Two polls tonight – Opinium in the Observer have topline figures of CON 34%, LAB 33%, LDEM 8%, UKIP 13%, GRN 7% (tabs). All par for the course, and fieldwork was on the 24th and 25th March, so prior to the Paxman interviews on Thursday.

More intriguing is YouGov’s weekly poll for the Sunday Times. Fieldwork for this was done on Friday and Saturday, so was wholly after the Paxman interviews. Topline figures there are CON 32%, LAB 36%, LDEM 8%, UKIP 13%, GRN 6%. A four point Labour lead. I’ll write more tomorrow, but I’ll leave you with the usual caveats, it’s just one poll, it could be boost for Labour from the interviews or it could just be normal random variation, only time (and subsequent polling) will tell…

177 Responses to “Latest Opinium and YouGov polls”

1 2 3 4
  1. Thanks Oldnat,

    My quick model of the crossbreaks shows that SNP VI has dropped ths week, and Lab has increased.

    Before this week snp about 43, Lab about 27.
    This week snp=38, Lab=31.
    What can it all mean?

    Meanwhile, did we get a milibonce?

  2. @Oldnat,

    I was not addressing my original comment to anyone, just pointing out that if anyon selects to measure something, its good to stick to it — as you have done — and I agree that you have not used you figures to infer anything.

    I think we are all grateful for you collecting the crossbreaks, I think tht they can be used with caution to give an indication of the VI in Scotland.

    Thankyou for collectin the numbers for us.

  3. Andyo

    If you look at the weekly means, you’ll also note that mean SNP VI jumped by over 4 points a couple of weeks ago.

    One of the reasons I don’t provide commentary on my data, is that that single change wasn’t supporting evidence for any trend. The mean VI dropped back by almost exactly the same amount the next week.

  4. Oldnat

    Yes, it may all revert to mean again next week.
    Only time will tell.

  5. Andyo

    Thanks for the thanks! :-)

    I’ve always been chary of drawing conclusions from summary data (which is all we get from any poll) since I did a major analysis of Scottish exam results over an 8 year period.

    The apparent improvement in performance in certain subjects (as described by the then fashionable measures) turned out to be mainly caused by schools changing presentation policy or, cutting the subject from the curriculum altogether, in order to make their official statistics look better.

    In polling terms, it would only be by digging down to the microdata that the academics get, that would provide really useful material.

    However, as polling geeks, we still avidly scan the limited information we ordinary mortals get. :-(

  6. Re the SNP v Lab gap in Scotland.

    At what point on UNS do, say 10 seats, stay with Labour?

    What would a 10% gap deliver on UNS?

    Second, that evidence of the swing being bigger in seats with bigger Labour majorities can work both ways.

    At current swing levels it means the SNP are taking votes efficiently as they are taking less but enough in seats they have a smaller gap to close.

    I guess at some level of average swing, though, it may mean that they miss on a few with a below average swing whilst getting bigger swings in seats with bigger Lab leads and take neither.

    FPTP in operation.

    Of course, now evidence of a narrowing yet.

  7. Re close results, I thought I remembered from an old [1964?] Guinness Book of Records that the record was in pre-Tanzania Zanzibar, which Wiki confirms:
    under “Closest election”, the Afro-Shirazi Party (ASP) won 10 seats, and the Zanzibar Nationalist Party (ZNP) won 9. Specifically, the seat for the constituency of Chake-Chake was won by the margin of one vote, with 1,538 for the ASP and 1,537 for the ZNP

  8. @Andyo

    Maybe a look at the regression analysis for the crossbreaks, modellers?

    The last time I did this all party fits were flat since December. As individual crossbreak figures have an MoE in excess of 8, it is not at all easy to detect a shift using this approach.

    I doubt whether even @CMJ’s CUSUM methods would pick up any change here.

  9. DUP leader on Marr Show confirming he would talk first with whichever party has the most seats in GB.

  10. OLDNAT

    I agree with your comments. I think if you saw a You Gov change that was sustained over two or three weeks , only then might one be able to conclude anything.

  11. @Unicorn
    Thanks for looking into it for us.

  12. Its right to be cautious ,yougov are either picking up real movement or its a statistical illusion based on a house effect or a random movement.

    But as keynes said when the facts change I change my opinion.

    Its easy to be wise after the event.

    Also of interest is the tory lead in the midlands/wales which is still there even with labour on 36 and a lead of four.

  13. I note from today’s YG that 60% said that they didn’t see any of the Ch4/Sky non-debate – even clips on the news, but that only 38% say they are unlikely to watch the 7-way debate.

    It would be interesting if YG followed up the individuals who say they saw none of the first, but were likely to watch the 2nd.

    Did they actually watch the 7-way debate or subsequent news clips, or did they find something more interesting to do instead?

  14. The effect of the tv “debates” is clear in the YouGov cross breaks .

    No wonder DC was wary of them.

    The seven way show will be no easier for him.

    Pleased to see Jeremy Hunt off the leash at last & saying the right things .

    Not an auspicious start to the campaign for Cons. DC has to do better.
    They really should lay off the Milibashing now and just say what the differences in policy, and the future are.

  15. I think there’s a danger we’re getting too excited about one single poll.

    Interesting to see the Sun’s Tom Newton Dunn saying we should be wary of attributing labour’s 4 point lead to the paxo interview on thursday.

    Last week he ran a headline that the opinium 3 point tory lead was due to the budget, which of course no other polling evidence corroborated.

    The danger of rubbishing miliband constantly was that expectations of him were so low that he only had to perform moderately for people’s perception of him to improve.

    He has a chance to get more exposure in the next few weeks and if he can show he has passion, and stands up for the underdog against powerful vested interests, he should do quite well.

    If I were a Labour activist, I would worry about his fluency and ability to think on his feet, but he must be feeling more confident now after the paxo interview.

    For reasons that I have repeated ad nauseam, I have felt it more likely than not that Mili would end up in Downing St. this year. I don’t think the fundamentals have changed significantly.

  16. @ Colin

    It may be that DC benefits from being the one against six in the seven leaders debate. Certainly EM needs to be careful how he plays this one. One thing we can say, it has the potential to be mind-numbingly boring and unenlightening television.

  17. I haven’t read all preceding comments yet but I’ve CTRL+F “19%” and nobody has mentioned this

    The YouGov poll apparently over-sampled people who watched the televised interviews. 19% of the sample watched them compared to 7% of the electorate. Also another 21% watched clips.

    This info is in the tables

  18. @Andy S

    So when does “swingback” start to get turned off?

    This has already started to happen in both models. In the case of the Election Firecast model the relevant information was posted here on UKPR some time ago by @Chris Hanretty. From memory, he stated that the reversing weighting is gradually reduced over some time but becomes sharply accelerated over the last 50 days. My recollection is that it never gets switched off completely.

    In the case of Electionsetc, answers can be found in Stephen Fisher’s working paper providing the methodological foundations for his model. Table 1 and Fig 2 are perhaps the most helpful here. Swingback reduces gradually from as much as twenty months out, so in that sense the turn-off (or swingback modulation) started happening a long time ago. Past evidence has shown that there are differences even a few days before the vote so, again, the switch-off is never 100%.

  19. New Tread

  20. @Oldnat

    Could you please post a link to your Scottish data?

    Later I am happy to go over, but I am pushed for time to collate it.

  21. I fail to understand where this historical ‘swingback’ to the tories comes from. I’ve just had a quick glance at the last 5 elections in the Historical Polling section and the only one where the polls show there was any movement back to the tories was the 1992 election. And wasn’t this due to the so-called ‘Shy tories’? My understanding is that all the polling companies now factor this into their polls.

  22. Rojo

    Are the pollsters factoring-in the ‘shy Ukipers’ for this election?

  23. @ Rojo

    1997 too, surely. From memory, Labour’s percentage margin over the Tories was less than the polls predicted. Didn’t Labour get about 43% of the vote when the polls had been predicting 45-47, with the exception of the infamous ‘5% lead’ poll about a week out from election day? I am sure I remember thinking that, given the circumstances, 43% had been a modest return for Labour – a figure that gave the Tories hope (should they respond to the defeat sensibly).

  24. @Allan Christie

    “Slight bounce for Labour but I’ll still say the Tories will win the election.”

    On what basis? Anything scientific, or is it just a feeling?

  25. @ProfHoward

    “I see DUP has said that if they are in a pivotal position they will negotiate first with the party that has the most seats.”

    I don’t think that matters one whit….

    As far as I can see, this election will either result in a loose Labour-SNP coalition, or a marginal Labour majority. If this YouGuv poll is a trend, then we could see Labour one or two seats short of a majority, and they can then form a coalition with the Greens!

  26. @PeteB

    “Why do newspapers take political positions anyway? Is it just to further the owners’ agendas?”


  27. What do UKPR folk make of the claims by YouSay that they are able to provide an instant Twitter-based verdict on the outcome of the debates?

    They say their result show that Miliband ‘won’.

1 2 3 4