Sebastian Payne

Can we trust the Labour leadership polls?

Can we trust the Labour leadership polls?
Text settings

Is Jeremy Corbyn’s steaming ahead in the Labour leadership contest? Is he going to win on first preferences alone? The best quantitate answers we have come from two YouGov polls conducted for the Times — one in July and another last week. The first poll put Corbyn 17 points ahead of his nearest rival on first preferences and six points ahead in the final round. In the second poll, Corbyn was 31 points ahead on first preferences and 14 points in the final round.

Can these polls be trusted? Given that YouGov, like all the other pollsters, called the general election incorrectly, scepticism is rife in Westminster about any polling. On the World at One today, Peter Kellner of YouGov pointed out his company correctly called the Scottish referendum, the last London Mayoral election and 2010 Labour leadership contest and his pollsters are the best thing to judge the state of play:

'I don’t think anybody seriously says that Jeremy Corbyn, up to a week ago at any rate, was doing far better than anybody expected and was on first preferences out in the lead. What were the precise figures then, what are the precise figures now? I’m not sure.

'We felt our duty at YouGov, we’re the only organisation that can make an attempt at polling this electorate. This is an incredibly important moment in British political history and I think the best understanding we can have of that process and the people involved. It may it be perfect but it’s the best we have. I think it’s important to get it out into the public domain.'

Polling the Labour leadership electorate is not easy. The first poll had a sample size of ~1,000, the second one closer to 1,400. The July poll was weighted by YouGov to reflect the electorate at that time — approximately 390,000 people. The pollster used information in the public domain to take account of the makeup of party members who before May, new members who joined since the election and the union affiliates. YouGov also had a glimpse at party data for new full members by gender, age and region. The raw polling data was duly weighted for these groups, leading to the 53 per cent result for Corbyn in the final round.

But for the second poll, YouGov did not have data to take account of new, enlarged electorate of 610,000 people. They were only able to estimate the intentions of the new 220,000 people in each category, based upon the post-election signups. This lead to the 57 per cent lead for Corbyn in last week's poll.

Kelner said on World at One ‘a week ago, Jeremy Corbyn was in my view comfortably ahead. Is he now? Don’t know’. He is right to be unsure what is going on: the makeup of these 220,000 new members could follow what YouGov already knows. Or the makeup of the newest voters could be radically different. Plus, there has been a wave of high profile interventions from David Miliband, Gordon Brown, Neil Kinnock and Tony Blair. And again, we unsure whether this will push the electorate towards or further away from Corbyn. YouGov have done as much as possible to ensure their polls are accurate and reflect what's going on in Labour -- but it is a very difficult election to predict.