Renegotiation Versus Brexit: The Question of the UK's Constitutional Relationship with the EU

This article examines how the British public perceived UK Prime Minister David Cameron's plan to renegotiate his country's relationship with the EU. It asks whether attitudes towards renegotiation followed a similar pattern to attitudes towards Brexit. It asks: are preferences towards renegotiation and Brexit related, and did British citizens perceive them as conflicting or complementary? We modelled the similarities and differences between these two types of preferences, which allowed us to classify the attitudes into four patterns: unconditional europhiles, rejectionist eurosceptics, risk&#8208;averse eurosceptics and power&#8208;seeking eurosceptics. Using a large&#8208;N cross&#8208;sectional survey conducted in the UK in April 2015 (n&#8201;=&#8201;3000), our findings suggest that similar utilitarian concerns underpinned both types of preferences; but education and partisan cues differentiated them. Our findings have implications for understanding the result of the UK referendum. They also highlight the complex considerations that drive citizens&#8217; attitudes towards the EU and help us predict the scope of public acceptance of EU reform initiatives by other governments.


Introduction
In his January 2013 Bloomberg Speech, the UK Prime Minister David Cameron announced that if his party was re-elected to government following the 2015 general election he would renegotiate a new settlement for his country and subsequently hold an 'in-out' referendum on its European Union (EU) membership. This was the first time that a head of government of an EU Member State promised the public a renegotiation of their country's constitutional relationship with the EU since the latter was formally founded in 1993. The announcement drew comparisons with the renegotiation of the UK's European Economic Community membership in 1974, which was also followed by an in-out referendum. The Conservative Party's victory in the 2015 general election allowed the prime minster to renegotiate a new settlement for the UK, granting it special status within the EU, and paved the way towards a referendum on the UK's EU membership on 23 June 2016. Cameron's strategic calculation to promise this renegotiation may be interpreted as successful, given the public's apparent initial support for it and his party's surprise 2015 electoral victory, but at the same time it may be seen as a potential misjudgement in light of the referendum result in favour of Brexit (Curtice, 2016;Lynch, 2015, p. 199).
In this article we begin unravelling the puzzle of why, given the initial support for a renegotiation, Cameron's strategy failed when the public rejected the renegotiation and voted for Brexit. We aim to shed light on the drivers that shaped attitudes towards these two processes. More specifically, we make three contributions. First, although scholars have made advances to understanding EU level negotiations before the UK's referendum in 2016 (Oppermann, 2016) and voting behaviour during the referendum (Clarke et al., 2017;Hobolt, 2016), citizens' attitudes to the renegotiation remain under-researched. We still do not know how voters perceived Cameron's renegotiation promise in the first place and whether there was an initial relationship between individuals' attitudes towards renegotiation and preferences towards Brexit. Did British citizens view 'Brexit' and 'renegotiation' as complementary or as different from each other? If they were different, what led citizens to support or oppose them? By asking these questions we provide a nuanced understanding of how the British electorate made sense of a complex political deal and the challenges Cameron faced in trying to convince a majority of citizens to vote against Brexit.
Second, through our examination of the factors that shaped attitudes towards these two processes, we contribute to the comparative study of citizen preference formation towards European integration. We model the similarities but also the differences between these two types of preferences and show that, although the attitudes towards renegotiation (special status within the EU) and Brexit (special status outside the EU) overlap to some extent, they are distinct enough to allow us to identify four groups of UK citizens, which we term unconditional europhiles, rejectionist eurosceptics, risk-averse eurosceptics, and power-seeking eurosceptics.
Third, we posit that renegotiation and Brexit are hard and easy issues, respectively, requiring different cognitive engagement. We find support for the cognitive sophistication hypothesis. Education levels help explain conflicting preferences towards renegotiation and Brexit. Highly educated individuals supported renegotiation but opposed Brexit. We also find that cue-taking was significant. Support for renegotiation was predominantly found among right-wing party supporters, which suggests that renegotiation gave opportunities for Cameron to appeal to both Conservative and the UK Independence Party (UKIP) voters, capitalizing on citizens' attitudes to the renegotiation.
Brexit remains a unique event, since no other Member State has ever decided to exit the EU. 1 However, the sentiments that led to it are by no means a distinctively British phenomenon, with eurosceptic and EU-issue-focused entrepreneurs pushing for referendums in states across Europe. Our theoretical approach and classification has the potential to apply beyond the British context and helps us predict the scope of citizen support for EU reform initiatives and public acceptance of national government attempts to renegotiate international treaties elsewhere.

The Path to Renegotiation
In 1974 Harold Wilson's Labour government renegotiated the UK's terms of membership of the Common Market and in a referendum a year later a large majority of voters supported the country remaining in the European Economic Community. However, the UK remained Europe's 'awkward partner' (George, 1998;Glencross, 2016). Politicians seeking to reform the UK's relationship with the EU came to pledge more referendums on European issues. The Thatcher government criticized the direction of European integration, and in the mid-1990s, under John Major, the Conservative Party took a noticeable shift towards soft eurosceptic positions and promised a referendum should it recommend joining the single currency (Lynch and Whitaker, 2013a, p. 319). Blair's 'New Labour' governments showed signs of a positive approach to European integration but failed to foster a pro-European consensus, as the Conservative leader William Hague embraced eurosceptic positions to move his party closer to the median voter (Bale, 2006). Labour responded by downplaying party competition over European issues. It deferred to the general public with promises of a referendum should the UK adopt the single currency or the (failed) European Constitution (Oppermann, 2008, p. 156). As the Conservative Party's euroscepticism deepened, Hague's successors also committed to holding referendums on future EU treaties, including David Cameron's pledge to hold a referendum on the Lisbon Treaty in 2007.
Cameron has been criticized for pursuing an inconsistent strategy on Europe (Lynch and Whitaker, 2013a). In opposition he sought to lower the salience of Europe and focused on 'modernizing' his party by discussing political and social change (Lynch, 2015). When the Conservatives formed a coalition government in 2010 with the europhile Liberal Democrats, Cameron pursued a pragmatic strategy and appointed David Lidington, a pro-European Conservative, as Minister of State for Europe (Baldini, 2012). However, the sovereign debt crisis and the Conservative Party's loss of support to UKIP at the 2010 general election contributed to a hardening of euroscepticism among Conservative MPs (Heppell, 2013, p. 347). In response to these pressures, Cameron followed the well-trodden path of considering referendums on European integration (Lynch and Whitaker, 2013b). In 2011 the coalition passed the European Act that pledged a referendum on future European treaties.
In January 2013 Cameron attempted to manage intra-party conflicts by announcing that if his party were to win the general election in 2015 he would renegotiate the UK's terms of EU membership and put the outcome to the electorate in an in-out referendum (Startin, 2018, p. 456). The renegotiation appeared to be a carbon copy of the promise that Wilson made in 1974. Scholars noted that Cameron's strategy might provide an advantage in general elections, given the popularity of reforming the UK's relationship with the EU, and if voters concluded that only the Conservatives could deliver a renegotiation and referendum (Smith, 2016). In 2013 opinion polls also suggested that a majority of respondents would vote to remain in the EU should Cameron renegotiate the UK's terms of membership (Lynch, 2015).
Consistent with his Bloomberg speech, Cameron (2015) outlined four objectives for the renegotiation based in areas of economic governance, competitiveness, sovereignty and immigration. To this end, Cameron called for the limited repatriation of powers from Brussels, rights for Parliament to block EU legislation, guarantees that the UK could access the single market while remaining outside the eurozone, continued independence for the Bank of England in regulating the UK's financial system, and assurances that British taxpayers would not be liable to support the Euro as a currency. Perhaps most contentious were Cameron's calls to limit migration when countries join the EU and for restrictions on the rights of EU migrants to claim welfare benefits (Lynch, 2015).
The agreement reached with European Council President Donald Tusk in February 2016 met some of Cameron's demands. He delivered an exemption from the commitment to an 'ever closer union', increased the powers of national parliaments and made some progress in calls for cutting red tape to promote competitiveness within the EU through reducing the burden of regulation (Lynch, 2015;Smith, 2016). The deal proposed a temporary requirement that EU citizens should live in another Member State for 4 years before having full access to welfare benefits; although it fell short of Cameron's goal of ending the practice of sending child benefits overseas (Curtice, 2017). Overall, Cameron's new settlement did not satisfy influential Conservative politicians and was widely criticized in the British press, and ultimately featured very little during the referendum campaign (Hobolt, 2016).
However, we still do not know how, prior to the actual renegotiation outcome, voters perceived Cameron's renegotiation promise in the first place and whether there was initially a relationship between individuals' attitudes towards renegotiation and their preferences towards Brexit. It remains unclear as to whether voters viewed the two processes as conflicting or complementary and what factors shaped attitudes towards them. We explore these questions in the following sections.

A Model for Explaining Public Attitudes Towards EU Renegotiation and Brexit
We explain opinions on these two different aspects of the UK's constitutional relationship with the EU by drawing upon scholarly work on euroscepticism, which has focused on at least two main approaches; namely, identity and utilitarianism (Hobolt and de Vries, 2016). To do so, we modelled the common drivers of public opinion towards these two processes, while at the same time accounting for heterogeneity in citizens' preferences.

Political Identities
Identity theories posit that as the pace of EU integration increases, citizens may feel that their long-standing national identities are being challenged (for example, Carey, 2002;McLaren, 2004). Individuals who strongly identify with the nation-state and feel a strong bond with their community may view European integration as a threat to their identity. Conversely, their attachment to Europe may act as a buffer against hostility towards the EU (Wessels, 2007). We expected these emotive ties to have the same effect both on attitudes towards Brexit and renegotiation. Identity concerns related to the UK's relationship with the EU have been ascribed to a number of explanations, including the country's island location, imperial past, and attachment to the Anglo-American world (George, 1998). Being part of the EU has long been seen as an encroachment on national sovereignty and the UK Parliament's supreme legal authority. The Conservative party also employed a sovereignty-based rhetoric by presenting its renegotiation agenda as a plan for defending British sovereignty and the principle of subsidiarity vis-à-vis the EU's attempt to build an 'ever closer political union'. We thus hypothesize: Weak political attachment to Europe is likely to be associated with support for both Brexit and renegotiation (H1).

Economic Self-interest
Given the EU's strong emphasis on economic prosperity, scholars have also focused on whether public support for European integration is influenced by rational calculations (Anderson and Reichert, 1995;Gabel, 1998). Individuals may base their preferences on cost-benefit analyses of European integration. EU membership involves both winners and losers. While market liberalization associated with EU integration introduces opportunities for some citizens, it also increases competition for others. The extent to which an individual may perceive EU integration as an opportunity or a threat depends on their socioeconomic status. Citizens with high incomes and in highly skilled, managerial or professional positions may see a strong personal potential benefit in liberalized markets in goods, capital and labour. These individuals also tend to be more mobile and thus more likely to gain from EU freedom of movement. Conversely citizens in low-paid and low-skilled jobs tend to be less well equipped to adjust to competitiveness in an international marketplace and are more likely to contest the removal of national borders. Being the losers of economic transformations, these individuals are likely to view European integration as a source of insecurity and threat (Kriesi et al., 2006). We also suggest that similar cost-benefit calculations influence public opinion towards Brexit (Hobolt, 2016), and expect that Low skilled individuals with lower incomes who perceive they will have limited benefits from international cooperation are more likely to support Brexit (H2a).
This approach may also help us understand attitudes towards renegotiation, as Cameron's EU reform agenda had a strong economic component. In his 2013 Bloomberg speech he laid out his plan for safeguarding British economic interests related to the single market, competitiveness and red tape. His vision was to broker international trade agreements that would boost global trade and complete the single market in services. Renegotiation also involved a stipulation that the UK would not be committed to contributing to Eurozone bailouts, which was framed as a key economic opportunity that would decrease the costs of EU membership. Therefore, we hypothesize that, assuming that they would expect to derive some economic benefit from this process: Citizens who view their country's membership in terms of cost are more likely to seek renegotiation (H2b).
Cameron's renegotiation not only was justified as a move to defend UK economic interests abroad but also as a means of reducing the welfare costs of EU migration on the British state. We therefore hypothesize that individuals with low occupational skills are likely to support renegotiation (H2c).
given that such citizens tend to feel economically insecure (Walter, 2010) and are more likely to perceive immigration as a threat to their welfare (Lucassen and Lubbers, 2012). The relationship between income and attitudes to renegotiation is inconclusive. People with low incomes may expect an economic benefit from renegotiation and thus support it; at the same time individuals with high incomes may also see an economic opportunity in a new renegotiated UK status in the EU, especially given that renegotiation would not compromise their status quo.
The above hypotheses provide a good starting point in explaining whether individuals would support or oppose both the processes of renegotiation and Brexit. However, they do not help to explain why some individuals may support renegotiation but oppose Brexit and vice versa. To explain this, we consider the different nature of the two issues and the role of individuals' cognitive abilities and of cues in processing EU-related complexity during their decision-making calculus.

Renegotiation versus Brexit: Political Sophistication and Cues
Research differentiates between easy and hard political issues (Carmines and Stimson, 1980;Johnston and Wronski, 2015;Pollock et al., 1993). Easy issues have been on the political agenda a long time, tend to be symbolic rather than technical in nature, and can be easily communicated to the wider public. They tend to be rooted in existing prejudices and are more likely to elicit 'gut responses'. Making up one's mind on an easy issue is relatively straightforward because easy issues are more general and ends-oriented, that is, they relate to public policy ends rather than dealing with the means to achieve these ends. Hard issues, on the other hand, tend to be 'more difficult for individuals to tie to core values' (Pollock et al., 1993, p. 29) because they are technical and involve specialized knowledge. Deciding on a hard political issue therefore is the result of a reasoned attempt at understanding the issue in question and a sophisticated decision-making calculus, which require advanced cognitive skills.
We argue that whereas Brexit may be thought of as an easy issue, the question of renegotiation is a hard political issue. Brexit relates to policy ends, namely, leaving the EU, as opposed to policy means, which refer to an assessment of the different ways in which a country may achieve secession from the EU, as well as its political, social and economic implications. In addition, because the question of the UK leaving the EU has long featured in public and media debates in the UK, citizens might feel a level of familiarity with the issue, which is more likely to elicit affective responses (Vasilopoulou and Wagner, 2017). On the other hand, although Cameron's renegotiation had a strong symbolic element, as it represented his attempt to empower the UK by driving EU reform, in substance the issue is technical, involving a number of specific policies ranging from bureaucracy, trade and access to markets, competitiveness, immigration and the functioning of the EU regime. It is therefore hard to predict its real impact on the national economy and individuals' well-being, including their job prospects, incomes and purchasing power.
We posit that these substantive differences between the two types of issues may result in different cognitive processes taking place during the process of preference formation, with individuals' cognitive abilities serving to differentiate citizens' attitudes towards Brexit and renegotiation. On the one hand, the symbolic and ends-oriented character of Brexit could be a simple choice for individuals and one that does not necessitate a high level of 'cognitive mobilisation' (Inglehart, 1970). Individuals with low levels of education are more likely to prefer demarcation (Kriesi et al., 2006), and thus support the closing of borders and their country's exit from the EU. Making a conscious calculation of the costs and benefits derived from renegotiation, on the other hand, requires a level of sophistication in order to organize and process such information. Educated individuals are more likely to make a reasoned and thoughtful attempt to engage with the specificities of renegotiation and perceive the potential benefits or costs arising from it (Hakhverdian et al., 2013). A successful renegotiation potentially promises to improve the UK's status in the EU both in terms of power and in economic terms; but given that renegotiation is a non-costly process, an unsuccessful one is unlikely to compromise the status quo. We thus hypothesize that educational level is negatively associated with support for Brexit but positively associated with support for renegotiation (H3).
Heuristics also may play a role in shaping citizen preferences, helping the public to process information by simplifying the decision-making process. The literature suggests that elite positions may be employed as information shortcuts, influencing voters' preferences (Anderson, 1998;Sanchez-Cuenca, 2000). Applied to the questions of renegotiation and Brexit, we suggest that different partisan cues may be stronger in each issue. Brexit, being an easy issue, may be attached to citizens' core values and beliefs, which are also associated with their attitudes towards domestic elites (Hobolt and De Vries, 2016). UKIP and the Liberal Democrats have consistently been eurosceptic and europhile UK voices, respectively, sending clear cues to the electorate. The Conservative and Labour parties have been historically divided over the question of Europe. In fact, Cameron's promise to hold a referendum on Europe may be understood as an attempt to end internal party conflict on the EU issue (Lynch and Whitaker, 2013b). And, while the Labour party under Tony Blair initially sought to put forward a pro-EU stance, the party's EU position has been characterized as an attempt to contain the salience of the EU issue (Hertner and Keith, 2016;Oppermann, 2008). Given that their messages have been contradictory and often ambiguous, they are also unlikely to influence their voters in a coherent manner. We expect therefore that attitudes towards domestic elites matter, with UKIP voters more likely to support Brexit and Liberal Democrat voters more likely to oppose the same process (H4a). However, the hard nature of renegotiation makes it difficult for voters to understand where political parties stand, and even harder in a context where political parties did not necessarily have an EU reform plan in their agendas; or if they did, as arguably in the case of the Labour party, it did not coincide with the Conservative party's plan. The decision to renegotiate and the specific reform plan was explicitly tied to the Conservative party's deepening euroscepticism and in particular to the issues that Cameron was talking about at the time (mentioned in the previous section). We thus hypothesize that Conservative party voters are more likely to express support for renegotiation (H4b).
Our theoretical discussion above has suggested that if voters view the processes of renegotiation and Brexit as complementary then the same model related to identity and economic self-interest would serve to explain these attitudes. However, we have argued that Brexit and renegotiation are easy and hard political issues, respectively. We thus expect that there is scope for responses to these two questions not to overlap (Figure 1; see also Hobolt, 2014;Karp and Bowler, 2006).
The principal diagonal of Figure 1 presents preferences towards the two processes as complementary. Unconditional europhiles oppose both Brexit and renegotiation. The fact that they do not even wish to renegotiate the UK's EU membership qualifies them as unconditional. The logic behind opposition to renegotiation may involve acceptance of membership, or the opinion that an EU Member State should try to reform the EU by constant cooperation rather than putting forward specific red lines. Power-seeking eurosceptics, on the other hand, support both renegotiation and Brexit. These citizens perceive renegotiation as an opportunity to reassert the UK's power and influence before leaving the EU or feel that the best option is to exhaust all possibilities before taking drastic action, in order for the country to appear rational or orderly before ultimately changing its relationship with the EU through Brexit. The counter-diagonal illustrates citizen attitudes as conflicting. Rejectionist eurosceptics do not seek renegotiation but support Brexit. These citizens express an outright rejection of EU membership and do not view renegotiation as something that could change the UK's status within (or outside) the EU. Risk-averse eurosceptics are unhappy with the status quo and therefore support renegotiation as a potentially promising route to reforming the EU, but these citizens perceive Brexit as a risky option and therefore oppose it (this is akin to soft euroscepticism).
We posit that identity and utilitarianism drive citizen support for or opposition to both processes (unconditional europhiles and power-seeking eurosceptics). However, voters are also likely to distinguish between the two processes (rejectionist and risk-averse eurosceptics). We postulate that these conflicting preferences may be explained through individuals' political sophistication and cue-taking.

Methods and Data
We rely on data collected from a large-N cross-sectional online survey (n=3000) conducted in April 2015, prior to the 2015 UK general election. The survey was administered by Research Now. 2 Participants were recruited using a variety of actively managed online access panels. All participants were exposed to the survey in the same way. Informed consent was obtained through a disclaimer on the first page of the survey, which explained its purpose and ensured their anonymity. The sample is representative of the UK population in terms of gender, age and region (Appendix Table A.1). While online surveys tend not to be fully representative of the general population, online and face-to-face survey data tend to yield similar results both in terms of estimating coefficients and the overall predictive power of rival models (Sanders et al., 2007). The timing of the survey is ideal for testing the above expectations. First, given that Cameron tied his re-election campaign to renegotiation, this question featured prominently in public and media debates during that time. Second, the survey took place prior to both the renegotiation outcome and the referendum campaign, which arguably changed the context of preference formation. The timing allows us to tap into the formative stage of the discussions over the issue rather than predict vote choice in the referendum, which is outside the scope of this article.
For the outcome variables, we considered responses to the following two agreedisagree statements: (1) the UK should renegotiate the terms of its EU membership; and (2) irrespective of renegotiation, the UK should leave the EU. Responses to both questions were measured on a seven-point scale, ranging from 1 (strongly disagree) to 7 (strongly agree). Public opinion towards renegotiation tends to be supportive (mean = 5.19; median = 5) whereas preferences towards Brexit tend to be spread out (mean = 3.82; median = 4) (appendix Figure A.1).
To measure EU attachment (H1), we employed the agree-disagree question: 'You think of yourself as European', with scores ranging from one to seven, with higher values indicating weak attachment to Europe. We operationalized economic self-interest (H2) with reference to income, occupation skill and perceived benefit of EU membership. We relied on respondents' self-reported household income measured on a scale ranging from 1 = under £10,000 to 14 = £500,000 or more. We also included an occupation dummy for manual workers (skilled, semi-skilled or unskilled), with those working in professional and managerial positions as the reference category. Perceptions of EU membership (H2) were measured through the agree-disagree statement 'The UK has greatly benefited from being a member of the EU', with scores ranging from one to seven, with higher values indicating lower perceived benefits. Educational level (H3) was measured from one to four, with 1 indicating 'less than secondary school-leaving exams', 2 'secondary school-leaving exams (such as GCSEs or equivalent)', 3 'college entrance exams (such as A-levels, National Vocational Qualification, or equivalent)' and 4 'university (such as a degree or professional qualifications)'. To test the effect of party cues (H4) our models included a variable indicating how likely participants were to vote for each of these parties in the 2015 general election, including the Conservative, Labour, Liberal Democrat parties and UKIP (PTV=propensity to vote).
We controlled for a number of potential confounders. Research has shown that negative out-group bias in the form of intolerance and anti-immigration sentiment can be a strong predictor of opposition to the EU (De Vreese and Boomgaarden, 2005). We controlled for anti-immigration feelings using the agree-disagree statement 'The UK should have a restrictive policy on immigration', with scores ranging from one to seven, with higher values denoting anti-immigrant feelings. The literature also suggests that there is a relationship between citizens' evaluations of the political system and their attitudes towards the EU (Anderson, 1998;Sanchez-Cuenca, 2000). We controlled for citizen satisfaction with the way democracy works in the UK and the EU. Responses to both questions are coded from one to ten, with higher values denoting satisfaction. A control was used for self-placement on the left-right dimension, although there is no clear relationship between left-right political views and EU position in the literature (Marks and Steenbergen, 2002). Finally, other individual-level controls included age, gender and media use (TV, the internet and newspapers) (see Appendix Table A.2 on the descriptive statistics). Table 1 shows the broad patterns of preferences on renegotiation and Brexit by mapping respondents' answers to the two questions on the basis of the 2 × 2 framework presented in Figure 1. As answers to these questions are coded on a seven-point scale we have not recoded the middle neutral category, which appears on the table as 'not sure'. shows that prior to the UK 2015 general election there was greater support for renegotiation than for leaving the EU, at 66.7 and 33.4 per cent, respectively. In other words, at that time, the question of renegotiation resonated with a large majority of British citizens, almost double the number of those who wanted to leave the EU. We start by discussing the two groups of individuals who viewed the processes of renegotiation and Brexit as complementary. On the one hand, the percentage of unconditional europhiles, that is, those who were against both renegotiation and Brexit, was very small at approximately 8.3 per cent. On the other hand, power-seeking eurosceptics, who supported both renegotiation and Brexit, were the group with the highest number of respondents at approximately 29 per cent. Citizens who viewed these two processes as conflicting include the very small category of rejectionist eurosceptics, that is, the individuals who did not support renegotiation but wanted to leave the EU, at approximately 2.35 per cent of respondents; and the risk-averse eurosceptics, namely the individuals who supported renegotiation but opposed Brexit, at 23.4 per cent. Approximately 37 per cent of respondents were unsure about both processes. We explore variations in the attitudes towards the two processes through ordered logistic regression analyses (Table 2). We estimate two models of support for renegotiation and Brexit, which allow us to compare the drivers of these two types of preferences. Answers to the two questions are positively correlated, which suggests some complementarity, but the correlation is modest (0.38), which indicates that there is scope for heterogeneous opinion formation (Hobolt, 2014). While weak European identity is positively associated with support for renegotiation, it is not statistically related to support for  Brexit. This partially supports H1, which suggests that weak political attachment to Europe is likely to be associated with support for both Brexit and renegotiation. While Cameron's renegotiation plan was perceived as a question of identity, Brexit was not perceived as such prior to the start of the campaign (regarding identity and the referendum, see Curtice, 2017). We find support for the utilitarian hypothesis, as perceived cost from membership is associated with support for both Brexit and renegotiation. Skill, however, is only associated with support for Brexit, with manual workers more likely to be in favour of their country leaving the EU compared with individuals in professional and managerial positions. As expected, income only has an effect on support for Brexit, with citizens with higher income being less likely to seek this constitutional change for their country. In line with our expectations, it does not have an effect on support for renegotiation. We argued that renegotiation is unlikely to compromise the economic status quo of the country, and as a result it may be seen as an economic opportunity for individuals with different levels of wealth. In short, our findings support H2a and H2b but fail to support H2c.

Findings
In line with our hypothesis, educational level is negatively associated with support for Brexit but positively associated with support for renegotiation (H3). This entails that while individuals with high levels of sophistication did not support a change of the country's constitutional relationship with the EU, they were able to understand the complexity of renegotiation and perceived the potential benefits of such a process. Finally, our analyses support our expectations on political party cues (H4a and H4b). The individuals  Notes: Results from logistic regression analysis. Dependent variable: 1 risk-averse (support renegotiation but oppose Brexit) 0 unconditional europhiles and power-seeking eurosceptics (support or oppose both processes). Those in the middle of the scales were excluded from the analysis. Logistic regression coefficients are reported. Standard errors in parentheses. * P < 0.10, ** P < 0.05, *** P < 0.01 who were likely to vote for UKIP in the 2015 general election were also more likely to support Brexit. The opposite relationship is observed for Liberal Democrat voters who opposed Brexit. In line with our expectation, Conservative and Labour party cues were not clear to the voters, and thus did not have a statistically significant effect on public preferences towards Brexit. The question of renegotiation, on the other hand, found support among Conservative party voters. Interestingly, we also find that UKIP supporters backed renegotiation whereas Liberal Democrat voters did not. UKIP and Liberal Democrat voters fell neatly into the power-seeking eurosceptic and unconditional europhile categories, respectively. Crucially, and in line with our argument, a propensity to vote for the Labour party does not have a statistically significant effect on support for renegotiation. This suggests that these voters could have interpreted it as a 'Tory' renegotiation that would not necessarily focus on Labour's core issues, like social justice, and as such they did not have a clear preference on the initiative. It is also plausible that this was driven by a lack of willingness to support the Conservative party. Next, we assess the conflicting nature of the two types of attitudes towards renegotiation and Brexit. We do so by specifically testing our hypotheses on the risk-averse group of eurosceptics, that is, those who wanted to renegotiate the UK's relationship with the EU but did not want the country to leave the EU. Table 3 presents a logistic regression model where the dependent variable takes a binary form, with 1 denoting individuals who have conflicting attitudes, that is, they support renegotiation but oppose Brexit, and 0 indicating those who have complementary preferences; namely, they support or oppose both processes (unconditional europhiles and power-seeking eurosceptics). Our findings suggest that it is economic self-interest rather than identity that determines whether individuals would support renegotiation but oppose Brexit. Those who perceived the country's membership in terms of cost were less likely to be risk-averse eurosceptics. A similar relationship holds for manual workers. Crucially, and confirming our hypotheses H4a and H4b educational level and Conservative party support matter. First, individuals with high levels of cognitive sophistication are more likely to see the potential benefits of a renegotiation agenda, and thus support it, but at the same time perceive Brexit to be a risk. Second, Conservative voters are more likely to be risk-averse eurosceptics, suggesting that by and large, at that time, these citizens were convinced about the utility of Cameron's EU reform plan, which may also go some way to explaining the unprecedented 2015 general electoral result for the party. UKIP voters seemed to be unconvinced about the utility of negotiation if that was not also associated with Brexit.

Discussion
Our findings contribute to our understanding of attitudes towards the EU in the UK by revealing that there were stark divides related not just to support for Brexit (Clarke et al., 2017;Curtice, 2016;Hobolt, 2016) but also Cameron's renegotiation. We show that similar utilitarian concerns related to the benefits arising from EU membership underpinned attitudes to both renegotiation and Brexit; but educational level and partisan cues differentiated them. Although it is plausible that educated voters may have viewed Cameron's renegotiation as a political stunt, our findings indicate that, in fact, they were more likely to support renegotiation. These results confirmed our expectation that renegotiation and Brexit are respectively hard and easy political issues, requiring different cognitive processes during the process of preference formation. Educated voters were more likely to grasp and understand the question of renegotiation, which was a complicated issue that required a nuanced understanding of diplomacy and public policy. Support for renegotiation in 2015 was found among Conservative and UKIP supporters. Therefore, Cameron had an opportunity to appeal to Conservative voters but also branch out to UKIP supporters. However, he also faced significant challenges in winning support for the renegotiation from less educated citizens and supporters of other political parties, especially Labour and Liberal Democrat voters. Cameron faced a greater challenge than Harold Wilson had done in 1974 to win support for his renegotiation. European integration meant that a renegotiation had become a much harder issue for citizens to understand. While Wilson had support from the majority of the opposition Conservative voters and a pro-EU media landscape, Cameron found that the Labour Party's criticism of the renegotiation and low-key support for EU membership, coupled with media criticism of the renegotiation, did little to assist him (Smith, 2016, p. 4).
Our findings have further implications for understanding the result of the 2015 UK general election and why the referendum failed to deliver the result that Cameron wanted. Our findings help to resolve this puzzle in two ways. First, they highlight the fact that the initial support for renegotiation hid the reality that a significant proportion of voters wanted a renegotiation followed by Brexit, and were thus unlikely to be convinced of the results of a renegotiation. Second, our findings suggest that voters perceived the attempt to renegotiate the UK's EU membership in terms of utility. This chimes with findings that suggest there has been a rise in self-interest (Evans and Chzhen, 2013) in shaping voting behaviour in the UK in recent elections. As Startin (2018, p. 459) notes, the Remain camp ran a single issue campaign around a rational choice argument that was unsuccessful. While this may have seemed to be a shrewd strategy, when voters were so concerned about the perceived costs of EU membership, it was a major challenge to convince them that the UK had gained significant results from renegotiation. Ultimately, only a minority of voters thought the renegotiation had achieved this and, as Curtice (2016) notes, the process may have encouraged more voters to vote leave than remain.
Finally, our findings have implications for the wider study of attitudes towards European integration and international organizations beyond the UK. Cameron's renegotiation is not simply an isolated event, given that politicians are calling for renegotiations of EU membership in a number of Member States. Future research is needed to examine the applicability of our typology of attitudes towards renegotiation in other EU Member States and in studying attitudes to renegotiating international treaties in other contexts, such as the USA. Our findings support the cognitive sophistication hypothesis, which asserts that renegotiations are hard issues (being both complex and technical) which highly educated individuals are more likely to grasp and engage with. If national government and EU reform initiatives are less likely to find support among less educated voters, then political elites may encounter a need to simplify the concepts behind them in order to appeal to the wider electorate. Alternatively, politicians may have reason to present renegotiations along partisan lines. Here, future research should focus on the role of the media in mediating such complex issues as well as the costs and benefits of EU membership to different sections of the electorate. According to the benchmark theory (De Vries, 2017), events related to Brexit may be used as a point of comparison influencing public opinion on the EU among voters residing in different EU Member States. Further research is therefore needed to investigate whether educational levels shape citizenship awareness and knowledge of the UK's renegotiation elsewhere.

Supporting Information
Additional supporting information may be found online in the Supporting Information section at the end of the article.