Think about the party leaders asking the British public to re-elect them this week. Are they nervous or confident? Trustworthy or shifty? Out of touch or possessed of the common touch? Decisive or ditherer? Clever or clodhopper? Commanding or geeky? Nice or nasty?
Although often derided as trivial (“shouldn’t we concentrate on the policies?”), the focus on political leaders’ personalities actually is important. Elections choose governments, members of which face difficult decisions under high-pressure conditions. To get their policies through, leaders need to build support within their government; win over major interests affected by the policies; and win agreement from foreign governments.
They therefore require not only clear policy visions, but also quickness of mind, force of personality, a capacity to persuade, a degree of empathy with others, a willingness to take painful decisions and, not least, considerable inner reserves of self-confidence and resilience if they are to be effective. The personal is certainly political.
So is the 2015 election going to be a beauty contest between the leaders? If it is, then all the parties need to worry. An Ipsos MORI poll conducted between March 8-11 2015 shows why. Respondents were asked to say whether they liked or disliked the Conservatives, Labour, the Liberal Democrats and UKIP, and their respective leaders. In every case, considerably more voters disliked the party leader than liked them.
Even so, some leaders might still prove an electoral asset to their parties. Disliked though they are, the poll shows that on the eve of the election campaign, David Cameron and Nigel Farage were not as unpopular as their parties. But Ed Miliband and Nick Clegg were not only pretty unpopular in their own right, but they were also noticeably less popular than their parties (considerably so in the case of Miliband and Labour).
Over the course of the campaign, things have changed somewhat, as Ed Miliband in particular has generally performed better than expected during the campaign. YouGov polling shows the uptick: whereas at the start of March 2015, 21% thought he was doing a good job as Labour leader, by mid-April that had increased to 36% - hardly a ringing endorsement, but a distinct improvement on his previously dismal ratings (the equivalent figures for David Cameron are 44% and 47% respectively, while for Nick Clegg the uptick was from 19% to 27%).
This is linked to voter intention. As the Ipsos-MORI results show, in almost every case (the exception being Miliband and Labour), far more of those who like a leader than of those who dislike him intend to vote for his party. It seems to be an open-and-shut case: the more personable a party’s leader, the more votes the party stands to win.
But is it really so simple? A major complication is that we do not form our opinions of the party leaders in isolation from our views of their parties, creating a huge chicken-and-egg problem. What comes first: opinions of the leaders’ personalities or support for their parties? If the former, evaluations of leaders’ personalities really matter in elections. If the latter, they might well not matter at all.
Like for like
How can we find out which it is? Their other merits notwithstanding, surveys which interview people at just one point in time are not well suited to answering this question.
To do so, we need surveys which re-interview the same individuals on at least two occasions (in the jargon, a panel survey). Because the same people are being interviewed each time, we can see how views change. This is important if we want to figure out which came first - the “like the leader” chicken or the “support the party” egg.
The 2015 British Election Study provides one such survey. It interviews a very large sample of voters repeatedly over time. The first wave took place in February/March 2014; people were re-interviewed for a second time in May/June 2014, and then for a third time in September/October 2014. A further wave was conducted in March 2015, just before the election, and others will take place during the campaign, and immediately after the ballot.
The “probability of voting for party x” questions in the third wave of the survey give an estimate of voter intentions and is shown in the figure below: respondents were asked to rate how likely they were to vote for each party, on a scale from 0 (no chance of voting for a party) to 10 (certain to vote for it). According to the survey, all four of the parties competing throughout Britain turned significant numbers of voters off entirely.
Around a quarter of respondents said they would not consider voting Labour. A third each said they would not vote Conservative or Lib Dem, and 40% said they would not vote UKIP.
How was this related to perceptions of the party leaders? The next figure shows what people said about how much they liked each leader in the second wave of the survey, about five months before they gave the assessments discussed above of how likely they were to vote for each party. This is also measured on an 11-pint scale: deeply disliked leaders are scored at 0, while those who are most strongly liked are scored at 10.
The distribution is remarkably similar: all four leaders were deeply disliked by a significant minority of voters (most strikingly so in the cases of Clegg and Farage), but not all were liked to the same extent: Cameron came out narrowly ahead, with an average score of 3.9 to Milliband’s 3.5 and Farage’s 3.4. Clegg, meanwhile, was the least popular, with an average of 3.0.
Not surprisingly, there is a close correlation between how much individuals report liking a party’s leader in May/June 2014 and their self-reported probability of voting for the party in September/October 2014. The more they liked the leader at the earlier date, the more likely they thought they would be to vote for his party at the later.
But remember that people started out as supporters of particular parties and opponents of others – and this will colour their views of the party leaders. In wave one of the survey, in February/March 2014, respondents were asked which, if any, of the parties they identified with. Not surprisingly, this correlates strongly with what they thought of the party leaders five months later. As the figure below demonstrates, each party’s supporters in February/March rated its leader higher in May/June – usually substantially so – than they rated any of the other leaders.
So, voters are more inclined to vote for a party if they already like the leader. But they will be more inclined to think well of the leader if they already support the party. So far, so confusing.
The big two … and the rest
A bit of statistical manipulation can help clarify things. Using something called “multivariate regression”, we can see just how much one factor affects another. Using this tool, we can see that 65% of what made people more or less likely consider Conservative in September/October can be explained by how much they liked Cameron five months earlier. For Miliband and Labour voting, the figure is 54%. How much they liked Nick Clegg explained 45% of their likelihood of voting Liberal Democrat. And liking Nigel Farage in Spring 2014 accounted for 61% of the likelihood of voting for his party in the autumn.
Applying the same method, we can see how much people’s party allegiances at the start of 2014 affect how likeable they found the leaders in Spring 2014. Prior party allegiance explains 45% of how likeable they found Cameron, and 38% of how likeable they found Miliband.
In comparison, party allegiance is a less effective guide to what people thought of Clegg and Farage (in part because their parties had fewer adherents than was the case for the other two leaders): around 18% of the variation in attitudes towards them can be explained by party allegiances at the start of the year.
So people who start out as supporters of a party are predisposed to think better of its leader than are people who do not initially support the party. But of course if they start out as supporters of a party, their chances of ending up voting for it also go up. If we want to see how much evaluations of the leaders affect support for their parties, therefore, we need to remove the influence of their initial party allegiances on both their views of the leaders and their eventual probability of voting for the party (if we don’t, we end up back in the chicken-and-egg world where we started).
A bit of statistical wizardry allows us to do this. When we do take into account the tendency for people who start out supporting a party to both like its leader and to be likely to vote for it, we find that leader evaluations do still influence the likelihood of voting for parties. But the effect of leader evaluations on the chances of voting for a party are much weaker when we control for prior partisanship than when we do not. For instance, when we control for prior partisanship, whether or not people like David Cameron goes from explaining 65% of whether they will consider voting Conservative to just 15%. The fall in the capacity of feelings about Ed Miliband to explain how likely people are to vote Labour is similar – down from 54% to 14%.
For Ed Miliband and David Cameron, therefore, persuading voters to like them personally will help their parties’ chances of winning more votes (good news so far for Ed Miliband, who seems to have had a good election campaign to date). But for both, this is dwarfed by the importance of prior support for their parties, both on perceptions of their personalities and on prospects of voting for their parties.
Crunch time for Clegg and Farage
Even when partisanship is taken into account, almost 30% of individuals’ chances of voting Liberal Democrat can be accounted for by whether they like Clegg (twice as much as for the Conservatives or Labour). And 43% of individuals’ perceived chances of voting UKIP can be accounted for by whether they like Farage.
Part of the issue here is that, for smaller parties, the leader bears a larger part of the burden of representing the party (and, in the absence of sustained media coverage of the party, voters are likely to have less firm views on what the party stands for).
So yes, personalities do matter. But they matter more for less prominent parties. The Lib Dems’ misfortune is that their leader is currently deeply unpopular. The challenge for UKIP, meanwhile, is that their leader is almost the only figure in the party who is easily recognised by voters.
Hard Evidence is a series of articles in which academics use research evidence to tackle the trickiest public policy questions.