Private Populus poll shows Labour’s lead evaporating
Under the British Polling Council’s disclosure rules Populus have released the full table for the Conservative party’s private polling that was mentioned in the Telegraph this morning. The topline results, with changes from the last Populus poll, are CON 36% (+3), LAB 37% (-2), LDEM 16% (+1). It was carried out between the 25th and 28th August.
It is obviously a sharp drop in the Labour lead and, for the Conservatives, this is back to the sort of levels of support they were recording in Populus polls prior to Gordon Brown becoming leader. The only reason they are not ahead is that Gordon Brown has won back some of the support Labour had lost to the Liberal Democrats.
Private polls for political parties carry a certain mystique, not entirely deserved. Mainstream political parties don’t generally commission polls to deceive, they are after accurate polling information to inform strategic decisions about what messages to run with, what demographic groups and areas to target. They are after accuracy, not flattery (and if they wanted someone to tell them they were doing wonderfully regardless of reality they wouldn’t need to cough up hundreds of thousands of pounds a year to do it.)
Prior to the BPC disclosure rules you had to take internal polls leaked or briefed by political parties somewhat sceptically - after all, you never knew for sure of they were what they claimed to be - had they used different wording? Or had different weighting for turnout applied that we were used to? Exactly when had they been carried out? In this case, Populus have released the full tables so we know exactly what we are looking at - the only substantial difference is that there is no sign of Populus’s normal topline adjustment, and that wouldn’t necessarily be reflected in the tables anyway (and doesn’t necessary make the slightest difference to the topline figures). This poll can be treated pretty much in the same way as a normal Populus poll.
So, why such a contrast between the Populus and YouGov polls? Well, we’re used to some differences between their figures - no other pollsters has shown the sort of 9 and 10 point Labour leads that YouGov reported - but there must be something else here. Both polls were carried out over the bank holiday weekend and they are notorious for producing rather strange samples (many readers will remember the Populus tracker poll from the last election which carried on over a bank holiday weekend immediately before election day and produced a ridiculously large Labour lead because of a strange bank holiday sample). The Populus poll was also conducted marginally later than the YouGov one - the sample periods are similar, but the majority of responses to YouGov polls are received on the first day, so in effect many responses to Populus’s poll will be from a day or two later when the effect of the reaction to Rhys Jones’s murder in Liverpool would have begun to have an effect.
More simply it could just be that one of the polls is an outlier. As luck would have it we’ve got another Populus poll due next week, this time for the Times (Communicate and MORI also have polls due). IF Populus’s Times poll echoes the findings of this poll then firstly, I think we would be able to forget any possiblity of an autumn election, and secondly it looks like the Conservatives are back in the game again.
Filed under: Populus, Voting Intention
Well done Anthony for your successful efforts in getting the figures released.
Well done Anthony but note the sample size was only just over 500 , normally it is around 1,500 .
Very well done, Anthony.
Mark, I thought that too when I first looked, but there are 524 people with voting intentions, after all the turnout weighting and the exclusion of don’t knows and won’t votes. I think it’s actually the equivalent of a sample size of 1000.
Anthony the last normal Populus poll has 854 Con/Lab/LibDem voters from a sample of 1,511 , this poll has 473 from a sample of 836 on the same proportion .
Surely this poll more than YouGov should be treated with caution, being conducted over a bank holiday weekend?
Trying not to be cynical but has there ever been private polling for a party that doesn’t show them in a better light.
The Conservatives would not let their man in the Telegraph mention their private polling if it was disastrous. Plus would the party keep them on as private pollsters if their polls were not favourable.
Well done Anthony.
At the very least, its food for thought for Brown/Labour.
We wait for next weeks polls with great interest.
Mark - but who uses a sample size of 854?
John - they were both done over the bank holiday weekend. Perhaps YouGov were a bit less vulnerable with a lot of the work done on Friday, but if there really is an effect on sample quality from bank holiday weekends then YouGov wouldn’t necessary have escaped - e.g. people setting off for a long weekend on Friday evening rather than sitting on their computer, people doing their packing rather than checking email, etc.
Tony - no, they don’t let those ones out
(though I’m told this one did escape by accident rather than being deliberately leaked). And yes, they would keep them on. Parties pay pollsters hundreds of thousands of pounds a year for accurate results, not flattery. Flattery would be far cheaper.
Tony: ‘The Conservatives would not let their man in the Telegraph mention their private polling if it was disastrous.’
But that doesn’t make the poll invalid or a rogue. What might is while most pollsters have Labour 5/6% ahead this is a reverse YouGov result.
Well done on getting those results so fast Anthony.
[...] for Labour 31 08 2007 A new opinion poll from Populus (hat tip: Conservative Home and UK Polling Report) commissioned by the Tory Party shows that Labour is on 37% (-2), Conservatives 36% (+3) and Lib [...]
Anthony , of course noone picks a sample size of 836 , the aim was presumably a sample of 1,000 but the bank holiday polling probably reduced the response figure to make it an even smaller sample than expected .
Incidently I was asked to take part in the Yougov poll on Wednesday so unless another Yougov poll is in the offing , the polling went on well past the bank holiday .
There might well be another poll for someone else - if you got the same questions then it probably just means someone forgot to close the survey! The data was downloaded on Tuesday so no data after that would have ended up in the results.
Don’t know if all the questions were the same but I did not even receive the Email asking me to take part till Wed evening . There were a number of strange multi choice questions some of which none of the possible answers fitted my desired response !!!???
The emails all go out at once when the survey goes live, so unless your email is playing up that’s an entirely different poll.
No don’t have a problem with my Email so sounds as if it is a different survey to the DT one .
Er, aren’t we ALWAYS told never to trust a poll that is conducted over the August bank holiday weekend? It regularly turns up bizarre results, which is why pollsters now generally avoid it.
This poll is to be confined to some dark shelf to gather dust, until or unless a proper poll corroborates it.
Apologies for a cross-post, but I’d be interested in Anthony’s view of a comment I’ve made on pb.com:
Some important new light on that Populus poll in the Sunday Times (Peter Riddell):
“The leaked Populus poll was part of a rolling, tracker poll that is updated every day or two. The Tories seemed to have leaked a particularly favourable result for them, given their desire to avoid an early election.”
pb regulars may remember the problems with tracker polls in the past: if this is correct, the result needs to be taken with a large pinch of salt, as it exploits the margin of error. If the real lead is say 4%, then a daily tracker may vary from 1% to 7% - if you then take the most favourable finding, you give a false impression. I’m not sure, if that is the case, that Populus has met the spirit of the BPC guidelines by presenting it on their website as a normal poll.
There’s nothing intrinsically wrong with something from a tracker poll - the problems in a traker poll are largely that if you get one days duff data then it sticks around for three days polluting your poll. Here we don’t have the luxury of that problem
The margin of error on a tracker poll with a sample size of 750 or so is exactly the same as the margin of error on a standalone poll of 750 - the fact that its part of a tracker poll is irrelevant. It’s just a poll with a relatively small sample size.
It could indeed be just a jolly good poll for the Tories out of a series of less favourable ones - in one sense it almost certainly is, after all, we can assume that Populus’s private polls have been showing pretty much the same as their published ones so the Conservative private polling in the past month has probably been showing a substantial Labour lead.
What this probably isn’t is a blip that went straight back down again - one good poll plucked from the middle of a series of bad ones - simply because when it scampered out of CCHQ it would have been the most up to date data they had available: fieldwork finished on Tuesday, Populus knock up some tables for them on Wednesday, Benedict Brogan knew about it on Thursday.
We haven’t got long to ponder about it anyway, given there is Populus poll of normal size and standard timings due in a matter of days.
Odd sample but still can squeeze out something of note.
The breakdown by sex
Tory:
Male 34%
Female 37%
Lab:
Male 36%
Female 37%
Lib:
Male 16%
Female 16%
It cannot be coincidence Cameron is young and good looking and is ahead on the ladies count. More impressive because ladies are typically socialist - Brown also is at 37%.
This says Cameron is not macho enough for men, or at least when polled men don’t want to admit to voting for Cameron. 4% of the male respondents chose BNP.
This latest private poll for the Conservatives may or may not be totally accurate - but it does fit in with the general trend in all the polls of a closing of the gap between Labour and the Tories as i predicted would happen during September and have the Tories in the lead in the polls by October at the latest. Putting to bed finally the notion that Brown would call an election this year - I still believe it will be 2009 at the earliest.
The average of all 3 polls published is a Labour lead of 4.6% - that is a considerable drop from say 2 to 4 weeks ago and prior to the Cameron headlines on crime and immigration .
NICK PALMER MP :-
I can understand you being very concerned at the Conservative Party private poll - with only a 2% lead at the last election in Broxtowe you are in a very vulnerable seat indeed . If the polls saw a breakdown of say
C.37% (+1) L.35% (-2) LibDdem 18% and that was reflected in a general election you would lose your seat !!
As the ‘ping’ from my blog at 8.18pm indicates, the Populus poll, according to my analysis of the regional variations, shows major Tory advance in the South and Midlands and Labour not really moving forward anywhere apart from Scotland, does not augur well for a snap poll - despite all of the rumours elsewhere of an election being called on Tuesday.
Anthony , I am a bit confused by your last post . You seem now to accept that the Populus Conservative poll is part of a regular tracker though no less valid for that reason .
However you cannot know it was a blip amongst a series of bad ones because it was released before presumably the figures from the next 1 or 2 polls in this series were known .
Of more interest to me is - Is this tracker part of Populus normal polling procedure paid for by themselves od by the Conservative Party . If the former , did someone at Populus then go to the Conservatives and say they had a poll favourable for them , would you like to buy it . Would that scenario be part of the normal polling industry scene ?
Mark, that’s exactly what I mean - when it was released there weren’t any polls after it so while it could have been a blip going down straight afterwards, no one could have known!
No, that wouldn’t be part of the industry at all. YouGov sometimes do regular internal polls for panel care purposes - we need to try and keep panellists enthused and happy - but apart from occassionally testing things or doing something on spec to land a client it would be madness for a non-panel pollster to do polls without a paying client, each one would be like burning a couple of thousand quid for no gain. Doing it as a tracker would be certifiable, burning a couple of thousand quid every day or two for fun.
Populus are the Conservative party’s regular contracted pollster - they’ll be doing this every day or two on a long term contract. Michael Ashcroft is in charge of Tory polling and this is the same sort of daily poll he had Populus carry out for him privately prior to the last election.
Mountjoy - regional breaks are small enough on a normal sized poll. On a poll this size they are minute, you can’t draw and reliable conclusions at all from the breaks.
Anthony , thanks for that clarification , though Mori do their monthly polls without a specific client and then usually but not always successfully manage to sell them .
MORI do it as a shop-window. Sometimes those polls aren’t sold, they are given free. MORI do it for the publicity, to try and get a new contract and to make sure their 20-odd year record of unbroken monthly polls remains unbroken.
If Populus were doing this them would be doing 30 times more polls, unlike MORI they don’t have their own fieldforce, they need to pay ICM for each one. They clearly aren’t getting publicity out of it, since it is secret and they don’t need to use it as a shop window to land a client because (a) they’ve already got the Times (who will have an exclusivity deal) and the Conservative party and (b) they’ve already got the Times as a shop window.
Ok fair comment but this raises a number of points .
The poll is as you say in fact one of a series of trackers paid for by the Conservative Party and normally kept secret .
Why leak this one and not the previous one or the one before that . Will they leak the next one or only ones that are favourable ? Apart from that it is likely the previous trackers were not as good for the Conservatives as this one , why leak a good one . To boost Conservative morale ? To try and reduce the enthusiasm for an early general election ? To reduce criticism of Cameron ?
[...] think this is plausible for a whole host of reasons but there is a very interesting blog post by Anthony Wells at UK polling report that gives a very interesting picture. In essence he seems to be saying that private polling by the [...]
Well I’ll put in my standard North of the Border contribution with the normal caveats about samples size (a paltry 53).
Again another indication of the SNP at over 30% compared to under 20% at the last election, It’s also I think got Tories down at 8% which would see there vote halved….
Now I don’t want to get carried away but given that it has been constant since the Scottish election in May there is the possibility, and I’d stress only that, that there has actually been a real shift in Scottish politics.
For over two decades the line from Labour, Tory and LibDem has consistently been, if the SNP ever formed a government then the earth would rupture and Scotland would slip in to the void.
Well it hasn’t happened and in fact people are saying,
“Hey these people aren’t bad and there actually better than the guys who said they would bring about the apocalypse”
So the question is,
“Has the difference between the reality of the SNP in power and the dire warnings not so much boosted the SNP as undermined the credibility of the opposition parties”?
Oddly enough the new (New) Labour leader in Scotland Wendy Alexander is quoted in the press today as being shocked by the depth of feeling she has encountered from Labour members about the lack of affordable housing. (She’s on a Brown style listening tour where she lectures people).
Now that for me is pretty much an indication of how far New Labour in Scotland has lost touch with it’s own core support in following the Blair agenda.
We also had the Labour party commenting on a report that figures from the last decade may have understated growth in Scottish services (while the UK leisure and retail sector grew by 25-35% over the figures for Scotland was stated at between 0-5%) to show that the SNP message on slow Scottish growth is a myth.
What they haven’t been able to answer is the question of why given 11 years of Labour government and 8 in coalition in Holyrood they never noticed the flaw in the statistics they were producing.
I don’t mention these to score political points but rather to show the nature of the problem, I am not sure about.
Have the SNP got off to a very good start (The Nationalist Bounce) while the Scottish opposition parties are still all over the place, or has there been the once in a generation shift.
Regardless of what the polls are showing I have a feeling we won’t know until the SNP face a real challenge or the election.
On election night in May Alex Salmond said, Scottish Politics would never be the same again, so who thinks he was right?
Peter.
Mark - the Conservative party claim they didn’t leak it deliberately…well, erm….I think everything obvious from here. Private polls don’t normally show much different from published polls, so their previous polls would have been showing the same sort of figures as Populus’s newspaper polls. If they did leak it then they wanted more positive coverage and to boost morale - “don’t worry about the YouGov poll, our tracker shows that we’ve turned the corner and we’re catching on them”. It ain’t rocket science.
I doubt we’ll see any more private polling, but Populus’s Times poll on Tuesday will tell us if that was just a blip.
Labour will hold on May 2008 the General Election or Even May 2009?
Wales. Has anyone noticed this?
Very interesting.
Unless I’m very much mistake, Tories are AHEAD of Labour in Wales with a much better vote share there than Northern England. Very surprising.
However, this trend isn’t just speculation. It seems to be borne out by the facts. The Welsh Assembly elections recently had some very promising results for the Tories, with around half-a-dozen near misses. In surprising places like Gower and Newport too.
It could be taken as evidence that Tories are no longer “toxic” in Wales, as they obviously still are in urban areas of Northern England, and have made a comeback - even though they’ve dissolved to nothing in Scotland.
It could be that Wales is becoming more affluent, prosperous and middle-class and finally swinging right after being a Labour stronghold for decades.
Rob B . I have no idea where you get the wrong impression from that the Conservatives lead in Wales . Most pollsters do not split out Wales from other regions . The only one that does is Mori . The figures from their June/July polls on very small samples so M of E is very high give Lab 44 LibDem 17 Plaid 15 Con 15 in Wales . The one clear thing you can say is that Labour are well in the lead in Wales .
This was posted in the political-betting blog;
http://observer.guardian.co.uk/politics/story/0,,2160780,00.html
The Observer thinks next weeks poll will carry on from the Tories internal Populus poll and show an upturn in Tory support.
I got it from the detailed breakdown on the populus poll:
http://www.populuslimited.com/uploads/download_pdf-280807-The-Conservative-Party-Voting-Intention.pdf
Yes, try not to be too smug, you are right on the breakdown of the poll.
I misread “Wales and Southwest” as “Wales” and the sample base was quite small, but there is still quite a lot of evidence the Tories are doing rather well in Wales and have recovered substantially since 1997, with strong potential for growth in future.
I can see Wales returning a similar number of Tory MPs in future as in the 80s/90s (ie. 10-15) where there seems little chance of any recovery ever happening in Scotland, and little evidence of progress in Northern England, so far.
I think you only need to look at the last elections in Wales this year - the Tories pushed the Liberals right back. Labour are no longer the main party there - they certainly don’t have enough seats in the assembly to have overall control .
As for Scotland - where on earth people are getting figures stating the Tories are a spent force in Scotland and North of England . The Tories made significant gains in votes this year in Scotland . In the North of England the Tories grabbed one council after another / in my own constituency Stockton South - Labour were virtually anihilated in the council elections and in the 2005 general election their majority was halved !!
Remember - Stockton South is one of the deciding constituencies when the Conservatives win an election - if they win that seat they’ve won the election !
Mike , you may be correct about the recent Welsh Assembly elections but Rob was saying the regional breakdowns showed the Conservatives in the lead in Wales which is manifestly untrue . I would place no faith whatsover in these regional subsamples but if as 1 or 2 Conservative posters on here they are all important , then they show Labour well in the lead in Wales .
Incidently the regional figure for Scotland on the Comm Res latest poll has LibDems on a paltry 8% in Scotland and the Conservatives on an even worse 4% .
Mike - I’ve already admitted I got it wrong. I wasn’t try to be misleading, I just misread the statistic.
Why do you always feel the need to rub it in and be all pleased with yourself?
I disagree with you about all regional subsamples being worthless too. Yes, they are more inaccurate, and have a larger +/- error, but if they show clear trends, they are a useful indicator of what’s happening in the regions.
I’d be very wary about regional breaks. They have small sample sizes and polls aren’t weighted within region, only for the country as a whole. There might, for example, be far too few older people in the London part of the sample but far too many in other parts of the country, balancing each other out. The sample as a whole is representative, but the breaks aren’t necessarily. If you want to try and discern regional trends the best way to do it is what ICM and the Guardian did a week or two back, combine a large number of recent polls together so you’ve got big enough sample sizes to give meaningful figures.
With a bit of jiggery pokery you can extract separate figures for Wales & SW:
Wales: Lab 40%, PC 24%, Con 20%, LD 8% oth 8%
SW: Con 37%, Lab 31%, LD 24%, oth 9%
sub-subsample so even more care should be taken, but these look about right.
I remember two very inaccurate London polls.
Before June 1987 41% Con 36% Labour
which turned out 46.5% to 31.5% in London.
Then in the 1990 local elections, a poll came out just before saying there would be a 9-10% direct swing from Conservative to Labour across London compared to May 1986. In fact, both parties increased their votes compared to the 1986 locals and there was a swing of about 0.6% to the Tories.