Categories
Options
politicalbetting.com » Blog Archive » The great methodology divide: All the CON leads are from ph
politicalbetting.com » Blog Archive » The great methodology divide: All the CON leads are from phone polls – all but one of the LAB leads are from online surveys
The latest Ashcroft weekly phone poll is out and show a move back to CON and a 5% decline in the LAB vote. The figures and trend are in the chart above.
0
This discussion has been closed.
Comments
Mr Farage would simply be “Blighty!”, probably Margate or Southend because “they adore him down there” or (the view from Muswell Hill) “Benidorm.
Somewhere tacky and loud with egg and chips”. Mr Clegg would be somewhere “nice and inoffensive”, or possibly, since he must feel beleaguered, a distant location like the Caribbean “where people don’t know him so he won’t get hassled all the time”.
Mr Miliband? A place “where the traffic is terrible, because he doesn’t have any sense of direction.” Alternatively “the Moon, his own little world,” or more charitably “somewhere misunderstood – a really nice place but no-one goes there.”
11th Jan: Con +6
18th Jan: Con +1
25th Jan: Tie
1st Feb: Tie
8th Feb: Con +3
15th Feb: Lab +1
22nd Feb: Lab +4
1st Mar: Con +3
Could be current sitn is narrow tory lead: a point or so.
This is one of them "untested" pollsters at a GE, right?
The 15 Feb Ashcroft Monday poll was a ICM phone poll (we know this from OGH). ICM phone has been tested in a previous GE.
Are you seriously saying that Lord A is arbitrarily switching polling companies to produce what purports to be a uniform series of polls?
I think it was not, as Mike said it did not ask the VI question first, and Lord A's weekly polls asks VI first.
13 telephone polls is a reasonable sample. We can probably say with some confidence that with the telephone companies, crossover has now taken place.
@TSEofPB: The Tories have achieved an average lead with the phone pollsters http://t.co/lVNuhfNC47
https://www.ipsos-mori.com/Assets/Docs/Polls/feb2015web4_Qs11_15.pdf
Still miles behind in terms of what they need for a majority and probably struggling to get most seats on these numbers although the fundamentals have probably changed enough that they won't need quite as much as an 11% or so lead for a majority.
On balance still more upside for the Tories in the campaign, so I would probably go with most seats and most votes, but short of a majority, if predicting the outcome at this point.
Any ideas about the relative demographics of those responding by phone, and those more likely to respond online?
I'm inputting Con 32.7; Lab 33.3 (30 Scotland) Lib Dem 7.8 (5 Scotland); SNP (40 Scotland, 3.4% overall)
yields
Con 277
Lab 273
LD 25
SNP 48
If you look at the 2010 polls in terms of who was closest to the Con - Lab gap (clearly the most important feature of the polls) the phone polls gave an average Con lead of 8.4%, the online polls gave an average Con lead of 8.2%. Not much difference, but the online polls were closest. Excluding Angus Reid - which was a huge outlier - and the online performance improves to give a gap of 7.4%, just 0.1% out!
The problem that we have is that we now have new pollsters and changes in methodology so it is difficult to be confident about reading past performance into the future.
There are lots of reasons to think one method is likely to be more accurate than the other - we will only find out which is correct on May 8th.
One last point - if there is a difference in likelihood to vote this may be less of an issue for Labour in marginal seats, where the campaign is fought hard, than in 'safe' seats. If so it could help maintain the efficiency of the Labour vote.
https://en.wikipedia.org/wiki/United_States_presidential_election,_1936#Campaign
http://www.portsmouth.co.uk/news/politics/ukip-announces-its-contender-for-fareham-mp-1-6608081
Isn't she a councillor in Surrey?
I would surmise the reason is that internet panels will tend to attract people who are most strongly committed to politics.
And as a good "scientist", you know that accurate measurements needs to be backed up with knowing what the hell you are actually measuring?
Ghania Emwazi realised that the the knife-wielding executioner who appeared in video showing murder of US journalist James Foley was her son Mohammed "
http://www.telegraph.co.uk/news/worldnews/islamic-state/11444736/Mother-of-Mohammed-Emwazi-knew-he-was-Jihadi-John-from-the-outset.html
https://twitter.com/Sunil_P2/status/572053013654847488
That's still 3.6m votes.
The one thing the polls are showing at the moment is.
No one has "sealed the deal" for a large proportion of the electorate.
As for you absurd 'peak kipper ramblings' UKIP have peaked three times so far since the middle of 2012 and every time their vote share has remained higher once it has peaked than it was prior to the increase and it has then increased in the run up to the next election from that new base. Last year around at this time they were on a rough average of around 12. Now they are on 14. Thats hardly bad news for UKIP. The polls treatment of UKIP has nothing to do with my disaffection.
I could go on about how each of the sampling methods (phone/internet) excludes at least 9 million people or how panel based polling is questionable because panels are volunteer based and therefore have their own agenda for joining a polling panel (so the polls from them are not random per se) or indeed how the increasing regional variations in polling could undermine the national weighting approach of pollsters or indeed how many of the weighting systems do not handle step changes in voter opinion very well but I will not labour such points or about the validity of seemingly heavily weighted polls to get the right answer but I will not labour these points. Simply put I am increasingly sceptical about the integrity of polling simply because the circumstances have changed and the pollsters lack the wherewithal to adapt in a practical way.
That's my entirely unscientific analysis as well.
Given the confusion and pointless obscurantism on who is actually polling, the bouncy castle effect and the November Hallam error corrected in February, I'd like to see his Lordship taken out of the averages.
"The poll with the lowest Labour score is the correct one".
That's right, @Shadsy of this parish - as big a steer as any you'll find that that's the most likely result.
It's particularly annoying, because living in the unexpectedly marginal Solihull, I was trying to convince him to actually vote UKIP in order to deprive the tories of a vote, increasing the chances of the LD's holding on and the country returning a labour led government.
For full disclosure, I'm a member of the green party.
FPTP rocks!
"Six arrested after Carmarthen school child abuse claims" (Beeb)
"Try chanting it out loud. It sadly doesn't work with three syllables.."
Let's hope MORI come out with some unbelievable numbers then
http://www.sportingindex.com/spread-betting/politics/british/mm4.uk.meeting.5052257/uk-general-election-name-the-government
We should get TNS, either tomorrow or Wednesday, which will presumably show a large Labour lead.
http://www.theguardian.com/politics/2015/feb/26/ed-balls-plays-down-prospect-of-labour-deal-with-snp
The calculation is slightly different if you have a credit account though.
Presumably not 2010
http://www.bbc.co.uk/news/world-middle-east-31699632