Poll puts Jeremy Corbyn on course for leadership victory – should we trust it?

A YouGov poll has put Jeremy Corbyn on course for a decisive victory in the Labour leadership contest. Tuesday’s poll, commissioned by The Times, has him 32 points ahead of his rival – 17 points up on a previous poll three weeks ago. But following a general election in which almost all polls were proved wrong, should we believe it? How strong is the evidence that Corbyn is ahead?

First, it is worth stressing that the election result should not discredit all polling. The big UK polling companies seriously misjudged the gap between the Conservatives and Labour – probably because of problems caused by differential turnout, as a preliminary investigation into what went wrong suggests – but the pre-election polls provided a good guide as to what was going to happen to the Liberal Democrat, Ukip and Green share of the vote, and in Scotland, too, they were spot on.

Trying to predict the outcome of an internal party election through polling is a much more tricky. A vast amount of information is available about the population as a whole, and hence finding a representative sample of them (the key to a successful poll) is relatively straightforward. But only about 400,000 people will be eligible to vote in the Labour leadership contest, their details are not available to outsiders, and information about who they are is more limited.

Several polling companies have published polls during the contest, releasing findings showing which leadership candidate the Labour-supporting respondents prefer. These polls have shown Andy Burnham coming top.

Facebook Twitter Pinterest

But the Labour-voting subsets are small, and Labour voters are not necessarily representative of the members, affiliated trade unionists and registered supporters who will actually decide the contest.

YouGov is the only polling company that has polled actual members. It is able to do this because it has a 500,000-strong panel of respondents who take part in its surveys, many of whom are eligible to vote in the leadership contest.

A poll last month showed Corbyn 17 points ahead of Burnham in the first round, and on course to win by 53% to 47% after second preferences were taken into account. Tuesday’s poll suggests Corbyn has extended his lead, and that he now has 53% of first-preference votes, meaning that he could win outright on the first ballot.

The first YouGov poll, which shocked Westminster, prompted some critics to argue that the YouGov panel could be unrepresentative because its members self-select: they are the kind of people who choose to take part in internet polling surveys.

But YouGov has been using this polling strategy very successfully for years, and at the time Peter Kellner, the YouGov president, pointed out in 2010 it predicted correctly that Ed Miliband would beat his brother for the leadership very narrowly.

Kellner says new information not available five years ago should make its sample even more representative of those voting in this election. Labour has released data about the age, gender and regional location of its members, and YouGov has weighted its figures accordingly. It has also chosen its sample to reflect the fact that almost half the people voting will be people who have signed up in some way since May.

Another adjustment has been made to reflect the fact that more than half the YouGov respondents who said they voted in the 2010 contest said they voted for Ed Miliband, even though in reality more individual party members voted for his brother. However, as Kellner says, this adjustment could understate Corbyn’s support. “If, as some people have suggested, more of David’s supporters have left the party since 2010 than Ed’s, that slight adjustment might be wrong; Mr Corbyn’s support could be slightly higher than we think,” he said.

It would be easier to discount the YouGov findings if they were contradicted by other evidence about Labour members’ voting intentions. But this is not the case, because other information confirms the impression that Corbyn is well ahead.

The leadership campaigns (unlike YouGov) do have access to full membership lists, and they have been contacting people by phone to try to drum up support. They have not published figures, but last month the New Statesman reported that two surveys conducted by Corbyn’s opponents showed him well ahead. More recently, the Daily Mirror reported on another “poll”, which seems to have been conducted by a rival leadership campaign, giving Corbyn a 20-point lead.

Corbyn also received more nominations from constituency Labour parties than any other candidate (152, against 111 for Burnham, 109 for Cooper and 18 for Liz Kendall). CLPs don’t ballot all their members before they decide who to nominate, and the decision is just taken by those who turn up to the relevant meeting, but these figures give some insight into whom activists want.

Social media, anecdotal evidence and the vast turnout at Corbyn’s rallies, where Corbyn has often found himself addressing crowds in the street because the meeting rooms cannot contain everyone who wants to hear him, also point to the Corbyn surge being real.

It is possible to read too much into this. In his anti-Corbyn blog on Monday, Alastair Campbell, Tony Blair’s former communications chief, recalled that Michael Foot thought he was on course to win the 1983 general election because so many people were turning up to hear him speak at meetings during the campaign. Foot thought those crowds were indicative of the public mood. They weren’t.

Equally, social media can be a poor guide to voting intention, as the SNP discovered last September. If Twitter were an accurate guide to public opinion, Scotland would now be negotiating independence.

Twitter and rally turnout are only really indicative of activist opinion. But party leadership contests, unlike general elections, are decided by activist opinion, which is yet another reason why the YouGov findings seem highly credible.