## YouGov/Sun – CON 32, LAB 41, LD 8, UKIP 11

The Sun have tweeted tonight’s YouGov voting intention figures – toplines are CON 32%, LAB 41%, LD 8%, UKIP 11%. The Labour lead is up to nine points, the biggest for three weeks or so, and suggests a positive reaction to the conference announcements.

Usual caveats for any poll showing interesting movements apply, especially for the sometimes up and down polls of conference season (some years it doesn’t have much effect on polls, some years they are up and down like a rollercoaster after each conference). Last week we had a narrowing of the Labour lead following the Lib Dem conference, this week a boost in the Labour lead following their conference. Next up we have the Conservative conference – will the Labour conference boost be sustained so they end up the net beneficiaries of conference season? Or will it fade away again once the immediate publicity passes? Or indeed will the Conservatives get their own conference boost next week?

norboldBack from Brighton in time to be first.

Red RagClose yet again – I predicted 40/31 when it was 41/32….got the 9% lead right .

norboldAnyway it’s all go here. We have a council by-election tomorrow as well…..

chatterclassI just don’t know whether the speech would have that much effect. Not because I don’t think it was a good speech-it was. But because most people that I spoke to today hadn’t seen it, or even seen news clips of it.

While the yougov poll covers people across the political spectrum and is weighted, I am not entirely convinced that online polling, of people who have already agreed to be surveyed, is truely representative.

howardI’ve repeated this post as it referred to the thread which did not exist then (!!).

After Ed’s speech (notice how I am apparently cosying up?) I predicted 42 by Friday, so well on course, unless the above posts are spoofs.

Protecting bankers’ bonuses is sure however to be a counter measure that should restore the balance (not!).

Rest assured however, that DC has a cunning plan for next week.

If I knew what it was, it would not be sufficiently cunning, would it?

howardI see Mandy is determined to scotch EM’s revival in its tracks.

http://www.theguardian.com/politics/2013/sep/25/peter-mandelson-ed-miliband-energy-plan

Perhaps DC can just rely on such as he, to do the job for him.

Chuckle.

rosieanddaisieLds are doooooooooooooooooooooooooooooomed.

They have lost their niche, protest position.

James EI doubt whether Mandleson’s disapproval will do Ed Milliband any harm at all.

richard in norwayFluke

R HuckleThe problem polling wise for the Tories is that if they are seen to back business interests e.g energy companies, then this will not be popular with most people. Somehow the Tories will have to combat Labours energy prize freeze, perhaps by setting up an independent review on the energy markets and how to offer consumers the best energy deals available.

The sale of the Royal Mail is also unpopular with many people, who are concerned that prices will increase and services will be reduced. Again more work needed by the government to reassure people, otherwise coalition parties may suffer in polling compared to Labour who are against.

HS2 is another potential problem for the coalition parties, as is the bedroom tax issue. Both of these policies are not liked by voters according to polling. Labour are still supportive of HS2, if the cost can be kept down, but it sounds as if they could change their mind by May 2015.

Labour have picked on a few issues at their conference, which chime with voters. As has been pointed out elsewhere, Labour appear to have been reading the polling, to decide on the issues the wanted to speak about at their conference.

couper2802The Sun never normally tweets a good Lab lead – very strange

richard in norwayI see that the lovely former member for Corby has applied for US citizenship, what a loss for the country!!

Allan Christie“The Sun have tweeted tonight’s YouGov voting intention figures – toplines are CON 32%, LAB 41%, LD 8%, UKIP 11%”

_________

I wonder if this tweet will have an affect on the markets tomorrow?

rosieanddaisieac

“I wonder if this tweet will have an affect on the markets tomorrow?”

What’s your forecast?

ChatterclassMaybe Sun tweeted early cause Mike Smithson of politicalbetting asked them to on twitter, as first poll conducted post speech.

mrnameless“The Sun never normally tweets a good Lab lead – very strange”

—

Murdoch doesn’t like to back a loser. If it looks as though Ed’s on an unstoppable rise by late next year, we may see a pretty huge about-face.

Phil HainesAW is working overtime. New thread, polling average and graphs updated, all by 11pm.

It may in time prove to be no more than a bounce, but in the meantime this satisfies Labour hopes of their conference. And even bounces shouldn’t be taken for granted, for the Lib Dems didn’t get one.

R Huckle“@PHIL HAINES

AW is working overtime. New thread, polling average and graphs updated, all by 11pm.”

Perhaps Anthony has been spurred on by the good poll for Labour !!

mrnamelessEd’s really pushing Murdoch, isn’t he?

http://metro.co.uk/2013/09/25/ed-miliband-calls-for-no-more-page-3-4111307/

Michael ElliottI think that Murdoch himself has hinted that Page 3 might be for the chop (though, even if he hadn’t, I wouldn’t be surprised at Miliband supporting the campaign).

drunkenscouserI’m listening to the Radio 4 midnight news now and all the drivel from Mandelson and Digby Jones. I think the fact someone like Digby Jones could be a minister in the last government showed just how corrupted the Labour party became in the Blair/Brown years.

I have to say I’ve not been impressed with the BBC’s coverage the last couple of days. Their presenters have been rolling out every trite old cliche going. I suppose when someone tries to challenge the dominant ideology of the last few decades the reaction of many of the blathering bobbleheads in the Westminster Industrial Complex was never going to be anything but apoplexy.

I wasn’t around in the 70s but it wouldn’t surprise me if Thatcher and hew followers got similar treatment when they were trying to distance themselves from the post war consensus. These days the Very Serious People say to Ed ‘But how can you win if you don’t doff your cap the Big Leccy’, while they probably wheeled out similar lazy thought bubbles to Thatcher about confronting the unions.

Amber StarAnd even bounces shouldn’t be taken for granted, for the Lib Dems didn’t get one.

———–

Neither did UKIP, which was not what I’d expected. After the ‘fun’ at the May locals, I expected Labour’s conference to be cold shouldered by the media whilst colourful kippers were given all the coverage. That’s not how things turned out & the kipper conference was a bit of a flounder. Maybe it’s Farage who’s carp.

oldnatAmber

Stop posting red herrings!

Roger MexicoFrom the Guardian article already refered to:

Mandelson fears his own carefully crafted legacy of “industrial activism” built up during his two years as business secretary is under threat, and that Miliband’s party conference speech in which he made the energy price pledge was driven by politics as much as economics.I wish I could explain exactly, the many ways in which I find this funny.

Amber Star@ Roger Mexico

I wish I could explain exactly, the many ways in which I find this funny.

————–

I laughed so much I got hiccups.

CharlesIn the last thread we discussed among other things the MOE of ‘the lead’. Someone (I think Martyn or Roger Mexico) has given us the correct formulas.

The formula for MOE of an individual percentage is: 1.96*sqrt((p)(1-p)/n).

The formula for the difference between two percentages (p and q) is

1.96*sqrt(((p+q)+(p-q)**2)/(n-1))

(**2 is meant to be squared!)

So if the sample size (n) is a thousand and the percentage is 35 percent the Margin of error for an individual percentage is around three percent. If there was a dead heat between conservatives and labour the Margin of Error would be around 5.2 per cent.

(Those interested would be wise to check my mathematics which are never as accurate as one would wish. And I apologize to any dogs who may be still awake since I know that distress at such figures. Everyone else will no doubt be lulled to sleep but there is an interest since it suggests if the Labour lead is say 4 per cent we should not be surprised if polls suggest a 1.5 per cent lead or a 6.5 per cent lead which with rounding could easily be 1 per cent or 7 per cent.)

CharlesI should have said “labour and conservatives were tied on 35 per cent”

Michael ElliottAh, cheers Charles (and whoever provided the formula).

Michael ElliottAnyone know whence the “1.96” arises?

Stan JI suppose, in a weird round a bout way, the Lib Dems could get a bounce from the Tory conference.

But I think I’ll go with Rosie’s extensive wall of text analysis and declare..they are dooooooooooooomed.

Martyn@Charles

I’m 60 miles away from my books and so cannot check properly. But I did google your equation and this blog entry http://www.r-bloggers.com/margin-of-error-and-comparing-proportions-in-the-same-sample/ seems to agree with you.

Michael Elliott“they are dooooooooooooomed.”

TLDR.

RichardEveryone keeps suggesting a UKIP/Conservative pact. What about a Lib Dem/ Conservative merger instead. The left wing Lib Dems have already left to Labour, and the right wing Conservatives have left for UKIP.

A new merged Lib Dem/ Conservative party focused mainly on the economy is the only way I can see the conservatives survive long term. Moving to the right and trying to get those that have moved to UKIP back again, even if successful is a doomed long term strategy, given the demographics of UKIP and right wing support.

And its attractive for the Lib Dems as they are dooooooooooooooooomed like in Germany.

Neil AEurope would be a deal-breaker I suspect.

Most Tories, even relatively woolly liberal ones like me, are firmly eurosceptic.

Top HatThe formula for finding the standard deviation ? is ? = sqrt((p-(p^2))/n), provided n isn’t close to the population size (which, for YouGov polling, it won’t be). To find the margin of error for 95% confidence, you then have to work out how many standard deviations corresponds to 95% confidence (it’s 1.96). If you wanted a margin of error for a 99% confidence, you’d use 2.58 instead.

Having said that, you can’t extrapolate directly from the MoE to the lead. That is, if Labour is on 40%, and the Conservatives were on 30%, and just for the sake of the numbers the MoE is 3%, that doesn’t mean you can say the Labour lead is between 4% and 16% with 95% confidence, as the likelihood that all of the error in correctly measuring the Labour proportion of the population will have been attributed to the Conservatives (and vice versa) is particularly low. There is some maths you can do to work out the MoE of the lead to a 95% confidence interval, but because there are multiple different samples it’s much more complicated.

Top HatApologies. The “?” were originally sigma symbols.

Martyn@MICHAEL ELLIOTT

“…Anyone know whence the “1.96? arises?…”A “cold reading” is an anodyne statement that is usually true. So for fortune telling, a cold reading is “you had some trouble when you were in your teens” A cold reading for statistics is “it’s the central limit theorem”.

The answer to your question is “it’s the central limit theorem”…:-)

The CLT is the big thing in stats. If you take a sample from a population, and take the mean of that sample, is the sample mean the same as the population mean? If it isn’t the same, then how different is it? The CLT tells us that if you take a squillion samples, then

regardless of the population distribution, the distribution of the means of those samples is a normal curve.And we know the shape of the normal curveA normal curve has 95% of its values in a band that’s 2*1.96 standard deviations wide. In the middle of that band is the sample mean. So 95% of the time, the population mean will be within a band that is the sample mean +/- 1.96 standard deviations.

Michael Elliott“you then have to work out how many standard deviations corresponds to 95% confidence (it’s 1.96)”

I see…thanks!

Michael ElliottThanks Martyn, too. (I spent the five minutes preceding my last post trying to work out how to get my iPhone to copy and paste one line of Top Hat’s text…)

Martyn@Charles, @TopHat

Or to simplfy things enormously, assume the covariance between the two percentages is zero (it isn’t), and the solution collapses down to “the moe of the lead is twice the moe of the individual numbers”

Or, as somebody pointed out the last time we did this, it should be sqrt(2) instead of “twice”, but b***er it, it’s close enough..:-)

Martyn@Michael Elliot

You’re welcome

rosieanddaisieDad says you mean it’s close enough for jazz.

Top HatIt’s not even sqrt(2) unless you assume you’re equally likely to mistake a Labour voter for a Conservative voter as you are to mistake them for a Liberal Democrat voter. If you take away that assumption, the maths becomes extremely horrible very quickly. However, as you say, the “bugger it, twice is good enough” assumption does help.

Michael ElliottMartyn: thanks also for that explanation of the Central Limit Theorem. I’m fairly ignorant when it comes to statistics.

Martyn@RosieAndDaisie

Tell your dad I always called it “good enough for government work”..:-)

Martyn@TopHat

Fair point.

John PilgrimHOWARD

“I see Mandy is determined to scotch EM’s revival in its tracks.”

He would, wouldn’t he?

There may be a difference in the Labour Conference bounce: the main reaction will be to what came across as real and understandable policy announcements, not necessarily to the speech, which most people won’t have seen, or to the razmatazz.

On the other hand, it may be more difficult – though they will do it nonetheless – for the media to portray EM as indecisive or a weak leader. In that respect, seeing off the old guard may have been just what he needs.

John PilgrimAMBER

“kipper conference was a bit of a flounder. Maybe it’s Farage who’s carp.”

Farage may be a dab hand with the media, but perhaps that was his sole advantage. I think they may have haddock, except for a few floating bloaters..

Charles@Martyn Thank you and apologies for failing to note that it was almost certainly you who provided the formula! I am also sorry that you seem to be as insomniac as I am.

I do actually have the paper that you provided and have checked the result in a different way (which is about as good as having it checked by a six year old but basically seemed to work out). But unfortunately although I have the paper I do not have the web reference that you gave and from which I got it.

My problem in being more helpful is that I can’t seem to copy the formulas into this space and square roots and squares and so on come out looking nothing like their originals. But the relevant bits of English from the paper are as follows:

“The problem is that in political horse-race polls there is usually less interest in the proportion supporting a single candidate than in the lead one candidate has over another,

the difference between the two proportions. When reporters (and political scientists) try to explain this they usually run into trouble because they don’t use the correct formula

for the confidence interval of the difference of proportions.

A voter preference survey question will always have at least three categories and often more. There will be response categories for the Republican and Democratic candidates

plus at least an undecided category. There may also be categories for candidates for third parties and perhaps an “other” category as well. In presidential or other primary

contests, there will usually be a number of candidates, often considerably greater than two. When responses fall into more than two discrete categories the proportions have a

multinomial distribution. …..

The correct formula for the variance of the difference of two multinomial proportions for candidates 1 and 2, p1 and p2, is (adapted from Kish, Survey Sampling, 1965, p. 498-

501) is Var(p1 – p2) =((p1 + p2) – (p1 – p2)**2)/(n-1)”

(**2 means squared! and the formulas I gave above do unless I got the brackets wrong follow from this in a couple of lines I cannot copy )

“Note that it doesn’t matter what candidates 3 . . . k have. We only need the proportions for the pair of candidates we care about in the formula.”

All of which seems to apply to the situation we face at the moment where there are a number of parties and we are basically interested in the difference between two of them.

Which is about as good as I can do at the moment or you perhaps until you get back to your books!

HarryVery odd – if you go to the YouGov site, the latest result is:

Update: Labour lead at 5 – Latest YouGov/The Sun results 24th September – Con 34%, Lab 39%, LD 10%, UKIP 10%;

There’s no sign of the poll AW refers to.