Our most trusted polling organisation

[Deleted]

I've only specifically included polling organisations which have reported voting intention figures in the past three months, but to be accommodating I have included an others option. I'm not sure who to vote for yet, but will do so after some deliberation.

[Andrew_S]

I can't recall which organisation was closest to the 2010 election result but whichever it was would win my vote.

[Sibboleth]

Mar 25, 2013 21:55:24 GMT Andrew_S said:

I can't recall which organisation was closest to the 2010 election result but whichever it was would win my vote.

Some Indian outfit or other. Which says everything.

[Deleted]

Yougov for the current national opinion. ICM for getting the final result correct.

[Deleted]

Final polls by polling organisation before the 2010 general election, with differences from the actual GB vote share:

Ipsos-MORI: C 36 (-1), Lab 29 (-1), LD 27 (+3)
YouGov: C 35 (-2), Lab 28 (-2), LD 28 (+4)
Harris: C 35 (-2), Lab 29 (-1), LD 27 (+3)
Populus: C 37 (=), Lab 29 (-1), LD 27 (+3)
Angus Reid: C 36 (-1), LD 29 (+5), Lab 24 (-6)
Opinium: C 35 (-2), Lab 27 (-3), LD 26 (+2)
ComRes: C 37 (=), Lab 28 (-2), LD 28 (+4)
ICM: C 36 (-1), Lab 28 (-2), LD 26 (+2)
TNS-BMRB: C 33 (-4), LD 29 (+5), Lab 27 (-3)
BPIX: C 34 (-3), LD 30 (+6), Lab 27 (-3)

Actual GB vote share: C 37, Lab 30, LD 24
Actual UK vote share: C 36, Lab 29, LD 23

These figures should not necessarily be taken as gospel, of course. Subsequent methodology changes and accuracy in other elections ought to be taken into account.

[Sibboleth]

I bet half of them just make their numbers up anyway. There have been numerous scandals of this sort in the U.S.A.

[Sibboleth]

Of course it's easily done as political types are the most gullible people on the planet.

[Deleted]

Mar 26, 2013 0:41:41 GMT swanarcadian said:

Final polls by polling organisation before the 2010 general election, with differences from the actual GB vote share:

Ipsos-MORI: C 36 (-1), Lab 29 (-1), LD 27 (+3)
YouGov: C 35 (-2), Lab 28 (-2), LD 28 (+4)
Harris: C 35 (-2), Lab 29 (-1), LD 27 (+3)
Populus: C 37 (=), Lab 29 (-1), LD 27 (+3)
Angus Reid: C 36 (-1), LD 29 (+5), Lab 24 (-6)
Opinium: C 35 (-2), Lab 27 (-3), LD 26 (+2)
ComRes: C 37 (=), Lab 28 (-2), LD 28 (+4)
ICM: C 36 (-1), Lab 28 (-2), LD 26 (+2)
TNS-BMRB: C 33 (-4), LD 29 (+5), Lab 27 (-3)
BPIX: C 34 (-3), LD 30 (+6), Lab 27 (-3)

Actual GB vote share: C 37, Lab 30, LD 24
Actual UK vote share: C 36, Lab 29, LD 23

These figures should not necessarily be taken as gospel, of course. Subsequent methodology changes and accuracy in other elections ought to be taken into account.

interesting figures in that the LD was all over the shop but the others fairly accurate.

As to the poll I can not answer As I simply trust phone polls more than internet ones.

[Deleted]

For me it's very close between ICM and YouGov. I have voted ICM but could possibly be persuaded to change my mind.

[marksenior]

It can only be ICM . Yougov as an online pollster are ruled out as their constant sampling errors are never fully corrected in the subsequent weighting adjustments . Ipsos Mori could be considered in contention if their final VI figure based on only those absolutely certain to vote is ignored and you take the figures in the data tables for those with 6-10 or 8-10 certainty to vote . However these will tend to overstate Labour as unlike ICM they do not use past vote weighting to correct for the tendency of phone polls to over sample Labour voters .

[Tony Otim]

Where's the option for whichever one tells me what I want to hear/what I think the situation really is

[Deleted]

Mar 26, 2013 9:51:41 GMT marksenior said:

It can only be ICM . Yougov as an online pollster are ruled out as their constant sampling errors are never fully corrected in the subsequent weighting adjustments . Ipsos Mori could be considered in contention if their final VI figure based on only those absolutely certain to vote is ignored and you take the figures in the data tables for those with 6-10 or 8-10 certainty to vote . However these will tend to overstate Labour as unlike ICM they do not use past vote weighting to correct for the tendency of phone polls to over sample Labour voters .

However in the example above both ICM and YG were almost similar and both over estimated you guys by the same amount. IF that is still continuing then I would be worried for the current poll.s

[The Bishop]

My answer to this is "anybody but ComRes"

(I see that nobody here has bothered posting their latest - very telling)

[marksenior]

Mar 26, 2013 10:17:40 GMT @ianrobo said:

Mar 26, 2013 9:51:41 GMT marksenior said:

It can only be ICM . Yougov as an online pollster are ruled out as their constant sampling errors are never fully corrected in the subsequent weighting adjustments . Ipsos Mori could be considered in contention if their final VI figure based on only those absolutely certain to vote is ignored and you take the figures in the data tables for those with 6-10 or 8-10 certainty to vote . However these will tend to overstate Labour as unlike ICM they do not use past vote weighting to correct for the tendency of phone polls to over sample Labour voters .

However in the example above both ICM and YG were almost similar and both over estimated you guys by the same amount. IF that is still continuing then I would be worried for the current poll.s

That depends on whether the over estimate in the LD's in 2010 was caused by sampling errors or last minute decisions by voters to change their vote or not vote at all . A very late swing or change can never be picked up by pollsters .
Incidently , your statement that ICM and Yougov were almost similar and both over estimated LD's by the same amount is factually incorrect and probably wishful thinking on your part . Yougov over estimated LD by 4 and ICM by 2 .

[Deleted]

it does seem that there was a late swing of LD-LAB ( around 2%) which was enough it seems to cause the hung parliament

[marksenior]

Mar 26, 2013 0:41:41 GMT swanarcadian said:

Final polls by polling organisation before the 2010 general election, with differences from the actual GB vote share:

Ipsos-MORI: C 36 (-1), Lab 29 (-1), LD 27 (+3)
YouGov: C 35 (-2), Lab 28 (-2), LD 28 (+4)
Harris: C 35 (-2), Lab 29 (-1), LD 27 (+3)
Populus: C 37 (=), Lab 29 (-1), LD 27 (+3)
Angus Reid: C 36 (-1), LD 29 (+5), Lab 24 (-6)
Opinium: C 35 (-2), Lab 27 (-3), LD 26 (+2)
ComRes: C 37 (=), Lab 28 (-2), LD 28 (+4)
ICM: C 36 (-1), Lab 28 (-2), LD 26 (+2)
TNS-BMRB: C 33 (-4), LD 29 (+5), Lab 27 (-3)
BPIX: C 34 (-3), LD 30 (+6), Lab 27 (-3)

Actual GB vote share: C 37, Lab 30, LD 24
Actual UK vote share: C 36, Lab 29, LD 23

These figures should not necessarily be taken as gospel, of course. Subsequent methodology changes and accuracy in other elections ought to be taken into account.

I believe the last Populus poll was 37/28/27

[Tony Otim]

Mar 26, 2013 10:49:32 GMT @ianrobo said:

it does seem that there was a late swing of LD-LAB ( around 2%) which was enough it seems to cause the hung parliament

There would probably still have been a hung parliament. The extra seat the Conservatives would have taken off Labour would have been balanced by a few they wouldn't have taken off the LDs.

[Sibboleth]

If that were the case then LibDem performance in constituencies where postal voters make up an unusual high share of the electorate ought to have been significantly better than elsewhere. This was not the case. 'Cleggmania' (hahahahaha: explaining that to people in the future is going to be so much fun) never actually happened.

[Andrew_S]

ICM is winning at the moment. Just checked on their latest poll: it was Lab 39%, Con 31%, LD 15%, UKIP 7%. That's UKIP's worst showing in any recent poll.

[Deleted]

Mar 26, 2013 19:41:05 GMT Andrew_S said:

ICM is winning at the moment. Just checked on their latest poll: it was Lab 39%, Con 31%, LD 15%, UKIP 7%. That's UKIP's worst showing in any recent poll.

Infinitely more credible than the 17% we saw recently though.