Scholarly article on topic 'Junior coalition parties in the British context: Explaining the Liberal Democrat collapse at the 2015 general election'

Junior coalition parties in the British context: Explaining the Liberal Democrat collapse at the 2015 general election Academic research paper on "Political Science"

CC BY
0
0
Share paper
Academic journal
Electoral Studies
OECD Field of science
Keywords
{Coalition / "Junior parties" / "Valence politics" / Trust / Incumbency / Campaigning}

Abstract of research paper on Political Science, author of scientific article — Craig Johnson, Alia Middleton

Abstract The Liberal Democrats’ performance in the 2015 general election provides an opportunity to examine the only case in the post-war period of a national junior coalition partner in British politics. Comparative research highlights competence, trust and leadership as three key challenges facing junior coalition parties. This article uses British Election Study data to show that the Liberal Democrats failed to convince the electorate on all three counts. The article also uses constituency-level data to examine the continued benefits of incumbency to the party and the impact of constituency campaigning. It finds that while the incumbency advantage remained for the Liberal Democrats, it was ultimately unable to mitigate the much larger national collapse.

Academic research paper on topic "Junior coalition parties in the British context: Explaining the Liberal Democrat collapse at the 2015 general election"

Contents lists available at ScienceDirect

Electoral Studies

journal homepage: www.elsevier.com/locate/electstud

Junior coalition parties in the British context: Explaining the Liberal Democrat collapse at the 2015 general election

Craig Johnson a *, Alia Middleton b

a Newcastle University. Politics Building, 40-42 Great North Road, Newcastle University, Newcastle upon Tyne, NE1 7RU, UK b University of Surrey. Faculty of Arts and Social Sciences, AC Building, University of Surrey, Guildford, GU2 7XH, UK

CrossMark

ARTICLE INFO

Article history:

Received 2 December 2015

Received in revised form

16 May 2016

Accepted 31 May 2016

Available online 2 June 2016

Keywords: Coalition Junior parties Valence politics Trust

Incumbency Campaigning

ABSTRACT

The Liberal Democrats' performance in the 2015 general election provides an opportunity to examine the only case in the post-war period of a national junior coalition partner in British politics. Comparative research highlights competence, trust and leadership as three key challenges facing junior coalition parties. This article uses British Election Study data to show that the Liberal Democrats failed to convince the electorate on all three counts. The article also uses constituency-level data to examine the continued benefits of incumbency to the party and the impact of constituency campaigning. It finds that while the incumbency advantage remained for the Liberal Democrats, it was ultimately unable to mitigate the much larger national collapse.

© 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license

(http://creativecommons.org/licenses/by/4.0/).

1. Introduction

At the 2010 British general election, the Liberal Democrats won 23 percent of the vote and 57 seats. It was an electorally relevant share of seats that, following five days of negotiations, facilitated the Conservative-Liberal Democrat coalition. Despite arguably satisfying the office-seeking and policy-seeking criteria associated with coalition theory (Bale, 2012), the electoral prospects for the Liberal Democrats were bleak. The comparative literature on coalitions suggests that junior coalition parties tend to struggle in subsequent elections, and often face an uphill battle to get noticed by the electorate (Bolleyer, 2008; Dunphy and Bale, 2011). The challenge for junior parties in coalition is to be competent in government, while maintaining party distinctiveness and popular leadership (Boston and Bullock, 2012; Paun and Munro, 2013). The case of the Liberal Democrats provides a unique opportunity to apply the comparative literature on junior coalition parties to the British context.

Following heavy defeats in local, sub-national and national elections during the 2010 parliament, the Liberal Democrats ran a

* Corresponding author. E-mail addresses: craig.johnson1@newcastle.ac.uk (C. Johnson), a.middleton@ surrey.ac.uk (A. Middleton).

highly defensive campaign ahead of the 2015 general election (Coetzee, 2015). It was unsuccessful. The Liberal Democrats won 7.9 percent of the vote and just 8 seats. In urban areas of northern England where they had built up support as the opposition to Labour in both local and Westminster elections, they were heavily beaten. Standing against the Conservatives in the south west of England, long-standing Liberal Democrat MPs were wiped out. In Scotland, along with Labour and the Conservatives, they lost heavily to the Scottish National Party. Their efforts over a generation to win an electorally relevant share of seats have now collapsed and will take a monumental effort to rebuild.

This article examines the reasons behind the Liberal Democrats' collapse at the 2015 general election. The first section places the Liberal Democrats' experience within the comparative literature on junior coalition parties, and outlines various explanations for their collapse. The second section analyses the Liberal Democrats' particularly defensive campaign strategy in the 2015 general election, and examines the continued importance of incumbency to the party. It finds that while the incumbency effect still reaps benefits for the Liberal Democrats, it was not enough to withstand the fall in the national vote. The third section utilises data from the 2015 British Election Study to identify the individual reasons behind voting (or not) for the Liberal Democrats. It finds that the Liberal Democrats suffered badly due to a lack of perceived competence

http://dx.doi.org/10.1016/j.electstud.2016.05.007

0261-3794/© 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).

and influence, and badly lost the trust of their electorate.

This article contributes not only to analysis of the Liberal Democrats, but also informs broader comparative literature in two key respects. First, the 2010-2015 Conservative-Liberal Democrat coalition was the first formal UK coalition in the post-war period, and examining the Liberal Democrats' collapse in the 2015 general election informs broader analysis of junior parties in coalition. It explores whether junior parties must always suffer in coalition and the key tasks for them to overcome the challenges they face. Second, the article also examines the continuing success of constituency campaigning and incumbency strategies, shown to be important both to the Liberal Democrats but also to political parties more broadly (Fisher et al., 2011, 2015; Smith, 2013). How successful was this in the 2015 general election? This article examines these questions.

2. The comparative perspective and electoral context

The comparative literature on coalitions works from two main criteria. First, as parties are office-seeking, they should share as many of the spoils of office with as few parties as possible (Riker, 1962). When achieved, this is known as the minimum-winning coalition: a government that has an overall majority (winning) amongst the smaller number of parties (minimum). Second, parties are also policy-seeking, and look to form coalitions that broadly coincide with their principles and policy programmes (Axelrod, 1970). When achieved alongside office-seeking priorities, the resulting government is the minimum-winning connected coalition.

When the Liberal Democrats joined the Conservatives in coalition following the 2010 general election, they arguably satisfied these two main criteria (Bale, 2012). Needing 326 MPs to pass a Queen's Speech and budget, Labour, with 258 MPs, could not even begin to think about governing without the support of the Liberal Democrats' 57 MPs. Meanwhile, the Conservatives on their own with 307 MPs would have also found themselves short should a 'rainbow coalition' of the centre-left (including the Liberal Democrats) try and defeat them. Even if the Conservatives could pass a Queen's Speech, they would have been unable to govern comfortably. The subsequent Conservative-Liberal Democrat coalition, with a de-facto majority of 80, was therefore the minimum-winning coalition. The Conservative-Liberal Democrat coalition also arguably satisfied the policy-seeking requirement. The Liberal Democrats' professionalisation in recent years has been accompanied by a more equidistant strategy that made a coalition with the Conservatives easier to navigate (Evans and Sanderson-Nash, 2011).

Boston and Bullock (2012) note the tension that exists between governmental unity and party distinctiveness. This tension is particularly strong for junior parties in coalition without a history of government and coalitional compromise (Bolleyer, 2008), such as the Liberal Democrats. How did the Liberal Democrats perform in the 2010 parliament in this regard? Bennisterand Heffernan (2015) argued that the Conservative-Liberal Democrat coalition was defined by consensus and compromise, with both sides operating from a position of mutual trust and respect. Despite doubts that it would be the case (Bale, 2012), the coalition lasted the full five years with remarkably few threats to its existence. What of party distinctiveness? An analysis of the 2010 coalition agreement between the two parties showed that both parties satisfied the main commitments of their manifesto, and the Liberal Democrats in particular were in a position to deliver much of the minutiae of their manifesto (Quinn et al., 2011). However in doing so, they gave ground to the Conservatives on the key issues of deficit reduction, tax, welfare, education, health, immigration and defence policy (Bale and Webb, 2015). As Bale (2012: 328) wryly notes, the

coalition agreement showed 'what happens when vegetarians negotiate with carnivores'.

The unity/distinctiveness dilemma invites a key question in relation to how junior parties are perceived in coalition: who do voters blame or credit for a coalition government's performance? Comparative evidence suggests that the lead governing party within a coalition, the 'proposer' of the coalition and the party with the 'chief executive', will bear most of the responsibility with voters for government performance (Duch et al., 2015; Fisher and Hobolt, 2010). This suggests that in the UK context of 2010-2015, the Conservatives were more likely to be credited or blamed by the electorate than the Liberal Democrats for the government's performance. Indeed, while the Liberal Democrats can claim successful implementation of some of their policies1, polling conducted in 2014 suggested that the Liberal Democrats' key policy delivered in government, such as increasing the income tax threshold, was primarily associated by the voters with the Conservatives (Bennett,

2014). As Behr (2014) argues, 'one difference between opposition and government for Liberal Democrats has been that, before coalition, no one noticed what they said; now, no one notices what they do'. In short, by prioritising coalition unity in the early years of the coalition, they sacrificed party distinctiveness (McEnhill,

2015).

The Liberal Democrats' lack of influence and distinctiveness suggests that their collapse can be explained by their voters feeling betrayed by the party (Cutts and Russell, 2015). More generally, Muller and Strom (1999) highlight the importance of maintaining party legitimacy, while Dunphy and Bale (2011) suggest that parties risk losing their identity in coalition with a larger party. Dommett (2013) suggests that a number of decisions by the Liberal Democrats created a schism between their rhetoric as an opposition party and their rhetoric in government, fostering a perception of distrust and betrayal.

Junior coalition parties also face a number of other challenges. First, they must appear competent as a party in government (Muller and Strom, 1999). Clarke et al. (2009) argue that competence has increasingly shaped party competition in Britain. Green (2015) argues that the Liberal Democrats' perceived competence on their key policy issues increased support for the party. However as this support has been gained based on difference from the two major parties (Green and Hobolt, 2008), the argument arises that they might have lost competence on key issues after aligning themselves too closely to the Conservatives in coalition. Indeed, on the three key valence issues ahead of the 2015 general election (the economy, immigration and health care), the Conservatives were more trusted than the Liberal Democrats (YouGov, 2015).

As well as perception of competence of parties' key policy issues, the competence and popularity of party leaders is also argued to be important. Stokes (1992) argues that an assessment of a party's competence is shaped 'from its experience with the parties and the leaders, and the results they achieve, over time'. The assessment of a party leader is thus something to be considered alongside assessment of a broader political party (Clarke et al., 2009; Whiteley et al., 2013). At the 2010 general election, party leader Nick Clegg was a source of electoral advantage for the Liberal Democrats (Cutts, 2012; Middleton, 2015). However, his popularity plunged throughout the 2010 parliament, and he was regarded as

1 The threshold at which people start paying income tax was £7475 in 2010-2011, and now stands at £10,600 (as of the financial year 2015-2016). The pupil premium ring-fences money in the education budget to go directly to the most deprived schools in the country. The coalition government abolished identity cards and child detention, radically reformed the pensions system, and implemented much of the Liberal Democrats' environmental policy.

25.00%

20.00%

15.00%

10.00%

5.00% 0.00%

Fig. 1. Liberal democrat poll ratings, 2010-2015. Note: National opinion polling data between May 2010 and May 2015, taken from Mark Pack (2015) national opinion polling database.2

the least trusted leader of the then three main political parties ahead of the 2015 general election (Denver, 2015).

The Liberal Democrats' poll rating dipped quickly and then continually fell during the 2010 parliament (see Fig. 1), and the 2015 general election saw the near-wipeout of Liberal Democrat MPs. They received just 7.9 percent of the vote, a drop of 15.2 percentage points on their performance in 2010. They returned just 8 MPs, compared with 57 MPs in 2010. Of the 49 seats lost, 28 were lost to the Conservatives, 12 to Labour and 9 to the Scottish National Party. They made no gains, and their vote share fell in every constituency across the country. Whilst in 2010 the Liberal Democrats did not lose a single deposit3, in 2015 they lost 340 deposits, costing them £170,000. The next section explains the constituency factors behind the party's collapse.

3. Liberal Democrat local performance

This section examines the locally targeted strategy of the Liberal Democrat campaign. Despite the fall in the party's national support, various academic forecasting models predicted that the party would do better than a uniform national swing based on opinion polls would suggest (Monk and Lambert, 2015). This was based on the party's ability to target seats through locally focused campaigns (Cutts, 2014; Fieldhouse et al., 2006). Local campaigning is no longer an irrelevant sideshow (Butler and Kavanagh, 1988) to the national campaign, but a way in which political parties in the UK and beyond can alter local results (Jacobson, 2006; Pattie and Johnston, 2009; Whiteley and Seyd, 2003). Such a targeted strategy was the only realistic option for the party. Many of the Liberal Democrats' MPs had been respected community representatives and locally popular for a long period of time (Smith, 2013), and their local electoral and organisational strength in these constituencies post-2010 suggested that they might be able to mitigate the effects of a collapse in their national vote share (Johnson, 2014). As the Liberal Democrats' election strategist Ryan Coetzee (2015) has outlined, the party embarked on a targeted consolidation strategy, focusing almost entirely on seats where they had the incumbent MP.

How defensive was the Liberal Democrat campaign in the 2015 general election? One method of answering this question is to

2 It should be noted that there was a major polling error over the 2010 parliament. While the estimates for the Liberal Democrats were broadly correct, we do not know that this was correct throughout the entire parliament.

3 Candidates lose their deposit (£500) when they win less than five per cent of the vote in a constituency election.

examine the visits made by party leaders to constituencies. During election campaigns party leaders tour the country, visiting constituencies and candidates. Political parties 'weaponise' their leaders at the local level, bringing the national campaign if not on to people's doorsteps, then on to the streets that surround them. Leader Nick Clegg, accompanied by the media, spent the weeks of the campaign touring the country in his battle bus (accidentally killing a pigeon along the way), donning hard hats and high-vis jackets as the occasion demanded. The bus broke down on more than one occasion and the occupants (including a heavy Press contingent) headed off to the local pub. Such visits made by party leaders can have a significant impact on party vote share at the local level (Carty and Eagles, 2005; Holbrook, 2002), with Nick Clegg's visits during the 2010 general election campaign boosting the Liberal Democrats' vote share on average by 1.5 percentage points (Middleton, 2015).

Table 1 identifies the number of constituencies visited by the three main party leaders during the 2015 general election campaign, and what percentage were to seats held by their own party. By isolating the incumbency of the constituencies visited, it is possible to identify expansionist visits (to seats held by other parties) and defensive visits (to seats already held). The evidence from Table 1 confirms that Nick Clegg ran the most defensive visit strategy of all three party leaders, with almost three quarters of his visits being to seats held by the Liberal Democrats. In the final three days of campaigning alone (4-6th May), all bar one of the 15 constituencies he visited were being defended by his own party. Contrast this to 2010, when just 23.9 percent of his total number of visits were to seats his party already held (Middleton, 2015).

As indicated by Nick Clegg's visits to constituencies the party already held, the Liberal Democrats were running a defensive electoral strategy. However, to understand the impact that this defensive strategy had on local results, we must consider the local electoral contexts and explore the types of seats where the Liberal Democrats lost votes. Table 2 breaks down the falls in the party's vote share according to whether the Liberal Democrats held the seat or were the second-placed party. In seats held by the Liberal Democrats, their vote share dropped on average three percentage points less than where they were placed second, although this was still insufficient to retain 49 of their seats.

To examine these seats in more detail, Table 2 also identifies the vulnerability of the constituency (Jacobson, 1987) by isolating safe and marginal constituencies. Once these two incumbency scenarios are disaggregated according to whether the seat was marginal or safe, the falls in vote share take on an interesting pattern. In the constituencies that the party already held, they did worse in those that were safe. The largest drop in a safe Liberal Democrat-held constituency came in Bristol West, where the incumbent Stephen

Table 1

Visits by party leaders during the 2015 short campaign.

Number of constituencies visited Visits in 2015 to constituencies held by their own party (%) Change from 2010 (%)

David Cameron 69 56.5 +49.2

Ed Miliband 51 33.3 -53.8

Nick Clegg 44 70.4 +46.5

Source: Leadership visits to constituencies between 30 March 2015 and 6 May 2015. Data from 2010 retrieved from Middleton (2015).

Table 2

Liberal Democrat vote share change according to incumbency and marginality.

Percentage point change 2010—2015

Overall -15.1

Liberal Democrat held -15.7

Marginal -15.0

Safe -16.3

Liberal Democrat challenger -18.7

Marginal -22.7

Safe -18.0

Liberal Democrat held, Conservative challenger -16.4

Liberal Democrat held, Labour challenger -15.0

Liberal Democrat held, SNP challenger -3.3a

Seats won in 2005 but lost in 2010 -21.7

Source: 2015 British Election Study Constituency Dataset.

N.B: Safe/marginal is classified according to the percentage majority of the constituency going into the 2015 election. Marginal seats have a majority of <9.99 percentage points, safe seats >10.00 points. For a discussion of the appropriateness of this measure see Cornford and Dorling (1997).

a Gordon was the only constituency where the Scottish National Party was in second place prior to the 2015 election. Overall, in seats the SNP gained from the Liberal Democrats, their vote share fell on average by 10.1 percentage points.

Williams saw a drop in his vote of 29.2 percentage points, losing his seat and coming third. However, the reverse is true for the seats in which the party were second, with the Liberal Democrat vote share falling on average 4.7 percentage points more in marginal seats where they were in second place. This raises the possibility that the party may have been punished by voters according to their distance to power: the party's vote share fell less where they were more vulnerable (both in their marginal seats or a distantly-placed second), yet in seats where they were close to winning (those they held safely or where their second-placed challenger stood a reasonable chance) the drop in their vote was greater.

As Table 2 demonstrates, the Liberal Democrats did particularly badly in seats where their challengers were the Conservatives. It would appear that the 'Black Widow Effect' for the junior coalition partner came true (Bale, 2012). Research by Green and Prosser (2015) shows that Labour were the largest beneficiary of the Liberal Democrat collapse in terms of votes across all constituencies in Britain, but that this predominantly served to take votes away from the Liberal Democrats in seats where they were being challenged by the Conservatives. Paradoxically, Labour's gain in Liberal Democrat votes was the Conservatives' gain in Liberal Democrat seats. Perhaps unsurprisingly the Liberal Democrats did particularly badly in the 13 seats they won in the 2005 general election but lost in the 2010 election, with the party's vote share in these seats dropping by an average of 21.7 percentage points. Considering that the party had lost these seats when they were relatively popular, rebuilding the party vote share in these seats in the post-coalition environment seems to have been a tall order.

Examinations of candidate incumbency have identified two key points at which it impacts party electoral performance: when an MP retires and when an MP defends their first election as the incumbent (Norris et al., 1992). In constituencies where a sitting MP retires, their party's vote share performs comparatively worse than in other constituencies where the party is standing as either incumbent or challenger. This phenomenon is known as the retirement slump (Cain et al., 1987) and is attributed to the loss of

the personal vote cultivated by the retiring MP over the course of their career. In the case of the Liberal Democrats, whose incumbency advantage over both the Conservatives and Labour is comparatively large (Smith, 2013), the loss of an incumbent MP through retirement is likely to hit the party's local vote share particularly hard. To examine whether MP retirement impacted Liberal Democrat local performance in 2015, Table 3 compares vote share in seats where the incumbent Liberal Democrat MP retired with seats where the incumbent Liberal Democrat MP fought again. A clear and significant difference can be found, with Liberal Democrat vote share falling by an average of 21.8 percentage points in seats where the sitting MP retired, compared with a fall of 14.3 percentage points where the sitting MP remained. This shows a clear difference: the party suffered particularly heavily where the personal vote of the retiring MP was lost, performing 7.5 percentage points worse on average.

This is not to suggest that Liberal Democrat MPs who retired would have won had they remained a candidate (the sheer scale of defeat for many Liberal Democrat MPs suggests that only Don Foster in Bath and Sir Menzies Campbell in Fife North East presented any realistic possibility). However, in some seats the damage might have been countered enough by an incumbent MP seeking

Table 3

Liberal Democrat vote share change for retiring MPs and first time MPs.

% Change 2010-2015

Retiring

MP stood down (11) -21.8

MP did not stand down (46) -14.3

Independent samples t-test 3.386** First-time incumbency

First-time incumbent MP (10) -10.9

MP elected before 2010(36) -16.7

Independent samples t-test -2.084*

Source: 2015 British Election Study Constituency Dataset. NB: *p < 0.05; **p < 0.01; ***p < 0.001.

C. Johnson, A. Middleton / Electoral Studies 43 (2016) 63-71 67

Table 4 Difference between first-time incumbent MPs and other seats, 2015 general election.

Liberal Democrat (%) Conservative (%) Labour(%)

First-time incumbent bonus over all candidates +5.8 First-time incumbents bonus over retiring MPs +10.9 +3.2 +2.8 ±0 +0.3

Source: 2015 British Election Study Constituency Dataset.

re-election to keep the seats in Liberal Democrat hands. Perhaps more importantly, the eight Liberal Democrat seats that the party retained might have changed hands had their candidates retired. Five of the remaining eight Liberal Democrat MPs' majorities are within the 7.5 per cent average drop; of the other three, both Mark Williams' majority in Ceredigion and Norman Lamb's majority in Norfolk North are only 8.2 percentage points.4 On this basis, the incumbency factor was not strong enough to stop the Liberal Democrats losing the majority of their seats, but appears strong enough to have stopped them losing even more.

The second point at which constituency incumbency impacts party performance is the 'sophomore surge' observed in incumbents defending their first election (Gelman and King, 1990). These candidates typically do better than both the party's longer-serving MPs and their yet-to-be-elected candidates; they spend their first term building and consolidating their personal vote in their constituency, with the pay-off at their first election as the incumbent. In the case of the Liberal Democrats, first time incumbents tend to do particularly well (Cutts, 2012) compared to their other candidates. This was reflected in 2015, as Table 3 indicates, the Liberal Democrats' ten first-time incumbents performed better (or less worse) than their longer serving counterparts. Where an MP was fighting their first election as the incumbent, the Liberal Democrat vote share fell by an average of 10.9 percentage points, compared to an average of 16.7 points for the party's MPs elected before 2010. Yet all first-time Liberal Democrat MPs who stood in 2015 lost their seat; the sophomore surge was insufficient to protect them. This failure of what is generally seen as a beneficial surge to the vote can be explained by looking at the types of constituencies that these first-time MPs were defending. All, bar one (Cambridge, where Julian Huppert was defending a 13.2 percentage point majority), were marginal. The first time incumbents, while insulated from the severity of the drop in the Liberal Democrat vote share, suffered more proportionally in their marginal constituencies than their longer-standing colleagues.

Table 4shows the electoral bonus for Liberal Democrat first-time incumbent MPs compared with all Liberal Democrat candidates, and retiring candidates for the party, and compares this with the bonus for first-time incumbent MPs in the Conservative and Labour parties. The Liberal Democrats' first-time incumbency bonus over all the party's candidates and also candidates where the Liberal Democrat MP had retired remains strong, with an advantage of 5.8 and 10.9 percentage points respectively. In short, the Liberal Democrats' first-time incumbency bonus was strong, and particularly strong relative to other parties. Of course, the Liberal Democrats' problem at the next election is that they currently have no first-time incumbents. For the Liberal Democrats' future electoral prospects, it would appear vital that their eight remaining incumbent MPs stand as candidates at the next general election to avoid the retirement slump. The Liberal Democrats' experience suggests that incumbency remains one source of electoral advantage for political parties, but as Smith (2013) concludes, its effect should not

4 The remaining Liberal Democrat MP, Tim Farron, has a majority of 18.3 percentage points in Westmoreland and Lonsdale.

be exaggerated. It is one factor of many in determining electoral advantage or disadvantage, and in the Liberal Democrats' case was overwhelmed by national factors. The next section examines such factors in more detail.

4. Individual predictors of Liberal Democrat support

The previous section used aggregate level data to examine the constituency campaigning strategy and incumbency benefits for the Liberal Democrats. While it highlights the continued importance of local campaigning and incumbency for the party, it does not explain the individual reasons behind the Liberal Democrats' collapse. This section addresses those, examining the explanations for challenges outlined for junior coalition parties in the comparative literature, using data from the pre- and post-election waves of the 2015 British Election Study to produce a binary logistic regression model. The dependent variable is a measure asking whether or not respondents voted Liberal Democrat in the 2015 general election. Given the large sample size of the 2015 British Election Study (N = 30027), it is possible to focus on a specific subset of cases and still have a large sample. This study focuses solely on those respondents that voted Liberal Democrat in the 2010 general election, and the reasons behind their support (or lack of it) for the party in the 2015 general election. This is done for two reasons. First, the sheer scale of the Liberal Democrats' defeat suggests that they failed to pick up many new voters. Indeed, their vote share did not increase in a single constituency. Second, explaining the party's collapse should examine why people stopped voting for the party, as opposed to looking at those people who did not vote for the party more broadly. Extensive analysis of this broader question has been conducted before (see Cutts, 2012), and understanding why parties lose support is an important question to be addressed (Johnston and Pattie, 2011). The scale of the 2015 British Election Study sample allows much more targeted analysis, still leaving a sample size of 2167 that can be included in the regression.

As examined in the comparative section, there are a range of explanations as to why junior coalition parties struggle in elections. Analysis of valence issue politics suggests that junior coalition partners find it hard to convince supporters of their influence in coalition (Duch et al., 2015; Dunphy and Bale, 2011; Muller and Strom, 1999). In this context, the Conservatives' dominance within the coalition suggests that the Liberal Democrats may not have appeared a credible governing party with their electorate. The model tests this in three different ways. First, the model incorporates a binary variable measuring whether (1) or not (0) an individual respondent thought that a party other than a Liberal Democrats were best on the most important issue to them. The valence model suggests that if voters feel that a party other than the Liberal Democrats to be the best on their most important issue, then they will be less likely to vote for them. Second, the model includes a variable asking whether respondents felt the economy was getting better or worse. If the Liberal Democrats' 2010 voters accorded them credit for being in government, they should be more likely to support them if they felt the economy was improving. Third, the model includes a similar variable asking whether

respondents felt that the education system was getting better or worse. Throughout the parliament, the Liberal Democrats stressed their positive influences on education policy: if their voters give them credit for this, they should be more likely to have supported them in 2015.

The Liberal Democrats' difficulty in convincing supporters of their influence within the coalition might have fostered a sense of betrayal amongst their supporters (Cutts and Russell, 2015; Dommett, 2013). To measure perception of betrayal, the model makes use of the variable 'anger at the Liberal Democrats'. Surveys of political attitudes increasingly include questions on emotions in relation to political action. In particular, anger is a decisive negative emotion (Lockerbie, 1993) and arises when an external actor can be blamed (Wagner, 2014). The sheer scale of the local, sub-national and national electoral collapse suggests an attribution of responsibility to the Liberal Democrats by their former supporters, and anger as an emotional response to betrayal is argued here to be the most appropriate variable to account for any breakdown in trust in the Liberal Democrats. As Cutts and Russell (2015: 80) argue, 'those voters ... who had supported the Liberal Democrats ... to keep the Conservatives out were left angry and betrayed'. It is expected that anger with the Liberal Democrats, coded (1), to be a significant and negative predictor of voting Liberal Democrat at the 2015 general election.

The importance of party leadership to a party's electoral fortunes has also been widely noted (Clarke et al., 2009; Stokes, 1992). Positive attitudes to Nick Clegg were a big indicator of support for the Liberal Democrats at the 2010 general election (Cutts, 2012; Middleton, 2015). However, he was perceived by the electorate to be the most incompetent of the three main leaders for most of the 2010 parliament (Denver, 2015). The model incorporates attitudes to party leaders, measured on a 0-10 thermometer scale. Whether respondents liked or disliked Nick Clegg is included in the model, as is the same measure for David Cameron and Ed Miliband.

Other factors highlighted as indicating levels of support for political parties are also included in the model. Partisan identification has previously been shown to be a key indicator of support (or lack of it) for political parties. Previous research has shown that if people develop an attachment to a party, they are more likely to support it (Holmberg, 2003). Respondents to the British Election Study were asked which political party, if any, they identify with. It is expected that those who identify with the Liberal Democrats, coded as (1), to be more likely to have supported the party at the 2015 general election.

The previous section demonstrated the continued importance of incumbency to the Liberal Democrats, so the model controls for whether or not the respondent was based in a constituency with a Liberal Democrat MP. Given the importance of local campaigning to the party (Cutts, 2014), the model also controls for whether or not a respondent recalls being contacted by the Liberal Democrats during the election campaign.

The model also accounts for attitudes on and perceptions of areas of British politics that might indicate support for the Liberal Democrats. Whether or not respondents think coalitions (1) are more effective than single-party government (0) is included. Given that the Liberal Democrats' only realistic hope of national office is in coalition with a major party, it is expected that support for coalitions as an effective form of government would make people more likely to support the Liberal Democrats. However, the strong contextual nature of local support for the Liberal Democrats suggests that support for coalitions might not play a significant part.

Given the Liberal Democrats' historical ability to win support based on their policy stance (Green, 2015), the model also includes key policy areas asked in the British Election Study that might indicate support for the party. First, given the prominence of debate

surrounding the European Union in British politics between 2010 and 2015, and the Liberal Democrats' strong support of the European Union (Goes, 2015), voting intention (1 = remain, 0 = leave) in a referendum on the European Union is included in the model. Second, the model also includes a variable to test the effect of the Liberal Democrats' policy shift on tuition fees. Prior to the 2010 general election, the Liberal Democrats made the most of a National Union of Students campaign to abolish tuition fees. Many of their parliamentary candidates, including party leader Nick Clegg, proudly held a card in front of the cameras pledging to abolish tuition fees. Whatever its immediate electoral benefit, it was fundamentally naive. The Conservatives' intention was always to implement the recommendations of the Browne review (increasing tuition fees to £9000), and the Liberal Democrats went back on their promise. However, a number of Liberal Democrat MPs rebelled and voted against the move. The model questions whether or not a rebelling MP (1) was more successful in holding on to Liberal Democrat voters than a non-rebelling MP (0). There is debate within the academic literature as to the effect of tuition fees on the Liberal Democrats' support (Curtice, 2015a; Cutts and Russell, 2015). Comparative evidence suggests that while voters do hold their MPs accountable, it is general, and MPs tend not to be rewarded for rebellion on a 'particularly salient single issue' (Vivyan and Wagner, 2012: 259). Whether or not rebelling MPs performed better will contribute further to this discussion. Finally, the model includes the respondent's age, sex, and whether or not they have a degree as socio-demographic control variables.

Table 5 shows the results of a logistic regression model predicting support for the Liberal Democrats in the 2015 general election amongst respondents who supported the party in the 2010 general election.5 Together the overall model worked well, with a pseudo R-square of 0.51. Firstly, the argument that the Liberal Democrats' collapse can be explained due to a perceived lack of competence is supported. Those that felt a party other than the Liberal Democrats were the best on the most important issue in British politics to them were less likely to support the party. The very small odds ratio highlights the strength of this negativity. The problems that the party had to be noticed in coalition appear to have done the party great damage in the 2015 general election. Alongside this, whether or not respondents felt that the economy or the education system got better under the coalition government made them no more likely to vote Liberal Democrat in 2015. This supports comparative arguments that junior coalition parties struggle to have influence in government (Duch et al., 2015; Dunphy and Bale, 2011; Muller and Strom, 1999).

Those who liked then Liberal Democrat leader Nick Clegg were more likely to support the Liberal Democrats; however the odds ratio's proximity to one suggests that his positive impact was much smaller than in 2010 (Cutts, 2012; Middleton, 2015). Attitudes to David Cameron had no significant effect on Liberal Democrat support, while those 2010 Liberal Democrat voters who liked Ed Miliband were less likely to support the Liberal Democrats at the 2015 general election.

The explanation that the Liberal Democrats lost support due a perception of betrayal is also strongly supported. Those who were angry with the Liberal Democrats were less likely to vote for them in the 2015 general election. Again, the small odds ratio highlights the strength of this negativity. The distrust that once benefitted the Liberal Democrats at the expense of the Labour and Conservative

5 Results are presented using odds ratios, which predict the probability of an event occurring. Higher than 1 and the effect is positive. Lower than 1 and the effect is negative. The greater the distance from 1 (either way), the greater the effect of the independent variable.

Table 5

Logistic regression detailing predictors of 2015 Liberal Democrat support.

Odds ratio

Competence and trust Other parties best on most important issue

The economy is improving Education system is improving Nick Clegg David Cameron Ed Miliband

Angry with the Liberal Democrats

Party identification, incumbency and campaigning Identifies with the Liberal Democrats Liberal Democrat incumbent Liberal Democrat contact before election

Attitudes and perceptions Coalitions more effective than single-party govt Support Britain's membership of the EU

Liberal Democrat MP rebelled on tuition fees Socio-demographic factors Age Sex Degree Constant

0.21"" 1.01 1.11 1.22*** 0.96 0.93** 0.35***

3.45*** 6.68*** 3.07***

1.11 1.68** 0.68

1.02*** 0.90 1.00 0.02***

Nagelkerke Pseudo R2 = 0.51 -2 Log Likelihood = 1616.779

Source: 2015 British Election Study (pre- and post-election waves). (Fieldhouse et al., 2015).

N = 2167 (Sample of 2010 Liberal Democrat voters).

Bold figures denote significant effects: *p < 0.05; **p < 0.01; ***p < 0.001.

Dependent variable: Voted Liberal Democrat in 2015 general election.

Source: 2015 British Election Study (pre- and post-election waves). (Fieldhouse et al., 2015).

parties has now engulfed them too. While they can claim a series of individual policy success and limitations on their Conservative coalition partners, they ultimately failed to distinguish themselves as the junior coalition partner. As Dommett (2013) argues, the effort to rebuild trust with voters will be a long and painful process.

The key lessons from the comparative literature on junior coalition parties is that they must appear competent and influential in government, maintain trust with their supporters and make the most of their leadership (Duch et al., 2015; Dunphy and Bale, 2011; Paun and Munro, 2013). On all three issues, the Liberal Democrats were unable to convince the electorate. Their perceived incompetence and lack of influence cost them heavily, those who were angry with the party deserted them and Nick Clegg was not the overwhelmingly positive contributor to Liberal Democrat support that he was in 2010. Together, these three issues did great damage to the Liberal Democrats in the 2015 general election.

What was the effect of other explanations of support for political parties? Party identification continues to be a strong indicator of party support. Those who identified with the Liberal Democrats were more likely to vote for the party in 2015. This was a strong indicator of Liberal Democrat support. There were statistically significant predictors relating to the 2010—2015 parliament. As expected, being in a Liberal Democrat incumbent constituency made a respondent more likely to support the Liberal Democrats in 2015. Respondents who recalled being contacted by the Liberal Democrats during the four weeks prior to the 2015 general election were also more likely to support the party.6 With strong respective odds ratios, incumbency and campaigning continues to benefit the party.

Attitudes to and perceptions of policies also had a mixed effect

6 It was suggested that we compared this effect with an earlier pre-election contact variable. This was done, but it did not change the results, and heavily reduced the number of respondents, so the model stands with the post-election variable.

on support for the Liberal Democrats. Those who would vote to stay in the European Union were also more likely to support the Liberal Democrats in the 2015 general election. However, whether or not a respondent preferred coalitions to single-party government made no difference. On the issue of tuition fees, if a respondent lived in a constituency with a Liberal Democrat MP that rebelled on tuition fees they were no more likely to support the party at the general election. Those MPs who rebelled on tuition fees do not appear to have reaped any electoral reward. While this does not settle the debate about the effect of tuition fees on the party more broadly (see Curtice, 2015a; Cutts and Russell, 2015), it supports evidence that rebelling on a single salient issue does not lead to subsequent electoral reward (Vivyan and Wagner, 2012). Finally, older respondents were more likely than younger respondents to support the Liberal Democrats. Whether a respondent was male or female made no difference, nor did whether or not they had a degree. In short, the Liberal Democrats failed to convince the electorate on the key issues of competence, distinctiveness and leadership.

5. Conclusion

The Liberal Democrats' performance in the 2015 general election provides an opportunity to examine the only case in the post-war period of a national junior coalition partner in British politics. The comparative literature on coalitions suggested that they faced a series of challenges: they needed to appear competent in government; they needed to make a positive and distinctive contribution to the country that was recognised by the electorate; and they needed to make the most of political leadership to promote their participation in coalition. Most evidence suggests that junior coalition parties struggle to overcome these challenges, although some positive examples exist (Bale, 2012; Paun and Munro, 2013). Alongside this, the Liberal Democrats' have enjoyed particular electoral success due to incumbency and constituency campaigning (Cutts, 2014; Smith, 2013). The 2015 general election provided the party with a huge, uphill challenge. Due to the fall in their national

popularity, the party embarked on a highly defensive strategy, focusing almost entirely on their incumbent seats.

Using aggregate-level and individual-level data, this article has examined the Liberal Democrats' success in combatting the challenges outlined above. At the aggregate level, the article finds that incumbency was unable to stem the far greater national tides hitting Liberal Democrat support. The first-time incumbency 'sophomore' bonus remained, and was stronger for the Liberal Democrats than the Conservative and Labour parties, but was also unable to protect them. While their first-time incumbent MPs were able to build up levels of support in their first five years in office, they lacked the time to build up substantial enough majorities to withstand the heavy fall in national vote share. However, without the incumbency factor the return of seats might have been even worse. The party did particularly badly where their MPs had stepped down. Future research should continue to examine the positive impact of incumbency for political parties. For the Liberal Democrats, it appears vital that their current small crop of MPs stay to contest their seats at the next general election.

Individual level analysis shows that the main challenges set out for the Liberal Democrats as a junior coalition party ultimately defeated them. First, they were unable to convince their 2010 voters that they were a credible coalition partner that would be competent in government. Second, they failed to make a contribution to government so distinctive from their Conservative partners that cut through to their electorate, and perception of betrayal as a consequence cost them badly. Third, their reliance on then leader Nick Clegg was a huge boost for the Liberal Democrats in 2010, but he was unable to have such a strong positive effect in 2015. Together, these three factors provide key explanations of the Liberal Democrats' collapse in the 2015 general election.

The continuing likelihood of hung parliaments following future general elections in Britain (Curtice, 2015b) means that cooperation between political parties is not off the table, and the Liberal Democrats' experience is a valuable lesson for analysis both of British politics and more comparative work on coalitions. In future, junior coalition parties in Britain must find a way to balance unity within coalition alongside trust and competence with the electorate. In many respects, the Liberal Democrats were well prepared for coalition. They produced alongside the Conservatives a detailed and binding coalition agreement, held important cabinet posts and implemented much of their manifesto. However, they ultimately failed to convince the electorate of their influence and their experience in government stands as a monumental electoral failure. Their local, sub-national and now national electoral base has been fundamentally weakened, and it will take a monumental effort to rebuild it.

Acknowledgements

The authors are very grateful to Dr Louise Thompson, Dr Alistair Clark, Dr Nick Randall and the members of the North East Research Development group, as well as to the journal's editorial team and anonymous referees, for their assistance and comments. Craig Johnson would also like to acknowledge the financial support of the Economic and Social Research Council, Grant Reference ES/ J500082/1. British Election Study data are publicly available, and leadership visits data can be collected from Dr Alia Middleton.

References

Axelrod, R., 1970. Conflict of Interest. Markham, Chicago.

Bale, T., 2012. The black Widow effect: why Britain's conservative—liberal democrat

coalition might have an unhappy ending. Parliam. Aff. 65 (2), 323—337. Bale, T., Webb, P., 2015. The Conservatives: their sweetest victory? Parliam. Aff. 68

(s1), 41—53.

Behr, R., 2014. Coalition has given us a narrower, angrier Tory party. The Guardian. http://www.theguardian.com/commentisfree/2014/nov/05/conservative-lib-dems-norman-baker?CMP=twt_gu (accessed 20 10.15.).

Bennett, A., 2014. Barely anyone knows what the Liberal Democrats' policies are. Huffington Post. http://www.huffingtonpost.co.uk/2014/10/13/libdems-nick-clegg-policies_n_5976254.html (accessed 03.10.15.).

Bennister, M., Heffernan, R., 2015. The limits to prime ministerial autonomy: cameron and the constraints of coalition. Parliam. Aff. 68 (1), 25-41.

Bolleyer, N., 2008. The organizational costs of public office. In: Deschouwer, K. (Ed.), New Parties in Government: In Power for the First Time. Routledge, London, pp. 17-44.

Boston, J., Bullock, D., 2012. Multi-party governance: managing the unity-distinctiveness dilemma in executive coalitions. Party Polit. 18 (3), 349-368.

Butler, D., Kavanagh, D., 1988. The British General Election of 1987. Macmillan, Basingstoke.

Cain, B., Ferejohn, J., Fiorina, M., 1987. The Personal Vote: Constituency service and Electoral Independence. Harvard University Press, Cambridge.

Carty, R.K., Eagles, D.M., 2005. Politics is Local: National Politics at the Grassroots Level. Oxford University Press, Oxford.

Clarke, H.D., Sanders, D., Stewart, M.C., Whiteley, P., 2009. Performance Politics and the British Voter. Cambridge University Press, Cambridge.

Coetzee, R., 2015. Causes and implications of the 2015 general election result, Inside Politics. http://inside-politics.org/2015/05/23/causes-and-implications-of-the-liberal-democrats-2015-election-result/ (accessed 03.10.15.).

Cornford, J., Dorling, D., 1997. Crooked margins and marginal seats. Journal of Elections. Public Opin. Parties 7 (1), 74-90.

Curtice, J., 2015a. The liberal Democrats and the 2015 election. J. Lib. Hist. 88 (3), 64-69.

Curtice, J., 2015b. A return to normality? How the electoral system operated. Parliam. Aff. 68 (s1), 25-40.

Cutts, D., 2012. Yet another false dawn? An examination of the liberal Democrats' performance in the 2010 general election. Br. J. Polit. Int. Relat. 14 (1), 96-114.

Cutts, D., 2014. Local elections as a 'stepping stone': does winning Council seats boost the liberal Democrats' performance in general elections? Polit. Stud. 62 (2), 361-380.

Cutts, D., Russell, A., 2015. From coalition to catastrophe: the electoral meltdown of the Liberal Democrats. Parliam. Aff. 68 (s1), 70-87.

Denver, D., 2015. The results: how Britain voted. Parliam. Aff. 68 (s1), 5-24.

Dommett, K., 2013. A miserable little compromise? Exploring liberal democrat fortunes in the Uk coalition. Polit. Q. 84 (2), 218-227.

Duch, R., Przepiorka, W., Stevenson, R.T., 2015. Responsibility attribution for collective decision makers. Am. J. Political Sci. 59 (2), 372-389.

Dunphy, R., Bale, T., 2011. The radical left in coalition government: towards a comparative measurement of success and failure. Party Polit. 17 (4), 488-504.

Evans, E., Sanderson-Nash, E., 2011. From sandals to suits: professionalisation, coalition and the liberal Democrats. Br. J. Polit. Int. Relat. 13 (4), 459-473.

Fieldhouse, E., Cutts, D., Russell, A., 2006. Neither North nor south: the liberal democrat performance in the 2005 general election. Journal of ElectionsD Public Opin. Parties 16 (1), 77-92.

Fieldhouse, E., Green, J., Schmitt, H., Evans, G., Eijk, C.v.d, 2015. British Election Study. British Election Study. http://www.britishelectionstudy.com/ (accessed 10 10.15.).

Fisher, J., Cutts, D., Fieldhouse, E., 2011. The electoral effectiveness of constituency campaigning in the 2010 British general election: the 'triumph' of Labour? Elect. Stud. 30 (4), 816-828.

Fisher, J., Fieldhouse, E., Johnston, R., Pattie, C., Cutts, D., 2015. Is all campaigning equally positive? the impact of district level campaigning on voter turnout at the 2010 British general election. Party Politics, EarlyView 22 (2), 215-226.

Fisher, S., Hobolt, S.B., 2010. Coalition government and electoral accountability. Elect. Stud. 29 (3), 358-369.

Gelman, A., King, G., 1990. Estimating incumbency advantage without bias. Am. J. Political Sci. 34 (4), 142-164.

Goes, E., 2015. The Liberal Democrats and the coalition: driven to the edge of Europe. Polit. Q. 86 (1), 93-100.

Green, J., 2015. Party and voter incentives at the crowded centre of British politics. Party Polit. 21 (1), 80-99.

Green, J., Hobolt, S.B., 2008. Owning the issue agenda: party strategies and vote choices in British elections. Elect. Stud. 27 (3), 460-476.

Green, J., Prosser, C., 2015. Learning the right lessons from Labour's 2015 defeat. Juncture 22 (2), 131-141.

Holbrook, T., 2002. Did the whistle-stop campaign matter. Political Sci. 35 (1), 59-66.

Holmberg, S., 2003. Are political parties necessary? Elect. Stud. 22 (2), 287-299.

Jacobson, G., 1987. The marginals never vanished: incumbency and competition in elections to the U.S. House of Representatives, 1952-82. Am. J. Polit. Sci. 31 (1), 125 -141.

Jacobson, G.C., 2006. Campaign spending effects in US senate elections: evidence from the national annenberg election survey. Elect. Stud. 25 (2), 195-226.

Johnson, C., 2014. The importance of local parties and incumbency to the electoral prospects of the Liberal Democrats. Politics 34 (3), 201-212.

Johnston, R.J., Pattie, C.J., 2011. Where did Labour's votes go? Valence politics and campaign effects at the 2010 British general election. Br. J. Polit. Int. Relat. 13 (3), 283-303.

Lockerbie, B., 1993. Economic dissatisfaction and political alienation in Western Europe. Eur. J. Political Res. 23 (3), 281 -293.

McEnhill, L., 2015. Unity and distinctiveness in Uk coalition government: lessons for Junior partners. Polit. Q. 86 (1), 101—109.

Middleton, A., 2015. The effectiveness of leader visits during the 2010 general election campaign. Br. J. Polit. Int. Relat. 17 (2), 244—259.

Monk, T., Lambert, H., 2015. Can any party win a majority? http://www.may2015. com/category/seat-calculator/ (accessed 25.06.15.).

Muller, W.C., Strom, K., 1999. Policy, Office or Voters? How Political Parties in Western Europe Make Hard Decisions. Cambridge University Press, Cambridge.

Norris, P., Vallance, E., Lovenduski, J., 1992. Do candidates make a difference? gender, race, ideology and incumbency. Parliam. Aff. 45 (2), 496—517.

Pack, M., 2015. Opinion polls. http://www.markpack.org.uk/opinion-polls/ (accessed 23.09.15.).

Pattie, C.J., Johnston, R.J., 2009. Still talking, but is anyone listening? the changing face of constituency campaigning in Britain, 1997—2005. Party Polit. 15 (4), 411—434.

Paun, A., Munro, R., 2013. Endgames: Lessons for the Lib Dems in the final phase of coalition Institute for Government. http://www.instituteforgovernment.org.uk/ publications/endgames-lessons-lib-dems-final-phase-coalition (accessed 20.06.15.).

Quinn, T., Bara, J., Bartle, J., 2011. The UK coalition agreement of 2010: who won?

Journal of elections, Public Opin. Parties 21 (2), 295—312.

Riker, W., 1962. The Theory of Political Coalitions. Yale University, New Haven.

Smith, T.H., 2013. Are you sitting comfortably? Estimating incumbency advantage in the UK: 1983—2010 — a research note. Elect. Stud. 32 (1), 167—173.

Stokes, D., 1992. Valence politics. In: Kavanagh, D. (Ed.), Electoral Politics. Clarendon Press, Oxford.

Vivyan, N., Wagner, M., 2012. Do voters reward rebellion? the electoral accountability of MPs in Britain. Eur. J. Political Res. 51 (2), 235—264.

Wagner, M., 2014. Fear and anger in great Britain: blame assignment and emotional reactions to the financial crisis. Polit. Behav. 36 (3), 683—703.

Whiteley, P., Clarke, H.D., Sanders, D., Stewart, M.C., 2013. Affluence, Austerity and Electoral Change in Britain. Cambridge University Press, Cambridge.

Whiteley, P.F., Seyd, P., 2003. How to win a landslide by really trying: the effects of local campaigning on voting in the 1997 British general election. Elect. Stud. 22 (2), 301—324.

YouGov, 2015. Party trust on most important issues. YouGov. https:// d25d2506sfb94s.cloudfront.net/cumulus_uploads/document/p897bk8f84/YG-Archives-Pol-Trackers-lssues(1)-Best-Party-on-lssue-130415.pdf (accessed 23.10.15.).