Fear, anger and enthusiasm about the EU:
Effects of emotional reactions on public preferences towards European integration
Sofia Vasilopoulou, Department of Politics, University of York, York, UK
Markus Wagner, Department of Government, University of Vienna, Vienna, Austria
How do emotions affect public opinion on the EU? This article advances existing literature
that focuses on cue-taking, utilitarianism and identity by arguing that emotional reactions are
important to understanding citizen EU attitudes. This is because discrete emotions such as
fear, anger and enthusiasm affect how individuals deal with threats and how they seek out,
process and use information. We hypothesise that, compared to anxious citizens, those angry
with the EU are more likely to wish to leave the EU, less receptive to cost-benefit
considerations and less nuanced in their opinions about integration. Our analyses, carried out
using a survey conducted in the UK in April 2015 (n=3.000), support our hypotheses. These
results help us predict the effectiveness of political strategies, e.g. in referendum campaigns.
Keywords: Anger, emotional reactions, enthusiasm, Euroscepticism, fear
Sofia Vasilopoulou, Department of Politics, University of York, Heslington, York, YO10
What explains public preferences towards European integration? A review article of EU
public opinion identifies three prominent approaches: cue taking, utilitarianism and identity
(Hobolt and de Vries, 2016). The first approach postulates that citizens evaluate the EU
through proxies, such as national institutions (e.g. Anderson, 1998; Armingeon and Ceka,
2014) or party cues (Steenbergen et al., 2007). The utilitarian approach posits that support for
integration is a function of economic utility related to either the individual’s cost-benefit
analysis of European integration or the country’s economy (e.g. Anderson and Reichert, 1995;
Gabel and Whitten, 1997; Gomez, 2015). The identity approach suggests that European
identity may act as a buffer against hostility towards the EU. In contrast, exclusive
conceptions of national identity and perceptions of threat posed by other cultures are
associated with opposition to the EU (e.g. Carey, 2002; Hooghe and Marks, 2009; McLaren,
2002; see also Kentmen-Cin and Erisen (2017) on anti-immigrant attitudes).
This article advances the literature by taking into account individuals’ emotional
reactions to the EU. Individuals’ discrete emotions such as anger, fear and enthusiasm affect
how they deal with threats, how they form preferences, and how they seek out, process and
use information. Anger activates a different mental system from fear (Cacioppo et al., 1999;
MacKuen et al., 2010; Marcus and MacKuen, 1993; Weber, 2012). Anger activates the
approach/disposition system, lowers risk estimates, and leads individuals to try to remove
perceived threats while relying on instinctive routines. In contrast, fear is related to the
avoidance/surveillance system, makes individuals more risk-averse and cautious, and leads to
an increased search for information (Brader et al., 2010; Carver and Harmon-Jones, 2009;
Druckman and McDermott, 2008; Lerner and Keltner, 2000, 2001; Valentino et al., 2009,
2011; Smith et al., 2008). Enthusiasm is different from anger and fear due to its positive
valence. In terms of its behavioural consequences, enthusiasm is associated with positive
goal-oriented behaviour, but tends to be more similar to anger than to fear in that it tends to
reinforce people’s habitual behaviours and existing political choices (e.g. partisanship),
minimises perceived risks and reduces reliance on information (Isbell et al., 2006; Lerner and
Keltner, 2001; Marcus et al., 2000; Valentino et al., 2008; see also Ladd and Lenz, 2008).
Despite the fact that emotions are a key topic in the literature on political behaviour,
only one published study has examined their role in understanding EU attitudes, specifically
in the context of vote choice in the Irish referendum on the fiscal compact treaty (Garry,
2014). The study finds that emotional reactions tend to condition the extent to which voting is
driven by attitudes towards domestic politics. Risky framing matters for vote choice: because
voting against the treaty was successfully painted as the riskier option, fearful citizens were
more likely to vote yes, with angry voters more likely to choose no. Garry’s analysis shows
that instrumental, issue-based considerations such as the perceived economic consequences of
the vote mattered more to fearful voters, while angry voters decided more on the basis of
‘second-order’ concerns such as partisanship or government performance.
Our argument builds directly on Garry (2014) and extends his argument by
considering the direct and moderating impact of emotional reactions on preferences towards
European integration. We expect emotions and affect to be ‘both a motivational component
underlying information processing strategies and a direct source of information that
individuals consult in making social judgments’ (Erişen, 2013: 117). Emotions concerning the
EU are likely to be influential at three stages of opinion formation: as a direct source of
opinions, as a moderator of cognitive considerations and as a moderator of information-
seeking strategies. First, emotional reactions exert a distinct direct influence on support for
integration because different emotional reactions influence how individuals deal with
perceived threats (Druckman and McDermott, 2008). We argue that those who experience
anger should be the most likely to want to leave the EU. In contrast, fearful voters should be
those who most want to renegotiate their country’s relationship with the EU. Enthusiastic
citizens should unsurprisingly be least in favour of either option since continued membership
on current terms will be attractive. Second, emotional reactions affect how individuals use
other considerations to form opinions on EU integration (e.g. Garry, 2014; Valentino et al.,
2011). We posit that instrumental considerations are more directly related to support for EU
membership for fearful citizens, while underlying affect towards the EU is more important for
those experiencing anger. Finally, emotional reactions also influence how individuals seek out
and process information (e.g. Brader et al., 2008; Valentino et al., 2008; Merolla and
Zechmeister, 2009). Because angry voters tend to be less active in finding and considering
information, we suggest that they will have less varied and nuanced opinions regarding
European integration than fearful voters.
Using evidence from an online survey conducted in the UK in April 2015 (n=3.000),
our study has a three-fold contribution. First, we put forward an overall theoretical model of
how emotional reactions affect public preferences on the EU, showing that emotions are
important to understanding three distinct stages of opinion formation. In proposing this
model, we offer an entirely new perspective by considering how emotional reactions affect
the extent to which opinions on EU integration are varied and nuanced. Second, we examine
general preferences concerning a member state’s relationship with the EU rather than voting
decisions in a referendum. By incorporating emotionality into our empirical analysis we
demonstrate that discrete emotional reactions have divergent effects on people’s preferences
towards European integration. Finally, we also examine the distinct effects of enthusiasm,
showing that its impact is more similar to that of anger than to that of fear. Our findings are
robust because our data allow us to incorporate and control for the major explanations in the
literature. In sum, we provide an important complement to existing work explaining support
for EU integration (Hobolt and de Vries, 2016), and contend that future work explaining
preferences over EU integration should integrate emotional reactions.
Emotional reactions and opinions about European integration
Early research into emotional reactions either used one simple positive-negative dimension or
a two-dimensional valence model (Marcus, 2003; Marcus et al., 2000). More recently,
researchers have argued that distinct emotions may characterise individuals’ reactions (e.g.
Huddy et al., 2007; Larsen and McGraw, 2011; Petersen, 2010; Smith et al., 2008; Valentino
et al. 2008, 2011; Weber, 2012; see also Conover and Feldman, 1986). Appraisal theories of
emotions suggest that various distinct emotions arise due to individuals’ cognitive
interpretation of the situation they experience (e.g. Roseman, 1991; Smith and Ellsworth,
1985; for a review, see Brader and Marcus, 2013). Neuroscientific theories based on the
neural processes that generate emotional responses have also increasingly turned towards a
three-dimensional model of emotions that distinguishes between fear, anger and enthusiasm
(Brader and Marcus, 2013; Lerner and Keltner, 2001). Research in political psychology
building on both appraisal and neural process theories has therefore focused in particular on
the causes and consequences of anger, fear1 and enthusiasm as distinct emotions (Carver and
Harmon-Jones, 2009; Conover and Feldman, 1986; Huddy et al., 2005, 2007; Lerner and
Keltner, 2000, 2001; Petersen, 2010; Smith and Ellsworth, 1985; Valentino et al., 2011).
Applied to political phenomena, distinct emotions cause different types of actions and
attitudes: fear leads to caution and openness and anger to more confident and aggressive
responses (Druckman and McDermott, 2008; MacKuen et al., 2010). For instance, whereas
fear and anger both tend to increase intolerance and threat perceptions (e.g. Erisen and
Kentmen-Cin, 2017), they lead to distinct ways of reacting to terrorist threats (Lerner et al.,
2003). Fearful citizens are more wary of the increased risk and turn to isolationism, while
angry citizens are less risk-averse and tend to favour pro-active intervention (Huddy et al.,
2007). Anxious citizens are also more likely to seek out new information and process this
carefully (Brader et al., 2008; Marcus et al., 2000; Merolla and Valentino et al., 2008;
Zechmeister, 2009). Finally, enthusiasm increases the willingness to participate in politics, but
like anger it also increases less careful forms of information-processing (Brader, 2006;
Marcus et al., 2000).
The process of European integration is likely to elicit distinct emotional responses
from citizens. The EU is a complex institutional framework that has expanded its
jurisdictional authority over a number of key policy areas. In terms of the negative emotions
elicited by the EU, we might consider aspects such as beliefs about the negative impact of the
EU, perceptions of one’s influence over political decisions and assumptions about the
motivations of EU actors. For some citizens, transposing sovereignty to the EU may therefore
be perceived as a threat to domestic economic stability, cultural homogeneity, and national
identity and sovereignty. For other individuals, the EU may inspire enthusiasm by appealing
to common identities, solidarity and belonging as well as projects associated with economic
prosperity. Those who view the EU with enthusiasm may associate it with positive past
experiences as well as a feeling of future positive impact, coupled with perceptions of
influence and control over those circumstances.
We suggest that these emotional reactions can subsequently influence support for
European integration at three stages of the opinion formation process. First, emotional
reactions can have a direct influence on attitudes towards EU membership; second, emotional
reactions can affect the impact of other evaluations of the EU on opinions and attitudes about
integration; and finally, emotional reactions can influence how individuals seek out and
process information about the EU (Figure 1).
Figure 1 about here
Our first set of hypotheses concerns how emotional reactions directly affect support
for EU integration, so whether and how individuals want to re-order their country’s
relationship with the EU.2 Angry citizens should be more in favour of cutting ties with the
EU. This is because angry individuals are more confrontational and less risk-averse. Anger
leads to more confident and aggressive responses to a given threat that are meant to remove it
completely (Druckman & McDermott, 2008). In contrast, anxious citizens are more open to
compromise than angry individuals (Huddy et al., 2005; MacKuen et al., 2010). Anxious
citizens also tend to perceive higher levels of risk, coupled with a greater sense of uncertainty
and lack of control (Huddy et al., 2005). Hence, they are also generally more risk-averse,
preferring caution over radicalism. Anxious citizens should therefore be more likely to
support re-negotiation of their country’s relationship with the EU. Finally, while enthusiasm
is generally linked to political interest and mobilisation (Marcus et al., 2000), its status as a
positive emotion means that these citizens should not want to leave the EU or renegotiate the
terms of membership. Our first three hypotheses are thus:
H1: Anxious citizens are more likely to favour renegotiation than angry citizens (1a), and
angry citizens are more likely to favour leaving the EU than anxious citizens (1b), while
enthusiastic citizens are less likely to favour renegotiation or leaving the EU than anxious or
angry citizens (1c).
Our second set of hypotheses concerns the effect of emotional reactions on how
individuals use broader evaluations of the EU to form their opinion about European
integration. Anger and enthusiasm lead citizens to take detailed and balanced information less
into account: their opinions are formed more heuristically (Garry, 2014; MacKuen et al.,
2010). In contrast, anxious individuals are less likely to rely on habitual responses and more
likely to form their judgements based on careful consultation of the available information
(Garry, 2014; MacKuen et al., 2010; Valentino et al., 2009). The surveillance system
activated by anxiety leads citizens to ‘stop and seek [relevant] information’ (Merolla and
Zechmeister, 2009). As such, the opinions of fearful citizens should be based more on
If different emotions lead to different kinds of information-seeking and information-
processing on European issues, then the resulting EU attitudes should also vary in their
relationship with utilitarian concerns and affective evaluations. The EU stances of fearful
citizens should be more strongly linked to cost-benefit analyses of the EU. We know that
utilitarianism in the form of perceived costs and benefits of European integration provides a
robust explanation of people’s EU attitudes, with those who perceive that integration benefits
themselves and/or their country more likely to support the EU (e.g. Gabel, 1998; Gabel and
Whitten, 1997). In contrast, angry and enthusiastic citizens may base their stance less on their
perceptions of the costs and benefits of the EU. Such citizens should base their overall
opinions more on affect-based evaluations of the EU (Garry, 2014; MacKuen et al., 2010).
When asked about their opinion regarding political matters, angry and enthusiastic individuals
tend to rely on heuristics based on their overall views (Brader and Marcus, 2013: 185).
H2: The attitudes of angry and enthusiastic citizens towards the EU are less conditional on
utilitarian considerations concerning EU integration than those of anxious citizens (2a), and
the attitudes of angry and enthusiastic citizens towards the EU are more conditional on affect-
based considerations concerning EU integration than those of anxious citizens (2b).
Finally, emotional reactions may affect how individual seek out and process
information. On the one hand, anger and enthusiasm lead individuals to have stronger
opinions and to be less open in their search for more information. There is evidence that anger
in particular leads individuals to seek out less information and to look for sources that tend to
confirm their pre-existing opinions (MacKuen et al., 2007, Valentino et al., 2008, 2009). In
contrast, anxiety and fear lead individuals to engage in more careful information-seeking
behaviour (Brader et al., 2008; Huddy et al. 2007; Marcus et al., 2000, Valentino et al., 2008).
This is because this emotion leads individuals to be more deliberate and open in forming
opinions. Applied to the EU, this would mean that angry citizens focus more on negative
news and spend less time searching for information than anxious citizens; in contrast,
enthusiastic citizens will focus on positive news, but will also work less hard at integrating
We do not test directly how emotions moderate information-seeking and information-
processing. Instead, we take advantage of the empirical consequences that should be
observable if emotions moderate citizens’ cognitive behaviour. We argue that the impact of
emotions on information-seeking influences how nuanced and differentiated citizen opinions
are. The EU is a complex institutional arrangement that covers a wide array of different policy
areas. It is reasonable to support EU integration in some areas more than in others. However,
emotions will affect the extent to which individuals want, look for and process information
about the EU. If anger or enthusiasm lead to less information-seeking relevant to integration,
citizens associating the EU with those emotions will have more uniform opinions about EU
integration across policy areas. In contrast, anxiety is likely to lead individuals to seek out
more information, which should also lead fearful individuals to be less uniform in their
support or opposition across policy areas. Of course, some anxious voters will still develop
uniform views on the EU, but their tendency to seek out more information should on average
counteract these tendencies, at least compared to angry and enthusiastic voters.
H3: Angry and enthusiastic citizens’ opinion on European integration is likely to be more
uniform across policy areas than that of anxious citizens.
Data and methods
We rely on data collected from a large-n cross-sectional online survey (n=3000) conducted in
the UK in April 2015, about 14 months before the Brexit referendum. The UK provides an
ideal environment to test the effect of emotional reactions on public preferences towards
European integration as the debate in light of the Brexit referendum increased the
politicisation of the EU, defined as ‘higher levels of salience, polarisation of opinion and the
expansion of actors and audiences involved in EU issues’ (De Wilde et al., 2016: 3).
ResearchNow recruited our participants using online panels of approximately 500000
respondents from a wide variety of Internet sites to avoid the bias associated with limited
source recruitment. Our survey is representative of the British population in terms of gender,
age and region (see Online appendix). Online surveys tend not to be fully representative of
populations in terms of education and income; however, in-person and online survey data tend
to yield similar results both in terms of estimating parameters and the overall explanatory
power of competing models (e.g. Sanders et al., 2007), so we consider this to be a limited
problem for our analysis. We control for education and income in our models, also accounting
for the fact that the less-educated and less well-off citizens voted in favour of Brexit (Hobolt,
Measures of EU support
Our analyses use two questions that measure different aspects of opinion about the EU. First,
we capture an individual’s willingness to renegotiate EU membership based on answers to the
agree-disagree statement ‘The UK should renegotiate the terms of its EU membership’.
Second, we examine an individual’s opinion on whether the UK should leave the EU based on
the agree-disagree statement ‘Irrespective of renegotiation, the UK should leave the EU’.
Answers to both questions were measured on a seven-point Likert scale ranging from 1
(‘strongly disagree’) to 7 (‘strongly agree’). The points in between were labelled: ‘disagree’,
‘slightly disagree’, ‘neither agree nor disagree’, ‘slightly agree’ and ‘agree’. While the
majority of respondents wish to renegotiate the terms of the UK’s EU membership, opinion is
much more varied when it comes to leaving the EU (Figure 2).
[Figure 2 about here]
To measure the degree of attitude uniformity across EU policies, we asked
respondents a series of questions to capture how much authority they think the EU should
have in 19 policy areas. Specifically, we asked ‘To what extent do you agree or disagree that
the EU should have more authority over the EU Member States in the following policy
areas’.3 Agreement was again measured on seven-point scales. To assess attitude uniformity
we calculated the standard deviation of each individual’s responses across these 19 policy
Measures of emotional reactions
Our survey asked: ‘Which, if any, of the following words describe your feelings about
Britain’s membership of the EU (choose up to four words).’ As in the British Election Study
(e.g. Whiteley et al., 2014), respondents could choose a maximum of four emotions from a list
of nine: angry, disgusted, uneasy, afraid, happy, hopeful, confident, proud and indifferent.
The most prominent reaction to Britain’s membership of the EU is ‘uneasy’ at 49.3 per cent
followed by ‘hopeful’ at 25.3 per cent. The least chosen emotion is ‘proud’, at 7.4 per cent
(Figure 3). About a quarter of respondents say that they feel ‘indifferent’ about Britain’s EU
[Figure 3 about here]
We combine these emotions into four groups (Figure 3).4 We categorised individuals
as afraid if they ticked either uneasy or afraid or both (Wagner 2014) and as angry if they
ticked either angry or disgusted or both.5 Individuals are categorised as enthusiastic if they
choose ‘happy’, ‘hopeful’, ‘confident’ or ‘proud’ (Marcus and Mackuen, 1993). While these
terms may seem to tap into different emotions, they tend to be very similar in terms of their
consequences (Brader and Marcus 2013: 175). We measure indifference as those people who
ticked that box; most people who are indifferent chose no other emotions.
The design of the survey question means that respondents can state that they have
different feelings at the same time, e.g. anger and fear. This is implied by and consistent with
current approaches to studying emotions (Carver and Harmon-Jones 2009: 197; Tellegen et al.
1999). In our survey, 36 per cent reported feeling fear and not anger, 6 per cent reported
feeling anger but not fear, and 16 per cent reported anger and fear.6
In addition to examining the direct impact of these emotions (H1a-H1c), we also aim
to test whether the presence of certain emotions moderates the impact of other attitudes on
preferences concerning European integration. First, we hypothesised that the effect of cost-
benefit considerations varies by emotional reaction (H2a). To test this, we interact our
emotion indicators with a variable measuring the perceived costs and benefits of membership.
This is assessed using answers to the agree-disagree statement ‘Britain has greatly benefited
from being a member of the EU’; higher levels of this seven-point variable indicate lower
Second, we suggested that the effect of the general underlying stance towards the EU
also differs by emotional state (H2b). To test this, we interact our emotion indicators with a
variable measuring overall support for the EU; this is measured by extracting the first
component of a principal components analysis of three items: first, how much integration the
respondent would like to see, measured on a 0-10 scale where 0 means ‘the integration of
Europe has already gone too far’ and 10 that ‘European integration should be pushed further’;
second, trust in EU institutions, measured on a 0-10 scale where 0 means ‘do not trust at all’
and 10 means ‘trust completely’; and, third, satisfaction with democracy in the EU, measured
on a 0-10 scale where 0 means ‘completely dissatisfied’ and 10 means ‘completely satisfied’.
This variable captures a respondent’s overall positive or negative evaluation of the EU.
To assess the effects of emotions, our models have to include important confounders.
Emotional reactions towards the EU may be caused by characteristics and attitudes that also
predict attitudes towards the EU. For instance, an individual who opposes immigration may
be more likely to be both angry at Britain’s EU membership and more likely to support
Britain leaving the EU, without anger having a direct effect on the outcome variable at all.
First, partisanship and party cues are pivotal in structuring people’s EU attitudes (e.g.
Steenbergen et al., 2007) but also influence emotions (Ladd and Lenz, 2008). We include
variables for party sympathy for Conservatives, UKIP, Labour, Liberal Democrats, Greens
and any other party. Respondents were asked to assess, on a scale of 0-10: ‘How likely is it
that you would ever vote for this party?’. This is known as the propensity-to-vote question;
results do not differ if we use a standard party identification question instead.
Another potential concern is that our emotions measures are proxies for attitudes
concerning EU integration. We therefore include those attitudes in our models to account for
these confounders. We control for the items that make up our EU opinion scale used to test
H2, i.e. how much integration the respondent would like to see, trust in national and EU
institutions (e.g. see Anderson, 1998; Armingeon and Ceka, 2014); and a respondent’s
satisfaction with democracy in the EU. We also include a measure of European identity in our
models, with higher values indicating strong feelings of European identity (e.g. Carey, 2002;
Our controls also comprise variables related to the utilitarian approach to European
integration (e.g. Gabel, 1998; Gabel and Whitten, 1997; Gomez, 2015). These include
questions that capture an individual’s personal economic evaluation and her assessment of the
country’s economic situation. These are measured from 1 to 5, with lower values indicating
improvement compared to 12 months ago. We control for the respondent’s attribution of
responsibility to the EU for her country’s economic situation, measured on a scale from 0 to
10, with higher values indicating full attribution of responsibility to the EU. Next, we know
that cognitive mobilisation influences an individual’s view of European integration (Inglehart,
1970), so we control for education levels. This four-point variable, with 1 ‘less than secondary
school exams’, 2 ‘secondary school exams (e.g. GCSEs or equivalent)’, 3 ‘college exams (e.g.
A-Levels, NVQ, or equivalent)’ and 4 ‘university (e.g. Degree, Professional qualifications)’.
We also add controls that relate to ideology (Hooghe and Marks, 2009), including an
individual’s self-placement of the left-right dimension; attitudes on state intervention and
redistribution, with higher values indicating leftist attitudes; attitudes on immigration, with
larger values denoting right-wing attitudes; and attitudes towards addressing climate change,
with higher values indicating a positive stance. Finally, other individual-level controls include
gender, age, income and the frequency of following current affairs on TV, the internet and in
Direct effects on support for EU membership
Table 1 below reports our findings from three models using ordinary least squares (OLS)
regression to predict attitudes towards renegotiation (Model 1) and leaving the EU (Models 2
and 3). First, we test whether anxious citizens are more likely to favour renegotiation than
angry citizens. We expect this because anxious and fearful citizens tend to be more risk-
averse and more willing to accept compromise compared to angry citizens who tend to be
much more confrontational. Model 1 supports this hypothesis. In addition to emotional
reactions, Model 1 includes our set of controls, including attitudes towards leaving the EU; it
is necessary to control for this to isolate preferences for renegotiation as distinct from leaving
[Table 1 about here]
Model 1 provides evidence that fearful and anxious citizens wish to renegotiate the
terms and conditions of the country’s EU membership, with enthusiastic citizens less likely to
do so. Note that the variables for each emotional reaction can change independently from each
other, as different emotions can co-occur. The effect for anger is not statistically significant; it
is, however, statistically significantly different from the effect of anxiety (p<0.001). The
effects of emotions on preferences over renegotiation are shown in Figure 4, which shows the
predicted level of support for renegotiating EU membership on a 1-7 scale. The Figure shows
that anxious respondents most strongly support renegotiation, followed by angry and
enthusiastic citizens. Perhaps angry and enthusiastic voters are more likely to have already
made their mind up with regard to the UK’s EU membership, so that renegotiation no longer
matters for their decision. Indifferent voters are similar to angry and enthusiastic voters.
[Figure 4 about here]
This resonates well with our findings in Model 2, where we test our second
hypothesis, i.e. that angry citizens are more likely to favour leaving the EU than anxious
citizens. In Model 2, we see a strong relationship between anger and the wish to
fundamentally change the UK’s constitutional relationship with the EU. Enthusiastic citizens
are less likely to want to leave the EU. Fear has a small but statistically significant positive
effect on wanting to leave the EU compared to enthusiasm; moreover, anxious citizens are
statistically significantly less likely to support leaving than angry citizens. Indifferent voters
are similar to anxious voters.
Our findings related to the relationship between emotions on the one hand and
willingness to renegotiate and preference for leaving the EU on the other are robust. Our
models include a number of variables controlling for the main confounders of the relationship
between emotional reactions and attitudes towards the EU. Party sympathy does not seem to
have a strong effect across the models, with two exceptions: those who support the openly
Eurosceptic United Kingdom Independent Party (UKIP) are more likely to wish to leave the
EU; and those who support the Conservatives are more likely to support renegotiation. Our
variables relating to the EU also have a mixed impact. For example, lower perceived benefits
decrease the willingness to renegotiate and increase support for leaving the EU. The
attribution of responsibility variable is positive and significant across models, i.e. the more an
individual attributes responsibility to the EU for the UK’s economic situation, the more likely
they are to support both renegotiation and withdrawal. Higher EU trust, satisfaction with EU
democracy and support for deeper integration also decrease the desire for renegotiation, but
these variables have a far weaker effect on wanting to the leave the EU. Finally, we find that
there is a relationship between willingness to renegotiate and leaving the EU, i.e. individuals
who want to renegotiate are more likely to wish to leave the EU and vice versa.
Our results show that higher levels of education mean a voter is more likely to wish to
renegotiate the UK’s EU membership but less likely to wish to leave (see Hobolt 2016).
Among educated voters, a preference for renegotiation does not necessarily lead to a desire
for withdrawal. Subjective perceptions of one’s personal economic situation compared to last
year are associated with a wish to leave the EU, but there is no effect on willingness to
renegotiate. Negative perceptions of the country’s economic situation are related with less
support for renegotiation, but have only a weak positive effect on wish for withdrawal.
Ideological dimensions relating to the economy and immigration are also important, and we
note that the immigration scales in general have a larger impact than the economic scales.
Finally, higher levels of trust in the UK government increase the desire both to renegotiate
and to leave the EU.
Emotions as moderators of cost-benefit and affective considerations
Our second set of hypotheses concerned how emotions moderate the impact of general EU
attitudes and cost-benefit perceptions. To test these hypotheses, we interact cost-benefit
perceptions and overall EU attitudes with our emotion indicators. Figure 5 shows the marginal
effect of each attitude for angry, fearful, enthusiastic and indifferent voters, respectively.9 The
dependent variable is willingness to leave the EU. The top half of the Figure shows the
marginal effect of general EU attitudes depending on emotional reactions, while the bottom
half shows the marginal effect of cost-benefit perceptions. In order to read Figure 5 correctly,
note that, unlike in Figure 4, the x-axis is the predicted marginal effect of cost-benefit
perceptions and general attitudes; in Figure 4, the x-axis is the predicted level of the
Turning first to cost-benefit perceptions (bottom half of Figure 5), we can see that the
marginal effect of this variable is clearly larger for anxious than for angry or enthusiastic
voters. While cost-benefit perceptions are important for voters with all emotional reactions,
the effect is about twice as large for anxious than for angry or enthusiastic individuals.
Indifferent voters are between anxious voters on the one hand and angry or enthusiastic voters
on the other. Overall, H2a is confirmed.
The top half of Figure 5 shows that the effect of general EU evaluations is negative for
angry individuals, who are the only group for whom the marginal effect is statistically
significantly smaller than 0. Enthusiasm appears to increase the positive effect of general EU
attitudes. The moderating effect of enthusiasm also statistically differs from that of anxiety
(F=6.09, p=0.01), which itself is only slightly positive and far from statistical significance.
Finally, there is no clear moderating effect of indifference. We can confirm H2b: overall, it
seems that angry and enthusiastic individuals are indeed more likely to rely on heuristic
decision-making and therefore on general EU opinions, while anxious citizens base their
decisions more on cost-benefit considerations.
[Figure 5 about here]
Effects on information-seeking
To test the third hypothesis, we present indicative evidence that fear and anger differ in their
effects on information-seeking and opinion formation. In Table 2, Model 1 predicts the
standard deviation of opinions concerning integration in nineteen policy areas. We include our
controls differently in this model: we control for the extremity of opinion, since those with
extreme opinions should have lower variation in opinions about EU policy areas. Turning to
the results, we can see that fearful citizens have a higher variation in opinions than angry
citizens, confirming H3, i.e. that fearful citizens are less uniform across policy areas. The
effects of anger and fear are statistically significantly different (F=30.5, p<0.001). This is in
line with literature that suggests that angry citizens tend to rely on heuristics based on their
overall views and as such seek less information.
[Table 2 about here]
However, we also partly disconfirm H3 in that we find that enthusiasm is associated
with increased variation in opinion, and its impact is in fact similar to that of fear. Hence,
while angry citizens have by far the most uniform views, enthusiasm and fear are both
associated with less heterogeneous opinions. Perhaps the increased interest in and affect
towards the EU implied by enthusiasm are just as important for information-seeking as
enthusiasm’s tendency to limit careful information-processing. Overall, it seems that anger
most clearly leads to a lack of careful information-seeking and more uniform opinion across
different EU policy areas.
In this article, we have put forward and tested a theoretical framework of how emotional
reactions affect public preferences on the EU at three stages of opinion formation. First, fear
increases citizens’ desire to renegotiate their country’s constitutional relationship with the EU,
while anger increases their willingness to leave the EU. Second, the attitudes of angry and
enthusiastic citizens towards the EU are less conditional on the perceived benefits of EU
integration than those of anxious citizens. Moreover, angry voters’ stances on EU
membership are related more to underlying attitudes towards the EU than among fearful and
enthusiastic voters. Finally, indicative evidence shows that angry citizens tend to express less
nuanced and varied views on specific polices related to the EU; we suggested that this is due
to differences in information-seeking and information-processing.
Our findings are significant as the EU is currently highly contested across EU member
states. The continued inability of the EU to resolve the economic and refugee crises may have
contributed to feelings of anger and fear among EU citizens, and these feelings have also been
strategically exploited by populist politicians. In order to explain support or opposition to the
EU, we need to consider how these emotional considerations may affect public opinion and
behaviour. Doing so may help us go some way toward understanding the increasing success
of Eurosceptic parties in the 2014 European Parliament elections as well as the UK’s decision
to leave the EU.
Our research also has implications for the persuasiveness and effectiveness of EU-
related campaigns. Campaigns that elicit anger are more likely to solidify existing Eurosceptic
attitudes, whereas pro-EU groups that create enthusiasm will mobilise citizens in favour of
European integration. However, when discussions about the EU elicit anxiety, citizens are
more likely to have nuanced positions on specific EU policies and base their EU opinion on
cost-benefit considerations. A campaign that encourages fear and anxiety may mean that
voters pay closer attention to the debate and decide more cautiously, so both sides will have to
marshal good and convincing evidence for their position.
Our analysis also sheds light on the Brexit referendum. The two official campaigns
‘Britain Stronger In Europe’ and ‘Vote Leave’ set the agenda by transforming the debate into
a battle between two issues: economy versus immigration (Hobolt, 2016). Whereas the
rhetoric was negative in both camps, being referred to as ‘Project Fear’ and ‘Project Hate’
respectively by the other side, they each elicited different negative emotions. On the one hand,
Britain Stronger In framed its campaign in terms of risk and risk avoidance. Its posters
included phrases such as ‘Leave Europe and we will lose our seat at the table’, ‘Leave and
there is no going back’ and ‘Leaving Europe would be a leap in the dark’. On the other hand,
Vote Leave messages such as ‘Turkey (population 76 million) is joining the EU’ and ‘Let’s
give our NHS the £350 billon the EU takes every week’ elicited anger, driven by a sense of
collective threat. Anger was further aroused through the portrayal of the EU as an illegitimate
source of this threat. Therefore, by evoking feelings of anxiety, the Remain camp sought to
persuade voters to maintain the status quo (as the less risky option), whereas by eliciting
anger, the Vote Leave camp enabled more risk-taking behaviours. Interestingly, neither
campaign tried hard to evoke positive emotions.
A factor relevant for future research is the effect of these emotions on turnout. Fear
and anxiety tend to be associated with avoidance strategies in order to minimise the
experienced threat whereas anger may transform a stimulus into a reaction (Halperin &
Pilskin, 2015). Enthusiasm tends to be linked to political interest and mobilisation (Brader and
Marcus, 2013). It is plausible that anxiety is not a good motivator to turn out, whereas anger
and enthusiasm may induce citizens to vote (Brader, 2006; Valentino et al., 2011). Future
research should therefore examine whether citizens who expressed fear about EU membership
were less likely to go to the ballot box as opposed to angry or enthusiastic citizens. If this is
empirically verified, then the Vote Leave campaign not only persuaded citizens in terms of its
argument but also might have encouraged people to participate in the vote. Put differently, if
the Remain campaign had managed to create more enthusiasm about the EU, it might have
more strongly mobilised UK citizens.
Future research should also take a comparative perspective. Our theoretical predictions
apply to the UK, and similar arguments have been tested in Ireland (Garry, 2014). In both
cases, the EU issue is relatively salient due to the success of UKIP (Britain) and frequent
referendums (Ireland). However, we still do not know whether emotions affect public opinion
on the EU where Euroscepticism is low, e.g. in Spain and Portugal, or in newer member states
such as in Central and Eastern Europe. Overall, research that takes into account individuals’
emotions on the EU promises to improve our understanding of how and when people support
and oppose European integration.
This article draws upon research funded by an ESRC Future Leaders grant, ES/N001826/1.
We are grateful to the journal's referees for comments.
1. In line with other studies (e.g. Valentino et al. 2008), we use the term fear and anxiety
2. For an argument on emotions and EU support based on a two-dimensional valence
conception of emotions, see Erişen and Kentmen-Cin (2015).
3. The nineteen policy areas are: agriculture and food, asylum seekers, competition and
business regulation, defence policy, digital security and data protection, education,
employment and social affairs, energy, environment and climate change, foreign and security
policy, health, justice, fundamental rights and equality, labour market, monetary policy,
overseas aid / development policy, sustainable development, taxation, trade and transport.
4. We do not conduct measurement tests on our coding of emotions. Doing so is difficult
because some respondents may check only ‘angry’ or ‘disgusted’ without being necessarily
less angry than those who select both emotions. In addition, it is quite common for people to
feel several emotions at the same time, so the co-occurrence of emotions does not mean that
they are the same phenomenon. The validity of our measures should instead be assessed by
their differential effects on attitudes and behavior.
5. While treating ‘disgust’ as an indicator for anger is a common approach (Conover and
Feldman, 1986; MacKuen et al., 2010; Valentino et al., 2011), some research argues that these
two emotions are distinct (Hutcherson and Gross, 2011).
6. Only one per cent of respondents reported feeling both enthusiasm and anger; 7 per cent
reported feeling both fear and enthusiasm. In our analysis, we always includes all four sets of
emotions, so we control for the effects of the other emotions when examining the impact of
one emotion. In robustness checks (see the Online appendix), we included interactions
between fear and anger in order to check whether voters who are only angry differ from those
who are angry and afraid. No substantively relevant differences were found.
7. This wording does not ask directly about the costs of membership. We assume that those
who perceive high costs see low benefits and would give low scores on the seven-point
8. Our models include a large set of control variables. Robustness checks (see the Online
appendix) indicate that models controlling just for gender, age, education and income produce
substantively the same results, although effect estimates are naturally larger in the simpler
models. Our models are likely to be conservative estimates of the effects of emotional
reactions to the EU. For information on descriptive statistics, see Online appendix.
9. We set other emotions to 0 if another emotion is treated as present (i.e. 1). All other
variables are held at their observed values.
Anderson CJ and Reichert, MS (1995) Economic Benefits and Support for Membership in the
E.U.: A Cross-National Analysis. Journal of Public Policy 15(03): 231-249.
Anderson CJ (1998) When in Doubt, Use Proxies: Attitudes toward Domestic Politics and
Support for European Integration. Comparative Political Studies 31(5): 569–601.
Armingeon K and Ceka B (2014) The loss of trust in the European Union during the great
recession since 2007: The role of heuristics from the national political system. European
Union Politics 15(1): 82–107.
Brader T (2006) Campaigning for hearts and minds: How emotional appeals in political ads
work. Chicago: University of Chicago Press.
Brader T and Marcus G (2013) Emotion and political psychology. In: Huddy, L, Sears D O
and Levy JS (eds) The Oxford handbook of political psychology. OUP: Oxford, pp.165-
Brader T, Valentino N, and Suhay E (2008) What triggers public opposition to immigration?
Anxiety, group cues, and immigration threat. American Journal of Political Science 52(4):
Cacioppo JT, Gardner WL, and Berntson GG (1999) The affect system has parallel and
integrative processing components: Form follows function. Journal of Personality and
Social Psychology 76(5): 839–855.
Carey S (2002) Undivided loyalties: is national identity an obstacle to European integration?
European Union Politics 3(4): 387–413.
Carver CS and Harmon-Jones E (2009) Anger is an approach-related affect: evidence and
implications. Psychological bulletin 135(2): 183-204.
Conover PJ and Feldman S (1986) Emotional reactions to the economy: I'm mad as hell and
I'm not going to take it anymore. American Journal of Political Science 30(1): 50-78.
De Wilde P, Laupold A and Schmidtke H (2016) Introduction: the differentiated politicisation
of European governance. West European Politics 39(1): 3-22.
Druckman JN and McDermott, R (2008) Emotion and the framing of risky choice. Political
Behavior 30(3): 297-321.
Erişen C (2013) Emotions as a determinant in Turkish political behavior. Turkish studies
Erişen C and Kentmen-Cin C (2017) Tolerance and perceived threat toward Muslim
immigrants in Germany and the Netherlands. European Union Politics Epub ahead of print
11 November 2016. DOI: 10.1177/1465116516675979.
Erişen C and Kentmen-Cin C (2015) Utilitarian Considerations versus Emotions as the
Foundations of the EU Public Opinion, paper presented at the Midwest Political Science
Association Annual Meeting, 16 April.
Gabel M (1998) Interests and Integration: Market Liberalizaiton, Public Opinion, and
European Union. Ann Arbor: University of Michigan Press.
Gabel M (2000) European integration, voters and national politics. West European Politics
Gabel M and Whitten GD (1997) Economic Conditions, Economic Perceptions and Public
Support for European Integration. Political Behavior 19(1): 81–96.
Garry J (2014) Emotions and voting in EU referendums. European Union Politics 15(2): 235-
Gomez R (2015) The Economy Strikes Back: Support for the EU during the Great Recession.
Journal of Common Market Studies 53(3): 577-592.
Halperin E and Pilskin R (2015) Emotions and Emotion Regulation in Intractable Conflict:
Studying Emotional Processes Within a Unique Context. Advances in Political Psychology
Hobolt SB (2016) The Brexit vote: a divided nation, a divided continent. Journal of European
Public Policy 23(9): 1259-1277.
Hobolt SB and De Vries C (2016) Public support for European integration, Annual Review of
Political Science 19. ISSN 1545-1577
Hooghe L and Marks G (2009) A Postfunctionalist Theory of European Integration: From
Permissive Consensus to Constraining Dissensus, British Journal of Political Science 39:
Huddy L, Feldman, S, Taber C and Lahav G. (2005) Threat, anxiety, and support of
antiterrorism policies. American Journal of Political Science 49(3): 593-608.
Huddy L, Feldman S and Cassese E (2007) On the distinct political effects of anxiety and
anger. In: Neuman R, Marcus GE, MacKuen M, and Crigler A (eds) The affect effect:
Dynamics of emotion in political thinking and behavior, Chicago University Press, pp.
Hutcherson CA and Gross JJ (2011) The moral emotions: A social–functionalist account of
anger, disgust, and contempt. Journal of Personality and Social Psychology 100(4): 719-
Inglehart R (1970) Cognitive Mobilization and European Identity. Comparative Politics 3(1):
Isbell LM, Ottati VC and Burns KC (2006) Affect and politics: Effects on judgment,
processing, and information seeking. In: Redlawsk, DP (ed) Feeling politics, New York:
Palgrave, pp. 57–86.
Kentmen-Cin C and Erişen C (2017) Towards a comprehensive understanding of how anti-
immigration attitudes shape opposition to European integration: A critical assessment,
European Union Politics Epub ahead of print 8 February 2017. DOI:
Ladd JM and Lenz GS (2008) Reassessing the role of anxiety in vote choice. Political
Psychology, 29(2), 275-296.
Larsen JT and McGraw AP (2011) Further evidence for mixed emotions. Journal of
Personality and Social Psychology 100(6), 1095-1110.
Lerner JS and Keltner D (2000) Beyond valence: Toward a model of emotion-specific
influences on judgement and choice. Cognition and Emotion 14: 473–493.
Lerner JS and Keltner D (2001) Fear, anger, and risk. Journal of personality and social
psychology 81(1): 146-159.
Lerner JS, Gonzalez RM, Small, DA and Fischhoff B (2003) Effects of fear and anger on
perceived risks of terrorism a national field experiment. Psychological science 14(2): 144-
MacKuen M, Marcus GE, Neuman WR and Keele L (2007) The third way: The theory of
affective intelligence and American democracy. In: Neuman WR, Marcus GE, Crigler AN
and MacKuen M (eds) The affect effect: Dynamics of emotion in political thinking and
behavior, Chicago: University of Chicago Press, pp. 124-151.
MacKuen M, Wolak J, Keele L and Marcus G (2010) Civic engagements: Resolute
partisanship or reflective deliberation. American Journal of Political Science 54(2): 440–
Marcus G and MacKuen M (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 (2003) The psychology of emotion and politics. In: Huddy L, Sears D and Jervis
R (eds) Oxford handbook of political psychology, Oxford: Oxford University Press, pp.
Marcus GE, Neuman WR and MacKuen M (2000) Affective intelligence and political
judgment. Chicago: University of Chicago Press.
McLaren LM (2002) Public support for the European Union: cost/benefit analysis or
perceived cultural threat? Journal of Politics 64(2): 551–566.
Merolla JL and Zechmeister EJ (2009) Terrorist threat, leadership, and the vote: Evidence
from three experiments. Political Behavior 31(4), 575-601.
Petersen MB (2010) Distinct emotions, distinct domains: Anger, anxiety and perceptions of
intentionality. The Journal of Politics 72(02): 357-365.
Roseman IJ (1991) Appraisal determinants of discrete emotions. Cognition & Emotion 5(3):
Sanders D, Clarke HD, Stewart M and Whiteley P (2007) Does Mode Matter For Modeling
Political Choice? Evidence From the 2005 British Election Study. Political Analysis
Smith CA and Ellsworth PC (1985) Patterns of cognitive appraisal in emotion. Journal of
personality and social psychology 48(4): 813-838.
Smith HJ, Cronin T and Kessler T (2008) Anger, fear, or sadness: Faculty members’
emotional reactions to collective pay disadvantage. Political Psychology 29(2): 221–246.
Steenbergen M, Edwards E, and De Vries C (2007) Who’s Cueing Whom? Mass–Elite
Linkages and the Future of European Integration, European Union Politics 8(1): 13–35.
Tellegen A, Watson D and Clark LA (1999) On the dimensional and hierarchical structure of
affect. Psychological Science 10(4): 297-303.
Valentino NA, Brader T, Groenendyk EW, Gregorowicz K and Hutchings VL (2011) Election
night’s alright for fighting: The role of emotions in political participation. Journal of
Politics 73(1): 156-170.
Valentino NA, Gregorowicz K and Groenendyk EW (2009) Efficacy, emotions and the habit
of participation. Political Behavior 31: 307–330.
Valentino NA, Hutchings V, Banks A and Davis A (2008) Is a worried citizen a good citizen?
Emotions, political information seeking and learning via the internet, Political Psychology
Wagner M (2014) Fear and anger in Great Britain: Blame assignment and emotional reactions
to the financial crisis. Political Behavior 36(3): 683-703.
Weber C (2012) Emotions, campaigns and political participation. Political Research
Quarterly 66(2): 414–428.
Whiteley P, Clarke HD, Sanders D and Stewart MC (2010) Government performance and life
satisfaction in contemporary Britain. Journal of Politics 72(3): 733-746.
Figure 1. Emotional reactions and opinion formation on EU integration
Figure 2. Attitudes towards renegotiation of the terms of the UK’s EU membership.
Source: Data from original survey of 3000 respondents conducted in the period 23/04/2015-
Figure 3. Emotional reactions to Britain's membership of the EU, raw and recoded.
Source: Data from original survey of 3000 respondents conducted in the period 23/04/2015-
Figure 4. Direct effect of emotional reactions on support for EU integration
Note: Scales of outcomes variables range from 1 to 7, with 1 labelled as ‘strongly disagree’ and 7 as
‘strongly agree’. Predicted levels of the outcome variable calculated by varying emotional reactions
while holding all other variables at their observed levels.
Figure 5. Emotions as a moderator of other considerations.
Note: Graph shows the marginal effect of cost-benefit perceptions and general EU attitudes for three
groups of voters, based on whether they declared feeling enthusiastic, angry and anxious. The
dependant variable is the individual’s position on leaving the EU, measured on a 1-7 scale. Bars show
95% confidence intervals. Marginal effects and confidence intervals calculated based on our results
from model 3, Table 1.
Table 1. The impact of emotional reactions on attitudes towards renegotiation and preferences for the
UK’s constitutional relationship with the EU.
DV: Leave EU
DV: Leave EU
Costs vs benefit (1-7)
Costs vs benefit * Fear
Costs vs benefit * Anger
Costs vs benefit * Enthusiasm
Costs vs benefit * Indifference
General EU attitude
General EU attitude * Fear
General EU attitude * Anger
General EU attitude * Enthusiasm
General EU attitude * Indifference
Leave EU (1-7)
Renegotiate EU membership (1-7)
Trust in EU (0-10)
Support for EU integration (0-10)
Satisfaction with democracy in EU (0-10)
PTV: Conservative (0-10)
PTV: Labour party (0-10)
PTV: Liberal Democrat party (0-10)
PTV: UKIP party (0-10)
PTV: Green party (0-10)
PTV: Other party (0-10)
Personal economic situation (1-5)
General economic situation (1-5)
EU: responsibility for economic situation (0-
European identity (0-7)
Economic ideology: For state intervention
Economic ideology: For redistribution (1-7)
Immigration: Restrictive policy (1-7)
Immigration: Against diversity (1-7)
Environment: Combat climate change (1-7)
Trust in UK government (0-10)
TV viewing (1-6)
Internet use (1-6)
Newspaper use (1-6)
Note: Standard errors in parentheses, * p < 0.05, ** p < 0.01, *** p < 0.001
Table 2. The effect of emotional reactions on opinion uniformity across policy areas
DV: Standard deviation in
Leave EU: extremism (0-3)
Renegotiate EU membership: extremism (0-3)
TV viewing (1-6)
Internet use (1-6)
Newspaper use (1-6)
Left-Right extremism (0-5)
PTV extremism: Conservative (0-5)
PTV extremism: Labour party (0-5)
PTV extremism: Liberal Democrat party (0-5)
PTV extremism: UKIP party (0-5)
PTV extremism: Green party (0-5)
PTV extremism: Other party (0-5)
Personal economic situation: extremism (0-2)
General economic situation: extremism (0-2)
EU: responsibility for economic situation: extremism (0-5)
Costs vs. benefits: extremism (0-3)
Economic ideology: For state intervention: extremism (0-3)
Economic ideology: For redistribution: extremism (0-3)
Immigration: Restrictive policy: extremism (0-3)
Immigration: Against diversity: extremism (0-3)
Environment: Combat climate change: extremism (0-3)
Satisfaction with democracy in EU: extremism (0-5)
European identity: extremism (0-3)
Trust in EU: extremism (0-5)
Trust in UK government: extremism (0-5)
Note: Standard errors in parentheses, * p < 0.05, ** p < 0.01, *** p < 0.001