Pages

How looking at bad polls can show Labour how to win the next election

By Leighton Vaughan Williams Professor of Economics and Finance & Director, Political Forecasting Unit at Nottingham Trent University Source
The electoral phenomenon that left David Cameron smiling will have huge implications for future battles. Ki Price/EPA

We have all now heard a multitude of theories to explain how and why the Conservative Party secured a surprise overall majority in the UK election, winning almost a hundred more seats than Labour. These theories are often centred around the ideas of leadership, economic competence and the ability to attract the “aspirational” voters of “middle England”. This conclusion would tell the Labour Party something very important about the ground their next leader will need to fight on, and indeed who that leader should be. But is this the whole picture?
Much mockery has been directed at pollsters following the election (and not only in the UK), but this question is where the opinion polls can in fact tell us something important. Discussion has been vigorous about weaknesses in survey design, but that in itself does not seem sufficient to explain the huge disparity in what actually happened at the polling stations (Tories ahead of Labour by 6.5%) and what happened in the polls (essentially tied).
We have heard about the phenomenon of the “shy Tory”, but I would argue that a big part of the reason for this disparity is what I term the “lethargic Labour” effect; the differential tendency of Labour supporters to stay at home compared to Tory supporters. “Lethargic” is a term I choose carefully for its association with apathy and general passivity, and it is a factor which I believe has huge implications for political strategy in the years ahead.

Lethal lethargy

To understand this, it is instructive to look to the now famous exit poll, which was conducted at polling stations with people who had actually voted. This was much more accurate than the other published polls, including those conducted on election day over the telephone or online, and showed a much lower Labour share of the vote. A dominant explanation for this disparity is that there was a significant difference in the number of those who declared they had voted Labour or that they would vote Labour and those who actually did vote.
This “lethargic Labour” effect is quite different to that of the “shy Tory” which was first advanced to help explain the polling meltdown of the 1992 general election, when the Conservative Party were similarly under-estimated in the opinion polls. The idea is that Tories in particular are shy of revealing their voting intention to pollsters. Yet in 2015 we would expect, if this were a real effect, to have seen it displayed in under-performance by the Tories in telephone polls compared to the relatively more anonymous setting of online polls. There is no such evidence, if anything the reverse being the case for much of the polling cycle.
Now, clearly, the idea of “lethargic Labour” supporters does not offer the whole explanation for the Tory victory. There is also a historically well-established tendency for a late swing on the day to incumbents, which cannot be blamed on the raw polls, but is sometimes built into poll-based forecasting models which can account for some of the differential. There is additionally late tactical switching to consider, where a voter, when face to face with an actual ballot paper, chooses to hinder their least preferred candidate.

Calling the odds, and calling the shots? cellanrCC BY-SA
Click to enlarge

As I have noted elsewhere, the betting markets significantly out-performed the polls and also a sophisticated model based on those polls which allowed for late swing, but they beat the latter somewhat less comprehensively, at least at constituency level. At national aggregated level, the betting markets beat both very convincingly, though the swing-adjusted polls performed rather better than the published polls.

Future-proofing the polls

So what does this tell us? It suggests that there was indeed a late swing to the Tories, which was picked up in the betting markets in advance of the actual poll. But the scale of the victory (at least compared to general expectations) was not fully anticipated by any established forecasting methodology. This suggests that there was an extra variable, which was not properly factored in by any forecasting methodology. This extra variable, I suggest, is the “lethargic Labour” supporter, who existed in far greater numbers than was generally supposed.
To the extent that this explanation of the Tory majority prevails, it has profound implications for the strategy of the Labour Party over the next few years in seeking to win office.
It tells us that if Labour is to win the next election, a strategy will have to be devised which motivates their own supporters to actually turn out and vote. If the party can’t do that, it doesn’t really matter how effective the leader is, how economically competent they are seen to be, how well they appeal to the “aspirational” voter. It is very unlikely that they will be able to win.