Tory YouGov lead down to 4%
December 22nd, 2006But it’s 45-32 on the Cameron-Brown forced choice
This month’s YouGov poll in the Daily Telegraph shows very little movement on a month ago. These are the shares - CON 37% (nc): LAB 33% (+1): LD 17% (+1). So the overall Tory lead is down a point with Labour and the Lib Dems both putting on a point.
But the named leader measure that YouGov uses - a forced choice asking “would you prefer to see after the next election, a Conservative Government led by Cameron or a Labour Government led by Brown” - shows a widening gap. These are the responses to that question for the past year:-
NOV 2005 CON 37: LAB 46 (LAB +9)
FEB 2006 CON 37: LAB 43 (LAB +6)
JUN 2006 CON 44: LAB 38 (CON +6)
AUG 2006 CON 43: LAB 36 (CON +7)
OCT 2006 CON 46: LAB 33 (CON +13)
NOV 2006 CON 43: LAB 34 (CON +9)
DEC 2006 CON 45: LAB 32 (CON +13)
So on this measure there has been a fall back in the Brown recovery that we saw in November. This question, of course, is forced and is not about voting intention but the huge falling off in support for a Labour Government under Brown in barely ten months should be worrying for the party. The Chancellor needs better numbers when questions like this are asked.
But there’s is a glimmer of good news for Gordon in another question - when asked, however, who would make “the best PM” respondents in the survey went Cameron 28%: Brown 27%: Campbell 6%..
So people are ready to rate Brown almost alongside Cameron when asked “who is best” but much prefer the younger man when pressed on what sort of Government they would like.
Will there be a Brown bounce when he eventually takes over - and in what direction will it go? If these and other poll findings are right then the best thing that can happen to Cameron’s Conservatives is for the Chancellor to take over from Blair. All hypothetical of course - but Brown needs better numbers to ease the jitters in his party.
UPDATE - New Mori Poll
The Ipsos-Mori poll for December is now up on the firm’s website and shows, with changes on last month CON 37 (+2): LAB 36 (+3): LD 18 (-2).
These are based on people saying they are “100% certain to vote” and Mori, as we have discussed at length here before, does not weight by past vote recall in order to ensure a politically balanced sample.
The voting figures are based on the responses of 1,075 “100% certains”. The non-voting question have Gordon Brown with a 42-42 satisfied-dissatisfied rating and Cameron with 28-33. It is not clear from the site what the base is for the non-voting questions.
Mori do not ask a named-leader voting intention question or a forced choice like in the YouGov poll.
As ever because of its sampling methodology I do not attach as much importance to Mori as ICM, Populus or YouGov.
Mike Smithson
MessageSpace Advertising
“If these and other poll findings are right then the best thing that can happen to Cameron’s Conservatives is for the Chancellor to take over from Blair”
Surely McD or MM becoming Labour leader would be better for Cameron’s Conservatives?
Or do you think Micheal Meacher’s Labour will poll better than Brown’s Labour?
Opik/Cheeky/Sian affair going on in the Mail…Lembit’s camp spins it’s not his all fault, but the relationship fell apart because Sian was too close to her gay friend
Lab Deputy Leadership race: interview with Cruddas
http://www.epolitix.com/EN/Interviews/200612/594e6fd0-1811-4e34-9c33-702b2d4d9eaf.htm
Yes, Mike, the ‘Brown bounce’ is a great imponderable.
You would normally expect a new Leader to get some sort of a poll lift but the situation with GB is unique. Nobody has ever been leader-in-waiting that long and he is so closely associated with TB and the whole Nulab thing that he is justly perceived as being in some respects a virtual leader already. This suggests any bounce may be small.
You can paint all sorts of possible scenarios, all perfectly plausible, but nobody knows what’s going to happen. I certainly don’t.
And I wouldn’t bet on it.
At times like this one misses Jack W’s ARSE. The polls seem to show Brownian motion, and I wonder how many of those voters who do express a preference start by saying “well I don’t know really…”
Conservatives will be disappointed that this does not reflect the apparent trend of the last ICM poll. But donç´t worry, lads - it is only yougov!
Given the scandals, I am surprised that Labour are doing so well.
2 “Nobody has ever been leader-in-waiting that long” Arguably both Callaghan and Eden were in a similar position - although a good part of Eden´s time as leader in waiting was in opposition.
Didn’t Populus showed a similar trend in the DC/GB question?
So I would suppose December has not been a good month for Gordon (or November has been a good month for him and the effect is already finished)
re 1. Actually Andrea I think that McDonnell would poll quite well. He’s personable and is quite good on TV - and he is English. He would also have the benefit of novelty.
I do have an interest - £6 staked on him for leader at an average price of 474/1.
5 Jeepers, Mike, why didn’t you let me know? You could have given me that £6.
Wouldn’t it be great if there was a “Brown trounce”? I think I’m going to copyright that
re 4. Yes. Populus recorded a 3% increase in the Tory margin with the named Brown-Cameron-Campbell voting intention question between November and December.
1 & 5 – One of the Scottish polls done by YouGov recently asked who people would most like to see as the next Labour leader. John McDonnell was given as an option and got around 2% IIRC. Which was about 2% more than I expected.
9. Max, I suppose the biggest surprise was that 2% knew who McD was, right?
It has been pointed out on here many times how good TB is, and how much he’ll be missed.
Govt performance has been poor. But for defending the indefensible, TB is the master. The tories have been accused of missing a series of open goals, which is fair comment. But with TB in goal, it has been tough to score, even from two yards.
It is comfortable to say that we don’t really know how plausible GB will be at justifying the unjustifiable. True. But this a betting website. And punters win money by making more right predictions than wrong ones. My cash says that GB will be significantly less effective, on bad days, than either TB or DC.
By PA Elections Editor
Tories gained overall control of Basingstoke and Deane Borough Council in Hampshire, winning a landslide by-election victory in a traditional Labour stronghold.
Candidate Robert Taylor took a Buckskin seat to give his party 30 councillors compared with 29 for the combined total of other parties and independents.
However there is one vacant Tory seat.
This is the first Conservative council by-election gain from Labour since the end of May.
I am genuinely intrigued by what the public are making of Cameron. It is being assumed that since he took over he has transformed Tory fortunes. From being 4 points behind Labour they are now 4 points ahead.
This of course ignores the serious problems Labour have faced (mainly of the leaderships own making) and it assumes that without the rebranding the parties would still be where they were.
So the startling figure that Blair leads Cameron by 27% to 26% as best Prime Minister is quite an eye-opener. That Blair has touched rock bottom can’t be in doubt. That Cameron isn’t even rated as highly as best PM is interesting.
This coupled with the Ipsos MORI report for the Fawcett society posted by Nick P must make Labour rather more content than they deserve to be
Tanks for the articles Mike. Well, looks like the polls are all over the place then. This poll and ICM poll are within each others margin of error indicating a probably Conservative lead of about 6%. Much done, much still to do.
RE 2, Peter the Punter, I agree on the unpredictability of the Brown bounce. Certainly from a betting point of view and the situation is unique as well, as you say.
Personally I don’t think the bounce will be big or sustained, but I would not bet on it
3. On the contrary, icm and yougov are picking up the exact same trend. Tories with a steady lead, Lab picking up one point. Icm had us gathering one extra point YG holding steady.
And if NuLab are happy with that, they are very easily pleased.
Excellent local result in Basingstoke.
Somebody post the full by-election results please!
LOL @ Andrea - following the Cheeky story closely are you? Looks like the Mail have swung behind Lembit now all the details are coming out:
http://www.dailymail.co.uk/pages/live/femail/article.html?in_article_id=424267&in_page_id=1879
This time the 3rd man was gay!
PS Sian was 44 when that picture was taken. Damn! That’s very fine for that age…
Politics is becalmed! Despite their chirpiness Conservative posters must be disappointed. Interesting piece in today’s Guardian by Polly, probably gives as good an assessment of the situation as you are likely to read. Perhaps Mike should start putting out odds on, whether Ming will last till the GE. Ming is 64 could be 67 when the day arrives, can’t see him lasting myself, the Libdems need a ‘young Turk’
16 “if NuLab are happy with that, they are very easily pleased.”
If any of the parties are content with these figures they are easily pleased. Except BNP perhaps, who are said to be at 4% in htis yougov poll. (This seems to me an indication that the BNP attempt to inflitrate the yogov panel is still bearing fruit).
17- here you go
Basingstoke DC Buckskin (Single member Labour seat 2004) result Lab 384 Con 251 Ind 152 - Candidates Con/Lab/LibDem/Green
2006 by election- Conservative gain from Labour Con 332 Lab 210 LibDem 123 Green 40 .
The last 4 polls give an average 3% lead to the Tories. I cant remember being more irritated with Labour and it’s leadership than at the moment and I sense that many other Labour voters are equally fed up.
Why aren’t the Tories benefiting? I wonder whether the perception that they’re without substance is really starting to take hold? There cant be a better time to be a thrusting Lib Dem!
16. 17% swing in Basingstoke….and that completes a miserable December for Labour, whose average vote share in the contests during the month was down 6% - worse even than in May.
NOt at all disappointed. All the polling suggests that when Brown takes over from Blair we will pass the 40% mark. It seems to me that only the residual appeal of Blair (pseudo-Tory) is holding onto that couple of percent of the electorate that used to vote Tory and havent yet taken to Cameron. When faced with the choice of Cameron or Brown I think they will come back to us.
The interesting thing is that several percent who hate Blair for Iraq etc and currently vote Brown, may go back to Labour when Brown takes over.
So the worse news is actually for the Lib Dems.
Rik W sees bleak future for LibDems, Shocker…
Rik W, how much money would you be prepared to put on, the Tories holding a sustained lead of 40%+ after the change?
17. MBoy…I suppose the difference is that Cheeky Girl accepted Lembit’s offer to sleep naked next to him, whilst Sian was usually busy making phone calls to her gay friend whilst during the night
Re Trident vote on Scottish Parliament. I was looking at the text of the various motions/amendments.
Nicola Sturgeon’s motion was: “That the Parliament notes the publication by the UK Government of its White Paper on the future of the Trident nuclear missile system on Monday 4 December 2006; recognises the need for a full debate to explore the military, economic and political consequences of Trident renewal and believes that a convincing case can be made, in military, economic and political terms, for the non-replacement of Trident, and calls on the UK Government not to go ahead at this time with the proposal in the White Paper”
It got just the votes of SNP, Greens, SSP, Solidarity, Canavan and Jean Turner + 4 Lab rebels.
2 MSPs abstained (1 Lab and 1 LD)
The one that almost made was the LD amendment: That this Parliament rejects the case made by the UK Government in its White Paper on the future of the Trident nuclear missile system published on 4 December 2006 that the decision on a replacement for Trident needs to be made in early 2007 and calls on the UK Government not to go ahead with the proposals in the White Paper at this time.”
It got the votes of the ones who voted for Sturgeon’s motion + Libdem MSPs. Except the 4 Labour rebels who didn’t vote for LD amendment.
In the end it was defeated 62 to 57
Talking of thrusting Lib Dems…..what irony that the third man in Lembic’s love triangle is called Stifyn!
PA cont…
The by-election was caused by the resignation of Labour’s former group leader David Potter.
It was originally thought that there would not be a contest under what is known as the sixth-month rule because the seat is up for election next May. However, a closer examination of rules showed the vacancy had occurred before the critical deadline.
Analysis of nine comparable by-elections in December suggests a projected 14.6% nationwide Tory lead over Labour.
A calculation based on four wards where all three parties fought both times gives a projected line-up of: C 43.4%, Lab 28.1%, Lib Dem 21.5%.
There are barely four months until May’s marathon mid-term English council, Scottish Parliament and Welsh Assembly contests, likely to be the last under Tony Blair’s leadership.
RESULT: Basingstoke and Deane Borough - Buckskin: C 332, Lab 210, Lib Dem 123, Green 40. (June 2004 - Lab 384, C 251, Ind 152). C gain from Lab. Swing 17.1% Lab to C.
23.”All the polling suggests that when Brown takes over from Blair we will pass the 40% mark.”
which polling?
Populus used to show Con over 40% with Brown (and LD at very low %), but in the last couple of months they had the tories in high 30s (and LD back at reasonable numbers…high 10s)
Mike S, I wonder how immigration/crime is playing out in these Tory advances. Polls show the issue is at or near the top of people’s priorities - and with this string of appalling stories - of illegal but non-deported immigrants slaughtering policemen and nursing mothers in the streets - I think it must be helping the Conservative cause.
Labour talk tough on immigration but people simply don’t believe them: Labour have presided over the collapse of the asylum system, the tripling or immigraton to record levels, and the absurdity of human rights laws meaning we can’t deport murderers. Blair can waffle all he likes about the need for integration but people aren’t that gullible any more - they see what is before their eyes.
So they will vote for the party that is traditionally more robust in this area.
I think immigration/crime will be perhaps the major issue in the next G/E. And I think it will therefore assist the right, no matter how much Labour posture. The question is: how many people will go straight to the BNP rather than stopping at the Tories?
I can envisage the BNP doubling their vote at the GE, where they stand.
LOL @27
Annoying that the Lab rebels didnt vote for the LibDem ammnt - that would have been enough and would have been nationally significant for Trident in the UK.
ARTICLE UPDATE - New MORI Poll
26/31. Margo MacDonald and Campbell Martin (former SNP) voted for both SNP motion and LD amendment.
Monteith (former Tory) voted for LD amendment and against SNP motion
30. The escaped in a veil case encapsulates the lunacy of our current asylum/immigration system rather well. A murderer escapes wearing his sister’s veil - which he is not obliged to remove due to PC considerations. He escapes to a country to which the Home Office said he could not be deported - on the grounds it might endanger his safety.
9. The 2% quite possibly thought the option was Jack McConnell!
34 - Sorry to be pedantic, but is there any evidence anywhere that this “murderer” did “escape” wearing his sister’s veil? There is no evidence for this supposition whatsoever, but it is being touted round the newsrooms as if it was Holy Writ. Personally, I think he escaped to Hartlepool as an Elvis Presley impersonater, a theory which has just as much evidence as the other.
Incidentally, after the Basingstoke result last night, I think that this has been the first year that I can remember that the Conservatives have won more local government by elections than the Liberal Democrats.
test
36. The veil story is irrelevant. Ignore it.
The guy who killed the policewoman was a foreign criminal. He was slated for deportation to Som@lia; but he was saved by UK judges under human rights laws, and the deportation was stopped.
Then he killed the policewoman - and fled to Som@lia.
This government’s record on immigration/crime is disgusting. They talk the talk - and then they betray the British people.
37 Not quite true Augustus 2004 had Con +11 LibDem + 8 This year’s final figure is Con + 18 LibDem + 17 .
Anyone know the Yougov sample size?
Watch out for a large increase in LD support in the next opinion poll. The Lembit story gives welcome publicity to the LDs. Nothing wrong appears to have been done its just as tittle-tattle and helps to remind tabloid readers that the LDs are out there and are “human”. Remember the poll boost after the “Paddy Pantsdown” headlines?
Perhaps Lembit could fix up Gabby’s sister with Ming on a double date?
42.”Perhaps Lembit could fix up Gabby’s sister with Ming on a double date? ”
Maybe Mum Cheeky Girl would be more appropriate!
I think the sister is fixed up with Liam Bryne
Come on He old enough to be her Grandfather (her mum that is)
Thanks, Mark S. As I said, “the first year that I can remember”. These “senior” moments (if you will forgive the phrase) are becoming more of a problem!
21. I might describe lembit as a thrusting libdem at the moment.
30. the funny thing is a labour friend of mine thinks that Labour are seen as competent on the immigration issue….couldn’t believe how out of touch he was.
I’m very encouraged that after failing to arouse significant posting in the past following my comments on opinion polls and elections, at last a get a string of responses to comments about Lembit. The cheek of it!
42. Opik needs to watch his step. Its absurd to suggest he doesn’t use his position to better understand what is going on (because, lets be honest, he can’t actually influence it, he’s an LD) but in these ridiculously over sensitive times asking questions, not for cash, can get you in the soup of apparent impropriety.
I mean, fat old pervert, womean bothering politicians seem to to attract less attention these days even though they should be beaten.
“Use his position” nudge-nudge, wink-wink, say no more.
48. Mike Sole. Maybe you didn’t get many replies to your past comments, because people were agreeing with what you said!
In this case, the matter is more controversial!
44.”I think the sister is fixed up with Liam Bryne ”
Peter Pigeon, why do you want to ruin a happy family?
http://www.ivydene2.co.uk/liambyrne/pressroom/thumb-liamandfamily.jpg
52 looks like he is married to Sian Lloyd
50. Holy good God, someone is Christmas demob happy today…..can we put you down as a writer for the new movie ‘Carry On Up Westminster’?
41. RodCrosby, the yougov sample size was 1,874
I just can’t take the Cheekygate scandal seriously.
Now there is a sentence I never thought I would use.
Re. Polls. Was braced for worse regarding Labour’s rating. Not really reflecting my personal experience on the ground. But do they ever?
My data point of one in the South sees weaker, but still vibrant LibDems, very weak Labour vote but not wiped out and a Tory party still not noticeably stronger than 2005. None of the above party seems to be storming it.
So it’s MORI + 1
POPULUS + 1
Yougov + 4
And ICM + 8
It’s looking like ICM is a rogue poll.
I think that’s -1,-1.-4 and -8 Roger. There’s little positive about a Tory lead.
This is my last posting before Christmas.
Based on Andrea (51) I expect no comments.
MERRY CHRISTMAS TO ALL CONTRIBUTORS OF PB WHO HAVE GIVEN ME MANY ENJOYABLE READS SINCE I DISCOVERED THIS WEBSITE A FEW MONTHS AGO.
Well lets have a look at a fairly consistent thread in the polls, Cameron is consistently pounding Brown (no innuendos, please…)
What excuse are we going to make for that?
Before i go off to the gym, then ruin it by boozing it up all afternoon and evening, Santa obsessives should check out http://www.noradsanta.org . The versions of christmas songs are particularly bad but its otherwise an inspired site. I believe Tower Hamlets has an equivalent http://www.christmasdoesntexist.org
…………..
Thanks Andrea. (polite request: please can we have sample size included with every poll thread…)
59. well here’s a response merry christmas back at you….just a thought, can we expect a significant reshuffle before Blair goes….last throw of the dice you know
Seems to be status quo ante pretty much in the polls (if we assume the 8% lead was not a rogue, but just on the upper end of the actual range). If anything I’m reasonably happy with that- given the pile ‘o steaming **** we’ve waded through over the last month.
On the question of the Brown bounce. That depends on what he does as PM. If he walks out of No 10 on the first day and says “everything is perfect I’m keeping everything the same” there won’t be. If he does stuff that people like there will be. To a certain extent his challenge will be to show he is changing things.
ICM had the BNP at 3% so anyinfiltration is marginal at best. The unpalatable truth is that many people are going to vote for the BNP at the earliest opportunity.
Poor lib dem figures and, I’ll say it again, this is for one reason only. Lib dems are not attacking labour anywhere near enough, in frantically trying to stop Cameron the left flank is being left exposed. The votes are there to pick off from labour and, in doing so, a hung parliament could be there for the taking. A by-election win in a labour seat could help but if an illiberal government can’t be attacked successfully then something is wrong in terms of strategy.
Whoops, I meant Populus and 2% not ICM and 3% (thinking SNP!), the point still holds in that people are more likely to avoid hiding their support for them in an online survey. Local election results (if they’re good for anything) bear the 4% figure out.
63. All the polls seem to be showing not much change month to month at the moment, it’s just that the different companies disagree on what the levels of support are. The gaps recorded by each firm are pretty consistent.
To return to an earlier topic of conversation, the most significant point is that the Conservatives for the second year in succession have captured more seats from the Liberal Democrats than they have from the Conservatives. Tories won 14 seats from the LDs compared to the LDs 11 seats. In the period 2002-2004 the Liberals consistently won more seats from the Tories than vice versa. Another sign that the Liberal Democrats are on the decline.
Best estimate at the current state of the parties. (prior to these 3 polls shown in brackets)
Con 37.2 (36.3)
Lab 33.1 (33.2)
LD 17.1 (18.7)
Tory Lead 4.1 (3.1)
Chance Tories ahead in votes 99.9% (99.6%)
Chance Tories would be largest party 9% (1%)
Chance Tories would have a majority 0% (0%)
Chance Labour would have a majority 7% (11%)
Chance of Hung Parliament 93% (89%)
Chance LDs would be Kingmakers 27% (15%)
re 61. As a general rule, Rod, ICM usually operate with a 1,000 sample, the Populus Times survey has 1,500 while YouGov and Mori have about 2,000.
Communicate Research, I think, uses a 1,000 sample.
What is more critical than sample size is the date of the survey. Quite often the dispairty between polls can be put down to the fieldwork dates. Mori polls tend to emerge some time after the fieldwork took place. This latest survey closed just two days after Populus and five days before ICM.
67. ‘Another sign that the Liberal Democrats are on the decline.’
Let all the church bells ring out in rejoicing!
69. Thanks, I now have exact sample sizes and dates of all polls going back to Dec 2005. My figures are based on the Kalman filter in which both sample size and polling date are critical to estimating party support at a particular point in time. I’ve submitted an article which tracks Tory support since DC became leader. If Mike publishes it, you may judge on the efficacy of the technique….
If you look just at Lib Dem-Tory contests in 2006 (there were about 70) the swing to the Tories is around 3.4%…on that basis about a dozen Lib Dem seats would be at risk based on Anthony Wells’ notionals - Romsey, Cheltenham, Eastleigh, St Austell & Newquay, Westmorland, Carshalton, Chippenham, Hereford, York Outer, Torbay and Sutton would all fall….with Richmond on a knife edge. Not a bad start…
Cameron has been pushing the word “change” a lot. the Tory party has changed etc. i think this is part of a language strategy whereby he is is using the word to prepare the electorate so that Come the election expect the slogan around the same theme of “time for a change” or “be part of the change”.
This line can also attack Lib Dems, In that if you want change a vote for the Lib Dems is a waste. could put the squeeze on especially if there are signs of a Lib Lab pact being prepared or speculated upon. anyway we will see.
72 So if that’s 12 seats in the bag for the Tories. You only have to win another 120 or so to get a majority next time. Falls to just under 100 if you count boundary changes.
Hmmmm… Warm glow
72. Correction - there have been around 90 Lib Dem-Tory contests, not 70
Well, all polls are hypothetical really, but we do get to test their hypotheses once every four years or so.
So, there are two questions to consider against the various pollster’s methodologies - one relates to the previous thread really.
1) past vote recall weightings. Have the methodologies changed since May 2005? If not, were the past-weight pollsters more accurate?
2) voting intention to named leaders/parties. Was this also done before the real poll, and if so, did it converge with the un-prompted question? Whether it moved towards the unprompted one, or the other way round would also be interesting, but I would expect that would be very much determined by individual leader’s personalities and how party marketing was carried out.
(sorry- I started typing this a lot earlier when there were only about 15 responses so far, but got distracted by pesky work)
re 71. what sample sizes are you using for Mori? The headline voting intention figures are restricted to those saying they are “100% certain” and are usually about half the total sample.
I would love you to run the figures with just the polls that weight to ensure a politically balanced sample and are members of the BPC. That means ICM, Populus, YouGov and NOP if they return to regular polling.
My view of the data you sent me is that it is distorted by MORI with its big ups and downs and your reliance on its large sample - half of which is not used for the headline voting figures. You also include BPIX which is not a BPC member and does not make information available about its methods.
Michael Meacher is creating excitement among conspiracy theory websites:
Propagandamatrix leads with the story, includes a video clip of Meacher’s conspiracy theory and describes Brown as a pawn to the elite who rule the world:
“What are Meacher’s chances of overtaking Bilderberg establishment pawn Gordon Brown and becoming Prime Minister? Slim to none - the British hierarchy would sooner hijack the outcome rather than see a fiercely anti-empire, anti-Neo-Con Prime Minister.”
http://www.propagandamatrix.com/articles/december2006/211206primeminister.htm
26. Andrea and 31. MBoy
So, what you are saying is that if the 4 Labour rebels had also voted for the LD amendment then that would have passed, and thus had a wider impact on UK politics - with the Scottish Parliament officially opposed to Blair’s plan!! I find it highly noteworthy that they chose to back the SNP motion, but not the almost identical LD amendment. These Labour MSPs really do absolutely detest the LDs, and are going to go to any lengths to diminish them or trip them up. To back the SNP is political suicide for anyone in the Labour party - so it shows how pissed off these guys are with the Lib Dems that they chose to back the SNP motion but not the LD one.
To put the scale of yesterday’s Labour Trident rebellion into context: those 4 MSPs (including a minister) represent 8% of the parliamentary party. Imagine if 8% of the Westminster parliamentary Labour Party, including a minister, had rebelled against the government in a vote yesterday. That would have been 29 Labour rebels.
78. Meacher is clearly a fruitcake but can’t see him donning a turquoise tracksuit and shouting about flesh eating lizards just yet.
79. This story from the ‘pretendy parliament’ is strangely reminiscent of the 1980s, when loony left councils used to declare nuclear free zones…
All of the polls show one thing, the Libdems will be the party which will have to do some hard thinking over the xmas period. I’m sure that Libdem MP’s and party bigwigs will be meeting in non-smoke filled rooms to discuss the Ming factor. More than anyone, the Libdems need a vigerous party leader, someone who keeps them in the public eye, a Paddy Ashdown clone. The Liberal Democrats cannot go into the next GE with Ming at the helm! Get those knives sharpened guys, ‘The quiet man is not here to stay’ March is a pretty good month for assassinations, if its good enough for Julias its good enough for Ming.
sorry vigorous
78. + 80. - ha, excellent stuff - we need more posts like this.
I was unaware that Gordo was making waves amongst the Bilderberg lot. I thought his best fwend Mandy was well advanced on that hierarchy.
RE 63, Britspin, question is what happens if Brown gets in and does things people don’t like?
77. Mike, for the current update I used a Mori sample of 1075 - as quoted by yourself - although tracking back through my dataset I notice that previously I have used the larger sample.
The dataset uses polls from BPIX, Communicate, ICM, Ipsos/Mori, Populus and YouGov.
I’ll rerun the data if you like. Which surveys would you like me to include? Polling house bias is very difficult to identify, and there may even be an industry-wide bias.
A reduced dataset will be automatically reduce confidence in any output. The Kalman smoothing algorithm itself quickly discards old data and takes account of peturbations.
79. Stuart Dickson, the “who has voted for” list can be seen here:
http://www.scottish.parliament.uk/business/officialReports/meetingsParliament/or-06/sor1221-01.htm
(go down to “Decision Time” section)
Rod a more accurate way to measure support was suggested by the statistician Graeme Archer at Conservative Home
He said:
+++++++++++++++++++++++++++++++++======
I’ve just had an aha moment in the bath, preparing myself to face the west end. It’s a Christmas present for Tim about the CH “Poll of polls”. I’m not going to rant again, but the basic reason that the CH figure is a useless estimator, as presented, is that all the different polls are lumped together. Now no poll is an unbiased estimator of true party support, by dint of the methodology used to construct the polls. So averaging them all doesn’t remove these distortions, leading to a good (unbiased) estimator with smaller variance than the individual poll numbers, it actually probably distorts the truth by adding all the systematic biases together (every month we probably get a bit further from the truth). There’s no law of large numbers which will make the CH poll of polls as it stands ever trend towards the truth. OK that’s the ranting bit, here’s the aha bit.
If we can assume that the systematic bias within each poll is the same each month, then each poll is giving an estimate of tory support which is T+b (say), where T is the true level of Tory support we’re after (”after” in the sense of “we want to estimate well”) and b is that poll’s systematic bias. If I subtract this month’s poll estimate from last month’s poll estimate, then assuming the bias in each month is the same (which I think is fair since the methodology doesn’t alter, I don’t think, month to month) I get an estimator of the change in support for the Tories which is unbiased.
Now, if I did that for every poll I would have a set of unbiased estimates of the change in Tory support from last month to this.
Since these estimates are unbiased, it would now make sense to average them a la CH poll-of-polls, but we wouldn’t get a (dreadful) estimate of Tory support now, instead we would get an excellent estimator of the change in Tory support from last month to this.
Tim, if you went to your excel sheet and for each poll individually subtract the latest estimate of Tory (or each party) support from that pertaining 12 months ago, and then average them, you could tell us a very good estimate of the change in support for each party over the last 12 months.
John Lewis here I come!
+++++++++++++++++++++++++++++++++++++++
This method of poll averaging gives:
30.5% - (251 seats)
37.5% - (307 seats)
21.6% - (59 seats)
Lab, Tory, LD, sorry
67 & 72
I’m more interested in how LDs have done against Lab in this year’s by-elections, and how that translates into gains from Lab at a putative GE. Anyone care to crunch the figures?
87. True enough, but where do you get your unbiased baseline figure from to add the +/- changes to?
90 The actual figure at the general election was taken as the baseline and the plus/minus shift from the GE to the ICM date was then added.
Then you can add the change in support and average it out for a month to month shift. Does that make sense?
I posted this in the previous thread, but as it mentions MORI it’s probably worth repeating here…
I’m not convinced about this idea of ‘misremembering’. Obviously some people are confused or are sufficiently indifferent to politics for their vote not to register when asked later how they voted. But some people will deliberately say Labour as they will wish to be seen backing the winner last time.
I’m not sure how you can quantify these people, but we all know that they exist - in sport for example the thousands of ‘gloryhunting’ Man Utd (and now Chelsea) fans who have only a passing interest in football but want to be seen backing the winners.
Assuming that there exists even one or two percent of voters behaving in this way (I’d suspect it is more) then it has implications for pollsters. They shouldn’t be treated as real Labour voters and therefore adjusted to fit to make them ‘political representative’.
It’s just one of the problems you get when you move away from a ‘pure’ market research approach. At some point the pollsters assumptions become more important than the data - something that you cannot ever accuse MORI of, which is why I’ll go against the grain on this site and say I’d rather see slightly more volatile, but statisticaly valid polls rather than pollsters best guesses as to how representative the people they talk to are.
30. The best BNP recruiter at the moment is the Labour party. Their immigration policies are proving to be a disaster.
Check this for a recruitment poster:
http://www.telegraph.co.uk/news/main.jhtml?xml=/news/2006/12/22/ngang22.xml
27.
“the third man in Lembic’s love triangle is called Stifyn!”
obviously the common thread that binds the Lib Dems!
Charlie loved to down a Stifyn
Simon ran round to swiftly to sink a Stifyn
Mark did all sorts of strange thinks we’d rather not think about let alone talk about to get a Stifyn
Lembit would rather his lover avoid a Stifyn!
Surely, all we need for Xmas is for the Labour and Tory Parties to adopt one each of the Cheeky Girls as their party leaders? They would be greater intellectually and more honest than the present incumbents and, overnight, there would be no problem with the thought of ‘getting into bed with the Lib Dems’!
94. It would of been great if Have I Got News For You was on tonight with all these Lempit/cheeky stories around
I wonder if HIGNFY is on at all over Xmas?
BB
85. have rerun my dataset again for ICM, Populus and YouGov ONLY.
It gives (with comparison to total dataset)
Con 37.1 (-0.1)
Lab 32.6 (-0.5)
LD 17.2 (+0.1)
Lead 4.5 (+0.4)
Apart from these slight differences, looking back over the DC timeframe, it suggests the Tories low in April was 34.1% (versus 32.7%) and their high in June was 38.7% (versus 39.4%) There is no discernable difference in the trend.
giving… (compared to total dataset)
Chance Tories ahead in votes 99.99% (99.9%)
Chance Tories would be largest party 14% (9%)
Chance Tories would have a majority 0% (0%)
Chance Labour would have a majority 3% (7%)
Chance of Hung Parliament 97% (93%)
Chance LDs would be Kingmakers 38% (27%)
97 Thanks Rod - that’s interesting.
The PBC Polling average based on end month figures has the following Tory lead/deficits. all fairly similar.
Oct-05 -7.0
Nov-05 -4.3
Dec-05 0.0
Jan-06 -1.0
Feb-06 2.0
Mar-06 -1.3
Apr-06 1.0
May-06 6.0
Jun-06 3.3
Jul-06 3.7
Aug-06 6.0
Sep-06 2.7
Oct-06 6.0
Nov-06 4.3
Dec-06 4.3
Time to go offline for Christmas: Hope those on pb.com have a good one and all the best for the New Year - to ‘all’ the contributers. We may have our politcal differences but I always enjoy reading the various takes across the political spectrum.
And Mike; as ever, thank you for providing such an outstanding political site through the year.
Mike, the Samplemiser gives the following figures for the lead on the exact last day of each month(except for this month obviously). This is the filtering method(forward-looking only). The figures in brackets are the equivalents after the smoothing algorithm has “looked-back” and adjusted the data in the light of new information. Both these figures are for the entire dataset of polls….
Oct-05 -6.7 (-5.8)
Nov-05 -5.4 (-2.0)
Dec-05 1.3 (0.3)
Jan-06 0.3 (0.1)
Feb-06 0.3 (0.2)
Mar-06 -1.0 (-0.9)
Apr-06 1.8 (3.4)
May-06 7.9 (6.6)
Jun-06 4.5 (4.7)
Jul-06 4.5 (5.1)
Aug-06 7.1 (5.2)
Sep-06 2.2 (2.5)
Oct-06 5.6 (4.4)
Nov-06 4.5 (4.3)
Dec-06 4.1 (4.1)
Interesting poll work chaps. You all have good Christmass now
Unfortunately Rod, the results you are coming up with are patently ludicrous, aren’t they? Chance of a Tory majority 0%? Chance of most seats 14%? The markets are discounting crazy figures like that.
Well, this is the difference between the statistical properties of a snapshot in time, as compared to projecting results forward.
102. I think Rod’s figures are suggesting that the Conservatives would have a nil per cent chance of an overall majority if the polls were at the same level on the day of the election and had eliminated political bias (ie were only subject to normal sampling bias), rather than that there is no chance of a Conservative overall majority come what may. Any movement in the precentage shares would affect Rod’s ‘chance of x happening’ figures as well.
Interesting stuff, and thanks for doing the maths to put it into an easily readable format.
103 - Indeed. Rod, perhaps you could demonstrate what the results would be given current polling figures and a standard deviation of say 2.5% (or even a range of standard deviations) to allow for changes in opinion between now and the next GE?
Calculating changes of hung parliaments, majorities etc based on current polls is largely pointless because it fails to show prospective swings in opinion from an election campaign - something that Chirac learnt too well. Something people need to remember here is that Cameron’s biggest criticism is “no real policies” and that problem will surely evaporate during a campaign.
104. That’s exactly what I’m saying - or equivalently, it’s election day tomorrow, and these are the final polls… (assumed unbiased)
105. Wouldn’t it be easier to just monitor the figures as they change, and give you regular updates?! I’ve done enough number-crunching for this week!
But again, even if it’s election day today yada yada these probabilities are… outside the bounds of probability, aren’t they?
Chance of most Tory seats at 14%?
Everybody on the doorstep knows that is nonsense.
Or to put it another way from the AP, re: Basingstoke and December by-elections:
“Analysis of nine comparable by-elections in December suggests a projected 14.6% nationwide Tory lead over Labour.
A calculation based on four wards where all three parties fought both times gives a projected line-up of: C 43.4%, Lab 28.1%, Lib Dem 21.5%”
Populus’s chief downplayed on this site the difference likely to be generated by his firm’s methodology as compared with ICM. He reckoned 1% Labour down and 1% Tory up for ICM versus Populus. On that basis either the most recent ICM or the Populus poll are a little out of whack. I think even Mike would concede that ICM have not now got a pro Labour bias in their DNA!
107 - Rod, the number-crunching is much appreciated! However from a betting / predictive perspective saying that the chance of a Conservative majority is 0% is clearly wrong as they could quite easily garner more support in the interim (or indeed lose it). For your (already very useful) work to translate into a betting guide some correction needs to be made to take into account future changes in opinion.
For what it’s worth the obvious plays based on your info as well as other factors (e.g. regional swing) are to back the No Overall Majority at 2.4 and lay Conservative Majority at 3.6 - however it’s more than possible that conventional bookies may chalk up 11/8 or bigger about the hung parliament at some stage so it may be worth waiting.
Check out my “No Overall Majority” thread on the betfair Politics forum for some further thoughts on this subject.
Once again, thanks Rod for all your efforts on this subject. May I also take the chance to wish all frequenters of this site a very Merry Christmas and thank Mike (and those who help him from time to time) for all the effort they put in. I usually only lurk here but it’s always worth a read.
re 68. Rod, how do you calculate your percentages. It seems to be an ideal way of presenting the data for a betting site. I.e. it’s not worth betting on the Tories winning a majority at less than 12-1 etc.
“Analysis of nine comparable by-elections in December suggests a projected 14.6% nationwide Tory lead over Labour.”
I might as well say
“Analysis of partisanship on PB.C indicates at Tory landslide….”
You do it your way, I’ll do it mine. We’ll know who’s right one day…
The problem appears to be some people just conceptually understand chance or probability.
Does it sound better if I say the chance of the Tories having most seats(14%) is greater than the chance of you throwing 3 coins which all land heads-up (12.5%)?
In other words, small, but far from impossible….
Here’s a question. If the Tories were 6% ahead in the polls, and assuming 6% is what they require to be the largest party, what would be the chance they WOULD IN FACT BE THE LARGEST PARTY?
112. That should say “just DON’T conceptually understand…”
110. See my response at 107. For the umpteenth time, I am NOT saying thinks won’t or can’t change. They almost certainly will. I’m just trying to give a probability approach to the CURRENT position. The low probabilities are perhaps better understood as being reflections of the anti-Tory bias in the electoral system, and the current Lab/Con dynamic moving into what I have termed the “Grand Canyon” which comprises hung-parliament territory….
108 - Sadly, Commentator, I think Rod’s figures are accurate (or at least as accurate as the opinion polls, which is as good as we’ll get). I know it’s hard to credit than there are still some Labour voters left, but that’s because you’re a Conservative - indeed a Conservative who’s fairly interested in politics - and as such find the mindset of the died-in-the-wool Labour voter - or even the Labour voter through blind ignorance - fairly difficult to grasp. People rarely change their minds, and people know far, far, far less about politics than even in our most generous moments either of us can seriously grasp. That’s not to say people are stupid, just that they don’t follow politics as a spectator sport. Look at the number of people who don’t realise that a by-election is going on around them, for example.
Moreover, it’s quite hard to credit how seriously stacked against the Conservatives the electoral system is. though we know it, we harbour a hope that come election time the pro-Labour bias currently in the system will be suddenly swept away by a revolution of people wanting to rid themselves of Labour. It might happen. But we can’t really bank on it.
I say none of this implying any criticism, as I know you’re well-informed - I just suspect that on this occasion you’re as guilty of trying to wish away the hurdles as I tend to be.
If it’s of any consolation, if the Conservatives can start to get upwards of 43%, the electoral system starts to swing back to massively favour the blues again (and this is similarly true of the Lib Dems - it’s not an inherently biased system, it’s just weighted towards parties getting around 43% and above).
If it’s any consolation
110 - Aaron “The obvious plays … are to back the No Overall Majority at 2.4 and lay Conservative Majority at 3.6.”
That’s exactly what I’ve done, Aaron, and not because of some sentimental allegiance. I’ve been persuaded by what I think we can now term ‘The Crosby Thesis’. Rod has done us all a favour by providing a well researched statistical argument in support of a view I had reached by more intuitive means. If the Conservatives are to win an outright majority, they have to perform extraordinarily well because of certain technical features of the election and the electoral system. In boxing parlance, they have to knock ‘em out to draw.
Commentator may be aghast but others, like David H and, I believe, Benedict the Blog, appear to have a more realistic grasp of the huge electoral task facing their Party. Its chances of success will not be improved by rubbishing the Crosby principle because its implications are unpalatable.
As regards betting on an HP, I can see the merit in waiting becuase there must be a good chance of a price drift before the GE. Personally I always like to take a good price when I see one. If it drifts, I can always go into the market again.
So far I have staked £330 at 2.4 on an HP with a further lay of £50 at 3.6 against an overall Tory majority. That takes me close to my usual maximum for this type of bet but I may well break the rules on this one as the GE gets closer.
Rod has given the punters here a valuable steer. We would be foolish to ignore it.
111. I’m not sure Rod’s figures are useful in that way, any more than polls in the standard format are useful. Their value lies in translating the percentages at the current level into real-life events - but are in no way predictive of what might happen in 2009 or 2010 except to illustrate how the battlefield lies.
Rod’s data is undoubtedly quality information, but is only of value when tied to a judgement about what is likely to happen over the next three years or so. For example, if you believe that the Conservative lead will still be at about the present level at the election, then 6/1 for the Conservatives to be the largest party would be par for the course, so the current odds would imply the lay to be a good bet. But if you expect the Conservative revival to continue to the 6% Rod quotes in post 112 (if that figure’s right), then it would be better to back.
116. Sorry, should have said “what will happen” towards the end of the first paragraph. They obviously do illustrate what might happen if the shares are the same then as they are now.
O/T - Horses Over The Christmas Period
Apologies to the non-racing types on the site but it has become customary for me to give occasional big race tips here.
Kauto Star should hack up in the King George on Boxing Day but at prohibitive odds he is certainly no bet. Of much more interest is one I have heard very good reports of and which runs in The Welsh National at Chepstow on the 27th. It is *Lou de Moulin Mas* and is still available at 8-1 with Paddy Power, which represents great each way value. Take the price now because the horse is almost certain to run whereas a number of other fancied runners may be withdrawn. Anyway, ‘Lou’ will certainly not be bigger than eights on the day.
Happing Punting…and may the horse be with you!
Annyone hazard a guess at an answer to my question at the end of 112?
119 Rod, I am sure you are going to tell me I am wrong if I say 50% but I’ll put that forward anyway.
You have my sympathy trying to explain probability to people who are only interested in who will win. I experience similar frustration explaining value to novice punters.
Some people just don’t get it.
Without putting to much of a burden on Rod it would be interesting to put the 2004/5 polls ahead of the General Election into his calculation and see how good it would have been.
My guess is that it would have taken out the “noise” of those excessive Labour over-statements from Populus and Mori and come in with Labour 38% to the Tories 32% and the LDs on accurate 23%.
120 - I agree, 50%. The ensemble of results from the same poll methodology in the same circumstance should be normally distributed, so the actual result is equally likely to fall above the mean as it is below.
(Assuming the poll is at the same time as the election - perhaps an exit poll - and there is no systematic error.)
120. Correct-a-mundo!!! In other words, when the polls show you’re “there” -(”there” being whatever, largest party, majority, etc)-, they are really saying you only have a 50% chance of being “there”, - there’s also a 50% chance that you’re NOT “there” yet!!
Some pundits here seem to equate these thresholds with 100% probabilities- they couldn’t be more wrong….
Stick with your intuition, ptp. You have it right…
121. It would be interesting to see what % chance Rod’s methodology gave the Tories of winning 33 seats at the GE based on the final opinion polls as well.
126. It would. Rod?
Rod, Thanks for the work. Interesting. As Peter the Punter suggests I take your analysis as the current position, and accept that. We have not done anything like enough work to win a GE now, or even come close. My optomism is based on looking at how things are likely to pan out over time, leading me to be optomistic albeit with caveats.
Commentator, please bear in mind that the numbers reflect the current poll position.
I woudl be interested, as Mike is, to compare your now refined method to the 2005 GE, Rod. I appreaciate it is Christmass, and you have much else to do. Any time yesterday is fine
(Actualy why don’t you use it as a must writed excuse to get away from the inlaws as and when you need to:) )
damn, something’s stopping me posting.. (test)
Benedict, I don’t think they do. I am more convinced by the simple method of bias-elimination proposed by Rod’s fellow statistician Graeme Archer at ConHome listed in post 87 with results at post 88. That does not give us a majority but clearly indicates a very significant largest seats margin. That’s my intuitive sense, and Rod since he seems to accept the theory of post 87 a few posts down will have to walk me through why it isn’t accurate?
Re 129, Commentator, they are both valid ways of looking at thins and both leads to results. It is not an exact science. We will only know which one is best when an election immeadiatly follows the prediction, and it might not be best next time around.
125. I’m sure. There was an interesting reason why the Tories won more seats than expected in 2005.
www britishpollingcouncil org/john-curtice.pdf
You will notice how JC suggests adopting a similar probabilistic approach to the marginal distrubtion(pages 6-7), as I have attempted with the overall forecast. Perhaps combining the two will be useful. However, on a preliminary examination of the Wells’ estimates, the distribution is much more uniform than it was in 2005. The caveat being that boundary changes make it only an estimate - we don’t know what the real distribution is. So, perhaps there is hope for the Tories after all, that they could do a little better than we estimate. But don’t all get too excited. Even in 2005 the skewed distribution was only worth another 9 seats to the Tories! Apart from this oddity, Curtice explores uniform swing and found it operated almost exactly as predicted….
Benedict I am no statistician. But one method, Rod’s, puts the chances of the Tories winning the greatest number of seats at a mere 14%. The other statistician’s (simple to understand) method gives the Tories 307 seats to Labour’s 251. I was never a maths whizz, but I do know they can’t both be right! And I would love Rod to explain why Graeme is wrong (when you get the time Rod, adoesn’t have to be tonight!).
Commentator I would love to respond, but I hear a pint calling my name, and it would be impolite and most improper to refuse such a call
131. Rod, It was 14 extra seats to the Tories in 2005 compared with the uniform swing prediction, and random variations cannot account for all of them.
Variations from uniform swing since 1987:
1987 CON +2 LAB -8 ALL +4
1992 CON -25 LAB +24 LD +1
1997 CON -43 LAB +24 LD +18
2001 CON -14 LAB +11 LD +4
2005 CON +14 LAB -14 LD 0
Clearly these are some systematic factors here.
129. It’s not bias elimination. It’s simply stating an apparent bleeding obvious, - that the CHANGES from poll-to-poll by a polling house should have NO MORE BIAS in them than ALREADY EXISTS.
I doubt the person you mention is a statistician…
You’ve still got to get an unbiased baseline to work from, and I doubt if such thing exists. If you get your baseline wrong, your overall error could end up larger than if you’d done nothing.
btw, 131 didn’t like the dots in the URL for some reason, so you’ll have to put them back in…
must get ready for my girlfriend now. Merry Xmas to you all, and I won’t forget some of the suggestions on the way forward. Bear with me, it can take hours to do the number crunching. BTW Mike, that article I sent you was two days solid work. It wouldn’t hurt to put it up for 10 minutes in the middle of the night sometime when no-one’s looking, would it?
134 That’s an interesting little table, Kevin L. It suggests CON did a good targeting job last time and LD usually do so, but failed in 2005.
134. It was 8 actually, page 3, para 3, which is the relevant figure to consider for the Lab/Con battleground…
136. Or that the LDs do better than expected when their national vote is FALLING….
136/ The Lib Dems did very poorly in Con/LD marginals - they would have gained 4 on a uniform swing but instead made a net loss of two. This was offset by gains on above average swings in Lab/LD seats. The difference in the Lib Dem performance in the two battlegrounds explains almost half the Conservative and Labour variations.
134. The random factors are real political events on the ground, related to the constituency rather than the national picture. They could include differential turnout in seats held by different parties, tactical voting, the personal vote of a candidate, an in/effective campaign, demographic change and local government factors to name a few. All these would affect people’s votes but wouldn’t be picked up on a UNS ‘map’ and so could appear to be random ‘noise’.
We need to remember that the mathematical models are