We surveyed Labour members just after the 2015 General Election, and then ran a second survey in May this year so we could capture those who joined the party after the election....
By our reckoning, Labour’s leadership contest is going to be decided, for the most part, by less than 400,000 mainly middle-class university graduates. Nearly half of these members – unlike many of Labour’s voters – live in London and the South of England. Some 75 per cent of Labour members are ABC1 voters, and 57 per cent of them have a degree. Around 15 per cent live in London and 32 per cent live in other parts of the South of England. Only 28 per cent live in the party’s northern heartlands and 20 per cent in Wales and the Midlands, where (think, Nuneaton) any party wanting to win a general election desperately needs to win over voters.
There’s been some confusion about a new poll from Ipsos Mori (https://www.ipsos-mori.com/…/Over-half-say-Theresa-May-has-…).
The media has run with what Mori calls the “headline voting intention” figures, which give a 1-point lead to the Tories (Labour: 35%, Tories: 36%), but the firm also published a second set of figures for “all giving a voting intention”, which gives Labour a 5-point lead (Labour: 38%, Tories: 33% – see page 3 of its charts: https://www.ipsos-mori.com/…/…/Polls/pm-july-2016-charts.pdf).
So why the two different results? And which is more “accurate”? To find out, you have to understand how Mori came up with the numbers in the first place.
All pollsters use what’s called “weighting” to create a more representative sample than just the random people who answered the survey. They usually use three different kinds of weightings: demographics, political affiliation and turnout.
Demographic weighting means getting the right ratio of women to men, for example, or rich to poor. If the actual sample has too many women in it, then the weight of their answers will be reduced to compensate.
Political affiliation weighting is about compensating for, say, too many Labour voters in the sample. If a larger proportion of people say they voted Labour at the last election than actually did, their answers will be scaled down accordingly. The big problem with this is “false recall”, in which people pretend to have voted for a party (usually Labour) when they in fact didn’t vote, or pretend that they hadn’t voted when in fact they had.
The third weight is for turnout, which means that, if someone says they are 100% certain to vote, then their answers carry a 100% weight, while the answers of someone who says they’re only 50% likely to vote weighs only half as much. (For more on all these weightings, check out http://www.morningstaronline.co.uk/a-6722-Polling-headlines…)
Mori, like all reputable pollsters, publishes some of the data they collect in their surveys – the “tables” (https://www.ipsos-mori.com/…/…/Polls/pm-july-2016-tables.pdf). For the poll in question, tables 5 and 6 show us the results of the first two weightings – demographics and affiliation – giving Labour a 1-point lead (Labour: 26%, Tories: 25%). This figure could be described as a “snapshot” rather than a “prediction”.
If you apply the third weighting – turnout – then this moves more into predictive territory. The results can be seen in tables 1 and 2, and show Labour ahead by 5 points (Labour 38%, Tories: 33%). This is the figure that most pollsters call the “headline”, but it has been buried by the media despite appearing alongside what Mori now calls the “headline” figure in its charts.
Mori’s “headline” figure itself is arrived at by a more mysterious process, and we have to go back to 2015 to understand what it’s all about.
Pollsters messed up last year’s general election because they were caught out by the “shy Tory” phenomenon, in which many people who eventually voted for the Tories pretended they weren’t going to. However, because pollsters don’t like the idea that people might be lying to them, many of them deny that this was why they called it wrong. Instead, they insist that Labour voters overestimated their own likelihood of voting, and have now adapted their algorithms to take this dubious conclusion into account.
This is why Mori says it has now introduced a new “headline” definition, which only takes into account those people who say they’re very likely to vote. The exact methodology is unclear, but we do know that, from a total sample of 1021, only 640 people are actually included in this sub-sample. This is how it gets the Tory lead of 1 point, by excluding over a third of the original sample – including many who say they are likely but not certain to vote.
Mori justifies this by saying that applying this extra filter retrospectively makes the polls proceeding the general election more accurate. Which is fine, but there are many other ways you can weight the data to do the same thing – some logical, others not so much. Even Mori admit that this is unsatisfactory, calling it an “interim measure” while they carry out “an internal review into improving the accuracy of our polls”.
What this basically means is that Labour voting intention is suppressed in the “headline” figure, as Labour supporters appear to say they’re generally less likely to vote than Tories – unsurprising given the ongoing sabotage from many MPs and the attempted coup against Jeremy Corbyn. It’s also not clear whether a previous lack of enthusiasm, real or imagined, among Labour supporters to actually vote for Ed Miliband suppresses this figure still further. But this could all change – Labour supporters might actually become much more enthusiastic about voting as the Corbyn project develops, fatally undermining the logic behind this new methodology.
In conclusion, it looks like the official “headline” figure has less significance than the other result highlighted by Mori, although given the demonstrable and self-confessed flaws in the firm’s methodology it might be wise not to take any of its numbers too seriously. It is useful, however, to compare it to last month’s result (https://www.ipsos-mori.com/…/Immigration-is-now-the-top-iss…) which saw a 2-point lead for Labour (Labour: 36%, Tories: 34%).
So, according to Mori, Labour has more than doubled its lead despite the coup chaos, which blows a huge hole in the poisonous narrative that Corbyn is “unelectable”. The numbers, certainly in this case, say otherwise.
From Charley Allan. See his Tweet and Follow him on Twitter, here:
Red Labour's photo.
plus the 36%35% is based literally on 1 single voter (227/226) which seemingly gives the Tories an extra 1% of the vote... which seems bizarre to say the least.
What authentic version of the Labour Party is Corbyn fighting for? Presumably one that existed before he entered the House of Commons in 1983, given that he was one of the top ten Labour rebels even in the 1983-87 Parliament. It is, I think, from the foundation myth of the Labour Party as a movement of idealists and working people, finding solidarity in the struggle for their rights, that he derives his chief inspiration. Corbyn’s hero is Keir Hardie. Yet Hardie first ran for Parliament on the slogan ‘A vote for Hardie is a vote for Gladstone,’ successfully argued for the party to be called ‘Labour’ rather than ‘Socialist’ for fear of alienating potential supporters, and refused to back campaigns for the extension of the franchise because he was more anxious to secure practical reforms within the existing system than to fritter away his energy on constitutional struggles, even if it meant leaving some working men without the vote. There has never been a Labour Party that has not made compromises in the hope of improving its chances at electoral success. There have always been refuseniks too. But Labour at its most radical won in 1945 after spending more than a decade painstakingly stitching together a body of support; later serving as chancellor was Stafford ‘Austerity’ Cripps, who for most of the 1930s had been a stalwart of the extreme left.
The Corbynite refusal to compromise – compromise is Blairite revanchism – ignores the existence of vast tracts of common ground. As several observers have pointed out, John McDonnell’s ‘new economics’, with its emphasis on the use of strategic investment and higher wages to create the returns needed to reduce the deficit without further squeezes on spending, bears a striking resemblance to the policy advocated by Ed Balls. Though Ed Miliband’s reluctance to break with the language of austerity was frustrating, it was a (failed) strategic decision rather than an ideological choice: during last year’s general election campaign the Institute for Fiscal Studies pointed out that under Balls’s plans the party could have ended cuts in 2016 and still met its deficit targets. Margaret Hodge, who tabled the motion of no confidence, is another MP attacked as an unregenerate Blairite, but she spent the last Parliament as chair of the Public Accounts Committee pouring scorn on the PFI deals and outsourcing to the private sector that were Blair’s lasting domestic legacy, and was in large part responsible for publicising tax evasion as a national concern. Only last summer Angela Eagle was John McDonnell’s choice for deputy leader of the party.
Corbyn supporters increasingly resemble devout Brexiters, insistent on a golden future that is in contradiction to all known facts. They appear to believe that Jeremy Corbyn can win a general election without the support of his parliamentary colleagues, without the backing of the majority of Labour councillors, without support in the national media, without needing to demonstrate competence in his office, without even average personal approval ratings, without public confidence in his economic policies and without anything close to a Labour lead in the polls. The fear, based on current projections, must be that the British left will bury itself for good in 2020 (or earlier, if Theresa May chooses to junk the Fixed-Term Parliaments Act and hold a snap election). Those who talk of a ‘marathon and not a sprint’, of a long-term strategy of renewal that will bear fruit in 2025, do not seem to consider the possibility that Labour's position could get much worse at the next election: a marathon gets a hell of a lot longer if you lose your legs halfway through.
Well, no, thats not true, its based one a more/less likely curve, the validity of which is questioned, plus the 36%35% is based literally on 1 single voter (227/226) which seemingly gives the Tories an extra 1% of the vote... which seems bizarre to say the least.
Amused still by the vociferousness of the anti-corbyn camp. All this talk of 'competence' or 'electability'...
Three months ago, I stood on a boat with Nigel Farage as it was chased up and down the Thames by an enraged Bob Geldof on a pleasure cruiser. At the time, I felt reasonably certain that politics in 2016 was unlikely to throw up an occasion more surreal.
But perhaps I spoke too soon, because this afternoon Jeremy Corbyn called a press conference in London to announce that he’d secured the endorsement of Birmingham-born 1980s pop-reggae act UB40.
I went along, if only to prove to myself that it was really happening. Remarkably, it was. There, sitting beneath a large sign that read “UB4Corbyn”, were the Labour leader and five members of UB40. (There used to be others, but they left to form a separate band some years ago after an acrimonious split. I believe this is what is known in literary circles as “a metaphor”.)
“Thank you to UB40 for this incredible endorsement,” began Mr Corbyn, in all seriousness. The men from UB40 nodded graciously.