Sunday, October 19, 2014

Do leadership ratings matter more in the run-up to an election?

If the public find Ed Miliband so unimpressive, why is his party ahead in the polls? This question, pondered by Anthony Wells among others, is tricky. He calls it the “Ed Miliband paradox”, and says:

Given Labour are ahead now, I think the question is whether perceptions of the opposition and the choice of Prime Minister increase in importance as the election approaches and voting intention becomes less of a way of people indicating their opinion of the government, and more a choice between two alternatives.

I don’t know what will happen over the next seven months. But I can look at the past and see what happened to other leaders.

If perceptions of party leaders become more important in the run-up to an election. we would expect to see a swing to the party with the more popular leader. We can test this theory by looking at Ipsos MORI’s archive of decades of voting intention and leadership rating polls.

In the charts below, the solid lines are how many people said they would vote Labour or Conservative, and the dotted lines are how many people said they were satisfied with each party leader. Note that the final pre-election polls didn’t ask about leader ratings, so the dotted lines stop a bit short.

1979 election

Jim Callaghan’s lead over Margaret Thatcher held pretty constant for about a year and a half, while the Tory lead over Labour bobbed about with no real pattern. The theory would have predicted a Labour recovery, but instead we got the Winter of Discontent, which hit Callaghan’s ratings as well as Labour’s. After that, nothing much changed.

The theory gets no support from 1979, but arguably events got in the way.

1983 election

The Falklands war gave Thatcher and the Tories a big boost. In the year between then and the election, the lines move around a bit but nothing really changed.

The theory gets no support from 1983, although conceivably the war made voters care more about leadership a year early.

1987 election

For about a year, Neil Kinnock’s ratings were better than Thatcher’s, and his party tended to have a modest lead. But both of these things changed. The Tory vote recovery looks to have started a bit before Thatcher’s personal recovery, while Kinnock and Labour fell in tandem. There are no grounds for inferring that the change in leader ratings caused the change in voting intention.

The theory gets no support from 1987.

1992 election

This chart is a bit shorter, because John Major only became Prime Minister at the end of 1990. After his honeymoon – aided by the Gulf war – his personal ratings drifted down, but the parties’ positions didn’t change significantly. The polls before this election were badly wrong, of course, but I’m assuming the wrongness was consistent across this period.

The theory gets no support from 1992.

1997 election

Here we have Tony Blair consistently miles ahead of Major, but voting intention actually shifts a bit from Labour to Tory.

The theory gets no support from 1997.

2001 election

Here, Blair remains well ahead of William Hague, apart from during the fuel protests, after which he and his party recover their previous standing There is a slight trend from Labour to Tory, although this has pretty much stopped before the last year of the parliament.

The theory gets no support from 2001.

2005 election

Another shorter chart, as Michael Howard became Tory leader in late 2003. This is a closer contest, but there is little discernible trend in voting intention, despite Blair pulling ahead of Howard in satisfaction ratings.

The theory gets no support from 2005.

2010 election

This chart is also a bit shorter, because Ipsos MORI changed its methodology in June 2008. David Cameron is consistently ahead of Gordon Brown (although the gap narrows during the financial crisis). Despite this, in the final year there is a decent swing from Tory to Labour. As in 1987 and 1979, this movement in voting intention accompanies movement in leader ratings, but from this we can conclude nothing about causation.

The theory gets no support from 2010.

And that’s that. Eight elections, no support for the theory that perceptions of party leaders become more important for voting intention in the run-up to an election.

You can make various excuses and add caveats and say that many of these elections didn’t provide circumstances that made a good test of the theory. Fair enough; I’m not claiming to have disproved the theory, just to have shown that no evidence supports it. But if you want to hang onto some version of this theory, all you have to base it on is a hunch.

Recent history

All I note from this is that when Miliband’s ratings were better, so were Labour’s. In the last year and a half, both have fallen: it could be that one is driving the other, or it could be that other things are driving both.

I share the view that Miliband is painfully unimpressive. I share the view that leader ratings matter. But other things matter too, and maybe Labour has other strengths – or the Tories other weaknesses – to outweigh this. I don’t know. We’ll see.