sociology working papers · an analysis of english voters using the ... on the eve of the 7 may...

24
1 Sociology Working Papers Paper Number 2016-02 May 2016 Manor Road Building, Oxford OX1 3UQ Why did Labour fail to perform better in the 2015 General Election? An analysis of English voters using the 2015 British Election Study * Aaron Reeves 1,2, Martin McKee 2 , and David Stuckler 2,3 1 - International Inequalities Institute, London School of Economics and Political Science 2 - Department of Sociology, University of Oxford 3 - Department of Public Health & Policy, London School of Hygiene & Tropical Medicine Abstract Contrary to expectations, Labour failed to perform much better in the 2015 General Election than they did in 2010. Here, we evaluate the British Election Survey to examine some of the explanations for their 2015 performance. Negative perceptions of Ed Miliband had a modest negative effect on Labour’s vote share. There is no evidence Labour fared worse because they moved to the left, but they did suffer from a perceived lack of competence. Labour’s anti-austerity message led to some gains, but this was undermined by concerns about their competence and their perceived failures on the economy and immigration. Keywords: Competence; Economy; General Election 2015; Immigration; Labour party * Acknowledgements: We would like to thank Steve Fisher for comments on an earlier version of this paper. Corresponding Author: Aaaron Reeves, International Inequalities Institute, London School of Economics and Political Science, Houghton Street, London, WC2A 2AE.

Upload: lamdang

Post on 12-Jun-2018

213 views

Category:

Documents


0 download

TRANSCRIPT

1

Sociology Working Papers

Paper Number 2016-02

May 2016

Manor Road Building, Oxford OX1 3UQ

Why did Labour fail to perform better in the 2015 General Election?

An analysis of English voters using the 2015 British Election Study* †

Aaron Reeves1,2, Martin McKee2, and David Stuckler2,3

1 - International Inequalities Institute, London School of Economics and Political Science 2 - Department of Sociology, University of Oxford 3 - Department of Public Health & Policy, London School of Hygiene & Tropical Medicine

Abstract

Contrary to expectations, Labour failed to perform much better in the 2015 General

Election than they did in 2010. Here, we evaluate the British Election Survey to

examine some of the explanations for their 2015 performance. Negative perceptions of

Ed Miliband had a modest negative effect on Labour’s vote share. There is no evidence

Labour fared worse because they moved to the left, but they did suffer from a perceived

lack of competence. Labour’s anti-austerity message led to some gains, but this was

undermined by concerns about their competence and their perceived failures on the

economy and immigration.

Keywords: Competence; Economy; General Election 2015; Immigration; Labour party

* Acknowledgements: We would like to thank Steve Fisher for comments on an earlier version of this paper. † Corresponding Author: Aaaron Reeves, International Inequalities Institute, London School of Economics and Political Science, Houghton Street, London, WC2A 2AE.

2

Introduction

On the eve of the 7 May 2015 UK General Election, few pollsters and election scholars

anticipated that the Conservative party would win a majority of Parliamentary seats (Cowley

and Kavanagh 2016, Geddes and Tonge 2015). Polls in the lead up to the election consistently

indicated that the next parliament would, again, be formed by a coalition government. Results

from the face-to-face exit poll conducted on the day of the election were the first sign that

earlier predictions were wrong; surprising many by predicting a considerable Conservative

majority (Denver 2015). One Labour staffer recalled that on the night of election people waited

anxiously for the results from the exit poll and once they were announced: “I looked round and

saw rows of people all with their hands over their mouths… It was if the whole room was

witnessing some terrible tragedy unfold before their eyes” (Helm 2015). In now famous words,

a former LibDem leader said that if the result was true he would eat his hat. The exit poll was

incorrect but only in under-estimating the extent of the swing to the Conservative party. They

won 330 seats (gaining 24), while Labour lost 26 seats despite an increase in vote share of

approximately 1.5%.

In the aftermath of the election, there has been a series of studies investigating both why

the polls were so inaccurate and why Labour did not fare better. Turning to the first issue, there

is now compelling evidence that online polling data oversampled Labour supporters and that

younger Labour supporters were far less likely to actually turn out to vote on election day

(Council 2015).

However, the second issue – why Labour did not win more seats – remains puzzling.

Although the polls were biased towards Labour, they were not prima facie implausible. The

UK Coalition government had not delivered on a number of its election promises. They had

presided over a period of high unemployment and a slow economic recovery from the Great

Recession (Taylor 2014). Contrary to what they promised, they had failed to reduce the public

debt despite austerity programmes and had not curbed immigration (Travis 2016, Kirkup

2015). Further, the Coalition government implemented a range of unpopular policies – such as

the ‘Bedroom tax’ and working capability assessments – whilst reducing government spending,

leading to growing economic hardship (Browne and Elming 2015). Taken together, conditions

appeared favourable for the Labour party but they failed to make significant gains in England.

Numerous explanations have been proposed for Labour’s weak performance in England.

(Labour’s loss in Scotland is relatively straightforward) (Geddes and Tonge 2015). One set of

explanations proposes that Labour lost because it moved to the political left, alienating

3

moderate voters who had become disillusioned with the actions of the Liberal Democrats when

they were the minority party in the coalition government (Cowley and Kavanagh 2016).

Another, related argument, centres on the economy (Gamble 2015). Perceived fiscal

responsibility - and specifically the need to reduce the deficit - remained a salient issue for

many voters. Many people accepted the narrative, developed by the Conservative-led coalition

government, that Labour was responsible for the financial crisis by overspending, variously

presented in ways that resonated with individuals managing family finances as “failing to fix

the roof when the sun was shining” or “maxing out on the national credit card” (Cruddas 2015).

Labour initially left this narrative unchallenged as they were preoccupied by the party’s lengthy

leadership campaign after their defeat in 2010, allowing the view that they were fiscally

irresponsible to gain traction. When Labour did eventually offer an alternative narrative, they

were perceived as being defensive and in denial.

Leadership has also been offered as a potential explanation for Labour’s weak performance,

with particular focus on the perceived weakness of Labour’s leader, Ed Miliband (Fielding

2015, Clarke et al. 2011, Lenz and Lawson 2011). In the media, one commentator noted that

Miliband “suffered a pounding that was reminiscent of the bitterly personal attacks on Neil

Kinnock in 1992 and Michael Foot in 1983” (Greenslade 2015). Given this context, it is not

surprising that many in the media have claimed that it was Miliband’s failures as a leader that

affected Labour’s chances (Fielding 2015). Even among Labour party members, there has been

a strong perception that Labour lost because people did not think Miliband would ‘make a good

Prime Minister’. In short, the British public may have back Miliband’s policies ‘but not the

man’ (Kellner 2014).

Immigration was another key issue in this election (Geddes and Tonge 2015). Some have

speculated that Labour failed to address concerns about their record on immigration, with large

numbers of Eastern Europeans coming to the UK after European Union enlargement in 2004,

during the New Labour administration. Immigrants were often perceived as competitors for

jobs and housing by white working class communities that were already suffering stagnating

wages and long-term unemployment (Ford and Goodwin 2014). These communities formed a

large part of Labour’s core constituencies and so this influx of people rendered the party

especially vulnerable to anti-immigrant sentiment advanced, albeit often implicitly, by UKIP.

A final set of explanations emphasises competence rather than values or policies (Geddes

and Tonge 2015, Clarke et al. 2009). In this view, people prefer leaders who are able to get

things done even if they do not entirely agree with their policies or even their ideals (Whiteley

4

et al. 2013). Such people would prefer an effective Conservative candidate over an idealistic

but ineffective Labour candidate, even if they agreed with the Labour candidate’s policies.

People may be opposed to austerity, growing inequality, and the rising cost of living but they

may also believe that Labour will be unable to address these problems. This disconnect between

ideals and implementation can be described as the ‘competence gap’.

The extent to which these factors explain Labour’s performance at the 2015 general election

remains unclear. Previous analyses have often consulted focus groups, especially drawn from

marginal seats (Beckett 2016). To be sure, these studies can yield important insights, but they

are also partial and unrepresentative of the entire country. Moreover, they do not always tell a

consistent story. For example, some stress the perceived personal failures of Ed Miliband,

others highlight the failure to present a coherent and consistent message, while still others focus

on Labour’s perceived profligate spending prior to the economic crisis (Cruddas 2015, Hunter

2015). In short, there is little consensus about why Labour failed to gain more seats in England.

These are important questions and a comprehensive answer must include any relevant data

that exist. In this paper we split our analysis into four parts. First, we look at transitions and

issues; examining those who moved to and away from the Labour Party between the 2010 and

2015 elections and the issues they cite as being most important for the 2015 election. Next, we

look at leadership; testing whether Ed Miliband’s ‘likeability’ affected the outcome. Third, we

consider perceptions; exploring both whether Labour moved to the left and whether Labour

suffered from a competence gap (a disconnect between Labour’s policy promises and their

perceived ability to deliver on those policy promises). Finally, we unpack the big issues in the

2015 election; attempting to understand how the economy, immigration, and welfare

influenced voting behaviour.

To examine these questions we take advantage of the most recent wave of the British

Election Study. This survey provides quarterly data from February 2014 through to the election

in mid-2015, and includes data on how individuals voted in the 2010 general election. Using

this data, we aim to shed light on why Labour failed to perform better in the 2015 General

Election?

5

Methods

Source of Data

Data were taken from the British Election Study (BES), an online survey administered by

YouGov over six different sampling periods: Wave 1 - 20th February 2014 and 9th March

2014; Wave 2 between 22nd May 2014 and 25th June 2014; Wave 3 between 19th September

2014 and 17th October 2014; Wave 4 between 4th March 2015 and 30th March 2015; Wave 5

between 31st March 2015 and 6th May 2015; and Wave 6 (a post-election survey) between 8th

May 2015 and 26th May 2015.(Fieldhouse et al. 2015) Approximately, 30,027 respondents

participated in the survey in wave 6.

We use the longitudinal component of the data, using responses in previous waves to

understand voting decisions in wave 6, the post-election wave of the study. All analyses are

weighted using the full sample weights. While these data exhibit similar biases towards the

Labour party as other online surveys, these known biases do not impede our ability to

investigate determinants of vote choice and transition (internal validity).

Stratified Design

Initially, we disaggregated the data analysis to investigate determinants of vote choice among

three main groups: 1) non-voters in 2010, 2) people who voted for parties other than Labour in

2010, and 3) 2010 Labour voters. This enables us to then examine the determinants of

transitions between 2010 and 2015, such as leaving Labour for the Conservatives or for UKIP.

The analysis proceeded in four parts. In the first stage we mapped these transitions in

relation to issues perceived to be salient by voters. This was measured using questions asking

respondents what was the most salient issue to them: ‘As far as you're concerned, what is the

SINGLE MOST important issue facing the country at the present time?’ (emphasis in the

original). This was an open-ended question and respondents were able to provide any answer.

For example, one respondent said the ‘2 billion in welfare cuts’ while another said that it was

‘a benefit[s] reliant culture’. These responses were then coded using a machine learning

procedure based on previous surveys, supplemented by hand-coding. This coding procedure

generated 32 different responses including the economy, unemployment, the NHS,

immigration, among other issues. We only examine the largest 17 issues. To examine these

transitions we estimated the following equations:

6

Eq. 1a: Pr(LabourVotei, wave 6 | Did not votei, 2010) = α + βzMostImportantIssuei,wave 5 + β18agei,

wave 6 + β19agesqi, wave 6 + β20genderi wave 6 + εi

Eq. 1b: Pr(LabourVotei, wave 6 | Did not vote Labouri, 2010) = α + βzMostImportantIssuei,wave 5 +

β18agei, wave 6 + β19agesqi, wave 6 + β20genderi wave 6 + εi

Eq. 1c: Pr(LabourVotei, wave 6 | LabourVotei, 2010) = α + βzMostImportantIssuei,wave 5 + β18agei,

wave 6 + β19agesqi, wave 6 + β20genderi wave 6 + εi

Eq. 1d: Pr(ConservativeVotei, wave 6 | LabourVotei, 2010) = α + βzMostImportantIssuei,wave 5 +

β18agei, wave 6 + β19agesqi, wave 6 + β20genderi wave 6 + εi

Eq. 1e: Pr(UKIPVotei, wave 6 | LabourVotei, 2010) = α + βzMostImportantIssuei,wave 5 + β18agei,

wave 6 + β19agesqi, wave 6 + β20genderi wave 6 + εi.

Here i is individual, wave 6 is data collected after the election, wave 5 is data collected

between 31st March 2015 and 6th May 2015, and 2010 measures who people voted for at the

2010 election. βz is a vector of coefficients for each of the 17 categories in the question

pertaining to the ‘most important issue’. Age, agesq, and gender are socio-demographic

controls that are measured post-election but are assumed to be exogenous with respect to the

other predictors in the model. All analyses are restricted to people living in England and to

those who voted in the 2015 election.

In the second stage of the analysis, we examined whether the ‘likeability’ of Ed Miliband

and David Cameron was associated with voting decisions. Here, building on the earlier models,

we estimate two equations:

Eq. 2: Pr(LabourVotei, wave 6) = α + β1LikeMilibandi, wave 5 + β2LikeCameroni, wave 5 +

βzMostImportantIssuei,wave 5 + β20agei, wave 6 + β21agesqi, wave 6 + β22genderi wave 6 +

βmPartyVotei,2010 + εi.

These models are the same as above except that they do not restrict the sample to

respondents who voted for particular parties in 2010. Instead we include a direct measure of

the party people voted for (or not at all) in 2010 as a series of dummy variables (βm). We also

add a measure of whether respondents liked (or disliked) Miliband and Cameron (measured on

7

a scale from 0-10, 0 = strongly dislike and 10 = strongly like). In eq 2, we predict the probability

of voting Labour in 2016 given how much respondents (dis)like Miliband and Cameron in the

months before the election. We expect β1 > 0 and β2 < 0 but what is of particular interest here

is the size of the association between liking Miliband and voting Labour compared with the

size of the association between the measures of respondents’ most important issues. To make

this comparison we standardize all predictors by dividing them by two times their standard

deviation, so that they all variables (binary and scalar) are measured on the same scale (Gelman

2008, Gelman and Hill 2007). We then visualise the size of the coefficients using forest plots.

In a final step we estimate the impact of likeability n the election outcome by calculating the

shift in Labour’s vote share if Miliband had the same distribution of likeability as Cameron.

In the third stage of the analysis, we assessed perceptions of the Labour party, exploring

both whether people believed it had moved to the Left and whether the Labour party was

harmed by a perceived lack of competence on key issues. Initially, we measured public

perceptions of themselves along with the Labour and Conservative parties on a left-right scale

from 1 (Left) to 10 (Right). Next, we evaluated whether a ‘competence gap’ affected support

for Labour. The competence gap is the disconnect between promises and the probability of

success. In political terms, people may be sympathetic to Labour’s ideals but may not believe

it would be successful at achieving those ideals. For example, a respondent may agree that

inequality is the biggest problem facing society but they may not believe Labour are going to

address this problem successfully. The BES contains questions on immigration and inequality,

asking respondents whether they believe Labour will try to reduce a) immigration and b)

inequality and whether they Labour will be successful at reducing a) immigration and b)

inequality. Placing these measures in a 2x2 table and taking inequality as our example, we can

observe three main groups of people: 1) respondents who believe Labour will not try to nor

succeed at reducing inequality, 2) respondents who believe Labour will try but fail to reduce

inequality, and 3) respondents who believe Labour will try to and succeed at reducing

inequality. (Almost no one believes Labour will not try to reduce inequality but will succeed

anyway.) We compare the likelihood of supporting Labour given respondent’s answers to these

two questions.

In the fourth stage of the analysis, we examine three specific issues that appear to have been

particularly salient for English voters: the economy, immigration, and welfare. In the first case,

we consider the how peoples’ views on austerity influence their likelihood of voting Labour.

Second, we explore the impact on voting preference of their views on the economic benefits of

8

immigration. Third, we estimate the likelihood of people supporting Labour according to how

much they believe that welfare recipients receive and whether they feel that this is too much or

too little.

Results

At the most superficial level, it is clear that Labour lost the election because it attracted an

insufficient numbers of voters. The Party attracted about 16% of those who did not vote Labour

in 2010 (Figure 1). As shown in the diagram below, about three-quarters of those who voted

Labour in 2010 remained loyal to the Labour party. The remaining group, about one-quarter,

ultimately dictated the outcome of the General Election. We evaluate the salient issues

associated with each of the main transitions sequentially.

Figure 1: Who people voted for in 2015, conditional on whether they voted Labour in 2010

Note: The proportions estimated here are likely to be slight over-estimates because these data are based on a

YouGov survey which over-estimated support for Labour.

Salient issues associated with transitions to and away from the Labour Party

Labour’s largest gains in vote share came among those who did not vote in the 2010 General

Election but who voted in 2015. Among this group, 36% voted Labour in 2015 compared with

27% who voted Conservative.

9

Figures 2 and 3 show forest plots evaluating the associations of vote choice in 2015 among

non-voters in 2010 (figure 2) and non-Labour voters in 2010 (figure 3), with 17 issues reported

to be salient. As shown in figure 2, among non-voters who did vote in 2015, people concerned

about poverty and living standards were more likely to vote Labour in 2015.

Figure 2: People who did not vote in 2010, what made them more or less likely to vote Labour in 2015?

Notes: Size of the circle represents the number of people who believe that each item is the most important issue.

For example, 30% of respondents believe the economy is the most important issue while less than 1% thought

welfare fraud was the most important issue.

Figure 3: People who voted against Labour in 2010, what made them more or less likely to vote Labour in

2015?

Notes: Size of the circle represents the number of people who believe that each item is the most important issue.

For example, 30% of respondents believe the economy is the most important issue while less than 1% thought

welfare fraud was the most important issue.

10

Turning to the sub-group of those who voted for other parties in 2010, of those who

switched to Labour in 2015 from another party, the majority came from the Liberal Democrats

(76%), followed by the Conservatives (21%) and UKIP (3%). Members of this group were

considerably more likely to vote for Labour when their main salient issue was unemployment,

the NHS, housing, poverty and living standards, or social inequalities. In contrast, people

concerned about immigration, the economy in general, Europe, and terrorism were less likely

to transition to Labour.

Finally we looked at issues correlated with remaining Labour (figure 4). As shown in figure

4, this group tended to be concerned about the NHS, education, unemployment, and living

standards. Taken together, this analysis suggested that the NHS, poverty and living standards,

and social inequalities were areas which attracted people to Labour.

Figure 4: People who voted for Labour in 2010, what made them more or less likely to stay with Labour in

2015?

Notes: Size of the circle represents the number of people who believe that each item is the most important issue.

For example, 30% of respondents believe the economy is the most important issue while less than 1% thought

welfare fraud was the most important issue.

Having observed issues associated with transitions to Labour, we then looked at issues

corresponding to voting for different parties. Of those who voted Labour in 2010 but then left

Labour in 2015, most went to the Conservatives (40.2%), UKIP (28.4%), and Liberal

Democrats (14.4%) (figure 5). We focus this analysis on the two main transitions, to the

Conservative Party and to UKIP.

11

Figure 5: Who did people vote for if they left Labour between 2010 and 2015

Figures 6 and 7 shows forest plots depicting the statistical correlates of these transitions to

the Conservatives and UKIP, respectively. The main issues driving people from Labour to the

Conservatives were immigration, Europe, and the economy (figure 6) while immigration was

the sole issue driving people from Labour to UKIP (figure 7).

Figure 6: Why people left Labour (for the Conservatives) between 2010 and 2015?

Notes: Size of the circle represents the number of people who believe that each item is the most important issue.

For example, 30% of respondents believe the economy is the most important issue while less than 1% thought

welfare fraud was the most important issue.

12

Figure 7: Why people left Labour (for UKIP) between 2010 and 2015?

Notes: Size of the circle represents the number of people who believe that each item is the most important issue.

For example, 30% of respondents believe the economy is the most important issue while less than 1% thought

welfare fraud was the most important issue.

These findings are interesting because, although UKIP seeks to define itself as being

opposed to EU membership, and so should logically attract those who see this as the greatest

concern, in fact many with these concerns have moved to the Conservatives. In contrast, while

UKIP’s leaders have stressed that they are not opposed to immigration per se, with some

arguing that reduced immigration from the rest of the EU would free up opportunities to attract

migrants from the Commonwealth, it is this issue that is most important for those who move

from Labour to UKIP.

The ‘likeability’ of Ed Miliband

Did Labour lose the election because Ed Miliband was unpopular? Data from the BES do not

suggest that Miliband was clearly less popular than David Cameron. In fact, David Cameron

appears to have been more polarising that Ed Miliband. For example, ~25% of people ‘strongly

disliked’ David Cameron while ~22% said the same of Ed Miliband. At the same time, David

Cameron had slightly higher positive sentiment than Ed Miliband. This is important because

the likeability of leaders is strongly correlated with whether people vote for a given party. But

did it affect the outcome of the election? To test this we first explore the extent to which positive

sentiment about a leader predicts vote behaviour. Then, given these predictions, we estimate

13

the change in the vote share for the Labour party if Ed Miliband had the same likeability as

David Cameron (holding all else the same).

Figure 8 shows estimates of the probability of voting Labour in 2015 given how much

people liked or disliked Ed Miliband and/or David Cameron. These models include an

interaction term between how much people like the party leader and their view of the most

important issue. This interaction allows us to visualise the magnitude of the association

between likeability and voting given whether people also believe the economy, for example, is

the most important issue in the 2015 election (Figure 8).

Figure 8: How likeability of the leader affects voting behaviour in 2015

Notes: Size of the circle represents a measure of the precision of the estimate, more precise estimates have

larger circles. All variables are standardized following Gelman, i.e., dividing each variable by two times their

standard deviation, so that a one-unit increase is equivalent to a two standard deviation increase. For example,

for the like Miliband variable this is equivalent to moving from 4 (the mean) to 10 on a scale measured from 0

to 10 (10 = strongly like Miliband).

Likeability matters for both candidates, although people who dislike David Cameron

are more likely to vote Labour. (Results are similar for voting Conservative, except the

coefficients are reversed). People who express positive sentiments about a leader are also more

likely to vote for them. Yet, the size of the association between likeability and voting is striking,

and much larger than any of the associations between issues and voting Labour. Recall that

each of these variables is measured on the same scale and so that a one-unit change is the same

for each variable.

14

If likeability is equally important for both leaders, then how many people like a

particular leader is what matters. We therefore calculate the proportion of people we would

expect to vote Labour given how many people ‘like’ Miliband and the proportion of those

people (given our model predictions) who then vote Labour. For example, 22% of people

‘strongly dislike’ Miliband and, according to our model, only 8% of this group would vote

Labour. Given these figures we can estimate that, all else being equal, 1.76% of these 22%

voted Labour (22 * 0.08). Summing across all possible responses we can calculate the

proportion of voters that our model predicts are attributable to Miliband’s ‘likeability’. Using

this approach, our model predicts that, based on likeability alone, Miliband would win ~26.6%

of the vote in England. (He, in fact, won 31.6% and so, unsurprisingly, other factors also

influence the vote share of Labour in England). In contrast, if we ascribe to Miliband the

likeability of Cameron then we estimate that Labour’s vote share would have risen to ~28%; a

1.5 percentage point increase. This crude approach is almost certainly an over-estimate and so,

while 1.5 percentage points is not an insignificant figure, even this inflated figure would still

have been insufficient to win Labour the election. Importantly, this analysis does not tell us the

counterfactual, i.e., what would have happened had Labour chosen a different leader in 2010

instead of Ed Miliband nor can we reliably estimate what would have happened had Miliband

been more popular than Cameron. Crucially, what our analysis suggests is that ambivalent

feelings toward Ed Miliband are unlikely to have lost Labour the election.

Perceptions of the Labour party

We tested the hypothesis that Labour lost votes because it moved to the Left by evaluating

whether voters perceived a shift to the left on the political spectrum during the data collection

period.

Figure 9 charts respondents’ perceptions of Labour and Conservative parties on a left-

right scale in all six waves. As the figure shows, there were no clear shifts in how people viewed

the Labour Party, Conservative Party, or themselves on this ideological spectrum. Consistent

with previous findings, respondents place themselves closer to the Labour party on average

than the Conservative party even though the Conservative party clearly managed to increase

their share of the vote.

15

Figure 9: Where people place themselves and the two main parties on the left-right scale?

Notes: Models also adjust for age, age-squared, and gender.

Next we examined whether the public believed Labour was fiscally irresponsible. In

March 2015, people were asked how well different parties would perform in handling the

economy. Approximately 50% of people believed Labour would have been handling the

economy badly. In contrast, only 34% of people believed the Coalition government was

handling the economy badly. The survey data confirm that a majority of respondents blame

Labour for the increased national debt, a figure almost the same as those who, consistent with

the consensus among leading economists, blame the banks (here taken as shorthand for the

global financial system). Interestingly, although the debt burden increased markedly under the

Coalition government, the Conservatives have remained largely immune from blame.

Table 1. Who is responsible for the UK’s debt?

Responsible for the debt? No Yes

Labour 34% 55%

UK Banks 35% 54%

Conservatives 66% 23%

Notes: 11% respond ‘don’t know’.

Labour’s failure to perform better in 2015 could also be explained by a perceived lack

of competence. People may have sympathised with Labour’s policies but did not believe they

16

were able to make a difference in these policy areas. This ‘competence gap’ would be important

if voters prefer a competent leader, even if their policies are not perfectly aligned. We examine

this competence gap on two key issues: reducing immigration and inequality (Figure 10).

Around one quarter of the electorate thought that immigration was the most important issue in

the 2015 election. At the same time, the majority of voters did not believe Labour would try to

reduce immigration, and only 30% of this group were likely to vote for Labour. This is

unsurprising but there is another important story here. Among those who did think Labour

would try to reduce immigration there is a striking difference between those who thought

Labour would succeed in doing so and those who thought Labour would not succeed. For

example, 35% of those who believed Labour would try to reduce immigration but would not

succeed voted Labour in 2015. In contrast, more than 50% of those who believed Labour would

try and successfully reduce immigration voted Labour in 2015. We see a similar story on

inequality for the Labour party. People who believe Labour will try to reduce inequality but

will be unsuccessful are much less likely to vote Labour. Labour had a message that resonated

with many people but the party was less able to convince the electorate they could combine

policy with effectiveness.

Figure 10: Labour is most successful among those who understand their policies and believe they will be

effective

Notes: Models adjust for age, age-squared, sex, whether you voted Labour in 2010 and the variables measuring

the most important issues. Models are weighted. Size of the circle represents the number of people in each

category.

Unpacking the three big issues: the economy, welfare, and immigration

The economy and migration were the top two reasons people left Labour. Welfare, too,

was closely connected to Labour’s ‘excessive’ spending prior to the election. But this masks

some complexity in patterns of support for Labour.

17

The economy and austerity

Given Labour’s perceived fiscal irresponsibility, it is unsurprising that people who are most

concerned about the economy are less likely to support Labour in 2015. However, this is not

true for all people (Figure 11). Those who believe the cuts have gone too far are highly likely

to support Labour (~80%), even if they also believe the economy is the most important issue

for the election. In short, there were a group of voters who were concerned about the economy

and did not trust the Conservative party to fix these problems. This group were relatively small

but suggest that Labour failed to create a persuasive narrative that linked austerity to the

economic difficulties people had been facing.

Figure 11: Voters who believe the economy is the most important issue and that the cuts have gone too far are

highly likely to support Labour.

Notes: Models also adjust for age, age-squared, and gender.

Migration and economy: Making the economic case for immigration

Those people most concerned about immigration are less likely to support Labour, but

their concerns are not primarily economic. In contrast, people who are worried about issues

unrelated to immigration (such as the NHS or the economy) are more likely to vote Labour.

Importantly, this group – whose primary concern is not immigration – become more likely to

support Labour if they also believe that immigration is good for the economy (Figure 12). The

18

people most concerned about immigration are not moved in their support for Labour by this

economic argument, either positively or negatively. In short, making the economic case for

immigration may not have hurt Labour among people who are very concerned about

immigration but it may benefitted them among those who were not primarily concerned about

immigration. Importantly, the vast majority of people (77%) did not view immigration as the

most important issue in 2015.

Figure 12: Making the economic case for immigration

Notes: Models also adjust for age, age-squared, and gender.

Welfare: It is the perceptions of welfare rather than the actual amount that matters.

Labour’s perceived economic profligacy is rooted in their welfare spending. And so some have

argued that Labour’s failure to address the ‘scroungers’ debate did not help them win support

for their anti-austerity policies in England. Over 40% of people believe welfare is too generous

while only 20% believe it is too little. Among both groups, over 60% of people do not know

how much, on average, people actually receive. Uncertainty about how much people receive is

not necessarily the problem.

In fact, the actual amount received, in pounds, matters less than the overall perception

of welfare spending (Figures 13-14). Among those who voted Labour in 2010, people were

most likely to remain Labour supporters if they thought welfare spending was too low or about

right, regardless of the amount they believed people on welfare actually receive. Among people

who switched to Labour, people were most likely to make this switch if they thought welfare

was too low, again, regardless of the amount.

19

Figure 13: Probability of voting Labour in 2015 given you voted for Labour in 2010, by attitudes toward

welfare.

Notes: Models also adjust for age, age-squared, and gender.

Figure 14: Probability of switching to Labour in 2015 from other parties, given attitudes toward welfare.

Notes: Models also adjust for age, age-squared, and gender.

20

Conclusion

The Conservative party obtained an unexpected majority during the 2015 general election in

the UK (Cowley and Kavanagh 2016). Their victory was unanticipated because the polls

predicted another very close race but also because the polls were plausible given high profile

failings of the Coalition government on the economy, on immigration, and on deficit reduction.

We explore different explanations for Labour’s failure to win a larger share of the vote in

England by drawing on the British Election Study to explore transitions to and from Labour

between 2010 and 2015. We find that Labour gained support among people who believed that

the single most important issue facing the country was poverty and living standards, the NHS,

or unemployment. However, Labour lost support to the Conservatives among those most

concerned about the economy and to UKIP among those most concerned about immigration.

There are important limitations to our analysis. First, data collection only covers a

relatively short period of time and it is possible that Labour moved to the left before data

collection began, and so this data simply missed that transition. Second, voting behaviour in

2010 was collected based on a retrospective question asked between 2014 and 2015. This data

collection procedure has known measurement error, increasingly the likelihood that people

recall voting for the winning party (and forgetting voting for now unpopular parties, such as

the Liberal Democrats). Third, we cannot capture the timing of the transitions. This limits our

ability to isolate when or why people changed their vote. Fourth, we cannot measure

respondent’s attitudes or perceptions of the most important issues in 2010. In short, it is

impossible to tell whether people changed their attitudes and their votes or just their votes.

Fifth, there are known problems with online sampling procedures which tend to overestimate

support for Labour and which draw from a sample of highly politically engaged people. These

limitations raise concerns about the generalisability of our findings from these data to wider

population.

We found evidence that contradicts reports from smaller, focus group studies. One is

that negative perceptions of Ed Miliband had a small effect on Labour’s vote share but not as

large as others have suggested (Clarke et al. 2009). It is, however, difficult to disentangle

whether perceptions of Miliband would have been different if the print media coverage of

Miliband had been less negative and less personal. Approximately 95% of tabloid editorials

were anti-Labour in the run-up to the election, with many of those directly aimed at Miliband.

Media coverage of the Labour leader may have had some impact on the outcome of the election

and so Miliband the man may not have been a problem, rather it was the party and the policies

21

he represented. In short, would Miliband have fared better if he and David Cameron had

switched parties and policies? Our analysis suggests that this would have yielded relatively

little change in vote share.

We also found consistent evidence that perceptions of the Labour party clearly mattered

(Gamble 2015), albeit not in ways that are commonly argued. For example, we find no evidence

that Labour lost support because they moved to the left. In fact, across the study period there

was almost no movement in Labour’s position on the left-right ideological spectrum. Yet,

Labour’s perceived fiscal extravagance did matter (Gamble 2015). In the public’s imagination,

Labour remained as responsible for the financial crisis as the bankers, a group who are still

incredibly unpopular in the UK. People believed that Labour were responsible for the debt and

that they would have handled the economy badly if they had been in power. Labour also

suffered from a competence gap; a perceived disconnect between the policy promises of the

party and their ability to deliver on those promises. Our results suggest that people who

believed Labour would try but fail to reduce inequality were as likely to vote Labour as those

who believed would not try nor succeed at reducing inequality. In short, both promises and

performance would be necessary for a Labour victory.

Issues played a key role – particularly around the economy, welfare, and immigration.

However, the common assertion that the Conservatives gained support among people

concerned about the economy obfuscates important complexity. Our data indicate that Labour

did lose votes in those concerned about the economy and immigration. However, those who

were concerned about the economy and believed cuts had gone too far were more likely to

support Labour. Additionally, people who did not think immigration was the ‘most important’

issue could be persuaded by the economic case for migrants. Finally, people who believed that

social security was inadequate are more likely to vote Labour, regardless of the amount they

believe people receive. In short, the link between support for Labour and these three key issues

are complex and are rarely unidirectional.

Taken together our results help account for why Labour failed to perform better given

otherwise favourable political and economic conditions for a challenger party. Taken together,

our results indicate that Labour’s shortcomings were an inability: 1) to build a strong economic

narrative that countered the Conservatives story of deficit-reduction, 2) to convince the

electorate they were able to deliver on their election promises, 3) to address tensions around

immigration, particularly the economic case for immigration in communities experiencing both

the costs and benefits of migrants, and 4) to develop a competing case for social security, one

22

that was grounded in the lived experience of everyday people affected by the Great Recession

and subsequent austerity. Addressing these challenges may not have changed the outcome of

the election, but they offer a partial explanation for why Labour did not perform better in 2015

and why they did not make some gains against a first-term incumbent party that had failed to

achieve many of their original manifesto promises.

23

References

Beckett, D. M. 2016. Learning the lessons from defeat: Taskforce report. London: The

Labour Party.

Browne, J. and Elming, W. 2015. The Effect of the Coalition’s Tax and Benefit Changes on

Household Incomes and Work Incentives. In: Studies, I. F. F. (ed.). London: IFS.

Clarke, H., Sanders, D., Stewart, M. and Whiteley, P. 2011. Valence Politics and Electoral

Choice in Britain, 2010. Journal of Elections, Public Opinion & Parties, 21, 237-253.

Clarke, H. D., Kornberg, A. and Scotto, T. J. 2009. Making political choices : Canada and

the United States, Toronto, University of Toronto Press.

Council, B. P. 2015. General Election: 7 May 2015 [Online]. Available:

http://www.britishpollingcouncil.org/general-election-7-may-2015/ [Accessed

11/03/2016.

Cowley, P. and Kavanagh, D. 2016. The British General Election of 2015, London, Palgrave

Macmillan.

Cruddas, J. 2015. Labour lost because voters believed it was anti-austerity [Online].

Available: http://labourlist.org/2015/08/labour-lost-because-voters-believed-it-was-

anti-austerity/ [Accessed 5 Aug 2015.

Denver, D. 2015. The Results: How Britain Voted. Parliamentary Affairs, 68, 5-24.

Fieldhouse, E., Green, J., Evans, G., Schmitt, H. and Eijk, C. V. D. 2015. British Election

Study: Longitudinal Waves 1-6 In: Bes (ed.).

Fielding, S. 2015. ‘Hell, No!’ Labour's Campaign: The Correct Diagnosis but the Wrong

Doctor? Parliamentary Affairs, 68, 54-69.

Ford, R. and Goodwin, M. J. 2014. Revolt on the Right: Explaining Support for the Radical

Right in Britain, London, Routledge.

Gamble, A. 2015. The Economy. Parliamentary Affairs, 68, 154-167.

Geddes, A. and Tonge, J. (eds.) 2015. Britain Votes 2015, Oxford: Oxford University Press.

Gelman, A. 2008. Scaling regression inputs by dividing by two standard deviations. Statistics

in Medicine, 27, 2865-2873.

24

Gelman, A. and Hill, J. 2007. Data analysis using regression and multilevel/hierarchical

models, Cambridge, Cambridge University Press.

Greenslade, R. 2015. Yes, rightwing newspaper coverage did cause Ed Miliband's downfall.

Guardian, 11 May 2015.

Helm, T. 2015. After the exit poll, a tsunami raged across the political map. The Observer, 10

May 2015.

Hunter, P. 2015. Red alert: why Labour lost and what needs to change? In: Institute, T. S.

(ed.).

Kellner, P. 2014. Miliband: we back his policies but not the man [Online]. Available:

https://yougov.co.uk/news/2014/05/12/miliband-we-back-his-policies-not-man/

[Accessed 11/03/2016.

Kirkup, J. 2015. David Cameron and the national debt monster in three charts. The Daily

Telegraph, 12 Jan 2015.

Lenz, G. S. and Lawson, C. 2011. Looking the Part: Television Leads Less Informed Citizens

to Vote Based on Candidates' Appearance. American Journal of Political Science, 55,

574-589.

Taylor, J. B. 2014. The Role of Policy in the Great Recession and the Weak Recovery.

American Economic Review, 104, 61-66.

Travis, A. 2016. Net migration to UK higher than when coalition took office. Guardian, 26

Feb 2016.

Whiteley, P., Clarke, H. D., Sanders, D. and Stewart, M. C. 2013. Affluence, austerity and

electoral change in Britain, New York, Cambridge University Press.