Labour are on course for a blow-out victory in the coming European elections, at least if the first poll of the contest, by Hanbury Strategy for the think tank Open Europe, is correct. They say that if the election were held tomorrow, Labour would finish miles ahead of the rest, with 37 per cent of the vote, the Conservatives in second place on 23 per cent, the Brexit party on 10 per cent, the Liberal Democrats on eight per cent, Ukip on seven per cent, and Change UK, the SNP and the Greens all on four per cent.
As ever, we should treat any single poll with a healthy degree of scepticism (last month’s shock Kantar poll was a good example of this; it showed a big Tory lead that looks to have since reverted to the mean, producing a small Labour lead, which is what we’d expect given the general polling average and the result of the Newport West by-election). However, because it is the first poll of the contest, the Hanbury poll is likely to be used as a baseline for expectations – so it is worth looking at in a little more detail.
If borne out, the poll would be remarkable: it would be the best performance by any political party since the introduction of a proportional election system for the European parliamentary elections, and would mean that Labour would be doing better at a European election than a Westminster one. This feels unlikely, as the general trend – which makes a lot of sense when you think about it – is that in first past the post elections people pick a big party that can actually win, whereas in proportional elections they feel more able to vote for whichever party they actually want.
The problems of political polling are well-advertised, but it is worth revisiting them to understand some of the reasons we should be iffy about this poll. One Conservative MP in a marginal seat has a good analogy for the broader problem of reading across from polls and elections in off-years: most people follow politics in the same way they follow international football, that is, with limited interest until a World Cup or a general election rolls around.
Now, yes, before a World Cup you will have a series of preconceptions about the various teams. I went into the last World Cup associating Uruguay with Luis Suarez, and assuming I’d be rooting for whoever they faced, while France’s long association with Arsenal meant I expected to be backing them in most of their games. But when it came to it I found Uruguay pretty likable, and France rather dull. A similar effect happens during elections.
Most people haven’t really heard of Nigel Farage’s new Brexit party, or Change UK, the new anti-Brexit force. As a result, they could do a great deal better or a great deal worse than these polls suggest.
But the more important thing speaks to the wider difficulties of opinion polling, which is that it is getting harder and harder for pollsters to talk to people who follow politics only casually, because increasingly small numbers of people have landline phones and it is more difficult to get people who follow politics casually to take opinion poll surveys.
For traditional polling, this could make it harder to accurately gauge the correct level of support because Labour support is increasingly concentrated among the young, people who live in cities and people with degrees, all of who are, for various reasons, more likely to have smartphones and to refrain from having or using a landline.
Hanbury surveys use smartphones to find respondents and conduct their surveys, which probably means that they are better at locating Labour supporters than traditional polling. But that cuts both ways: they are probably worse at locating supporters of the right-wing parties than a traditional pollster. Remember that while smartphone usage is near-universal among the under 45s, just half of the over 50s use a smartphone.
Hanbury tries to fix the problem through weighting, but I am dubious that it can be overcome. The 50 per cent of the over-50s with a smartphone are, obviously, different from the 50 per cent without, just as you can create a statistically representative sample of the whole country using just Ocado customers but they are still going to be a little bit odd because they are Ocado customers.
(To take the example further, notionally, Labour could produce representative surveys by polling its own half-a-million strong membership and weighting them accordingly – but a Labour member, aged over 50, who votes for the Conservative party doesn’t give us a very good grip on what the average Conservative voter aged over 50 thinks.)
Of course, if your sample has too many or too few Labour voters then everything else will be off. While this poll likely tells us a great deal about what Labour-inclined voters think, I wouldn’t use it to tell us much about what voters will actually do on 23 May. The same holds for pretty much any other poll. The one thing I would say is I wouldn’t use this as a yardstick for any of the parties’ performances, or draw any sweeping conclusions as a result.
Content from our partners
Related
