UK General Election - May 7th 2015

[🦀🎂🦀🎂]

https://www.youtube.com/watch?v=m5XoO9RQYtY

If this is the response of Scottish Labour, then we should probably look more at their chances of losing Glasgow North East as well than their chances of regaining the lead in some of the other seats.

First of all, it is quite clearly a lie and one which can easily be shown to be just that. For example by looking at the actions and statements by Labour after the 2010 election.
But also it suggests that no matter how favourable the composition of parliament might be for Labour, they will refuse to form a government if they are not the single biggest party. I'm not really sure they thought this through.

Scottish Labour has a curious way of dealing with nationalists. Namely, they seem to implicitly say that they suck and nobody would choose to vote for them over the Nats ... but you should vote for them anyway to suck it to Tories. Hardly the most inspiring of messages.

[YL]

Speaking of large swings, we have a Survation poll of Sheffield Hallam: Lab 33 LD 23 Con 22 Green 12 UKIP 9.

Take with appropriate quantities of salt, especially the ward crossbreaks, which are hilarious, though given the methodology (no reallocation of don't knows) the headline figures tell a similar story to the other polls of the constituency (including the ICM/Oakeshott one Survation themselves publicly criticised).

Survation's previous constituency polls - before by-elections - have been pretty poor, especially when compared to the Ashcroft ones. I wouldn't read too much into this.

Anthony Wells has written an article on UKPollingReport about the Hallam polls, specifically the Ashcroft and Survation ones and the methodological differences between them.  As he says, most of the difference between their headline figures is down to the reallocation of don't knows, because there are a lot of 2010 Clegg voters in Hallam who are now telling pollsters they don't know.  Survation didn't do any reallocation of don't knows at all, and Ashcroft, unusually, reallocated all don't knows back to their old parties.  As I said above, once you take account of this, the polls aren't telling very different stories.

Wells doesn't mention the ICM/Oakeshott poll.  That poll had fairly similar methodology to the Survation poll, though its question was a little less constituency-specific and its sample size was smaller, and it produced pretty similar figures.

I still tend to think that Clegg will just about hold on.  Many of those don't knows probably will go back to him, he'll squeeze the Tories, the local Lib Dem party is very well organised (though Labour seems to be getting its act together too), the demographics aren't those of a Labour seat, and there's always a bit of a suspicion that constituency polls are prone to dodgy samples.  But I think it's fair to say that he has problems.

[Insula Dei]

Speaking of large swings, we have a Survation poll of Sheffield Hallam: Lab 33 LD 23 Con 22 Green 12 UKIP 9.

Take with appropriate quantities of salt, especially the ward crossbreaks, which are hilarious, though given the methodology (no reallocation of don't knows) the headline figures tell a similar story to the other polls of the constituency (including the ICM/Oakeshott one Survation themselves publicly criticised).

Survation's previous constituency polls - before by-elections - have been pretty poor, especially when compared to the Ashcroft ones. I wouldn't read too much into this.

Anthony Wells has written an article on UKPollingReport about the Hallam polls, specifically the Ashcroft and Survation ones and the methodological differences between them.  As he says, most of the difference between their headline figures is down to the reallocation of don't knows, because there are a lot of 2010 Clegg voters in Hallam who are now telling pollsters they don't know.  Survation didn't do any reallocation of don't knows at all, and Ashcroft, unusually, reallocated all don't knows back to their old parties.  As I said above, once you take account of this, the polls aren't telling very different stories.

Wells doesn't mention the ICM/Oakeshott poll.  That poll had fairly similar methodology to the Survation poll, though its question was a little less constituency-specific and its sample size was smaller, and it produced pretty similar figures.

I still tend to think that Clegg will just about hold on.  Many of those don't knows probably will go back to him, he'll squeeze the Tories, the local Lib Dem party is very well organised (though Labour seems to be getting its act together too), the demographics aren't those of a Labour seat, and there's always a bit of a suspicion that constituency polls are prone to dodgy samples.  But I think it's fair to say that he has problems.

The thing is that 'Nick Clegg could lose his seat' is the sort of line that will inevitably wind up living a life of its own, and that polls like the Survation one will influence the course of the election within the seat itself.

(Btw, iirc even the Ashcroft poll had a generic Labour candidate  up like 10-15 percentage points over a generic LibDem candidate in Hallam, and it was only after being explicitly prompted to consider the situation in their own constituencythat those interrogated broke narrowly in favour of Clegg.)

[EPG]

I can't be bothered to watch any electoral propaganda video by anyone (life really is too short), but it is a fact that the SNP often relied on Tory votes at Holyrood when they were a narrow minority government (2007-11).

But that just tells us that Scottish Labour wanted to bring down the SNP government in the late 2000s, which had nothing to do with David Cameron.

[YL]

(Btw, iirc even the Ashcroft poll had a generic Labour candidate  up like 10-15 percentage points over a generic LibDem candidate in Hallam, and it was only after being explicitly prompted to consider the situation in their own constituencythat those interrogated broke narrowly in favour of Clegg.)

Sort of.  The initial figures showing the Lib Dems on only 17% (and in third; the Tories were on 23%, 10 points behind Labour) were without reallocation and on the "standard" voting question (the one Ashcroft asks first).  Both changes were needed to put Clegg ahead; essentially asking the constituency-specific question caused a swing from Con to Lib Dem without changing the Labour figure much, and the 100% reallocation of don't knows helped the Lib Dems considerably.

In other constituency poll news, we now have one for East Belfast, conducted by LucidTalk (anyone have any idea about their track record?) for the Belfast Telegraph (article here).  Excluding don't knows and non-voters, the DUP are on 34.4% (yay, decimal points...) and Alliance on 28.7%.  Could be worse for Alliance, actually.

[Dr. Cynic]

I'm betting another election before 2020, even if there is a coalition. Possibly even in 2016.

I find it very hard to see that if Labour gets the largest number of seats, that there will be enough LibDems to form a govt. with and if the Nats decimate Labour in Scotland, how willing will they be to have a long term coalition agreement?

[Clyde1998]

In other constituency poll news, we now have one for East Belfast, conducted by LucidTalk (anyone have any idea about their track record?) for the Belfast Telegraph (article here).  Excluding don't knows and non-voters, the DUP are on 34.4% (yay, decimal points...) and Alliance on 28.7%.  Could be worse for Alliance, actually.

Full List without Don't Knows:
DUP - 34.4% (+1.6%)
Alliance - 28.7% (-8.5%)
UUP - 14.6% (-6.6% compared to Ulster Conservatives and Unionists)
PUP - 6.5% (+6.5%)
TUV - 2.8% (-2.6%)
Sinn Fein - 1.8% (-0.6%)
UKIP - 1.6% (+1.6%)
Greens - 1.5% (+1.5%)
SDLP - 0.8% (-0.3%)
Others - 7.5% (+7.5%)
Don't Know - 38.3%

I've never heard of LucidTalk.

Massive 38.3% undecided (or not voting) that could cause a major change in these results. Wouldn't be a major shock if the DUP re-gain this seat though - they'd held it since 1979 before the 2010 election.

[EPG]

Please don't believe any opinion poll of party support conducted in Northern Ireland!

[🦀🎂🦀🎂]

Will Long be scalped by Fleggate then? Sad.

[Clyde1998]

Please don't believe any opinion poll of party support conducted in Northern Ireland!

I'm not really - I just said I wouldn't be surprised if it happened.

Northern Ireland is tricky politically, as people will vote on religious lines rather than policy lines.

[Filuwaúrdjan]

Constituency poll + Northern Ireland poll = LOLOLOLOLOLOLOLOL

[EPG]

Please don't believe any opinion poll of party support conducted in Northern Ireland!

I'm not really - I just said I wouldn't be surprised if it happened.

Northern Ireland is tricky politically, as people will vote on religious lines rather than policy lines.

Ah, that would be fine! It's easier to measure religion than class. The tricky parts are tactical voting, and turnout propensities that are very difficult to measure: hardly any of the usual methods work.

[Helsinkian]

The Telegraph reports that Ashcroft has revised some of their earlier constituency polls done in November, because there was a "mistake in the data".

According to the revised results, Sheffield Hallam was:

Labour 30%
Lib Dem 27%
Conservatives 19%
UKIP 13%
Greens 10%

And South Thanet was:

Conservatives 33%
UKIP 32%
Labour 26%

[You kip if you want to...]

There's been 3 polls of Hallam in the last year, none with Clegg ahead. Maybe Labour actually has the edge there.

[Clyde1998]

There's usually a swing back to the status quo - usually by telling people the things that they've achieved. I don't think Clegg would want to remind people of what he's achieved though...

[You kip if you want to...]

There's usually a swing back to the status quo - usually by telling people the things that they've achieved. I don't think Clegg would want to remind people of what he's achieved though...

Tactical voting is one thing, but when a party leader's only ticket to reelection in the own back garden is tactical voting... well, that's really something.

[Clyde1998]

There's usually a swing back to the status quo - usually by telling people the things that they've achieved. I don't think Clegg would want to remind people of what he's achieved though...

Tactical voting is one thing, but when a party leader's only ticket to reelection in the own back garden is tactical voting... well, that's really something.

It's interesting that Labour are likely to gain the seat - when they've never won the seat before and haven't finished second since 1979. However, the rest of Sheffield are held by Labour.

If there was tactical voting to get Clegg out - you'd think that people would vote for the Conservatives over Labour.



According to the Ashcroft poll (page 10): http://lordashcroftpolls.com/wp-content/uploads/2015/02/Sheffield-Hallam.pdf

The Lib Dem 2010 vote splits:
49% - Lib Dem
24% - Labour
10% - UKIP
10% - Green
7% - Conservative

Labour are gaining Lib Dem votes at a rate of 7:2 compare to the Conservatives.

However, using the percentages (under the 2010 vote column) - around 60% (or ~11,000) of those who didn't vote (~18,200) last time need to vote for the headline figures to be accurate. For the Lib Dems to be ahead - less than 20% (~3,500) need to vote.

Although there's probably another reason for that...
[most likely me multiplying the Lib Dem 2010 votes by the aforementioned percentages wrong]

[Filuwaúrdjan]

But, again, let's not over-analyse constituency polling.

[EPG]

SNP at Westminster will cast votes on nominally-English laws that may have some impact on Scottish national or economic interests.

This is a difficult line for Sturgeon to walk. It is a necessary step towards her party's supporting a Labour-led government, but also a movement away from principled abstention, because practically everything voted on at Westminster can have an impact on Scottish budgets through the Barnett formula.

(It is a bit like the harm principle. Taken too simply, it fails because my well-being can be indirectly hurt by someone else's self-harmful behaviour.)

[YL]

The Ashcroft Hallam poll, and the South Thanet and Doncaster North ones done at the same time, were clearly carried out incompetently, and you wonder what else might be wrong with them.  The correction announced yesterday is, I understand, that the 100% reallocation of don't knows has been changed to something more normal.  Ashcroft has said he won't be using the polling company which did them again.

That said, the messages from the three Hallam polls are all quite similar.  Unless there's some systematic sampling problem in the constituency which they've all had, there's some information in them. FWIW the only constituency poll I'm aware of in Hallam for a previous election was in 1997, and it predicted the big swing from the Tories to the Lib Dems pretty accurately.  The record of constituency polling in the UK since 1997 is mixed, but it's not totally awful.

[🦀🎂🦀🎂]

Has someone calculated how accurate constituency polling has been for by-elections this parliament?

[EPG]

Heywood and Middleton was really bad; Ukip nearly won but that was far from clear from polls.

[You kip if you want to...]

Has someone calculated how accurate constituency polling has been for by-elections this parliament?

http://en.wikipedia.org/wiki/Opinion_polling_in_United_Kingdom_constituencies,_2010%E2%80%9315

Hit and miss.

[Phony Moderate]

Labour are about to wheel A. C. L. Blair out again, it seems.

[EPG]

Perhaps they think he can win back the nuns' vote.