Like other polling companies, YouGov has started monitoring voting intentions ahead of next year’s elections to the European Parliament (EP). And, like other companies, we have found that Ukip’s support is significantly higher when people are asked for their EP vote than when they are asked how they would vote in a general election.
In the past week, Ukip’s EP support has been reported by ComRes to be 23% and by Survation to be 22%. YouGov has sought EP voting intentions twice, once using YouGov’s normal method, and once using the same method as ComRes and Survation. These put Ukip’s support at 17% and 19% respectively.
At this stage, with 17 months to go until the elections, voting figures are of little predictive use. In the past, Ukip has added very significantly to its support during the election campaign itself. All that can be said at this stage is that for Ukip to have support in the region of 20% long before the campaign has begun is excellent news for them. They could easily go on to win 25-30% next year, and very possibly top the poll.
However, there is a tricky methodological issue that needs to be addressed; and it was to illuminate this that YouGov has asked EP voting intention two different ways. To explain why this matters, we must delve into what happened at the last two EP elections in 2004 and 2009.
Back in 2004, when we first polled EP elections, we gave respondents a list of all the parties standing in the election, much as people would see on the ballot paper. This seems intuitively correct, but when the results came in we found our final survey, conducted on polling day for Sky News, overestimated the level of support for Ukip by four percentage points, and underestimated the Conservatives by five points. In other words, our total for the two right-of-centre parties was pretty accurate, but our figures for them individually lay outside the normal margin of error.
When it came to polling for the 2009 European elections therefore we took a different “two-stage” approach, similar to the way YouGov polls for Westminster elections. We reminded people that EP elections are fought under a proportional system, where minor parties do better than general elections, but still offered people the initial choice of the traditional big parties - Labour, the Conservatives, the Liberal Democrats and the nationalist parties in Scotland and Wales - and then the choice of 'other.' People who clicked other were then offered a further list including the Greens, Ukip, the BNP and all the other minor parties standing.
This may on the face of it seem odd or even unfair - Ukip, the Greens and the BNP all return MEPs at European elections, why should they be classed as 'other' parties? In practice however it worked, and produced far more accurate results. Indeed, YouGov’s final poll, for the Daily Telegraph, produced the most accurate figures of any polling company. Our average error was just one point and we got both the Tory and Ukip shares of the vote right to within two points. Again, the combined total was virtually spot on. (In fact, we overestimated Ukip’s support by one point and understated the Tories by two). And this was, don’t forget, an election in which Ukip came second, yet the two stage approach still accurately predicted their support.
In preparation for next year’s European elections we have tested both approaches. On two separate polls we asked voting intention as a single list, and as a two-stage question:
Using our usual two-stage prompting, Ukip are doing seven points better in the EP election than in the Westminster election. Using a single list, Ukip are doing eleven points better. This suggests prompting for Ukip in a question on European voting intention would give them a boost of about four points, entirely in line with our experience in 2004 and 2009.
What about the ComRes and Survation polls? Both of their surveys used the single-list method, with Ukip listed along with the main parties (as well as Greens and BNP). So their figures are comparable to YouGov’s 19% share. To put it another way, on a like-with like basis, we all put Ukip’s EP support at 21% plus or minus two.
What about the comparative uplift compared with the general election voting questions? ComRes’s survey did not provide comparative general election voting figures, but Survation did. Its general election figure for Ukip was 16%; so UKIP’s EP uplift was six points. However, the key thing here is that Survation also include Ukip in their main list when they ask their general election voting question. That is why their Ukip figures tend to be higher than those of other polling companies.
In other words, when Survation asks their general election and EP voting intentions in essentially the same way (single lists both times), Ukip receives a six point uplift; when YouGov asks both questions the same way as each other (two-part lists both times), we find a seven point uplift – much the same as Survation.
What, then, is the best approach going forward? YouGov’s view is that the difference in Ukip’s uplifts – seven points, using our tried-and-tested method, 11 points when switching to a single list for the EP voting intention – mean that the effect of prompting still works in much the same way it did in 2004 and 2009. So we reckon that our two-stage approach is still the more likely to produce accurate results.
However, it is worth stressing the near certainty that the actual EP election results will differ, probably massively, from these figures so long before the contest. Were I David Cameron, I wouldn’t breathe a sigh of relief about outperforming Ukip at the European elections just yet. Unless he manages to turn back the Ukip tide, starting with his speech on Europe on Friday, I would be surprised if the Conservatives remain ahead of Ukip when all the EP election votes are counted in June next year.