Political polling

[kabbes]

Confirmation bias, or the "halo and horns" effect is a well established psychological trick we play on ourselves.

 

[butchersapron]

Eh, even Ashcroft has discussed the concept that the Tories are basically in wish fulfilment mode and will jump on any positive news,

I can't remember his exact phrase, 'conformation bias'?

It's comfort polling as Santino said - and he aimed it at Tories who fall for it. He wasn't applauding it.

 

[bi0boy]

Rogue poll, obviously.

Obviously?

http://fivethirtyeight.com/features/heres-proof-some-pollsters-are-putting-a-thumb-on-the-scale/

 

[danny la rouge]

http://www.tnsglobal.com/uk/press-release/snp-increases-its-lead-in-latest-tns-poll

"More than half of adults in Scotland who are certain to vote in the May 7 general election (52%), said they would vote SNP, against 24% backing Labour.

In the event of a hung parliament, 44% of Scottish voters would prefer a Labour-led government against 15% backing a Tory-led government. The most popular single option is a Labour-SNP coalition


A sample of 978 adults aged 18+ was interviewed across Scotland over the period March 18th-April 8th 2015."

 

[killer b]

Obviously?

http://fivethirtyeight.com/features/heres-proof-some-pollsters-are-putting-a-thumb-on-the-scale/

interesting piece, thanks.

 

[emanymton]

I know someone who is transferring their vote to their 2nd/holiday home in order to vote LibDem in a seat where they might win.

When you say 'know' I trust it is a euphemism for 'beat to death with a shovel'?

 

[weepiper]

It has 35% for con in scotland and 12% for lab - poll of 44 people. So 16 people against 5 people.

well that's clearly bollocks. Too small a sample size. Tories consistently get around 15% here. It's hardly changed at all - all the movement has been between Labour, the Libs and SNP. This post is more directed at treelover than you butchers but your post has the numbers in it

 

[belboid]

well that's clearly bollocks. Too small a sample size. Tories consistently get around 15% here. It's hardly changed at all - all the movement has been between Labour, the Libs and SNP. This post is more directed at treelover than you butchers but your post has the numbers in it

it was a poll of 90, breaking 16-9 - about right in terms of proportion of Scots to include in a UK wide poll, but quite useless in getting meaningful cross-tabs for any individual part of the UK. According to the poll there are 0 LibScum in Wales, but I, sadly, expect that not to be the case.

(full tabs)

 

[butchersapron]

**

 

[Dogsauce]

Why would anyone's morale be boosted by a blatantly flawed poll? They aren't idiots.

Because if it's played up enough in some bits of the press it might make a Tory victory look possible, and those who'd jumped ship to UKIP on the basis that the Tories were on a loser anyway might see some point in voting for them again.

On the other hand it might motivate Labour voters/supporters to avoid a Tory govt, so this is just speculation.

 

[danny la rouge]

**

Keep the poll in the freezer for a month?

 

[brogdale]

Calm down dears....

Labour and the Conservatives are tied at 33% in this week’s Ashcroft National Poll, conducted over the past weekend. Both parties are down since the last ANP two weeks ago (Labour by one point, the Tories by three); UKIP and the Lib Dems are each up three points at 13% and 9% respectively, with the Greens down one at 6% and the SNP unchanged at 4%.

 

[brogdale]

....the daily YouGov poll for the Sun has topline figures of:-

CON 33%, LAB 34%, LD 8%, UKIP 13%, GRN 6%
– a one point Labour lead. Putting all four polls together that ICM poll looks very much like an outlier...

Here.

 

[Fez909]

Here.

Did you read bi0boy's link?

 

[brogdale]

Did you read bi0boy's link?

Yes.

 

[brogdale]

Ashcroft's latest marginal polling takes a "..look further down the Conservative defence list to see whether there could be some surprises in seats which ought, on paper, to be safer for the Tories."

He finds that...

I found the Conservatives ahead in five of the ten seats: Cleethorpes (though by only by two points), Dover, Dudley South, Harlow and North East Somerset, where Jacob Rees-Mogg is sixteen points clear of Labour despite the Liberal Democrat vote falling by more than half.

However, I found ties in Rossendale & Darwen (where the Tory vote was unchanged since 2010 at 42%, but Labour were up ten points) and South Ribble.

I found Labour leads in three seats, albeit well within the margin of error: Crewe & Nantwich (by three points), Finchley & Golders Green (by two points) and Milton Keynes South (by two points).

 

[butchersapron]

...and the 3 labour gains are target seats #69, 86 and 89. Av swing 3.5% tory to lab across the polled seats.

 

[butchersapron]

Obviously?

http://fivethirtyeight.com/features/heres-proof-some-pollsters-are-putting-a-thumb-on-the-scale/

The opposite to herding is happening in this elections polling - we're having large leads for labour side by side with larger leads for the tories. And the UK pollsters have been pretty accurate on UK elections. Whilst that author's attempts at UK polling saw him scurrying back over the water sharpish.

 

[butchersapron]

One of these marginals that ashcroft has labour leading in being thatchers seat of course.

 

[brogdale]

One of these marginals that ashcroft has labour leading in being thatchers seat of course.

Yep, but with slight (but significant) boundary changes since 1997. Lost some Northern wards and gained some more southerly ones from Hendon...making it more marginal.

 

[nino_savatte]

Why is it that a number of Tory politicians are being paid by polling companies?
Philip Davies has been paid for his services on 18 occasions by ComRes.
http://searchthemoney.com/profile/156/8?p0=1#tabsx-1

Priti Patel too.
http://searchthemoney.com/profile/481/8

ComRes isn't the only company to do this, YouGov have paid Davies over £1,000 over the course of 4 years.

Are Labour MPs also working for polling companies? How does this affect objectivity and, more importantly, validity?

 

[The Boy]

How does this affect objectivity and, more importantly, validity?

Unless they're being paid to actually carry out field work or number crunching, then not very much.

 

[chilango]

YouGov have paid Davies over £1,000 over the course of 4 years.

That's one fuck of a lot of surveys he must have filled in.

 

[brogdale]

That's one fuck of a lot of surveys he must have filled in.

 

[brogdale]

Guardian piece on different results produced by phone and online polling.

http://www.theguardian.com/politics/2015/apr/13/telephone-internet-polling-different-results

 

[brogdale]

Lighting failure, broken-down bus....and now this! Quite a day for the LDs.

The latest ComRes / ITV News Battleground poll has focused in on the Liberal Democrat heartland – 14 currently Liberal Democrat seats in the South West (where their Coalition partners, the Conservatives, are in second) - and reveals a 13-point swing from the Liberal Democrats to the Conservatives. If a 13 point swing was seen in each constituency, it could hand the Conservatives control of all of these seats.

In further bad news for Nick Clegg, three in ten voters in these seats (29%) say that Mr Clegg puts them off voting for the Liberal Democrats. Strikingly, one in three voters in these Liberal Democrat seats (32%) say that the party has become irrelevant, including one in five (21%) people who voted Liberal Democrat in 2010.

 

[brogdale]

Really interesting post from YG's Anthony regarding the discrepancy between the ComRes (LD wipe-out) LD -> Con swing of 13% and Ashcroft's constituency polling pointing to a 4% swing.

 

[brogdale]

 

[brogdale]

...and, as Smithson says...

After the debate about phone polls and internet ones today’s Ipsos-MORI survey for the Standard has LAB up a touch and still in the lead. This contrasts sharply with. Monday’s ICM phone poll with its 6% CON lead and the ComRes phone poll series. There is a small ratings boost for Ed but he continues to trail behind Dave. This doesn’t seem to have affected the voting numbers. On a general level the election is now only 20 days away and at some stage the Tories need that long hoped for cross over.

people are voting now.

 

[brogdale]

That's an England only 4.5% Con -> Lab swing. (In 2010 the vermin had a vote lead in England of 11.4%.)