All of the pollsters can

[Deleted]

Fuck right off.

Useless fucking cunts.

[Pete Whitehead]

Well the GB share they're showing now on BBC is more or less what the pollsters said, but I can't believe its correct

[Deleted]

As the Forum Pollster, I feel as though I should offer my resignation. I'm not with the others, honest...

In all seriousness, I think the pollsters made the mistake of focusing way too much on the potential national vote share, which evidently is of no relevance any more when you consider how much variation there is these days from one constituency to the next. All you have to go on is UNS. And we were bombarded with these VI polls on almost a daily basis over the past five years.
And Ashcroft made the error of failing to take into account the incumbency factor in his constituency polls, not that that held much sway in most parts of the country. It certainly did help incumbent Conservative MPs.

It is 1970 and 1992 all over again. When the polls are proved wrong like this, it's usually the Conservatives who end up pleasantly surprised. It's turning into a tradition and seems to happen every 20-25 years.

FWIW, here was our own Forum-wide result, which I removed from the General Election thread to ease the heavy traffic. There were a record 118 votes cast:

Lab 33
Con 23
Lib Dem 22
UKIP 15
Greens 9
SNP 5
Plaid Cymru 5
Communist 2
TUSC 1
Liberal 1
PSP 1
Pirates 1

[The Bishop]

The polls were right in Scotland of course <harrumph>

Isn't turnout only slightly up on 2010? That must be part of the explanation if so - many were predicting it to be up on 1997.

That would surely have benefited non-Tory parties in E & W - both Labour and UKIP especially were reliant on the support of non-voters last time.

[marksenior]

I don't think the national VI polls were that far out and the general error may be attributable to last minute changes of mind amongst voters .
The interpretation of the polls into a result forecast was awry .
The Ashcroft constituency polls can be considered a failure though .

[greenchristian]

May 8, 2015 11:24:48 GMT marksenior said:

I don't think the national VI polls were that far out and the general error may be attributable to last minute changes of mind amongst voters .
The interpretation of the polls into a result forecast was awry.

Agreed. And a decent model was pretty difficult to create.

The Ashcroft constituency polls can be considered a failure though.

True. Though I think Bristol West turned out to be reasonably accurate, so some of them were OK. Perhaps going through them all there's some room to work out lessons for future constituency polls.

[Davıd Boothroyd]

... hold a polling inquest rather like 1992.

www.britishpollingcouncil.org/general-election-7-may-2015/

[greatkingrat]

May 8, 2015 11:24:48 GMT marksenior said:

I don't think the national VI polls were that far out and the general error may be attributable to last minute changes of mind amongst voters .
The interpretation of the polls into a result forecast was awry .
The Ashcroft constituency polls can be considered a failure though .

It is difficult to rate the constituency polls as many were taken months before the election so wouldn't pick up any late swings. They did a pretty good job in Scotland of picking up the SNP surge.

The main errors were in the Con/LD marginals. What is interesting there is that the polls had two questions, one general voting intention, and one asking about your specific constituency. The second set of numbers were the ones most commonly reported, but the first question ended up much closer to the correct outcome

For example in Cornwall North
Headline VI - Con 40%, LD 29%
Constituency VI - LD 38%, Con 36%
Actual - Con 45%, LD 31%

[Richard Allen]

May 8, 2015 14:11:15 GMT greatkingrat said:

May 8, 2015 11:24:48 GMT marksenior said:

I don't think the national VI polls were that far out and the general error may be attributable to last minute changes of mind amongst voters .
The interpretation of the polls into a result forecast was awry .
The Ashcroft constituency polls can be considered a failure though .

It is difficult to rate the constituency polls as many were taken months before the election so wouldn't pick up any late swings. They did a pretty good job in Scotland of picking up the SNP surge.

The main errors were in the Con/LD marginals. What is interesting there is that the polls had two questions, one general voting intention, and one asking about your specific constituency. The second set of numbers were the ones most commonly reported, but the first question ended up much closer to the correct outcome

For example in Cornwall North
Headline VI - Con 40%, LD 29%
Constituency VI - LD 38%, Con 36%
Actual - Con 45%, LD 31%

But not in all cases, see Sheffield Hallam.

[Deleted]

Hats off to John Curtice by the way.

[Devil Wincarnate]

Survation claim that they had a poll indicating it, but dismissed it as an outlier:

survation.com/snatching-defeat-from-the-jaws-of-victory/

[Deleted]

May 8, 2015 14:53:09 GMT swanarcadian said:

Hats off to John Curtice by the way.

May he live forever

[Khunanup]

The main problem with polls is that they believe that people aren't lying to them. One of the lessons from this campaign is that people lie, and do so consistently in certain circumstances.

And the general public has the temerity to label the whole political class as dishonest...

[Deleted]

What I don't understand is why people lie to an online form, and indeed apparently lied to Yougov about how they voted on the day (but not to the exit pollsters).

It seems much likelier that online panels especially are just missing a certain cohort. The phone pollsters got it wrong too but were notably showing moves to the Tories from about Feb onwards and did register some big Tory leads during the campaign that were close to the final result.

[The Bishop]

I repeat, a large part of it was getting turnout wrong - none predicted it would be only 1% up on 2010 (so any "lies" may have been on whether they would vote, not how)

That surely benefited the Tories significantly......

[carlton43]

As you know I don't set much store by individual polls and doubt the efficacy of most of the weighting methodology; and I was well slagged off for being so undcientific as to say so! What I paid attention to was the Poll of Polls graph and that showed a constant movement by the Conservatives to close the gap and it was correct.

You chaps keep telling me these are tools and they are only snapshots NOT forecasts? So why are you concerned to outraged over the 'inaccuracy'. They may well have been right on the day? I think a lot of people changed their mind in the polling both at the thought of the big picture. They thought Cameron and Osborne were a safer pair of hands than Miliband and Balls (especially Balls) and they decided not to flirt with a good Labour candidate or UKIP but to swallow hard and vote Conservative. The polls were probably mainly right all the time. The exit poll was again right on the money because it was a large sample and a snapshot ON THE DAy of the vote.

[dizz]

I enjoyed the Constituency polling and hope it returns. Whilst it didn't pick-up the Lab to Con swing on (or close to) the day itself, it was an indicator of where Lab was doing better e.g. Chester & equally worse e.g. Somerset North East. The Bristol West poll was accurate & the Scottish ones in general too e.g the swing against Lab was greater in its safer seats. Also that Brighton Pavilion was not going to be gained etc.

[Tony Otim]

May 9, 2015 10:36:34 GMT dizz said:

I enjoyed the Constituency polling and hope it returns. Whilst it didn't pick-up the Lab to Con swing on (or close to) the day itself, it was an indicator of where Lab was doing better e.g. Chester & equally worse e.g. Somerset North East. The Bristol West poll was accurate & the Scottish ones in general too e.g the swing against Lab was greater in its safer seats. Also that Brighton Pavilion was not going to be gained etc.

The Scottish polling was better in the Lab-SNP seats than those involving the LDs - the LDs performed better vs polls in Gordon and Inverness but significantly worse in BRS.

[Deleted]

May 8, 2015 16:35:55 GMT doktorb🏳️‍🌈🏳️‍⚧️ said:

May 8, 2015 14:53:09 GMT swanarcadian said:

Hats off to John Curtice by the way.

May he live forever

I'd like him to join this forum.

[Andrew_S]

Interesting that there was an Ashcroft national poll from 24-26 April which gave Con 36%, Lab 30%, UKIP 11%, LD 9%. This was very close to the final result:

en.wikipedia.org/wiki/Opinion_polling_for_the_2015_United_Kingdom_general_election