respondents aged
18 years and older. All respondents were Irish citizens listed
on the electoral register. The poll was carried out between
24 and 31 July 2008, slightly over a month after the referendum
had taken place. Details and bivariate analyses can
be found in IMS (2008) and a more extensive multivariate
1 Or framing and cueing, respectively, in the language of political
studies.
2 A somewhat complicating factor is that at times these perceptions
might be influenced by attitudes. A study by Lord et al. demonstrates how
subjects interpret neutral or contradicting evidence in such a way that it
confirms their prior beliefs. They adjust their evaluation of the quality,
biasedness, logic, and conclusions all on the basis of their prior attitude
towards the object under study (Lord et al., 1979).
J.A. Elkink, R. Sinnott / Electoral Studies 38 (2015) 217e225 219
analysis in Sinnott et al. (2009). The appendix provides the
precise coding of all variables used.3
Table 1 provides an overview of key Treaty items
emphasised by the two campaigns and translated into a
survey question. Although most key campaign items have
been included in the question, it is not an exhaustive list of
issues raised during the campaign. The respondents were
asked which of the items were part of the Treaty with the
option to select the ‘don’t know’ answer category in each
case. The responses to this question are presented in the
first two columns of the table. These answers can thus be
seen as the perceptions of the Treaty contents. For most
items it is quite clear whether these were or were not part
of the Treaty. The third column lists the correct answers.
The final column lists the items that are indeed in the
Treaty and those that are not. As we argue here, the
campaign focused somewhat less on influencing the attitude
towards particular items, but rather on influencing the
perception that particular items were or were not part of
the Treaty. An evaluation, either positive or negative, was
generally only implied. The final column lists the particular
items that the campaigns suggested were part of the Treaty.
Although the evaluation of the items differs significantly,
both campaigns admitted clearly that the loss of an Irish
commissioner was indeed part of the Treaty.4
We argue that an important part of the effect of objective
knowledge is the susceptibility of respondents with
low levels of knowledge to the (potentially misleading)
campaign messages of the different campaigns. That for a
large proportion of the electorate, campaigns can influence
the vote in the referendum not only by promoting a
particular evaluation of the issue to be voted on, but also
particular perceptions of the issue itself, of what it is one is
voting on. For the statistical evaluation of this proposition,
we measure the alignment of respondents with particular
campaign messages through additive scales. Both the
measures of objective knowledge of the Treaty and the
measures of campaign effects are based on the data
described in Table 1. The objective knowledge scale reflects
the number of correct answers on items 1e5, 7, 8, 10, and
11. The NO campaign scale counts the number of YES answers
on items 2e6 and the YES campaign scale counts the
number of YES answers on items 7e11.5 The two scales on
the campaigns do not summarise the campaigns as such d
they summarise how the campaigns provided different
perceptions of the contents of the Treaty.
Cronbach’s alpha can be used to assess whether these
are indeed reasonable scales to use as measures in subsequent
regression analysis. We obtain for the knowledge
scale a ¼ 0.74; for the NO campaign items a ¼ 0.76 and for
the YES campaign items a ¼ 0.71. Typically, an alpha of
approximately 0.6 is seen as the threshold to accept a scale
as sufficiently additive e the items in the scale are suffi-
ciently correlated to consider them indicators of the same
underlying concept.
We use these additive scales in the regression analysis in
the next section to assess the relative explanatory power of
objective knowledge of the Treaty contents e the truth
value of the perceptions of the Treaty e and of the alignment
of the perceptions of the Treaty with the messages of
the two campaigns. An important caveat is necessary,
however, namely that there is the potential for endogeneity
in this model. While we take the alignment of voters with
particular campaign messages to be an explanatory variable
for their vote choice, it is possible that for some voters
the decision to vote YES or NO was made early in the
campaign and that this has lead them to follow a particular
campaign messages more closely, which lead to them
aligning their perceptions with the campaign messages.
Both in terms of the information they pay attention to and
in line with cognitive dissonance theory, such reverse
causality is a genuine possibility.6
5. Explaining vote choice in the referendum
We are primarily interested in the role of knowledge in
the referendum. As is clear from Fig. 2, the level of objective
knowledge of the contents of the Treaty has a substantial
effect on the individual’s voting behaviour. This figure
Table 1
Perceptions of the contents of the Lisbon Treaty. The question is: “Which
of the following do you think are included in the Lisbon Treaty?”.
% Yes % No Correct Campaign
1. Loss of Irish Commissioner
for 5 out of every 15 years
65 9 Correct Ambiguous
2. Ending of Ireland’s right to
decide its own corporate
tax rate
43 22 Incorrect No campaign
3. The introduction of
conscription to a
European army
33 37 Incorrect No campaign
4. The reduction of Ireland’s
voting strength in the
Council of Ministers
48 18 Correct No campaign
5. The end of Ireland’s
control over its
policy on abortion
34 33 Incorrect No campaign
6. The erosion of Irish
neutrality
42 30 Ambiguous No campaign
7. Improved efficiency of EU
decision-making
56 15 Correct Yes campaign
8. Strengthening Europe’s
role in the world
61 12 Correct Yes campaign
9. Improved protection of
workers’ rights
49 19 Ambiguous Yes campaign
10. Strengthening the role
ofnational parliaments
in EU decision-making
43 20 Correct Yes campaign
11. The Charter of
Fundamental Rights
36 14 Correct Yes campaign
3 The questionnaire was designed by a working group comprising
officials from the Irish Department of Foreign Affairs and researchers
from the survey agency (Millward Brown-IMS). Professor Richard Sinnott
acted as technical advisor to the working group.
4 This despite the fact that even under the already adopted Nice Treaty,
Ireland will lose its Commissioner, albeit later.
5 Note that for the knowledge scale, the ‘don’t know’ category is taken
as someone who lacks knowledge, so it is combined with the incorrect
answer, while for the campaign scales, the ‘don’t know’ category is
treated as missing data.
6 We thank the anonymous reviewer for this excellent point.
220 J.A. Elkink, R. Sinnott / Electoral Studies 38 (2015) 217e225
presents the predicted values of the probability of a NO
vote for a particular configuration of the independent variables
(see the caption). The predictions are based on the
logistic regression reported as Model 3 in Table 2 in the
appendix, which includes a large number of control variables,
the measure of knowledge as the number of correct
perceptions of the contents of the Treaty, and subjective
knowledge, but which excludes the two separate
dimensions discussed above. Whereas a knowledgeable
voter who is dissatisfied with the government has slightly
less than a fifty per cent chance of voting NO, a similar voter
with much less knowledge of the Treaty contents has a
probability of voting NO of about ninety per cent. The effect
is strong for both people who are satisfied with the
government and people who are not. While the secondorder
effects of closeness to the opposition and dissatisfaction
with the government are important factors in
explaining the vote, political knowledge played a more
influential role in this referendum.
As suggested in the theoretical discussion above, the
story is somewhat more complicated than the results in
Fig. 2 suggest. The causal diagram in Fig. 1 functions as a
useful guide for a more nuanced interpretation of the
regression results, which are graphically presented in
Fig. 3.
7 Two separate models were estimated to take account
of the different causal paths to be evaluated. When
estimating a causal effect, it is generally advised to control
for any potential confounding factors, but not to control for
any mediating variables (Pearl, 2000; Morgan and Winship,
2007). As can be easily determined on the basis of the
causal diagram, to estimate the effect of objective knowledge
on the vote, we do not want to control for subjective
knowledge, but to estimate the effect of subjective
knowledge, we do want to control for the level of objective
knowledge. For exactly this reason, the two models were
estimated separately. Although Fig. 3 only reports the coefficients
of interest to the theoretical discussion in this
paper, a large set of control variables were included in the
model. Numerical estimates of all variables are reported in
Table 2 in the appendix. Since all variables have been
standardised, we can directly compare the magnitudes of
the effects of each variable.8
It is clear from these estimates that subjective knowledge
of the Treaty has a significant effect on the vote. Objective
knowledge also has a significant effect on the probability of
voting NO, but once we control for subjective knowledge,
this effect is reduced and is statistically insignificant. This
suggests that the impact of objective knowledge is indeed
02468
0.0 0.2 0.4 0.6 0.8 1.0
Low objective knowledge of the Lisbon treaty
Probability of NO vote
Satisfied
Dissatisfied
Fig. 2. Effect of objective knowledge of the Lisbon Treaty and dissatisfaction
with the government on the NO vote. Estimates are based on Model 3 of Table
2 in the appendix. The grey shaded area is a histogram of the objective
knowledge measure. The lines represent the median voter, which is an
hypothetical lower middle class female, between 35 and 49 years old, with
secondary education, who supports an opposition party, had a median score
on the anti-immigration and Irish neutrality scales, who does not consider her
own economic situation to be bad and who does not identify as “Irish only”.
Fig. 3. Absolute values of logistic regression coefficients and their standard
errors, based on models 5 (solid lines) and 6 (dashed lines) in Table 2. The
models contain the same control variables (not shown) and differ only in
terms of the inclusion of the two subjective knowledge variables. Horizontal
lines reflect the 95% confidence intervals, while the vertical markers reflect
the 90% confidence intervals. The full regression table can be found in
Table 2 in the appendix.
7 See Kastellec and Leoni (2007) for an argument in favour of
presenting regression results in this manner.
8 A logistic regression can be perceived of as a linear relation of the
independent variables with a latent dependent variable, whereby the
probability of a 1 on the dependent variable is a non-linear function of
the latent variable (King, 1998). The reported coefficients reflect in all
cases the effect of an increase of two standard deviations (Gelman, 2007)
in the independent variable on the latent dependent variable.
J.A. Elkink, R. Sinnott / Electoral Studies 38 (2015) 217e225 221
through the level of subjective knowledge (Arrow 3 in the
causal diagram in Fig. 1) d the awareness of the voter of his
or her lack of knowledge d and the two dimensions of the
information from the campaign (Arrow 6).
The alignment of the perceptions of the contents of the
Treaty with that of either side of the campaign has a highly
significant effect on the NO vote, even stronger than that of
objective and subjective knowledge variables.9 The more
the voter’s perceptions of the contents of Treaty are in line
with the messages of the NO campaign, the more likely this
voter is to vote NO; the more the perceptions are in line
with the YES campaign, the more likely this voter is to vote
YES. This suggests campaigns can not only influence attitudes
towards the European Union or towards the Treaty,
they can also influence knowledge or perceptions of the
Treaty. Separating out the two dimensions of perceptions
or knowledge of the Treaty thus clearly improves our understanding
of the voting behaviour in this referendum.
The regression results strongly suggest a more complicated
causal explanation than a straightforward relationship between
knowledge and voting behaviour.
Although we have clear results from the regression
analysis that our separation of the two dimensions has
been fruitful, an important caveat regarding this interpretation
is in order. Whereas the existence of a relationship
between objective and subjective knowledge and the reported
voting behaviour is clearly supported by the statistical
results, the support for the relationship with the
campaign depends on observations about the campaign
that go beyond the individual-level data. Casual observation
of the campaign does suggest the focus on perceptions
instead of attitudes and the regression analysis suggests
that these two dimensions do indeed matter.10 We still
have very limited evidence, however, that it really was the
campaign messages that brought about this effect. We have
limited statistical information about the sources of the
perceptions of the respondents d we infer influence of the
campaign from alignment of the attitudes of the respondent
with the messages of the campaign d and we have
limited empirical evidence of how we attributed particular
items to particular campaigns. Furthermore, we do not
directly measure the attitudes of respondents towards the
issues they do or do not perceive to be part of the Treaty.
For this reason, we are not able to directly measure the
interactive effect between perceptions and attitudes as
suggested in our causal diagram in Fig. 1. Our discussion
makes theoretical sense and our empirical findings are in
line with our expectations, but further research is needed
to confirm the theoretical claims made.
6. Conclusion
The Irish referendum on the Lisbon Treaty provides an
excellent opportunity to study voting behaviour in referendums.
Preliminary analyses of a post-referendum opinion
survey suggest that, among other factors, political knowledge
played a crucial role (IMS, 2008; Sinnott et al., 2009).
This article proposes a clear distinction between perceptions
of the referendum issue d what do voters think is
part of the Treaty d and attitudes towards the referendum
issue d how do voters evaluate the elements they think are
part of the Treaty. Although these two components are very
closely related to each other, they are conceptually distinct.
Acknowledging that these are separate components of the
way knowledge and attitudes interact to influence voting
behaviour, we can conceive of a political campaign in a
referendum that emphasises either one of the two components.
The campaigns in the Irish referendum indeed
appear to focus more on influencing the perceptions of the
contents of the Treaty, than on influencing the attitudes
towards these contents. Whereas in some referendums the
issue at hand might be clearly defined and debate centres
on the positive or negative evaluation of this issue, in the
campaign on the Lisbon Treaty, the main debate was on
what was and what was not part of the Treaty.
Including these scales as separate explanatory variables
in a multivariate regression model that measure the extent
to which a respondent is aligned with a particular
campaign in terms of the perceptions of the contents of the
Treaty leads to the conclusion that these alignments are
strong predictors of the vote in the referendum. The more a
voter’s perceptions of the contents of the Treaty are aligned
with the NO campaign, the more likely the voter is to vote
NO, and vice versa for the YES campaign.
While there are good reasons to expect a direct effect of
a lack of subjective knowledge of the referendum issue at
hand on the vote choice in a referendum (see, e.g., Binzer
Hobolt, 2009), the impact of objective knowledge is indirect.
The indirect effect proceeds through various channels,
not least its impact through subjective knowledge, among
which is the particular susceptibility of uninformed voters
to misleading campaign messages. The analysis demonstrates
this effect in the context of the 2008 Irish referendum
on the Lisbon Treaty.
A rational voter can decide not to invest extensively in
acquiring the knowledge to make a fully informed decision
in a referendum as complex as that on the Treaty of Lisbon
and instead to rely on cues from members of the political
elite. The risk in doing so, however, is that these elites will
have particular political interests that might run counter to
those of the voter (Lupia and McCubbins, 1998).
This article makes two contributions to the resolution of
these issues. The first contribution lies in providing a more
nuanced account of the interaction between knowledge,
attitudes, campaigns and voting behaviour in referendums.
The second contribution focuses on the policymaker’s point
of view and argues that governments need to develop ways
and means of ensuring that electorates have access to real
debate and deliberation in a way that enhances knowledge
of the often complex issues that are set before them in the
form of the referendum.
9 These inferences are, of course, based on largely the same information,
namely that recorded in the question presented extensively in Table
1. That this does not lead to unacceptable levels of multicollinearity is
evidenced by the fact that the coefficients on both the objective knowledge
of the Treaty variable and on the campaign variables are statistically
significant. Furthermore, the significant decrease in the Akaike Information
Criterium, by more than twice the number of variables added
(Gelman and Hill, 2007, 525), suggests that the inclusion of the campaign
variables is a significant improvement of the model. 10 Furthermore, a latent trait analysis not reported in this paper provides
supporting evidence that the differentiation between the two dimensions
in the perceptions of the Treaty is reasonable.
222 J.A. Elkink, R. Sinnott / Electoral Studies 38 (2015) 217e225
Acknowledgements
Previously presented at the New Directions in Referendums:
Politics and Campaigns conference, Dublin, May 9,
2013; 1st Annual General Conference of the European Political
Science Association, Dublin, June 16e18, 2011; and
the ECPR Joint Sessions of Workshops, Lisbon, April 2009.
The authors are grateful to Elisabeth Gidengil, Paul Kellstedt,
Theresa Reidy, Jane Suiter and other workshop participants,
as well as the anonymous reviewer, for their
comments.
Appendix
Own economic situation bad
“What about your own economic situation these days?
Would you say it is … ? ” (very good/fairly good/fairly bad/
very bad/don’t know). Values “fairly bad” and “very bad”
are coded as 1, others as 0.
Dissatisfied with government
“Overall are you generally satisfied or dissatisfied with
the way the government is running the country? ” (very
satisfied/quite satisfied/quite dissatisfied/very dissatisfied/
don’t know). Values “very dissatisfied” and “quite dissatisfied”
are coded as 1, others as 0.
Feels close to opposition
“Do you feel close to any of the political parties? ” and if
no, “Do you feel yourself a little closer to one of the political
parties than the others? ” (FF/FG/Labour/Green/Sinn Fein/
PDs/Other/Not close to any). Values “FF”, “PDs” and “Green”
are coded as 1, others as 0.
Identifies as Irish only
“In the near future, do you see yourself as … ? ” (Irish
only/Irish and European/European and Irish/European only/
don’t know). Value “Irish only” is coded as 1, the rest as 0.
Anti-immigration attitude
“Using the card provided please indicate whether
Ireland is made a worse or a better place to live in by people
coming to live here from other countries? ” (1 worse place
to live … 10 better place to live). The variable is entered in
reverse order, 1 ¼ 10, 2 ¼ 9, etc.
Pro-neutrality attitude
“I have a number of statements here that people
sometimes make. I would like you to indicate on this scale
which of each pair of opposing statements comes closest to
your view. A score of one would indicate that you agree
Table 2
Logistic regression explaining the NO vote. Standard errors in parentheses with statistically significant (a ¼ .05) coefficients in bold print. Coefficients are
standardised as suggested by Gelman (2007). The coefficients on age are relative to the 65þ group; those on class relative to the upper middle class; and
those on education relative to those with only primary education.
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6
Constant 0.02 (0.31) 0.08 (0.35) 0.08 (0.36) 0.11 (0.36) 0.05 (0.37) 0.02 (0.37)
Female 0.17 (0.13) 0.07 (0.14) 0.04 (0.15) 0.04 (0.15) 0.00 (0.16) 0.00 (0.15)
Age 18e24 0.37 (0.28) 0.51 (0.31) 0.40 (0.32) 0.42 (0.32) 0.35 (0.33) 0.33 (0.33)
Age 25e34 0.64 (0.23) 0.67 (0.26) 0.63 (0.27) 0.63 (0.27) 0.54 (0.28) 0.54 (0.28)
Age 35e49 0.24 (0.21) 0.37 (0.24) 0.37 (0.25) 0.37 (0.25) 0.33 (0.25) 0.34 (0.25)
Age 50e64 0.23 (0.21) 0.30 (0.24) 0.32 (0.24) 0.30 (0.25) 0.25 (0.25) 0.28 (0.25)
Lower middle class 0.24 (0.20) 0.15 (0.22) 0.11 (0.23) 0.09 (0.23) 0.10 (0.24) 0.12 (0.24)
Skilled worker 0.78 (0.22) 0.62 (0.25) 0.48 (0.26) 0.43 (0.26) 0.46 (0.27) 0.51 (0.27)
Unskilled worker 0.85 (0.24) 0.58 (0.27) 0.47 (0.28) 0.39 (0.28) 0.44 (0.29) 0.53 (0.28)
Large farmer 0.22 (0.30) 0.48 (0.35) 0.53 (0.38) 0.55 (0.38) 0.48 (0.39) 0.49 (0.39)
Small farmer 0.75 (0.50) 0.37 (0.57) 0.55 (0.60) 0.48 (0.60) 0.44 (0.62) 0.52 (0.62)
Secondary education 0.14 (0.25) 0.23 (0.29) 0.32 (0.30) 0.37 (0.30) 0.41 (0.31) 0.34 (0.31)
Third level education 0.42 (0.28) 0.05 (0.33) 0.16 (0.34) 0.28 (0.34) 0.31 (0.35) 0.17 (0.34)
Own economic situation bad 0.76 (0.15) 0.43 (0.18) 0.31 (0.18) 0.29 (0.18) 0.30 (0.19) 0.33 (0.19)
Dissatisfied with government 0.82 (0.13) 0.70 (0.15) 0.76 (0.16) 0.77 (0.16) 0.75 (0.16) 0.74 (0.16)
Feels close to opposition 0.73 (0.12) 0.86 (0.14) 0.92 (0.15) 0.91 (0.15) 0.94 (0.15) 0.96 (0.15)
Identifies as Irish only 0.49 (0.15) 0.34 (0.16) 0.35 (0.16) 0.44 (0.17) 0.41 (0.16)
Pro-neutrality attitude 0.78 (0.14) 0.76 (0.15) 0.80 (0.15) 0.72 (0.15) 0.68 (0.15)
Anti-immigration attitude 0.56 (0.17) 0.39 (0.17) 0.37 (0.17) 0.31 (0.18) 0.33 (0.18)
EU membership a good thing ¡1.27 (0.20) ¡1.11 (0.20) ¡1.09 (0.20) ¡1.00 (0.20) ¡1.03 (0.20)
Too many issues decided by EU 0.64 (0.15) 0.63 (0.15) 0.62 (0.15) 0.50 (0.16) 0.52 (0.16)
EU means low wages 0.69 (0.19) 0.62 (0.19) 0.62 (0.19) 0.54 (0.19) 0.54 (0.19)
Low objective knowledge of EU 0.27 (0.18) 0.17 (0.19) 0.18 (0.19) 0.30 (0.19)
Low objective knowledge of treaty 1.35 (0.19) 1.25 (0.19) 0.52 (0.31) 0.70 (0.30)
Low subjective knowledge of EU 0.06 (0.20) 0.05 (0.20)
Low subjective knowledge of treaty 0.48 (0.20) 0.60 (0.21)
Positive on NO campaign items 1.18 (0.18) 1.12 (0.18)
Positive on YES campaign items ¡0.87 (0.28) ¡0.82 (0.28)
Number of observations 1301 1254 1254 1246 1246 1254
Akaike information criterion 1557 1288 1223 1215 1172 1184
J.A. Elkink, R. Sinnott / Electoral Studies 38 (2015) 217e225 223
fully with the statement on the left. A score of nine would
indicate that you agree fully with the statement on the
right. Of course your view could be somewhere in between.
Also of course there may be issues that you have no
particular view on. If so, please just say this and we will
move on to the next item.” (1 “Ireland should do everything
it can to strengthen its neutrality even if this means being
less involved in European Union cooperation on foreign
and defence policy” … 9 “Ireland should be willing to
accept limitations on its neutrality so that it can be more
fully involved in European Union cooperation on foreign
and defence policy”). Values are in reverse order, 1 ¼ 9,
2 ¼ 8, etc.
EU membership a good thing
“Generally speaking, do you think that Ireland’s membership
of the European Union (European Community) is …
? ” (“A good thing”/“A bad thing”/“Neither good nor bad”/
“Don’t know”). The first category is coded as 1, the rest as 0.
Too many issues decided by EU
“There has been a lot of discussion recently about the
European Union. Some people say that too many issues are
decided on by the European Union, others say that more
issues should be decided on by the European Union. Which
of the following statement comes closest to your view? ”
(“Too many issues are decided by the European Union”/
“The number of issues decided on by the European Union at
present is about right”/“More issues should be decided by
the European Union”/“I have not really thought about it”/“It
depends on the issue”/“Don’t know”). Those that selected
the first option were coded as 1, others as 0.
EU means low wages
“What does the European Union mean to you personally?
” (many answer categories, among which “Lower
wage rates”). Those that included “Lower wage rates” in
their selection were coded as 1, others as 0.
Low objective knowledge of Treaty
Number of items deemed to be correct in Table 1 that
are perceived to be in the Treaty; ‘don’t know’ taken as an
incorrect answer.
Low objective knowledge of EU
“For each of the following statements about the European
Union could you please tell me whether you think it is
true or false? ” (“The EU currently consists of fifteen
Member States”/“Switzerland is a member of the European
Union”/“Every six months, a different Member State becomes
the President of the Council of the European Union”/
“The members of the European Parliament are directly
elected by the citizens of the EU”). A four point score is
generated, from 1 (4 correct answers) to 4 (1 correct
answer). All respondents had at least one correct answer.
Low subjective knowledge of Treaty
“By the date of the referendum (12 June), how good was
your understanding of the issues involved? Please use this
card to choose the phrase that applies best to you.” (“I had a
good understanding of what the Treaty was all about”/“I
understood some of the issues but not all that was
involved”/“I was only vaguely aware of the issues
involved”/“I did not know what the Treaty was about at
all”/don’t know). The resulting variable is a four point scale
from good understanding to not knowing what the Treaty
was about at all.
Low subjective knowledge of EU
“And how about the European Union in general? Using
this scale, how much do you feel you know about the European
Union, its policies, its institutions? ” (1 nothing at all
… 10 know a great deal).
Positive on NO campaign items
Number of items associated with the NO campaign in
Table 1 that are perceived to be in the Treaty; ‘don’t know’
taken as missing data.
Positive on YES campaign items
Number of items associated with the YES campaign in
Table 1 that are perceived to be in the Treaty; ‘don’t know’
taken as missing data.
Abstained
“On the 12 June last, a referendum was held on the
Treaty of Lisbon. As you may remember, many people did
not vote in that referendum. How about you? Did you vote
in the referendum on the Treaty of Lisbon that was held in
June? ” (“Yes e voted”/“Did not vote”). In all analyses, this is
used as a filter variable, and only those who voted are taken
into account.
Voted NO
“How did you vote in that referendum e in favour or
against the Lisbon Treaty? ” (“Voted e In favour”/“Voted e
Against”). This is the dependent variable in the regression
analyses.
References
Ajzen, I., Fishbein, M., 1977. Attitude-behavior relations: a theoretical
analysis and review of empirical research. Psychol. Bull. 84 (5),
888e918.
Alvarez, R.M., 1997. Information and Elections. University of Michigan
Press.
Basinger, S.J., Lavine, H., 2005. Ambivalence, information, and electoral
choice. Am. Polit. Sci. Rev. 99, 169e184.
Binzer Hobolt, S., 2005. When Europe matters: the impact of political
information on voting behaviour in EU referendums. J. Elections
Public Opin. Parties 15 (1), 85e109.
Binzer Hobolt, S., 2009. Europe in Question: Referendums on European
Integration. Oxford University Press, New York.
224 J.A. Elkink, R. Sinnott / Electoral Studies 38 (2015) 217e225
Bowler, S., Donovan, T., 1994. Information and opinion change on ballot
propositions. Polit. Behav. 16 (4), 411e435.
Campbell, A., Converse, P.E., Miller, W.E., Stokes, D., 1960. The American
Voter. University of Chicago Press, Chicago.
De Vries, C.E., Van der Brug, W., Van Egmond, M.H., Van der Eijk, C., 2011.
Individual and contextual variation in EU issue voting: the role of
political information. Elect. Stud. 30 (1), 16e28.
Delli Carpini, M.X., Keeter, S., 1996. What Americans Know about Politics
and Why it Matters. Yale University Press.
Druckman, J.N., 2004. Political preference formation: competition,
deliberation, and the (ir)relevance of framing effects. Am. Polit. Sci.
Rev. 98 (4), 671e686.
Druckman, J.N., Nelson, K.R., 2003. Framing and deliberation: how citizens’
conversations limit elite influence. Am. J. Polit. Sci. 47 (4),
729e745.
Ehin, P., 2001. Determinants of public support for EU membership: data
from the Baltic countries. Eur. J. Polit. Res. 40, 31e56.
Franklin, M., 2002. Learning from the Danish case: a comment on Palle
Svensson’s critique of the Franklin thesis. Eur. J. Polit. Res. 41,
751e757.
Franklin, M., Marsh, M., Wlezien, C., 1994. Attitudes towards Europe and
referendum votes: a response to Siune and Svensson. Elect. Stud. 13,
117e121.
Gabel, M.J., 1998. Economic integration and mass politics: market liberalisation
and public attitudes in the European Union. Am. J. Polit. Sci.
42 (3), 936e953.
Gamble, A., 2006. The European disunion. Br. J. Polit. Int. Relat. 8 (1), 34e49.
Garry, J., Marsh, M., Sinnott, R., 2005. ‘Second-order’ versus ‘issue-voting’
effects in EU referendums. Eur. Union Polit. 6 (2), 201e221.
Gelman, A., 2007. Scaling regression inputs by dividing by two standard
deviations. Stat. Med. 27 (15), 2865e2873.
Gelman, A., Hill, J., 2007. Data Analysis Using Regression and Multilevel/
Hierarchical Models. Analytical Methods for Social Research. Cambridge
University Press, Cambridge.
Grynaviski, J.D., Corrigan, B.E., 2006. Specification issues in proximity
models of candidate evaluation (with issue importance). Polit. Anal.
14, 393e420.
Hix, S., 2006. Why the EU Needs Left-right Politics. In: Politics: the Right or
the Wrong Sort of Medicine for the EU?. Notre Europe Policy Paper 19.
Hooghe, L., Marks, G., 2009. A postfunctionalist theory of European
integration: from permissive consensus to constraining dissensus. Br.
J. Polit. Sci. 39 (1), 1e23.
IMS, 2008. Post Lisbon Treaty Referendum Research Findings (Technical
report Millward Brown-IMS).
Kastellec, J.P., Leoni, E.L., 2007. Using graphs instead of tables in political
science. Perspect. Polit. 5 (4), 755e771.
King, G., 1998. Unifying Political Methodology. The Likelihood Theory of
Statistical Inference. University of Michigan Press.
Krosnick, J.A., 1988. Attitude importance and attitude change. J. Exp. Soc.
Psychol. 24, 240e255.
Laffan, B., 2008. Reflection: three readings of the failure to ratify the
constitutional treaty. In: Brown, T. (Ed.), Lisbon: What the Reform
Treaty Means. Institute of International and European Affairs, Dublin,
pp. 60e70.
Lord, C.G., Ross, L., Lepper, M.R., 1979. Biased assimilation and attitude
polarization: the effects of prior theories on subsequently considered
evidence. J. Personal. Soc. Psychol. 37 (11), 2098e2109.
Lupia, A., 2006. How elitism undermines the study of voter competence.
Crit. Rev. 18 (1), 217e232.
Lupia, A., McCubbins, M.D., 1998. The Democratic Dilemma: Can Citizens
Learn What They Need to Know? Cambridge University Press, New
York.
Miller, J.M., Krosnick, J.A., 2000. News media impact on the ingredients of
presidential evaluations: politically knowledgeable citizens are
guided by a trusted source. Am. J. Polit. Sci. 44 (2), 301e315.
Morgan, S.L., Winship, C., 2007. Counterfactuals and Causal Inference.
Methods and Principles for Social Research. Cambridge University
Press, New York.
Nie, N.H., Verba, S., Petrocik, J.R., 1979. The Changing American Voter.
Harvard University Press Cambridge.
Palfrey, T.R., Poole, K.T., 1987. The relationship between information,
ideology, and voting behavior. Am. J. Polit. Sci. 31 (3), 511e530.
Pearl, J., 2000. Causality: Models, Reasoning, and Inference. Cambridge
University Press, Cambridge.
Perloff, R.M., 2003. The Dynamics of Persuasion. Communication and
Attitudes in the 21st Century, second ed. Lawrence Erlbaum Associates,
Mahwah, NJ.
Quinlan, S., 2009. The Lisbon treaty referendum 2008. Ir. Polit. Stud. 24
(1), 107e121.
Reif, K., Schmitt, H., 1980. Nine second order national elections: a conceptual
framework for the analysis of European election results. Eur. J.
Polit. Res. 8 (1), 3e44.
Sinnott, R., 2001. Attitudes and Behaviour of the Irish Electorate in the
Referendum on the Treaty of Nice (Technical report, European Commission
Representation in Ireland).
Sinnott, R., 2003. Attitudes and Behaviour of the Irish Electorate in the
Second Referendum on the Treaty of Nice (Technical report, European
Commission Representation in Ireland).
Sinnott, R., Elkink, J.A., O’Rourke, K., McBride, J., 2009. Attitudes and
Behaviour in the Referendum on the Treaty of Lisbon (Technical
report, Department of Foreign Affairs).
Suiter, J., Reidy, T., 2013. It’s the campaign learning stupid, an examination
of a volatile Irish referendum. Parliam. Aff. (published online).
Van Apeldoorn, B., 2009. The contradictions of ’embedded neoliberalism’
and Europe’s multi-level legitimacy crisis: the European project and
its limits. In: Van Apeldoorn, B., Drahokoupil, J., Horn, L. (Eds.), Contradictions
and Limits of Neoliberal European Governance. From
Lisbon to Lisbon. Palgrave MacMillan, pp. 21e43.
Zaller, J., 1992. The Nature and Origins of Mass Opinion. Cambridge University
Press, Cambridge.
J.A. Elkink, R. Sinnott / Electoral Studies 38 (2015) 217e225 225
Political knowledge and campaign effects in the 2008 Irish
referendum on the Lisbon Treaty
Johan A. Elkink*
, Richard Sinnott
University College Dublin, Belfield, Dublin 4, Ireland
article info
Article history:
Available online 25 February 2015
Keywords:
Referendums
Political knowledge
European integration
Campaigns
Lisbon Treaty
Ireland
abstract
This article makes a distinction between the attitude component of campaigns and the
knowledge component and argues that a campaign that influences knowledge of a proposal
can be quite successful in influencing the vote. On June 12th 2008 Irish voters voted
against ratification of the Lisbon Treaty. Perceptions, and especially misperceptions, of the
Treaty played an important role in the referendum. The campaign focused in particular on
influencing voters’ perceptions of, rather than attitudes towards the Treaty. This article
examines the interaction between knowledge, campaigns, and perceptions of the Treaty, in
the referendum.
© 2015 Elsevier Ltd. All rights reserved.
1. Introduction
The nature of European Union referendums, as distinct
from general elections and from clearly single issue referendums,
provides valuable material with which to further
our understanding of how voters make their decisions on
whether and how to vote. EU referendums are also typically
quite consequential, making it important not only to
understand them from a scientific, but also from a policymaking
point of view.
Voting behaviour in EU referendums is generally
understood either in terms of satisfaction or dissatisfaction
with the government of the day e the so-called secondorder
perspective (Reif and Schmitt, 1980; Franklin et al.,
1994; Franklin, 2002); or in terms of the attitudes of
voters towards European integration (Garry et al., 2005)
and towards European policies (Hix, 2006; Laffan, 2008;
Gamble, 2006); or in terms of the utilitarian benefits one
expects to obtain from European integration (Gabel, 1998;
Ehin, 2001; Van Apeldoorn, 2009); or, finally, in terms of
the effect of knowledge of the issue at hand or of the
European Union in general (Binzer Hobolt, 2005). Previous
studies of the Irish referendums suggest that political
knowledge was crucial in determining the outcome of the
referendums on the Nice and Lisbon treaties (Sinnott, 2001,
2003; Sinnott et al., 2009).
This article provides an in-depth examination of the role
of political knowledge in the 2008 Irish referendum on the
Treaty of Lisbon. It argues that a central feature of the
campaign was the emphasis on perceptions of the contents
of the Treaty, rather than evaluations of these contents.
Although these two components of a referendum campaign
are inseparably linked, they are nevertheless conceptually
distinct.
Referendum campaigns can be distinguished by the
extent to which they emphasise either the knowledge or
the attitude component. On the basis of a statistical analysis
of the results from an opinion survey conducted shortly
after the referendum it is apparent that knowledge played a
crucial role (Sinnott et al., 2009). This research shows that
perceptions of the contents of the Treaty can be separated
into two groups, one corresponding to issues emphasised
by the NO campaign and one to issues emphasised by the
YES campaign. The alignment of what voters perceive to be
the contents of the Treaty with either the NO or the YES
campaign is shown to be an important factor in accounting
* Corresponding author.
E-mail addresses: jos.elkink@ucd.ie (J.A. Elkink), richard.sinnott@ucd.ie
(R. Sinnott).
Contents lists available at ScienceDirect
Electoral Studies
journal homepage: www.elsevier.com/locate/electstud
http://dx.doi.org/10.1016/j.electstud.2015.02.003
0261-3794/© 2015 Elsevier Ltd. All rights reserved.
Electoral Studies 38 (2015) 217e225
for the referendum result, as opposed to the overall level of
knowledge of the Treaty contents or the negative or positive
evaluation of these Treaty contents.
The next section of this article briefly describes the 2008
referendum campaign, thus providing the context in which
the empirical analysis is pursued. Section 3 sets out the
causal mechanisms that we believe connect political
knowledge to voting behaviour. Section 4 introduces the
survey data and the measurement of the key variables.
Section 5 tests the causal claims in a series of regression
analyses. Section 6 summarises the findings and draws some
tentative conclusions. The appendices contain definitions of
the variables and further details on the empirical analysis.
2. The Irish context
The two sides of the campaign in the 2008 referendum
were of very different composition. Whereas the YES
campaign was forced to defend the Treaty in its entirety
and had to have an answer to every possible criticism of the
Treaty, each different actor in the NO campaign could
simply focus on a specific aspect of the Treaty. Quinlan
(2009: 109) succinctly describes the NO campaign:
“As in previous European referendums, the no side was
a diverse coalition[, including] … Sinn Fein [as] the only 
party in the Oireachtas to call for a no vote …, the Peace
and Neutrality Alliance (PANA) … the People’s
Campaign was a broad coalition which included [several
MEPs] … Meanwhile, sustained opposition to the Treaty
on the right was provided by Coir  … But the most vocal
opposition to the Treaty came from the think-tank Libertas.
Founded by Galway businessman Declan Ganley,
Libertas opposed Lisbon because in its view it did not
provide a transparent democratic Europe, weakened
Irish power within the EU and opened the way for Irish
corporate tax rates to be interfered with by Brussels.”
The YES campaign on the other hand consisted of almost
all major political parties (with the Green Party formally
staying neutral), the Irish Alliance for Europe and a number
of business interest groups (Quinlan, 2009: 108e109). In
addition to the protagonists on either side, there was the
role of the Referendum Commission, which was established
by statute to “explain the subject matter to the
public, … promote awareness of the referendum and …
encourage people to vote” (Quinlan, 2009: 109).
While this is at first sight a neutral assignment, in a
referendum in which lack of knowledge of the Treaty
contents is an explicit issue, the task of increasing
“awareness” easily becomes indistinguishable from the
political campaigns.
While the YES campaign was emphasising the increased
efficiency and improved decision-making in the European
Union to be brought about by the Lisbon Treaty, the NO
campaign actively attempted to influence specific perceptions
of the Treaty, often by emphasising items that were
not actually present in the Treaty text, but that are important
to Irish voters. For example, one section of the NO
campaign emphasised the risk that the Lisbon Treaty would
lead to the establishment of a European army with
conscription for Irish citizens (Quinlan, 2009: 111).
3. Knowledge and referendum voting behaviour
Knowledge played a crucial role in determining the
outcome of the Lisbon Treaty referendum. This is clear from
the subjective evaluations of a nationally representative
sample of voters. When NO voters in a post-referendum
poll were asked why they voted NO, 46% of them
mentioned something related to lack of information,
knowledge, or understanding (Sinnott et al., 2009: 13).
Subsequent statistical analyses of the survey data using
appropriate control variables confirms that political
knowledge was a crucial factor in determining the referendum
outcome (Sinnott et al., 2009), which would not
surprise even a casual observer of the campaign.
The idea that knowledge affects voting behaviour is
hardly new (Carpini and Keeter, 1996; see also Bowler, this
issue). In the 2008 Irish referendum, however, the role of
knowledge was a complicated one, involving campaign
effects, perceptions and misperceptions of the contents of
the Treaty, and subjective evaluations of the level of
knowledge of the Treaty. Fig. 1 provides a graphical
depiction of our key causal argument. The difference between
subjective knowledge d the extent to which a voter
feels informed about the issue at hand in the referendum d
and objective knowledge d the actual level of knowledge
of the issue at hand, or of the European Union more
generally d will be a crucial distinction in what follows.
The primary mechanism through which subjective
knowledge can be expected to have an impact on vote
choice (Arrow 4) is through risk aversion. The more uncertain
a respondent feels about the impact of voting for a
referendum proposal, the more likely the voter is to support
the status quo and to vote against the proposal (Suiter
and Reidy, 2013). Binzer Hobolt (2009: 40e57) provides a
formal model elaborating on this idea. In this argument, the
level of objective knowledge is of less importance;
although it can of course be assumed that objectively less
knowledgeable voters are also likely to feel less knowledgeable
(Arrow 3). A feeling of a lack of knowledge was
widespread in the 2008 referendum. Indeed some politicians
exacerbated this feeling by making statements such
as the admission by Taoiseach (prime minister) Brian
Cowen on television that he had not read the Treaty “cover
to cover” or the remark by Irish EU Commissioner Charlie
Fig. 1. Causal model examining the effect of knowledge on the vote (YES or
NO). Objective knowledge is here conceptualised as perceptions that are
correct, hence is part of the perceptions variable in this diagram.
218 J.A. Elkink, R. Sinnott / Electoral Studies 38 (2015) 217e225
McCreevy a week later that any voter who read the Treaty
was “insane” (Quinlan, 2009: 113).
The impact of objective knowledge is likely to be more
indirect, as there is no reason a priori to assume that higher
levels of knowledge lead to stronger support for European
integration. Furthermore, lack of knowledge of politics in
general, of the European Union, or of the Lisbon Treaty does
not necessarily render the voter incapable of casting a
rational choice: “if there are multiple informational pathways
to a competent vote (e.g., if interest-group endorsements,
effective heuristics, etc., lead a voter to cast the
same vote as she would have if better informed), then
voters need not use the same information that political
elites use” (Lupia, 2006: 222; Bowler, this issue).
This reliance on informed sources to make a decision
does depend on the trustworthiness of the source. In other
words, the information can be either correct or incorrect
and therefore such heuristics contain a risk for the voter d
the interests of the sender of information might be in
conflict with those of the receiver (Lupia and McCubbins,
1998). An objectively less knowledgeable voter is thus
relying more on campaign messages from trusted sources
than a more knowledgeable voter, and is more susceptible
to potentially misleading campaign messages about the
referendum issue at hand.
Less knowledgeable voters are therefore more dependent
on high quality information, but are likely to in fact
rely on more easily accessible sources (Bowler and
Donovan, 1994). Hooghe and Marks (2009: 13) make a
similar argument in particular in the context of public
opinion on Europe, which they argue “is particularly susceptible
to construction: i.e. priming (making a consideration
salient), framing (connecting a particular
consideration to a political object) and cueing (instilling a
bias)” (see also, Druckman, 2004; Druckman and Nelson,
2003; Miller and Krosnick, 2000).
The finding that knowledge is important, even when
controlling for well-known effects such as the popularity of
the government of the day or attitudes towards the European
Union in general, has implications for the kind of
campaigns that could be effective in a referendum. If
knowledge matters, campaigns can either focus on changing
the knowledge or perceptions voters have of the issue
at hand or they can concentrate on the attitudes these
voters have towards what they perceive to be the issue. We
are thus making a distinction between the attitude and the
knowledge components of campaigns.1
An obvious criticism of this distinction is that these two
components can never be observed entirely independently.
Any campaign that emphasises how, for example, corporate
taxes are part of the Lisbon Treaty, also implies that this is a
negative thing. In other words, such a campaign influences
perceptions and attitudes simultaneously. It can be argued
that these are still distinct, if generally coinciding, components.
The point becomes clearer when we compare the
Irish referendum on the Lisbon Treaty with the Irish referendum
on abortion in March 2002. In a referendum on
abortion, the issue at hand is clear to voters. Campaigns
focus on influencing voters’ evaluations of whether allowing
abortion is a good or a bad thing, so they attempt to
influence attitudes rather than perceptions. In the referendum
on the Lisbon Treaty, however, the emphasis was
very much on influencing perceptions of what was in the
Treaty rather than the evaluations of these. For example,
the NO campaign would argue that the Treaty could lead to
an increase in Irish corporate tax rates. On both sides of the
campaign, many voters would oppose such an increase, but
it is the perception that this is part of the referendum decision
that the campaign tried to influence, rather than
whether such increase would be a good or bad thing for
Ireland. It was implied that it is a bad thing, but that was
not the contentious issue.
Since ‘perceptions’ and ‘attitudes’ are often confused, it
is crucial that we are specific about what we mean by
these terms. According to the expectancy-value approach
in social psychology, attitudes are “a multiplicative combination
of (a) strength of beliefs that an object has certain
attributes and (b) evaluations of these attributes” (Perloff,
2003: 46; see also Ajzen and Fishbein, 1977). In this
article we will refer to ‘perceptions’ or ‘knowledge’ when
talking about the beliefs that an object, in this case the
Treaty, has certain attributes. In short, whereas ‘perceptions’
refers to what the voters think, rightly or wrongly, is
contained in the Treaty, ‘knowledge’ refers to the extent to
which these perceptions were correct. We refer to ‘attitudes’
when talking about the evaluations of these
perceived attributes.
An example would be the campaign item related to the
possibility of an increase in the Irish corporate tax rates. We
make a clear distinction between the perception that the
Lisbon Treaty does indeed imply future increases in tax
rates and the evaluating attitude the respondent has towards
the idea of such increase.2 There is an extensive
literature in the US context outlining the conditional and
nonlinear nature of issue voting (e.g., Grynaviski and
Corrigan, 2006) and the impact of political knowledge
therein (Campbell et al., 1960; Nie et al., 1979; Palfrey and
Poole, 1987; Krosnick, 1988; Zaller, 1992; Alvarez, 1997;
Basinger and Lavine, 2005; De Vries et al., 2011).
4. Knowledge and the campaign messages
The main data for the following empirical analysis is a
national opinion poll conducted by Millward Brown IMS on
behalf of the Department of Foreign Affairs of the Republic
of Ireland. Data were collected from 2101 respondents aged
18 years and older. All respondents were Irish citizens listed
on the electoral register. The poll was carried out between
24 and 31 July 2008, slightly over a month after the referendum
had taken place. Details and bivariate analyses can
be found in IMS (2008) and a more extensive multivariate
1 Or framing and cueing, respectively, in the language of political
studies.
2 A somewhat complicating factor is that at times these perceptions
might be influenced by attitudes. A study by Lord et al. demonstrates how
subjects interpret neutral or contradicting evidence in such a way that it
confirms their prior beliefs. They adjust their evaluation of the quality,
biasedness, logic, and conclusions all on the basis of their prior attitude
towards the object under study (Lord et al., 1979).
J.A. Elkink, R. Sinnott / Electoral Studies 38 (2015) 217e225 219
analysis in Sinnott et al. (2009). The appendix provides the
precise coding of all variables used.3
Table 1 provides an overview of key Treaty items
emphasised by the two campaigns and translated into a
survey question. Although most key campaign items have
been included in the question, it is not an exhaustive list of
issues raised during the campaign. The respondents were
asked which of the items were part of the Treaty with the
option to select the ‘don’t know’ answer category in each
case. The responses to this question are presented in the
first two columns of the table. These answers can thus be
seen as the perceptions of the Treaty contents. For most
items it is quite clear whether these were or were not part
of the Treaty. The third column lists the correct answers.
The final column lists the items that are indeed in the
Treaty and those that are not. As we argue here, the
campaign focused somewhat less on influencing the attitude
towards particular items, but rather on influencing the
perception that particular items were or were not part of
the Treaty. An evaluation, either positive or negative, was
generally only implied. The final column lists the particular
items that the campaigns suggested were part of the Treaty.
Although the evaluation of the items differs significantly,
both campaigns admitted clearly that the loss of an Irish
commissioner was indeed part of the Treaty.4
We argue that an important part of the effect of objective
knowledge is the susceptibility of respondents with
low levels of knowledge to the (potentially misleading)
campaign messages of the different campaigns. That for a
large proportion of the electorate, campaigns can influence
the vote in the referendum not only by promoting a
particular evaluation of the issue to be voted on, but also
particular perceptions of the issue itself, of what it is one is
voting on. For the statistical evaluation of this proposition,
we measure the alignment of respondents with particular
campaign messages through additive scales. Both the
measures of objective knowledge of the Treaty and the
measures of campaign effects are based on the data
described in Table 1. The objective knowledge scale reflects
the number of correct answers on items 1e5, 7, 8, 10, and
11. The NO campaign scale counts the number of YES answers
on items 2e6 and the YES campaign scale counts the
number of YES answers on items 7e11.5 The two scales on
the campaigns do not summarise the campaigns as such d
they summarise how the campaigns provided different
perceptions of the contents of the Treaty.
Cronbach’s alpha can be used to assess whether these
are indeed reasonable scales to use as measures in subsequent
regression analysis. We obtain for the knowledge
scale a ¼ 0.74; for the NO campaign items a ¼ 0.76 and for
the YES campaign items a ¼ 0.71. Typically, an alpha of
approximately 0.6 is seen as the threshold to accept a scale
as sufficiently additive e the items in the scale are suffi-
ciently correlated to consider them indicators of the same
underlying concept.
We use these additive scales in the regression analysis in
the next section to assess the relative explanatory power of
objective knowledge of the Treaty contents e the truth
value of the perceptions of the Treaty e and of the alignment
of the perceptions of the Treaty with the messages of
the two campaigns. An important caveat is necessary,
however, namely that there is the potential for endogeneity
in this model. While we take the alignment of voters with
particular campaign messages to be an explanatory variable
for their vote choice, it is possible that for some voters
the decision to vote YES or NO was made early in the
campaign and that this has lead them to follow a particular
campaign messages more closely, which lead to them
aligning their perceptions with the campaign messages.
Both in terms of the information they pay attention to and
in line with cognitive dissonance theory, such reverse
causality is a genuine possibility.6
5. Explaining vote choice in the referendum
We are primarily interested in the role of knowledge in
the referendum. As is clear from Fig. 2, the level of objective
knowledge of the contents of the Treaty has a substantial
effect on the individual’s voting behaviour. This figure
Table 1
Perceptions of the contents of the Lisbon Treaty. The question is: “Which
of the following do you think are included in the Lisbon Treaty?”.
% Yes % No Correct Campaign
1. Loss of Irish Commissioner
for 5 out of every 15 years
65 9 Correct Ambiguous
2. Ending of Ireland’s right to
decide its own corporate
tax rate
43 22 Incorrect No campaign
3. The introduction of
conscription to a
European army
33 37 Incorrect No campaign
4. The reduction of Ireland’s
voting strength in the
Council of Ministers
48 18 Correct No campaign
5. The end of Ireland’s
control over its
policy on abortion
34 33 Incorrect No campaign
6. The erosion of Irish
neutrality
42 30 Ambiguous No campaign
7. Improved efficiency of EU
decision-making
56 15 Correct Yes campaign
8. Strengthening Europe’s
role in the world
61 12 Correct Yes campaign
9. Improved protection of
workers’ rights
49 19 Ambiguous Yes campaign
10. Strengthening the role
ofnational parliaments
in EU decision-making
43 20 Correct Yes campaign
11. The Charter of
Fundamental Rights
36 14 Correct Yes campaign
3 The questionnaire was designed by a working group comprising
officials from the Irish Department of Foreign Affairs and researchers
from the survey agency (Millward Brown-IMS). Professor Richard Sinnott
acted as technical advisor to the working group.
4 This despite the fact that even under the already adopted Nice Treaty,
Ireland will lose its Commissioner, albeit later.
5 Note that for the knowledge scale, the ‘don’t know’ category is taken
as someone who lacks knowledge, so it is combined with the incorrect
answer, while for the campaign scales, the ‘don’t know’ category is
treated as missing data.
6 We thank the anonymous reviewer for this excellent point.
220 J.A. Elkink, R. Sinnott / Electoral Studies 38 (2015) 217e225
presents the predicted values of the probability of a NO
vote for a particular configuration of the independent variables
(see the caption). The predictions are based on the
logistic regression reported as Model 3 in Table 2 in the
appendix, which includes a large number of control variables,
the measure of knowledge as the number of correct
perceptions of the contents of the Treaty, and subjective
knowledge, but which excludes the two separate
dimensions discussed above. Whereas a knowledgeable
voter who is dissatisfied with the government has slightly
less than a fifty per cent chance of voting NO, a similar voter
with much less knowledge of the Treaty contents has a
probability of voting NO of about ninety per cent. The effect
is strong for both people who are satisfied with the
government and people who are not. While the secondorder
effects of closeness to the opposition and dissatisfaction
with the government are important factors in
explaining the vote, political knowledge played a more
influential role in this referendum.
As suggested in the theoretical discussion above, the
story is somewhat more complicated than the results in
Fig. 2 suggest. The causal diagram in Fig. 1 functions as a
useful guide for a more nuanced interpretation of the
regression results, which are graphically presented in
Fig. 3.
7 Two separate models were estimated to take account
of the different causal paths to be evaluated. When
estimating a causal effect, it is generally advised to control
for any potential confounding factors, but not to control for
any mediating variables (Pearl, 2000; Morgan and Winship,
2007). As can be easily determined on the basis of the
causal diagram, to estimate the effect of objective knowledge
on the vote, we do not want to control for subjective
knowledge, but to estimate the effect of subjective
knowledge, we do want to control for the level of objective
knowledge. For exactly this reason, the two models were
estimated separately. Although Fig. 3 only reports the coefficients
of interest to the theoretical discussion in this
paper, a large set of control variables were included in the
model. Numerical estimates of all variables are reported in
Table 2 in the appendix. Since all variables have been
standardised, we can directly compare the magnitudes of
the effects of each variable.8
It is clear from these estimates that subjective knowledge
of the Treaty has a significant effect on the vote. Objective
knowledge also has a significant effect on the probability of
voting NO, but once we control for subjective knowledge,
this effect is reduced and is statistically insignificant. This
suggests that the impact of objective knowledge is indeed
02468
0.0 0.2 0.4 0.6 0.8 1.0
Low objective knowledge of the Lisbon treaty
Probability of NO vote
Satisfied
Dissatisfied
Fig. 2. Effect of objective knowledge of the Lisbon Treaty and dissatisfaction
with the government on the NO vote. Estimates are based on Model 3 of Table
2 in the appendix. The grey shaded area is a histogram of the objective
knowledge measure. The lines represent the median voter, which is an
hypothetical lower middle class female, between 35 and 49 years old, with
secondary education, who supports an opposition party, had a median score
on the anti-immigration and Irish neutrality scales, who does not consider her
own economic situation to be bad and who does not identify as “Irish only”.
Fig. 3. Absolute values of logistic regression coefficients and their standard
errors, based on models 5 (solid lines) and 6 (dashed lines) in Table 2. The
models contain the same control variables (not shown) and differ only in
terms of the inclusion of the two subjective knowledge variables. Horizontal
lines reflect the 95% confidence intervals, while the vertical markers reflect
the 90% confidence intervals. The full regression table can be found in
Table 2 in the appendix.
7 See Kastellec and Leoni (2007) for an argument in favour of
presenting regression results in this manner.
8 A logistic regression can be perceived of as a linear relation of the
independent variables with a latent dependent variable, whereby the
probability of a 1 on the dependent variable is a non-linear function of
the latent variable (King, 1998). The reported coefficients reflect in all
cases the effect of an increase of two standard deviations (Gelman, 2007)
in the independent variable on the latent dependent variable.
J.A. Elkink, R. Sinnott / Electoral Studies 38 (2015) 217e225 221
through the level of subjective knowledge (Arrow 3 in the
causal diagram in Fig. 1) d the awareness of the voter of his
or her lack of knowledge d and the two dimensions of the
information from the campaign (Arrow 6).
The alignment of the perceptions of the contents of the
Treaty with that of either side of the campaign has a highly
significant effect on the NO vote, even stronger than that of
objective and subjective knowledge variables.9 The more
the voter’s perceptions of the contents of Treaty are in line
with the messages of the NO campaign, the more likely this
voter is to vote NO; the more the perceptions are in line
with the YES campaign, the more likely this voter is to vote
YES. This suggests campaigns can not only influence attitudes
towards the European Union or towards the Treaty,
they can also influence knowledge or perceptions of the
Treaty. Separating out the two dimensions of perceptions
or knowledge of the Treaty thus clearly improves our understanding
of the voting behaviour in this referendum.
The regression results strongly suggest a more complicated
causal explanation than a straightforward relationship between
knowledge and voting behaviour.
Although we have clear results from the regression
analysis that our separation of the two dimensions has
been fruitful, an important caveat regarding this interpretation
is in order. Whereas the existence of a relationship
between objective and subjective knowledge and the reported
voting behaviour is clearly supported by the statistical
results, the support for the relationship with the
campaign depends on observations about the campaign
that go beyond the individual-level data. Casual observation
of the campaign does suggest the focus on perceptions
instead of attitudes and the regression analysis suggests
that these two dimensions do indeed matter.10 We still
have very limited evidence, however, that it really was the
campaign messages that brought about this effect. We have
limited statistical information about the sources of the
perceptions of the respondents d we infer influence of the
campaign from alignment of the attitudes of the respondent
with the messages of the campaign d and we have
limited empirical evidence of how we attributed particular
items to particular campaigns. Furthermore, we do not
directly measure the attitudes of respondents towards the
issues they do or do not perceive to be part of the Treaty.
For this reason, we are not able to directly measure the
interactive effect between perceptions and attitudes as
suggested in our causal diagram in Fig. 1. Our discussion
makes theoretical sense and our empirical findings are in
line with our expectations, but further research is needed
to confirm the theoretical claims made.
6. Conclusion
The Irish referendum on the Lisbon Treaty provides an
excellent opportunity to study voting behaviour in referendums.
Preliminary analyses of a post-referendum opinion
survey suggest that, among other factors, political knowledge
played a crucial role (IMS, 2008; Sinnott et al., 2009).
This article proposes a clear distinction between perceptions
of the referendum issue d what do voters think is
part of the Treaty d and attitudes towards the referendum
issue d how do voters evaluate the elements they think are
part of the Treaty. Although these two components are very
closely related to each other, they are conceptually distinct.
Acknowledging that these are separate components of the
way knowledge and attitudes interact to influence voting
behaviour, we can conceive of a political campaign in a
referendum that emphasises either one of the two components.
The campaigns in the Irish referendum indeed
appear to focus more on influencing the perceptions of the
contents of the Treaty, than on influencing the attitudes
towards these contents. Whereas in some referendums the
issue at hand might be clearly defined and debate centres
on the positive or negative evaluation of this issue, in the
campaign on the Lisbon Treaty, the main debate was on
what was and what was not part of the Treaty.
Including these scales as separate explanatory variables
in a multivariate regression model that measure the extent
to which a respondent is aligned with a particular
campaign in terms of the perceptions of the contents of the
Treaty leads to the conclusion that these alignments are
strong predictors of the vote in the referendum. The more a
voter’s perceptions of the contents of the Treaty are aligned
with the NO campaign, the more likely the voter is to vote
NO, and vice versa for the YES campaign.
While there are good reasons to expect a direct effect of
a lack of subjective knowledge of the referendum issue at
hand on the vote choice in a referendum (see, e.g., Binzer
Hobolt, 2009), the impact of objective knowledge is indirect.
The indirect effect proceeds through various channels,
not least its impact through subjective knowledge, among
which is the particular susceptibility of uninformed voters
to misleading campaign messages. The analysis demonstrates
this effect in the context of the 2008 Irish referendum
on the Lisbon Treaty.
A rational voter can decide not to invest extensively in
acquiring the knowledge to make a fully informed decision
in a referendum as complex as that on the Treaty of Lisbon
and instead to rely on cues from members of the political
elite. The risk in doing so, however, is that these elites will
have particular political interests that might run counter to
those of the voter (Lupia and McCubbins, 1998).
This article makes two contributions to the resolution of
these issues. The first contribution lies in providing a more
nuanced account of the interaction between knowledge,
attitudes, campaigns and voting behaviour in referendums.
The second contribution focuses on the policymaker’s point
of view and argues that governments need to develop ways
and means of ensuring that electorates have access to real
debate and deliberation in a way that enhances knowledge
of the often complex issues that are set before them in the
form of the referendum.
9 These inferences are, of course, based on largely the same information,
namely that recorded in the question presented extensively in Table
1. That this does not lead to unacceptable levels of multicollinearity is
evidenced by the fact that the coefficients on both the objective knowledge
of the Treaty variable and on the campaign variables are statistically
significant. Furthermore, the significant decrease in the Akaike Information
Criterium, by more than twice the number of variables added
(Gelman and Hill, 2007, 525), suggests that the inclusion of the campaign
variables is a significant improvement of the model. 10 Furthermore, a latent trait analysis not reported in this paper provides
supporting evidence that the differentiation between the two dimensions
in the perceptions of the Treaty is reasonable.
222 J.A. Elkink, R. Sinnott / Electoral Studies 38 (2015) 217e225
Acknowledgements
Previously presented at the New Directions in Referendums:
Politics and Campaigns conference, Dublin, May 9,
2013; 1st Annual General Conference of the European Political
Science Association, Dublin, June 16e18, 2011; and
the ECPR Joint Sessions of Workshops, Lisbon, April 2009.
The authors are grateful to Elisabeth Gidengil, Paul Kellstedt,
Theresa Reidy, Jane Suiter and other workshop participants,
as well as the anonymous reviewer, for their
comments.
Appendix
Own economic situation bad
“What about your own economic situation these days?
Would you say it is … ? ” (very good/fairly good/fairly bad/
very bad/don’t know). Values “fairly bad” and “very bad”
are coded as 1, others as 0.
Dissatisfied with government
“Overall are you generally satisfied or dissatisfied with
the way the government is running the country? ” (very
satisfied/quite satisfied/quite dissatisfied/very dissatisfied/
don’t know). Values “very dissatisfied” and “quite dissatisfied”
are coded as 1, others as 0.
Feels close to opposition
“Do you feel close to any of the political parties? ” and if
no, “Do you feel yourself a little closer to one of the political
parties than the others? ” (FF/FG/Labour/Green/Sinn Fein/
PDs/Other/Not close to any). Values “FF”, “PDs” and “Green”
are coded as 1, others as 0.
Identifies as Irish only
“In the near future, do you see yourself as … ? ” (Irish
only/Irish and European/European and Irish/European only/
don’t know). Value “Irish only” is coded as 1, the rest as 0.
Anti-immigration attitude
“Using the card provided please indicate whether
Ireland is made a worse or a better place to live in by people
coming to live here from other countries? ” (1 worse place
to live … 10 better place to live). The variable is entered in
reverse order, 1 ¼ 10, 2 ¼ 9, etc.
Pro-neutrality attitude
“I have a number of statements here that people
sometimes make. I would like you to indicate on this scale
which of each pair of opposing statements comes closest to
your view. A score of one would indicate that you agree
Table 2
Logistic regression explaining the NO vote. Standard errors in parentheses with statistically significant (a ¼ .05) coefficients in bold print. Coefficients are
standardised as suggested by Gelman (2007). The coefficients on age are relative to the 65þ group; those on class relative to the upper middle class; and
those on education relative to those with only primary education.
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6
Constant 0.02 (0.31) 0.08 (0.35) 0.08 (0.36) 0.11 (0.36) 0.05 (0.37) 0.02 (0.37)
Female 0.17 (0.13) 0.07 (0.14) 0.04 (0.15) 0.04 (0.15) 0.00 (0.16) 0.00 (0.15)
Age 18e24 0.37 (0.28) 0.51 (0.31) 0.40 (0.32) 0.42 (0.32) 0.35 (0.33) 0.33 (0.33)
Age 25e34 0.64 (0.23) 0.67 (0.26) 0.63 (0.27) 0.63 (0.27) 0.54 (0.28) 0.54 (0.28)
Age 35e49 0.24 (0.21) 0.37 (0.24) 0.37 (0.25) 0.37 (0.25) 0.33 (0.25) 0.34 (0.25)
Age 50e64 0.23 (0.21) 0.30 (0.24) 0.32 (0.24) 0.30 (0.25) 0.25 (0.25) 0.28 (0.25)
Lower middle class 0.24 (0.20) 0.15 (0.22) 0.11 (0.23) 0.09 (0.23) 0.10 (0.24) 0.12 (0.24)
Skilled worker 0.78 (0.22) 0.62 (0.25) 0.48 (0.26) 0.43 (0.26) 0.46 (0.27) 0.51 (0.27)
Unskilled worker 0.85 (0.24) 0.58 (0.27) 0.47 (0.28) 0.39 (0.28) 0.44 (0.29) 0.53 (0.28)
Large farmer 0.22 (0.30) 0.48 (0.35) 0.53 (0.38) 0.55 (0.38) 0.48 (0.39) 0.49 (0.39)
Small farmer 0.75 (0.50) 0.37 (0.57) 0.55 (0.60) 0.48 (0.60) 0.44 (0.62) 0.52 (0.62)
Secondary education 0.14 (0.25) 0.23 (0.29) 0.32 (0.30) 0.37 (0.30) 0.41 (0.31) 0.34 (0.31)
Third level education 0.42 (0.28) 0.05 (0.33) 0.16 (0.34) 0.28 (0.34) 0.31 (0.35) 0.17 (0.34)
Own economic situation bad 0.76 (0.15) 0.43 (0.18) 0.31 (0.18) 0.29 (0.18) 0.30 (0.19) 0.33 (0.19)
Dissatisfied with government 0.82 (0.13) 0.70 (0.15) 0.76 (0.16) 0.77 (0.16) 0.75 (0.16) 0.74 (0.16)
Feels close to opposition 0.73 (0.12) 0.86 (0.14) 0.92 (0.15) 0.91 (0.15) 0.94 (0.15) 0.96 (0.15)
Identifies as Irish only 0.49 (0.15) 0.34 (0.16) 0.35 (0.16) 0.44 (0.17) 0.41 (0.16)
Pro-neutrality attitude 0.78 (0.14) 0.76 (0.15) 0.80 (0.15) 0.72 (0.15) 0.68 (0.15)
Anti-immigration attitude 0.56 (0.17) 0.39 (0.17) 0.37 (0.17) 0.31 (0.18) 0.33 (0.18)
EU membership a good thing ¡1.27 (0.20) ¡1.11 (0.20) ¡1.09 (0.20) ¡1.00 (0.20) ¡1.03 (0.20)
Too many issues decided by EU 0.64 (0.15) 0.63 (0.15) 0.62 (0.15) 0.50 (0.16) 0.52 (0.16)
EU means low wages 0.69 (0.19) 0.62 (0.19) 0.62 (0.19) 0.54 (0.19) 0.54 (0.19)
Low objective knowledge of EU 0.27 (0.18) 0.17 (0.19) 0.18 (0.19) 0.30 (0.19)
Low objective knowledge of treaty 1.35 (0.19) 1.25 (0.19) 0.52 (0.31) 0.70 (0.30)
Low subjective knowledge of EU 0.06 (0.20) 0.05 (0.20)
Low subjective knowledge of treaty 0.48 (0.20) 0.60 (0.21)
Positive on NO campaign items 1.18 (0.18) 1.12 (0.18)
Positive on YES campaign items ¡0.87 (0.28) ¡0.82 (0.28)
Number of observations 1301 1254 1254 1246 1246 1254
Akaike information criterion 1557 1288 1223 1215 1172 1184
J.A. Elkink, R. Sinnott / Electoral Studies 38 (2015) 217e225 223
fully with the statement on the left. A score of nine would
indicate that you agree fully with the statement on the
right. Of course your view could be somewhere in between.
Also of course there may be issues that you have no
particular view on. If so, please just say this and we will
move on to the next item.” (1 “Ireland should do everything
it can to strengthen its neutrality even if this means being
less involved in European Union cooperation on foreign
and defence policy” … 9 “Ireland should be willing to
accept limitations on its neutrality so that it can be more
fully involved in European Union cooperation on foreign
and defence policy”). Values are in reverse order, 1 ¼ 9,
2 ¼ 8, etc.
EU membership a good thing
“Generally speaking, do you think that Ireland’s membership
of the European Union (European Community) is …
? ” (“A good thing”/“A bad thing”/“Neither good nor bad”/
“Don’t know”). The first category is coded as 1, the rest as 0.
Too many issues decided by EU
“There has been a lot of discussion recently about the
European Union. Some people say that too many issues are
decided on by the European Union, others say that more
issues should be decided on by the European Union. Which
of the following statement comes closest to your view? ”
(“Too many issues are decided by the European Union”/
“The number of issues decided on by the European Union at
present is about right”/“More issues should be decided by
the European Union”/“I have not really thought about it”/“It
depends on the issue”/“Don’t know”). Those that selected
the first option were coded as 1, others as 0.
EU means low wages
“What does the European Union mean to you personally?
” (many answer categories, among which “Lower
wage rates”). Those that included “Lower wage rates” in
their selection were coded as 1, others as 0.
Low objective knowledge of Treaty
Number of items deemed to be correct in Table 1 that
are perceived to be in the Treaty; ‘don’t know’ taken as an
incorrect answer.
Low objective knowledge of EU
“For each of the following statements about the European
Union could you please tell me whether you think it is
true or false? ” (“The EU currently consists of fifteen
Member States”/“Switzerland is a member of the European
Union”/“Every six months, a different Member State becomes
the President of the Council of the European Union”/
“The members of the European Parliament are directly
elected by the citizens of the EU”). A four point score is
generated, from 1 (4 correct answers) to 4 (1 correct
answer). All respondents had at least one correct answer.
Low subjective knowledge of Treaty
“By the date of the referendum (12 June), how good was
your understanding of the issues involved? Please use this
card to choose the phrase that applies best to you.” (“I had a
good understanding of what the Treaty was all about”/“I
understood some of the issues but not all that was
involved”/“I was only vaguely aware of the issues
involved”/“I did not know what the Treaty was about at
all”/don’t know). The resulting variable is a four point scale
from good understanding to not knowing what the Treaty
was about at all.
Low subjective knowledge of EU
“And how about the European Union in general? Using
this scale, how much do you feel you know about the European
Union, its policies, its institutions? ” (1 nothing at all
… 10 know a great deal).
Positive on NO campaign items
Number of items associated with the NO campaign in
Table 1 that are perceived to be in the Treaty; ‘don’t know’
taken as missing data.
Positive on YES campaign items
Number of items associated with the YES campaign in
Table 1 that are perceived to be in the Treaty; ‘don’t know’
taken as missing data.
Abstained
“On the 12 June last, a referendum was held on the
Treaty of Lisbon. As you may remember, many people did
not vote in that referendum. How about you? Did you vote
in the referendum on the Treaty of Lisbon that was held in
June? ” (“Yes e voted”/“Did not vote”). In all analyses, this is
used as a filter variable, and only those who voted are taken
into account.
Voted NO
“How did you vote in that referendum e in favour or
against the Lisbon Treaty? ” (“Voted e In favour”/“Voted e
Against”). This is the dependent variable in the regression
analyses.
References
Ajzen, I., Fishbein, M., 1977. Attitude-behavior relations: a theoretical
analysis and review of empirical research. Psychol. Bull. 84 (5),
888e918.
Alvarez, R.M., 1997. Information and Elections. University of Michigan
Press.
Basinger, S.J., Lavine, H., 2005. Ambivalence, information, and electoral
choice. Am. Polit. Sci. Rev. 99, 169e184.
Binzer Hobolt, S., 2005. When Europe matters: the impact of political
information on voting behaviour in EU referendums. J. Elections
Public Opin. Parties 15 (1), 85e109.
Binzer Hobolt, S., 2009. Europe in Question: Referendums on European
Integration. Oxford University Press, New York.
224 J.A. Elkink, R. Sinnott / Electoral Studies 38 (2015) 217e225
Bowler, S., Donovan, T., 1994. Information and opinion change on ballot
propositions. Polit. Behav. 16 (4), 411e435.
Campbell, A., Converse, P.E., Miller, W.E., Stokes, D., 1960. The American
Voter. University of Chicago Press, Chicago.
De Vries, C.E., Van der Brug, W., Van Egmond, M.H., Van der Eijk, C., 2011.
Individual and contextual variation in EU issue voting: the role of
political information. Elect. Stud. 30 (1), 16e28.
Delli Carpini, M.X., Keeter, S., 1996. What Americans Know about Politics
and Why it Matters. Yale University Press.
Druckman, J.N., 2004. Political preference formation: competition,
deliberation, and the (ir)relevance of framing effects. Am. Polit. Sci.
Rev. 98 (4), 671e686.
Druckman, J.N., Nelson, K.R., 2003. Framing and deliberation: how citizens’
conversations limit elite influence. Am. J. Polit. Sci. 47 (4),
729e745.
Ehin, P., 2001. Determinants of public support for EU membership: data
from the Baltic countries. Eur. J. Polit. Res. 40, 31e56.
Franklin, M., 2002. Learning from the Danish case: a comment on Palle
Svensson’s critique of the Franklin thesis. Eur. J. Polit. Res. 41,
751e757.
Franklin, M., Marsh, M., Wlezien, C., 1994. Attitudes towards Europe and
referendum votes: a response to Siune and Svensson. Elect. Stud. 13,
117e121.
Gabel, M.J., 1998. Economic integration and mass politics: market liberalisation
and public attitudes in the European Union. Am. J. Polit. Sci.
42 (3), 936e953.
Gamble, A., 2006. The European disunion. Br. J. Polit. Int. Relat. 8 (1), 34e49.
Garry, J., Marsh, M., Sinnott, R., 2005. ‘Second-order’ versus ‘issue-voting’
effects in EU referendums. Eur. Union Polit. 6 (2), 201e221.
Gelman, A., 2007. Scaling regression inputs by dividing by two standard
deviations. Stat. Med. 27 (15), 2865e2873.
Gelman, A., Hill, J., 2007. Data Analysis Using Regression and Multilevel/
Hierarchical Models. Analytical Methods for Social Research. Cambridge
University Press, Cambridge.
Grynaviski, J.D., Corrigan, B.E., 2006. Specification issues in proximity
models of candidate evaluation (with issue importance). Polit. Anal.
14, 393e420.
Hix, S., 2006. Why the EU Needs Left-right Politics. In: Politics: the Right or
the Wrong Sort of Medicine for the EU?. Notre Europe Policy Paper 19.
Hooghe, L., Marks, G., 2009. A postfunctionalist theory of European
integration: from permissive consensus to constraining dissensus. Br.
J. Polit. Sci. 39 (1), 1e23.
IMS, 2008. Post Lisbon Treaty Referendum Research Findings (Technical
report Millward Brown-IMS).
Kastellec, J.P., Leoni, E.L., 2007. Using graphs instead of tables in political
science. Perspect. Polit. 5 (4), 755e771.
King, G., 1998. Unifying Political Methodology. The Likelihood Theory of
Statistical Inference. University of Michigan Press.
Krosnick, J.A., 1988. Attitude importance and attitude change. J. Exp. Soc.
Psychol. 24, 240e255.
Laffan, B., 2008. Reflection: three readings of the failure to ratify the
constitutional treaty. In: Brown, T. (Ed.), Lisbon: What the Reform
Treaty Means. Institute of International and European Affairs, Dublin,
pp. 60e70.
Lord, C.G., Ross, L., Lepper, M.R., 1979. Biased assimilation and attitude
polarization: the effects of prior theories on subsequently considered
evidence. J. Personal. Soc. Psychol. 37 (11), 2098e2109.
Lupia, A., 2006. How elitism undermines the study of voter competence.
Crit. Rev. 18 (1), 217e232.
Lupia, A., McCubbins, M.D., 1998. The Democratic Dilemma: Can Citizens
Learn What They Need to Know? Cambridge University Press, New
York.
Miller, J.M., Krosnick, J.A., 2000. News media impact on the ingredients of
presidential evaluations: politically knowledgeable citizens are
guided by a trusted source. Am. J. Polit. Sci. 44 (2), 301e315.
Morgan, S.L., Winship, C., 2007. Counterfactuals and Causal Inference.
Methods and Principles for Social Research. Cambridge University
Press, New York.
Nie, N.H., Verba, S., Petrocik, J.R., 1979. The Changing American Voter.
Harvard University Press Cambridge.
Palfrey, T.R., Poole, K.T., 1987. The relationship between information,
ideology, and voting behavior. Am. J. Polit. Sci. 31 (3), 511e530.
Pearl, J., 2000. Causality: Models, Reasoning, and Inference. Cambridge
University Press, Cambridge.
Perloff, R.M., 2003. The Dynamics of Persuasion. Communication and
Attitudes in the 21st Century, second ed. Lawrence Erlbaum Associates,
Mahwah, NJ.
Quinlan, S., 2009. The Lisbon treaty referendum 2008. Ir. Polit. Stud. 24
(1), 107e121.
Reif, K., Schmitt, H., 1980. Nine second order national elections: a conceptual
framework for the analysis of European election results. Eur. J.
Polit. Res. 8 (1), 3e44.
Sinnott, R., 2001. Attitudes and Behaviour of the Irish Electorate in the
Referendum on the Treaty of Nice (Technical report, European Commission
Representation in Ireland).
Sinnott, R., 2003. Attitudes and Behaviour of the Irish Electorate in the
Second Referendum on the Treaty of Nice (Technical report, European
Commission Representation in Ireland).
Sinnott, R., Elkink, J.A., O’Rourke, K., McBride, J., 2009. Attitudes and
Behaviour in the Referendum on the Treaty of Lisbon (Technical
report, Department of Foreign Affairs).
Suiter, J., Reidy, T., 2013. It’s the campaign learning stupid, an examination
of a volatile Irish referendum. Parliam. Aff. (published online).
Van Apeldoorn, B., 2009. The contradictions of ’embedded neoliberalism’
and Europe’s multi-level legitimacy crisis: the European project and
its limits. In: Van Apeldoorn, B., Drahokoupil, J., Horn, L. (Eds.), Contradictions
and Limits of Neoliberal European Governance. From
Lisbon to Lisbon. Palgrave MacMillan, pp. 21e43.
Zaller, J., 1992. The Nature and Origins of Mass Opinion. Cambridge University
Press, Cambridge.
J.A. Elkink, R. Sinnott / Electoral Studies 38 (2015) 217e225 225

Our academic experts are ready and waiting to assist with any writing project you may have. From simple essay plans, through to full dissertations, you can guarantee we have a service perfectly matched to your needs.

GET A 40% DISCOUNT ON YOU FIRST ORDER

ORDER NOW DISCOUNT CODE >>>> WELCOME40

 

 

Posted in Uncategorized

campaign effects and voter behaviour in referendums

Our academic experts are ready and waiting to assist with any writing project you may have. From simple essay plans, through to full dissertations, you can guarantee we have a service perfectly matched to your needs.

GET A 40% DISCOUNT ON YOU FIRST ORDER

ORDER NOW DISCOUNT CODE >>>> WELCOME40

European Union Politics
2014, Vol. 15(2) 235–254
! The Author(s) 2013
Reprints and permissions:
sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/1465116513514780
eup.sagepub.com
Article
Emotions and voting in
EU referendums
John Garry
Queen’s University Belfast, UK
Abstract
There is an emerging scholarship on the emotional bases of political opinion and behaviour
and, in particular, the contrasting implications of two distinct negative emotions –
anger and anxiety. I apply the insights in this literature to the previously unresearched
realm of the emotional bases of voting in EU referendums. I hypothesise that anxious
voters rely on substantive EU issues and angry voters rely on second-order factors
relating to domestic politics (partisanship and satisfaction with government). Focusing
on the case of Irish voting in the Fiscal Compact referendum, and using data from a
representative sample of voters, I find support for the hypotheses and discuss the
implications of the findings for our understanding of the emotional conditionality of
EU referendum voting.
Keywords
Emotions, EU referendums, Ireland, voting
Introduction
Much of the research to date on voting behaviour in EU referendums has focused
on the distinction between the ‘second-order national election’ approach and the
‘substantive issues’ approach (Glencross and Trechsel, 2011). According to the
former, originally elaborated by Reif and Schmitt (1980) to explain electoral behaviour
in European Parliament elections, citizens are driven by domestic political
factors when voting on an ostensibly European issue. Thus, citizens may use the
referendum as a vehicle for casting judgement on their national politicians: if
unsatisfied with the domestic government’s performance, they can ‘punish’ the
government by refusing to support the EU treaty it is advocating in the referendum;
if satisfied, they may reward the government by supporting it and voting yes
Corresponding author:
John Garry, Queen’s University Belfast, 25 University Square, Belfast BT7 1PA, UK.
Email: j.garry@qub.ac.uk
at the referendum (Franklin, 2002; Franklin et al., 1994, 1995). Voters may also be
driven by domestic factors in the sense of relying on their partisan attachments.
They may follow the line of their party when voting in the referendum, relying on
party cueing rather than an in-depth reflection upon issues relating to the specific
referendum under consideration (Hobolt, 2006, 2007).
In contrast, the substantive issues approach is more consistent with the notion of
deliberative democracy and assumes that citizens do actually engage with, and
reflect reasonably upon, European issues when voting in an EU referendum
(Siune and Svensson, 1993; Siune et al., 1994; Svensson, 1994, 2002). This may
occur at one of two levels. Voters may rely on their underlying ideological disposition
towards European integration, with pro-integrationists supporting a treaty
and eurosceptics opposing it. Voters may also be driven by their assessment of
aspects of the specific content of the treaty at issue: those who perceive positive
consequences resulting from treaty acceptance will favour the treaty while those
calculating negative consequences will be opposed (on this two-level distinction, see
Garry, 2013).
The implications of the second-order versus substantive issues debate are
important for understanding citizens’ individual level political behaviour in EU
referendums and also for normative debates surrounding the use (or non-use) of
referendums as a means of legitimising regional integration advance. If national
domestic political concerns, unrelated to the actual content of the treaty being
voted upon, drive behaviour then a normative defence of referendums as facilitators
of calm, reasoned and evidence-based reflection and decision making is diffi-
cult (see Moravcsik, 2006). However, the evidence to date suggests that citizens are
indeed more likely to rely on substantive issues than on second-order factors when
reaching judgement in an EU referendum context (see discussion in Garry, 2013).
This article seeks to advance knowledge in this field by investigating the emotional
conditions under which either second-order or substantive issues are relied
upon to drive voting. Previous work on conditionality has focused on the contextual
level (for instance, the more intense the EU referendum campaign, the greater
the role of substantive issue-voting and the less important the role of second-order
factors; Hobolt, 2006, 2007) or the individual level (for instance, it is argued that
issue voting in EU referendums is associated with relatively sophisticated voters
and second-order voting with less sophisticated voters; Hobolt, 2005, 2009). Here,
I focus on the individual level and investigate the previously unresearched role
played by emotions in driving the relative importance of second-order and issue
factors in EU referendum voting. Recent work on the emotional basis of political
behaviour has moved away from comparing positive to negative affect and instead
focuses on the distinct implications of different types of positive and negative
affect – in particular isolating the contrasting implications for political behaviour
of the two negative emotions of anger and anxiety (see Feldman et al., 2012 for an
overview). Notably, research from a neuropsychology perspective (Marcus, 2002;
Marcus et al., 2000) suggests that anxiety and fear are associated with a decrease in
reliance on habitual behaviour and an increase in the use of reason, information
236 European Union Politics 15(2)
search and reflective judgement. In direct contrast to anxiety, ‘anger may lead
people away from the kind of deliberation that many democratic theorists cherish’
(Valentino et al., 2008: 267) and is instead associated with behaviour driven by
long-standing political allegiance (MacKuen et al., 2010) or by the desire to blame
and punish (Petersen, 2010). Research by Druckman and McDermott (2008) also
suggests a direct link between the type of negative emotion and attitudes to ‘risky’
policies: angry citizens are likely to support a risky option while anxious citizens
will oppose it.
I apply these insights to the realm of EU referendum voting. First, I hypothesise
that anxious and fearful citizens are more likely than angry citizens to be issue
voters who learn about, reflect upon and vote on the basis of the implications of the
substantive content of the treaty. Second, I hypothesise that angry citizens are more
likely than anxious and fearful citizens to rely on second-order factors, either using
the referendum as a vehicle to pass judgement on their domestic government
(blaming and punishing their government if they are dissatisfied with its performance)
or relying on habitual partisan-based voting (in line with the recommendation
of their favoured party). Third, I hypothesise a direct effect of anger on
support for the referendum option that is portrayed in the campaign as ‘risky’
and a direct effect of anxiety on support for the particular referendum option
‘framed’ as ‘non-risky’.
I test these general hypotheses in the specific case of recent voting behaviour in
the Irish referendum on the ‘Fiscal Compact’ treaty,1 the only state to have a
referendum on the Treaty agreed to by all member states except the UK and the
Czech Republic in March 2012. Using data from a post-referendum voter survey,
I empirically demonstrate that issue voting (and particularly treaty-specific considerations)
is associated with anxiety while second-order voting (specifically, the
impact of government evaluation) is associated with anger. A strong direct relationship
also emerges between anger and voting for the ‘risky’ No option.
The emotional bases of voting in EU referendums
Emotions and voting
Party identification theory, which assumes that a tendency to vote for a particular
party is the result of an affective attachment to that party, has been the dominant
psychological approach in the electoral behaviour literature. Ever since it was originally
elaborated by Campbell et al. (1960), party identification theory has
informed almost every national and international election study and spawned hundreds
of analyses testing and contesting the approach. Although sometimes conceptualised
and operationalised in terms of both positive and negative affect
(Richardson, 1991), party identification is typically studied in terms of the
relationship between positive affect (positive party attachment) and vote choice.
Other psychological approaches to understanding voting have pitched general positive
affect against general negative affect, for example using thermometer scales
Garry 237
(how warm or cold do voters feel towards specified political actors?) or likeability
scales (to what extent do voters like or dislike certain actors?).
In contrast to these approaches, the influence of particular types of emotions on
vote choice has played – until very recently – a minor role. An influential point of
departure is the Theory of Affective Intelligence (TAI) developed by Marcus and
co-authors (Marcus and MacKuen, 1993; Marcus et al., 2000). The TAI draws
upon a neuro-psychological approach to emotions and specifies a number of emotional
subsystems that govern the extent to which humans rely upon either habit or
reason. The ‘disposition’ system induces enthusiasm that in turn triggers habitoriented
behaviour. In contrast, the ‘surveillance’ system monitors the environment
for potentially threatening stimuli; if such stimuli are detected, anxiety is induced.
Anxiety cuts off reliance on habit-related activity and instead prompts conscious
attention, information gathering and reasoning to deal with the potential threat.
Marcus et al. (2000) demonstrate that in relation to voting in US elections, anxiety
leads to a decrease in reliance on (habit-oriented) partisan-based behaviour and an
increase in reasoning and reliance on substantive political issues. While such findings
have attracted methodological criticism (Ladd and Lenz, 2008, 2011), Marcus
et al. (2011; see also Brader, 2011) have sought to demonstrate their robustness.
Thus, somewhat counter-intuitively given that it is a negative emotion, anxiety
may well enhance the quality of democracy as it is associated with facilitating citizens
to engage in evidence-based consideration of opposing points of view. Accordingly,
Valentino et al. (2008) investigate whether a worried citizen a good citizen. In their
analysis, the authors contrast anxiety with a second distinct negative emotion: anger.
They conclude that ‘concern about fear mongering in politics may be at least somewhat
overblown … outrage, perhaps, is more damaging than fear if we hope to foster
an informed citizenry’ (267). Similarly, MacKuen et al. (2010) argue that anger
(which they refer to as aversion) contrasts with anxiety in that the former operates
within the ‘disposition’ system and is consequently more likely than anxiety to be
associated with habit-oriented behaviour. The authors’ state:
A standard expectation is that when people get emotional about politics, their positions
become entrenched, resolute, and steadfast. Our theory predicts that this will be
true – for aversion but not for anxiety … people can engage either as deliberative or
partisan citizens, where the kind of citizenship practiced will depend upon whether
people react with anxiety or with anger. (MacKuen et al., 2010: 442)
Petersen (2010: 358) also directly contrasts anxiety and anger, suggesting that
the two emotions ‘are not only distinct but operate in distinct domains of political
life’. While, in line with TAI, he suggests that anxiety prompts the consideration of
issues, he draws on evolutionary psychology research to suggest that anger is a
moral emotion and is associated with the desire to attribute blame to (and then
punish) particular actors who are responsible for negative events.
In addition to these moderating (conditional or indirect) effects of negative
emotions on political opinion (i.e. anxiety is associated with issue-based
238 European Union Politics 15(2)
deliberation while anger is associated with habit-based partisanship and blaming/
punishing), it is also likely that there is a direct impact of negative emotion on
opinion formation. Druckman and McDermott (2008) argue that there is a direct
link between anger and support for ‘risky’ policies and a direct link between anxiety
and support for ‘non-risky’ policies. This is due to the different levels of certainty
associated with feelings of anger and anxiety. Drawing on the work of Lerner et al.
(2003) and Smith and Ellsworth (1985), Druckman and McDermott (2008: 301)
suggested that that ‘hostility or anger produces optimism about future outcomes
(due to certainty) and risk-seeking choices… In contrast, distress or fear leads to
pessimistic judgements (due to uncertainty) of the future and greater risk aversion’.
The authors conducted experiments inviting participants to choose from a set of
possible responses to disease outbreak and the need for investment, with the
response options varying in terms of certainty and riskiness. When the emotional
state of the respondents was related to option choice, they found that ‘anger
encourages greater risk seeking while distress encourages a more cautious
approach’ (Druckman and McDermott, 2008: 317).
Emotions and EU referendum voting
These insights map fairly neatly onto the second-order versus issues debate relating
to vote determinants in EU referendums. The ‘issues approach’ suggests that citizens
do rely upon EU related factors when deciding how to vote in an EU referendum
(Siune and Svensson, 1993; Siune et al., 1994; Svensson, 1994, 2002).
However, an important distinction within this approach relates to whether citizens
rely upon underlying ideological disposition or treaty-specific assessments (Garry,
2013). Voters who decide on the basis of underlying ideology may be distinguished
in terms of eurosceptics who are generally opposed to further EU integration and
euro-integrationists who are generally supportive of further unity. The scepticintegrationist
ideological distinction may have roots in a number of sources: economic
factors (Anderson and Reichert, 1996; Eichenberg and Dalton, 2007; Gabel,
1998a, 1998b), culture (de Vreese and Boomgaarden, 2005; McLaren, 2002) or
identity (Carey, 2002; McLaren, 2004). A somewhat different approach focuses
on citizens’ evaluations of the specific treaty under consideration rather than
their more long standing disposition towards the EU project. Citizens are assumed
to be information seekers who engage with and reflect upon debates relating to the
treaty at issue. If, after consideration of the specific content of the treaty, citizens
regard the consequences of acceptance of the treaty as positive, then they are
assumed to want to accept the treaty. Citizens who foresee deleterious consequences
in the event of acceptance are assumed to vote no.
The extent to which perceptions of the consequences of the treaty drive referendum
vote choice may plausibly be conditional upon the inducement of anxiety
during the campaign. The argument that it is citizens’ calculations regarding treaty
consequences that shape their vote choice essentially assumes that the treaty is a
potential threat: citizens’ attention to the treaty is triggered, leading to the search
Garry 239
for treaty-related information and reliance on reason and judgement designed to
minimise the perceived threat relating to either acceptance or rejection of the treaty.
Such information seeking and reasoning is consistent with the consequences of
anxiety induction, as described in the previous section.
The influence of underlying values – the sceptics versus integrationists ideological
distinction – on vote choice is also plausibly conditional upon anxiety.
Reliance upon EU ideological disposition represents citizen engagement with the
substantive domain of EU politics (rather than merely voting in line with your own
party or rewarding/punishing the government) and so is consistent with the induction
of anxiety and the consequent triggering of attention and reason. However, the
conditional effect is likely to be less strong than that relating to treaty-specific issues
given that the sceptic-enthusiast distinction is, to an extent, a general short-cut
heuristic rather than a direct engagement with and reflection upon the specific
issues relating to the treaty. Hence, both issue effects (ideology and treaty specific
considerations) are likely to be conditional upon the induction of anxiety; strongly
so, for treaty-specific issues that are indicative of attention and learning, and less
strongly so for ideology as it is indicative of reliance on an underlying disposition.
As far as ‘second-order’ voting is concerned, this approach has mainly been
studied in the context of European Parliament elections, with analysts investigating
the extent to which domestic political concerns drive voting in what is ostensibly an
‘EU’ election (Reif and Schmitt, 1980; for a recent analysis, see Hix and Marsh,
2011). The approach may also valuably be applied to EU referendums (see discussion
in de Vreese and Semetko, 2004: 701–702) in order to assess whether voting
behaviour is driven by the substantively important political issue (the EU referendum)
or domestic political concerns such as partisanship and evaluation of government
performance. In an EU referendum campaign, citizens in whom the treaty
referendum campaign induces anxiety are unlikely to rely upon second-order factors
when deciding how to vote. Notably, anxious citizens are unlikely to be influenced
by habit-oriented heuristics such as party cueing and partisanship. This is not
to argue that partisan-based voting will not be an important driver of voting,
consistent with previous work (Hobolt, 2006, 2007), but its level of importance
will depend upon emotional disposition. Anxious voters will be focused on the
substantive details of the campaign and unlikely to rely on the partisan heuristic.
Given that, according to TAI, anger (an example of aversion) is associated with the
habit oriented disposition system (rather than the surveillance system with which
anxiety is linked), angry citizens are hypothesised to be more likely than anxious
citizens to rely on partisanship when voting in the referendum (see discussion in
MacKuen et al., 2010: 441–442).
In addition to partisanship, the other aspect of second-order voting relates to the
use of referendums to pass judgement on the performance of the domestic government.
Anger is more likely than anxiety to prompt voters to ignore the substantive
EU issues in the campaign and simply use the referendum as a chance to express
their evaluation of the national government. Building on the work of Petersen
(2010) and Lazarus (1991), anger encourages voters to identify responsible actors
240 European Union Politics 15(2)
and pass judgement. During a referendum campaign, consistent with the secondorder
referendum voting approach, voters are likely to direct their anger at the
national government and to blame the government; they may hold it to account at
the time of referendum voting and punish it by voting against the government’s
preferred position on the referendum. Hence, previous work highlighting the
importance of domestic government evaluation in EU referendum vote choice
(Franklin, 2002; Franklin et al., 1994, 1995) is valid, but is expected to be dependent
upon the emotional make-up of the voter: angry voters are more likely than
anxious voters to operate in the moral domain of holding responsible and passing
judgement on the government.2
H1: Anxious citizens are more likely than angry citizens to be ‘issue’ voters who vote
on the basis of the substantive content of the EU Treaty.
H2: Angry citizens are more likely than anxious citizens to be ‘second-order’ voters
whose voting is driven by their attitude to domestic politics (government approval and
partisanship).
It is worth explicitly emphasising that the conditional nature of the hypotheses just
discussed prompts a consideration of what Petersen (2010: 359) notes as ‘the diffi-
culties associated with analyzing emotions and cognitions’. Petersen highlights
Huddy et al.’s statement (2007: 210) that ‘cognitions and emotions operate in a
recursive loop in which cognitions influence emotions which influence cognitions,
in turn—making it difficult to know whether cognitive appraisals are a cause or
consequence of differing emotional states’. Petersen (2010: 359) suggests that emotions
and cognitions co-occur and thus the aim of empirical analysis should be to
use interactive modelling to demonstrate that the effects of a specified explanatory
factor co-occur with a specified emotional state. Accordingly, the aim of the current
study may be seen as somewhat modest in terms of making causal claims: the
aim is to identify whether certain specified negative emotional states (anger or fear)
are associated with certain specified explanations of referendum voting: is anxiety
linked with issue-based voting and is anger linked with either partisan or blamebased
voting?
In addition to the moderating effects of anger and anxiety, there are also likely to
be direct effects. Druckman and McDermott’s (2008) findings regarding the direct
link between anger and choice of risky policy option (and anxiety and choice of
non-risky policy option) have implications for referendum vote choice. If voting in
a particular way in a referendum (e.g. voting No) is, during the referendum campaign,
strongly ‘framed’ or portrayed as a very risky option then angry voters will
tend to vote No. Similarly, if voting in a particular way (e.g. voting Yes) is characterised
during the campaign as the risk-free option, anxious voters will tend to
vote Yes.
H3: Angry citizens are more likely than anxious citizens to vote for the referendum
option which is characterised in the campaign as the ‘risky’ option.
Garry 241
Whether it is the Yes or No option that is ‘framed’ in the campaign as the risky
option is obviously dependent on the specific campaign. The Irish referendum
campaign is now briefly focused on in order to situate substantively and contextually
the generally applicable hypotheses that have just been explicated.
The case of Ireland and the ‘Fiscal Compact’ treaty
The European Fiscal Compact was designed to implement a set of financial rules in
participating states that would enhance the long-term survival chance of the Euro
currency, essentially placing constraints on states’ borrowing and spending activities.
In return for these accepted constraints, states would have access to a central
fund that would prevent complete state bankruptcy by providing emergency capital
in the event of a state being locked out of international capital markets.
At the time of the referendum, the effectively bankrupt Irish state was being kept
financially alive by a bailout from the combined forces of the IMF and the European
Union. Pro-treaty actors emphasised that an even worse economic catastrophe
would descend in the event of a rejection of the treaty, as rejection would prohibit
access to any further bailout funds. Pro-treaty actors argued that such a bailout was
looking ominously necessary as Ireland remained stubbornly isolated from international
funding markets. Anti-treaty actors emphasised that the main Irish parties
and the EU had generated the crisis and were not deserving of trust or support; they
argued that rejection of the treaty would not place Ireland in a precarious and vulnerable
economic position, but would rather put an end to the wrong-headed austerity
approach to economics pushed by the government and international lenders.
The Fine Gael/Labour coalition government portrayed itself as behaving
responsibly in the crisis but was seen by its detractors as behaving reprehensively,
breaking its general election promises related to obtaining an improved deal for
Ireland and instead imposing austerity measures upon an already suffering population.
Only one party was clearly opposed to the Treaty: Sinn Fe´in. It ran an
energetic anti-Treaty campaign that condemned the anti-growth pro-austerity
treaty and what it regarded as the scaremongering tactics of the government who
portrayed No voting as profoundly risky behaviour in a highly uncertain economic
climate. However, it was the Yes side that achieved victory: on polling day, 31 May
2012, Irish citizens voted yes by a margin 60.3% to 39.7%, with a turnout of 50%
(for a description of the campaign, see Fitzgibbon, 2013).
The Irish referendum on the Fiscal Treaty is arguably a good case to focus on to
demonstrate the effects of anxiety and anger on referendum vote choice. The referendum
was very plausibly a novel and threatening stimulus on the Irish political
environment. Ireland was already in a very precarious economic state and much of
the debate related to the strong possibility of a radical worsening of the situation,
with the spectre of Ireland literally running out of money to keep hospitals and
schools functioning. In such a campaign context, the inducement of anxiety in
many citizens is likely. This referendum represented not just another run of the
mill ordinary election or mundane referendum; the economic survival of the state
242 European Union Politics 15(2)
was at issue. This novel threat was likely to trigger, in many citizens, a discontinuation
of the ‘every day habit-based behaviour’ associated with the disposition
system and its replacement with a focused attention on the threat (the referendum)
and how to react to it in a way that could minimise reduction in well-being.
The nature of the Irish referendum campaign was also likely to incite anger at
those deemed responsible for the country’s disastrous position. Critics argued that
the government was not a passive actor in the Irish crisis; rather it played a key role
in exacerbating the economic difficulties. The government was formed in 2011 on
the back of election promises to renegotiate Ireland’s relationship with its international
lenders to achieve an amelioration of its position. It attracted much criticism
for failing to do so, leaving Labour’s general election campaign slogan ‘It’s
either Labour’s way or Frankfurt’s way’ looking like a slogan the party would
rather forget.
In short, there is good reason to expect many voters to be induced to be either
angry or anxious in the Irish ‘Fiscal Compact’ referendum campaign and this
provides fertile ground for assessing the possible (direct and moderating) effects
of such emotions on referendum vote choice. Indeed, in an interview with an Irish
Times journalist after the referendum, one close observer of the referendum campaign,
Gerard O’Neill of Ama´rach Research, interpreted the result of the referendum
and the nature of the campaign very much through the lens of the fear/anger
distinction:
The great struggle was between fear and anger – and fear won. Anger is about the past
and the present, about what has happened to people’s lives, about cuts and unemployment
and emigration … Fear is about the future – about what might happen to the
euro, to the world, to Ireland, to the economy. So you had that tension between anger
and the past and the present, and the referendum, which is about the future and
whether Ireland will have access to funds and that’s a vision of the future that motivates
fear. (Irish Times, 2 June 2012)
Methods
Experiments are particularly valuable with respect to identifying causal mechanisms
and have been used fruitfully with respect to the study of emotions and voting
(recent examples include Kiss and Hobolt, 2011, 2012; MacKuen et al., 2010). A
downside of experiments is that they are, by their nature, typically abstract and
divorced from the real world of campaigns and voting behaviour. Here, I rely on a
survey designed to capture voting in an actual EU referendum. While limited in
terms of unpacking causal order, survey datasets are useful in terms of identifying
conditional relationships via the use of interaction terms in appropriately structured
regression models (as emphasised by Petersen, 2010: 359) and have the
advantage of capturing important political events, such as voter behaviour in the
Irish ‘Fiscal Treaty’ referendum.
Garry 243
RedC polling company was commissioned to conduct a survey of voters (as
opposed to a survey of all citizens) in the direct aftermath of the referendum. A
representative sample of 1000 voters was achieved via telephone interviewing.3 In
order to generate a measure of voters’ perceptions of the likely economic consequences
of acceptance of, or rejection of, the treaty the following questions were
asked:
Now that Ireland has voted ‘Yes’ in favour of the Fiscal Treaty, I would like you to
think about how much you agree or disagree with each of the following statements
about voting in favour of the treaty. A ‘Yes’ vote will …
… make the Irish economy more stable and secure
… help Ireland avoid complete bankruptcy
(disagree strongly, disagree, agree, agree strongly, don’t know)
If Ireland had voted ‘No’ and rejected the Fiscal Treaty do you think this would have …
… led to great uncertainty about Ireland’s economic future
(disagree strongly, disagree, agree, agree strongly, don’t know)
Responses to these three items were summed and rescaled to run from 0 to 1, with
higher scores indicating strongly negative perceptions of the economic consequences
of treaty rejection. In order to measure citizens’ underlying values regarding
the EU, the following question was asked:
As regards the European Union in general, which of the following statements comes
closest to your view? Ireland should do all it can to unite fully with the European
Union VERSUS Ireland should do all it can to protect its independence from the
European Union
In order to tap voters’ satisfaction with the Fine Gael/Labour government,
respondents were asked:
How satisfied are you with the performance of the Fine/Gael Labour government:
very satisfied, fairly satisfied, not very satisfied or not at all satisfied?
In relation to measuring partisanship, respondents were asked which party they
would give their first preference vote to if a general election were to be held. Three
categories are generated: supporters of parties clearly in favour of the Treaty (Fine
Gael, Labour, Fianna Fa´il), supporters of the party clearly against the Treaty (Sinn
Fe´in),4 and other respondents.
In order to measure whether voters were ‘angry’ or ‘anxious/fearful’, the following
question was asked:
During the referendum campaign, when people were talking about the Fiscal Treaty,
some people said they were angry and other people said they were fearful. What about
you, which did you feel most – anger or fear?5
244 European Union Politics 15(2)
The word ‘fear’ was used rather than ‘anxious’ or ‘anxiety’ in order to make the
question as easy as possible for the respondent to answer. As noted by Bogdan
(2012), while fear and anxiety may not be precisely the same (Rhudi and Meagher,
2000), they are very similar (Davis, 1992). Furthermore, the terms are often used
interchangeably in the literature (e.g. Petersen, 2010: 360, 363; see also the discussion
in Druckman and McDermott, 2008: 317). The forced-choice nature of the
question may be of some concern. Again, the aim was to ask the question in as
simple and colloquial a way as possible. This style of question loses nuance (exactly
how fearful, exactly how angry?), but it does achieve the desired aim of a straightforward
categorisation of voters (see discussion of the measurement of emotions in
Marcus et al., 2006).
A number of questions were also asked to generate control variables that,
although not directly relevant for testing the hypotheses, have been shown to be
predictors of referendum vote choice in previous research. Social class was identified
using the usual social grade categories used in the UK (distinguishing between
professional (abc1) and manual (c2de) groups (as well as farmers)). The sex of the
respondent was noted and age was measured in categorical form. Some authors
also include a general measure of ideology, such as respondents’ self-placement on
a left–right scale, as a control variable. In the questionnaire used in this study, left–
right self-placement was not included. In any event, it may not have served as a
useful control variable given that the meaning of the measure is ambiguous
(Inglehart and Klingemann, 1976; Knutsen, 1997), meaning that it would not be
entirely clear what in fact left–right self-placement would be controlling for.
Results
In order to test the hypotheses regarding the impact of negative emotion on referendum
vote choice, a number of binary logistic regressions were conducted with
vote choice as the dependent variable (voted Yes rather than voted No). Table 1
reports four models, beginning with a regression including all predictors except the
emotion variable, then a regression including the emotion variable, followed by
analyses in which respondents are split by emotion. In model 1 of Table 1, the
‘issues’ variables (EU ideology and perceptions of consequences of the treaty) and
the second-order factors (partisanship and level of satisfaction with the domestic
government) are entered, in addition to controls (age, social class and sex). The
variables predict voting in the expected manner. Pro-integration voters and voters
who perceive negative economic consequences resulting from treaty rejection tend
to vote Yes and Sinn Fe´in supporters and voters who are dissatisfied with the
government tend to vote No. All variables run from 0 to 1 and a comparison of
coefficients indicates that issue-related factors, and in particular perceptions of
treaty consequences, are stronger predictors than second-order factors. In model
2, the variable capturing the distinction between angry and anxious voters is also
entered and is highly statistically significant. Fearful voters are more likely than
angry voters to support the treaty. This simple model confirms H3, which posited
Garry 245
that the option that was characterised in the campaign as ‘risky’ (voting No) was
associated with anger while the option that was characterised as relatively ‘riskfree’
(voting Yes) was associated with anxiety.
A preliminary investigation of the hypothesised conditioning effects of anger
and fear is conducted by re-running model 1 but this time splitting the analysis by
emotion type. In the regression containing only angry voters (model 3), a strong
Table 1. Binary logistic regression predicting Yes (rather than No) vote.
Model 1 Model 2
Model 3
(angry voters)
Model 4
(fearful voters)
Favours EU integration 1.685*** 1.629*** 1.202** 2.292***
(.287) (.293) (.416) (.473)
Perceives positive economic
consequence from acceptance
of the treaty
7.396*** 7.027*** 6.089*** 9.190***
(.669) (.671) (.873) (1.228)
Dissatisfied with the performance
of the government
2.003*** 1.681** 3.045*** .051
(.536) (.548) (.767) (.904)
(Supporter of a pro-Treaty party¼ref)
Sinn Fe´in supporter 1.065* 1.131* 1.545* 1.351+
(.451) (.471) (.768) (.715)
Other response to party support question .510+ .387 .427 .294
(.310) (.321) (.425) (.519)
Under 35 years old .151 .252 .161 .299
(.305) (.316) (.423) (.504)
Female .157 .002 .056 .194
(.290) (.300) (.421) (.468)
(Professional class¼ref)
Manual class .473 .642* 1.080* .238
(.292) (.302) (.436) (.460)
Farmers .249 .256 .093 .339
(.672) (.691) (.921) (1.208)
Fearful 1.121***
(.297)
Constant 3.125*** 3.650*** 2.010+ 5.151***
(.619) (.650) (.797) (1.133)
n 724 724 321 403
Cox & Snell r-square .565 .573 .554 .444
Models 1 and 2 include all voters who indicated that they were either fearful or angry. ‘Favours EU integration’
is a dichotomous variable whereby 1¼favours integration and 0¼opposes integration; ‘Perceives positive
economic consequences of treaty acceptance’ is a 0 to 1 scale, with higher values representing greater
perception of positive consequences’; ‘Dissatisfied with the performance of the government’ is a 0 to 1
scale, with higher values indicating greater dissatisfaction with government performance.
Coefficients are log odds (SE).
Statistical significance level: + .1; * .05; ** .01; *** .001.
246 European Union Politics 15(2)
and highly statistically significant relationship emerges between dissatisfaction with
the government and voting No. In striking contrast, for anxious voters (model 4),
there is no relationship at all between being dissatisfied and voting No. In addition,
the coefficients relating to the issues variables (ideology and perceptions of treaty
consequences) are stronger for anxious voters than for angry voters. While these
preliminary findings are in line with my expectations, the effect of partisanship on
referendum vote choice appears quite similar across emotional type, suggesting no
emotional conditioning effect.6
In order to test the conditioning effects formally, a further binary logistic regression
is conducted in which ‘issues’ variables, second-order variables and control
variables are entered along with the emotion variable and interactions between the
emotion variable and each of the issues and second-order variables (see Table 2).
H1 is supported. Perceptions of the economic consequences of the treaty is a
Table 2. Binary logistic regression predicting Yes (rather than No) vote.
Favours EU integration 1.171** (.402)
Perceives positive economic consequences from acceptance of treaty 6.012*** (.850)
Dissatisfied with the performance of the government 3.033*** (.758)
(Supporter of a pro-Treaty party¼ref)
Sinn Fe´in supporter 1.524* (.738)
Other response to party support question .403 (.418)
Fear 3.298** (1.286)
Fear Favours EU integration 1.066+ (.616)
Fear Perceives positive economic consequence from acceptance of treaty 3.173* (1.471)
Fear Dissatisfied with the performance of the government 3.256** (1.163)
Fear Sinn Fe´in supporter .183 (1.004)
Fear Other response to party support question .147 (.666)
Under 35 years old .204 (.321)
Female .090 (.310)
(Professional class ¼ ref)
Manual class .692* (.311)
Farmers .173 (.730)
Constant 1.940* (.759)
N 724
Cox & Snell r-square .582
Included are all voters who indicated that they were either fearful or angry. ‘Favours EU integration’ is a
dichotomous variable whereby 1¼favours integration and 0¼opposes integration; ‘Perceives positive economic
consequences of treaty acceptance’ is a 0 to 1 scale, with higher values representing greater perception
of positive consequences’; ‘Dissatisfied with the performance of the government’ is a 0 to 1 scale, with higher
values indicating greater dissatisfaction with government performance.
Coefficients are log odds (SE).
Statistical significance level: + .1; * .05; ** .01; *** .001.
Garry 247
stronger predictor of vote choice for fearful people than for angry people and the
interaction (fear perceives positive economic consequence from acceptance of the
treaty) is statistically significant at the .05 level. Similarly, ideological disposition
towards the EU (favouring unity) is a stronger predictor of voting Yes for fearful
people than for angry people. However, the strength of the conditionality, consistent
with expectations, is weaker than was the case for treaty-specific perceptions:
the fear favours EU integration interaction is statistically significant at .1 level
(although unusual to report effects at .1 level, it is deliberately done here as there is
an expectation that there will be an interaction effect for ideology but it will be
weaker than that for perceptions of consequences).
The impact of dissatisfaction with the performance of the government on referendum
vote choice is stronger for angry voters than for anxious voters to a highly
statistically significant extent. However, partisan-based voting is not stronger for
angry voters than for anxious voters. Thus, H2 is partially supported.7
The substantive size of these statistically significant interaction effects is illustrated
in Figure 1.8 The graphs plot the difference, between angry and fearful
voters, of a change in the probability of voting Yes given a shift in the predictor
variable. Figure 1(a) shows a large difference between the two emotional groups in
terms of the effect of dissatisfaction on voting Yes. Moving from a position of one
standard deviation below the mean on the ‘dissatisfied’ scale to a position one
standard deviation above the mean implies no decrease in the probability of
voting Yes for anxious voters (essentially a flat line in Figure 1(a)), but does
imply a large decrease in the probability of voting Yes for angry voters (a decrease
from .73 to .30). An increase in voters’ positive perceptions of the consequences of
acceptance of the Treaty (from one standard deviation below the mean of the scale
to one standard deviation above the mean) results in an increase in the probability
of voting Yes for both angry and anxious voters (see Figure 1(b)), but the increase
is greater for the latter than for the former (an increase of .58 compared to an
increase of .45). Finally, Figure 1(c) illustrates that a movement from a sceptic
position to a pro-integration position is related to an increase in the probability of
voting Yes for both types of emotional voters, but the increase is greater for anxious
voters than for angry voters.
Discussion and conclusions
The EU, and more specifically the Euro-zone, is in a state of prolonged economic
crisis. Many commentators believe that the only way for the Euro to survive is for
the EU to radically change in order to achieve a dramatic and speedy financial,
economic and political integration that would serve to solidify the currency.
Significant moves in an integrative direction would necessitate referendums in
many states and such referendums would likely occur in relatively turbulent
crisis-like conditions. It seems plausible to expect that emotions such as anger
and anxiety would play an important role in such campaigns. The Irish case studied
in this paper can shed some light on how emotions may influence European
248 European Union Politics 15(2)
citizens’ reactions to referendum campaigns and how they vote. Emotions and
cognition are often depicted as opposites, with reason-based thinking and behaviour
compromised by emotional factors. Recent research, however, shows that
emotions govern or regulate cognitive resources and that affect and rationality
(a)
(b)
(c)
_______
———
Probability of
vong Yes
Probability of
vong Yes
anger
fear
0.2
0.4
0.6
0.8
1
less dissafied more dissasfied
0.2
0.4
0.6
0.8
1
less posive more posive
0.2
0.4
0.6
0.8
1
an-integraon pro-integraon
Probability of
vong Yes
Figure 1. (a) Impact on probability of voting Yes of moving one standard deviation below the
mean on the ‘dissatisfied with the government’ scale to one standard deviation above the
mean. (b) Impact on probability of voting Yes of moving one standard deviation below the
mean in the ‘positive economic consequences of treaty acceptance’ scale to one standard
deviation above the mean. (c) Impact on probability of voting Yes of moving from an anti-integration
position to a pro-integration position on the dichotomous integration measure.
Garry 249
work in tandem with each other to aid human decision making, opinion formation
and behaviour (Feldman et al., 2012). The moderating effect of different types of
negative emotions on the determinants of vote choice in EU referendums highlights
an important development of the second-order versus issues debate. Issue voting
(and particularly treaty-specific considerations) is associated with anxiety while
second-order voting (specifically, the impact of government evaluation) is associated
with anger. The central argument in this paper is that emotional heterogeneity
is a conditioning influence on EU referendum vote choice and should be
integrated into future research on EU referendum voting.
The results may also have implications for how we think about the campaign
strategies of parties and governments during referendums. Emotional rhetoric
employed by competing political actors may hinge on possible direct emotional
effects resulting from risk-related framing of referendum options. If actors frame
the option advocated by their competitors as the inherently risky option, this may
prove to be a successful strategic ploy to generate and attract anxious voters.
Emotional rhetoric may also have indirect effects. Actors who are confident that
reasonable voters would agree with them may be incentivised to use campaigning
devices that raise as much anxiety as possible in the minds of voters (who then
calmly and rationally deliberate and arrive at the ‘right’ answer). Other actors may
offset this and seek to tap into an existing well of discontent with the government
by deploying anger-related rhetoric that seeks to highlight the possibility of blame
related voting. Hence, the role of emotions in referendum campaigning may be
deployed (indirectly) to raise (or lower) the salience of different drivers of referendum
vote choice.9
More generally, this study contributes to the emerging field of research on emotional
bases of political beliefs and behaviours (see Feldman et al., 2012 for overview).
The analysis in this article supports recent work that advocates investigating
the distinct implications of different positive and negative emotions, in contrast to
earlier work on emotion that focused on the negative versus positive emotion distinction.
In particular, this study complements previous research that argues that
two specific negative emotions (anger and anxiety) have important distinct implications.
The hypotheses tested here are not drawn from a single theoretical perspective
but are rather derived from work from a neuropsychological perspective,
an evolutionary psychology perspective and a risk-oriented perspective. What these
approaches share, however, is the assumption that anger and anxiety have different
implications for political behaviour. This assumption is empirically supported in
the application to the specific case of EU referendum voting behaviour studied in
this article. Further research, focusing on more cases and more time points, may
shed light on how generalisable these findings are.
Acknowledgements
I would like to thank Josie Knowles for commenting upon an earlier draft of this paper and
would also like to thank the editor and the anonymous reviewers. I would like to acknowledge
that the data used in this paper is part of a wider project with Michael Marsh.
250 European Union Politics 15(2)
Notes
1. The intergovernmental treaty is formally entitled the ‘Treaty on Stability, Coordination
and Governance in the Economic and Monetary Union’ and is typically referred to more
informally as the European Fiscal Compact or European Stability Treaty.
2. Thus, insofar as anger enhances second-order effects, there are two distinct possible roles
it could play: strengthening the effect of either partisan factors or factors relating to
(dis)satisfaction with the government. The aim here is simply to attempt to identify
empirically which particular moderating role, if any, anger plays.
3. RedC conducted eight representative pre-referendum polls that contain both vote intention
at the Fiscal Stability treaty and party vote intentions in the event of a general
election. In each of these surveys, respondents were asked if they were happy to be recontacted
for research by RedC in the future and approximately 65–70 percent of
respondents in the surveys stated that they were. This provided a pool of approximately
5500 people that could be followed up post the referendum. Demographic quotas were set
to ensure a nationally representative sample of voters: the quotas were based on responses
to the question ‘how likely are you to vote in the referendum?’ asked in the last two prereferendum
polls (to ensure the final sample was representative of those who actually
voted, rather than all potential voters). From the pool of 5500 people, 1000 interviews
with people who voted at the Fiscal Stability referendum were completed in the days
following the referendum vote.
4. It may possibly be regarded as a concern that the operationalisation of the party cues
hypothesis essentially compares Sinn Fe´in to the main Irish parties, with the potential
that what is being captured is the peculiarities of Sinn Fe´in rather than party cueing
effects per se. However, the peculiarities of Sinn Fe´in (which arguably currently relate to
its position as a working class protest party) are controlled for via the inclusion of social
class and level of government satisfaction, with the consequence that the ‘party’ variable
likely captures, in large part, party cueing effects.
5. Respondents were not explicitly offered further response categories. In addition to
respondents who stated that they were either angry (33.1%, n¼331) or fearful (42.2%,
n¼422), 22.1% of respondents responded that they were ‘neither’ angry nor fearful. A
small number of respondents (2%) also indicated that they were ‘both’ (and two respondents
were coded as ‘don’t know’). One concern may be that because respondents were not
explicitly offered the ‘neither’ or ‘both’ categories, respondents may have felt forced to
choose either ‘anger’ or ‘fear’, hence undermining the meaningfulness of the responses. A
range of analyses were conducted to assess the ‘neither’ respondents and it emerges that, in
line with expectations, those who chose ‘anger’ or ‘fear’ were the most distinct from each
other and ‘neither’ respondents seemingly situated between the two groups (see below).
6. A separate regression analysis was conducted including only those respondents who
indicated that they were ‘neither’ angry nor fearful (see Online Appendix A). In that
regression, the coefficients for the two issue variables (underlying attitude to EU unity
and perceptions of the economic consequences of the treaty) lie between those for the
‘angry’ and ‘fearful’ groups (compare Online Appendix A to models 3 and 4 in Table 1).
The same is the case for the dissatisfaction with government variable. This suggests that
the ‘neither’ respondents substantively lie between the ‘angry’ and ‘fearful’ respondents.
7. Further analysis was conducted using a different operationalisation of emotion. ‘Angry
versus not angry’ and ‘fearful versus not fearful’ categories were generated. (‘Not angry’
is made up of those who responded that they were ‘fearful’ plus those who said they were
Garry 251
‘neither’ angry nor fearful; similarly for ‘not fearful’). These were then used as interactions
(see results in Online Appendices B and C). Focusing on the two clear substantive
findings of the ‘angry’ versus ‘fearful’ analysis, it emerges that the results are substantively
the same in these new categorisations, although the statistical significance level falls
just outside the .05 cut-off for two of the effects (p¼.055 for the interaction between
‘fearful-not fearful’ and ‘perceptions of economic consequences’ and p¼.064 for the
interaction between angry-not angry and ‘perceptions of economic consequences).
Hence, the direct contrast between anger and anxiety produces similar but somewhat
stronger results than the ‘angry versus not angry’ and ‘fearful versus not fearful’ distinctions.
This is perhaps not surprising given that the literature summarised earlier emphasises
the direct contrast of these two negative emotions.
8. For Figure 1(a): assuming a voter who has a mean score on the perceptions of treaty
consequences measure and who holds a sceptical view on EU integration. For
Figure 1(b): assuming a voter who has a mean score on the dissatisfaction with government
measure and who holds a sceptical view on EU integration. For Figure 1(c): assuming
a voter who has mean scores on the perceptions of treaty consequences measure and
the dissatisfaction with government measure. In all three analyses, the voter is assumed to
be a supporter of a pro-Treaty party and is older, male and middle class.
9. This potentially raises some endogeneity concerns when modelling vote choice. If it is
political parties (via party strategists) that tailor a particular campaign such that certain
emotional reactions of specific groups are successfully triggered, researchers need to
carefully specify the role of ‘party’ and the role of ‘emotion’ in shaping voting behaviour.
References
Anderson C and Reichert S (1996) Economic benefits and support for membership in the
EU: A cross-national analysis. Journal of Public Policy 15(3): 231–249.
Bogdan A (2012) The anxious partisan: A test of affective intelligence theory in
RomaniaAvailable at: http://www.etd.ceu.hu/2012/bogdan_alexandra.pdf (accessed 4
October 2012) (2012, ).
Brader T (2011) The political relevance of emotions: ‘Re-assessing’ revisited. Political
Psychology 32(2): 327–346.
Campbell A, Converse PE, Miller WE, et al. (1960) The American Voter. New York: Wiley.
Carey S (2002) Undivided loyalties: Is national identity an obstacle to European integration?
European Union Politics 3(4): 387–413.
Davis M (1992) The role of the amygdala in fear and anxiety. Annual Reviews of
Neuroscience 15: 353–375.
de Vreese C and Boomgaarden H (2005) Projecting EU referendums: Fear of immigration
and support for European integration. European Union Politics 6(1): 59–82.
de Vreese C and Semetko H (2004) News matters: Influences on the vote in the Danish 2000
Euro referendum campaign. European Journal of Political Research 43(5): 699–722.
Druckman JN and McDermott R (2008) Emotion and the framing of risky choice. Political
Behaviour 30(3): 297–321.
Eichenberg R and Dalton R (2007) Post-Maastricht blues: The transformation of citizen
support for European integration, 1973–2004. Acta Politica 42(1–2): 128–152.
Feldman S, Huddy L and Cassesse E (2012) Emotional underpinnings of political behaviour.
In: Sun R (ed.) Grounding Social Sciences in Cognitive Sciences. Cambridge MA: MIT
Press, pp. 127–156.
252 European Union Politics 15(2)
Fitzgibbon J (2013) Referendum briefing: The referendum on the intergovernmental treaty
on stability, coordination and governance in the the economic and monetary union in
Ireland, 31 May 2012. Representation 49(2): 229–239.
Franklin M, Marsh M and McLaren L (1994) Uncorking the bottle: Popular opposition to
European unification in the wake of Maastricht. Journal of Common Market Studies
32(4): 455–472.
Franklin M, van der Eijk C and Marsh M (1995) Referendum outcomes and trust in government:
Public support for Europe in the wake of Maastricht. West European Politics
18(3): 101–117.
Franklin M (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.
Gabel M (1998a) Public support for European integration: An empirical test of five theories.
Journal of Politics 60(2): 333–354.
Gabel M (1998b) Interests and Integration: Market Liberalization, Public Opinion, and
European Union. Ann Arbor, MI: University of Michigan Press.
Garry J (2013) Direct democracy and regional integration: Citizens’ perceptions of treaty implications
and the Irish reversal on Lisbon. European Journal of Political Research 52(1): 94–118.
Glencross A and Trechsel A (2011) First or second order referendums? Understanding the
votes on the EU Constitutional Treaty in four EU member states. West European Politics
34(4): 755–772.
Hix S and Marsh M (2011) Second-order effects plus pan-European swings: An analysis of
European Parliament elections across time. Electoral Studies 30(1): 4–15.
Hobolt S (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 (2006) How parties affect vote choice in European integration referendums. Party
Politics 12(5): 623–647.
Hobolt S (2007) Taking cues on Europe? Voter competence and party endorsement in referendums
on EU integration. European Journal of Political Research 46(2): 151–182.
Hobolt S (2009) Europe in Question – Referendums on European Integration. Oxford: Oxford
University Press.
Huddy L, Feldman S and Cassese E (2007) On the distinct political effects of anger and
anxiety. In: Neumann R, Marcus G, Crigler A, et al. (eds) The Affect Effect. Cambridge:
Cambridge University Press, pp. 202–230.
Inglehart R and Klingemann HD (1976) Partisan identification, ideological preference and
the left-right dimension among western mass publics. In: Budge I, Crewe I and Fairlie D
(eds) Party Identification and Beyond. London: Wiley, pp. 243–273.
Irish Times (2012) The great struggle was between fear and anger – and fear has won.
Irish Times 2 June. Available at: http://www.irishtimes.com/news/the-great-strugglewas-between-fear-and-anger-and-fear-has-won-1.1063363
(accessed 8 October 2012).
Kiss Z and Hobolt S (2012) Negative campaigning, emotion and political participation. In:
EJECDEM Final Conference, Florence, June 28–30, 2012.
Kiss Z and Hobolt S (2011) The emotional voter: An experimental study of the moderating
effect of emotions on partisan behaviour. In: 6th ECPR General Conference, Reykjavik,
August 25–27, 2011.
Knutsen O (1997) The partisanship and the value-based component of left-right self-placement:
A comparative perspective. International Political Science Review 18(2): 191–225.
Ladd JM and Lenz GS (2008) Reassessing the role of anxiety in vote choice. Political
Psychology 29(2): 275–296.
Garry 253
Ladd JM and Lenz GS (2011) Does anxiety improve voters’ decision making? Political
Psychology 32(2): 347–361.
Lazarus RS (1991) Emotion and Adaptation. Oxford: Oxford University Press.
Lerner J, Gonzalez R, Small D and Fischoff B (2003) Effects of fear and anger on perceived
risk of terrorism: A national field experiment. Psychological Science 14(2): 144–150.
MacKuen M, Wolak J, Keele L and Marcus GE (2010) Civic engagements: Resolute partisanship
or reflective deliberation. American Journal of Political Science 54(2): 440–458.
Marcus GE (2002) The Sentimental Citizen: Emotion in Democratic Politics. University Park,
PA: Pennsylvania State University Press.
Marcus GE, MacKuen M and Neuman WR (2011) Parsimony and complexity: Developing
and testing theories of affective intelligence. Political Psychology 32(2): 323–326.
Marcus GE and MacKuen MB (1993) Anxiety, enthusiasm and the vote: The emotional
underpinnings of learning and involvement during presidential campaigns. American
Political Science Review 87(3): 672–685.
Marcus GE, Neuman WR and MacKuen M (2000) Affective Intelligence and Political
Judgement. Chicago: University of Chicago Press.
Marcus GE, MacKuen M, Wolak J, et al. (2006) The measure and mismeasure of emotion.
In: Redlawsk DP (ed.) Feeling Politics: Emotion in Political Information Processing.
New York: Palgrave Macmillan, pp. 31–47.
McLaren L (2002) Public support for the European Union: Cost/benefit analysis or
perceived cultural threat? Journal of Politics 64(2): 551–566.
McLaren L (2004) Opposition to European integration and fear of loss of national identity:
Debunking a basic assumption regarding hostility to the integration project. European
Journal of Political Research 43(6): 895–912.
Moravcsik A (2006) What can we learn from the collapse of the European Constitutional
Project? Politische Vierteljahresschrift 47(2): 219–241.
Petersen MB (2010) Distinct emotions, distinct domains: Anger, anxiety and perceptions of
intentionality. Journal of Politics 72(2): 357–365.
Reif K and Schmitt H (1980) Nine second order national elections. European Journal of
Political Research 8(1): 3–44.
Richardson B (1991) European party loyalties revisited. American Political Science Review
85: 751–775.
Rhudi JL and Meagher MW (2000) Fear and anxiety: Divergent effects on human pain
thresholds. Pain 84(1): 65–75.
Siune K and 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 and Tonsgaard O (1994) The EU: The Danes said ‘‘No’’ in 1992 and
‘‘Yes’’ in 1993 – How and why? Electoral Studies 13(2): 107–116.
Smith C and Ellsworth P (1985) Patterns of cognitive appraisal in emotion. Journal of
Personality and Social Psychology 48(4): 813–838.
Svensson P (1994) The Danish Yes to Maastricht and Edinburgh: The EC referendum of
May 1993. Scandinavian Political Studies 17(1): 69–82.
Svensson P (2002) Five Danish referendums on the European Community and European
Union: A critical assessment of the Franklin thesis. European Journal of Political
Research 41(6): 733–750.
Valentino NA, Hutchings L, Banks AJ, et al. (2008) Is a worried citizen a good citizen?
Emotions, political information seeking, and learning via the internet. Political
Psychology 29(2): 247–273.
254 European Union Politics 15(2)
Political knowledge and campaign effects in the 2008 Irish
referendum on the Lisbon Treaty
Johan A. Elkink*
, Richard Sinnott
University College Dublin, Belfield, Dublin 4, Ireland
article info
Article history:
Available online 25 February 2015
Keywords:
Referendums
Political knowledge
European integration
Campaigns
Lisbon Treaty
Ireland
abstract
This article makes a distinction between the attitude component of campaigns and the
knowledge component and argues that a campaign that influences knowledge of a proposal
can be quite successful in influencing the vote. On June 12th 2008 Irish voters voted
against ratification of the Lisbon Treaty. Perceptions, and especially misperceptions, of the
Treaty played an important role in the referendum. The campaign focused in particular on
influencing voters’ perceptions of, rather than attitudes towards the Treaty. This article
examines the interaction between knowledge, campaigns, and perceptions of the Treaty, in
the referendum.
© 2015 Elsevier Ltd. All rights reserved.
1. Introduction
The nature of European Union referendums, as distinct
from general elections and from clearly single issue referendums,
provides valuable material with which to further
our understanding of how voters make their decisions on
whether and how to vote. EU referendums are also typically
quite consequential, making it important not only to
understand them from a scientific, but also from a policymaking
point of view.
Voting behaviour in EU referendums is generally
understood either in terms of satisfaction or dissatisfaction
with the government of the day e the so-called secondorder
perspective (Reif and Schmitt, 1980; Franklin et al.,
1994; Franklin, 2002); or in terms of the attitudes of
voters towards European integration (Garry et al., 2005)
and towards European policies (Hix, 2006; Laffan, 2008;
Gamble, 2006); or in terms of the utilitarian benefits one
expects to obtain from European integration (Gabel, 1998;
Ehin, 2001; Van Apeldoorn, 2009); or, finally, in terms of
the effect of knowledge of the issue at hand or of the
European Union in general (Binzer Hobolt, 2005). Previous
studies of the Irish referendums suggest that political
knowledge was crucial in determining the outcome of the
referendums on the Nice and Lisbon treaties (Sinnott, 2001,
2003; Sinnott et al., 2009).
This article provides an in-depth examination of the role
of political knowledge in the 2008 Irish referendum on the
Treaty of Lisbon. It argues that a central feature of the
campaign was the emphasis on perceptions of the contents
of the Treaty, rather than evaluations of these contents.
Although these two components of a referendum campaign
are inseparably linked, they are nevertheless conceptually
distinct.
Referendum campaigns can be distinguished by the
extent to which they emphasise either the knowledge or
the attitude component. On the basis of a statistical analysis
of the results from an opinion survey conducted shortly
after the referendum it is apparent that knowledge played a
crucial role (Sinnott et al., 2009). This research shows that
perceptions of the contents of the Treaty can be separated
into two groups, one corresponding to issues emphasised
by the NO campaign and one to issues emphasised by the
YES campaign. The alignment of what voters perceive to be
the contents of the Treaty with either the NO or the YES
campaign is shown to be an important factor in accounting
* Corresponding author.
E-mail addresses: jos.elkink@ucd.ie (J.A. Elkink), richard.sinnott@ucd.ie
(R. Sinnott).
Contents lists available at ScienceDirect
Electoral Studies
journal homepage: www.elsevier.com/locate/electstud
http://dx.doi.org/10.1016/j.electstud.2015.02.003
0261-3794/© 2015 Elsevier Ltd. All rights reserved.
Electoral Studies 38 (2015) 217e225
for the referendum result, as opposed to the overall level of
knowledge of the Treaty contents or the negative or positive
evaluation of these Treaty contents.
The next section of this article briefly describes the 2008
referendum campaign, thus providing the context in which
the empirical analysis is pursued. Section 3 sets out the
causal mechanisms that we believe connect political
knowledge to voting behaviour. Section 4 introduces the
survey data and the measurement of the key variables.
Section 5 tests the causal claims in a series of regression
analyses. Section 6 summarises the findings and draws some
tentative conclusions. The appendices contain definitions of
the variables and further details on the empirical analysis.
2. The Irish context
The two sides of the campaign in the 2008 referendum
were of very different composition. Whereas the YES
campaign was forced to defend the Treaty in its entirety
and had to have an answer to every possible criticism of the
Treaty, each different actor in the NO campaign could
simply focus on a specific aspect of the Treaty. Quinlan
(2009: 109) succinctly describes the NO campaign:
“As in previous European referendums, the no side was
a diverse coalition[, including] … Sinn Fein [as] the only 
party in the Oireachtas to call for a no vote …, the Peace
and Neutrality Alliance (PANA) … the People’s
Campaign was a broad coalition which included [several
MEPs] … Meanwhile, sustained opposition to the Treaty
on the right was provided by Coir  … But the most vocal
opposition to the Treaty came from the think-tank Libertas.
Founded by Galway businessman Declan Ganley,
Libertas opposed Lisbon because in its view it did not
provide a transparent democratic Europe, weakened
Irish power within the EU and opened the way for Irish
corporate tax rates to be interfered with by Brussels.”
The YES campaign on the other hand consisted of almost
all major political parties (with the Green Party formally
staying neutral), the Irish Alliance for Europe and a number
of business interest groups (Quinlan, 2009: 108e109). In
addition to the protagonists on either side, there was the
role of the Referendum Commission, which was established
by statute to “explain the subject matter to the
public, … promote awareness of the referendum and …
encourage people to vote” (Quinlan, 2009: 109).
While this is at first sight a neutral assignment, in a
referendum in which lack of knowledge of the Treaty
contents is an explicit issue, the task of increasing
“awareness” easily becomes indistinguishable from the
political campaigns.
While the YES campaign was emphasising the increased
efficiency and improved decision-making in the European
Union to be brought about by the Lisbon Treaty, the NO
campaign actively attempted to influence specific perceptions
of the Treaty, often by emphasising items that were
not actually present in the Treaty text, but that are important
to Irish voters. For example, one section of the NO
campaign emphasised the risk that the Lisbon Treaty would
lead to the establishment of a European army with
conscription for Irish citizens (Quinlan, 2009: 111).
3. Knowledge and referendum voting behaviour
Knowledge played a crucial role in determining the
outcome of the Lisbon Treaty referendum. This is clear from
the subjective evaluations of a nationally representative
sample of voters. When NO voters in a post-referendum
poll were asked why they voted NO, 46% of them
mentioned something related to lack of information,
knowledge, or understanding (Sinnott et al., 2009: 13).
Subsequent statistical analyses of the survey data using
appropriate control variables confirms that political
knowledge was a crucial factor in determining the referendum
outcome (Sinnott et al., 2009), which would not
surprise even a casual observer of the campaign.
The idea that knowledge affects voting behaviour is
hardly new (Carpini and Keeter, 1996; see also Bowler, this
issue). In the 2008 Irish referendum, however, the role of
knowledge was a complicated one, involving campaign
effects, perceptions and misperceptions of the contents of
the Treaty, and subjective evaluations of the level of
knowledge of the Treaty. Fig. 1 provides a graphical
depiction of our key causal argument. The difference between
subjective knowledge d the extent to which a voter
feels informed about the issue at hand in the referendum d
and objective knowledge d the actual level of knowledge
of the issue at hand, or of the European Union more
generally d will be a crucial distinction in what follows.
The primary mechanism through which subjective
knowledge can be expected to have an impact on vote
choice (Arrow 4) is through risk aversion. The more uncertain
a respondent feels about the impact of voting for a
referendum proposal, the more likely the voter is to support
the status quo and to vote against the proposal (Suiter
and Reidy, 2013). Binzer Hobolt (2009: 40e57) provides a
formal model elaborating on this idea. In this argument, the
level of objective knowledge is of less importance;
although it can of course be assumed that objectively less
knowledgeable voters are also likely to feel less knowledgeable
(Arrow 3). A feeling of a lack of knowledge was
widespread in the 2008 referendum. Indeed some politicians
exacerbated this feeling by making statements such
as the admission by Taoiseach (prime minister) Brian
Cowen on television that he had not read the Treaty “cover
to cover” or the remark by Irish EU Commissioner Charlie
Fig. 1. Causal model examining the effect of knowledge on the vote (YES or
NO). Objective knowledge is here conceptualised as perceptions that are
correct, hence is part of the perceptions variable in this diagram.
218 J.A. Elkink, R. Sinnott / Electoral Studies 38 (2015) 217e225
McCreevy a week later that any voter who read the Treaty
was “insane” (Quinlan, 2009: 113).
The impact of objective knowledge is likely to be more
indirect, as there is no reason a priori to assume that higher
levels of knowledge lead to stronger support for European
integration. Furthermore, lack of knowledge of politics in
general, of the European Union, or of the Lisbon Treaty does
not necessarily render the voter incapable of casting a
rational choice: “if there are multiple informational pathways
to a competent vote (e.g., if interest-group endorsements,
effective heuristics, etc., lead a voter to cast the
same vote as she would have if better informed), then
voters need not use the same information that political
elites use” (Lupia, 2006: 222; Bowler, this issue).
This reliance on informed sources to make a decision
does depend on the trustworthiness of the source. In other
words, the information can be either correct or incorrect
and therefore such heuristics contain a risk for the voter d
the interests of the sender of information might be in
conflict with those of the receiver (Lupia and McCubbins,
1998). An objectively less knowledgeable voter is thus
relying more on campaign messages from trusted sources
than a more knowledgeable voter, and is more susceptible
to potentially misleading campaign messages about the
referendum issue at hand.
Less knowledgeable voters are therefore more dependent
on high quality information, but are likely to in fact
rely on more easily accessible sources (Bowler and
Donovan, 1994). Hooghe and Marks (2009: 13) make a
similar argument in particular in the context of public
opinion on Europe, which they argue “is particularly susceptible
to construction: i.e. priming (making a consideration
salient), framing (connecting a particular
consideration to a political object) and cueing (instilling a
bias)” (see also, Druckman, 2004; Druckman and Nelson,
2003; Miller and Krosnick, 2000).
The finding that knowledge is important, even when
controlling for well-known effects such as the popularity of
the government of the day or attitudes towards the European
Union in general, has implications for the kind of
campaigns that could be effective in a referendum. If
knowledge matters, campaigns can either focus on changing
the knowledge or perceptions voters have of the issue
at hand or they can concentrate on the attitudes these
voters have towards what they perceive to be the issue. We
are thus making a distinction between the attitude and the
knowledge components of campaigns.1
An obvious criticism of this distinction is that these two
components can never be observed entirely independently.
Any campaign that emphasises how, for example, corporate
taxes are part of the Lisbon Treaty, also implies that this is a
negative thing. In other words, such a campaign influences
perceptions and attitudes simultaneously. It can be argued
that these are still distinct, if generally coinciding, components.
The point becomes clearer when we compare the
Irish referendum on the Lisbon Treaty with the Irish referendum
on abortion in March 2002. In a referendum on
abortion, the issue at hand is clear to voters. Campaigns
focus on influencing voters’ evaluations of whether allowing
abortion is a good or a bad thing, so they attempt to
influence attitudes rather than perceptions. In the referendum
on the Lisbon Treaty, however, the emphasis was
very much on influencing perceptions of what was in the
Treaty rather than the evaluations of these. For example,
the NO campaign would argue that the Treaty could lead to
an increase in Irish corporate tax rates. On both sides of the
campaign, many voters would oppose such an increase, but
it is the perception that this is part of the referendum decision
that the campaign tried to influence, rather than
whether such increase would be a good or bad thing for
Ireland. It was implied that it is a bad thing, but that was
not the contentious issue.
Since ‘perceptions’ and ‘attitudes’ are often confused, it
is crucial that we are specific about what we mean by
these terms. According to the expectancy-value approach
in social psychology, attitudes are “a multiplicative combination
of (a) strength of beliefs that an object has certain
attributes and (b) evaluations of these attributes” (Perloff,
2003: 46; see also Ajzen and Fishbein, 1977). In this
article we will refer to ‘perceptions’ or ‘knowledge’ when
talking about the beliefs that an object, in this case the
Treaty, has certain attributes. In short, whereas ‘perceptions’
refers to what the voters think, rightly or wrongly, is
contained in the Treaty, ‘knowledge’ refers to the extent to
which these perceptions were correct. We refer to ‘attitudes’
when talking about the evaluations of these
perceived attributes.
An example would be the campaign item related to the
possibility of an increase in the Irish corporate tax rates. We
make a clear distinction between the perception that the
Lisbon Treaty does indeed imply future increases in tax
rates and the evaluating attitude the respondent has towards
the idea of such increase.2 There is an extensive
literature in the US context outlining the conditional and
nonlinear nature of issue voting (e.g., Grynaviski and
Corrigan, 2006) and the impact of political knowledge
therein (Campbell et al., 1960; Nie et al., 1979; Palfrey and
Poole, 1987; Krosnick, 1988; Zaller, 1992; Alvarez, 1997;
Basinger and Lavine, 2005; De Vries et al., 2011).
4. Knowledge and the campaign messages
The main data for the following empirical analysis is a
national opinion poll conducted by Millward Brown IMS on
behalf of the Department of Foreign Affairs of the Republic
of Ireland. Data were collected from 2101 respondents aged
18 years and older. All respondents were Irish citizens listed
on the electoral register. The poll was carried out between
24 and 31 July 2008, slightly over a month after the referendum
had taken place. Details and bivariate analyses can
be found in IMS (2008) and a more extensive multivariate
1 Or framing and cueing, respectively, in the language of political
studies.
2 A somewhat complicating factor is that at times these perceptions
might be influenced by attitudes. A study by Lord et al. demonstrates how
subjects interpret neutral or contradicting evidence in such a way that it
confirms their prior beliefs. They adjust their evaluation of the quality,
biasedness, logic, and conclusions all on the basis of their prior attitude
towards the object under study (Lord et al., 1979).
J.A. Elkink, R. Sinnott / Electoral Studies 38 (2015) 217e225 219
analysis in Sinnott et al. (2009). The appendix provides the
precise coding of all variables used.3
Table 1 provides an overview of key Treaty items
emphasised by the two campaigns and translated into a
survey question. Although most key campaign items have
been included in the question, it is not an exhaustive list of
issues raised during the campaign. The respondents were
asked which of the items were part of the Treaty with the
option to select the ‘don’t know’ answer category in each
case. The responses to this question are presented in the
first two columns of the table. These answers can thus be
seen as the perceptions of the Treaty contents. For most
items it is quite clear whether these were or were not part
of the Treaty. The third column lists the correct answers.
The final column lists the items that are indeed in the
Treaty and those that are not. As we argue here, the
campaign focused somewhat less on influencing the attitude
towards particular items, but rather on influencing the
perception that particular items were or were not part of
the Treaty. An evaluation, either positive or negative, was
generally only implied. The final column lists the particular
items that the campaigns suggested were part of the Treaty.
Although the evaluation of the items differs significantly,
both campaigns admitted clearly that the loss of an Irish
commissioner was indeed part of the Treaty.4
We argue that an important part of the effect of objective
knowledge is the susceptibility of respondents with
low levels of knowledge to the (potentially misleading)
campaign messages of the different campaigns. That for a
large proportion of the electorate, campaigns can influence
the vote in the referendum not only by promoting a
particular evaluation of the issue to be voted on, but also
particular perceptions of the issue itself, of what it is one is
voting on. For the statistical evaluation of this proposition,
we measure the alignment of respondents with particular
campaign messages through additive scales. Both the
measures of objective knowledge of the Treaty and the
measures of campaign effects are based on the data
described in Table 1. The objective knowledge scale reflects
the number of correct answers on items 1e5, 7, 8, 10, and
11. The NO campaign scale counts the number of YES answers
on items 2e6 and the YES campaign scale counts the
number of YES answers on items 7e11.5 The two scales on
the campaigns do not summarise the campaigns as such d
they summarise how the campaigns provided different
perceptions of the contents of the Treaty.
Cronbach’s alpha can be used to assess whether these
are indeed reasonable scales to use as measures in subsequent
regression analysis. We obtain for the knowledge
scale a ¼ 0.74; for the NO campaign items a ¼ 0.76 and for
the YES campaign items a ¼ 0.71. Typically, an alpha of
approximately 0.6 is seen as the threshold to accept a scale
as sufficiently additive e the items in the scale are suffi-
ciently correlated to consider them indicators of the same
underlying concept.
We use these additive scales in the regression analysis in
the next section to assess the relative explanatory power of
objective knowledge of the Treaty contents e the truth
value of the perceptions of the Treaty e and of the alignment
of the perceptions of the Treaty with the messages of
the two campaigns. An important caveat is necessary,
however, namely that there is the potential for endogeneity
in this model. While we take the alignment of voters with
particular campaign messages to be an explanatory variable
for their vote choice, it is possible that for some voters
the decision to vote YES or NO was made early in the
campaign and that this has lead them to follow a particular
campaign messages more closely, which lead to them
aligning their perceptions with the campaign messages.
Both in terms of the information they pay attention to and
in line with cognitive dissonance theory, such reverse
causality is a genuine possibility.6
5. Explaining vote choice in the referendum
We are primarily interested in the role of knowledge in
the referendum. As is clear from Fig. 2, the level of objective
knowledge of the contents of the Treaty has a substantial
effect on the individual’s voting behaviour. This figure
Table 1
Perceptions of the contents of the Lisbon Treaty. The question is: “Which
of the following do you think are included in the Lisbon Treaty?”.
% Yes % No Correct Campaign
1. Loss of Irish Commissioner
for 5 out of every 15 years
65 9 Correct Ambiguous
2. Ending of Ireland’s right to
decide its own corporate
tax rate
43 22 Incorrect No campaign
3. The introduction of
conscription to a
European army
33 37 Incorrect No campaign
4. The reduction of Ireland’s
voting strength in the
Council of Ministers
48 18 Correct No campaign
5. The end of Ireland’s
control over its
policy on abortion
34 33 Incorrect No campaign
6. The erosion of Irish
neutrality
42 30 Ambiguous No campaign
7. Improved efficiency of EU
decision-making
56 15 Correct Yes campaign
8. Strengthening Europe’s
role in the world
61 12 Correct Yes campaign
9. Improved protection of
workers’ rights
49 19 Ambiguous Yes campaign
10. Strengthening the role
ofnational parliaments
in EU decision-making
43 20 Correct Yes campaign
11. The Charter of
Fundamental Rights
36 14 Correct Yes campaign
3 The questionnaire was designed by a working group comprising
officials from the Irish Department of Foreign Affairs and researchers
from the survey agency (Millward Brown-IMS). Professor Richard Sinnott
acted as technical advisor to the working group.
4 This despite the fact that even under the already adopted Nice Treaty,
Ireland will lose its Commissioner, albeit later.
5 Note that for the knowledge scale, the ‘don’t know’ category is taken
as someone who lacks knowledge, so it is combined with the incorrect
answer, while for the campaign scales, the ‘don’t know’ category is
treated as missing data.
6 We thank the anonymous reviewer for this excellent point.
220 J.A. Elkink, R. Sinnott / Electoral Studies 38 (2015) 217e225
presents the predicted values of the probability of a NO
vote for a particular configuration of the independent variables
(see the caption). The predictions are based on the
logistic regression reported as Model 3 in Table 2 in the
appendix, which includes a large number of control variables,
the measure of knowledge as the number of correct
perceptions of the contents of the Treaty, and subjective
knowledge, but which excludes the two separate
dimensions discussed above. Whereas a knowledgeable
voter who is dissatisfied with the government has slightly
less than a fifty per cent chance of voting NO, a similar voter
with much less knowledge of the Treaty contents has a
probability of voting NO of about ninety per cent. The effect
is strong for both people who are satisfied with the
government and people who are not. While the secondorder
effects of closeness to the opposition and dissatisfaction
with the government are important factors in
explaining the vote, political knowledge played a more
influential role in this referendum.
As suggested in the theoretical discussion above, the
story is somewhat more complicated than the results in
Fig. 2 suggest. The causal diagram in Fig. 1 functions as a
useful guide for a more nuanced interpretation of the
regression results, which are graphically presented in
Fig. 3.
7 Two separate models were estimated to take account
of the different causal paths to be evaluated. When
estimating a causal effect, it is generally advised to control
for any potential confounding factors, but not to control for
any mediating variables (Pearl, 2000; Morgan and Winship,
2007). As can be easily determined on the basis of the
causal diagram, to estimate the effect of objective knowledge
on the vote, we do not want to control for subjective
knowledge, but to estimate the effect of subjective
knowledge, we do want to control for the level of objective
knowledge. For exactly this reason, the two models were
estimated separately. Although Fig. 3 only reports the coefficients
of interest to the theoretical discussion in this
paper, a large set of control variables were included in the
model. Numerical estimates of all variables are reported in
Table 2 in the appendix. Since all variables have been
standardised, we can directly compare the magnitudes of
the effects of each variable.8
It is clear from these estimates that subjective knowledge
of the Treaty has a significant effect on the vote. Objective
knowledge also has a significant effect on the probability of
voting NO, but once we control for subjective knowledge,
this effect is reduced and is statistically insignificant. This
suggests that the impact of objective knowledge is indeed
02468
0.0 0.2 0.4 0.6 0.8 1.0
Low objective knowledge of the Lisbon treaty
Probability of NO vote
Satisfied
Dissatisfied
Fig. 2. Effect of objective knowledge of the Lisbon Treaty and dissatisfaction
with the government on the NO vote. Estimates are based on Model 3 of Table
2 in the appendix. The grey shaded area is a histogram of the objective
knowledge measure. The lines represent the median voter, which is an
hypothetical lower middle class female, between 35 and 49 years old, with
secondary education, who supports an opposition party, had a median score
on the anti-immigration and Irish neutrality scales, who does not consider her
own economic situation to be bad and who does not identify as “Irish only”.
Fig. 3. Absolute values of logistic regression coefficients and their standard
errors, based on models 5 (solid lines) and 6 (dashed lines) in Table 2. The
models contain the same control variables (not shown) and differ only in
terms of the inclusion of the two subjective knowledge variables. Horizontal
lines reflect the 95% confidence intervals, while the vertical markers reflect
the 90% confidence intervals. The full regression table can be found in
Table 2 in the appendix.
7 See Kastellec and Leoni (2007) for an argument in favour of
presenting regression results in this manner.
8 A logistic regression can be perceived of as a linear relation of the
independent variables with a latent dependent variable, whereby the
probability of a 1 on the dependent variable is a non-linear function of
the latent variable (King, 1998). The reported coefficients reflect in all
cases the effect of an increase of two standard deviations (Gelman, 2007)
in the independent variable on the latent dependent variable.
J.A. Elkink, R. Sinnott / Electoral Studies 38 (2015) 217e225 221
through the level of subjective knowledge (Arrow 3 in the
causal diagram in Fig. 1) d the awareness of the voter of his
or her lack of knowledge d and the two dimensions of the
information from the campaign (Arrow 6).
The alignment of the perceptions of the contents of the
Treaty with that of either side of the campaign has a highly
significant effect on the NO vote, even stronger than that of
objective and subjective knowledge variables.9 The more
the voter’s perceptions of the contents of Treaty are in line
with the messages of the NO campaign, the more likely this
voter is to vote NO; the more the perceptions are in line
with the YES campaign, the more likely this voter is to vote
YES. This suggests campaigns can not only influence attitudes
towards the European Union or towards the Treaty,
they can also influence knowledge or perceptions of the
Treaty. Separating out the two dimensions of perceptions
or knowledge of the Treaty thus clearly improves our understanding
of the voting behaviour in this referendum.
The regression results strongly suggest a more complicated
causal explanation than a straightforward relationship between
knowledge and voting behaviour.
Although we have clear results from the regression
analysis that our separation of the two dimensions has
been fruitful, an important caveat regarding this interpretation
is in order. Whereas the existence of a relationship
between objective and subjective knowledge and the reported
voting behaviour is clearly supported by the statistical
results, the support for the relationship with the
campaign depends on observations about the campaign
that go beyond the individual-level data. Casual observation
of the campaign does suggest the focus on perceptions
instead of attitudes and the regression analysis suggests
that these two dimensions do indeed matter.10 We still
have very limited evidence, however, that it really was the
campaign messages that brought about this effect. We have
limited statistical information about the sources of the
perceptions of the respondents d we infer influence of the
campaign from alignment of the attitudes of the respondent
with the messages of the campaign d and we have
limited empirical evidence of how we attributed particular
items to particular campaigns. Furthermore, we do not
directly measure the attitudes of respondents towards the
issues they do or do not perceive to be part of the Treaty.
For this reason, we are not able to directly measure the
interactive effect between perceptions and attitudes as
suggested in our causal diagram in Fig. 1. Our discussion
makes theoretical sense and our empirical findings are in
line with our expectations, but further research is needed
to confirm the theoretical claims made.
6. Conclusion
The Irish referendum on the Lisbon Treaty provides an
excellent opportunity to study voting behaviour in referendums.
Preliminary analyses of a post-referendum opinion
survey suggest that, among other factors, political knowledge
played a crucial role (IMS, 2008; Sinnott et al., 2009).
This article proposes a clear distinction between perceptions
of the referendum issue d what do voters think is
part of the Treaty d and attitudes towards the referendum
issue d how do voters evaluate the elements they think are
part of the Treaty. Although these two components are very
closely related to each other, they are conceptually distinct.
Acknowledging that these are separate components of the
way knowledge and attitudes interact to influence voting
behaviour, we can conceive of a political campaign in a
referendum that emphasises either one of the two components.
The campaigns in the Irish referendum indeed
appear to focus more on influencing the perceptions of the
contents of the Treaty, than on influencing the attitudes
towards these contents. Whereas in some referendums the
issue at hand might be clearly defined and debate centres
on the positive or negative evaluation of this issue, in the
campaign on the Lisbon Treaty, the main debate was on
what was and what was not part of the Treaty.
Including these scales as separate explanatory variables
in a multivariate regression model that measure the extent
to which a respondent is aligned with a particular
campaign in terms of the perceptions of the contents of the
Treaty leads to the conclusion that these alignments are
strong predictors of the vote in the referendum. The more a
voter’s perceptions of the contents of the Treaty are aligned
with the NO campaign, the more likely the voter is to vote
NO, and vice versa for the YES campaign.
While there are good reasons to expect a direct effect of
a lack of subjective knowledge of the referendum issue at
hand on the vote choice in a referendum (see, e.g., Binzer
Hobolt, 2009), the impact of objective knowledge is indirect.
The indirect effect proceeds through various channels,
not least its impact through subjective knowledge, among
which is the particular susceptibility of uninformed voters
to misleading campaign messages. The analysis demonstrates
this effect in the context of the 2008 Irish referendum
on the Lisbon Treaty.
A rational voter can decide not to invest extensively in
acquiring the knowledge to make a fully informed decision
in a referendum as complex as that on the Treaty of Lisbon
and instead to rely on cues from members of the political
elite. The risk in doing so, however, is that these elites will
have particular political interests that might run counter to
those of the voter (Lupia and McCubbins, 1998).
This article makes two contributions to the resolution of
these issues. The first contribution lies in providing a more
nuanced account of the interaction between knowledge,
attitudes, campaigns and voting behaviour in referendums.
The second contribution focuses on the policymaker’s point
of view and argues that governments need to develop ways
and means of ensuring that electorates have access to real
debate and deliberation in a way that enhances knowledge
of the often complex issues that are set before them in the
form of the referendum.
9 These inferences are, of course, based on largely the same information,
namely that recorded in the question presented extensively in Table
1. That this does not lead to unacceptable levels of multicollinearity is
evidenced by the fact that the coefficients on both the objective knowledge
of the Treaty variable and on the campaign variables are statistically
significant. Furthermore, the significant decrease in the Akaike Information
Criterium, by more than twice the number of variables added
(Gelman and Hill, 2007, 525), suggests that the inclusion of the campaign
variables is a significant improvement of the model. 10 Furthermore, a latent trait analysis not reported in this paper provides
supporting evidence that the differentiation between the two dimensions
in the perceptions of the Treaty is reasonable.
222 J.A. Elkink, R. Sinnott / Electoral Studies 38 (2015) 217e225
Acknowledgements
Previously presented at the New Directions in Referendums:
Politics and Campaigns conference, Dublin, May 9,
2013; 1st Annual General Conference of the European Political
Science Association, Dublin, June 16e18, 2011; and
the ECPR Joint Sessions of Workshops, Lisbon, April 2009.
The authors are grateful to Elisabeth Gidengil, Paul Kellstedt,
Theresa Reidy, Jane Suiter and other workshop participants,
as well as the anonymous reviewer, for their
comments.
Appendix
Own economic situation bad
“What about your own economic situation these days?
Would you say it is … ? ” (very good/fairly good/fairly bad/
very bad/don’t know). Values “fairly bad” and “very bad”
are coded as 1, others as 0.
Dissatisfied with government
“Overall are you generally satisfied or dissatisfied with
the way the government is running the country? ” (very
satisfied/quite satisfied/quite dissatisfied/very dissatisfied/
don’t know). Values “very dissatisfied” and “quite dissatisfied”
are coded as 1, others as 0.
Feels close to opposition
“Do you feel close to any of the political parties? ” and if
no, “Do you feel yourself a little closer to one of the political
parties than the others? ” (FF/FG/Labour/Green/Sinn Fein/
PDs/Other/Not close to any). Values “FF”, “PDs” and “Green”
are coded as 1, others as 0.
Identifies as Irish only
“In the near future, do you see yourself as … ? ” (Irish
only/Irish and European/European and Irish/European only/
don’t know). Value “Irish only” is coded as 1, the rest as 0.
Anti-immigration attitude
“Using the card provided please indicate whether
Ireland is made a worse or a better place to live in by people
coming to live here from other countries? ” (1 worse place
to live … 10 better place to live). The variable is entered in
reverse order, 1 ¼ 10, 2 ¼ 9, etc.
Pro-neutrality attitude
“I have a number of statements here that people
sometimes make. I would like you to indicate on this scale
which of each pair of opposing statements comes closest to
your view. A score of one would indicate that you agree
Table 2
Logistic regression explaining the NO vote. Standard errors in parentheses with statistically significant (a ¼ .05) coefficients in bold print. Coefficients are
standardised as suggested by Gelman (2007). The coefficients on age are relative to the 65þ group; those on class relative to the upper middle class; and
those on education relative to those with only primary education.
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6
Constant 0.02 (0.31) 0.08 (0.35) 0.08 (0.36) 0.11 (0.36) 0.05 (0.37) 0.02 (0.37)
Female 0.17 (0.13) 0.07 (0.14) 0.04 (0.15) 0.04 (0.15) 0.00 (0.16) 0.00 (0.15)
Age 18e24 0.37 (0.28) 0.51 (0.31) 0.40 (0.32) 0.42 (0.32) 0.35 (0.33) 0.33 (0.33)
Age 25e34 0.64 (0.23) 0.67 (0.26) 0.63 (0.27) 0.63 (0.27) 0.54 (0.28) 0.54 (0.28)
Age 35e49 0.24 (0.21) 0.37 (0.24) 0.37 (0.25) 0.37 (0.25) 0.33 (0.25) 0.34 (0.25)
Age 50e64 0.23 (0.21) 0.30 (0.24) 0.32 (0.24) 0.30 (0.25) 0.25 (0.25) 0.28 (0.25)
Lower middle class 0.24 (0.20) 0.15 (0.22) 0.11 (0.23) 0.09 (0.23) 0.10 (0.24) 0.12 (0.24)
Skilled worker 0.78 (0.22) 0.62 (0.25) 0.48 (0.26) 0.43 (0.26) 0.46 (0.27) 0.51 (0.27)
Unskilled worker 0.85 (0.24) 0.58 (0.27) 0.47 (0.28) 0.39 (0.28) 0.44 (0.29) 0.53 (0.28)
Large farmer 0.22 (0.30) 0.48 (0.35) 0.53 (0.38) 0.55 (0.38) 0.48 (0.39) 0.49 (0.39)
Small farmer 0.75 (0.50) 0.37 (0.57) 0.55 (0.60) 0.48 (0.60) 0.44 (0.62) 0.52 (0.62)
Secondary education 0.14 (0.25) 0.23 (0.29) 0.32 (0.30) 0.37 (0.30) 0.41 (0.31) 0.34 (0.31)
Third level education 0.42 (0.28) 0.05 (0.33) 0.16 (0.34) 0.28 (0.34) 0.31 (0.35) 0.17 (0.34)
Own economic situation bad 0.76 (0.15) 0.43 (0.18) 0.31 (0.18) 0.29 (0.18) 0.30 (0.19) 0.33 (0.19)
Dissatisfied with government 0.82 (0.13) 0.70 (0.15) 0.76 (0.16) 0.77 (0.16) 0.75 (0.16) 0.74 (0.16)
Feels close to opposition 0.73 (0.12) 0.86 (0.14) 0.92 (0.15) 0.91 (0.15) 0.94 (0.15) 0.96 (0.15)
Identifies as Irish only 0.49 (0.15) 0.34 (0.16) 0.35 (0.16) 0.44 (0.17) 0.41 (0.16)
Pro-neutrality attitude 0.78 (0.14) 0.76 (0.15) 0.80 (0.15) 0.72 (0.15) 0.68 (0.15)
Anti-immigration attitude 0.56 (0.17) 0.39 (0.17) 0.37 (0.17) 0.31 (0.18) 0.33 (0.18)
EU membership a good thing ¡1.27 (0.20) ¡1.11 (0.20) ¡1.09 (0.20) ¡1.00 (0.20) ¡1.03 (0.20)
Too many issues decided by EU 0.64 (0.15) 0.63 (0.15) 0.62 (0.15) 0.50 (0.16) 0.52 (0.16)
EU means low wages 0.69 (0.19) 0.62 (0.19) 0.62 (0.19) 0.54 (0.19) 0.54 (0.19)
Low objective knowledge of EU 0.27 (0.18) 0.17 (0.19) 0.18 (0.19) 0.30 (0.19)
Low objective knowledge of treaty 1.35 (0.19) 1.25 (0.19) 0.52 (0.31) 0.70 (0.30)
Low subjective knowledge of EU 0.06 (0.20) 0.05 (0.20)
Low subjective knowledge of treaty 0.48 (0.20) 0.60 (0.21)
Positive on NO campaign items 1.18 (0.18) 1.12 (0.18)
Positive on YES campaign items ¡0.87 (0.28) ¡0.82 (0.28)
Number of observations 1301 1254 1254 1246 1246 1254
Akaike information criterion 1557 1288 1223 1215 1172 1184
J.A. Elkink, R. Sinnott / Electoral Studies 38 (2015) 217e225 223
fully with the statement on the left. A score of nine would
indicate that you agree fully with the statement on the
right. Of course your view could be somewhere in between.
Also of course there may be issues that you have no
particular view on. If so, please just say this and we will
move on to the next item.” (1 “Ireland should do everything
it can to strengthen its neutrality even if this means being
less involved in European Union cooperation on foreign
and defence policy” … 9 “Ireland should be willing to
accept limitations on its neutrality so that it can be more
fully involved in European Union cooperation on foreign
and defence policy”). Values are in reverse order, 1 ¼ 9,
2 ¼ 8, etc.
EU membership a good thing
“Generally speaking, do you think that Ireland’s membership
of the European Union (European Community) is …
? ” (“A good thing”/“A bad thing”/“Neither good nor bad”/
“Don’t know”). The first category is coded as 1, the rest as 0.
Too many issues decided by EU
“There has been a lot of discussion recently about the
European Union. Some people say that too many issues are
decided on by the European Union, others say that more
issues should be decided on by the European Union. Which
of the following statement comes closest to your view? ”
(“Too many issues are decided by the European Union”/
“The number of issues decided on by the European Union at
present is about right”/“More issues should be decided by
the European Union”/“I have not really thought about it”/“It
depends on the issue”/“Don’t know”). Those that selected
the first option were coded as 1, others as 0.
EU means low wages
“What does the European Union mean to you personally?
” (many answer categories, among which “Lower
wage rates”). Those that included “Lower wage rates” in
their selection were coded as 1, others as 0.
Low objective knowledge of Treaty
Number of items deemed to be correct in Table 1 that
are perceived to be in the Treaty; ‘don’t know’ taken as an
incorrect answer.
Low objective knowledge of EU
“For each of the following statements about the European
Union could you please tell me whether you think it is
true or false? ” (“The EU currently consists of fifteen
Member States”/“Switzerland is a member of the European
Union”/“Every six months, a different Member State becomes
the President of the Council of the European Union”/
“The members of the European Parliament are directly
elected by the citizens of the EU”). A four point score is
generated, from 1 (4 correct answers) to 4 (1 correct
answer). All respondents had at least one correct answer.
Low subjective knowledge of Treaty
“By the date of the referendum (12 June), how good was
your understanding of the issues involved? Please use this
card to choose the phrase that applies best to you.” (“I had a
good understanding of what the Treaty was all about”/“I
understood some of the issues but not all that was
involved”/“I was only vaguely aware of the issues
involved”/“I did not know what the Treaty was about at
all”/don’t know). The resulting variable is a four point scale
from good understanding to not knowing what the Treaty
was about at all.
Low subjective knowledge of EU
“And how about the European Union in general? Using
this scale, how much do you feel you know about the European
Union, its policies, its institutions? ” (1 nothing at all
… 10 know a great deal).
Positive on NO campaign items
Number of items associated with the NO campaign in
Table 1 that are perceived to be in the Treaty; ‘don’t know’
taken as missing data.
Positive on YES campaign items
Number of items associated with the YES campaign in
Table 1 that are perceived to be in the Treaty; ‘don’t know’
taken as missing data.
Abstained
“On the 12 June last, a referendum was held on the
Treaty of Lisbon. As you may remember, many people did
not vote in that referendum. How about you? Did you vote
in the referendum on the Treaty of Lisbon that was held in
June? ” (“Yes e voted”/“Did not vote”). In all analyses, this is
used as a filter variable, and only those who voted are taken
into account.
Voted NO
“How did you vote in that referendum e in favour or
against the Lisbon Treaty? ” (“Voted e In favour”/“Voted e
Against”). This is the dependent variable in the regression
analyses.
References
Ajzen, I., Fishbein, M., 1977. Attitude-behavior relations: a theoretical
analysis and review of empirical research. Psychol. Bull. 84 (5),
888e918.
Alvarez, R.M., 1997. Information and Elections. University of Michigan
Press.
Basinger, S.J., Lavine, H., 2005. Ambivalence, information, and electoral
choice. Am. Polit. Sci. Rev. 99, 169e184.
Binzer Hobolt, S., 2005. When Europe matters: the impact of political
information on voting behaviour in EU referendums. J. Elections
Public Opin. Parties 15 (1), 85e109.
Binzer Hobolt, S., 2009. Europe in Question: Referendums on European
Integration. Oxford University Press, New York.
224 J.A. Elkink, R. Sinnott / Electoral Studies 38 (2015) 217e225
Bowler, S., Donovan, T., 1994. Information and opinion change on ballot
propositions. Polit. Behav. 16 (4), 411e435.
Campbell, A., Converse, P.E., Miller, W.E., Stokes, D., 1960. The American
Voter. University of Chicago Press, Chicago.
De Vries, C.E., Van der Brug, W., Van Egmond, M.H., Van der Eijk, C., 2011.
Individual and contextual variation in EU issue voting: the role of
political information. Elect. Stud. 30 (1), 16e28.
Delli Carpini, M.X., Keeter, S., 1996. What Americans Know about Politics
and Why it Matters. Yale University Press.
Druckman, J.N., 2004. Political preference formation: competition,
deliberation, and the (ir)relevance of framing effects. Am. Polit. Sci.
Rev. 98 (4), 671e686.
Druckman, J.N., Nelson, K.R., 2003. Framing and deliberation: how citizens’
conversations limit elite influence. Am. J. Polit. Sci. 47 (4),
729e745.
Ehin, P., 2001. Determinants of public support for EU membership: data
from the Baltic countries. Eur. J. Polit. Res. 40, 31e56.
Franklin, M., 2002. Learning from the Danish case: a comment on Palle
Svensson’s critique of the Franklin thesis. Eur. J. Polit. Res. 41,
751e757.
Franklin, M., Marsh, M., Wlezien, C., 1994. Attitudes towards Europe and
referendum votes: a response to Siune and Svensson. Elect. Stud. 13,
117e121.
Gabel, M.J., 1998. Economic integration and mass politics: market liberalisation
and public attitudes in the European Union. Am. J. Polit. Sci.
42 (3), 936e953.
Gamble, A., 2006. The European disunion. Br. J. Polit. Int. Relat. 8 (1), 34e49.
Garry, J., Marsh, M., Sinnott, R., 2005. ‘Second-order’ versus ‘issue-voting’
effects in EU referendums. Eur. Union Polit. 6 (2), 201e221.
Gelman, A., 2007. Scaling regression inputs by dividing by two standard
deviations. Stat. Med. 27 (15), 2865e2873.
Gelman, A., Hill, J., 2007. Data Analysis Using Regression and Multilevel/
Hierarchical Models. Analytical Methods for Social Research. Cambridge
University Press, Cambridge.
Grynaviski, J.D., Corrigan, B.E., 2006. Specification issues in proximity
models of candidate evaluation (with issue importance). Polit. Anal.
14, 393e420.
Hix, S., 2006. Why the EU Needs Left-right Politics. In: Politics: the Right or
the Wrong Sort of Medicine for the EU?. Notre Europe Policy Paper 19.
Hooghe, L., Marks, G., 2009. A postfunctionalist theory of European
integration: from permissive consensus to constraining dissensus. Br.
J. Polit. Sci. 39 (1), 1e23.
IMS, 2008. Post Lisbon Treaty Referendum Research Findings (Technical
report Millward Brown-IMS).
Kastellec, J.P., Leoni, E.L., 2007. Using graphs instead of tables in political
science. Perspect. Polit. 5 (4), 755e771.
King, G., 1998. Unifying Political Methodology. The Likelihood Theory of
Statistical Inference. University of Michigan Press.
Krosnick, J.A., 1988. Attitude importance and attitude change. J. Exp. Soc.
Psychol. 24, 240e255.
Laffan, B., 2008. Reflection: three readings of the failure to ratify the
constitutional treaty. In: Brown, T. (Ed.), Lisbon: What the Reform
Treaty Means. Institute of International and European Affairs, Dublin,
pp. 60e70.
Lord, C.G., Ross, L., Lepper, M.R., 1979. Biased assimilation and attitude
polarization: the effects of prior theories on subsequently considered
evidence. J. Personal. Soc. Psychol. 37 (11), 2098e2109.
Lupia, A., 2006. How elitism undermines the study of voter competence.
Crit. Rev. 18 (1), 217e232.
Lupia, A., McCubbins, M.D., 1998. The Democratic Dilemma: Can Citizens
Learn What They Need to Know? Cambridge University Press, New
York.
Miller, J.M., Krosnick, J.A., 2000. News media impact on the ingredients of
presidential evaluations: politically knowledgeable citizens are
guided by a trusted source. Am. J. Polit. Sci. 44 (2), 301e315.
Morgan, S.L., Winship, C., 2007. Counterfactuals and Causal Inference.
Methods and Principles for Social Research. Cambridge University
Press, New York.
Nie, N.H., Verba, S., Petrocik, J.R., 1979. The Changing American Voter.
Harvard University Press Cambridge.
Palfrey, T.R., Poole, K.T., 1987. The relationship between information,
ideology, and voting behavior. Am. J. Polit. Sci. 31 (3), 511e530.
Pearl, J., 2000. Causality: Models, Reasoning, and Inference. Cambridge
University Press, Cambridge.
Perloff, R.M., 2003. The Dynamics of Persuasion. Communication and
Attitudes in the 21st Century, second ed. Lawrence Erlbaum Associates,
Mahwah, NJ.
Quinlan, S., 2009. The Lisbon treaty referendum 2008. Ir. Polit. Stud. 24
(1), 107e121.
Reif, K., Schmitt, H., 1980. Nine second order national elections: a conceptual
framework for the analysis of European election results. Eur. J.
Polit. Res. 8 (1), 3e44.
Sinnott, R., 2001. Attitudes and Behaviour of the Irish Electorate in the
Referendum on the Treaty of Nice (Technical report, European Commission
Representation in Ireland).
Sinnott, R., 2003. Attitudes and Behaviour of the Irish Electorate in the
Second Referendum on the Treaty of Nice (Technical report, European
Commission Representation in Ireland).
Sinnott, R., Elkink, J.A., O’Rourke, K., McBride, J., 2009. Attitudes and
Behaviour in the Referendum on the Treaty of Lisbon (Technical
report, Department of Foreign Affairs).
Suiter, J., Reidy, T., 2013. It’s the campaign learning stupid, an examination
of a volatile Irish referendum. Parliam. Aff. (published online).
Van Apeldoorn, B., 2009. The contradictions of ’embedded neoliberalism’
and Europe’s multi-level legitimacy crisis: the European project and
its limits. In: Van Apeldoorn, B., Drahokoupil, J., Horn, L. (Eds.), Contradictions
and Limits of Neoliberal European Governance. From
Lisbon to Lisbon. Palgrave MacMillan, pp. 21e43.
Zaller, J., 1992. The Nature and Origins of Mass Opinion. Cambridge University
Press, Cambridge.
J.A. Elkink, R. Sinnott / Electoral Studies 38 (2015) 217e225 225
Political knowledge and campaign effects in the 2008 Irish
referendum on the Lisbon Treaty
Johan A. Elkink*
, Richard Sinnott
University College Dublin, Belfield, Dublin 4, Ireland
article info
Article history:
Available online 25 February 2015
Keywords:
Referendums
Political knowledge
European integration
Campaigns
Lisbon Treaty
Ireland
abstract
This article makes a distinction between the attitude component of campaigns and the
knowledge component and argues that a campaign that influences knowledge of a proposal
can be quite successful in influencing the vote. On June 12th 2008 Irish voters voted
against ratification of the Lisbon Treaty. Perceptions, and especially misperceptions, of the
Treaty played an important role in the referendum. The campaign focused in particular on
influencing voters’ perceptions of, rather than attitudes towards the Treaty. This article
examines the interaction between knowledge, campaigns, and perceptions of the Treaty, in
the referendum.
© 2015 Elsevier Ltd. All rights reserved.
1. Introduction
The nature of European Union referendums, as distinct
from general elections and from clearly single issue referendums,
provides valuable material with which to further
our understanding of how voters make their decisions on
whether and how to vote. EU referendums are also typically
quite consequential, making it important not only to
understand them from a scientific, but also from a policymaking
point of view.
Voting behaviour in EU referendums is generally
understood either in terms of satisfaction or dissatisfaction
with the government of the day e the so-called secondorder
perspective (Reif and Schmitt, 1980; Franklin et al.,
1994; Franklin, 2002); or in terms of the attitudes of
voters towards European integration (Garry et al., 2005)
and towards European policies (Hix, 2006; Laffan, 2008;
Gamble, 2006); or in terms of the utilitarian benefits one
expects to obtain from European integration (Gabel, 1998;
Ehin, 2001; Van Apeldoorn, 2009); or, finally, in terms of
the effect of knowledge of the issue at hand or of the
European Union in general (Binzer Hobolt, 2005). Previous
studies of the Irish referendums suggest that political
knowledge was crucial in determining the outcome of the
referendums on the Nice and Lisbon treaties (Sinnott, 2001,
2003; Sinnott et al., 2009).
This article provides an in-depth examination of the role
of political knowledge in the 2008 Irish referendum on the
Treaty of Lisbon. It argues that a central feature of the
campaign was the emphasis on perceptions of the contents
of the Treaty, rather than evaluations of these contents.
Although these two components of a referendum campaign
are inseparably linked, they are nevertheless conceptually
distinct.
Referendum campaigns can be distinguished by the
extent to which they emphasise either the knowledge or
the attitude component. On the basis of a statistical analysis
of the results from an opinion survey conducted shortly
after the referendum it is apparent that knowledge played a
crucial role (Sinnott et al., 2009). This research shows that
perceptions of the contents of the Treaty can be separated
into two groups, one corresponding to issues emphasised
by the NO campaign and one to issues emphasised by the
YES campaign. The alignment of what voters perceive to be
the contents of the Treaty with either the NO or the YES
campaign is shown to be an important factor in accounting
* Corresponding author.
E-mail addresses: jos.elkink@ucd.ie (J.A. Elkink), richard.sinnott@ucd.ie
(R. Sinnott).
Contents lists available at ScienceDirect
Electoral Studies
journal homepage: www.elsevier.com/locate/electstud
http://dx.doi.org/10.1016/j.electstud.2015.02.003
0261-3794/© 2015 Elsevier Ltd. All rights reserved.
Electoral Studies 38 (2015) 217e225
for the referendum result, as opposed to the overall level of
knowledge of the Treaty contents or the negative or positive
evaluation of these Treaty contents.
The next section of this article briefly describes the 2008
referendum campaign, thus providing the context in which
the empirical analysis is pursued. Section 3 sets out the
causal mechanisms that we believe connect political
knowledge to voting behaviour. Section 4 introduces the
survey data and the measurement of the key variables.
Section 5 tests the causal claims in a series of regression
analyses. Section 6 summarises the findings and draws some
tentative conclusions. The appendices contain definitions of
the variables and further details on the empirical analysis.
2. The Irish context
The two sides of the campaign in the 2008 referendum
were of very different composition. Whereas the YES
campaign was forced to defend the Treaty in its entirety
and had to have an answer to every possible criticism of the
Treaty, each different actor in the NO campaign could
simply focus on a specific aspect of the Treaty. Quinlan
(2009: 109) succinctly describes the NO campaign:
“As in previous European referendums, the no side was
a diverse coalition[, including] … Sinn Fein [as] the only 
party in the Oireachtas to call for a no vote …, the Peace
and Neutrality Alliance (PANA) … the People’s
Campaign was a broad coalition which included [several
MEPs] … Meanwhile, sustained opposition to the Treaty
on the right was provided by Coir  … But the most vocal
opposition to the Treaty came from the think-tank Libertas.
Founded by Galway businessman Declan Ganley,
Libertas opposed Lisbon because in its view it did not
provide a transparent democratic Europe, weakened
Irish power within the EU and opened the way for Irish
corporate tax rates to be interfered with by Brussels.”
The YES campaign on the other hand consisted of almost
all major political parties (with the Green Party formally
staying neutral), the Irish Alliance for Europe and a number
of business interest groups (Quinlan, 2009: 108e109). In
addition to the protagonists on either side, there was the
role of the Referendum Commission, which was established
by statute to “explain the subject matter to the
public, … promote awareness of the referendum and …
encourage people to vote” (Quinlan, 2009: 109).
While this is at first sight a neutral assignment, in a
referendum in which lack of knowledge of the Treaty
contents is an explicit issue, the task of increasing
“awareness” easily becomes indistinguishable from the
political campaigns.
While the YES campaign was emphasising the increased
efficiency and improved decision-making in the European
Union to be brought about by the Lisbon Treaty, the NO
campaign actively attempted to influence specific perceptions
of the Treaty, often by emphasising items that were
not actually present in the Treaty text, but that are important
to Irish voters. For example, one section of the NO
campaign emphasised the risk that the Lisbon Treaty would
lead to the establishment of a European army with
conscription for Irish citizens (Quinlan, 2009: 111).
3. Knowledge and referendum voting behaviour
Knowledge played a crucial role in determining the
outcome of the Lisbon Treaty referendum. This is clear from
the subjective evaluations of a nationally representative
sample of voters. When NO voters in a post-referendum
poll were asked why they voted NO, 46% of them
mentioned something related to lack of information,
knowledge, or understanding (Sinnott et al., 2009: 13).
Subsequent statistical analyses of the survey data using
appropriate control variables confirms that political
knowledge was a crucial factor in determining the referendum
outcome (Sinnott et al., 2009), which would not
surprise even a casual observer of the campaign.
The idea that knowledge affects voting behaviour is
hardly new (Carpini and Keeter, 1996; see also Bowler, this
issue). In the 2008 Irish referendum, however, the role of
knowledge was a complicated one, involving campaign
effects, perceptions and misperceptions of the contents of
the Treaty, and subjective evaluations of the level of
knowledge of the Treaty. Fig. 1 provides a graphical
depiction of our key causal argument. The difference between
subjective knowledge d the extent to which a voter
feels informed about the issue at hand in the referendum d
and objective knowledge d the actual level of knowledge
of the issue at hand, or of the European Union more
generally d will be a crucial distinction in what follows.
The primary mechanism through which subjective
knowledge can be expected to have an impact on vote
choice (Arrow 4) is through risk aversion. The more uncertain
a respondent feels about the impact of voting for a
referendum proposal, the more likely the voter is to support
the status quo and to vote against the proposal (Suiter
and Reidy, 2013). Binzer Hobolt (2009: 40e57) provides a
formal model elaborating on this idea. In this argument, the
level of objective knowledge is of less importance;
although it can of course be assumed that objectively less
knowledgeable voters are also likely to feel less knowledgeable
(Arrow 3). A feeling of a lack of knowledge was
widespread in the 2008 referendum. Indeed some politicians
exacerbated this feeling by making statements such
as the admission by Taoiseach (prime minister) Brian
Cowen on television that he had not read the Treaty “cover
to cover” or the remark by Irish EU Commissioner Charlie
Fig. 1. Causal model examining the effect of knowledge on the vote (YES or
NO). Objective knowledge is here conceptualised as perceptions that are
correct, hence is part of the perceptions variable in this diagram.
218 J.A. Elkink, R. Sinnott / Electoral Studies 38 (2015) 217e225
McCreevy a week later that any voter who read the Treaty
was “insane” (Quinlan, 2009: 113).
The impact of objective knowledge is likely to be more
indirect, as there is no reason a priori to assume that higher
levels of knowledge lead to stronger support for European
integration. Furthermore, lack of knowledge of politics in
general, of the European Union, or of the Lisbon Treaty does
not necessarily render the voter incapable of casting a
rational choice: “if there are multiple informational pathways
to a competent vote (e.g., if interest-group endorsements,
effective heuristics, etc., lead a voter to cast the
same vote as she would have if better informed), then
voters need not use the same information that political
elites use” (Lupia, 2006: 222; Bowler, this issue).
This reliance on informed sources to make a decision
does depend on the trustworthiness of the source. In other
words, the information can be either correct or incorrect
and therefore such heuristics contain a risk for the voter d
the interests of the sender of information might be in
conflict with those of the receiver (Lupia and McCubbins,
1998). An objectively less knowledgeable voter is thus
relying more on campaign messages from trusted sources
than a more knowledgeable voter, and is more susceptible
to potentially misleading campaign messages about the
referendum issue at hand.
Less knowledgeable voters are therefore more dependent
on high quality information, but are likely to in fact
rely on more easily accessible sources (Bowler and
Donovan, 1994). Hooghe and Marks (2009: 13) make a
similar argument in particular in the context of public
opinion on Europe, which they argue “is particularly susceptible
to construction: i.e. priming (making a consideration
salient), framing (connecting a particular
consideration to a political object) and cueing (instilling a
bias)” (see also, Druckman, 2004; Druckman and Nelson,
2003; Miller and Krosnick, 2000).
The finding that knowledge is important, even when
controlling for well-known effects such as the popularity of
the government of the day or attitudes towards the European
Union in general, has implications for the kind of
campaigns that could be effective in a referendum. If
knowledge matters, campaigns can either focus on changing
the knowledge or perceptions voters have of the issue
at hand or they can concentrate on the attitudes these
voters have towards what they perceive to be the issue. We
are thus making a distinction between the attitude and the
knowledge components of campaigns.1
An obvious criticism of this distinction is that these two
components can never be observed entirely independently.
Any campaign that emphasises how, for example, corporate
taxes are part of the Lisbon Treaty, also implies that this is a
negative thing. In other words, such a campaign influences
perceptions and attitudes simultaneously. It can be argued
that these are still distinct, if generally coinciding, components.
The point becomes clearer when we compare the
Irish referendum on the Lisbon Treaty with the Irish referendum
on abortion in March 2002. In a referendum on
abortion, the issue at hand is clear to voters. Campaigns
focus on influencing voters’ evaluations of whether allowing
abortion is a good or a bad thing, so they attempt to
influence attitudes rather than perceptions. In the referendum
on the Lisbon Treaty, however, the emphasis was
very much on influencing perceptions of what was in the
Treaty rather than the evaluations of these. For example,
the NO campaign would argue that the Treaty could lead to
an increase in Irish corporate tax rates. On both sides of the
campaign, many voters would oppose such an increase, but
it is the perception that this is part of the referendum decision
that the campaign tried to influence, rather than
whether such increase would be a good or bad thing for
Ireland. It was implied that it is a bad thing, but that was
not the contentious issue.
Since ‘perceptions’ and ‘attitudes’ are often confused, it
is crucial that we are specific about what we mean by
these terms. According to the expectancy-value approach
in social psychology, attitudes are “a multiplicative combination
of (a) strength of beliefs that an object has certain
attributes and (b) evaluations of these attributes” (Perloff,
2003: 46; see also Ajzen and Fishbein, 1977). In this
article we will refer to ‘perceptions’ or ‘knowledge’ when
talking about the beliefs that an object, in this case the
Treaty, has certain attributes. In short, whereas ‘perceptions’
refers to what the voters think, rightly or wrongly, is
contained in the Treaty, ‘knowledge’ refers to the extent to
which these perceptions were correct. We refer to ‘attitudes’
when talking about the evaluations of these
perceived attributes.
An example would be the campaign item related to the
possibility of an increase in the Irish corporate tax rates. We
make a clear distinction between the perception that the
Lisbon Treaty does indeed imply future increases in tax
rates and the evaluating attitude the respondent has towards
the idea of such increase.2 There is an extensive
literature in the US context outlining the conditional and
nonlinear nature of issue voting (e.g., Grynaviski and
Corrigan, 2006) and the impact of political knowledge
therein (Campbell et al., 1960; Nie et al., 1979; Palfrey and
Poole, 1987; Krosnick, 1988; Zaller, 1992; Alvarez, 1997;
Basinger and Lavine, 2005; De Vries et al., 2011).
4. Knowledge and the campaign messages
The main data for the following empirical analysis is a
national opinion poll conducted by Millward Brown IMS on
behalf of the Department of Foreign Affairs of the Republic
of Ireland. Data were collected from 2101 respondents aged
18 years and older. All respondents were Irish citizens listed
on the electoral register. The poll was carried out between
24 and 31 July 2008, slightly over a month after the referendum
had taken place. Details and bivariate analyses can
be found in IMS (2008) and a more extensive multivariate
1 Or framing and cueing, respectively, in the language of political
studies.
2 A somewhat complicating factor is that at times these perceptions
might be influenced by attitudes. A study by Lord et al. demonstrates how
subjects interpret neutral or contradicting evidence in such a way that it
confirms their prior beliefs. They adjust their evaluation of the quality,
biasedness, logic, and conclusions all on the basis of their prior attitude
towards the object under study (Lord et al., 1979).
J.A. Elkink, R. Sinnott / Electoral Studies 38 (2015) 217e225 219
analysis in Sinnott et al. (2009). The appendix provides the
precise coding of all variables used.3
Table 1 provides an overview of key Treaty items
emphasised by the two campaigns and translated into a
survey question. Although most key campaign items have
been included in the question, it is not an exhaustive list of
issues raised during the campaign. The respondents were
asked which of the items were part of the Treaty with the
option to select the ‘don’t know’ answer category in each
case. The responses to this question are presented in the
first two columns of the table. These answers can thus be
seen as the perceptions of the Treaty contents. For most
items it is quite clear whether these were or were not part
of the Treaty. The third column lists the correct answers.
The final column lists the items that are indeed in the
Treaty and those that are not. As we argue here, the
campaign focused somewhat less on influencing the attitude
towards particular items, but rather on influencing the
perception that particular items were or were not part of
the Treaty. An evaluation, either positive or negative, was
generally only implied. The final column lists the particular
items that the campaigns suggested were part of the Treaty.
Although the evaluation of the items differs significantly,
both campaigns admitted clearly that the loss of an Irish
commissioner was indeed part of the Treaty.4
We argue that an important part of the effect of objective
knowledge is the susceptibility of respondents with
low levels of knowledge to the (potentially misleading)
campaign messages of the different campaigns. That for a
large proportion of the electorate, campaigns can influence
the vote in the referendum not only by promoting a
particular evaluation of the issue to be voted on, but also
particular perceptions of the issue itself, of what it is one is
voting on. For the statistical evaluation of this proposition,
we measure the alignment of respondents with particular
campaign messages through additive scales. Both the
measures of objective knowledge of the Treaty and the
measures of campaign effects are based on the data
described in Table 1. The objective knowledge scale reflects
the number of correct answers on items 1e5, 7, 8, 10, and
11. The NO campaign scale counts the number of YES answers
on items 2e6 and the YES campaign scale counts the
number of YES answers on items 7e11.5 The two scales on
the campaigns do not summarise the campaigns as such d
they summarise how the campaigns provided different
perceptions of the contents of the Treaty.
Cronbach’s alpha can be used to assess whether these
are indeed reasonable scales to use as measures in subsequent
regression analysis. We obtain for the knowledge
scale a ¼ 0.74; for the NO campaign items a ¼ 0.76 and for
the YES campaign items a ¼ 0.71. Typically, an alpha of
approximately 0.6 is seen as the threshold to accept a scale
as sufficiently additive e the items in the scale are suffi-
ciently correlated to consider them indicators of the same
underlying concept.
We use these additive scales in the regression analysis in
the next section to assess the relative explanatory power of
objective knowledge of the Treaty contents e the truth
value of the perceptions of the Treaty e and of the alignment
of the perceptions of the Treaty with the messages of
the two campaigns. An important caveat is necessary,
however, namely that there is the potential for endogeneity
in this model. While we take the alignment of voters with
particular campaign messages to be an explanatory variable
for their vote choice, it is possible that for some voters
the decision to vote YES or NO was made early in the
campaign and that this has lead them to follow a particular
campaign messages more closely, which lead to them
aligning their perceptions with the campaign messages.
Both in terms of the information they pay attention to and
in line with cognitive dissonance theory, such reverse
causality is a genuine possibility.6
5. Explaining vote choice in the referendum
We are primarily interested in the role of knowledge in
the referendum. As is clear from Fig. 2, the level of objective
knowledge of the contents of the Treaty has a substantial
effect on the individual’s voting behaviour. This figure
Table 1
Perceptions of the contents of the Lisbon Treaty. The question is: “Which
of the following do you think are included in the Lisbon Treaty?”.
% Yes % No Correct Campaign
1. Loss of Irish Commissioner
for 5 out of every 15 years
65 9 Correct Ambiguous
2. Ending of Ireland’s right to
decide its own corporate
tax rate
43 22 Incorrect No campaign
3. The introduction of
conscription to a
European army
33 37 Incorrect No campaign
4. The reduction of Ireland’s
voting strength in the
Council of Ministers
48 18 Correct No campaign
5. The end of Ireland’s
control over its
policy on abortion
34 33 Incorrect No campaign
6. The erosion of Irish
neutrality
42 30 Ambiguous No campaign
7. Improved efficiency of EU
decision-making
56 15 Correct Yes campaign
8. Strengthening Europe’s
role in the world
61 12 Correct Yes campaign
9. Improved protection of
workers’ rights
49 19 Ambiguous Yes campaign
10. Strengthening the role
ofnational parliaments
in EU decision-making
43 20 Correct Yes campaign
11. The Charter of
Fundamental Rights
36 14 Correct Yes campaign
3 The questionnaire was designed by a working group comprising
officials from the Irish Department of Foreign Affairs and researchers
from the survey agency (Millward Brown-IMS). Professor Richard Sinnott
acted as technical advisor to the working group.
4 This despite the fact that even under the already adopted Nice Treaty,
Ireland will lose its Commissioner, albeit later.
5 Note that for the knowledge scale, the ‘don’t know’ category is taken
as someone who lacks knowledge, so it is combined with the incorrect
answer, while for the campaign scales, the ‘don’t know’ category is
treated as missing data.
6 We thank the anonymous reviewer for this excellent point.
220 J.A. Elkink, R. Sinnott / Electoral Studies 38 (2015) 217e225
presents the predicted values of the probability of a NO
vote for a particular configuration of the independent variables
(see the caption). The predictions are based on the
logistic regression reported as Model 3 in Table 2 in the
appendix, which includes a large number of control variables,
the measure of knowledge as the number of correct
perceptions of the contents of the Treaty, and subjective
knowledge, but which excludes the two separate
dimensions discussed above. Whereas a knowledgeable
voter who is dissatisfied with the government has slightly
less than a fifty per cent chance of voting NO, a similar voter
with much less knowledge of the Treaty contents has a
probability of voting NO of about ninety per cent. The effect
is strong for both people who are satisfied with the
government and people who are not. While the secondorder
effects of closeness to the opposition and dissatisfaction
with the government are important factors in
explaining the vote, political knowledge played a more
influential role in this referendum.
As suggested in the theoretical discussion above, the
story is somewhat more complicated than the results in
Fig. 2 suggest. The causal diagram in Fig. 1 functions as a
useful guide for a more nuanced interpretation of the
regression results, which are graphically presented in
Fig. 3.
7 Two separate models were estimated to take account
of the different causal paths to be evaluated. When
estimating a causal effect, it is generally advised to control
for any potential confounding factors, but not to control for
any mediating variables (Pearl, 2000; Morgan and Winship,
2007). As can be easily determined on the basis of the
causal diagram, to estimate the effect of objective knowledge
on the vote, we do not want to control for subjective
knowledge, but to estimate the effect of subjective
knowledge, we do want to control for the level of objective
knowledge. For exactly this reason, the two models were
estimated separately. Although Fig. 3 only reports the coefficients
of interest to the theoretical discussion in this
paper, a large set of control variables were included in the
model. Numerical estimates of all variables are reported in
Table 2 in the appendix. Since all variables have been
standardised, we can directly compare the magnitudes of
the effects of each variable.8
It is clear from these estimates that subjective knowledge
of the Treaty has a significant effect on the vote. Objective
knowledge also has a significant effect on the probability of
voting NO, but once we control for subjective knowledge,
this effect is reduced and is statistically insignificant. This
suggests that the impact of objective knowledge is indeed
02468
0.0 0.2 0.4 0.6 0.8 1.0
Low objective knowledge of the Lisbon treaty
Probability of NO vote
Satisfied
Dissatisfied
Fig. 2. Effect of objective knowledge of the Lisbon Treaty and dissatisfaction
with the government on the NO vote. Estimates are based on Model 3 of Table
2 in the appendix. The grey shaded area is a histogram of the objective
knowledge measure. The lines represent the median voter, which is an
hypothetical lower middle class female, between 35 and 49 years old, with
secondary education, who supports an opposition party, had a median score
on the anti-immigration and Irish neutrality scales, who does not consider her
own economic situation to be bad and who does not identify as “Irish only”.
Fig. 3. Absolute values of logistic regression coefficients and their standard
errors, based on models 5 (solid lines) and 6 (dashed lines) in Table 2. The
models contain the same control variables (not shown) and differ only in
terms of the inclusion of the two subjective knowledge variables. Horizontal
lines reflect the 95% confidence intervals, while the vertical markers reflect
the 90% confidence intervals. The full regression table can be found in
Table 2 in the appendix.
7 See Kastellec and Leoni (2007) for an argument in favour of
presenting regression results in this manner.
8 A logistic regression can be perceived of as a linear relation of the
independent variables with a latent dependent variable, whereby the
probability of a 1 on the dependent variable is a non-linear function of
the latent variable (King, 1998). The reported coefficients reflect in all
cases the effect of an increase of two standard deviations (Gelman, 2007)
in the independent variable on the latent dependent variable.
J.A. Elkink, R. Sinnott / Electoral Studies 38 (2015) 217e225 221
through the level of subjective knowledge (Arrow 3 in the
causal diagram in Fig. 1) d the awareness of the voter of his
or her lack of knowledge d and the two dimensions of the
information from the campaign (Arrow 6).
The alignment of the perceptions of the contents of the
Treaty with that of either side of the campaign has a highly
significant effect on the NO vote, even stronger than that of
objective and subjective knowledge variables.9 The more
the voter’s perceptions of the contents of Treaty are in line
with the messages of the NO campaign, the more likely this
voter is to vote NO; the more the perceptions are in line
with the YES campaign, the more likely this voter is to vote
YES. This suggests campaigns can not only influence attitudes
towards the European Union or towards the Treaty,
they can also influence knowledge or perceptions of the
Treaty. Separating out the two dimensions of perceptions
or knowledge of the Treaty thus clearly improves our understanding
of the voting behaviour in this referendum.
The regression results strongly suggest a more complicated
causal explanation than a straightforward relationship between
knowledge and voting behaviour.
Although we have clear results from the regression
analysis that our separation of the two dimensions has
been fruitful, an important caveat regarding this interpretation
is in order. Whereas the existence of a relationship
between objective and subjective knowledge and the reported
voting behaviour is clearly supported by the statistical
results, the support for the relationship with the
campaign depends on observations about the campaign
that go beyond the individual-level data. Casual observation
of the campaign does suggest the focus on perceptions
instead of attitudes and the regression analysis suggests
that these two dimensions do indeed matter.10 We still
have very limited evidence, however, that it really was the
campaign messages that brought about this effect. We have
limited statistical information about the sources of the
perceptions of the respondents d we infer influence of the
campaign from alignment of the attitudes of the respondent
with the messages of the campaign d and we have
limited empirical evidence of how we attributed particular
items to particular campaigns. Furthermore, we do not
directly measure the attitudes of respondents towards the
issues they do or do not perceive to be part of the Treaty.
For this reason, we are not able to directly measure the
interactive effect between perceptions and attitudes as
suggested in our causal diagram in Fig. 1. Our discussion
makes theoretical sense and our empirical findings are in
line with our expectations, but further research is needed
to confirm the theoretical claims made.
6. Conclusion
The Irish referendum on the Lisbon Treaty provides an
excellent opportunity to study voting behaviour in referendums.
Preliminary analyses of a post-referendum opinion
survey suggest that, among other factors, political knowledge
played a crucial role (IMS, 2008; Sinnott et al., 2009).
This article proposes a clear distinction between perceptions
of the referendum issue d what do voters think is
part of the Treaty d and attitudes towards the referendum
issue d how do voters evaluate the elements they think are
part of the Treaty. Although these two components are very
closely related to each other, they are conceptually distinct.
Acknowledging that these are separate components of the
way knowledge and attitudes interact to influence voting
behaviour, we can conceive of a political campaign in a
referendum that emphasises either one of the two components.
The campaigns in the Irish referendum indeed
appear to focus more on influencing the perceptions of the
contents of the Treaty, than on influencing the attitudes
towards these contents. Whereas in some referendums the
issue at hand might be clearly defined and debate centres
on the positive or negative evaluation of this issue, in the
campaign on the Lisbon Treaty, the main debate was on
what was and what was not part of the Treaty.
Including these scales as separate explanatory variables
in a multivariate regression model that measure the extent
to which a respondent is aligned with a particular
campaign in terms of the perceptions of the contents of the
Treaty leads to the conclusion that these alignments are
strong predictors of the vote in the referendum. The more a
voter’s perceptions of the contents of the Treaty are aligned
with the NO campaign, the more likely the voter is to vote
NO, and vice versa for the YES campaign.
While there are good reasons to expect a direct effect of
a lack of subjective knowledge of the referendum issue at
hand on the vote choice in a referendum (see, e.g., Binzer
Hobolt, 2009), the impact of objective knowledge is indirect.
The indirect effect proceeds through various channels,
not least its impact through subjective knowledge, among
which is the particular susceptibility of uninformed voters
to misleading campaign messages. The analysis demonstrates
this effect in the context of the 2008 Irish referendum
on the Lisbon Treaty.
A rational voter can decide not to invest extensively in
acquiring the knowledge to make a fully informed decision
in a referendum as complex as that on the Treaty of Lisbon
and instead to rely on cues from members of the political
elite. The risk in doing so, however, is that these elites will
have particular political interests that might run counter to
those of the voter (Lupia and McCubbins, 1998).
This article makes two contributions to the resolution of
these issues. The first contribution lies in providing a more
nuanced account of the interaction between knowledge,
attitudes, campaigns and voting behaviour in referendums.
The second contribution focuses on the policymaker’s point
of view and argues that governments need to develop ways
and means of ensuring that electorates have access to real
debate and deliberation in a way that enhances knowledge
of the often complex issues that are set before them in the
form of the referendum.
9 These inferences are, of course, based on largely the same information,
namely that recorded in the question presented extensively in Table
1. That this does not lead to unacceptable levels of multicollinearity is
evidenced by the fact that the coefficients on both the objective knowledge
of the Treaty variable and on the campaign variables are statistically
significant. Furthermore, the significant decrease in the Akaike Information
Criterium, by more than twice the number of variables added
(Gelman and Hill, 2007, 525), suggests that the inclusion of the campaign
variables is a significant improvement of the model. 10 Furthermore, a latent trait analysis not reported in this paper provides
supporting evidence that the differentiation between the two dimensions
in the perceptions of the Treaty is reasonable.
222 J.A. Elkink, R. Sinnott / Electoral Studies 38 (2015) 217e225
Acknowledgements
Previously presented at the New Directions in Referendums:
Politics and Campaigns conference, Dublin, May 9,
2013; 1st Annual General Conference of the European Political
Science Association, Dublin, June 16e18, 2011; and
the ECPR Joint Sessions of Workshops, Lisbon, April 2009.
The authors are grateful to Elisabeth Gidengil, Paul Kellstedt,
Theresa Reidy, Jane Suiter and other workshop participants,
as well as the anonymous reviewer, for their
comments.
Appendix
Own economic situation bad
“What about your own economic situation these days?
Would you say it is … ? ” (very good/fairly good/fairly bad/
very bad/don’t know). Values “fairly bad” and “very bad”
are coded as 1, others as 0.
Dissatisfied with government
“Overall are you generally satisfied or dissatisfied with
the way the government is running the country? ” (very
satisfied/quite satisfied/quite dissatisfied/very dissatisfied/
don’t know). Values “very dissatisfied” and “quite dissatisfied”
are coded as 1, others as 0.
Feels close to opposition
“Do you feel close to any of the political parties? ” and if
no, “Do you feel yourself a little closer to one of the political
parties than the others? ” (FF/FG/Labour/Green/Sinn Fein/
PDs/Other/Not close to any). Values “FF”, “PDs” and “Green”
are coded as 1, others as 0.
Identifies as Irish only
“In the near future, do you see yourself as … ? ” (Irish
only/Irish and European/European and Irish/European only/
don’t know). Value “Irish only” is coded as 1, the rest as 0.
Anti-immigration attitude
“Using the card provided please indicate whether
Ireland is made a worse or a better place to live in by people
coming to live here from other countries? ” (1 worse place
to live … 10 better place to live). The variable is entered in
reverse order, 1 ¼ 10, 2 ¼ 9, etc.
Pro-neutrality attitude
“I have a number of statements here that people
sometimes make. I would like you to indicate on this scale
which of each pair of opposing statements comes closest to
your view. A score of one would indicate that you agree
Table 2
Logistic regression explaining the NO vote. Standard errors in parentheses with statistically significant (a ¼ .05) coefficients in bold print. Coefficients are
standardised as suggested by Gelman (2007). The coefficients on age are relative to the 65þ group; those on class relative to the upper middle class; and
those on education relative to those with only primary education.
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6
Constant 0.02 (0.31) 0.08 (0.35) 0.08 (0.36) 0.11 (0.36) 0.05 (0.37) 0.02 (0.37)
Female 0.17 (0.13) 0.07 (0.14) 0.04 (0.15) 0.04 (0.15) 0.00 (0.16) 0.00 (0.15)
Age 18e24 0.37 (0.28) 0.51 (0.31) 0.40 (0.32) 0.42 (0.32) 0.35 (0.33) 0.33 (0.33)
Age 25e34 0.64 (0.23) 0.67 (0.26) 0.63 (0.27) 0.63 (0.27) 0.54 (0.28) 0.54 (0.28)
Age 35e49 0.24 (0.21) 0.37 (0.24) 0.37 (0.25) 0.37 (0.25) 0.33 (0.25) 0.34 (0.25)
Age 50e64 0.23 (0.21) 0.30 (0.24) 0.32 (0.24) 0.30 (0.25) 0.25 (0.25) 0.28 (0.25)
Lower middle class 0.24 (0.20) 0.15 (0.22) 0.11 (0.23) 0.09 (0.23) 0.10 (0.24) 0.12 (0.24)
Skilled worker 0.78 (0.22) 0.62 (0.25) 0.48 (0.26) 0.43 (0.26) 0.46 (0.27) 0.51 (0.27)
Unskilled worker 0.85 (0.24) 0.58 (0.27) 0.47 (0.28) 0.39 (0.28) 0.44 (0.29) 0.53 (0.28)
Large farmer 0.22 (0.30) 0.48 (0.35) 0.53 (0.38) 0.55 (0.38) 0.48 (0.39) 0.49 (0.39)
Small farmer 0.75 (0.50) 0.37 (0.57) 0.55 (0.60) 0.48 (0.60) 0.44 (0.62) 0.52 (0.62)
Secondary education 0.14 (0.25) 0.23 (0.29) 0.32 (0.30) 0.37 (0.30) 0.41 (0.31) 0.34 (0.31)
Third level education 0.42 (0.28) 0.05 (0.33) 0.16 (0.34) 0.28 (0.34) 0.31 (0.35) 0.17 (0.34)
Own economic situation bad 0.76 (0.15) 0.43 (0.18) 0.31 (0.18) 0.29 (0.18) 0.30 (0.19) 0.33 (0.19)
Dissatisfied with government 0.82 (0.13) 0.70 (0.15) 0.76 (0.16) 0.77 (0.16) 0.75 (0.16) 0.74 (0.16)
Feels close to opposition 0.73 (0.12) 0.86 (0.14) 0.92 (0.15) 0.91 (0.15) 0.94 (0.15) 0.96 (0.15)
Identifies as Irish only 0.49 (0.15) 0.34 (0.16) 0.35 (0.16) 0.44 (0.17) 0.41 (0.16)
Pro-neutrality attitude 0.78 (0.14) 0.76 (0.15) 0.80 (0.15) 0.72 (0.15) 0.68 (0.15)
Anti-immigration attitude 0.56 (0.17) 0.39 (0.17) 0.37 (0.17) 0.31 (0.18) 0.33 (0.18)
EU membership a good thing ¡1.27 (0.20) ¡1.11 (0.20) ¡1.09 (0.20) ¡1.00 (0.20) ¡1.03 (0.20)
Too many issues decided by EU 0.64 (0.15) 0.63 (0.15) 0.62 (0.15) 0.50 (0.16) 0.52 (0.16)
EU means low wages 0.69 (0.19) 0.62 (0.19) 0.62 (0.19) 0.54 (0.19) 0.54 (0.19)
Low objective knowledge of EU 0.27 (0.18) 0.17 (0.19) 0.18 (0.19) 0.30 (0.19)
Low objective knowledge of treaty 1.35 (0.19) 1.25 (0.19) 0.52 (0.31) 0.70 (0.30)
Low subjective knowledge of EU 0.06 (0.20) 0.05 (0.20)
Low subjective knowledge of treaty 0.48 (0.20) 0.60 (0.21)
Positive on NO campaign items 1.18 (0.18) 1.12 (0.18)
Positive on YES campaign items ¡0.87 (0.28) ¡0.82 (0.28)
Number of observations 1301 1254 1254 1246 1246 1254
Akaike information criterion 1557 1288 1223 1215 1172 1184
J.A. Elkink, R. Sinnott / Electoral Studies 38 (2015) 217e225 223
fully with the statement on the left. A score of nine would
indicate that you agree fully with the statement on the
right. Of course your view could be somewhere in between.
Also of course there may be issues that you have no
particular view on. If so, please just say this and we will
move on to the next item.” (1 “Ireland should do everything
it can to strengthen its neutrality even if this means being
less involved in European Union cooperation on foreign
and defence policy” … 9 “Ireland should be willing to
accept limitations on its neutrality so that it can be more
fully involved in European Union cooperation on foreign
and defence policy”). Values are in reverse order, 1 ¼ 9,
2 ¼ 8, etc.
EU membership a good thing
“Generally speaking, do you think that Ireland’s membership
of the European Union (European Community) is …
? ” (“A good thing”/“A bad thing”/“Neither good nor bad”/
“Don’t know”). The first category is coded as 1, the rest as 0.
Too many issues decided by EU
“There has been a lot of discussion recently about the
European Union. Some people say that too many issues are
decided on by the European Union, others say that more
issues should be decided on by the European Union. Which
of the following statement comes closest to your view? ”
(“Too many issues are decided by the European Union”/
“The number of issues decided on by the European Union at
present is about right”/“More issues should be decided by
the European Union”/“I have not really thought about it”/“It
depends on the issue”/“Don’t know”). Those that selected
the first option were coded as 1, others as 0.
EU means low wages
“What does the European Union mean to you personally?
” (many answer categories, among which “Lower
wage rates”). Those that included “Lower wage rates” in
their selection were coded as 1, others as 0.
Low objective knowledge of Treaty
Number of items deemed to be correct in Table 1 that
are perceived to be in the Treaty; ‘don’t know’ taken as an
incorrect answer.
Low objective knowledge of EU
“For each of the following statements about the European
Union could you please tell me whether you think it is
true or false? ” (“The EU currently consists of fifteen
Member States”/“Switzerland is a member of the European
Union”/“Every six months, a different Member State becomes
the President of the Council of the European Union”/
“The members of the European Parliament are directly
elected by the citizens of the EU”). A four point score is
generated, from 1 (4 correct answers) to 4 (1 correct
answer). All respondents had at least one correct answer.
Low subjective knowledge of Treaty
“By the date of the referendum (12 June), how good was
your understanding of the issues involved? Please use this
card to choose the phrase that applies best to you.” (“I had a
good understanding of what the Treaty was all about”/“I
understood some of the issues but not all that was
involved”/“I was only vaguely aware of the issues
involved”/“I did not know what the Treaty was about at
all”/don’t know). The resulting variable is a four point scale
from good understanding to not knowing what the Treaty
was about at all.
Low subjective knowledge of EU
“And how about the European Union in general? Using
this scale, how much do you feel you know about the European
Union, its policies, its institutions? ” (1 nothing at all
… 10 know a great deal).
Positive on NO campaign items
Number of items associated with the NO campaign in
Table 1 that are perceived to be in the Treaty; ‘don’t know’
taken as missing data.
Positive on YES campaign items
Number of items associated with the YES campaign in
Table 1 that are perceived to be in the Treaty; ‘don’t know’
taken as missing data.
Abstained
“On the 12 June last, a referendum was held on the
Treaty of Lisbon. As you may remember, many people did
not vote in that referendum. How about you? Did you vote
in the referendum on the Treaty of Lisbon that was held in
June? ” (“Yes e voted”/“Did not vote”). In all analyses, this is
used as a filter variable, and only those who voted are taken
into account.
Voted NO
“How did you vote in that referendum e in favour or
against the Lisbon Treaty? ” (“Voted e In favour”/“Voted e
Against”). This is the dependent variable in the regression
analyses.
References
Ajzen, I., Fishbein, M., 1977. Attitude-behavior relations: a theoretical
analysis and review of empirical research. Psychol. Bull. 84 (5),
888e918.
Alvarez, R.M., 1997. Information and Elections. University of Michigan
Press.
Basinger, S.J., Lavine, H., 2005. Ambivalence, information, and electoral
choice. Am. Polit. Sci. Rev. 99, 169e184.
Binzer Hobolt, S., 2005. When Europe matters: the impact of political
information on voting behaviour in EU referendums. J. Elections
Public Opin. Parties 15 (1), 85e109.
Binzer Hobolt, S., 2009. Europe in Question: Referendums on European
Integration. Oxford University Press, New York.
224 J.A. Elkink, R. Sinnott / Electoral Studies 38 (2015) 217e225
Bowler, S., Donovan, T., 1994. Information and opinion change on ballot
propositions. Polit. Behav. 16 (4), 411e435.
Campbell, A., Converse, P.E., Miller, W.E., Stokes, D., 1960. The American
Voter. University of Chicago Press, Chicago.
De Vries, C.E., Van der Brug, W., Van Egmond, M.H., Van der Eijk, C., 2011.
Individual and contextual variation in EU issue voting: the role of
political information. Elect. Stud. 30 (1), 16e28.
Delli Carpini, M.X., Keeter, S., 1996. What Americans Know about Politics
and Why it Matters. Yale University Press.
Druckman, J.N., 2004. Political preference formation: competition,
deliberation, and the (ir)relevance of framing effects. Am. Polit. Sci.
Rev. 98 (4), 671e686.
Druckman, J.N., Nelson, K.R., 2003. Framing and deliberation: how citizens’
conversations limit elite influence. Am. J. Polit. Sci. 47 (4),
729e745.
Ehin, P., 2001. Determinants of public support for EU membership: data
from the Baltic countries. Eur. J. Polit. Res. 40, 31e56.
Franklin, M., 2002. Learning from the Danish case: a comment on Palle
Svensson’s critique of the Franklin thesis. Eur. J. Polit. Res. 41,
751e757.
Franklin, M., Marsh, M., Wlezien, C., 1994. Attitudes towards Europe and
referendum votes: a response to Siune and Svensson. Elect. Stud. 13,
117e121.
Gabel, M.J., 1998. Economic integration and mass politics: market liberalisation
and public attitudes in the European Union. Am. J. Polit. Sci.
42 (3), 936e953.
Gamble, A., 2006. The European disunion. Br. J. Polit. Int. Relat. 8 (1), 34e49.
Garry, J., Marsh, M., Sinnott, R., 2005. ‘Second-order’ versus ‘issue-voting’
effects in EU referendums. Eur. Union Polit. 6 (2), 201e221.
Gelman, A., 2007. Scaling regression inputs by dividing by two standard
deviations. Stat. Med. 27 (15), 2865e2873.
Gelman, A., Hill, J., 2007. Data Analysis Using Regression and Multilevel/
Hierarchical Models. Analytical Methods for Social Research. Cambridge
University Press, Cambridge.
Grynaviski, J.D., Corrigan, B.E., 2006. Specification issues in proximity
models of candidate evaluation (with issue importance). Polit. Anal.
14, 393e420.
Hix, S., 2006. Why the EU Needs Left-right Politics. In: Politics: the Right or
the Wrong Sort of Medicine for the EU?. Notre Europe Policy Paper 19.
Hooghe, L., Marks, G., 2009. A postfunctionalist theory of European
integration: from permissive consensus to constraining dissensus. Br.
J. Polit. Sci. 39 (1), 1e23.
IMS, 2008. Post Lisbon Treaty Referendum Research Findings (Technical
report Millward Brown-IMS).
Kastellec, J.P., Leoni, E.L., 2007. Using graphs instead of tables in political
science. Perspect. Polit. 5 (4), 755e771.
King, G., 1998. Unifying Political Methodology. The Likelihood Theory of
Statistical Inference. University of Michigan Press.
Krosnick, J.A., 1988. Attitude importance and attitude change. J. Exp. Soc.
Psychol. 24, 240e255.
Laffan, B., 2008. Reflection: three readings of the failure to ratify the
constitutional treaty. In: Brown, T. (Ed.), Lisbon: What the Reform
Treaty Means. Institute of International and European Affairs, Dublin,
pp. 60e70.
Lord, C.G., Ross, L., Lepper, M.R., 1979. Biased assimilation and attitude
polarization: the effects of prior theories on subsequently considered
evidence. J. Personal. Soc. Psychol. 37 (11), 2098e2109.
Lupia, A., 2006. How elitism undermines the study of voter competence.
Crit. Rev. 18 (1), 217e232.
Lupia, A., McCubbins, M.D., 1998. The Democratic Dilemma: Can Citizens
Learn What They Need to Know? Cambridge University Press, New
York.
Miller, J.M., Krosnick, J.A., 2000. News media impact on the ingredients of
presidential evaluations: politically knowledgeable citizens are
guided by a trusted source. Am. J. Polit. Sci. 44 (2), 301e315.
Morgan, S.L., Winship, C., 2007. Counterfactuals and Causal Inference.
Methods and Principles for Social Research. Cambridge University
Press, New York.
Nie, N.H., Verba, S., Petrocik, J.R., 1979. The Changing American Voter.
Harvard University Press Cambridge.
Palfrey, T.R., Poole, K.T., 1987. The relationship between information,
ideology, and voting behavior. Am. J. Polit. Sci. 31 (3), 511e530.
Pearl, J., 2000. Causality: Models, Reasoning, and Inference. Cambridge
University Press, Cambridge.
Perloff, R.M., 2003. The Dynamics of Persuasion. Communication and
Attitudes in the 21st Century, second ed. Lawrence Erlbaum Associates,
Mahwah, NJ.
Quinlan, S., 2009. The Lisbon treaty referendum 2008. Ir. Polit. Stud. 24
(1), 107e121.
Reif, K., Schmitt, H., 1980. Nine second order national elections: a conceptual
framework for the analysis of European election results. Eur. J.
Polit. Res. 8 (1), 3e44.
Sinnott, R., 2001. Attitudes and Behaviour of the Irish Electorate in the
Referendum on the Treaty of Nice (Technical report, European Commission
Representation in Ireland).
Sinnott, R., 2003. Attitudes and Behaviour of the Irish Electorate in the
Second Referendum on the Treaty of Nice (Technical report, European
Commission Representation in Ireland).
Sinnott, R., Elkink, J.A., O’Rourke, K., McBride, J., 2009. Attitudes and
Behaviour in the Referendum on the Treaty of Lisbon (Technical
report, Department of Foreign Affairs).
Suiter, J., Reidy, T., 2013. It’s the campaign learning stupid, an examination
of a volatile Irish referendum. Parliam. Aff. (published online).
Van Apeldoorn, B., 2009. The contradictions of ’embedded neoliberalism’
and Europe’s multi-level legitimacy crisis: the European project and
its limits. In: Van Apeldoorn, B., Drahokoupil, J., Horn, L. (Eds.), Contradictions
and Limits of Neoliberal European Governance. From
Lisbon to Lisbon. Palgrave MacMillan, pp. 21e43.
Zaller, J., 1992. The Nature and Origins of Mass Opinion. Cambridge University
Press, Cambridge.
J.A. Elkink, R. Sinnott / Electoral Studies 38 (2015) 217e225 225

Our academic experts are ready and waiting to assist with any writing project you may have. From simple essay plans, through to full dissertations, you can guarantee we have a service perfectly matched to your needs.

GET A 40% DISCOUNT ON YOU FIRST ORDER

ORDER NOW DISCOUNT CODE >>>> WELCOME40

 

 

Posted in Uncategorized