Policy, Office and Votes: Conservative MPs and the Brexit Referendum

Policy, Office and Votes: Conservative MPs and the Brexit Referendum Abstract The division within the Parliamentary Conservative Party (PCP) over Britain’s membership of the EU has been one of the most significant intra-party divisions in European political history. The 2016 Brexit referendum campaign offered a unique opportunity to consider legislative motivations as almost every MP declared a preference and frontbench MPs were free to back either side. This article uses logistic regression analysis in order to consider MPs’ motivations in terms of Müller and Strøm’s policy, office and votes trichotomy. It is argued that all three motivations affected MPs decision making on the EU referendum. However, vote-seeking motivations were less influential than either policy or office-seeking. Britain’s EU membership has been one of the most important issues in post-war Westminster politics. While divisions over Europe have at times affected both the major parties (Forster, 2002; Tzeglove, 2014), the European issue has had the most significant impact on the Conservative Party. The European issue has played a crucial role in at least three party leadership elections (Cowley and Garry, 1998; Heppell and Hill, 2008, 2010) and has caused many Parliamentary rebellions—particularly in the 1992 and 2010 Parliaments (Cowley and Norton, 1999; Cowley and Stuart, 2012; Hanretty et al., 2016). Since returning to power in 2010 a hard Eurosceptic element has emerged within the Parliamentary Conservative Party (PCP) (Heppell, 2013; Cowley and Stuart, 2012; cf. Szczerbiak and Taggart, 2008 for definitions of hard and soft-Euroscepticism). The emergence of this hard Eurosceptic group created a strong divide in the party between those who wished to leave the EU and those who, though in many ways Eurosceptic, nevertheless wanted to remain. The referendum campaign offered a unique opportunity to analyse attitudes within the PCP as the suspension of collective decision making meant that both front and backbench MPs were free to endorse either side, although the leadership had made it clear that it would rather MPs embraced remain1. This paper considers MPs’ referendum positions using multivariate logistic regression analysis. The data used is a bespoke dataset created by the author. Legislative motivations are analysed in terms of Müller and Strøm’s (1999) policy, office and votes trichotomy. It is argued that all three of these motivations affected MPs’ decision making on the EU referendum, although vote-seeking was of lesser importance than either policy or office-seeking. In terms of policy-seeking motivations, endorsing the leave campaign was strongly associated with social conservatism. The most likely reason for this association is that Euroscepticism within the Conservative party is influenced by traditional/authoritarian/nationalist values, particularly attitudes towards sovereignty and nationalism. The effect of office-seeking can be seen from the fact that an MP’s career status was an important predictor of preferences. The overwhelming majority of frontbench MPs who had the most incentives to be loyal to the leadership chose to follow the leadership in backing remain. The majority of experienced backbenchers, who had already resigned, been dismissed or overlooked for government office, opposed the leadership and backed leave. These MPs had the least incentives for loyalty to the leadership and indeed stood to potentially gain from the prospect of a post leave-vote Prime Ministerial resignation. MPs representing Eurosceptic constituencies were also found to be more likely to endorse leave, suggesting that vote-seeking motivations also affected MPs’ positions on the referendum. However, constituency opinion had a weaker effect on MPs’ referendum preferences than either personal ideology or office status. The article is structured as follows. Section 1 reviews the relevant literature concerning Euroscepticism and the PCP. Section 2 outlines the theory and hypotheses. Section 3 details the data and methods used. Section 4 presents the results which are discussed in Section 5 which concludes the article. 1. Euroscepticism and the PCP The PCP has always been, to a greater or less extent, divided on the issue of European integration since Britain first applied to join the EEC in 1961. The broad trend in this time is that a party largely supportive of the European project has become progressively more sceptical. In early 1960s, internal party sources estimated that only around 10% of Conservative MPs opposed entry into the Common Market (Crowson, 2007). In the 1970 Parliament, which passed the European Communities Act, it was estimated that nearly a third of Conservative MPs were either opposed to or doubtful about membership of the EEC (Norton, 1978). The majority of Conservative MPs were, however, supportive of European integration during 1960s and 1970s (Crowson, 2007). The late 1980s represented a turning point in the PCP as the Single European Act, Delors’ plan for a ‘Social Europe’, and Thatcher’s increased opposition to European integration produced higher levels of Euroscepticism within the PCP (Heppell et al., 2017; Fontana and Parsons, 2015). Garry’s (1995) survey research of the 1987 PCP suggested a party almost evenly split between pro and anti-integration MPs. Ideological mapping of the PCP by Heppell (2002, 2013) suggests that Eurosceptics have been a majority since 1992 and in opposition after 1997 the Europhiles became only a small minority of the party as the party was largely united in a Eurosceptic position. This Eurosceptic position was however, a soft Eurosceptic position which opposed further integration with Europe and advocated the repatriation of powers rather than withdrawal from the EU (Heppell et al., 2017). Euro-rejectionist sentiment during this period was rather muted as prior to the 2010 election only six Conservative MPs had publicly declared support for leaving the EU (Cowley and Stuart, 2010). A hard Eurosceptic element, demanding either withdrawal or a fundamental change in the UK relationship with the EU, did become more visible after the Conservatives returned to power in 2010. Heppell (2013) found that 26% of the 2010 PCPs were hard Eurosceptics. This created a divide in the Conservatives between a majority of the PCP supporting a soft-Eurosceptic reformist position and a notable minority advocating a hard-Eurosceptic rejectionist position. The number of hard Eurosceptics grew in the 2015 PCP as over 40% of Conservative MPs backed leave (Heppell et al., 2017). To summarize, the research literature suggests that the PCP has become progressively more Eurosceptic since the 1960s and that the 2015 PCP is the most Eurosceptic PCP to date. There have been only very limited attempts to understand the drivers of Euroscepticism with the PCP. Garry (1995) in his survey of the 1987 Parliament found that Eurosceptics tended to be both economically right-wing and socially conservative. The weakness of this study was that relying on survey data meant that the analysis included less than half of the PCP (45.1%). Heppell (2002) similarly found, for the 1992 Parliament, correlations between Euroscepticism and both economically dry and socially conservative positions. For the same Parliament, Berrington and Hague (1998) analysed attitudes towards the Maastricht Treaty using principal component analysis of Early Day Motions (EDMs). Their results showed numerous statistical differences between those for and against the Treaty, for example anti-Maastricht (i.e. Eurosceptic) MPs were more likely to be pro-death penalty and less likely to have been educated at a top public school. The use of EDM data however meant that Berrington and Hague’s sample was limited to only 155 MPs (46.1% of the total PCP) and included no frontbench MPs. For the 2015 Parliament, Heppell et al. (2017) found several statistical differences between remain and leave MPs. For instance, leave MPs were more likely to have served in military and to have opposed gay marriage, while remain MPs were more likely to have been ministers and attended an Oxbridge university. Heppell et al. (2017) used the same dependent variable and a similar methodology to this article; however, they do not directly address the issue of how MPs’ behaviour on Brexit may have been influenced by competing motivations. This article seeks to build upon the previous literature by creating a model to assess MPs’ motivations in terms of Müller and Strøm’s (1999) policy, office and votes trichotomy. The decision to formally suspend the Whip for the referendum campaigns presents a novel opportunity to be able to consider legislative motivations on both the back and frontbenches. By using this opportunity the article aims to contribute to a greater understanding of what motivates Euroscepticism within the PCP. 2. Theory and hypotheses The literatures on both the Conservative Party and legislative behaviour have informed the theory and hypotheses of this paper. Kam (2009) argues that there is consensus in the legislative behaviour literature around Müller and Strøm’s (1999) assertion that politicians seek a combination of three aims: policy, office and votes. In the long term, these three goals are not necessarily mutually exclusive and indeed may be mutually reinforcing. Re-election and promotion to ministerial office put an MP in a better position to influence policy. In Westminster systems particularly, government control of agenda setting means that MPs are much more likely to influence policy as ministers than as backbenchers (Benedetto and Hix, 2007). Moreover, Martin (2016), using data from Irish elections, argues that ministers can use their increased influence over policy to gain an electoral advantage over their co-partisans. He argues that ministers use ‘executive particularism’ (pork-barrelling) to specifically target government spending at their own constituencies. However, on individual policies there may be occasions when the three aims conflict. If on a particular policy the views of either the MP herself, her constituents, or the party leadership are not aligned the MP will have to choose to prioritise one aim over another. For instance if the party leadership and the MP hold different policy positions, the MP would either have to sacrifice her policy preferences on this particular issue in the hope of being able to influence other policy decisions in the future or sacrifice her chances of promotion to stay true to her policy preference. The EU referendum provides a good example of where conflict between these aims might arise as on the one hand there was a party leadership backing remain and on the other hand deep-seated Euroscepticism on the Conservative backbenches and in the grassroots party. Below, I consider each of the three motivations: policy, office and votes in order to generate the hypotheses for this study. 2.1 Policy-seeking When reviewing the recent history of the Conservative party and Euroscepticism, there is a strong case that politicians have been motivated more by their own policy agendas rather than electoral demand. The evidence for this is that since at least the 1990s the European issue has been of crucial importance for Conservative Party politics but of more limited salience for the wider electorate (Bale, 2006: Bartle, 2002). If policy-seeking was the main objective of Tory MPs concerning Europe, one would expect to find that ideology was highly important in predicting MPs’ positions. Apart from Europe, the other main policy divide in the PCP in recent times has been the social/moral policy divide which separates social conservatives from social liberals2 (Hayton, 2010). Social conservatives hold traditional/authoritarian policy positions, taking a more restrictive stance on issues such as abortion and gay rights, while advocating tougher punishment for criminals. Social liberals by contrast have a more permissive approach to issues such abortion and gay rights and a less punitive approach to crime and punishment. Previous empirical studies on Conservative MPs and grassroots members have shown strong statistical associations between Euroscepticism and social conservativism (Garry, 1995; Berrington and Hague, 1998; Heppell, 2002, 2013; Webb and Bale, 2014). Similar links between social conservatism and Euroscepticism have been found more broadly in European politics. For example Tillman (2013) finds, based on European Values Survey data, that authoritarian preferences over child rearing are linked to Euroscepticism and Prosser (2016) argues that party positions on European integration are primarily affected by the social liberal-conservative axis of politics. Specifically on the Brexit referendum, Kaufman (2016) finds strong statistical links between supporting harsher treatment of criminals and the probability of voting leave. The British Electoral Study (BES) data he uses shows that 71% of those who supported the death penalty indicated they would vote leave. This was a stronger predictor of voting intentions than education, income, social class or party support. The reason why statistical associations between social conservatism and Euroscepticism have been found cannot be established with any certainty. One plausible explanation for this is that positions on Europe could be related to the GAL (green/alternative/libertarian) versus TAN (tradition/authoritarian/nationalist) dimension of political contestation (Hooghe et al., 2002). Parties or politicians with a TAN orientation ‘combine support for traditional values, opposition to immigration and defence of the national community’ (Hooghe et al., 2002, p. 976). Hooghe et al. (2002) argue that Euroscepticism is often advocated by political parties with a TAN orientation as European integration is seen as a threat to the national sovereignty and community. Berrington and Hague (1998) allude to a connection between TAN policy positions and Euroscepticism amongst Conservative MPs in their study of Conservative MPs and the Maastricht Treaty. They find that pro-death penalty MPs were more likely to have opposed the Treaty. They connect this to the political psychology literature, arguing that attitudes towards harsher punishment of criminals were related to attitudes towards ethnocentrism, imperialism and intolerance (ibid). More recent contributions to the political psychology literature have also established a link between social conservatism and nationalism (Smith et al., 2011). Therefore, there could be psychological reasons underlying the relationship between social conservatism, nationalism and Euroscepticism. For Conservative MPs, a link between social conservatism and Euroscepticism could signify the importance of TAN values, particularly attitudes towards nationhood and sovereignty, as many in the literature have argued that nationalism is one of the main causes of Conservative Party Euroscepticism (Berrington and Hague, 1998; Turner, 2000; Crowson, 2007; Webb and Childs, 2011). The literature suggests, then, a strong relationship between social conservatism and Euroscepticism. This generates hypothesis 1 below: Hypothesis 1: Socially conservative MPs were more likely to endorse leave 2.2 Office-seeking motivations The Westminster system in which the party leadership controls ministerial appointments provides strong incentives for MPs seeking a ministerial career to obey the party whip (Benedetto and Hix, 2007; Kam, 2009). The literature has established that ministers and those MPs who still have a rational hope of climbing the ministerial ladder are less likely to rebel (Benedetto and Hix, 2007; Kam, 2009). This literature on rebellion is relevant because even though the EU referendum was formally unwhipped, and indeed six cabinet ministers joined leave, the official government position was for remain and the Prime Minister was seeking to persuade MPs to back remain. This means that although backing leave did not result in an automatic dismissal, as rebelling on a whipped vote would do, the Cameron leadership was more likely to reward those who backed remain. Possible evidence that remain MPs would be rewarded can be seen from the mini reshuffle following Iain Duncan-Smith’s resignation when the Prime Minister promoted only MPs who backed remain. This means that MPs who were climbing the ministerial ladder or who had ambitions of climbing the ladder could be expected to be more likely to back remain in order to satisfy Cameron. In contrast, MPs who had left office, through either dismissal or resignation, or who had been overlooked for a government post are predicted to have been less likely to endorse remain as they had less reason to believe that Cameron would reward them after the referendum. Benedetto and Hix (2007) found that ex-ministers and MPs overlooked for ministerial office are the most likely to rebel as they cannot be controlled by the promise of ministerial office. Kam (2009) in his comparative study of Westminster systems similarly argued that MPs who could no longer attain office were more likely to rebel. By the same logic MPs who have opted not to pursue a ministerial career, in order to focus on either constituency and/or select committee work, will also have been less inclined to be loyal to the leadership. Office-seeking calculations of MPs could also have been affected by the fact that a leave victory in the referendum was almost certain to result in a Prime Ministerial resignation. Backbenchers out of favour with the leadership, i.e. those sacked or overlooked for office, had a potential incentive to endorse leave as a new leader may give them a better chance of gaining a government position. Conversely, for frontbenchers, all of whom were appointed by Cameron, a change of leadership would make their positions less secure. Indeed, this is what happened after the leave vote: the new Prime Minister, Theresa May, brought in MPs who had been out of office for a long time such as Liam Fox and David Davis and dismissed long-serving ministers from the Cameron administration such as George Osborne and Nicky Morgan. From this discussion of office-seeking intentions, it is predicted that there will be three distinct groups within the PCP which vary in their likelihood of endorsing leave. The group most likely to back remain are frontbenchers. Frontbenchers had incentives to endorse remain as this would increase their chances of promotion under Cameron whilst a potential change of leader would threaten their position in the government. The group most likely to endorse leave are backbenchers elected before the 2015 election. These MPs have all been in Parliament long enough to have potentially been offered a frontbench position, as can be seen by the fact that a majority of Tory MPs first elected at the 2010 election, as well as Robert Jenrick who was elected in a 2014 by-election, were serving in government positions at the time of the referendum. This group of experienced backbenchers includes ex-ministers, those overlooked for government office and MPs who have chosen not to pursue a ministerial career. They have little incentive to be loyal to Cameron and may benefit from a change of leader in the event of a leave vote. Concerning MPs newly elected in 2015, this group is expected to have been more likely to endorse remain than the more experienced backbenchers as new MPs would have had a realistic potential of being offered a government post by Cameron had the remain side won the referendum campaign. However, new MPs are predicted to have been more likely to endorse leave than frontbenchers. The reason for this is that some newly elected MPs may have had no office-seeking ambitions and newly elected MPs with office-seeking motivations would be less likely to have been detrimentally affected by a change of leader than frontbench MPs. This analysis leads to hypotheses 2a–c below: Hypothesis 2: Office status was a strong predictor of an MP’s referendum position. 2a. Frontbench MPs were the most likely MPs to endorseremain. 2b. Backbench MPs elected before 2015 were the most likely MPs to endorse leave. 2c. Newly elected MPs were more likely to endorse leave than frontbench MPs but less likely to endorse leave than long serving backbench MPs. 2.3 Vote-seeking It is reasonable to suggest that vote-seeking and re-election is a concern for many MPs as most sitting MPs put themselves forward as candidates at the next election (Matland and Studlar, 2004). There is increasing evidence in the literature that British MPs’ roll-call records can affect their electoral performance (Pattie et al., 1994; Kam, 2009). Despite this the only study to specifically consider Conservative Eurosceptics concluded that the Europe issue did not affect their vote share (McAllister and Studlar, 2000). McAllister and Studlar found that declared Eurosceptics had a vote-share 2.3% higher than other Conservative MPs at the 1997 election but that almost all of this difference could be accounted for by incumbency and turnout differences across constituencies. However, of more importance than whether MP’s voting records affect their electoral vote share, is whether MPs think (rightly or wrongly) that voters react to their personal policy positions. Vote-seeking MPs would rationally choose their referendum position based on public opinion in their constituencies if they believed that this would affect their personal vote. A number of studies have considered whether MPs vote based on public opinion in their constituencies. For instance, Baughman (2004) found that MPs representing constituencies with a high number of Catholics were more likely to take conservative positions on abortion votes. Most significantly for this study, Hanretty et al. (2016) found that MPs’ positions in the 2010 Parliament were responsive to constituency opinion on three policy issues, one of which was Europe. They demonstrated that constituency disapproval of the EU was linked to a higher probability of a Conservative MP adopting Eurosceptic positions on both whipped and free votes. If MPs were vote-seeking, then, we should expect to see similar results to Hanretty et al. (2016) for the referendum, with MPs representing Eurosceptic constituencies being more likely to support leave. Hypothesis 3: There was a positive relationship between constituency Euroscepticism and the probability of MPs’ endorsing leave. The MPs who are most likely be most concerned with vote-seeking are those representing marginal and semi-marginal seats. As marginal seats can change hands on the basis of relatively small swings away from the incumbent, sitting MPs in marginal constituencies might be very attentive to opinion in their constituency. MPs representing safer seats have less reason to fear electoral defeat and may be less attentive to constituency opinion. This discussion leads to Hypothesis 4: Hypothesis 4: The relationship between constituency Euroscepticism and referendum position was stronger for MPs who represented marginal constituencies. 3. Data and methods This article uses a dataset of Conservative MPs elected at the 2015 general election. The dataset was created by the author. Logistic regression analysis is used in order to test the hypotheses identified in Section 2. As this study uses observational data, it makes no claim to have identified causal processes and the author accepts alternative interpretations of the findings are possible. However, by thoroughly and accurately presenting the statistical associations between the variables the author believes that this study can contribute to the literatures on legislative behaviour in British politics and Conservative Party Euroscepticism. Sections 3.1–3.6 below detail the variables used in the dataset; the statistical models used are explained in Section 3.7. 3.1 The dependent variable—MPs’ referendum endorsements The dependent variable is a binary variable which measures MPs’ publicly declared support in the referendum campaign. MPs who publicly declared support for leave were coded as 1 and MPs who publicly supported remain were coded 0. MPs who did not declare a position were treated as missing data. As only 6 (2%) MPs did not declare a preference, few MPs were excluded from the analysis. Table 1 displays the summary statistics of MPs’ referendum positions. It demonstrates that the majority of MPs did side with the Prime Minister and support remain. However, a substantial proportion (41%) of Conservative MPs backed leave. Table 1 Referendum positions of conservative MPs Stance  n (%)  Remain  189 (57)  Leave  135 (41)  Did not declare  6 (2)  Stance  n (%)  Remain  189 (57)  Leave  135 (41)  Did not declare  6 (2)  Sources: MPs’ personal websites, BBC News (http://www.bbc.co.uk/news/uk-politics-eu-referendum-35616946) and TheyWorkForYou.com. 3.2 Social conservatism A ‘social conservatism’ variable was developed in order to test Hypothesis 1. To generate the social conservatism variable a scale was created using a similar approach to that taken by Read and Marsh (1997), and Heppell (2002, 2013; Heppell and Hill, 2008, 2010). An ideological scale was created based on MPs’ positions on five issues: Gay Marriage Abortion Death Penalty Three-parent embryos Greater use of imprisonment as opposed to community sentences For each issue an MP was given 2 for a socially conservative stance, 0 for a socially liberal stance or 1 where either no data could be found or the MP was inconsistent on the issue. After the sum of the five issues was calculated, each MP had a score on a scale 0–10, with higher numbers denoting greater social conservativism. Data were gathered for each MP using a combination of their House of Commons voting record, record of signing relevant EDMs, membership of All Party Parliamentary Groups and public comment made by the MP3. The resulting summary statistics of social conservatism show that Conservative MPs were spread across the range of the scale, with three MPs being socially liberal on all five issues and seven MPs being socially conservative on every issue. Figure 1 above contains a histogram of the distribution of Conservative MPs on the social conservatism scale. The variable is skewed slightly to the conservative side of the spectrum with the mean score being 5.11. Figure 1 View largeDownload slide Histogram of MPs on the social conservatism scale. The social conservatism variable ranges 0 to 10 with higher values denoting stronger social conservatism. Social conservatism scores for each MP based on their positions on five issues: abortion, gay marriage, death penalty, use of three parent embryos and greater use of imprisonment as opposed to community sentences. For each issue, an MP was given 2 for a socially conservative stance, 0 for a socially liberal stance or 1 where either no data could be found or the MP was inconsistent on the issue. Figure 1 View largeDownload slide Histogram of MPs on the social conservatism scale. The social conservatism variable ranges 0 to 10 with higher values denoting stronger social conservatism. Social conservatism scores for each MP based on their positions on five issues: abortion, gay marriage, death penalty, use of three parent embryos and greater use of imprisonment as opposed to community sentences. For each issue, an MP was given 2 for a socially conservative stance, 0 for a socially liberal stance or 1 where either no data could be found or the MP was inconsistent on the issue. 3.3 Office-status To test Hypothesis 2, that office-status would affect referendum position, a categorical variable ‘office status’ was created with three categories: Experienced backbenchers (n = 125): This is the baseline category. MPs are included in this category if they were elected before the 2015 election. This group of MPs is on the backbenches through dismissal, resignation or being overlooked for government positions or choosing not to pursue a ministerial career. New MPs (n = 72): These MPs were elected for the first time in 2015. At the time of the referendum they were too inexperienced to have been offered a government post. Frontbenchers (n = 127): MPs were included in this category if they were a minister, whip or Parliamentary Private Secretary on the day of the referendum. 3.4 Constituency Euroscepticism A measure of constituency Euroscepticism was used in the model to examine whether MPs’ vote-seeking motivations affected their referendum positions. To measure constituency Euroscepticism, data from Hanretty et al. (2016) was used. They estimated Euroscepticism in each constituency using data from the BES 2010 Combined Internet Panel Survey which asked respondents if they disapproved of the UK’s EU membership. Hanretty et al. (2016) used multilevel regression and post-stratification analysis of the survey responses to obtain an EU disapproval percentage for each constituency. The mean value of this variable for Conservative constituencies was 53.6 (indicating that 53.6% of constituency residents disapproved of the EU) and the median was 54.5. Scores on the constituency Euroscepticism scale ranged from a low of 24.1 for Jane Ellison who represented Battersea in inner London to a high of 72.0 for Rebecca Harris who represented Castle Point in Essex. As the constituency Euroscepticism variable took account of numerous constituency demographic factors, such as percentage non-white, median education level and region, no other constituency demographic variables were included in the dataset for this study. 3.5 Seat marginality In order to test Hypothesis 4, that MPs in marginal seats were more responsive to constituency Euroscepticism, two variables were added to the dataset, one dummy variable measuring whether a seat was marginal or safe and an interaction variable of marginal seats and constituency Euroscepticism. There are no firm rules about the distinction between marginal, semi-marginal and safe-seats. Norris and Crewe (1994) defined a seat as either marginal or semi-marginal if the winning party gained a majority of 14.9% or lower over the second placed party. On this basis, this article uses the 14.9% winning margin as the threshold for coding seats as 1 in the marginal seats dummy variable. This results in 95 MPs (29%) being coded as 1, indicating that they represented marginal or semi-marginal seats, and 229 MPs (71%) being coded as 0, indicating that they represented comparatively safe seats. The interaction variable, which seeks to test whether MPs from marginal seats were more responsive to constituency Euroscepticism, consists of the marginal seat dummy variable multiplied by constituency Euroscepticism. As establishing a threshold between safe and marginal seats is to some extent arbitrary several different versions of the test were performed using different thresholds to check the validity of any findings.4 3.6 Control variables The dataset also includes control variables which could also have potentially predicted which side MPs endorsed in the referendum campaign. These variables are considered below: Age: Polling suggested that age was an important predictor of voting amongst the general public in the Brexit referendum with older generations voting more for leave and younger generations voting more for remain. It may have been the case that opinion trends within the PCP were similar to those in the wider public. Gender: A dummy variable was also included denoting an MP’s gender (female = 1). It might be case that female MPs are less likely to be pro-leave as many of the most prominent figures in Conservative Euroscepticism such as John Redwood, Bill Cash and Andrew Rosindell are male. Private school: This dummy variable denotes whether an MP attended a fee paying secondary school. A connection between MPs’ schooling and Europe was found by Berrington and Hague (1998) who found that MPs from the top public schools were less likely to take Eurosceptic positions. 3.7 The statistical models As the dependent variable is dichotomous, logistic regression models are used to analyse the data. The basic structure of the statistical models is:   Pr(leave=1)=logit−1(α+β1X1+β2X2+β3X3) Where α = the constant, βs = the variables and Xs = the logit coefficient for each variable. The baseline model (model 1) includes the social conservatism variable, office status variable and Constituency Euroscepticism for all MPs in the dataset. These are the three primary variables of interest in testing for MPs’ policy, office and vote-seeking intentions. The equation for this model is below:   Pr(leave=1)=logit−1(α+SocialConservatism1+OfficeStatus1+ConstituencyEuroscepticism1) Model 2 builds on the first model by including the three control variables listed in Section 3.6. The equation of this model is below:   Pr(leave=1)=logit−1(+SocialConservatism1+OfficeStatus1+ConstituencyEuroscepticism1+Age1+Gender1+PrivateSchool1) Model 3 tests whether the effect of constituency Euroscepticism differs for marginal and safe seats. This model adds the marginal seat dummy variable and an interaction variable: marginal × constituency Euroscepticism. The equation of this model is below:   Pr(leave=1)=logit−1(α+ SocialConservatism1+OfficeStatus1+ConstituencyEuroscepticism1+Age1+ Gender1+PrivateSchool1+MarginalSeat+MarginalSeat×ConstituencyEuroscepticism1) As a robustness check on the results, the linear probability (OLS) and probit versions of the models used in the statistical analysis are presented in the Appendix 1. 4. Results Table 2 below presents the results from the statistical analysis. The three models used in the table are described in Section 3.7. As logistic regression coefficients have no direct meaning, Table 2 presents the average marginal effects of the independent variables. The average marginal effects refer to the average effect of a 1 unit change in the independent variable. For example, Model 1 predicts that on average an increase of 1 on the social conservatism scale increases an MP’s probability of supporting leave by 0.05, and Model 2 predicts that the effect of being a frontbencher reduces the probability of supporting leave by 0.37 compared to an experienced backbencher. The Appendix 1 shows the logistic regression coefficients for the statistical models in Table 2. Table 2 Marginal effects on the probability of endorsing leave   Model 1 (Main variables)  Model 2 (includes demographic controls)  Model 3 (controls for constituency majority)  Social conservatism  0.050 (0.010)***  0.049 (0.011)***  0.050 (0.010)***  Office status reference category: experienced backbenchers  Newly elected MPs  −0.171 (0.069)**  −0.209 (0.087)**  −0.199 (0.090)**  Frontbenchers  −0.334 (0.058)***  −0.369 (0.061)***  −0.366 (0.061)***  Constituency Euroscepticisma  0.074 (0.031)**  0.058 (0.032)*  0.078 (0.038)**  Age    −0.004 (0.003)  −0.004 (0.003)  Gender (Female = 1)    −0.038 (0.068)  −0.042 (0.069)  Private school    −0.048 (0.050)  −0.047 (0.050)  Marginal constituency      0.345 (0.376)  Marginal × Constituency Euroscepticism      −0.007 (0.007)  Observations  324  309  309  Count R2  0.704  0.709  0.709  LR χ2b  66.17  68.28  69.91  p > χ2  >0.001  >0.001  >0.001    Model 1 (Main variables)  Model 2 (includes demographic controls)  Model 3 (controls for constituency majority)  Social conservatism  0.050 (0.010)***  0.049 (0.011)***  0.050 (0.010)***  Office status reference category: experienced backbenchers  Newly elected MPs  −0.171 (0.069)**  −0.209 (0.087)**  −0.199 (0.090)**  Frontbenchers  −0.334 (0.058)***  −0.369 (0.061)***  −0.366 (0.061)***  Constituency Euroscepticisma  0.074 (0.031)**  0.058 (0.032)*  0.078 (0.038)**  Age    −0.004 (0.003)  −0.004 (0.003)  Gender (Female = 1)    −0.038 (0.068)  −0.042 (0.069)  Private school    −0.048 (0.050)  −0.047 (0.050)  Marginal constituency      0.345 (0.376)  Marginal × Constituency Euroscepticism      −0.007 (0.007)  Observations  324  309  309  Count R2  0.704  0.709  0.709  LR χ2b  66.17  68.28  69.91  p > χ2  >0.001  >0.001  >0.001  Note: The numbers in the table are average marginal effects with robust standard errors in parentheses. The Count R2 refers to the proportion of correctly predicted outcomes from the model. *** p < 0.01, ** p < 0.05, * p < 0.1. aOne unit of constituency Euroscepticism = 10% disapproval of the EU, e.g. a constituency with 45.6% EU disapproval is coded as 4.56. bMeasure based on pseudolikelihood as robust standard errors have been used. The results found in the linear probability and probit models, presented in the Appendix 1, are qualitatively similar to those found in the logistic regression models. The results will now be considered in terms of policy, office and vote-seeking. 4.1 Policy-seeking The data provides strong evidence for a link between social conservatism and the probability of an MP endorsing leave. This relationship is highly statistically significant in all of the models. In addition to being statistically significant the relationship is also substantially significant as the models suggest that social conservatism has a strong effect on the probability of endorsing leave. Model 1 predicts that at the most socially conservative side of the scale an experienced backbencher has an 83% probability of supporting leave compared to 28% for a similar MP at the most socially liberal end of scale. Figure 2 plots this relationship. For frontbench MPs the model predicts that the probability of endorsing leave varies from 7% to 51% across the social conservatism scale. These results provide strong evidence to validate Hypothesis 1. This suggests that a TAN ideology among MPs could be motivating Euroscepticism. Figure 2 View largeDownload slide Marginal effect of social conservatism on the probability of endorsing leave. This is based on data from Model 1. The probabilities shown in the graph are for a backbench MP first elected before 2015. Social conservatism is measured on a scale of 1–10, see section 4.2. Figure 2 View largeDownload slide Marginal effect of social conservatism on the probability of endorsing leave. This is based on data from Model 1. The probabilities shown in the graph are for a backbench MP first elected before 2015. Social conservatism is measured on a scale of 1–10, see section 4.2. 4.2 Office-seeking Consistent with the predictions of Hypothesis 2, the results show substantial differences in referendum position between three career-status groups of MPs: experienced backbenchers, frontbenchers and new MPs. The difference between each of these three groups is statistically significant in all of the statistical models. The marginal effects indicate that the effect of office status on referendum position is strong. Model 3 suggests that being on the frontbench reduces the probability of supporting leave by 0.37 and being a new MP reduces the probability by 0.20, compared to an experienced backbencher. This suggests that the difference between being an experienced backbencher and a frontbencher is the equivalent of the effect of a 7.5 point difference (out of a possible 10) on the social conservatism scale and the difference between a new and experienced backbencher is the equivalent of a 4 point difference on the social conservatism scale. The summary statistics, see Figure 3 below, also demonstrate the sharp differences between the three groups. The only group with a majority supporting leave were the experienced backbenchers, 61% of whom supported leave. This is in contrast to less than a quarter (24%) of frontbench MPs supporting leave. Newly elected MPs, as predicted, occupy the middle group but are slightly closer to the frontbenchers with 38% supporting leave. Figure 3 View largeDownload slide Referendum support by office-status. Numbers in the bars refer to raw number of MPs. Experienced backbenchers refers to backbench MPs elected before 2015. Pearson χ2 (2 degrees of freedom) = 37.7; p < 0.001 Figure 3 View largeDownload slide Referendum support by office-status. Numbers in the bars refer to raw number of MPs. Experienced backbenchers refers to backbench MPs elected before 2015. Pearson χ2 (2 degrees of freedom) = 37.7; p < 0.001 These findings are consistent with the discussion in Section 2.2 which predicted that MPs’ office-seeking incentives, or lack thereof, to support the leadership would affect their referendum positions. 4.3 Vote-seeking The statistical analysis also provides evidence that vote-seeking motivations affected MPs’ positions on the referendum. All three models show a positive relationship between constituency Euroscepticism and the probability of an MP endorsing leave. This relationship is statistically significant to the 95% level in Models 1 and 3 and the 90% level in Model 2. These findings are in line with the predictions of Hypothesis 3. These data suggest that, across the range of the variable, constituency Euroscepticism could have had a strong effect on MPs’ referendum stance. Model 1 predicts that a frontbench MP representing a highly Eurosceptic constituency (70% against the EU) had a probability of endorsing leave of 0.37, more than three times higher than the probability, 0.11, of a frontbench MP representing a highly pro-EU constituency (25% against the EU). The same calculation for experienced backbenchers predicts that those representing a highly Eurosceptic constituency had a 0.72 chance of supporting leave compared to 0.35 for those representing a highly pro-EU constituency. There is, then, evidence to suggest that vote-seeking was an important factor affecting which side MPs backed in the referendum campaign. However, there are two caveats to this general finding. Firstly, there is no evidence that constituency Euroscepticism had a greater effect on MPs from marginal constituencies. The interaction variable of marginality and constituency Euroscepticism in Model 3 is not statistically significant. Several different thresholds for marginal constituencies were used in order to further test if an interaction could be found between constituency Euroscepticism and seat marginality5. None of these tests, however, produced a statistically significant result. Contrary to the predictions of Hypothesis 4, this suggests that MPs representing marginal seats were not more sensitive to constituency opinion than those representing safer seats. This could suggest that MPs from safe seats are equally as concerned with vote-seeking as those from marginal seats. The second caveat is that the effect of constituency Euroscepticism is weaker than both the effect of office status and social conservatism. All the models suggest that social conservatism is a better predictor of referendum preferences than constituency Euroscepticism. For example, Model 3 predicts that social conservatives in the most pro-EU seats were more likely to endorse leave and that social liberals in the most Eurosceptic constituencies were more likely to endorse remain. This suggests that when a potential conflict arose between an MP’s personal ideology and the opinion of their constituents, MPs were more likely to be guided by their own ideology. In relation to office-seeking, all the statistical models predict that a frontbench MP representing one of the most Eurosceptic constituencies was not statistically more likely to support leave than a backbench MP representing a strongly pro-EU constituency. This suggests that the effect of constituency Euroscepticism is relatively weak when compared to office status. In sum, the findings show a statistical link between constituency Euroscepticism and the likelihood of an MP adopting endorsing leave. However, no evidence was found that MPs in marginal constituencies were more responsive to constituency opinion. Furthermore, social conservatism and office status were stronger predictors of MPs’ referendum positions than constituency Euroscepticism. This suggests that whilst vote-seeking motivations may have been important for MPs they were of less importance than either ideology or office status. 5. Conclusion This article has sought to investigate why Conservative MPs endorsed different sides in the 2016 Brexit referendum. It has been argued that all three of Müller and Strøm’s (1999) aims: policy, office and votes, affected MPs’ positions on the referendum. However, vote-seeking motivations were of lesser importance than either policy or office-seeking motivations. Policy-seeking motivations appear to be highly influential as there is a very strong statistical relationship between social conservatism and endorsing leave. This relationship between social conservatism and Euroscepticism cannot be explained with any certainty. A plausible explanation is that MPs’ positions on the referendum are related to the TAN/GAL dimension of politics with TAN MPs being more likely to oppose the EU on the grounds that it threatens national sovereignty. This argument concerning the TAN/GAL political dimension and Euroscepticism is consistent with many in the literature who have argued that Conservative Euroscepticism is essentially nationalistic in nature (Sowemimo 1996; Berrington and Hague 1998; Crowson, 2007; Turner, 2000). Office-seeking motivations were also of crucial importance in influencing MPs’ position on the referendum. The statistics show notable differences between in the referendum positions of three distinct career status groups. Frontbench MPs were predominantly loyal to the Prime Minister in supporting the remain campaign. In contrast, a majority of experienced backbenchers endorsed leave. This is consistent with the theory that those who are climbing the ministerial ladder or those who have a realistic prospect of gaining a frontbench role are more likely to be loyal to the leader (Benedetto and Hix, 2007). Experienced backbenchers, on the other hand, have fewer incentives to be loyal as they may already have been dismissed from, or overlooked for, ministerial office (Benedetto and Hix, 2007). The differences between these groups of MPs may also have been affected by the fact that a post leave-vote leadership change would represent a potential career threat to frontbenchers but an opportunity for backbenchers out of favour with the Cameron leadership. In terms of vote-seeking, the models show a statistical association between constituency Euroscepticism and endorsing leave. This possibly indicates that MPs were seeking to improve their personal vote by following their constituents. However, there is no evidence that MPs representing marginal constituencies were more responsive to constituency opinion as might be expected if MPs were motivated by vote-seeking. Furthermore, the effect of constituency Euroscepticism is weaker than the effects of either social conservatism or office status. These results suggest that while vote-seeking may have been a motivation for MPs in the referendum, their personal policy preferences and career prospects had a stronger impact on their stance than constituency opinion. As the Brexit debate has several unique features such as the fact it divides parties, Conservative MPs have very strong views on Europe and that the issue was determined by referendum rather than Parliamentary vote, it is likely that the ordering of MPs’ priorities will be different for other issues. For example, on many routine areas of government policy like reform of the planning system, MPs may be less influenced by policy-seeking as most MPs will not have strongly held and long-established views on these issues. Moreover, on some issues MPs may be expected to place a higher priority on vote-seeking than in the Europe debate. As this debate was decided by a referendum, meaning the public not the politicians determined the result, MPs might reason that constituents were less likely to punish or reward them for their stance. The impact of vote-seeking on MPs could have also been supressed by the fact that most constituencies were divided. Only 82 (21%) Conservative MPs represented seats predicted to have 60% or more voting the same way in the referendum, meaning that most MPs would find themselves opposing at least 40% of their constituents whichever side they backed in the referendum. On issues where constituencies are highly united MPs may be more guided by vote-seeking considerations. For example, MPs representing constituencies on the Heathrow flight path or High Speed 2 rail route, whose residents could be expected to be united in opposition to these projects, might fear the electoral consequences if they were to take a different position to their constituents. The relative importance of vote-seeking to MPs could also be affected by the electoral system. The theoretical literature suggests that in single member plurality (SMP) systems, like the UK, MPs have more incentive to develop a personal vote than in party-centred systems such as closed-list PR where MPs are elected solely on the basis of their party label (Mitchell, 2000; Nielson 2003). However, empirical cross-national studies of both legislators’ perceptions (André et al., 2016) and voting behaviour (Söderlund, 2016) suggest that SMP systems provide less incentives for MPs to cultivate a personal vote than single-transferable vote (STV) systems where MPs have to compete against co-partisans to attract votes. The findings of this study, that there is evidence of vote-seeking but that the effect was weaker than either policy or office-seeking considerations, potentially conform to the literature as it is possible that under a closed list system the effect of vote-seeking could have been weaker and conversely under STV the effect of vote-seeking could have been stronger. Future research is needed on how legislators in different electoral systems prioritise their competing motivations on controversial topics like Europe. The conclusion of this article that policy-seeking and office-seeking are more important than vote-seeking suggests that the internal politics of the Conservative Party had a greater effect on MPs’ referendum positions than the wider electorate. This is not unexpected given that in the recent history of the Conservative Party, Europe has been one of the defining policy divides within the internal politics of the party but has been of only limited salience to the wider electorate (Bale, 2010). As this study has used observational data, the findings here should be treated with caution and the author accepts that alternative explanations could be reached from the data presented here. This paper makes no claim to have identified causal processes. Nevertheless, the author believes that this paper makes an important contribution to both the literature on the Conservative Party and Euroscepticism and the literature concerning legislative behaviour in British politics. Footnotes 1 The two referendum positions are referred hereafter as leave and remain. 2 Another policy divide which was not considered in this paper is the economic left/right divide. This was not considered as there is a general consensus in literature that the economic left (sometimes called wet) element of the party has declined substantially since the 1980s and 90s (Bale 2006; Heppell 2013). 3 The timeframe for the parliamentary activity (voting record, EDMS and APPGs) is from the MP’s first election to parliament and public comment takes account of any comments made by the MP since they became a parliamentary candidate. 4 Different versions of the test were tried using 4.9%, 9.9% and 19.9% majorities as the thresholds for marginal seats. 5 In the other tests 4.9%, 9.9% and 19.9% majorities were used as the thresholds for the marginal seat dummy variable Acknowledgements The author would like to thank David Hine, James Tilley, Raphael Heuwieser, Lewis Wilcocks and the two anonymous reviewers for their helpful comments and feedback. Bibliography André A., Depauw S., Martin S. ( 2016) ‘The Classification of Electoral Systems: Bringing Legislators Back In’, Electoral Studies , 42, 42– 53. Google Scholar CrossRef Search ADS   Bale T. ( 2006) ‘ Between a Soft and a Hard Place? The Conservative Party, Valence Politics and the Need for a New Eurorealism’, Parliamentary Affairs , 59, 385– 400. Google Scholar CrossRef Search ADS   Bale T. ( 2010) The Conservatives from Thatcher to Cameron , Cambridge, Polity Press. Baughman J. ( 2004) ‘ Party, Constituency, and Representation: Votes on Abortion in the British House of Commons’, Public Choice , 120, 63– 85. Google Scholar CrossRef Search ADS   Bartle J. ( 2002) ‘Why Labour Won Again’. In King A. (ed.) Britain at the Polls 2001 , London, Chatham House Publishers, pp. 164– 205. Benedetto G., Hix S. ( 2007) ‘ The Rejected, the Ejected, and the Dejected: Explaining Government Rebels in the 2001-2005 British House of Commons’, Comparative Political Studies , 40, 755– 781. Google Scholar CrossRef Search ADS   Berrington H., Hague R. ( 1998) ‘Europe, Thatcherism and Traditionalism: Opinion, Rebellion and the Maastricht Treaty in the Backbench Conservative Party, 1992–1994’, West European Politics , 21, 44– 71. Google Scholar CrossRef Search ADS   Cowley P., Garry J. ( 1998). ‘ The British Conservative Party and Europe: The Choosing of John Major.’, British Journal of Political Science , 28, 473– 499. Google Scholar CrossRef Search ADS   Cowley P., Norton P. ( 1999), ‘ Rebels and Rebellions: Conservative MPs in the 1992 Parliament’, The British Journal of Politics & International Relations , 1, 84– 105. Google Scholar CrossRef Search ADS   Cowley P., Stuart M. ( 2010) ‘ Where has all the trouble gone? British intra-party parliamentary divisions during the Lisbon ratification’, British Politics , 5, 133– 148. Google Scholar CrossRef Search ADS   Cowley P., Stuart M. ( 2012) ‘ The Cambusters: The Conservative European Union Referendum Rebellion of October 2011’, Political Quarterly , 83, 402– 406. Google Scholar CrossRef Search ADS   Crowson N. J. ( 2007) The Conservative Party and European Integration since 1945: At the Heart of Europe ? Abingdon, Routledge. Fontana C., Parsons C. ( 2015) ‘“One Woman's Prejudice”: Did Margaret Thatcher Cause Britain's Anti-Europeanism?’, JCMS: Journal of Common Market Studies , 53, 89– 105. Google Scholar CrossRef Search ADS   Forster A. ( 2002) Euroscepticism in Contemporary British Politics: Opposition to Europe in the British Conservative and Labour Parties Since 1945 , London, Routledge. Google Scholar CrossRef Search ADS   Garry J. ( 1995) ‘ The British Conservative Party: Divisions over European policy’, West European Politics , 18, 170– 189. Google Scholar CrossRef Search ADS   Hanretty C., Lauderdale B. E., Vivyan N. ( 2016) ‘Dyadic Representation in a Westminster System’, Legislative Studies Quarterly , Advanced Access published online: 01 October 2016, DOI: 10.1111/lsq.12148. Hayton R. ( 2010) ‘ Conservative Party Modernisation and David Cameron’s Politics of the Family’, Political Quarterly , 81, 492– 500. Google Scholar CrossRef Search ADS   Heppell T. ( 2002) ‘ The Ideological Composition of the Parliamentary Conservative Party 1992–97’, British Journal of Politics and International Relations , 4, 299– 324. Google Scholar CrossRef Search ADS   Heppell T. ( 2013), ‘ Cameron and Liberal Conservatism: Attitudes within the Parliamentary Conservative Party and Conservative Ministers’, The British Journal of Politics & International Relations , 15, 340– 361. Google Scholar CrossRef Search ADS   Heppell T., Hill M. ( 2008), ‘ The Conservative Party Leadership Election of 1997: An Analysis of the Voting Motivations of Conservative Parliamentarians’, British Politics , 3, 63– 91. Google Scholar CrossRef Search ADS   Heppell T., Hill M. ( 2010) ‘ The Voting Motivations of Conservative Parliamentarians in the Conservative Party Leadership Election of 2001’, Politics , 30, 36– 51. Google Scholar CrossRef Search ADS   Heppell T., Crines A., Jeffery D. ( 2017) ‘The United Kingdom Referendum on European Union Membership: The Voting of Conservative Parliamentarians’ JCMS: Journal of Common Market Studies . doi: 10.1111/jcms.12529. Hooghe L., Marks G., Wilson C. ( 2002) ‘ Does Left/Right Structure Party Positions on European Integration?’, Comparative Political Studies , 35, 965– 89. Google Scholar CrossRef Search ADS   Kam C. J. ( 2009) Party Discipline and Parliamentary Politics , Cambridge, Cambridge University Press. Google Scholar CrossRef Search ADS   Kaufman E. ( 2016) ‘Brexit: it’s NOT the Economy, Stupid’, 10 Gower Street: The Birkbeck Politics Department Blog, accessed at: http://10-gower-street.com/2016/07/08/brexit-its-not-the-economy-stupid/ on 17 September 2016. Martin S. ( 2016) ‘ Policy, Office and Votes: The Electoral Value of Ministerial Office’, British Journal of Political Science , 46, 281– 296. Google Scholar CrossRef Search ADS   Matland R. E., Studlar D. T. ( 2004) ‘ Determinants of Legislative Turnover: A Cross-National Analysis’, British Journal of Political Science , 34, 87– 108. Google Scholar CrossRef Search ADS   McAllister I., Studlar D. T. ( 2000) ‘ Conservative Euroscepticism and the Referendum Party in the 1997 British General Election’, Party Politics , 6, 359– 371. Google Scholar CrossRef Search ADS   Mitchell P. ( 2000) ‘ Voters and their Representatives: Electoral Institutions and Delegation in Parliamentary Democracies’, European Journal of Political Research , 37, 335– 351. Müller W. C., Strøm K. ( 1999) ‘Political Parties and Hard Choice’. In Müller W. C., Strøm K. (eds) Policy, Office, or Votes? How Political Parties in Western Europe Make Hard Choices , New York, Cambridge University Press, pp. 1– 35. Google Scholar CrossRef Search ADS   Nielson D. L. ( 2003), ‘ Supplying Trade Reform: Political Institutions and Liberalization in Middle-Income Presidential Democracies’ American Journal of Political Science , 47, 470– 491. Google Scholar CrossRef Search ADS   Norris P, Crewe I. ( 1994) ‘ Did the British Marginals Vanish? Proportionality and Exaggeration in the British Electoral System Revisited.’ Electoral Studies , 13, 201– 221. Google Scholar CrossRef Search ADS   Norton P. ( 1978) Conservative Dissidents: dissent within the Parliamentary Conservative Party, 1970-74 , London, Temple Smith. Pattie C. J., Fieldhouse E. A., Johnston R. J. ( 1994) ‘ The Price of Conscience: the Electoral Correlates and Consequences of Free Votes and Rebellions in the British House of Commons, 1987–1992’, British Journal of Political Science , 24, 359– 380. Google Scholar CrossRef Search ADS   Prosser C. ( 2016) ‘ Dimensionality, Ideology and Party Positions Towards European Integration’, West European Politics , 39, 731– 754. Google Scholar CrossRef Search ADS   Read M., Marsh D. ( 1997) ‘ The Family Law Bill: Conservative Party splits and Labour Party Cohesion’, Parliamentary Affairs , 50, 263– 279. Google Scholar CrossRef Search ADS   Szczerbiak A., Taggart P. ( 2008) ‘Theorising Party-Based Euroscepticism: Problems of Definition, Measurement and Causality’. In Taggart P., Szczerbiak A. (eds) Opposing Europe: The Comparative Party Politics of Euroscepticism Volume 2 , Oxford, Oxford University Press, pp. 238– 262. Smith K. B., Oxley D. R., Hibbing M. V., Alford J. R., Hibbing J. R. ( 2011) ‘ Linking Genetics and Political Attitudes: Reconceptualizing Political Ideology’, Political Psychology , 32, 369– 397. Google Scholar CrossRef Search ADS   Söderlund P. ( 2016). ‘ Candidate-Centred Electoral Systems and Change in Incumbent Vote Share: A Cross-national and Longitudinal Analysis’, European Journal of Political Research , 55, 321– 339. Google Scholar CrossRef Search ADS   Sowemimo J. ( 1996) ‘ The Conservative Party and European Integration 1988-95’, Party Politics , 2, 77– 97. Google Scholar CrossRef Search ADS   Tillman E. R. ( 2013) ‘ Authoritarianism and Citizen Attitudes Towards European Integration’, European Union Politics , 14, 566– 589. Google Scholar CrossRef Search ADS   Turner J. ( 2000) The Tories and Europe , Manchester, Manchester University Press. Tzeglove E. ( 2014) ‘ Cross-Cutting Issues, Intraparty Dissent and Party Strategy: The Issue of European Integration in the House of Commons’, European Union Politics , 15, 3– 23. Google Scholar CrossRef Search ADS   Webb P. D., Bale T. ( 2014) ‘ Why Do Tories Defect to UKIP? Conservative Party Members and the Temptations of the Populist Radical Right’, Political Studies , 62, 961– 970. Google Scholar CrossRef Search ADS   Webb P. D., Childs S. ( 2011) ‘ Wets and Dries Resurgent? Intra-Party Alignments Among Contemporary Conservative Party Members’, Parliamentary Affairs , 64, 383– 402. Google Scholar CrossRef Search ADS   Appendix 1 Table A1 Logit regression coefficients of the probability of MPs backing leave   Model 1 (Main variables)  Model 3 (includes demographic controls)  Model 4 (controls for constituency majority)  Social conservatism  0.254 (0.060)***  0.256 (0.062)***  0.261 (0.063)***  Office status reference category: experienced backbenchers        Newly elected MPs  −0.762 (0.309)**  −0.943 (0.400)**  −0.901 (0.415)**  Frontbenchers  −1.582 (0.298)***  −1.771 (0.326)***  −1.766 (0.325)***  Constituency Euroscepticisma  0.374 (0.164)**  0.303 (0.180)*  0.404 (0.201)**  Age    −0.0196 (0.016)  −0.230 (0.163)  Gender (Female = 1)    −0.195 (0.357)  −0.218 (0.361)  Private school    −0.247 (0.262)  −0.246 (0.264)  Marginal constituency      1.798 (1.955)  Marginal × Constituency Euroscepticism      0.385 (0.363)  Constant  −2.921 (0.931)***  −1.360 (1.304)  −1.664 (1.433)  Observations  324  309  309  Count R2  0.704  0.709  0.709  LR χ2b  66.17  68.28  69.91  p > χ2  >0.001  >0.001  >0.001    Model 1 (Main variables)  Model 3 (includes demographic controls)  Model 4 (controls for constituency majority)  Social conservatism  0.254 (0.060)***  0.256 (0.062)***  0.261 (0.063)***  Office status reference category: experienced backbenchers        Newly elected MPs  −0.762 (0.309)**  −0.943 (0.400)**  −0.901 (0.415)**  Frontbenchers  −1.582 (0.298)***  −1.771 (0.326)***  −1.766 (0.325)***  Constituency Euroscepticisma  0.374 (0.164)**  0.303 (0.180)*  0.404 (0.201)**  Age    −0.0196 (0.016)  −0.230 (0.163)  Gender (Female = 1)    −0.195 (0.357)  −0.218 (0.361)  Private school    −0.247 (0.262)  −0.246 (0.264)  Marginal constituency      1.798 (1.955)  Marginal × Constituency Euroscepticism      0.385 (0.363)  Constant  −2.921 (0.931)***  −1.360 (1.304)  −1.664 (1.433)  Observations  324  309  309  Count R2  0.704  0.709  0.709  LR χ2b  66.17  68.28  69.91  p > χ2  >0.001  >0.001  >0.001  Note: The numbers in the table are logit regression coefficients with robust standard errors in parentheses. The Count R2 refers to the proportion of correctly predicted outcomes from the model. *** p < 0.01, ** p < 0.05, * p < 0.1. aOne unit of constituency Euroscepticism = 10% disapproval of the EU, e.g. a constituency with 45.6% EU disapproval is coded as 4.56. bMeasure based on pseudolikelihood as robust standard errors have been used. Table A2 Marginal effects on the probability of endorsing leave (Probit version)   Model 1 (Main variables)  Model 3 (includes demographic controls)  Model 4 (controls for constituency majority)  Social conservatism  0.049 (0.010)***  0.049 (0.011)***  0.049 (0.011)***  Office status reference category: experienced backbenchers  Newly elected MPs  −0.174 (0.070)**  −0.214 (0.087)**  −0.205 (0.090)**  Frontbenchers  −0.332 (0.058)***  −0.367 (0.061)***  −0.365 (0.061)***  Constituency Euroscepticisma  0.075 (0.032)**  0.060 (0.033)*  0.081 (0.038)**  Age    −0.004 (0.003)  −0.004 (0.003)  Gender (Female = 1)    −0.029 (0.069)  −0.034 (0.069)  Private school    −0.051 (0.050)  −0.051 (0.050)  Marginal constituency      0.382 (0.387)  Marginal × Constituency Euroscepticism      −0.008 (0.007)  Observations  324  309  309  Count R2  0.704  0.722  0.709  LR χ2b  65.78  67.78  69.56  p > χ2  >0.001  >0.001  >0.001    Model 1 (Main variables)  Model 3 (includes demographic controls)  Model 4 (controls for constituency majority)  Social conservatism  0.049 (0.010)***  0.049 (0.011)***  0.049 (0.011)***  Office status reference category: experienced backbenchers  Newly elected MPs  −0.174 (0.070)**  −0.214 (0.087)**  −0.205 (0.090)**  Frontbenchers  −0.332 (0.058)***  −0.367 (0.061)***  −0.365 (0.061)***  Constituency Euroscepticisma  0.075 (0.032)**  0.060 (0.033)*  0.081 (0.038)**  Age    −0.004 (0.003)  −0.004 (0.003)  Gender (Female = 1)    −0.029 (0.069)  −0.034 (0.069)  Private school    −0.051 (0.050)  −0.051 (0.050)  Marginal constituency      0.382 (0.387)  Marginal × Constituency Euroscepticism      −0.008 (0.007)  Observations  324  309  309  Count R2  0.704  0.722  0.709  LR χ2b  65.78  67.78  69.56  p > χ2  >0.001  >0.001  >0.001  Note: The numbers in the table are average marginal effects with robust standard errors in parentheses. The Count R2 refers to the proportion of correctly predicted outcomes from the model. *** p < 0.01, ** p < 0.05, * p < 0.1. aOne unit of constituency Euroscepticism = 10% disapproval of the EU, e.g. a constituency with 45.6% EU disapproval is coded as 4.56. bMeasure based on pseudolikelihood as robust standard errors have been used. Table A3 Probit regression coefficients of the probability of MPs backing leave   Model 1 (Main variables)  Model 3 (includes demographic controls)  Model 4 (controls for constituency majority)  Social conservatism  0.150 (0.035)***  0.151 (0.036)***  0.153 (0.037)***  Office status reference category: experienced backbenchers  Newly elected MPs  −0.477 (0.191)**  −0.590 (0.245)**  −0.567 (0.253)**  Frontbenchers  −0.948 (0.175)***  −1.059 (0.191)***  −1.059 (0.190)***  Constituency Euroscepticisma  0.228 (0.100)**  0.185 (0.102)*  0.253 (0.122)**  Age    −0.011 (0.009)  −0.014 (0.010)  Gender (Female = 1)    −0.091 (0.214)  −0.106 (0.215)  Private school    −0.159 (0.157)  −0.159 (0.157)  Marginal constituency      1.187 (1.205)  Marginal × Constituency Euroscepticism      −0.025 (0.022)  Constant  −1.753 (0.565)***  −0.811 (0.790)  −1.020 (0.865)  Observations  324  309  309  Count R2  0.704  0.722  0.709  LR χ2b  65.78  67.78  69.56  p > χ2  >0.001  >0.001  >0.001    Model 1 (Main variables)  Model 3 (includes demographic controls)  Model 4 (controls for constituency majority)  Social conservatism  0.150 (0.035)***  0.151 (0.036)***  0.153 (0.037)***  Office status reference category: experienced backbenchers  Newly elected MPs  −0.477 (0.191)**  −0.590 (0.245)**  −0.567 (0.253)**  Frontbenchers  −0.948 (0.175)***  −1.059 (0.191)***  −1.059 (0.190)***  Constituency Euroscepticisma  0.228 (0.100)**  0.185 (0.102)*  0.253 (0.122)**  Age    −0.011 (0.009)  −0.014 (0.010)  Gender (Female = 1)    −0.091 (0.214)  −0.106 (0.215)  Private school    −0.159 (0.157)  −0.159 (0.157)  Marginal constituency      1.187 (1.205)  Marginal × Constituency Euroscepticism      −0.025 (0.022)  Constant  −1.753 (0.565)***  −0.811 (0.790)  −1.020 (0.865)  Observations  324  309  309  Count R2  0.704  0.722  0.709  LR χ2b  65.78  67.78  69.56  p > χ2  >0.001  >0.001  >0.001  Note: The numbers in the table are probit regression coefficients with robust standard errors in parentheses. The Count R2 refers to the proportion of correctly predicted outcomes from the model. *** p < 0.01, ** p < 0.05, * p < 0.1. aOne unit of constituency Euroscepticism = 10% disapproval of the EU, e.g. a constituency with 45.6% EU disapproval is coded as 4.56. bMeasure based on pseudolikelihood as robust standard errors have been used. Table A4 Linear probability model (OLS) of MPs endorsing leave   Model 1 (Main variables)  Model 3 (includes demographic controls)  Model 4 (controls for constituency majority)  Social conservatism  0.051 (0.011)***  0.050 (0.012)***  0.051 (0.012)***  Office Status Reference Category: experienced backbenchers  Newly elected MPs  −0.180**  −0.215**  −0.205**    (0.071)  (0.088)  (0.091)  Frontbenchers  −0.331***  −0.365***  −0.361***    (0.058)  (0.062)  (0.062)  Constituency Euroscepticism‡  0.072 (0.031)**  0.057 (0.032)*  0.073 (0.038)**  Age    −0.004 (0.003)  −0.004 (0.003)  Gender (Female = 1)    −0.028 (0.068)  −0.029 (0.069)  Private School    −0.053 (0.052)  −0.053 (0.052)  Marginal constituency      0.275 (0.373)  Marginal × Constituency Euroscepticism      −0.006 (0.007)  Constant  −0.58 (0.174)  0.251 (0.264)  0.210 (0.282)  Observations  324  309  309  R2  0.189  0.203  0.206    Model 1 (Main variables)  Model 3 (includes demographic controls)  Model 4 (controls for constituency majority)  Social conservatism  0.051 (0.011)***  0.050 (0.012)***  0.051 (0.012)***  Office Status Reference Category: experienced backbenchers  Newly elected MPs  −0.180**  −0.215**  −0.205**    (0.071)  (0.088)  (0.091)  Frontbenchers  −0.331***  −0.365***  −0.361***    (0.058)  (0.062)  (0.062)  Constituency Euroscepticism‡  0.072 (0.031)**  0.057 (0.032)*  0.073 (0.038)**  Age    −0.004 (0.003)  −0.004 (0.003)  Gender (Female = 1)    −0.028 (0.068)  −0.029 (0.069)  Private School    −0.053 (0.052)  −0.053 (0.052)  Marginal constituency      0.275 (0.373)  Marginal × Constituency Euroscepticism      −0.006 (0.007)  Constant  −0.58 (0.174)  0.251 (0.264)  0.210 (0.282)  Observations  324  309  309  R2  0.189  0.203  0.206  Note: Robust standard errors in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1. aOne unit of constituency Euroscepticism = 10% disapproval of the EU, e.g. a constituency with 45.6% EU disapproval is coded as 4.56. © The Author 2017. Published by Oxford University Press on behalf of the Hansard Society; all rights reserved. For Permissions, please e-mail: journals.permissions@oup.com http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Parliamentary Affairs Oxford University Press

Policy, Office and Votes: Conservative MPs and the Brexit Referendum

Loading next page...
 
/lp/ou_press/policy-office-and-votes-conservative-mps-and-the-brexit-referendum-EDedkAYkf2
Publisher
Oxford University Press
Copyright
© The Author 2017. Published by Oxford University Press on behalf of the Hansard Society; all rights reserved. For Permissions, please e-mail: journals.permissions@oup.com
ISSN
0031-2290
eISSN
1460-2482
D.O.I.
10.1093/pa/gsx010
Publisher site
See Article on Publisher Site

Abstract

Abstract The division within the Parliamentary Conservative Party (PCP) over Britain’s membership of the EU has been one of the most significant intra-party divisions in European political history. The 2016 Brexit referendum campaign offered a unique opportunity to consider legislative motivations as almost every MP declared a preference and frontbench MPs were free to back either side. This article uses logistic regression analysis in order to consider MPs’ motivations in terms of Müller and Strøm’s policy, office and votes trichotomy. It is argued that all three motivations affected MPs decision making on the EU referendum. However, vote-seeking motivations were less influential than either policy or office-seeking. Britain’s EU membership has been one of the most important issues in post-war Westminster politics. While divisions over Europe have at times affected both the major parties (Forster, 2002; Tzeglove, 2014), the European issue has had the most significant impact on the Conservative Party. The European issue has played a crucial role in at least three party leadership elections (Cowley and Garry, 1998; Heppell and Hill, 2008, 2010) and has caused many Parliamentary rebellions—particularly in the 1992 and 2010 Parliaments (Cowley and Norton, 1999; Cowley and Stuart, 2012; Hanretty et al., 2016). Since returning to power in 2010 a hard Eurosceptic element has emerged within the Parliamentary Conservative Party (PCP) (Heppell, 2013; Cowley and Stuart, 2012; cf. Szczerbiak and Taggart, 2008 for definitions of hard and soft-Euroscepticism). The emergence of this hard Eurosceptic group created a strong divide in the party between those who wished to leave the EU and those who, though in many ways Eurosceptic, nevertheless wanted to remain. The referendum campaign offered a unique opportunity to analyse attitudes within the PCP as the suspension of collective decision making meant that both front and backbench MPs were free to endorse either side, although the leadership had made it clear that it would rather MPs embraced remain1. This paper considers MPs’ referendum positions using multivariate logistic regression analysis. The data used is a bespoke dataset created by the author. Legislative motivations are analysed in terms of Müller and Strøm’s (1999) policy, office and votes trichotomy. It is argued that all three of these motivations affected MPs’ decision making on the EU referendum, although vote-seeking was of lesser importance than either policy or office-seeking. In terms of policy-seeking motivations, endorsing the leave campaign was strongly associated with social conservatism. The most likely reason for this association is that Euroscepticism within the Conservative party is influenced by traditional/authoritarian/nationalist values, particularly attitudes towards sovereignty and nationalism. The effect of office-seeking can be seen from the fact that an MP’s career status was an important predictor of preferences. The overwhelming majority of frontbench MPs who had the most incentives to be loyal to the leadership chose to follow the leadership in backing remain. The majority of experienced backbenchers, who had already resigned, been dismissed or overlooked for government office, opposed the leadership and backed leave. These MPs had the least incentives for loyalty to the leadership and indeed stood to potentially gain from the prospect of a post leave-vote Prime Ministerial resignation. MPs representing Eurosceptic constituencies were also found to be more likely to endorse leave, suggesting that vote-seeking motivations also affected MPs’ positions on the referendum. However, constituency opinion had a weaker effect on MPs’ referendum preferences than either personal ideology or office status. The article is structured as follows. Section 1 reviews the relevant literature concerning Euroscepticism and the PCP. Section 2 outlines the theory and hypotheses. Section 3 details the data and methods used. Section 4 presents the results which are discussed in Section 5 which concludes the article. 1. Euroscepticism and the PCP The PCP has always been, to a greater or less extent, divided on the issue of European integration since Britain first applied to join the EEC in 1961. The broad trend in this time is that a party largely supportive of the European project has become progressively more sceptical. In early 1960s, internal party sources estimated that only around 10% of Conservative MPs opposed entry into the Common Market (Crowson, 2007). In the 1970 Parliament, which passed the European Communities Act, it was estimated that nearly a third of Conservative MPs were either opposed to or doubtful about membership of the EEC (Norton, 1978). The majority of Conservative MPs were, however, supportive of European integration during 1960s and 1970s (Crowson, 2007). The late 1980s represented a turning point in the PCP as the Single European Act, Delors’ plan for a ‘Social Europe’, and Thatcher’s increased opposition to European integration produced higher levels of Euroscepticism within the PCP (Heppell et al., 2017; Fontana and Parsons, 2015). Garry’s (1995) survey research of the 1987 PCP suggested a party almost evenly split between pro and anti-integration MPs. Ideological mapping of the PCP by Heppell (2002, 2013) suggests that Eurosceptics have been a majority since 1992 and in opposition after 1997 the Europhiles became only a small minority of the party as the party was largely united in a Eurosceptic position. This Eurosceptic position was however, a soft Eurosceptic position which opposed further integration with Europe and advocated the repatriation of powers rather than withdrawal from the EU (Heppell et al., 2017). Euro-rejectionist sentiment during this period was rather muted as prior to the 2010 election only six Conservative MPs had publicly declared support for leaving the EU (Cowley and Stuart, 2010). A hard Eurosceptic element, demanding either withdrawal or a fundamental change in the UK relationship with the EU, did become more visible after the Conservatives returned to power in 2010. Heppell (2013) found that 26% of the 2010 PCPs were hard Eurosceptics. This created a divide in the Conservatives between a majority of the PCP supporting a soft-Eurosceptic reformist position and a notable minority advocating a hard-Eurosceptic rejectionist position. The number of hard Eurosceptics grew in the 2015 PCP as over 40% of Conservative MPs backed leave (Heppell et al., 2017). To summarize, the research literature suggests that the PCP has become progressively more Eurosceptic since the 1960s and that the 2015 PCP is the most Eurosceptic PCP to date. There have been only very limited attempts to understand the drivers of Euroscepticism with the PCP. Garry (1995) in his survey of the 1987 Parliament found that Eurosceptics tended to be both economically right-wing and socially conservative. The weakness of this study was that relying on survey data meant that the analysis included less than half of the PCP (45.1%). Heppell (2002) similarly found, for the 1992 Parliament, correlations between Euroscepticism and both economically dry and socially conservative positions. For the same Parliament, Berrington and Hague (1998) analysed attitudes towards the Maastricht Treaty using principal component analysis of Early Day Motions (EDMs). Their results showed numerous statistical differences between those for and against the Treaty, for example anti-Maastricht (i.e. Eurosceptic) MPs were more likely to be pro-death penalty and less likely to have been educated at a top public school. The use of EDM data however meant that Berrington and Hague’s sample was limited to only 155 MPs (46.1% of the total PCP) and included no frontbench MPs. For the 2015 Parliament, Heppell et al. (2017) found several statistical differences between remain and leave MPs. For instance, leave MPs were more likely to have served in military and to have opposed gay marriage, while remain MPs were more likely to have been ministers and attended an Oxbridge university. Heppell et al. (2017) used the same dependent variable and a similar methodology to this article; however, they do not directly address the issue of how MPs’ behaviour on Brexit may have been influenced by competing motivations. This article seeks to build upon the previous literature by creating a model to assess MPs’ motivations in terms of Müller and Strøm’s (1999) policy, office and votes trichotomy. The decision to formally suspend the Whip for the referendum campaigns presents a novel opportunity to be able to consider legislative motivations on both the back and frontbenches. By using this opportunity the article aims to contribute to a greater understanding of what motivates Euroscepticism within the PCP. 2. Theory and hypotheses The literatures on both the Conservative Party and legislative behaviour have informed the theory and hypotheses of this paper. Kam (2009) argues that there is consensus in the legislative behaviour literature around Müller and Strøm’s (1999) assertion that politicians seek a combination of three aims: policy, office and votes. In the long term, these three goals are not necessarily mutually exclusive and indeed may be mutually reinforcing. Re-election and promotion to ministerial office put an MP in a better position to influence policy. In Westminster systems particularly, government control of agenda setting means that MPs are much more likely to influence policy as ministers than as backbenchers (Benedetto and Hix, 2007). Moreover, Martin (2016), using data from Irish elections, argues that ministers can use their increased influence over policy to gain an electoral advantage over their co-partisans. He argues that ministers use ‘executive particularism’ (pork-barrelling) to specifically target government spending at their own constituencies. However, on individual policies there may be occasions when the three aims conflict. If on a particular policy the views of either the MP herself, her constituents, or the party leadership are not aligned the MP will have to choose to prioritise one aim over another. For instance if the party leadership and the MP hold different policy positions, the MP would either have to sacrifice her policy preferences on this particular issue in the hope of being able to influence other policy decisions in the future or sacrifice her chances of promotion to stay true to her policy preference. The EU referendum provides a good example of where conflict between these aims might arise as on the one hand there was a party leadership backing remain and on the other hand deep-seated Euroscepticism on the Conservative backbenches and in the grassroots party. Below, I consider each of the three motivations: policy, office and votes in order to generate the hypotheses for this study. 2.1 Policy-seeking When reviewing the recent history of the Conservative party and Euroscepticism, there is a strong case that politicians have been motivated more by their own policy agendas rather than electoral demand. The evidence for this is that since at least the 1990s the European issue has been of crucial importance for Conservative Party politics but of more limited salience for the wider electorate (Bale, 2006: Bartle, 2002). If policy-seeking was the main objective of Tory MPs concerning Europe, one would expect to find that ideology was highly important in predicting MPs’ positions. Apart from Europe, the other main policy divide in the PCP in recent times has been the social/moral policy divide which separates social conservatives from social liberals2 (Hayton, 2010). Social conservatives hold traditional/authoritarian policy positions, taking a more restrictive stance on issues such as abortion and gay rights, while advocating tougher punishment for criminals. Social liberals by contrast have a more permissive approach to issues such abortion and gay rights and a less punitive approach to crime and punishment. Previous empirical studies on Conservative MPs and grassroots members have shown strong statistical associations between Euroscepticism and social conservativism (Garry, 1995; Berrington and Hague, 1998; Heppell, 2002, 2013; Webb and Bale, 2014). Similar links between social conservatism and Euroscepticism have been found more broadly in European politics. For example Tillman (2013) finds, based on European Values Survey data, that authoritarian preferences over child rearing are linked to Euroscepticism and Prosser (2016) argues that party positions on European integration are primarily affected by the social liberal-conservative axis of politics. Specifically on the Brexit referendum, Kaufman (2016) finds strong statistical links between supporting harsher treatment of criminals and the probability of voting leave. The British Electoral Study (BES) data he uses shows that 71% of those who supported the death penalty indicated they would vote leave. This was a stronger predictor of voting intentions than education, income, social class or party support. The reason why statistical associations between social conservatism and Euroscepticism have been found cannot be established with any certainty. One plausible explanation for this is that positions on Europe could be related to the GAL (green/alternative/libertarian) versus TAN (tradition/authoritarian/nationalist) dimension of political contestation (Hooghe et al., 2002). Parties or politicians with a TAN orientation ‘combine support for traditional values, opposition to immigration and defence of the national community’ (Hooghe et al., 2002, p. 976). Hooghe et al. (2002) argue that Euroscepticism is often advocated by political parties with a TAN orientation as European integration is seen as a threat to the national sovereignty and community. Berrington and Hague (1998) allude to a connection between TAN policy positions and Euroscepticism amongst Conservative MPs in their study of Conservative MPs and the Maastricht Treaty. They find that pro-death penalty MPs were more likely to have opposed the Treaty. They connect this to the political psychology literature, arguing that attitudes towards harsher punishment of criminals were related to attitudes towards ethnocentrism, imperialism and intolerance (ibid). More recent contributions to the political psychology literature have also established a link between social conservatism and nationalism (Smith et al., 2011). Therefore, there could be psychological reasons underlying the relationship between social conservatism, nationalism and Euroscepticism. For Conservative MPs, a link between social conservatism and Euroscepticism could signify the importance of TAN values, particularly attitudes towards nationhood and sovereignty, as many in the literature have argued that nationalism is one of the main causes of Conservative Party Euroscepticism (Berrington and Hague, 1998; Turner, 2000; Crowson, 2007; Webb and Childs, 2011). The literature suggests, then, a strong relationship between social conservatism and Euroscepticism. This generates hypothesis 1 below: Hypothesis 1: Socially conservative MPs were more likely to endorse leave 2.2 Office-seeking motivations The Westminster system in which the party leadership controls ministerial appointments provides strong incentives for MPs seeking a ministerial career to obey the party whip (Benedetto and Hix, 2007; Kam, 2009). The literature has established that ministers and those MPs who still have a rational hope of climbing the ministerial ladder are less likely to rebel (Benedetto and Hix, 2007; Kam, 2009). This literature on rebellion is relevant because even though the EU referendum was formally unwhipped, and indeed six cabinet ministers joined leave, the official government position was for remain and the Prime Minister was seeking to persuade MPs to back remain. This means that although backing leave did not result in an automatic dismissal, as rebelling on a whipped vote would do, the Cameron leadership was more likely to reward those who backed remain. Possible evidence that remain MPs would be rewarded can be seen from the mini reshuffle following Iain Duncan-Smith’s resignation when the Prime Minister promoted only MPs who backed remain. This means that MPs who were climbing the ministerial ladder or who had ambitions of climbing the ladder could be expected to be more likely to back remain in order to satisfy Cameron. In contrast, MPs who had left office, through either dismissal or resignation, or who had been overlooked for a government post are predicted to have been less likely to endorse remain as they had less reason to believe that Cameron would reward them after the referendum. Benedetto and Hix (2007) found that ex-ministers and MPs overlooked for ministerial office are the most likely to rebel as they cannot be controlled by the promise of ministerial office. Kam (2009) in his comparative study of Westminster systems similarly argued that MPs who could no longer attain office were more likely to rebel. By the same logic MPs who have opted not to pursue a ministerial career, in order to focus on either constituency and/or select committee work, will also have been less inclined to be loyal to the leadership. Office-seeking calculations of MPs could also have been affected by the fact that a leave victory in the referendum was almost certain to result in a Prime Ministerial resignation. Backbenchers out of favour with the leadership, i.e. those sacked or overlooked for office, had a potential incentive to endorse leave as a new leader may give them a better chance of gaining a government position. Conversely, for frontbenchers, all of whom were appointed by Cameron, a change of leadership would make their positions less secure. Indeed, this is what happened after the leave vote: the new Prime Minister, Theresa May, brought in MPs who had been out of office for a long time such as Liam Fox and David Davis and dismissed long-serving ministers from the Cameron administration such as George Osborne and Nicky Morgan. From this discussion of office-seeking intentions, it is predicted that there will be three distinct groups within the PCP which vary in their likelihood of endorsing leave. The group most likely to back remain are frontbenchers. Frontbenchers had incentives to endorse remain as this would increase their chances of promotion under Cameron whilst a potential change of leader would threaten their position in the government. The group most likely to endorse leave are backbenchers elected before the 2015 election. These MPs have all been in Parliament long enough to have potentially been offered a frontbench position, as can be seen by the fact that a majority of Tory MPs first elected at the 2010 election, as well as Robert Jenrick who was elected in a 2014 by-election, were serving in government positions at the time of the referendum. This group of experienced backbenchers includes ex-ministers, those overlooked for government office and MPs who have chosen not to pursue a ministerial career. They have little incentive to be loyal to Cameron and may benefit from a change of leader in the event of a leave vote. Concerning MPs newly elected in 2015, this group is expected to have been more likely to endorse remain than the more experienced backbenchers as new MPs would have had a realistic potential of being offered a government post by Cameron had the remain side won the referendum campaign. However, new MPs are predicted to have been more likely to endorse leave than frontbenchers. The reason for this is that some newly elected MPs may have had no office-seeking ambitions and newly elected MPs with office-seeking motivations would be less likely to have been detrimentally affected by a change of leader than frontbench MPs. This analysis leads to hypotheses 2a–c below: Hypothesis 2: Office status was a strong predictor of an MP’s referendum position. 2a. Frontbench MPs were the most likely MPs to endorseremain. 2b. Backbench MPs elected before 2015 were the most likely MPs to endorse leave. 2c. Newly elected MPs were more likely to endorse leave than frontbench MPs but less likely to endorse leave than long serving backbench MPs. 2.3 Vote-seeking It is reasonable to suggest that vote-seeking and re-election is a concern for many MPs as most sitting MPs put themselves forward as candidates at the next election (Matland and Studlar, 2004). There is increasing evidence in the literature that British MPs’ roll-call records can affect their electoral performance (Pattie et al., 1994; Kam, 2009). Despite this the only study to specifically consider Conservative Eurosceptics concluded that the Europe issue did not affect their vote share (McAllister and Studlar, 2000). McAllister and Studlar found that declared Eurosceptics had a vote-share 2.3% higher than other Conservative MPs at the 1997 election but that almost all of this difference could be accounted for by incumbency and turnout differences across constituencies. However, of more importance than whether MP’s voting records affect their electoral vote share, is whether MPs think (rightly or wrongly) that voters react to their personal policy positions. Vote-seeking MPs would rationally choose their referendum position based on public opinion in their constituencies if they believed that this would affect their personal vote. A number of studies have considered whether MPs vote based on public opinion in their constituencies. For instance, Baughman (2004) found that MPs representing constituencies with a high number of Catholics were more likely to take conservative positions on abortion votes. Most significantly for this study, Hanretty et al. (2016) found that MPs’ positions in the 2010 Parliament were responsive to constituency opinion on three policy issues, one of which was Europe. They demonstrated that constituency disapproval of the EU was linked to a higher probability of a Conservative MP adopting Eurosceptic positions on both whipped and free votes. If MPs were vote-seeking, then, we should expect to see similar results to Hanretty et al. (2016) for the referendum, with MPs representing Eurosceptic constituencies being more likely to support leave. Hypothesis 3: There was a positive relationship between constituency Euroscepticism and the probability of MPs’ endorsing leave. The MPs who are most likely be most concerned with vote-seeking are those representing marginal and semi-marginal seats. As marginal seats can change hands on the basis of relatively small swings away from the incumbent, sitting MPs in marginal constituencies might be very attentive to opinion in their constituency. MPs representing safer seats have less reason to fear electoral defeat and may be less attentive to constituency opinion. This discussion leads to Hypothesis 4: Hypothesis 4: The relationship between constituency Euroscepticism and referendum position was stronger for MPs who represented marginal constituencies. 3. Data and methods This article uses a dataset of Conservative MPs elected at the 2015 general election. The dataset was created by the author. Logistic regression analysis is used in order to test the hypotheses identified in Section 2. As this study uses observational data, it makes no claim to have identified causal processes and the author accepts alternative interpretations of the findings are possible. However, by thoroughly and accurately presenting the statistical associations between the variables the author believes that this study can contribute to the literatures on legislative behaviour in British politics and Conservative Party Euroscepticism. Sections 3.1–3.6 below detail the variables used in the dataset; the statistical models used are explained in Section 3.7. 3.1 The dependent variable—MPs’ referendum endorsements The dependent variable is a binary variable which measures MPs’ publicly declared support in the referendum campaign. MPs who publicly declared support for leave were coded as 1 and MPs who publicly supported remain were coded 0. MPs who did not declare a position were treated as missing data. As only 6 (2%) MPs did not declare a preference, few MPs were excluded from the analysis. Table 1 displays the summary statistics of MPs’ referendum positions. It demonstrates that the majority of MPs did side with the Prime Minister and support remain. However, a substantial proportion (41%) of Conservative MPs backed leave. Table 1 Referendum positions of conservative MPs Stance  n (%)  Remain  189 (57)  Leave  135 (41)  Did not declare  6 (2)  Stance  n (%)  Remain  189 (57)  Leave  135 (41)  Did not declare  6 (2)  Sources: MPs’ personal websites, BBC News (http://www.bbc.co.uk/news/uk-politics-eu-referendum-35616946) and TheyWorkForYou.com. 3.2 Social conservatism A ‘social conservatism’ variable was developed in order to test Hypothesis 1. To generate the social conservatism variable a scale was created using a similar approach to that taken by Read and Marsh (1997), and Heppell (2002, 2013; Heppell and Hill, 2008, 2010). An ideological scale was created based on MPs’ positions on five issues: Gay Marriage Abortion Death Penalty Three-parent embryos Greater use of imprisonment as opposed to community sentences For each issue an MP was given 2 for a socially conservative stance, 0 for a socially liberal stance or 1 where either no data could be found or the MP was inconsistent on the issue. After the sum of the five issues was calculated, each MP had a score on a scale 0–10, with higher numbers denoting greater social conservativism. Data were gathered for each MP using a combination of their House of Commons voting record, record of signing relevant EDMs, membership of All Party Parliamentary Groups and public comment made by the MP3. The resulting summary statistics of social conservatism show that Conservative MPs were spread across the range of the scale, with three MPs being socially liberal on all five issues and seven MPs being socially conservative on every issue. Figure 1 above contains a histogram of the distribution of Conservative MPs on the social conservatism scale. The variable is skewed slightly to the conservative side of the spectrum with the mean score being 5.11. Figure 1 View largeDownload slide Histogram of MPs on the social conservatism scale. The social conservatism variable ranges 0 to 10 with higher values denoting stronger social conservatism. Social conservatism scores for each MP based on their positions on five issues: abortion, gay marriage, death penalty, use of three parent embryos and greater use of imprisonment as opposed to community sentences. For each issue, an MP was given 2 for a socially conservative stance, 0 for a socially liberal stance or 1 where either no data could be found or the MP was inconsistent on the issue. Figure 1 View largeDownload slide Histogram of MPs on the social conservatism scale. The social conservatism variable ranges 0 to 10 with higher values denoting stronger social conservatism. Social conservatism scores for each MP based on their positions on five issues: abortion, gay marriage, death penalty, use of three parent embryos and greater use of imprisonment as opposed to community sentences. For each issue, an MP was given 2 for a socially conservative stance, 0 for a socially liberal stance or 1 where either no data could be found or the MP was inconsistent on the issue. 3.3 Office-status To test Hypothesis 2, that office-status would affect referendum position, a categorical variable ‘office status’ was created with three categories: Experienced backbenchers (n = 125): This is the baseline category. MPs are included in this category if they were elected before the 2015 election. This group of MPs is on the backbenches through dismissal, resignation or being overlooked for government positions or choosing not to pursue a ministerial career. New MPs (n = 72): These MPs were elected for the first time in 2015. At the time of the referendum they were too inexperienced to have been offered a government post. Frontbenchers (n = 127): MPs were included in this category if they were a minister, whip or Parliamentary Private Secretary on the day of the referendum. 3.4 Constituency Euroscepticism A measure of constituency Euroscepticism was used in the model to examine whether MPs’ vote-seeking motivations affected their referendum positions. To measure constituency Euroscepticism, data from Hanretty et al. (2016) was used. They estimated Euroscepticism in each constituency using data from the BES 2010 Combined Internet Panel Survey which asked respondents if they disapproved of the UK’s EU membership. Hanretty et al. (2016) used multilevel regression and post-stratification analysis of the survey responses to obtain an EU disapproval percentage for each constituency. The mean value of this variable for Conservative constituencies was 53.6 (indicating that 53.6% of constituency residents disapproved of the EU) and the median was 54.5. Scores on the constituency Euroscepticism scale ranged from a low of 24.1 for Jane Ellison who represented Battersea in inner London to a high of 72.0 for Rebecca Harris who represented Castle Point in Essex. As the constituency Euroscepticism variable took account of numerous constituency demographic factors, such as percentage non-white, median education level and region, no other constituency demographic variables were included in the dataset for this study. 3.5 Seat marginality In order to test Hypothesis 4, that MPs in marginal seats were more responsive to constituency Euroscepticism, two variables were added to the dataset, one dummy variable measuring whether a seat was marginal or safe and an interaction variable of marginal seats and constituency Euroscepticism. There are no firm rules about the distinction between marginal, semi-marginal and safe-seats. Norris and Crewe (1994) defined a seat as either marginal or semi-marginal if the winning party gained a majority of 14.9% or lower over the second placed party. On this basis, this article uses the 14.9% winning margin as the threshold for coding seats as 1 in the marginal seats dummy variable. This results in 95 MPs (29%) being coded as 1, indicating that they represented marginal or semi-marginal seats, and 229 MPs (71%) being coded as 0, indicating that they represented comparatively safe seats. The interaction variable, which seeks to test whether MPs from marginal seats were more responsive to constituency Euroscepticism, consists of the marginal seat dummy variable multiplied by constituency Euroscepticism. As establishing a threshold between safe and marginal seats is to some extent arbitrary several different versions of the test were performed using different thresholds to check the validity of any findings.4 3.6 Control variables The dataset also includes control variables which could also have potentially predicted which side MPs endorsed in the referendum campaign. These variables are considered below: Age: Polling suggested that age was an important predictor of voting amongst the general public in the Brexit referendum with older generations voting more for leave and younger generations voting more for remain. It may have been the case that opinion trends within the PCP were similar to those in the wider public. Gender: A dummy variable was also included denoting an MP’s gender (female = 1). It might be case that female MPs are less likely to be pro-leave as many of the most prominent figures in Conservative Euroscepticism such as John Redwood, Bill Cash and Andrew Rosindell are male. Private school: This dummy variable denotes whether an MP attended a fee paying secondary school. A connection between MPs’ schooling and Europe was found by Berrington and Hague (1998) who found that MPs from the top public schools were less likely to take Eurosceptic positions. 3.7 The statistical models As the dependent variable is dichotomous, logistic regression models are used to analyse the data. The basic structure of the statistical models is:   Pr(leave=1)=logit−1(α+β1X1+β2X2+β3X3) Where α = the constant, βs = the variables and Xs = the logit coefficient for each variable. The baseline model (model 1) includes the social conservatism variable, office status variable and Constituency Euroscepticism for all MPs in the dataset. These are the three primary variables of interest in testing for MPs’ policy, office and vote-seeking intentions. The equation for this model is below:   Pr(leave=1)=logit−1(α+SocialConservatism1+OfficeStatus1+ConstituencyEuroscepticism1) Model 2 builds on the first model by including the three control variables listed in Section 3.6. The equation of this model is below:   Pr(leave=1)=logit−1(+SocialConservatism1+OfficeStatus1+ConstituencyEuroscepticism1+Age1+Gender1+PrivateSchool1) Model 3 tests whether the effect of constituency Euroscepticism differs for marginal and safe seats. This model adds the marginal seat dummy variable and an interaction variable: marginal × constituency Euroscepticism. The equation of this model is below:   Pr(leave=1)=logit−1(α+ SocialConservatism1+OfficeStatus1+ConstituencyEuroscepticism1+Age1+ Gender1+PrivateSchool1+MarginalSeat+MarginalSeat×ConstituencyEuroscepticism1) As a robustness check on the results, the linear probability (OLS) and probit versions of the models used in the statistical analysis are presented in the Appendix 1. 4. Results Table 2 below presents the results from the statistical analysis. The three models used in the table are described in Section 3.7. As logistic regression coefficients have no direct meaning, Table 2 presents the average marginal effects of the independent variables. The average marginal effects refer to the average effect of a 1 unit change in the independent variable. For example, Model 1 predicts that on average an increase of 1 on the social conservatism scale increases an MP’s probability of supporting leave by 0.05, and Model 2 predicts that the effect of being a frontbencher reduces the probability of supporting leave by 0.37 compared to an experienced backbencher. The Appendix 1 shows the logistic regression coefficients for the statistical models in Table 2. Table 2 Marginal effects on the probability of endorsing leave   Model 1 (Main variables)  Model 2 (includes demographic controls)  Model 3 (controls for constituency majority)  Social conservatism  0.050 (0.010)***  0.049 (0.011)***  0.050 (0.010)***  Office status reference category: experienced backbenchers  Newly elected MPs  −0.171 (0.069)**  −0.209 (0.087)**  −0.199 (0.090)**  Frontbenchers  −0.334 (0.058)***  −0.369 (0.061)***  −0.366 (0.061)***  Constituency Euroscepticisma  0.074 (0.031)**  0.058 (0.032)*  0.078 (0.038)**  Age    −0.004 (0.003)  −0.004 (0.003)  Gender (Female = 1)    −0.038 (0.068)  −0.042 (0.069)  Private school    −0.048 (0.050)  −0.047 (0.050)  Marginal constituency      0.345 (0.376)  Marginal × Constituency Euroscepticism      −0.007 (0.007)  Observations  324  309  309  Count R2  0.704  0.709  0.709  LR χ2b  66.17  68.28  69.91  p > χ2  >0.001  >0.001  >0.001    Model 1 (Main variables)  Model 2 (includes demographic controls)  Model 3 (controls for constituency majority)  Social conservatism  0.050 (0.010)***  0.049 (0.011)***  0.050 (0.010)***  Office status reference category: experienced backbenchers  Newly elected MPs  −0.171 (0.069)**  −0.209 (0.087)**  −0.199 (0.090)**  Frontbenchers  −0.334 (0.058)***  −0.369 (0.061)***  −0.366 (0.061)***  Constituency Euroscepticisma  0.074 (0.031)**  0.058 (0.032)*  0.078 (0.038)**  Age    −0.004 (0.003)  −0.004 (0.003)  Gender (Female = 1)    −0.038 (0.068)  −0.042 (0.069)  Private school    −0.048 (0.050)  −0.047 (0.050)  Marginal constituency      0.345 (0.376)  Marginal × Constituency Euroscepticism      −0.007 (0.007)  Observations  324  309  309  Count R2  0.704  0.709  0.709  LR χ2b  66.17  68.28  69.91  p > χ2  >0.001  >0.001  >0.001  Note: The numbers in the table are average marginal effects with robust standard errors in parentheses. The Count R2 refers to the proportion of correctly predicted outcomes from the model. *** p < 0.01, ** p < 0.05, * p < 0.1. aOne unit of constituency Euroscepticism = 10% disapproval of the EU, e.g. a constituency with 45.6% EU disapproval is coded as 4.56. bMeasure based on pseudolikelihood as robust standard errors have been used. The results found in the linear probability and probit models, presented in the Appendix 1, are qualitatively similar to those found in the logistic regression models. The results will now be considered in terms of policy, office and vote-seeking. 4.1 Policy-seeking The data provides strong evidence for a link between social conservatism and the probability of an MP endorsing leave. This relationship is highly statistically significant in all of the models. In addition to being statistically significant the relationship is also substantially significant as the models suggest that social conservatism has a strong effect on the probability of endorsing leave. Model 1 predicts that at the most socially conservative side of the scale an experienced backbencher has an 83% probability of supporting leave compared to 28% for a similar MP at the most socially liberal end of scale. Figure 2 plots this relationship. For frontbench MPs the model predicts that the probability of endorsing leave varies from 7% to 51% across the social conservatism scale. These results provide strong evidence to validate Hypothesis 1. This suggests that a TAN ideology among MPs could be motivating Euroscepticism. Figure 2 View largeDownload slide Marginal effect of social conservatism on the probability of endorsing leave. This is based on data from Model 1. The probabilities shown in the graph are for a backbench MP first elected before 2015. Social conservatism is measured on a scale of 1–10, see section 4.2. Figure 2 View largeDownload slide Marginal effect of social conservatism on the probability of endorsing leave. This is based on data from Model 1. The probabilities shown in the graph are for a backbench MP first elected before 2015. Social conservatism is measured on a scale of 1–10, see section 4.2. 4.2 Office-seeking Consistent with the predictions of Hypothesis 2, the results show substantial differences in referendum position between three career-status groups of MPs: experienced backbenchers, frontbenchers and new MPs. The difference between each of these three groups is statistically significant in all of the statistical models. The marginal effects indicate that the effect of office status on referendum position is strong. Model 3 suggests that being on the frontbench reduces the probability of supporting leave by 0.37 and being a new MP reduces the probability by 0.20, compared to an experienced backbencher. This suggests that the difference between being an experienced backbencher and a frontbencher is the equivalent of the effect of a 7.5 point difference (out of a possible 10) on the social conservatism scale and the difference between a new and experienced backbencher is the equivalent of a 4 point difference on the social conservatism scale. The summary statistics, see Figure 3 below, also demonstrate the sharp differences between the three groups. The only group with a majority supporting leave were the experienced backbenchers, 61% of whom supported leave. This is in contrast to less than a quarter (24%) of frontbench MPs supporting leave. Newly elected MPs, as predicted, occupy the middle group but are slightly closer to the frontbenchers with 38% supporting leave. Figure 3 View largeDownload slide Referendum support by office-status. Numbers in the bars refer to raw number of MPs. Experienced backbenchers refers to backbench MPs elected before 2015. Pearson χ2 (2 degrees of freedom) = 37.7; p < 0.001 Figure 3 View largeDownload slide Referendum support by office-status. Numbers in the bars refer to raw number of MPs. Experienced backbenchers refers to backbench MPs elected before 2015. Pearson χ2 (2 degrees of freedom) = 37.7; p < 0.001 These findings are consistent with the discussion in Section 2.2 which predicted that MPs’ office-seeking incentives, or lack thereof, to support the leadership would affect their referendum positions. 4.3 Vote-seeking The statistical analysis also provides evidence that vote-seeking motivations affected MPs’ positions on the referendum. All three models show a positive relationship between constituency Euroscepticism and the probability of an MP endorsing leave. This relationship is statistically significant to the 95% level in Models 1 and 3 and the 90% level in Model 2. These findings are in line with the predictions of Hypothesis 3. These data suggest that, across the range of the variable, constituency Euroscepticism could have had a strong effect on MPs’ referendum stance. Model 1 predicts that a frontbench MP representing a highly Eurosceptic constituency (70% against the EU) had a probability of endorsing leave of 0.37, more than three times higher than the probability, 0.11, of a frontbench MP representing a highly pro-EU constituency (25% against the EU). The same calculation for experienced backbenchers predicts that those representing a highly Eurosceptic constituency had a 0.72 chance of supporting leave compared to 0.35 for those representing a highly pro-EU constituency. There is, then, evidence to suggest that vote-seeking was an important factor affecting which side MPs backed in the referendum campaign. However, there are two caveats to this general finding. Firstly, there is no evidence that constituency Euroscepticism had a greater effect on MPs from marginal constituencies. The interaction variable of marginality and constituency Euroscepticism in Model 3 is not statistically significant. Several different thresholds for marginal constituencies were used in order to further test if an interaction could be found between constituency Euroscepticism and seat marginality5. None of these tests, however, produced a statistically significant result. Contrary to the predictions of Hypothesis 4, this suggests that MPs representing marginal seats were not more sensitive to constituency opinion than those representing safer seats. This could suggest that MPs from safe seats are equally as concerned with vote-seeking as those from marginal seats. The second caveat is that the effect of constituency Euroscepticism is weaker than both the effect of office status and social conservatism. All the models suggest that social conservatism is a better predictor of referendum preferences than constituency Euroscepticism. For example, Model 3 predicts that social conservatives in the most pro-EU seats were more likely to endorse leave and that social liberals in the most Eurosceptic constituencies were more likely to endorse remain. This suggests that when a potential conflict arose between an MP’s personal ideology and the opinion of their constituents, MPs were more likely to be guided by their own ideology. In relation to office-seeking, all the statistical models predict that a frontbench MP representing one of the most Eurosceptic constituencies was not statistically more likely to support leave than a backbench MP representing a strongly pro-EU constituency. This suggests that the effect of constituency Euroscepticism is relatively weak when compared to office status. In sum, the findings show a statistical link between constituency Euroscepticism and the likelihood of an MP adopting endorsing leave. However, no evidence was found that MPs in marginal constituencies were more responsive to constituency opinion. Furthermore, social conservatism and office status were stronger predictors of MPs’ referendum positions than constituency Euroscepticism. This suggests that whilst vote-seeking motivations may have been important for MPs they were of less importance than either ideology or office status. 5. Conclusion This article has sought to investigate why Conservative MPs endorsed different sides in the 2016 Brexit referendum. It has been argued that all three of Müller and Strøm’s (1999) aims: policy, office and votes, affected MPs’ positions on the referendum. However, vote-seeking motivations were of lesser importance than either policy or office-seeking motivations. Policy-seeking motivations appear to be highly influential as there is a very strong statistical relationship between social conservatism and endorsing leave. This relationship between social conservatism and Euroscepticism cannot be explained with any certainty. A plausible explanation is that MPs’ positions on the referendum are related to the TAN/GAL dimension of politics with TAN MPs being more likely to oppose the EU on the grounds that it threatens national sovereignty. This argument concerning the TAN/GAL political dimension and Euroscepticism is consistent with many in the literature who have argued that Conservative Euroscepticism is essentially nationalistic in nature (Sowemimo 1996; Berrington and Hague 1998; Crowson, 2007; Turner, 2000). Office-seeking motivations were also of crucial importance in influencing MPs’ position on the referendum. The statistics show notable differences between in the referendum positions of three distinct career status groups. Frontbench MPs were predominantly loyal to the Prime Minister in supporting the remain campaign. In contrast, a majority of experienced backbenchers endorsed leave. This is consistent with the theory that those who are climbing the ministerial ladder or those who have a realistic prospect of gaining a frontbench role are more likely to be loyal to the leader (Benedetto and Hix, 2007). Experienced backbenchers, on the other hand, have fewer incentives to be loyal as they may already have been dismissed from, or overlooked for, ministerial office (Benedetto and Hix, 2007). The differences between these groups of MPs may also have been affected by the fact that a post leave-vote leadership change would represent a potential career threat to frontbenchers but an opportunity for backbenchers out of favour with the Cameron leadership. In terms of vote-seeking, the models show a statistical association between constituency Euroscepticism and endorsing leave. This possibly indicates that MPs were seeking to improve their personal vote by following their constituents. However, there is no evidence that MPs representing marginal constituencies were more responsive to constituency opinion as might be expected if MPs were motivated by vote-seeking. Furthermore, the effect of constituency Euroscepticism is weaker than the effects of either social conservatism or office status. These results suggest that while vote-seeking may have been a motivation for MPs in the referendum, their personal policy preferences and career prospects had a stronger impact on their stance than constituency opinion. As the Brexit debate has several unique features such as the fact it divides parties, Conservative MPs have very strong views on Europe and that the issue was determined by referendum rather than Parliamentary vote, it is likely that the ordering of MPs’ priorities will be different for other issues. For example, on many routine areas of government policy like reform of the planning system, MPs may be less influenced by policy-seeking as most MPs will not have strongly held and long-established views on these issues. Moreover, on some issues MPs may be expected to place a higher priority on vote-seeking than in the Europe debate. As this debate was decided by a referendum, meaning the public not the politicians determined the result, MPs might reason that constituents were less likely to punish or reward them for their stance. The impact of vote-seeking on MPs could have also been supressed by the fact that most constituencies were divided. Only 82 (21%) Conservative MPs represented seats predicted to have 60% or more voting the same way in the referendum, meaning that most MPs would find themselves opposing at least 40% of their constituents whichever side they backed in the referendum. On issues where constituencies are highly united MPs may be more guided by vote-seeking considerations. For example, MPs representing constituencies on the Heathrow flight path or High Speed 2 rail route, whose residents could be expected to be united in opposition to these projects, might fear the electoral consequences if they were to take a different position to their constituents. The relative importance of vote-seeking to MPs could also be affected by the electoral system. The theoretical literature suggests that in single member plurality (SMP) systems, like the UK, MPs have more incentive to develop a personal vote than in party-centred systems such as closed-list PR where MPs are elected solely on the basis of their party label (Mitchell, 2000; Nielson 2003). However, empirical cross-national studies of both legislators’ perceptions (André et al., 2016) and voting behaviour (Söderlund, 2016) suggest that SMP systems provide less incentives for MPs to cultivate a personal vote than single-transferable vote (STV) systems where MPs have to compete against co-partisans to attract votes. The findings of this study, that there is evidence of vote-seeking but that the effect was weaker than either policy or office-seeking considerations, potentially conform to the literature as it is possible that under a closed list system the effect of vote-seeking could have been weaker and conversely under STV the effect of vote-seeking could have been stronger. Future research is needed on how legislators in different electoral systems prioritise their competing motivations on controversial topics like Europe. The conclusion of this article that policy-seeking and office-seeking are more important than vote-seeking suggests that the internal politics of the Conservative Party had a greater effect on MPs’ referendum positions than the wider electorate. This is not unexpected given that in the recent history of the Conservative Party, Europe has been one of the defining policy divides within the internal politics of the party but has been of only limited salience to the wider electorate (Bale, 2010). As this study has used observational data, the findings here should be treated with caution and the author accepts that alternative explanations could be reached from the data presented here. This paper makes no claim to have identified causal processes. Nevertheless, the author believes that this paper makes an important contribution to both the literature on the Conservative Party and Euroscepticism and the literature concerning legislative behaviour in British politics. Footnotes 1 The two referendum positions are referred hereafter as leave and remain. 2 Another policy divide which was not considered in this paper is the economic left/right divide. This was not considered as there is a general consensus in literature that the economic left (sometimes called wet) element of the party has declined substantially since the 1980s and 90s (Bale 2006; Heppell 2013). 3 The timeframe for the parliamentary activity (voting record, EDMS and APPGs) is from the MP’s first election to parliament and public comment takes account of any comments made by the MP since they became a parliamentary candidate. 4 Different versions of the test were tried using 4.9%, 9.9% and 19.9% majorities as the thresholds for marginal seats. 5 In the other tests 4.9%, 9.9% and 19.9% majorities were used as the thresholds for the marginal seat dummy variable Acknowledgements The author would like to thank David Hine, James Tilley, Raphael Heuwieser, Lewis Wilcocks and the two anonymous reviewers for their helpful comments and feedback. Bibliography André A., Depauw S., Martin S. ( 2016) ‘The Classification of Electoral Systems: Bringing Legislators Back In’, Electoral Studies , 42, 42– 53. Google Scholar CrossRef Search ADS   Bale T. ( 2006) ‘ Between a Soft and a Hard Place? The Conservative Party, Valence Politics and the Need for a New Eurorealism’, Parliamentary Affairs , 59, 385– 400. Google Scholar CrossRef Search ADS   Bale T. ( 2010) The Conservatives from Thatcher to Cameron , Cambridge, Polity Press. Baughman J. ( 2004) ‘ Party, Constituency, and Representation: Votes on Abortion in the British House of Commons’, Public Choice , 120, 63– 85. Google Scholar CrossRef Search ADS   Bartle J. ( 2002) ‘Why Labour Won Again’. In King A. (ed.) Britain at the Polls 2001 , London, Chatham House Publishers, pp. 164– 205. Benedetto G., Hix S. ( 2007) ‘ The Rejected, the Ejected, and the Dejected: Explaining Government Rebels in the 2001-2005 British House of Commons’, Comparative Political Studies , 40, 755– 781. Google Scholar CrossRef Search ADS   Berrington H., Hague R. ( 1998) ‘Europe, Thatcherism and Traditionalism: Opinion, Rebellion and the Maastricht Treaty in the Backbench Conservative Party, 1992–1994’, West European Politics , 21, 44– 71. Google Scholar CrossRef Search ADS   Cowley P., Garry J. ( 1998). ‘ The British Conservative Party and Europe: The Choosing of John Major.’, British Journal of Political Science , 28, 473– 499. Google Scholar CrossRef Search ADS   Cowley P., Norton P. ( 1999), ‘ Rebels and Rebellions: Conservative MPs in the 1992 Parliament’, The British Journal of Politics & International Relations , 1, 84– 105. Google Scholar CrossRef Search ADS   Cowley P., Stuart M. ( 2010) ‘ Where has all the trouble gone? British intra-party parliamentary divisions during the Lisbon ratification’, British Politics , 5, 133– 148. Google Scholar CrossRef Search ADS   Cowley P., Stuart M. ( 2012) ‘ The Cambusters: The Conservative European Union Referendum Rebellion of October 2011’, Political Quarterly , 83, 402– 406. Google Scholar CrossRef Search ADS   Crowson N. J. ( 2007) The Conservative Party and European Integration since 1945: At the Heart of Europe ? Abingdon, Routledge. Fontana C., Parsons C. ( 2015) ‘“One Woman's Prejudice”: Did Margaret Thatcher Cause Britain's Anti-Europeanism?’, JCMS: Journal of Common Market Studies , 53, 89– 105. Google Scholar CrossRef Search ADS   Forster A. ( 2002) Euroscepticism in Contemporary British Politics: Opposition to Europe in the British Conservative and Labour Parties Since 1945 , London, Routledge. Google Scholar CrossRef Search ADS   Garry J. ( 1995) ‘ The British Conservative Party: Divisions over European policy’, West European Politics , 18, 170– 189. Google Scholar CrossRef Search ADS   Hanretty C., Lauderdale B. E., Vivyan N. ( 2016) ‘Dyadic Representation in a Westminster System’, Legislative Studies Quarterly , Advanced Access published online: 01 October 2016, DOI: 10.1111/lsq.12148. Hayton R. ( 2010) ‘ Conservative Party Modernisation and David Cameron’s Politics of the Family’, Political Quarterly , 81, 492– 500. Google Scholar CrossRef Search ADS   Heppell T. ( 2002) ‘ The Ideological Composition of the Parliamentary Conservative Party 1992–97’, British Journal of Politics and International Relations , 4, 299– 324. Google Scholar CrossRef Search ADS   Heppell T. ( 2013), ‘ Cameron and Liberal Conservatism: Attitudes within the Parliamentary Conservative Party and Conservative Ministers’, The British Journal of Politics & International Relations , 15, 340– 361. Google Scholar CrossRef Search ADS   Heppell T., Hill M. ( 2008), ‘ The Conservative Party Leadership Election of 1997: An Analysis of the Voting Motivations of Conservative Parliamentarians’, British Politics , 3, 63– 91. Google Scholar CrossRef Search ADS   Heppell T., Hill M. ( 2010) ‘ The Voting Motivations of Conservative Parliamentarians in the Conservative Party Leadership Election of 2001’, Politics , 30, 36– 51. Google Scholar CrossRef Search ADS   Heppell T., Crines A., Jeffery D. ( 2017) ‘The United Kingdom Referendum on European Union Membership: The Voting of Conservative Parliamentarians’ JCMS: Journal of Common Market Studies . doi: 10.1111/jcms.12529. Hooghe L., Marks G., Wilson C. ( 2002) ‘ Does Left/Right Structure Party Positions on European Integration?’, Comparative Political Studies , 35, 965– 89. Google Scholar CrossRef Search ADS   Kam C. J. ( 2009) Party Discipline and Parliamentary Politics , Cambridge, Cambridge University Press. Google Scholar CrossRef Search ADS   Kaufman E. ( 2016) ‘Brexit: it’s NOT the Economy, Stupid’, 10 Gower Street: The Birkbeck Politics Department Blog, accessed at: http://10-gower-street.com/2016/07/08/brexit-its-not-the-economy-stupid/ on 17 September 2016. Martin S. ( 2016) ‘ Policy, Office and Votes: The Electoral Value of Ministerial Office’, British Journal of Political Science , 46, 281– 296. Google Scholar CrossRef Search ADS   Matland R. E., Studlar D. T. ( 2004) ‘ Determinants of Legislative Turnover: A Cross-National Analysis’, British Journal of Political Science , 34, 87– 108. Google Scholar CrossRef Search ADS   McAllister I., Studlar D. T. ( 2000) ‘ Conservative Euroscepticism and the Referendum Party in the 1997 British General Election’, Party Politics , 6, 359– 371. Google Scholar CrossRef Search ADS   Mitchell P. ( 2000) ‘ Voters and their Representatives: Electoral Institutions and Delegation in Parliamentary Democracies’, European Journal of Political Research , 37, 335– 351. Müller W. C., Strøm K. ( 1999) ‘Political Parties and Hard Choice’. In Müller W. C., Strøm K. (eds) Policy, Office, or Votes? How Political Parties in Western Europe Make Hard Choices , New York, Cambridge University Press, pp. 1– 35. Google Scholar CrossRef Search ADS   Nielson D. L. ( 2003), ‘ Supplying Trade Reform: Political Institutions and Liberalization in Middle-Income Presidential Democracies’ American Journal of Political Science , 47, 470– 491. Google Scholar CrossRef Search ADS   Norris P, Crewe I. ( 1994) ‘ Did the British Marginals Vanish? Proportionality and Exaggeration in the British Electoral System Revisited.’ Electoral Studies , 13, 201– 221. Google Scholar CrossRef Search ADS   Norton P. ( 1978) Conservative Dissidents: dissent within the Parliamentary Conservative Party, 1970-74 , London, Temple Smith. Pattie C. J., Fieldhouse E. A., Johnston R. J. ( 1994) ‘ The Price of Conscience: the Electoral Correlates and Consequences of Free Votes and Rebellions in the British House of Commons, 1987–1992’, British Journal of Political Science , 24, 359– 380. Google Scholar CrossRef Search ADS   Prosser C. ( 2016) ‘ Dimensionality, Ideology and Party Positions Towards European Integration’, West European Politics , 39, 731– 754. Google Scholar CrossRef Search ADS   Read M., Marsh D. ( 1997) ‘ The Family Law Bill: Conservative Party splits and Labour Party Cohesion’, Parliamentary Affairs , 50, 263– 279. Google Scholar CrossRef Search ADS   Szczerbiak A., Taggart P. ( 2008) ‘Theorising Party-Based Euroscepticism: Problems of Definition, Measurement and Causality’. In Taggart P., Szczerbiak A. (eds) Opposing Europe: The Comparative Party Politics of Euroscepticism Volume 2 , Oxford, Oxford University Press, pp. 238– 262. Smith K. B., Oxley D. R., Hibbing M. V., Alford J. R., Hibbing J. R. ( 2011) ‘ Linking Genetics and Political Attitudes: Reconceptualizing Political Ideology’, Political Psychology , 32, 369– 397. Google Scholar CrossRef Search ADS   Söderlund P. ( 2016). ‘ Candidate-Centred Electoral Systems and Change in Incumbent Vote Share: A Cross-national and Longitudinal Analysis’, European Journal of Political Research , 55, 321– 339. Google Scholar CrossRef Search ADS   Sowemimo J. ( 1996) ‘ The Conservative Party and European Integration 1988-95’, Party Politics , 2, 77– 97. Google Scholar CrossRef Search ADS   Tillman E. R. ( 2013) ‘ Authoritarianism and Citizen Attitudes Towards European Integration’, European Union Politics , 14, 566– 589. Google Scholar CrossRef Search ADS   Turner J. ( 2000) The Tories and Europe , Manchester, Manchester University Press. Tzeglove E. ( 2014) ‘ Cross-Cutting Issues, Intraparty Dissent and Party Strategy: The Issue of European Integration in the House of Commons’, European Union Politics , 15, 3– 23. Google Scholar CrossRef Search ADS   Webb P. D., Bale T. ( 2014) ‘ Why Do Tories Defect to UKIP? Conservative Party Members and the Temptations of the Populist Radical Right’, Political Studies , 62, 961– 970. Google Scholar CrossRef Search ADS   Webb P. D., Childs S. ( 2011) ‘ Wets and Dries Resurgent? Intra-Party Alignments Among Contemporary Conservative Party Members’, Parliamentary Affairs , 64, 383– 402. Google Scholar CrossRef Search ADS   Appendix 1 Table A1 Logit regression coefficients of the probability of MPs backing leave   Model 1 (Main variables)  Model 3 (includes demographic controls)  Model 4 (controls for constituency majority)  Social conservatism  0.254 (0.060)***  0.256 (0.062)***  0.261 (0.063)***  Office status reference category: experienced backbenchers        Newly elected MPs  −0.762 (0.309)**  −0.943 (0.400)**  −0.901 (0.415)**  Frontbenchers  −1.582 (0.298)***  −1.771 (0.326)***  −1.766 (0.325)***  Constituency Euroscepticisma  0.374 (0.164)**  0.303 (0.180)*  0.404 (0.201)**  Age    −0.0196 (0.016)  −0.230 (0.163)  Gender (Female = 1)    −0.195 (0.357)  −0.218 (0.361)  Private school    −0.247 (0.262)  −0.246 (0.264)  Marginal constituency      1.798 (1.955)  Marginal × Constituency Euroscepticism      0.385 (0.363)  Constant  −2.921 (0.931)***  −1.360 (1.304)  −1.664 (1.433)  Observations  324  309  309  Count R2  0.704  0.709  0.709  LR χ2b  66.17  68.28  69.91  p > χ2  >0.001  >0.001  >0.001    Model 1 (Main variables)  Model 3 (includes demographic controls)  Model 4 (controls for constituency majority)  Social conservatism  0.254 (0.060)***  0.256 (0.062)***  0.261 (0.063)***  Office status reference category: experienced backbenchers        Newly elected MPs  −0.762 (0.309)**  −0.943 (0.400)**  −0.901 (0.415)**  Frontbenchers  −1.582 (0.298)***  −1.771 (0.326)***  −1.766 (0.325)***  Constituency Euroscepticisma  0.374 (0.164)**  0.303 (0.180)*  0.404 (0.201)**  Age    −0.0196 (0.016)  −0.230 (0.163)  Gender (Female = 1)    −0.195 (0.357)  −0.218 (0.361)  Private school    −0.247 (0.262)  −0.246 (0.264)  Marginal constituency      1.798 (1.955)  Marginal × Constituency Euroscepticism      0.385 (0.363)  Constant  −2.921 (0.931)***  −1.360 (1.304)  −1.664 (1.433)  Observations  324  309  309  Count R2  0.704  0.709  0.709  LR χ2b  66.17  68.28  69.91  p > χ2  >0.001  >0.001  >0.001  Note: The numbers in the table are logit regression coefficients with robust standard errors in parentheses. The Count R2 refers to the proportion of correctly predicted outcomes from the model. *** p < 0.01, ** p < 0.05, * p < 0.1. aOne unit of constituency Euroscepticism = 10% disapproval of the EU, e.g. a constituency with 45.6% EU disapproval is coded as 4.56. bMeasure based on pseudolikelihood as robust standard errors have been used. Table A2 Marginal effects on the probability of endorsing leave (Probit version)   Model 1 (Main variables)  Model 3 (includes demographic controls)  Model 4 (controls for constituency majority)  Social conservatism  0.049 (0.010)***  0.049 (0.011)***  0.049 (0.011)***  Office status reference category: experienced backbenchers  Newly elected MPs  −0.174 (0.070)**  −0.214 (0.087)**  −0.205 (0.090)**  Frontbenchers  −0.332 (0.058)***  −0.367 (0.061)***  −0.365 (0.061)***  Constituency Euroscepticisma  0.075 (0.032)**  0.060 (0.033)*  0.081 (0.038)**  Age    −0.004 (0.003)  −0.004 (0.003)  Gender (Female = 1)    −0.029 (0.069)  −0.034 (0.069)  Private school    −0.051 (0.050)  −0.051 (0.050)  Marginal constituency      0.382 (0.387)  Marginal × Constituency Euroscepticism      −0.008 (0.007)  Observations  324  309  309  Count R2  0.704  0.722  0.709  LR χ2b  65.78  67.78  69.56  p > χ2  >0.001  >0.001  >0.001    Model 1 (Main variables)  Model 3 (includes demographic controls)  Model 4 (controls for constituency majority)  Social conservatism  0.049 (0.010)***  0.049 (0.011)***  0.049 (0.011)***  Office status reference category: experienced backbenchers  Newly elected MPs  −0.174 (0.070)**  −0.214 (0.087)**  −0.205 (0.090)**  Frontbenchers  −0.332 (0.058)***  −0.367 (0.061)***  −0.365 (0.061)***  Constituency Euroscepticisma  0.075 (0.032)**  0.060 (0.033)*  0.081 (0.038)**  Age    −0.004 (0.003)  −0.004 (0.003)  Gender (Female = 1)    −0.029 (0.069)  −0.034 (0.069)  Private school    −0.051 (0.050)  −0.051 (0.050)  Marginal constituency      0.382 (0.387)  Marginal × Constituency Euroscepticism      −0.008 (0.007)  Observations  324  309  309  Count R2  0.704  0.722  0.709  LR χ2b  65.78  67.78  69.56  p > χ2  >0.001  >0.001  >0.001  Note: The numbers in the table are average marginal effects with robust standard errors in parentheses. The Count R2 refers to the proportion of correctly predicted outcomes from the model. *** p < 0.01, ** p < 0.05, * p < 0.1. aOne unit of constituency Euroscepticism = 10% disapproval of the EU, e.g. a constituency with 45.6% EU disapproval is coded as 4.56. bMeasure based on pseudolikelihood as robust standard errors have been used. Table A3 Probit regression coefficients of the probability of MPs backing leave   Model 1 (Main variables)  Model 3 (includes demographic controls)  Model 4 (controls for constituency majority)  Social conservatism  0.150 (0.035)***  0.151 (0.036)***  0.153 (0.037)***  Office status reference category: experienced backbenchers  Newly elected MPs  −0.477 (0.191)**  −0.590 (0.245)**  −0.567 (0.253)**  Frontbenchers  −0.948 (0.175)***  −1.059 (0.191)***  −1.059 (0.190)***  Constituency Euroscepticisma  0.228 (0.100)**  0.185 (0.102)*  0.253 (0.122)**  Age    −0.011 (0.009)  −0.014 (0.010)  Gender (Female = 1)    −0.091 (0.214)  −0.106 (0.215)  Private school    −0.159 (0.157)  −0.159 (0.157)  Marginal constituency      1.187 (1.205)  Marginal × Constituency Euroscepticism      −0.025 (0.022)  Constant  −1.753 (0.565)***  −0.811 (0.790)  −1.020 (0.865)  Observations  324  309  309  Count R2  0.704  0.722  0.709  LR χ2b  65.78  67.78  69.56  p > χ2  >0.001  >0.001  >0.001    Model 1 (Main variables)  Model 3 (includes demographic controls)  Model 4 (controls for constituency majority)  Social conservatism  0.150 (0.035)***  0.151 (0.036)***  0.153 (0.037)***  Office status reference category: experienced backbenchers  Newly elected MPs  −0.477 (0.191)**  −0.590 (0.245)**  −0.567 (0.253)**  Frontbenchers  −0.948 (0.175)***  −1.059 (0.191)***  −1.059 (0.190)***  Constituency Euroscepticisma  0.228 (0.100)**  0.185 (0.102)*  0.253 (0.122)**  Age    −0.011 (0.009)  −0.014 (0.010)  Gender (Female = 1)    −0.091 (0.214)  −0.106 (0.215)  Private school    −0.159 (0.157)  −0.159 (0.157)  Marginal constituency      1.187 (1.205)  Marginal × Constituency Euroscepticism      −0.025 (0.022)  Constant  −1.753 (0.565)***  −0.811 (0.790)  −1.020 (0.865)  Observations  324  309  309  Count R2  0.704  0.722  0.709  LR χ2b  65.78  67.78  69.56  p > χ2  >0.001  >0.001  >0.001  Note: The numbers in the table are probit regression coefficients with robust standard errors in parentheses. The Count R2 refers to the proportion of correctly predicted outcomes from the model. *** p < 0.01, ** p < 0.05, * p < 0.1. aOne unit of constituency Euroscepticism = 10% disapproval of the EU, e.g. a constituency with 45.6% EU disapproval is coded as 4.56. bMeasure based on pseudolikelihood as robust standard errors have been used. Table A4 Linear probability model (OLS) of MPs endorsing leave   Model 1 (Main variables)  Model 3 (includes demographic controls)  Model 4 (controls for constituency majority)  Social conservatism  0.051 (0.011)***  0.050 (0.012)***  0.051 (0.012)***  Office Status Reference Category: experienced backbenchers  Newly elected MPs  −0.180**  −0.215**  −0.205**    (0.071)  (0.088)  (0.091)  Frontbenchers  −0.331***  −0.365***  −0.361***    (0.058)  (0.062)  (0.062)  Constituency Euroscepticism‡  0.072 (0.031)**  0.057 (0.032)*  0.073 (0.038)**  Age    −0.004 (0.003)  −0.004 (0.003)  Gender (Female = 1)    −0.028 (0.068)  −0.029 (0.069)  Private School    −0.053 (0.052)  −0.053 (0.052)  Marginal constituency      0.275 (0.373)  Marginal × Constituency Euroscepticism      −0.006 (0.007)  Constant  −0.58 (0.174)  0.251 (0.264)  0.210 (0.282)  Observations  324  309  309  R2  0.189  0.203  0.206    Model 1 (Main variables)  Model 3 (includes demographic controls)  Model 4 (controls for constituency majority)  Social conservatism  0.051 (0.011)***  0.050 (0.012)***  0.051 (0.012)***  Office Status Reference Category: experienced backbenchers  Newly elected MPs  −0.180**  −0.215**  −0.205**    (0.071)  (0.088)  (0.091)  Frontbenchers  −0.331***  −0.365***  −0.361***    (0.058)  (0.062)  (0.062)  Constituency Euroscepticism‡  0.072 (0.031)**  0.057 (0.032)*  0.073 (0.038)**  Age    −0.004 (0.003)  −0.004 (0.003)  Gender (Female = 1)    −0.028 (0.068)  −0.029 (0.069)  Private School    −0.053 (0.052)  −0.053 (0.052)  Marginal constituency      0.275 (0.373)  Marginal × Constituency Euroscepticism      −0.006 (0.007)  Constant  −0.58 (0.174)  0.251 (0.264)  0.210 (0.282)  Observations  324  309  309  R2  0.189  0.203  0.206  Note: Robust standard errors in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1. aOne unit of constituency Euroscepticism = 10% disapproval of the EU, e.g. a constituency with 45.6% EU disapproval is coded as 4.56. © The Author 2017. Published by Oxford University Press on behalf of the Hansard Society; all rights reserved. For Permissions, please e-mail: journals.permissions@oup.com

Journal

Parliamentary AffairsOxford University Press

Published: Jan 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off