All of the pollsters can

[The Bishop]

May 14, 2015 2:07:03 GMT manchesterman said:

Seems that the old "rule" still stands that a party cannot win where it is neither the "most trusted to take care of the economy" nor "has the best-suited PM". Parties can (and often do) win where they lead in one of the 2 but never win where they lead in neither.

Though it should be noted Blair trailed on the former in 1997, and wasn't that far ahead on the latter.

I honestly think other things were at least as important this time.

[woollyliberal]

May 11, 2015 7:34:33 GMT swindonlad said:

May 10, 2015 6:56:45 GMT hedgehog said:

I dont see how you could stop people carrying out a survey of voting intentions, however the number of national polls did seem crazy.

Just glancing at the last 3 constituency polls, Swindon South, Sheffield Hallam and Thanet South, it struck me how accurate they were.

I dont see where this pollster bashing is coming from, they were opinion polls, last minute switching or not bothering to go and vote cant be blamed on the pollsters.

Looking at South Swindon (which I know best) the last Ashcroft poll was C 37 L 36 (+1), final figures C46 L 35 (+11), some difference

Mind you at 10pm both parties were still thinking it was very close

I think it was a last minute decision, the reports coming back from the polling stations (not so much the VIs but the response to the tellers) in the marginal wards were good but it was not being picked up before that

Something similar happened in my ward. Canvass data was very close but the result wasn't. There were a lot of ballot papers where voters had started to vote for us then put a cross elsewhere. People were changing their minds at the last possible minute.

[woollyliberal]

May 14, 2015 10:27:17 GMT @trident said:

May 13, 2015 20:03:07 GMT cogload said:

Are we to see the end of the YouGov daily then?

If so, I won't miss it. I think that with polling, quantity has come at the expense of quality.

With YouGov, quantity had no bearing on quality. The methodology was the same, and therefore the quality. Daily weekly or monthly it would have made no difference. In fact quantity is YouGov's strength. It allows you to filter out noise and get to the "real" output.

The election was won and lost in part about a week before polling day and in part in the polling booth. Interestingly, Paddy Ashdown told a conference PPC meeting that this would happen. I noticed the change from the tone of canvassing on the doorstep. YouGov didn't pick this up. Survation did, but didn't publish. The monthly polls had less chance of identifying it. No poll in the world can detect a change of heart with pencil in hand.

[Arthur Figgis]

Man, man, your time is sand, your ways are leaves upon the sea...

[Robert Waller]

Woollyliberal's last sentence and Carlton's most recent post (on this thread) sum up my point above more articulately than I did (or perhaps had time to, on this occasion).
I believed that when I was a practising pollster - one of the reasons why I managed to win my bets that year despite being part of the pollster's 'Waterloo' - and still do, although poll techniques have improved vastly since my day.
This time, everything said to me that the C would win and Labour would get fewer seats than 2010 except for what I read. What I did not anticipate (that is, got wrong) was the Lib Dem collapse. That's why I lost money overall this time betting.

incidentally, I know in 1992 that the remaining error in the exit poll (for which I was chiefly responsible) was partly due to its difference from the pre-election polls, so we toned its simplistic seat forecast down from "C largest party and with a good shout of an overall majority" to fit the expected 'hung parliament' narrative. I do not know if anything like this happened this time, or whether this might account for the broadcasters' and commentators' reluctance to see the message of the real results that was so apparent early on.

[carlton43]

This has been a fascinating election (aren't they all though?) for the core structural reasons for success and failure. The analysts have suggested these components:-

1) Inaccurate polls causing mood and vote change
2) Shy Tories
3) Shy UKIP
4) SNP surge take Labour seats
5) SNP surge alters English voting intentions to support Conservatives and block Labour
6) LD vote collapse more extreme caused distortions to models
7) UKIP late fade
8) Lazy Labour voter syndrome.
9) LD lost votes not mainly transferred to Labour.

I think all of those were genuine inputs to the mix. Some were trivial, some important and only three critical.

A) The loss of Labour seats direct to SNP. Dished Labour's chance of outright win. But neutral to Conservative outright win.
B) The fear of SNP undue influence on minority Labour administration caused significant vote change to Conservative
C) The LD lost votes did not mainly go Labour with rest NOTA, but mainly UKIP and partially Conservative as well as Labour.

These feature coalesce to a simple and definite Conservative majority administration and lack of UKIP seats won, and more LD lost seats, and fewer Labour wins in England. The election proved a perfect scenario for SNP and Conservatives and few saw that in the making.

[woollyliberal]

May 14, 2015 13:19:20 GMT carlton43 said:

A) The loss of Labour seats direct to SNP. Dished Labour's chance of outright win. But neutral to Conservative outright win.
B) The fear of SNP undue influence on minority Labour administration caused significant vote change to Conservative
C) The LD lost votes did not mainly go Labour with rest NOTA, but mainly UKIP and partially Conservative as well as Labour.

These feature coalesce to a simple and definite Conservative majority administration and lack of UKIP seats won, and more LD lost seats, and fewer Labour wins in England. The election proved a perfect scenario for SNP and Conservatives and few saw that in the making.

Pretty much this.

[Richard Allen]

May 14, 2015 10:59:56 GMT woollyliberal said:

May 14, 2015 10:27:17 GMT @trident said:

If so, I won't miss it. I think that with polling, quantity has come at the expense of quality.

With YouGov, quantity had no bearing on quality. The methodology was the same, and therefore the quality. Daily weekly or monthly it would have made no difference. In fact quantity is YouGov's strength. It allows you to filter out noise and get to the "real" output.

The election was won and lost in part about a week before polling day and in part in the polling booth. Interestingly, Paddy Ashdown told a conference PPC meeting that this would happen. I noticed the change from the tone of canvassing on the doorstep. YouGov didn't pick this up. Survation did, but didn't publish. The monthly polls had less chance of identifying it. No poll in the world can detect a change of heart with pencil in hand.

This is why I deeply dislike postal voting on demand and any attempts to make voting "easier". Actively walking into a polling station on election day tends to concentrate the mind and make people think seriously about the decision they are making.

[mrhell]

All the data tables showed the LD vote was the least secure of the four parties. That's partly why the polls overestimated our vote. I just hoped throughout the campaign that it would be OK.

The polls showed a strong Tory lead vs Labour among 65+ voters (we are also weaker in that demographic). Given the differential turnout advantage vs younger voters that partly explains the Tories doing better than expected.

I couldn't understand the SNP performance given their weakness among 65+ voters. I now understand. Likelihood to vote figures in the data tables were lower among younger people in Scotland but nothing like the same as England. I can only assume that the high turnout of younger people in the referendum vote has carried over.

I don't think the polls were wrong but the analysis of them was wrong.

[Robert Waller]

I think Carlton's analysis above is very much on the right lines; the other six factors identified lack significant impact.
Mr Hell's point also seems pertinent. I also noted the age breaks in the poll cross tables. Did Scotland not have a turnout around 5% higher than the rest of Britain?

[carlton43]

May 14, 2015 18:47:44 GMT Robert Waller said:

I think Carlton's analysis above is very much on the right lines; the other six factors identified lack significant impact.
Mr Hell's point also seems pertinent. I also noted the age breaks in the poll cross tables. Did Scotland not have a turnout around 5% higher than the rest of Britain?

Thanks Robert.

I omitted one strand in
10)  Green Parties surge.

This was not critical as I see it but certainly a feature. I think they probably suffered a late fade along with UKIP when many focused on the main event of Con-Lab or Dave-Ed? I would not wish to think Green friends had been snubbed! Especially as it is more significant than at least three features I did list. It was a simple oversight first time round.

Richard's point on postal votes is interesting to add in to the meld. This must have been the highest postal vote with many cast and posted before the Election Endgame was in play. There are two features to that observation. One I had not sufficiently considered being the difference of voting as a communal event, in person, at a venue , on the appointed day, with all the attendant psychology of importance and serious joint venture....against a lesser civic duty, ahead of the event, in privacy of own home, with no particular imperatives of 'moment' and 'frisson'! Rather the difference of watching Arsenal live abroad on Sky opposed to in the stands at Arsenal! Witness my neighbours who had both promised a UKIP vote. His was duly delivered by post for UKIP two weeks early. She changed her mind on the day in the ballot box and tried to save Charlie Kennedy! Really fascinating.

Two...Many postal voters also missed the End Game of the pumped up Dave, the SNP fear angst, and the polarization. I would wager that UKIP and the Greens had a better proportion of the postals than the ballot boxes for that reason. If we had had an all personal ballot on the day, I think the Conservative majority and certainly the vote, would have been higher still.

[Andrew_S]

May 14, 2015 18:47:44 GMT Robert Waller said:

I think Carlton's analysis above is very much on the right lines; the other six factors identified lack significant impact.
Mr Hell's point also seems pertinent. I also noted the age breaks in the poll cross tables. Did Scotland not have a turnout around 5% higher than the rest of Britain?

Yes just about 5 points higher according to the BBC website.

[Andrew_S]

Lynton Crosby takes aim at the "Westminster village" including Tories:

www.telegraph.co.uk/news/general-election-2015/11608648/Lynton-Crosby-Betrayal-of-British-voters.html

"Lynton Crosby: Betrayal of British voters

Mr Cameron’s Australian election strategist suggests that opinion polls be banned during the final three weeks of future campaigns"

[carlton43]

May 15, 2015 22:47:54 GMT Andrew_S said:

Lynton Crosby takes aim at the "Westminster village" including Tories:

www.telegraph.co.uk/news/general-election-2015/11608648/Lynton-Crosby-Betrayal-of-British-voters.html

"Lynton Crosby: Betrayal of British voters

Mr Cameron’s Australian election strategist suggests that opinion polls be banned during the final three weeks of future campaigns"

Well the Strine Bas'Turd should mind his manners and mind his own business.

I believe in Free Speech and freedom of action. If I want to take polls, or to pay for polls or to publish polls that is my absolute right and it MUST not be breached. Politicians are allowed to use rhetoric and to be shall we say economical with the truth, newspapers may come out in favour of a party and write leaders exhorting us to support one party, people are permitted to bang on doors and solicit votes and too try and change opinion, and we strain at the gnat of 'perhaps a poll might influence opinion'? Well, yes it might. that is what the whole bloody campaign is about !!! But if the poll is conducted on a proper basis with all workings shown it is merely a bit of semi-scientific evidence...And why should we be denied that interest and that 'evidence' because of fears it might alter mood or vote. Everything else in the campaign is doing so as well. Do we ban official statistics, stock exchange reports, bad sports results and expressions by foreigners as well?

This is advanced after every election and must be bashed on the head until it be dead. 

[mrhell]

I haven't listened to this but there's a Polling Matters podcast with interviews with various pollsters www2.politicalbetting.com/index.php/archives/2015/05/16/ge2015-the-inquest-a-special-podcast-in-collaboration-with-polling-matters/

[manchesterman]

May 14, 2015 13:19:20 GMT carlton43 said:

This has been a fascinating election (aren't they all though?) for the core structural reasons for success and failure. The analysts have suggested these components:-

1) Inaccurate polls causing mood and vote change
2) Shy Tories
3) Shy UKIP
4) SNP surge take Labour seats
5) SNP surge alters English voting intentions to support Conservatives and block Labour
6) LD vote collapse more extreme caused distortions to models
7) UKIP late fade
8) Lazy Labour voter syndrome.
9) LD lost votes not mainly transferred to Labour.

I think all of those were genuine inputs to the mix. Some were trivial, some important and only three critical.

A) The loss of Labour seats direct to SNP. Dished Labour's chance of outright win. But neutral to Conservative outright win.
B) The fear of SNP undue influence on minority Labour administration caused significant vote change to Conservative
C) The LD lost votes did not mainly go Labour with rest NOTA, but mainly UKIP and partially Conservative as well as Labour.

These feature coalesce to a simple and definite Conservative majority administration and lack of UKIP seats won, and more LD lost seats, and fewer Labour wins in England. The election proved a perfect scenario for SNP and Conservatives and few saw that in the making.

I havent visited this thread since my return from the USA a few weeks ago. I think Carlton has hit many of the key factors why the pollsters seem to have all got it so badly wrong.

THere is one other possible element I would like to throw into the pot.  From what I recall, in many of the polls there were a fairly high number of "undecideds" "don't knows" etc - many of these just won't have voted but many of them would have decided how to vote in the final days of the campaign.  Is it not feasible that simply the vast majority of this group just voted Conservative?

In other words take a poll which was 34% Tory 34% Labour etc.  We had all been extrapolating from that that the result would be neck and neck between the 2 main parties. But say there were 12% dont knows, let's say4% of them didnt vote at all, but then 7% of the remaining 8% voted Tory with the other 1% scattered amongst the other parties; that alone would more or less account for the final result in terms of percentage lead for the Conservatives.

So maybe the polls were right in terms of those who were certain to vote for Party X (and maybe those voters didnt change at the last minute), but just assumed in their calculations that the undecideds would break evenly?

[The Bishop]

And, as mentioned before, both Tories and SNP got their vote out massively on the day. Other parties, less so.

[Deleted]

Don't forget the first time incumbency factor. The vast majority of Conservative MPs who first gained their seats in 2010 increased their majorities, This is the sort of thing that national opinion polls can't detect, and something Ashcroft didn't adequately address in his constituency polls.

[mick745]

I had felt for most of the last parliament that the only the conservatives would be capable of winning in 2015. This was based on just one factor cameron v miliband. However, i waivered in this view in the last few weeks before the election because we kept being told repeatedly it would be a hung parliament. I should have stuck to my instinct as there never really seemed to be a great appetite for change. There was always more of a '92' mood but the polls did not reflect what i was feeling. From now on i will follow my instincts and disregard the polls.

However, up to poling day i though lib dems would end up third in the popular vote. Got that wrong.

It has never been explained to me why conservative voters would be more reluctant to admit who they voted for than voters from other parties. I would have thought in every election voters supporting the government would be 'shyer' than those supporting the main opposition parties. Even in 2005, after the invasion of iraq, the polls called the labour victory okay.

[manchesterman]

People like to think of themselves as caring for others, altruistic etc.

Without getting into a debate with the left & right supporters on here, in very very broad terms, people vote Conservative out of self-interest, people vote Labour or LD and a few other parties, out of a sense of fairness and helping others. (I know this is very simplistic, but it is how many "non-political" people think about the parties). So people who voted say Labour when asked about it, proudly say they voted Labour with the accompanying warm glow of satisfaction. Those who vote Tory can only do so out of a sense of self-interest (please dont give me any of that 'trickle down' economics voodoo nonsense!) and, because everyone thinks of themselves as caring & altruistic, even if they arent, sometimes it is easier to tell the pollster you voted for some other party or are an 'undecided' - which brings me back to the point I made yesterday