Through December there were some somewhat contradictory polls – we saw YouGov putting the Conservatives way up at 45%, then an ICM poll showing Labour recovering. That was followed by a YouGov poll that also showed Labour recovering…but that was taken in the few days before Christmas when it’s had to believe a reliable sample could have been drawn.

It looked like Labour might have been recovering. Populus’s poll this month didn’t show the same sort of recovery in Labour’s support, but it did show the Tory lead falling thanks to them loosing support to the Liberal Democrats. Now two new weekend polls, one from Ipsos MORI for the Sun and one by YouGov for the Sunday Times suggest the Conservative lead is back up into double figures.

MORI’s topline figures, with changes from last month, are CON 42%(nc), LAB 32%(-3), LD 15%(+1). YouGov’s figures are CON 43%(+3), LAB 33%(-2), LDEM 14%(-1) – though that is comparing things to the poll done at Christmastime, comparing it to the previous YouGov poll the Conservatives are unchanged, Labour up 2, the Lib Dems down 2.

We can’t tell exactly what happened – it could have been that ICM’s poll and the Christmas YouGov polls were just blips or artefacts of the seasonal timing, and that actually the picture is pretty stable with the Tory party stable at around 40%, Labour recovering ever so slightly but still in the low thirties, Alternatively it could be that Labour had been recovering, but have been put back in their box thanks to the Hain funding row which has been ticking over during the week – both MORI and YouGov’s fieldwork was done mid-week. It’s now pretty irrelevant, whether there was a recovery or not, these two polls suggest Labour are back down in the low thirties.

On the Lib Dem front, there’s a contrast here between the Populus poll which showed them three points up and the MORI and YouGov polls which show no Clegg boost.

UPDATE: There is also a new ICM poll in the Sunday Telegraph. The topline figures with changes from the December ICM poll are CON 40%(+1), LAB 33%(-1), LDEM 18% (nc), so again the Tories up around 40%, Labour down in the low thirties but recovering slightly (remember the changes here are comparing the poll to the ICM figures that showed a big increase last month – 33% is still an improvement on Labour’s November’s figures) and no obvious boost for the Liberal Democrats.

UPDATE 2: Some more interesting stuff in the polls – the full tables of the YouGov poll are up here, I’ll have a proper look tomorrow.

90 Responses to “Tories back to 10 points ahead”

1 2
  1. What are the Yougov figures?

  2. Sean – didn’t have them earlier. Now up in the article.

  3. Terrible polls for Labour really. I think this is showing the folly of polling over the Christmas period, but these post Christmas polls are looking very much like the pre Christmas polls, with the polls that were conducted over the Holiday itself, seeming to be rogue’s.

    Labour haven’t had a bad Christmas period, and I’m not sure the Hain thing has been going on long enough to really filter into the polling data? So whats going on?

    Mt guess is, theres two issues at work, and both will fill Labour with horror;

    1. Brown has been all over the TV this week. Theres increasing evidence that Brown is proving a turn-off, and the more the public see of him, the worse Labour’s position get’s. Thats the opposite of the Cameron effect, where the more people see of him, the better the Conservative position gets. If that is going on (and the evidence os growing) how on earth will Labout get through a general election campaign with Brown in the spotlight every day for 3-4 weeks?

    2. Labour are flatlining at around their core support, and nothing they or the other parties do really boost’s their position. I’ve suspected for a couple of weeks now, that Labour are seemingly stuck at between 30-33% for the forseeable future. Whether the Tories go up or down. Whether the Lib-Dem’s go up or down. Labour’s position seems to be stuck on the low 30’s.

  4. Anthony , the new layout on Yougov’s website is ok but unfortunately now most of the links to the detailed data in the polls don’t work , hope that they have not been lost permanently .

  5. Mark. Deary me – I’ll flag it up on Monday. It looks like you can work round it though, change the “/archives/pdf/” in the address to “/uk/archiveupload/” and they turn up.

  6. The “economy” responses look rather partisan with both Conservative and LibDem voters predicting a recession in sizeable numbers (29 and 25% respectively, a further 27 and 25% also saying it won’t grow at all) whereas Labour voters still expect some growth in fairly high numbers

  7. Well I said that the polls at the turn of the year looked suspect. And indeed we now have retrospective errors on them: both 2.2%. The Weighted Moving Average is 41:33:16 and it won’t be long before it gets to 44:30:16 I suspect with this absurd fiasco of Peter Hain (and as Guido Fawkes points out even Gordon Brown’s Leadership Campaign is in breach of Labour Party Rules) and the general miasma of incompetence.

  8. As ever I look at the Scottish figures first and even with the usual caveats about sample size the 7% for the LibDems looks awful. On a UK scale that is equivalent to about 10%….

    As with previous YouGov figures it shows Labour and the SNP neck and neck mid thirties, although 18% for the Tories looks higher than recent polls but that could just be sample error.


  9. Good results for the Tories.
    Perhaps the sudden narrowing over Christmas/New Year were away from the trend, but my gut feel is we (in the Tories) need to move the agenda on a bit to a wide range of domestic policy to get into the mid 40s consistently.
    Gaffs and scandals are not enough.
    To be fair, a lot of policy work has been done, but often I doubt the public knows we are doing it.

  10. Still looks to me like the political situation is fairly stable, rather than the received wisdom of being volatile. Again, its steady post-Conference not just over the last couple of months but also matching what people were saying they would think this time last year.

    The good news for the Tories (and bad for Labour) is that this makes it look like our current lead is fairly strong and not flakey. However the concern is where do we go from here? If the ~40% who said a year ago they’d support Cameron’s Tories v Brown’s Labour are the ~40% now, etc, etc for the other parties . . . then what’s it going to take to cause a shift (any direction) in the polls.

    If GIN’s #1 theory is correct then that must fill Labour MPs with trepidation. If the situation is like this at the start of the eventual election campaign, I just don’t see Brown improving the situation during it when up against Cameron daily.

  11. As for detailed policy work, I disagree with the idea of announcing stuff now. Anything good will be nicked by the magpie’s in the government and thus unavailable for the eventual manifesto, anything unsuccessful will be attacked and used as a weapon (potentially in 2 years time still). The risks outweigh the gains IMO.

    The work should be done, but the results need not be announced prematurely. Besides, the opposition should not just be planning to be a successful opposition, but a successful government. If Cameron is to be our next PM then lets hope he gets there knowing what to do to move this country forwards. That doesn’t mean announcing daily “headline grabbing initiatives” Blair-style a couple of years before he’s elected.

  12. For all the doubters in my posting after the first POLLS of the new year especially by Populus I have been proved right again (which i said was a rogue POLL) . So many with little faith in my forecasts and predictions .

    I don’t know why the Tory followers panic at the sight of a rogue POLL and the Labour followers leap with joy at one .

    I agree with GIN in his comments – the Labour mishaps are helping the Tory boost slightly , but i honestly think that it has been proved over the last 4 weeks that even with no bad news coming in for Labour over the Xmas period and Brown getting more exposure in the last 2 weeks with his statements ,if this is the best they and the Liberals can do , they are gonna find it very hard to pull back up in the POLLS once the rot sets in – these POLLS show the trend now till the next election – with of course the odd rogue POLL showing through to worry some.

    You say that a lot of policy needs to come from the Tories – they can’t give out all their policies at once or too early – we have seen how Labour try to pinch them everytime – i believe it is working in the Tories favour that the public see Labour as without any new ideas and the Tories with new ideas that are filtered through gradually .

  13. Having been a labour voter party supporter since 1979 I think Gordon Brown is an embarracement to the country and his party i find him far to, false,elusive,opportunistic, and out of touch for me he and his spineless allies removed Tony Blair for this!!!Im sorry but for the first time and it hurts to say but its Cameron for me at the next election so promise me David you WILL GET A GRIP ON;Tax,benifits including foreign nationals,europe and its influence on our laws ,council tax spending , public sector pay and pension settlements,,health and saftey laws,the human rights act,immigration,pc culture and so on please set us free from this awful staight jacket of a society we live in today if you can you will have my vote for life well we can all dream cant we!!

  14. Cllr PETER CAIRNS :-

    You wrote “As with previous YouGov figures it shows Labour and the SNP neck and neck mid thirties, although 18% for the Tories looks higher than recent polls but that could just be sample error”

    Why do you find the Tory figure of 18% a sample error , but all the other figures especially for your party correct ?

    I have said before – the Tory revival is not just an English thing – even Scotland as part of the UK wants change , but not the type that the SNP are offering.

    I still believe that with a Liberal & Labour collapse at the next general election in Scotland the Tories could well take the 11 seats I have previously quoted !!

  15. With regards to the polls I agree with GIN we will not see the the full extent of the Tory lead over Labour until Brown and Cameron go head to head we had a small preview of that over the election that never was and the swing to Cameron was quite astonishing and Im sure the Labour party are only to aware of this so I think Labour will set their sights this year and maybe next waiting patiently for Cameron to slip up in one form or another then go for a snap election while Cameron and the Tories are trying to recover when that will be is anyones guess but I think it will be Labours only chance but even if Labour are ahead in the polls when the election is called I still think the head to head will destroy Brown and the Labour Party

  16. Mike, I am curious if your “Tory followers who panic” remark actually refers to anyone here or elsewhere in particular. I saw no such panic by anyone.

  17. Not good for Labour, but I guess the only glimmer could that if the Xmas polls were rogues there is a (very) slight recovery. The LD’s have not yet had an impact – that is something we need to wait for, as well as seeing if the Hain story has affected these polls – my guess is it hasn’t, and this is the underlying position. If it has had any impact, this mornings Tory donor story may be interesting – I don’t think it’s particularly important in itself, but any linkage of the Tories to the super rich may help to puncture some of DC’s ‘man of the people image’.
    The figures quoted above on the views of the economy split by party allegience are interesting. If a recession does occur, Labour is probably finished. If things turn out better than many expect, what will be the response of the Tory/LD voters who currently expect a recession? It may be that this is the underlying story of why the polls are where they are, and the ultimate result may depend on whether the more dire predictions are played out. My guess is that things won’t be as bad as many expect, and this factor may help Labour to close the gap in 9 – 12 months, but I wouldn’t bet my mortgage on it.

  18. Usual responses on here from those who pick and choose the poll figures they like . Look at the detailed figures from the Yougov poll . The unweighted data shows Con 36% Lab 26% LibDems 10% Others 7% . This purports to show that turnout out will be 79% at the next GE some 10-12% higher than all other pollsters and these extra voters are all going to vote Conservative . This may be true of a self selected panel more interested in politics than the electorate as a whole , but is manifestly not a representative opinion poll of the whole electorate .

  19. Mark Senior, you obviously have a thing about YouGov polling. As a member of the YouGov UK panel one would advise you to heed their results more then any others.

    YouGov has been the only polling organisation to inquire about my views, and one is – by habit – a political beast that would like one’s views to be reflected. One suspects many fellow-members hold a similar viewpoint. As such, at a election, we are more likely to vote. Other “random” polling samples always seem to miss me by.

    Add to which one likes the unweighted values. It matches ones forecast for Labour support. One would certainly regard YouGov’s analysis has more bi-partisan then the BBC’s George Osbourne [non-] story which it attempts to lead on today!

  20. Gin’s idea about GB’s adverse personal effect is interesting.

    Both the Times & Telegraph Polls asked opinion on “Welfare to Work” policies and got massive positive responses.Although both Labour & Conservative have policies on this it would usually be associated with the right of centre party I think. So we have yet another area of traditional Conservative thinking-like Immigration control and Crime reduction where the public expresses concern and approval for action.

    It is no wonder then that Brown and Clegg are shifting policy rightward-but if the public really wants these sort of policies why vote Labour or Lib Dem?
    Seems to me that there is a slow tide of opinion on “the state of the nation” which is moving from left to right. Question is who is trusted to deliver?
    Brown’s credibility credentials have collapsed,and taken together with Gin’s idea, if/when the public seriously buy in to Cameron’s then I think Labour have had it.

    This tip-over point is close I think-but not quite there yet.

  21. peter,
    as ever you are trying to down play the conservatives in scotland.their last poll was 19%.

    one day the snp(if the are not lefties)will work out the tories are their best vehicle to power.politics is not as grown up in scotland as norther ireland where the nationalists and the unionists work together very successfully.the reactionary parts of the snp need to grow up.

  22. Mark – I’m quite bemused by your recent obsession with unweighted figures. Weighted figures are what matter, the unweighted figures are a sign only of what would happen if pollsters didn’t properly weight their sample. YouGov often have to weight downwards older voters, but no sampling technique gives perfect samples just like that – ICM for example often have to weight up under 24s very heavily to get a representative sample.

    The better a sampling technique is then the less weighting should be required, and large amounts of weighting technically reduce the effetive sample size and increase the volatility of a poll. But for our purposes it’s largely an academic point – we can tell from experience how volatile different pollsters are and as long as they are all correctly weighted to make up for the biases of the different sampling techniques, the raw unweighted figures won’t effect the final figures.

    I don’t know what actual reason you are so certain the YouGov panel is more politically aware thn samples obtained using other techniques (well, beyond as you say, those people who desperately want to find fault with polls tht give low ratings to their party of choice). The overwhelming majority of panelists are recruited pro-actively to make up a panel that is balanced overall – very few just join through the website. At some points that will involve recruiting people with little interest in current affairs through advertising or surveys on entirely none political things like celebrities or sport – at other points that’s involved specifically recruiting people interested in current affairs because the panel was politically aware enough.

    The really politically and socially marginalised are unlikely to be included in any poll, however it is conducted. Remember response rates are about 1 in 6 or something, so those uninterested in current affairs are able to massively self-deselect themselves from phone polls.

    The difference in implied likelihood to vote is largely down to what the different pollsters do with the data and what questions they ask. YouGov polls have a proportion of don’t knows and won’t votes somewhere in the low 20s, ICM’s are normally in the high twenties or low thirties (though ICM assume that is partially down to the spiral of silence and reallocate many of those don’t knows). The differences in implied turnouts are much larger but are because YouGov assume all those people will actually vote, while other pollsters ask them their likelihood of voting and exclude various proportions of people who they think will not actually vote.

  23. Mark
    I really think it is time you stopped your vendetta against YouGov-they are a fine highly rated polling organisation with a good track record. Find another subject please.

  24. Mark
    Analysis of all the published polls since the beginning of 2006 shows that the only difference between the 52 YouGov results and the 118 from the other pollsters (on aggregate) is that the YouGov are somewhat more accurate (StD of their their “error” from the Weighted Moving Average is 2.1 rather than 2.7 – the mean “error” is 0.1 in both cases).

  25. I always thought we applied weighting to get us from people who say they will vote (for whoever) to “actual” votes in the notional election we asking about. As far as I’m aware these two things are quite different and have been so for many years.

    As far as Labour’s chances go, it seems that at last Labour’s competence on the economy is slipping:

    Looks like steadily rising oil prices will ruin any chance of a Labour comeback. But, would anyone else do better?

  26. I think Mike does have a point that some Tory supporters panic at the site of a worse poll or a rogue poll – (probably an unrepresentative sample of anoraks).

    I seem to spend a certain amount of my life trying my best to calm people down on Conservative Home who seem to demand that every single poll is as good or better than the last one – and was slightly worried before Xmas that there would be a load of vitriolic posts on there criticising the national party because the next polls weren’t 45%!

    Perhaps there is a nervousness that the lead may be related partly to scandals and gaffs. I agree with the other point that the Tories shouldn’t release all their policies to have them pinched – the trick is to set out clear aims without publishing hostages to fortune.

  27. I am pretty sure that turnout at the next General Election will be in the mid to high 60s.

    That seems a reasonable bet for an election which will be close.

    Perhaps it will rise 7 or 8 points in marginal seats, and increase about 2 points in safe Labour seats.

    I don’t have any statistical evidence for this – it’s a gut feel – but it looked like turnout hit rock bottom around 2002 (following the miserable turnout in the 2001 General Election) and has been on a gently rising trend since.

  28. Mike,

    It’s always difficult to put any real weight on the Scottish intentions because we are talking about samples usually under 200 and sometimes half that.

    The Scottish sample is only 193 in this poll so you really can’t say that the Tories going up to 18% from 16% is a trend.

    However the LibDems showing 7% when the lowest I’ve seen previously I think was 11% seems really low.

    It’s the same with my comment on Labour and the SNP.

    They seem both to be in the mid thirties and where as I suspect Labour are ahead, they could be anywhere between 32-40% and the SNP between 28-36%, so I wouldn’t really say anything other than we are clearly the two main parties in Scotland both probably polling more than the Tories and LibDems combined.

    What we could have is the possibility that the switch to Clegg the LibDems in Scotland have become less attractive to Scots Tories who defected when they had a Scot in charge and the likes of Michael Howard reminded people to much of the Thatcher years.

    It’s possible but I wouldn’t predict it on one poll with a sample of 193…..


  29. Someone earlier said this was a terrible poll for Labour. That’s not really true, since their vote is only down 3% on the last election. It’s more of an excellent poll for the Tories, taking most of their support from the Lib Dems, who are still down 9% on 2005.

  30. Andy,

    That’s what makes the 7% for the LibDems in Scotland stand out.

    In the last Westminster election they were on just over 21%in Scotland , which means that, in theory, they could have lost two thirds of their vote and be down 14%.

    If that’s true and they are still dropping , they are in free fall……


  31. I have a theory that a lot of well-off core Conservative supporters must have been on holiday over the Christmas period – thus accounting for their support dipping and now rising again.

    I am surprised how badly the LibDems are doing. So far they have received no bounce at all – if anything they are doing even worse.

  32. Clearly – a good poll for the Tories.

    Regarding one or 2 posts about the LDs.
    I agree that they are doing badly and, new leader or not, I still say they are going to get badly squeezed in the next election.

    If there is any sort of a major mood for change in the electorate this will hurt the LDs. If you want to remove the Government the best way to do that is to vote Tory.

    The LDs have benefitted in previous elections from Tactical anti-Tory voting. I don’t feel that sentiment will be there next time.

    Unless Cameron slips on a major banana skin between now and the next election (and there is plenty of time for it) the Tories will take a lot of seats back from the LDs.

  33. If the Tories can hold 42% and look to go higher this year then I think they will have an unassailable lead whenever the elction is.

    A word about gaffs. The polling for these polls looks to have taken place before much of the news about the Hain issue broke. The Hain Gaff is another gift to the Tories and bad for Labour.

    Time will tell but the Osbourne donation issue looks nowhere near as serious as the Hain episode.

    For a start the money was given to the Tory party (not to Osboune directly) then chanelled to Osbourne to run his shadow chancellor office. Secondly, the money has at least been declared (although possibly not totally in the way it should have been.)

  34. “If there is any sort of a major mood for change in the electorate this will hurt the LDs. If you want to remove the Government the best way to do that is to vote Tory.”

    KTL, I think you’re probably right about that. When Labour won in 1997 they lost no seats to the LibDems at all – they gained Rochdale and Inverness, Nairn and Lochaber East. In fact, even in 1992 Labour they lost no seats to the LibDems.

  35. If Osborne can produce emails saying he was told it was not a matter for the Register of Interests by that very department – having complied with all other rules, he should come out fine.. He can clearly show he had talks with them
    BUT Labour will try and muddy the issue and what is more may succeed if all journalists show the quality [or lack of it] that Martha Kearney showed on Newsnight on Friday. She didn’t allow a Welsh Nat MP to reply to charges that he was ‘dispicable’ for saying Hain’s credibility was still very much in question and allowed the ‘Hain man’ to use up what time was left with a repeat of ‘I have no explanation of what happened but I am sure all is well’.

    I am not a Welsh Nat, but he was treated disgracefully.

    I also note that the Times / Sun have gone soft on Brown recently. The talk on some websites is that the review of Murdoch’s ITV shares has led to a softening of Murdoch’s approach to Brown. We will see if the combined effect of the BBC and Murdoch [and the Mail] press is to keep Brown afloat or whether he will continue to decline.
    The Gov has not had a bad press in 2008. ‘Hain’ has an effect on us political anoraks, but doesn’t seriously affect most voters [more of a drip drip in the background]. It will be interesting to see if the PM’s high profile and a kind press continues to fail to help.
    If he continues to stagnate does it mean the public will have passed a tipping point or at least that the Tory lead is made up of ‘solids’?
    Political Betting have been having an interesting discussion about whether Brown’s ‘this is not a relaunch’ relaunch and his daily appearances are actually detrimental in themselves.
    On a previous thread I [half] jokingly said a particular poll where the Tories had gone down a bit and Labour had gone up a bit was because Cameron had not been in the news and neither had Brown!
    Many a true thing is said in jest.

  36. Sally C – I take your point that “Hain” won’t affect voters intentions in general and will be more of “drip drip in the background”. I agree with this.

    The problem for the Government is that there have been a lot of “drip drip” stories in the last few months and it is cumulative effect of all these stories together that is important on voters intentions not one story in particular.

    The biggest “drip drip” story that is hurting the Government is still Northern Rock which has been going on for a few months now without a decent solution for the taxpayer in sight. Now the NR Pension fund says they have a £100M hole in the Pension Fund and are asking for money. Where will this one end ?

  37. I have to say my own opinion is (unfortunately)somewhere close to Philip Johnson’s i.e the results are pretty stable Labour 32ish, Tories 41ish with the Liberals maybe moving around a little more between 14 and 19.

    I can see the sense in Cameron not making too many big policy announcements in case they get stolen inheritance-tax-style but I think it would be a mistake to keen too silent. I think most people generally vote for someone, not against someone. If by election time they don’t like Brown but they’re not too bothered about Cameorn either thay simply won’t vote at all. Mike,

    I can’t quite remember what your prediction was – you suggested I should cut and paste it for later use but I forgot. Could you remind us what it was again? I know it was for the first week in February but I can’t remember what the percentages were.

  38. Sorry, presses enter before reading that through. THe second paragraph was supposed to be addressed to Mike Richardson.

  39. Sorry to have upset the Yougov supporters with my observations earlier today . As a point of interest I am a member of the Yougov panel to which helps to show that it’s representatives are more politically motivated than the population as a whole .
    Yes Anthony , I accept that the weighting SHOULD eliminate any discrepancies in the unweighted figures because of a non representative sample . However the unweighted figures from all the polls show that the Yougov sample responses always have a greater response from over 55 ABC1s than they should do .
    Further the essence of getting an accurate opinion poll of the country as a whole is to have a sample which is representative of the country as a whole . The panel used by Yougov is clearly NOT representative of the country as a whole . It’s members are clearly more interested in politics than the population at large witnessed by the fact that nearly 80% of them say they will vote in a GE whereas other pollsters find a figure in the mid to high 60%s and the figure at the last GE was of course in the low 60%s .
    Yougov may produce useful polls in measuring the political temperature of that part of the population that are politically motivated but it requires a false assumption that the non politically motivated part of the population less interested in politics will behave in the same way to use them as measures of overall national public opinion .
    I would welcome someone from Yougov able to give some evidence that their panel is rather more representative of the whole population than I have stated .

  40. Mark,

    As the polls are weighted to give as accurate a prediction of the intentions of those who will vote as possible it doesn’t matter in the slightest if the group you started with is 80% or indeed 405 likely to vote, as it’s only the ones who will vote that you will base your prediction on.

    You seem to be moving towards saying YouGov can’t be relied on to give an accurate election prediction because it doesn’t the voting intentions of people who probably won’t vote, that’s just daft.

    If you did a 100% poll of every voter in the country and everyone responded, which would be closer to the actual result?, The figure for the 100% asked or the one for the 60% who would vote?

    YouGov like everyone else filters to focus in on the voters and the bit they cut out be it big or small is just chaff.

    It’s not about testing the mood of the nation for the sake of it ,it’s about accurately trying to predict what will actually happen.

    The proof that YouGov’s panel and methodology works is pretty obviously that they have a proven track record of getting poll results pretty damned close to the actual election results. I suspect that if you compared all the polsters unweighted results, weighted ones and actual election results you’d find that weighted ones are more accurate.

    That’s why people weight results…..


  41. Peter , no I am not saying that Yougov cannot be relied on to give an accurate voting prediction because they do not poll the people who do not vote . I am saying they do not give an accurate voting prediction because they do not poll a representative sample of the whole population .
    I recall that one of the first ever opinion polls in the USA in the 1940’s by Gallup had a massive sample number ( 40,000? ) but was way out on the final result because it only sampled people who read the Readers Digest and as it turned out their voting intentions were not the same as the rest of the USA population .
    Noted your comments re the Yougov Scottish subsample . FWIW with the low subsample numbers , Populus had Conservatives at 11% and LibDems at 13% which does not agree with your hypothesis .

  42. Peter , one further point , I have nowhere said that unweighted polls are more accurate only that looking at the unweighted data can give some clues as to whether the sampling is a good representation of the population as a whole .

  43. If I were the Labour Party person in charge of boosting their figures I would
    1. Have less of Brown and more of the other members of the Cabinet
    2. Have fewer messianic policies for 50 years time but more policies for today such as how to clean hospitals
    3. Admit the whole process of electing a Deputy Leader was flawed
    4.Make Vincent Cable an offer to be Chancellor

  44. Mark,

    But representing the views of the population as a whole is pointless if it doesn’t give you the result you want, one that will reflect the final outcome. The objective is accuracy.

    If you asked everybody in a factory lossing money that could be closing should it close the majority would usually say “No we want to keep our jobs”.

    But if you asked the members of the board you may well get a different result and it’s that result that will be closest to the final outcome. It’s the board you sample because they are the ones who will make the decision that you are interested in.

    It’s the reverse of the RD Poll in 40″s American, there they polled the wrong sample and got the wrong result. The key lesson from that isn’t to poll everybody but to poll the right people.

    As I said previously the touch stone of that is how close you get to the final result, and so far YouGov is up there with the best.

    You can never know till the votes are counted and you are constantly reviewing and checking your methodology, but I think you’ll find lessons have been learned and things moved on a bit in the last 60 years.

    As to Scotland, different polsters with different methodologies produce different results in their UK polls, I think MORI had the Tories at under 10% last month but the number of identified Tories was only about twenty.

    The issue with the YouGov 7%, is that it looks exceptionally low compared to previous YouGov polls over the last year or so using the same methodology and indeed the same panel.

    The difference is that as Anthony is pointed out as YouGov use the Scottish TV region as a results break but don’t try to balance within regions then the very weighting system that means it’s UK samples are accurate can’t be used in the sub samples.

    That’s why Anthony really only comments when there is a full quality Scottish poll, he’s got a reputation to think about, me I don’t give a monkey’s. For all there limitations sub sets are all I usually have and I find them interesting.


  45. Mark –

    From the top:

    1) You don’t know what the demographic make up of the YouGov panel is. No figures have been made public for a long time.

    2) It is irrelevant anyway what the make up of the overall panel is. To give an example, imagine there was a panel that was perfectly representative, obviously you can get a very good sample from it. Now imagine that an extra 50,000 older, working class men from Wales joined. The panel as a whole would now be grossly skewed towards older working class welshmen…but you still have all the people you need for a perfect sample, there is just more choice when generating the quota of older working class welshmen. The overall make up of the panel doesn’t matter, it only matters that there are enough people of every social group and classification to allow a representative sample to be drawn (though for a business, it’s obviously advantagous if lots of discreet samples can be drawn at the same time so you can have lots of clients!)

    3) The current YouGov sampling frame appears to get more responses from over 55s than are needed, so they have to be weighted down. Again, this doesn’t matter beyond slightly decreasing the effective sample size. Imagine 75 fewer over 55s checked their email that day, would the sample then be wonderful? Well, that’s the effect that weighting them down has.

    4) There is no sampling technique that gives perfect results. Look at any tables and you’ll find some demographics needed a lot of weighting to get right. Phone polling tends to get far too few young people and needs to weight them upwards, there is no point comparing things to the platonic perfect sample cos it doesn’t exist (I haven’t checked, but I expect you’ll find MORI need the lowest level of weighting since they have the most control over their fieldwork, interviewers pick people to meet demographic quotas and can stop interviewing when they fill them. YouGov can invite people to match the profile, but if more respond then they have to weigh them down). The degree of weighting that YouGov uses isn’t higher than other pollsters – I just gave you an example from ICM’s weighting of under 24s last month, ICM very regularly apply a weight around 1.4 or so to under 24s, a greater weight than that YouGov use on over 55s.

    5) There is no comparable data to demonstrate that YouGov respondents are more likely to vote than respondents in phone polls. The pollsters ask in entirely different ways, and there is very likely to be an interviewer effect. With the exception of YouGov pollsters ask people to give likelihood on a scale of 1-10, with the lowest number meaning definitely not voting. YouGov only determine not voting by including a not voting option when asking about voting intention.

    If anyone came on here trying to claim there had been a rise or fall in likely turnout by comparing a figure taken from an interveiwer question asking directly about likelihood to vote using a ten point scale, to a figure obtained online by taking won’t votes from a voting intention question then they would rightly be laughed out of the place.

    YouGov very rarely ask likelihood to vote, When they do the figures are not unduly high. For example, in the final polls before the 2005 election YouGov found 55% of people said they were 10/10 certain to vote, ICM found 62% and MORI found 70%. There’s a big spread there, causes by various reasons like how postal votes were included (thats a major reason why YouGov is lower, the percentage didn’t include postal voters), exact question wording, interviewer effect, question order and so on. Wording matters a lot in this question – it’s why ICM always preface the question with the words to the effect that “many people have said they wouldn’t vote in an election” – in any other question we’d dismiss that as having the potential to skew responses, but in this case pollsters try their best to get people to admit they may not vote. What differences there are between pollsters are as likely to be down to wording as sampling.

    6) There is no evidence to suggest YouGov samples are anymore unrepresentative of the population than those obtained by other pollsters (beyond, of course, the obvious of having internet access)

    If you want an answer from YouGov then feel free to email Peter, he doesn’t comment here.

    I feel unhappy put in the position of defending YouGov here since I don’t write the blog from that standpoint. I hope I would defend ICM or Populus against unfounded sillyness in the same way, and while it’s probably obvious that I’m a support of past vote weighting so don’t favour MORI’s methodology, I hope I give a fair explanation of their case when that arises.

    I’ve said many, many times I really don’t give a stuff what polls people believe, if they want to believe voodoo polls, or press-the-red-button polls on Sky, (or the findings of Scottish sub-breaks with tiny sample sizes Peter!) or whatever else then that is their call; no skin off my nose. My aim here is only to help people understand why there are differences between the polls, what the genuine differences and question marks are so they can base their opinions on something more than a visceral dislike of any pollster that give their party bad ratings.

  46. I suspect past vote weighting sounds valid, but in practice can be a bed of nails.
    If people are shifting their view they may well deny what they voted before.
    A lot of (ex) Tories would have denied they voted as such in 1992 – after Sept. 92.

    And some people have really atrocious memories – no head for details – and can’t remember.

  47. JJB – “False recall”. I’ve talked about it here at length in the past, but it’s probably time i did another post on it (especially if relatively long time readers like you can’t remember the last one).

    I think pollsters univerally accept that it is a factor theses days – the evidence from panel surveys is pretty conclusive that the same people will change their response to past vote questions – the dispute amongst pollsters is more how stable it is over time. ICM and Populus take the view that the pattern of false recall changes only gradually over time, so it is safe to weight by past vote adjusted to take into account an estimated level of false recall. Ipsos MORI take the view that isn’t stable and can change suddenly in response to events, so that weighting by past vote might actual cancel out and hide genuine changes in opinion.

    FWIW, I think the main drivers of it are social pressure to claim you did actually vote, which tends to push up the people who claim to have voted Labour, and the way people report tactical voting. If you were a Labour supporter who voted Lib Dem at the last election because you were in a CON-LD marginal, then I suspect many would answer Labour if asked who they voted for at the last election because Labour was the party they really supported, the Lib Dem vote was just because of local circumstances…hence false recall produces a lower level of Lib Dem support than they actually got.

  48. The headline for the MORI poll in the SUn on Saturday was interesting- It was something like ‘Battler Brown stays ahead’. This was the case because Brown was 3% ahead of Cameron on one of the head-to head clashes. Labour were 10 points behind. Pretty good headline for a 10 point deficiet.

  49. Lukw – swings and roundabouts really. The Times poll showed the Tory lead halved, but headlined it as being bad news for Labour because Brown’s own figures had slumped. Their stablemate the Sun showed a great big 10 point Tory lead, but headlined it positively for Labour because Brown stayed just ahead of Cameron in their figures.

    A reminder to always look at the figures themselves rather than someone else’s analysis ;)

  50. LUKW.
    I note your comments about The Sun. Brown has put Murdoch’s holdings at ITV back on the table for discussion and so the paper will row in behind him for a while. Polls are like all statistics, they can and are being used selectively. In some cases, I would suggest this may come about from direct pressure from the Govt. If Murdoch is forced by the Government review to sell his shares in ITV he is said to lose 200 million?
    Murdoch claims that he does not influence The Times. We will see.

1 2