Tonight’s YouGov poll for the Sun has topline figures of CON 32%, LAB 34%, LDEM 6%, UKIP 15%, GRN 7%. All looks normal by YouGov’s recent standards (the two polls showing UKIP up at 17% and 18% that we saw immediately after the attacks in Paris seem to have gone away again – perhaps it was a Paris effect, perhaps it was just a random blip).

A word about the debates and polling on them. If the debates go ahead, they obviously have the potential to have a big impact upon public opinion and the election result – we saw that in 2010 and the boost they gave the Liberal Democrats. Think in particular of the potential impact for Nigel Farage or (if she ends up being included) Natalie Bennett – it’s the challenger candidates they are a real opportunity for. If the debates end up going ahead, but without David Cameron, that will probably have an impact too. It would really highlight his non-participation.

I’d be more wary of whether there will actually be much impact if they end up NOT going ahead, if it ends with them not happening. We’ve seen several polls on how people think it’s bad that David Cameron might not take part and so on… but regular readers will be very familiar with the flow of polls showing that people don’t like David Cameron doing this and don’t like Ed Miliband doing that which make not an iota of difference to voting intention or to people’s attitudes towards the leaders. Most Westminster stories make no appreciable difference to anything in terms of public opinion. A lot of the time, most people probably don’t even notice the story – they give an opinion because we pollsters have prodded them with a question, it doesn’t mean they care.

In this case, I think people are at least aware of what’s happening, it’s not a pure Westminster bubble story. Without giving any information about what stance the leaders had taken YouGov asked a question on Sunday on whether people thought each leader did or did not want a debate. Obviously lots of people said don’t know, but on balance people thought Miliband, Clegg and Farage did want a debate, and by 51% to 22% people thought Cameron was trying to avoid one. So, it has got through to the public.

Whether it makes any difference to their opinions of David Cameron is a different matter though. The measure to look for here – if it happens – won’t be the sort of questions saying “Does blocking the debates make Cameron look bad?”, “Is Cameron a coward?” and so on. You’ll get people cynical about politicians or hostile to David Cameron saying yes anyway. It will be whether David Cameron’s actual ratings go down – does not taking part in a debate damage his lead on being Prime Ministerial? Does it make his approval ratings worse and make him look less of a leader? Does it make his (already bad) ratings on being out of touch worse? Does it damage the Conservative party’s voting intention at all? Those are the things that count and the things to watch, if they don’t go down then, frankly, the bullet will have been dodged.

113 Responses to “Latest YouGov poll, and some thoughts on debate polling”

1 2 3
  1. Unicorn,

    It might seem that way, but a MoE interval doesn’t guarantee anything about the degree of variance within that interval.

  2. I have to say I’m quite surprised that we’re seeing no rise in Green support in relation to their increasing coverage. UKIP seems to do well with increased media coverage, but thus far it’s seeming to make no difference to the Greens.

  3. @ Bill Patrick

    “..but a MoE interval doesn’t guarantee anything about the degree of variance within that interval.”

    Not sure what you are suggesting here. ‘MoE interval’ is just an informal expression for the confidence interval (normally 95% in polling circles) that is drawn around a published polling figure. It is directly related to the sampling variance, subject to distributional assumptions.

    If the variance is lower than assumed, then so too is the MoE.

  4. David,

    Much of their rise has been thanks to people my age on social media, who don’t read papers or watch TV news. They divide their time between evangelising the Greens and shouting at me.

  5. We should not assume folk are voting SNP on the basis of the economics of independence. I’d say it’s more about “standing up for Scotland” at Westminster. Labour has been rendered toxic for many because it got too close to the Tories during the #indyref. It should be crystal clear Indy is not happening but Scottish interests now have higher salience among those energised to vote. Those who got off their backsides in September to vote No will not need to bother in May as Indy is not on the ballot paper but the energised 45 remain keen to assert their unhappiness with Westminster. Voter registration shouldn’t be a problem either!

  6. To be clear talking about oil prices is fighting the last war. This is all about standing up for Scotland, which is why Murphy is trying to show he is the boss.

  7. Alec

    I wondered if the comment was from you on Wings – I wasn’t sure if you were still banned. Oddly enough Stu always seems more balanced in commenting on his own polls than he is on other people’s, possibly because he devotes more time to understanding them rather than looking for a pre-determined result. So hopefully there won’t be too much spin (he’s currently having a mini-breakdown on discovering that the Scots are no more liberal than the rest of the UK).

    In this case he may have actually believed what he said was true. Certainly there’s enough people around claiming that the Conservatives are (and ever more will be) dominant in England and only these pesky Celts allow Labour to impose their alien ideas on the trusty yeomen. It’s the whole idea behind all this EVEL nonsense for example. So he may have just been following received opinion.

    Anyway the current spin for the SNP should be different from the pre-Referendum one, as the bogey-man is no longer an all-Conservative government at Westminster which only Yes can free you from. Instead voters have to be tempted with the prospect of a powerful SNP bloc pledged to get the best deal for Scotland. For that you require Labour and Tory to be roughly equal so the desirable balance of power can be achieved. If a Conservative government is inevitable it hardly matters who you send to Westminster.

    As to expecting political repercussions for the SNP from the oil price situation – that simply isn’t going to happen. Indeed they are pointing to current job losses as illustrating how the union isn’t the promised haven that it was said to be. As I keep on pointing out, voters regard oil as still valuable and price fluctuations as temporary. And pointing to the structural instability will merely give rise to the cry of “And whose fault is that then?”.

    The economic rights and wrongs don’t matter – as an electoral topic it blows up in the face of those who try to use it against the SNP. Saying that people would have been in trouble if they had voted Yes is irrelevant because they didn’t and they won’t be voting on that in May.

  8. Bill Patrick

    For a given distribution, yes it does. (There are distributions out there that require more statistics to be defined, once they are all defined you still know everything about a distribution)

    Knowing a normal distribution’s mean and SD tells you everything about the distribution. MOE is an example of such a derived statistic.

    If you want to do the maths more formally you’d want to use a multinomial distribution. To account for the fact VI percentages are not independent, the sum has to be 100.

    The tricky part seems to be figuring out how the effect of weighting approximates a random sample. It might be “on average” unbiased (unlikely to be perfect as we see variations between pollsters, the chances that one is perfect is unlikely) however the distribution generated could be quite strange compared to a theoretical multinomial distribution it purports to approximate. Even AW is pretty cynical about polls being true random samples.

    It could be that different methodologies lead to differences in the expected MOE and Unicorn is right and Yougov’s choice of methodology means the weird “samples” get made a lot more “typical” than for other pollsters, this would be an artifact of the methodology and not mean yougov were “better” as a certain amount of sampling error should exist, if you has someway of creating a perfect random sample in which everyone in the country had an equal chance of being sampled.

    It’s an interesting point raised that individual pollsters might have an (unintentional) bias in the mean for each party, the variation of their results might well not conform to the theoretically calculated variations either. We like to close our eyes and pretend that the number 3% is an absolute. It could be vary a lot depending on methodology.

  9. new thread

  10. @ Steve

    To be clear talking about oil prices is fighting the last war. This is all about standing up for Scotland, which is why Murphy is trying to show he is the boss.

    To be clear: Talking about oil prices is fighting the current war. Here’s why:
    A government – with a majority – which is committed to helping Scotland can do things immediately. A government which is relying on the SNP for support will have to hold off helping until all their ‘own’ legislation has been passed to be sure that the SNP can’t keep upping the ante.

  11. Unicorn

    Day-to-day variance will be smaller than margin of error because it is the difference between two random values likely to be within a range of the ‘true’ value plus and minus the MoE 95% of the time. But these values are not spread evenly across that range but clustered about the midpoint – the ‘true’ value – in a bell curve. So any pair are more likely to be nearer each other than the MoE (which what would be more or less the case if they were evenly spread) because both are more likely to be near the centre and so each other. There will however be a direct relationship between the variance and the MoE – the smaller one is the smaller the other will be.

    So YouGov’s day-to-day variance will be smaller because its MoEs are smaller. The main reason for this will be that their samples are usually bigger than most other pollsters and so MoE is smaller. And with the distance between most polls being only a day, any actual change in the ‘true’ value (short- or long-term) are less likely to happen between successive polls and going to be smaller when they do occur. So the main component in change will be this random motion.

    The variation for the Greens (and Lib Dems) will be less than that for Con and Lab because of the relationship with MoE as well. MoE is based not just on the size of the sample but on the percentage you are estimating (p)[1]. Most MoEs quoted are actually assuming a value of p=0.5 (50%) and there isn’t much difference if it’s 40% or even 30%. However if you get down to around 10% there is a noticeable difference and MoE is smaller[2]. So the variation will be correspondingly smaller.

    [1] In theory it’s also dependant on the size of the population you are sampling from, but with polls that’s always large enough that you can ignore that.

    [2] For example for a sample of 1000, at 50% MoE is +/-3.1. For 30% it’s +/-2.84 – not that different. But for 10% it’s +/-1.86.

    A government – with a majority – which is committed to helping Scotland can do things immediately. A government which is relying on the SNP for support will have to hold off helping until all their ‘own’ legislation has been passed to be sure that the SNP can’t keep upping the ante.

    Clearly put and possibly true. Arguably not the subtlest way to “entice” potential voters to LiS, though.

  13. @ Amber Star- I am merely pointing out that there aren’t many votes in telling the 45 they should be relieved it was a No vote. The SNP is offering to head south and bring home some prime political pork and few trust Lab is willing and able to deliver cash or powers. Jim Murphy realises his highly effective 2008 arc of insolvency line won’t work as Indy is not happening. Instead he needs to sell himself as the man who can bring power and money back from London.

1 2 3