Political polling

[kabbes]

The problem in 2015 is that every single pollster understated the Tory vote share, which they've pretty much always done since the 70s:
General Election: 7 May 2015

5/8 were within 4% of the Tory vote. Isn't that their margin for error? The other 3 weren't that far away either.

So yes, they were underestimating the Tories but not by that huge a margin. I think the problem is as much about the communication and interpretation of the polls as the actual percentages. There is all kind of model and parameter error in there, which will tend to systematically under- or over-estimate certain results, as well as random result error. This type of uncertainty is really not properly communicated. And then the way this gets interpreted as uniform swing is just all kinds of wrong and should really be binned.

 

[bi0boy]

5/8 were within 4% of the Tory vote. Isn't that their margin for error? The other 3 weren't that far away either.

So yes, they were underestimating the Tories but not by that huge a margin. I think the problem is as much about the communication and interpretation of the polls as the actual percentages. There is all kind of model and parameter error in there, which will tend to systematically under- or over-estimate certain results, as well as random result error. This type of uncertainty is really not properly communicated. And then the way this gets interpreted as uniform swing is just all kinds of wrong and should really be binned.

The most reliable information that has been obtainable from the polls for yonks is that the Conservative vote share will be greater than indicated. Of course this time their methodology adjustments might have worked, but I wouldn't count on it.

 

[kabbes]

The most reliable information that has been obtainable from the polls for yonks is that the Conservative vote share will be greater than indicated. Of course this time their methodology adjustments might have worked, but I wouldn't count on it.

Trufax. There still comes a point, though, where reactive systems overreact/fail to account for new circumstance and it all goes the other way.

 

[Dogsauce]

That YouGov data was collected over seven days, so represents quite a long time period relative to other polls - there could have been some movement back towards the Tories since (think I've seen this on one poll).

 

[phillm]

I suspect as ever a lot of tory voters will stay in the 'underground' and not make it known and a good few more will hold their nose and vote Tory on the day.

 

[chilango]

Door to door polling in Bristol NW this evening

 

[littlebabyjesus]

That YouGov data was collected over seven days, so represents quite a long time period relative to other polls - there could have been some movement back towards the Tories since (think I've seen this on one poll).

The way I'm looking at polling is to track each company and look at the trends within their series of polls. Clearly their methodologies and assumptions are different, leading to very different leads for Con, but the trends within each line are relatively consistent.

Looking at the most recent ones here, the drift is still towards Labour with ICM, Youguv, Survation. The very latest one, Kantar, has drifted a tiny bit back to tory after drifting a long way towards Lab, but that's one smallish poll with only a tiny change, so you can't read anything into that on its own. There's no evidence in there really of any drift back to tory - most polling companies are still showing drift to lab.

Person on wikipedia is doing a good job here, btw, updating with every new poll virtually straight away. It's my go-to source for a glance at current polling.

 

[William of Walworth]

Looking at the most recent ones here, the drift is still towards Labour with ICM, Youguv, Survation. The very latest one, Kantar, has drifted a tiny bit back to tory after drifting a long way towards Lab, but that's one smallish poll with only a tiny change, so you can't read anything into that on its own. There's no vidence in there really of any drift back to tory - most polling companies are still showing drift to lab.
.
Person on wikipedia is doing a good job here, btw, updating with every new poll virtually straight away. It's my go-to source for a glance at current polling.

That's an excellent page isn't it?
Thanks for putting it up , I wasn't aware of it.

 

[JimW]

Wait, Keith Allen has a nightclub in Stroud?!

It was news to me too, though I knew he lived local. Bought an old gig venue at the top of town, think he's still doing it up.

 

[redsquirrel]

Sheffield Hallam too! Oh god that would be good

Oh fucking please that would be great (though I echo J Ed's scepticism)

 

[redsquirrel]

One thing that YouGov tool does confirm is that you're a fucking idiot if you're going to vote LD to keep the Tories out.

 

[littlebabyjesus]

One thing that YouGov tool does confirm is that you're a fucking idiot if you're going to vote LD to keep the Tories out.

Unless you're in Richmond. Or some bits of Cornwall. There are a few tory/ld marginals about.

 

[redsquirrel]

Unless you're in Richmond. Or some bits of Cornwall. There are a few tory/ld marginals about.

Richmond possibly, Cornwell - no. Look at the map, not a Con/LD marginal seat. Voting LD either makes it more likely that the Tories will take the seat, or the seat is such a safe Tory one that it's irrelevant - in which case why vote for yellow filth.

EDIT: On Sheffield Hallam, while I still think Clegg will survive, one thing that does need to be considered is that he was saved in 2015 by Tory voters (5000 votes down on 2010), and helped by the fact that the Tories basically didn't run a candidate against him. This time the Tories will be running a campaign and I can see some of those Tory voters going back to blue, so if Labour could keep the vote it got from 2015 then he might be in some trouble.

 

[Raheem]

 

[JTG]

Richmond possibly, Cornwell - no. Look at the map, not a Con/LD marginal seat. Voting LD either makes it more likely that the Tories will take the seat, or the seat is such a safe Tory one that it's irrelevant - in which case why vote for yellow filth.

EDIT: On Sheffield Hallam, while I still think Clegg will survive, one thing that does need to be considered is that he was saved in 2015 by Tory voters (5000 votes down on 2010), and helped by the fact that the Tories basically didn't run a candidate against him. This time the Tories will be running a campaign and I can see some of those Tory voters going back to blue, so if Labour could keep the vote it got from 2015 then he might be in some trouble.

I think Clegg's gone. I really do

 

[JTG]

CON: 42% (-1)
LAB: 39% (+3)
LDEM: 7% (-2)
UKIP: 4% (-)

YouGov 30/31 May

 

[chilango]

I think Clegg's gone. I really do

Is he 2017's Portillo moment?

I think I said that in 2015 too

 

[J Ed]

According to Goodwin, anything under 9 is potentially hung parliament territory.

 

[JTG]

Is he 2017's Portillo moment?

I think I said that in 2015 too

What's been said is right - he was saved by Tory voters in 2015, this time a few of those will go back to the Tories and he'll be sunk by the rising red tide

 

[J Ed]

Is he 2017's Portillo moment?

I think I said that in 2015 too

Christ I remember walking back through Sheffield city centre in 2015 after seeing the exit polls, thinking to myself that at least Clegg might lose his seat

 

[chilango]

What's been said is right - he was saved by Tory voters in 2015, this time a few of those will go back to the Tories and he'll be sunk by the rising red tide

Yes I did...

Was going to post this in the polling thread, but thought it merited its own thread.

So, which "big names" could lose their seats in May?

Obvs the dream is Clegg. But are there any others to hope for?

 

[bemused]

YouGov is either going to look like geniuses or half-wits in just over a week.

 

[J Ed]

Look at Matthew Goodwin's twitter. He reckons Corbyn can do it. Matthew Goodwin (@GoodwinMJ) on Twitter

 

[JimW]

If Labour go over 40% will Hilary Benn and the rest do the decent thing with the brandy and service revolver?

 

[JTG]

If Labour go over 40% will Hilary Benn and the rest do the decent thing with the brandy and service revolver?

No, they'll suddenly become available for Cabinet positions once more and demand their 'experience' is called upon

 

[agricola]

No, they'll suddenly become available for Cabinet positions once more and demand their 'experience' is called upon

They'll get them too, if he wins.

 

[redsquirrel]

I think Clegg's gone. I really do

I hope you're right

 

[JTG]

More YouGov:

On which party has the best policies for you and your family:
LAB: 35%
CON: 29%
LDEM: 6%
UKIP: 4%
[DK]: 26%

On which party has the best policies for the country:
LAB: 33%
CON: 32%
LDEM: 6%
UKIP: 4%
[DK]: 25%

"Don't know" quite high in those but Labour would appear to have convinced a far higher % of their voters that they've got the goods than the Tories have

 

[free spirit]

The You Gov poll is now showing Labour in the lead in London, the north and now the midlands & Wales.

The Tory support is heavily concentrated in the 'rest of the south' region where they're beating Labour 38% to 25% (overall figures), and Scotland where this won't count for anything because the SNP are trouncing both parties.

So Labour are now in the lead in 3 out of 5 regions, the tories only in one (though it's a big region).

Also Corbyn is now preferred as Prime Minister by Lib Dem voters both past and present, whereas they were heavily in favour of May at the start of this election.

And there's been a general reduction in the tory standing on all issues, along with a big swing in the proportion who view Labour campaign as being more positive than negative, and the reverse for the Tories who've now viewed as running a negative campaign.

And this bit's looking pretty good for Labour as well.

 

[JTG]

This is interesting:



Remainers almost completely uninterested in the Liberal Democrats