How big will the Tory campaign bounce be?

[innocentabroad]

Governing parties usually recover somewhat during election campaigns - alas, I'm not wonkish enough to produce any statistics showing by how much - but the next one will be different in two ways. The outgoing government is a coalition, and the PM has surrendered his prerogative of calling the election in favour of a fixed date.

What movement in the polls do members of this forum expect during the campaign itself? If we can reach a consensus here on this question - and of course we may not be able to - then we can apply it to poll results as they appear.

[Tony Otim]

I'm not sure that fixed term parliaments will have all that much effect on how much the government recovers during the campaign.

I'd expect some movement back to the Conservatives but maybe not more than a point or two. NOt sure what will happen to the LDs (or to UKIP for that matter). A lot will depend on their position at the start of the campaign which will in turn depend a lot on if people are feeling that the economy is starting to recover.

[Deleted]

Very hard to say. With two five-point leads in a row I am increasingly confident that we will have the biggest share of the vote in 2015 though.

[Clarko]

I think it's far too early to say what the polls will look like in 18 months' time, but I'd expect a degree of 'leveling' between the two main parties and for the Lib Dems to recover somewhat (not to the degree of 2010, but 3 or 4 points up from the 10-12 points they're currently getting).

I wouldn't be awfully shocked if Labour and Conservative polled roughly the same - I still think it'll be another hung parliament.

[Deleted]

Jul 2, 2013 12:39:24 GMT @trident said:

Jul 2, 2013 12:21:31 GMT joe said:

Very hard to say. With two five-point leads in a row I am increasingly confident that we will have the biggest share of the vote in 2015 though.

Making such a sweeping prediction based on two YouGov polls is ridiculous - they aren't backed up by the other pollsters and the ballot box is telling a different story (just look at the weekly by-election threads).

This is all coming down to getting the vote out in the right places. In those places in general our voting is doing well. Here in the West Mids I expect us to win at least 10 seats if the election was held tomorrow.

Elections will all come down to events of course.

[russthomas]

Jul 2, 2013 11:59:02 GMT innocentabroad said:

What movement in the polls do members of this forum expect during the campaign itself? If we can reach a consensus here on this question - and of course we may not be able to - then we can apply it to poll results as they appear.

The May 2015 scenario is highly dependent upon a large number of big unknowns - like what UKIP are doing at the time of the run up to the election.

The 2014 Euro MEP and Urban Council elections on the same day, are the first crucial factor. Some newspapers say UKIP are favourite for most seats, maybe 25. Last poll I saw was LAB 23.7%, UKIP 23.3%, Undecided 22% - which could swing either way, but most probably LAB & UKIP 22 each, CON 13, LIB 6. It is tempting to expect a high UKIP MEP vote to reflect in the councillor seats, but the Urban constituencies are vastly different demographics to the Rural Councils in 2013. If LAB come out with most MEPs, and UKIP gain few councillors, this could well remove much of the media publicity that UKIP need to keep rolling all the way through to 2015, after the run of stories in 2014 about Romanian and Bulgarian immigrants.

The second big factor is how many seats UKIP attack, and what funds they have to spend. Last week Nigel said they were hoping for 50 to 60 MPs, who would hold the Balance of Power, like LIB did in 2010. Problem is, there are almost zero chances that there will be any such balanced positions. Barring any vast change in fortunes, LAB are looking at a safe win, with somewhere between 340 and 370 seats, with the bulk of the rest split between CON and UKIP. This could be anywhere between LAB 370, CON 210, UKIP 20, all the way to LAB 350, CON 120, UKIP 130. All parties will have figured this out by late 2014 - but what is not clear is Nigel/Paul tactics., Do they pour ALL their resources into contending 90 constituencies, virtually guaranteeing 60 wins by healthy margins. OR do they spread the troops thinner, seriously contend 200 seats with selected candidates, and hope for 140 wins by narrower margins, and become the opposition. Is Nigel being economical with the truth being relayed to the journalists, so that he can keep plying local newspapers with the same old "Our MP target" story until 2015, adding a few more onto the target each time ?

Other big factors:- Could the Tories get a leader who could re-unite the party ? Boris doesn't seem interested. Are there scandals yet to be released - if The Guardian have loads of GCHQ emails, are they going to drip feed stuff from them, like The Telegraph did with the expenses scandal ? Green shoots of economic recovery may yet arise - the recent GDP figures all look suspiciously flat line - maybe they are saving some goodies for a late surge. They go bananas about a few minus 0.1%'s. What will be the impact of a plus 0.3% ? (At any point in time, the most recent figures are highly dependent upon Treasury Estimates, plus or minus 0.3%, which later get corrected as actual data comes in.)

During the campaign, many of these factors will be having an influence, or just be coming into play. It seems way too early, just now, to see which will be important. Maybe this time next year things will be clearer.

[Crimson King]

All v interesting (although I disagree with your assesment that the activities of UKIP are the crucial factor) but not really answering the question about movement during the campaign.
Obviously events (dear boy) are crucial but if we discount the possibility of events during the campaign then I would agree that Lab and Con (presuming they are at 1st and second place in the polls in whatever order) will move up and down by a few points to become closer together, I suspect the further apart they start the bigger the movement will be.
If they start closer than 2% apart then a switch or movement apart would be possible.
I would expect the Lib Dems to move up during the campaign but perhaps as not as much as the traditional (ie before last time) gain.
However it is unlikely that (given the big differences between different pollsters) that we will be able to agree on what 'the polls' say at the start of the campaign, so it is unlikely that current speculation will be very helpful

[Deleted]

I'm not convinced the election campaign in itself will necessarily be a game changer for the Conservatives - there are two years for the party to recover from mid-term blues.

Some food for thought: at this point in the previous Parliament - July 2008 - the average voting intention figures were Con 45, Lab 26, LD 17. In the general election, the actual GB vote share was Con 37, Lab 30, LD 24. It is very tempting simply to extrapolate these changes onto current poll ratings to get our prediction for 2015, but obviously it is not as straightforward as that. This time, we have the impact of UKIP to consider. Their ratings might have slipped back a little from 8 weeks ago but we don't know how great the impact the European Parliament elections will be. Back in 2008 they were only polling 4 or 5% on a good day.

I agree with those who say it's all down to events. In the last Parliament, I believe the expenses scandal and Cleggmania had an impact which otherwise would have kept the Lib Dem share at about 18% rather than 24, and the Tories and Labour up 2 or 3 points apiece.

[Deleted]

Jul 2, 2013 12:39:24 GMT @trident said:

Jul 2, 2013 12:21:31 GMT joe said:

Very hard to say. With two five-point leads in a row I am increasingly confident that we will have the biggest share of the vote in 2015 though.

Making such a sweeping prediction based on two YouGov polls is ridiculous - they aren't backed up by the other pollsters and the ballot box is telling a different story (just look at the weekly by-election threads).

I trust polls far more than a few scattered by-elections. Its pretty clear the gap has been closing in said polls even if the true lead is over 5. 

The only other one out recently ive seen is opinium on Lab+10 whose method is over-friendly to UKIP on a rather implausible 19.

[Devonian]

Jul 2, 2013 13:48:24 GMT russthomas said:

The 2014 Euro MEP and Urban Council elections on the same day, are the first crucial factor. Some newspapers say UKIP are favourite for most seats, maybe 25. Last poll I saw was LAB 23.7%, UKIP 23.3%, Undecided 22% - which could swing either way, but most probably LAB & UKIP 22 each, CON 13, LIB 6. It is tempting to expect a high UKIP MEP vote to reflect in the councillor seats, but the Urban constituencies are vastly different demographics to the Rural Councils in 2013. If LAB come out with most MEPs, and UKIP gain few councillors, this could well remove much of the media publicity that UKIP need to keep rolling all the way through to 2015, after the run of stories in 2014 about Romanian and Bulgarian immigrants.

The second big factor is how many seats UKIP attack, and what funds they have to spend. Last week Nigel said they were hoping for 50 to 60 MPs, who would hold the Balance of Power, like LIB did in 2010. Problem is, there are almost zero chances that there will be any such balanced positions. Barring any vast change in fortunes, LAB are looking at a safe win, with somewhere between 340 and 370 seats, with the bulk of the rest split between CON and UKIP. This could be anywhere between LAB 370, CON 210, UKIP 20, all the way to LAB 350, CON 120, UKIP 130. All parties will have figured this out by late 2014 - but what is not clear is Nigel/Paul tactics

I rather suspect that Farage and the UKIP leadership are not clear on what their tactics will be. He has said that the plan is to use bases in local government to target seats so everything depends on how well they do in the local elections next year. If they fall flat in the local elections their targets will be much smaller in number. If they do better than expected that would give them huge extra momentum to target many more seats. I don't get the impression that Farage intends to work out the details of the strategy until after those elections.

I would agree that the inner city council elections in London, Birmingham are likely to prove slim pickings for UKIP (and indeed any party that isn't Labour). However remember that there will be a huge number of district council seats up for grabs as well urban seats in the likes of outer London, Essex, North east Lincolnshire etc where they're likely to have more of a chance. These will then give UKIP its target seat list. In particular where they gain district seats in the same Westminster constituency as they already have County seats will be likely targets.

[innocentabroad]

The replies so far seem to suggest (once I scrape away the partisan ramping) that we are in indeed in uncharted territory.

[russthomas]

Jul 2, 2013 16:11:36 GMT joe said:

The only other one out recently ive seen is Opinium on Lab+10 whose method is over-friendly to UKIP on a rather implausible 19.

1) Opinium explicitly concede that their poll is severely unfriendly to UKIP and LAB, and over friendly to CON and especially LIB - in that they redistribute undecideds using a high proportion of last election %. UKIP had 3%, LIB 23% in 2010, there is no logic whatsoever to apply these same %s again.
2) Opinium also embed a considerable bias against parties outside CONLIBLAB - they are not included in the first choice options. A responder has to specifically name GRN, SNP, BNP, UKIP whoever, in order to give them a vote.
3) There is a bias against voters uncertain whether they will be able to get to a polling station on the day, anyone less than 50% certain is discounted entirely. This undervalues the votes of the poor, elderly, disabled and those living at a distance from the nearest polling station - which from age, health and income demographics would more likely be LAB, UKIP, SNP and PC voters, rather than CON or LIB.
4) Almost all polls fail to handle the silly protest element - MRLP and pals regularly clock over 5% in total in ballots, but very rarely does anyone admit that this is what they will do, to a pollster. People intending to protest often often pick a really tiny party for polls, but register their silly protest in the ballot box, so in polls somewhat overstating MIN candidates and severely understating OTH.  We haven't yet had a GE where UKIP were possible seat winners, so the likelihood of former MRLP votes now switching to UKIP is unknown, but not improbable.
5) Switchers from LIB protest votes to UKIP are probably already reflected in the polls, but all polls also fail to take into account the number of people who are forced to vote tactically. In the past this has usually meant choosing the most likely winner out of LAB or LIB to ensure that CON cannot win - but now LIB have thrown away those tactical voters, who are people who vote Anti CON going to choose ? LAB and UKIP will probably benefit equally, but each will now win seats, that the natural poll % would place them second. So the CON poll % is an overestimate of seat winning power.
6) The dramatic fragmentation of CON supporters between i) Toffs and Chums. ii) In EU rural. iii) Out of EU urban. iv) Immigrant worried, but not yet switched to UKIP in the polls ... now means that a CON national poll % is only going to fully apply if a Tory candidate precisely fits the seat demographic. A Europhile is not going to reach the national poll % in an urban seat, or a wealthy leafy suburb. Similarly a Eurosceptic won't appeal to usual Tory voters in a rural seat. It is nigh on impossible to pick a candidate that all local Tory party faithful will vote for... unless CON manage to find a leader who can re-unite the party.

Overall I would venture that the above reasons indicate that Opinium polls (and Survation) undercount LAB and UKIP by about 1% each, and overcount CON and LIB by about 1% each, from what people would vote for, if there were an election tomorrow - which is what they attempt to measure. Obviously there are the mid term blues to consider, there may be a bounce back in the election, but there are also Cash for Questions bribery scandals to come, which will have the opposite effect.

[russthomas]

Jul 2, 2013 11:59:02 GMT innocentabroad said:

Governing parties usually recover somewhat during election campaigns - alas, I'm not wonkish enough to produce any statistics showing by how much - but the next one will be different in two ways. The outgoing government is a coalition, and the PM has surrendered his prerogative of calling the election in favour of a fixed date.

What movement in the polls do members of this forum expect during the campaign itself? If we can reach a consensus here on this question - and of course we may not be able to - then we can apply it to poll results as they appear.

Pure bounce back would transfer about 1% back from UKIP to CON. LAB would be unaffected, because the mid term anti coalition feeling is expressed in the polls by switching to an almost Tory light blue protest type party. LIB would gain nothing because that protest switch is permanent

But this effect will almost certainly be massively overpowered by the effect of CON fragmentation during the campaign itself - nowhere can pick a Tory candidate that can collect the votes of all the local party faithful. At very best three quarters of national poll % Tory support in any one seat will refuse to support a candidate with views totally opposite to theirs. A reasonable estimate of this effect in terms of seat winning in marginals is:-

CON -6%, LAB +2%, LIB +1%, UKIP +2%, NAT +1%

Where CON had 40% plus in 2010 will remain CON, but on slimmer majorities.

[Deleted]

wasn;t the last marginal poll Ashcroft did very damaging to Tories though when compared with nationwide polls ?

[Devonian]

Jul 2, 2013 19:07:35 GMT russthomas said:

Jul 2, 2013 16:11:36 GMT joe said:

The only other one out recently ive seen is Opinium on Lab+10 whose method is over-friendly to UKIP on a rather implausible 19.

1) Opinium explicitly concede that their poll is severely unfriendly to UKIP and LAB, and over friendly to CON and especially LIB - in that they redistribute undecideds using a high proportion of last election %. UKIP had 3%, LIB 23% in 2010, there is no logic whatsoever to apply these same %s again.

I think that this is a very important point. Looking at the details of some recent opinion polls I notice that the percentages of 2010 Lib Dem voters who fall into the don't know category is much higher that the percentage of the Conservative and Labour 2010 who fall into the same category. It would be reasonable to assume that a good portion of those people are thinking along the lines of those people you see commenting on newspaper message boards 'I voted Lib Dem in 2010 NEVER AGAIN'.

Added to that the collapse of the Lib Dem vote may very well be having a significant effect on the percentage of people simply lying about who they voted for in 2010. Left wingers who voted LD in 2010 and who feel tricked and might simply claim they voted Labour or did not vote. Lying to pollsters may also be effecting the UKIP vote as the deeply unfashionable UKIP suffer from their own version of the 'Shy Tory Effect'. It doesn't seem unlikely that these factors combined add up to several percentage points in the polls making the predictions of election results much more difficult.

[Deleted]

The YG polls are interesting as the Labour vote not moved but the Tories seen a recovery, is this because as we discussed before the no prompting for UKIP ?

my guess is that in reality we still at 38/30 for Lab/Con nothing has really happened to see the Tories move around 5% recently

[Tony Otim]

Jul 2, 2013 20:00:20 GMT Devonian said:

Jul 2, 2013 19:07:35 GMT russthomas said:

1) Opinium explicitly concede that their poll is severely unfriendly to UKIP and LAB, and over friendly to CON and especially LIB - in that they redistribute undecideds using a high proportion of last election %. UKIP had 3%, LIB 23% in 2010, there is no logic whatsoever to apply these same %s again.

I think that this is a very important point. Looking at the details of some recent opinion polls I notice that the percentages of 2010 Lib Dem voters who fall into the don't know category is much higher that the percentage of the Conservative and Labour 2010 who fall into the same category. It would be reasonable to assume that a good portion of those people are thinking along the lines of those people you see commenting on newspaper message boards 'I voted Lib Dem in 2010 NEVER AGAIN'.

Added to that the collapse of the Lib Dem vote may very well be having a significant effect on the percentage of people simply lying about who they voted for in 2010. Left wingers who voted LD in 2010 and who feel tricked and might simply claim they voted Labour or did not vote. Lying to pollsters may also be effecting the UKIP vote as the deeply unfashionable UKIP suffer from their own version of the 'Shy Tory Effect'. It doesn't seem unlikely that these factors combined add up to several percentage points in the polls making the predictions of election results much more difficult.

I may be wrong, but I thought this was actually working the other way round - pollsters were picking up more people who said they voted LD in 2010 than actually did. In other words, possibly left leaning voters who considered voting LD, actually stuck with Labour, but now say they voted LD in order to be able to say they'd never do it again.

And are UKIP that unfashionable - they seem to be all the media want to talk about half the time.

[Deleted]

Jul 2, 2013 19:07:35 GMT russthomas said:

Jul 2, 2013 16:11:36 GMT joe said:

The only other one out recently ive seen is Opinium on Lab+10 whose method is over-friendly to UKIP on a rather implausible 19.

1) Opinium explicitly concede that their poll is severely unfriendly to UKIP and LAB, and over friendly to CON and especially LIB - in that they redistribute undecideds using a high proportion of last election %. UKIP had 3%, LIB 23% in 2010, there is no logic whatsoever to apply these same %s again.
2) Opinium also embed a considerable bias against parties outside CONLIBLAB - they are not included in the first choice options. A responder has to specifically name GRN, SNP, BNP, UKIP whoever, in order to give them a vote.
3) There is a bias against voters uncertain whether they will be able to get to a polling station on the day, anyone less than 50% certain is discounted entirely. This undervalues the votes of the poor, elderly, disabled and those living at a distance from the nearest polling station - which from age, health and income demographics would more likely be LAB, UKIP, SNP and PC voters, rather than CON or LIB.
4) Almost all polls fail to handle the silly protest element - MRLP and pals regularly clock over 5% in total in ballots, but very rarely does anyone admit that this is what they will do, to a pollster. People intending to protest often often pick a really tiny party for polls, but register their silly protest in the ballot box, so in polls somewhat overstating MIN candidates and severely understating OTH.  We haven't yet had a GE where UKIP were possible seat winners, so the likelihood of former MRLP votes now switching to UKIP is unknown, but not improbable.
5) Switchers from LIB protest votes to UKIP are probably already reflected in the polls, but all polls also fail to take into account the number of people who are forced to vote tactically. In the past this has usually meant choosing the most likely winner out of LAB or LIB to ensure that CON cannot win - but now LIB have thrown away those tactical voters, who are people who vote Anti CON going to choose ? LAB and UKIP will probably benefit equally, but each will now win seats, that the natural poll % would place them second. So the CON poll % is an overestimate of seat winning power.
6) The dramatic fragmentation of CON supporters between i) Toffs and Chums. ii) In EU rural. iii) Out of EU urban. iv) Immigrant worried, but not yet switched to UKIP in the polls ... now means that a CON national poll % is only going to fully apply if a Tory candidate precisely fits the seat demographic. A Europhile is not going to reach the national poll % in an urban seat, or a wealthy leafy suburb. Similarly a Eurosceptic won't appeal to usual Tory voters in a rural seat. It is nigh on impossible to pick a candidate that all local Tory party faithful will vote for... unless CON manage to find a leader who can re-unite the party.

Overall I would venture that the above reasons indicate that Opinium polls (and Survation) undercount LAB and UKIP by about 1% each, and overcount CON and LIB by about 1% each, from what people would vote for, if there were an election tomorrow - which is what they attempt to measure. Obviously there are the mid term blues to consider, there may be a bounce back in the election, but there are also Cash for Questions bribery scandals to come, which will have the opposite effect.

1) They may say that (where?) but Opinium is slightly more friendly to labour and far more friendly to ukip than average (See PollingReport current home page).  Redistribution has been proven to work in both high (1997+2010) and low swing elections (2001, 2005 to a certain extent) and improve the model.

Any serious modelling will help the government as compared to the raw data, whoever is in government. If you want raw data rather than a manipulation fair enough, but these manipulations have been proven to help predictions and surely predicting the future is why polls are commissioned.

2) This is a way of modelling the inevitable GE squeeze. Not correcting for this at all would be also be damaging for the model.

3) Given that we should expect the turnout to be 80 odd percent from the polls and we see nowhere near this in reality once again it is necessary to alter the model to try to match reality. Those who say they probably wont vote almost certainly wont.

4) MRLP at 5%? Um, where? They don't stand in very many seats and tend to be well under 1%. This is a very random comment and total howling at the moon nonsense.

5) If you previously voted LD and now say you will vote UKIP or labour this is in the model already. Your point?

6) Nonsense, or at the very least applies equally to labour or the LDs (but not much to any party).

[Tony Otim]

Jul 2, 2013 19:07:35 GMT russthomas said:

3) There is a bias against voters uncertain whether they will be able to get to a polling station on the day, anyone less than 50% certain is discounted entirely. This undervalues the votes of the poor, elderly, disabled and those living at a distance from the nearest polling station - which from age, health and income demographics would more likely be LAB, UKIP, SNP and PC voters, rather than CON or LIB.

A lot of elderly and disabled voters will have postal votes which they will use and so won't be screened out by the certainty to vote criteria.

It's also true that poorer voters are less likely to vote at the moment, so a certain amount of screening out makes the poll more accurate rather than less.

[Deleted]

Jul 2, 2013 19:07:35 GMT russthomas said:

6) The dramatic fragmentation of CON supporters between i) Toffs and Chums. ii) In EU rural. iii) Out of EU urban. iv) Immigrant worried, but not yet switched to UKIP in the polls ... now means that a CON national poll % is only going to fully apply if a Tory candidate precisely fits the seat demographic. A Europhile is not going to reach the national poll % in an urban seat, or a wealthy leafy suburb. Similarly a Eurosceptic won't appeal to usual Tory voters in a rural seat. It is nigh on impossible to pick a candidate that all local Tory party faithful will vote for... unless CON manage to find a leader who can re-unite the party.

Genuinely curious -- is there evidence, anecdotal or otherwise, to suggest that (sub)urban Conservative voters are substantially more anti-EU than rural Conservative voters? If anything, that's the opposite of what I would have expected, though I realise that Edinburgh Tories might well be a class apart from the rest.